2019-01-02 06:33:26 -05:00
|
|
|
// Copyright 2019 The Hugo Authors. All rights reserved.
|
2013-07-04 11:32:55 -04:00
|
|
|
//
|
2015-11-23 22:16:36 -05:00
|
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
2013-07-04 11:32:55 -04:00
|
|
|
// you may not use this file except in compliance with the License.
|
|
|
|
// You may obtain a copy of the License at
|
2015-11-23 22:16:36 -05:00
|
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
2013-07-04 11:32:55 -04:00
|
|
|
//
|
|
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
// See the License for the specific language governing permissions and
|
|
|
|
// limitations under the License.
|
|
|
|
|
|
|
|
package hugolib
|
|
|
|
|
|
|
|
import (
|
2015-06-21 07:08:30 -04:00
|
|
|
"bytes"
|
2023-02-11 10:20:24 -05:00
|
|
|
"context"
|
2023-10-25 05:31:01 -04:00
|
|
|
"errors"
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
"fmt"
|
2020-12-02 07:23:25 -05:00
|
|
|
"html/template"
|
|
|
|
"path"
|
|
|
|
"reflect"
|
|
|
|
"regexp"
|
|
|
|
"sort"
|
2019-01-02 06:33:26 -05:00
|
|
|
"strconv"
|
2020-12-02 07:23:25 -05:00
|
|
|
"strings"
|
|
|
|
"sync"
|
2019-01-02 06:33:26 -05:00
|
|
|
|
2018-11-26 05:01:27 -05:00
|
|
|
"github.com/gohugoio/hugo/common/herrors"
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
"github.com/gohugoio/hugo/common/types"
|
2018-10-21 06:20:21 -04:00
|
|
|
|
2018-10-17 06:57:09 -04:00
|
|
|
"github.com/gohugoio/hugo/parser/pageparser"
|
2019-01-02 06:33:26 -05:00
|
|
|
"github.com/gohugoio/hugo/resources/page"
|
2018-10-17 06:57:09 -04:00
|
|
|
|
2018-07-06 08:12:10 -04:00
|
|
|
"github.com/gohugoio/hugo/common/maps"
|
2018-11-01 06:28:30 -04:00
|
|
|
"github.com/gohugoio/hugo/common/text"
|
|
|
|
"github.com/gohugoio/hugo/common/urls"
|
2017-06-13 12:42:45 -04:00
|
|
|
"github.com/gohugoio/hugo/output"
|
2017-05-06 14:15:28 -04:00
|
|
|
|
2017-06-13 12:42:45 -04:00
|
|
|
bp "github.com/gohugoio/hugo/bufferpool"
|
|
|
|
"github.com/gohugoio/hugo/tpl"
|
2014-03-31 13:23:34 -04:00
|
|
|
)
|
2013-07-04 11:32:55 -04:00
|
|
|
|
2018-11-01 06:28:30 -04:00
|
|
|
var (
|
2024-11-13 05:07:32 -05:00
|
|
|
_ urls.RefLinker = (*ShortcodeWithPage)(nil)
|
|
|
|
_ types.Unwrapper = (*ShortcodeWithPage)(nil)
|
|
|
|
_ text.Positioner = (*ShortcodeWithPage)(nil)
|
|
|
|
_ maps.StoreProvider = (*ShortcodeWithPage)(nil)
|
2018-11-01 06:28:30 -04:00
|
|
|
)
|
|
|
|
|
2016-03-24 09:11:04 -04:00
|
|
|
// ShortcodeWithPage is the "." context in a shortcode template.
|
2013-07-04 11:32:55 -04:00
|
|
|
type ShortcodeWithPage struct {
|
2022-03-17 17:03:27 -04:00
|
|
|
Params any
|
2015-11-20 19:59:54 -05:00
|
|
|
Inner template.HTML
|
2019-01-02 06:33:26 -05:00
|
|
|
Page page.Page
|
2016-03-08 14:56:24 -05:00
|
|
|
Parent *ShortcodeWithPage
|
2018-12-21 03:51:15 -05:00
|
|
|
Name string
|
2015-11-20 19:59:54 -05:00
|
|
|
IsNamedParams bool
|
2018-04-23 02:09:56 -04:00
|
|
|
|
2018-05-27 15:34:05 -04:00
|
|
|
// Zero-based ordinal in relation to its parent. If the parent is the page itself,
|
|
|
|
// this ordinal will represent the position of this shortcode in the page content.
|
2018-04-23 02:09:56 -04:00
|
|
|
Ordinal int
|
|
|
|
|
2022-05-30 14:42:46 -04:00
|
|
|
// Indentation before the opening shortcode in the source.
|
|
|
|
indentation string
|
|
|
|
|
|
|
|
innerDeindentInit sync.Once
|
|
|
|
innerDeindent template.HTML
|
|
|
|
|
2018-11-01 05:39:44 -04:00
|
|
|
// pos is the position in bytes in the source file. Used for error logging.
|
|
|
|
posInit sync.Once
|
|
|
|
posOffset int
|
2018-11-01 06:28:30 -04:00
|
|
|
pos text.Position
|
2018-11-01 05:39:44 -04:00
|
|
|
|
2024-11-13 05:07:32 -05:00
|
|
|
store *maps.Scratch
|
2013-07-04 11:32:55 -04:00
|
|
|
}
|
|
|
|
|
2022-05-30 14:42:46 -04:00
|
|
|
// InnerDeindent returns the (potentially de-indented) inner content of the shortcode.
|
|
|
|
func (scp *ShortcodeWithPage) InnerDeindent() template.HTML {
|
|
|
|
if scp.indentation == "" {
|
|
|
|
return scp.Inner
|
|
|
|
}
|
|
|
|
scp.innerDeindentInit.Do(func() {
|
|
|
|
b := bp.GetBuffer()
|
|
|
|
text.VisitLinesAfter(string(scp.Inner), func(s string) {
|
|
|
|
if strings.HasPrefix(s, scp.indentation) {
|
|
|
|
b.WriteString(strings.TrimPrefix(s, scp.indentation))
|
|
|
|
} else {
|
|
|
|
b.WriteString(s)
|
|
|
|
}
|
|
|
|
})
|
|
|
|
scp.innerDeindent = template.HTML(b.String())
|
|
|
|
bp.PutBuffer(b)
|
|
|
|
})
|
|
|
|
|
|
|
|
return scp.innerDeindent
|
|
|
|
}
|
|
|
|
|
2018-11-01 05:39:44 -04:00
|
|
|
// Position returns this shortcode's detailed position. Note that this information
|
|
|
|
// may be expensive to calculate, so only use this in error situations.
|
2018-11-01 06:28:30 -04:00
|
|
|
func (scp *ShortcodeWithPage) Position() text.Position {
|
2018-11-01 05:39:44 -04:00
|
|
|
scp.posInit.Do(func() {
|
2019-01-02 06:33:26 -05:00
|
|
|
if p, ok := mustUnwrapPage(scp.Page).(pageContext); ok {
|
|
|
|
scp.pos = p.posOffset(scp.posOffset)
|
|
|
|
}
|
2018-11-01 05:39:44 -04:00
|
|
|
})
|
|
|
|
return scp.pos
|
|
|
|
}
|
|
|
|
|
2016-03-24 09:11:04 -04:00
|
|
|
// Site returns information about the current site.
|
2019-01-02 06:33:26 -05:00
|
|
|
func (scp *ShortcodeWithPage) Site() page.Site {
|
|
|
|
return scp.Page.Site()
|
2016-03-14 09:10:15 -04:00
|
|
|
}
|
|
|
|
|
2018-11-01 06:28:30 -04:00
|
|
|
// Ref is a shortcut to the Ref method on Page. It passes itself as a context
|
|
|
|
// to get better error messages.
|
2022-03-17 17:03:27 -04:00
|
|
|
func (scp *ShortcodeWithPage) Ref(args map[string]any) (string, error) {
|
2019-01-02 06:33:26 -05:00
|
|
|
return scp.Page.RefFrom(args, scp)
|
Provide (relative) reference funcs & shortcodes.
- `.Ref` and `.RelRef` take a reference (the logical filename for a
page, including extension and/or a document fragment ID) and return
a permalink (or relative permalink) to the referenced document.
- If the reference is a page name (such as `about.md`), the page
will be discovered and the permalink will be returned: `/about/`
- If the reference is a page name with a fragment (such as
`about.md#who`), the page will be discovered and used to add the
`page.UniqueID()` to the resulting fragment and permalink:
`/about/#who:deadbeef`.
- If the reference is a fragment and `.*Ref` has been called from
a `Node` or `SiteInfo`, it will be returned as is: `#who`.
- If the reference is a fragment and `.*Ref` has been called from
a `Page`, it will be returned with the page’s unique ID:
`#who:deadbeef`.
- `.*Ref` can be called from either `Node`, `SiteInfo` (e.g.,
`Node.Site`), `Page` objects, or `ShortcodeWithPage` objects in
templates.
- `.*Ref` cannot be used in content, so two shortcodes have been
created to provide the functionality to content: `ref` and `relref`.
These are intended to be used within markup, like `[Who]({{% ref
about.md#who %}})` or `<a href="{{% ref about.md#who %}}">Who</a>`.
- There are also `ref` and `relref` template functions (used to create
the shortcodes) that expect a `Page` or `Node` object and the
reference string (e.g., `{{ relref . "about.md" }}` or `{{
"about.md" | ref . }}`). It actually looks for `.*Ref` as defined on
`Node` or `Page` objects.
- Shortcode handling had to use a *differently unique* wrapper in
`createShortcodePlaceholder` because of the way that the `ref` and
`relref` are intended to be used in content.
2014-11-24 01:15:34 -05:00
|
|
|
}
|
|
|
|
|
2018-11-01 06:28:30 -04:00
|
|
|
// RelRef is a shortcut to the RelRef method on Page. It passes itself as a context
|
|
|
|
// to get better error messages.
|
2022-03-17 17:03:27 -04:00
|
|
|
func (scp *ShortcodeWithPage) RelRef(args map[string]any) (string, error) {
|
2019-01-02 06:33:26 -05:00
|
|
|
return scp.Page.RelRefFrom(args, scp)
|
Provide (relative) reference funcs & shortcodes.
- `.Ref` and `.RelRef` take a reference (the logical filename for a
page, including extension and/or a document fragment ID) and return
a permalink (or relative permalink) to the referenced document.
- If the reference is a page name (such as `about.md`), the page
will be discovered and the permalink will be returned: `/about/`
- If the reference is a page name with a fragment (such as
`about.md#who`), the page will be discovered and used to add the
`page.UniqueID()` to the resulting fragment and permalink:
`/about/#who:deadbeef`.
- If the reference is a fragment and `.*Ref` has been called from
a `Node` or `SiteInfo`, it will be returned as is: `#who`.
- If the reference is a fragment and `.*Ref` has been called from
a `Page`, it will be returned with the page’s unique ID:
`#who:deadbeef`.
- `.*Ref` can be called from either `Node`, `SiteInfo` (e.g.,
`Node.Site`), `Page` objects, or `ShortcodeWithPage` objects in
templates.
- `.*Ref` cannot be used in content, so two shortcodes have been
created to provide the functionality to content: `ref` and `relref`.
These are intended to be used within markup, like `[Who]({{% ref
about.md#who %}})` or `<a href="{{% ref about.md#who %}}">Who</a>`.
- There are also `ref` and `relref` template functions (used to create
the shortcodes) that expect a `Page` or `Node` object and the
reference string (e.g., `{{ relref . "about.md" }}` or `{{
"about.md" | ref . }}`). It actually looks for `.*Ref` as defined on
`Node` or `Page` objects.
- Shortcode handling had to use a *differently unique* wrapper in
`createShortcodePlaceholder` because of the way that the `ref` and
`relref` are intended to be used in content.
2014-11-24 01:15:34 -05:00
|
|
|
}
|
|
|
|
|
2024-11-13 05:07:32 -05:00
|
|
|
// Store returns this shortcode's Store.
|
|
|
|
func (scp *ShortcodeWithPage) Store() *maps.Scratch {
|
|
|
|
if scp.store == nil {
|
|
|
|
scp.store = maps.NewScratch()
|
|
|
|
}
|
|
|
|
return scp.store
|
|
|
|
}
|
|
|
|
|
2016-03-24 09:11:04 -04:00
|
|
|
// Scratch returns a scratch-pad scoped for this shortcode. This can be used
|
|
|
|
// as a temporary storage for variables, counters etc.
|
2024-11-13 05:07:32 -05:00
|
|
|
// Deprecated: Use Store instead. Note that from the templates this should be considered a "soft deprecation".
|
2018-07-06 08:12:10 -04:00
|
|
|
func (scp *ShortcodeWithPage) Scratch() *maps.Scratch {
|
2024-11-13 05:07:32 -05:00
|
|
|
return scp.Store()
|
2015-04-19 08:50:24 -04:00
|
|
|
}
|
|
|
|
|
2016-03-24 09:11:04 -04:00
|
|
|
// Get is a convenience method to look up shortcode parameters by its key.
|
2022-03-17 17:03:27 -04:00
|
|
|
func (scp *ShortcodeWithPage) Get(key any) any {
|
2016-07-21 11:18:55 -04:00
|
|
|
if scp.Params == nil {
|
|
|
|
return nil
|
|
|
|
}
|
2014-02-25 23:57:31 -05:00
|
|
|
if reflect.ValueOf(scp.Params).Len() == 0 {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
|
|
|
var x reflect.Value
|
|
|
|
|
|
|
|
switch key.(type) {
|
|
|
|
case int64, int32, int16, int8, int:
|
|
|
|
if reflect.TypeOf(scp.Params).Kind() == reflect.Map {
|
2018-05-21 11:47:52 -04:00
|
|
|
// We treat this as a non error, so people can do similar to
|
|
|
|
// {{ $myParam := .Get "myParam" | default .Get 0 }}
|
|
|
|
// Without having to do additional checks.
|
|
|
|
return nil
|
2014-02-25 23:57:31 -05:00
|
|
|
} else if reflect.TypeOf(scp.Params).Kind() == reflect.Slice {
|
2015-08-07 13:21:26 -04:00
|
|
|
idx := int(reflect.ValueOf(key).Int())
|
|
|
|
ln := reflect.ValueOf(scp.Params).Len()
|
|
|
|
if idx > ln-1 {
|
2018-04-17 05:24:03 -04:00
|
|
|
return ""
|
2015-08-07 13:21:26 -04:00
|
|
|
}
|
|
|
|
x = reflect.ValueOf(scp.Params).Index(idx)
|
2014-02-25 23:57:31 -05:00
|
|
|
}
|
|
|
|
case string:
|
|
|
|
if reflect.TypeOf(scp.Params).Kind() == reflect.Map {
|
|
|
|
x = reflect.ValueOf(scp.Params).MapIndex(reflect.ValueOf(key))
|
|
|
|
if !x.IsValid() {
|
|
|
|
return ""
|
|
|
|
}
|
|
|
|
} else if reflect.TypeOf(scp.Params).Kind() == reflect.Slice {
|
2018-05-21 11:47:52 -04:00
|
|
|
// We treat this as a non error, so people can do similar to
|
|
|
|
// {{ $myParam := .Get "myParam" | default .Get 0 }}
|
|
|
|
// Without having to do additional checks.
|
|
|
|
return nil
|
2014-02-25 23:57:31 -05:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-09-29 08:51:51 -04:00
|
|
|
return x.Interface()
|
2014-02-25 23:57:31 -05:00
|
|
|
}
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
// For internal use only.
|
|
|
|
func (scp *ShortcodeWithPage) Unwrapv() any {
|
2019-01-02 06:33:26 -05:00
|
|
|
return scp.Page
|
2018-11-01 06:28:30 -04:00
|
|
|
}
|
|
|
|
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
// Note - this value must not contain any markup syntax
|
2019-01-02 06:33:26 -05:00
|
|
|
const shortcodePlaceholderPrefix = "HAHAHUGOSHORTCODE"
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
func createShortcodePlaceholder(sid string, id uint64, ordinal int) string {
|
|
|
|
return shortcodePlaceholderPrefix + strconv.FormatUint(id, 10) + sid + strconv.Itoa(ordinal) + "HBHB"
|
2019-01-02 06:33:26 -05:00
|
|
|
}
|
2013-07-04 11:32:55 -04:00
|
|
|
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
type shortcode struct {
|
2018-11-26 05:01:27 -05:00
|
|
|
name string
|
2022-03-17 17:03:27 -04:00
|
|
|
isInline bool // inline shortcode. Any inner will be a Go template.
|
|
|
|
isClosing bool // whether a closing tag was provided
|
|
|
|
inner []any // string or nested shortcode
|
|
|
|
params any // map or array
|
2018-11-26 05:01:27 -05:00
|
|
|
ordinal int
|
2019-01-02 06:33:26 -05:00
|
|
|
|
2022-05-28 07:18:50 -04:00
|
|
|
indentation string // indentation from source.
|
|
|
|
|
2020-07-03 12:02:32 -04:00
|
|
|
info tpl.Info // One of the output formats (arbitrary)
|
|
|
|
templs []tpl.Template // All output formats
|
2019-01-02 06:33:26 -05:00
|
|
|
|
|
|
|
// If set, the rendered shortcode is sent as part of the surrounding content
|
2022-05-28 05:01:47 -04:00
|
|
|
// to Goldmark and similar.
|
2019-01-02 06:33:26 -05:00
|
|
|
// Before Hug0 0.55 we didn't send any shortcode output to the markup
|
|
|
|
// renderer, and this flag told Hugo to process the {{ .Inner }} content
|
|
|
|
// separately.
|
2024-02-18 06:16:30 -05:00
|
|
|
// The old behavior can be had by starting your shortcode template with:
|
2019-01-02 06:33:26 -05:00
|
|
|
// {{ $_hugo_config := `{ "version": 1 }`}}
|
|
|
|
doMarkup bool
|
|
|
|
|
2022-05-28 05:01:47 -04:00
|
|
|
// the placeholder in the source when passed to Goldmark etc.
|
2019-01-02 06:33:26 -05:00
|
|
|
// This also identifies the rendered shortcode.
|
|
|
|
placeholder string
|
|
|
|
|
|
|
|
pos int // the position in bytes in the source file
|
|
|
|
length int // the length in bytes in the source file
|
2013-07-04 11:32:55 -04:00
|
|
|
}
|
|
|
|
|
2019-04-24 08:05:37 -04:00
|
|
|
func (s shortcode) insertPlaceholder() bool {
|
2019-12-28 06:07:23 -05:00
|
|
|
return !s.doMarkup || s.configVersion() == 1
|
|
|
|
}
|
|
|
|
|
2023-02-23 02:08:17 -05:00
|
|
|
func (s shortcode) needsInner() bool {
|
|
|
|
return s.info != nil && s.info.ParseInfo().IsInner
|
|
|
|
}
|
|
|
|
|
2019-12-28 06:07:23 -05:00
|
|
|
func (s shortcode) configVersion() int {
|
|
|
|
if s.info == nil {
|
|
|
|
// Not set for inline shortcodes.
|
|
|
|
return 2
|
|
|
|
}
|
|
|
|
|
|
|
|
return s.info.ParseInfo().Config.Version
|
2019-04-24 08:05:37 -04:00
|
|
|
}
|
|
|
|
|
Move the emoji parsing to pageparser
This avoids double parsing the page content when `enableEmoji=true`.
This commit also adds some general improvements to the parser, making it in general much faster:
```bash
benchmark old ns/op new ns/op delta
BenchmarkShortcodeLexer-4 90258 101730 +12.71%
BenchmarkParse-4 148940 15037 -89.90%
benchmark old allocs new allocs delta
BenchmarkShortcodeLexer-4 456 700 +53.51%
BenchmarkParse-4 28 33 +17.86%
benchmark old bytes new bytes delta
BenchmarkShortcodeLexer-4 69875 81014 +15.94%
BenchmarkParse-4 8128 8304 +2.17%
```
Running some site benchmarks with Emoji support turned on:
```bash
benchmark old ns/op new ns/op delta
BenchmarkSiteBuilding/TOML,num_langs=3,num_pages=5000,tags_per_page=5,shortcodes,render-4 924556797 818115620 -11.51%
benchmark old allocs new allocs delta
BenchmarkSiteBuilding/TOML,num_langs=3,num_pages=5000,tags_per_page=5,shortcodes,render-4 4112613 4133787 +0.51%
benchmark old bytes new bytes delta
BenchmarkSiteBuilding/TOML,num_langs=3,num_pages=5000,tags_per_page=5,shortcodes,render-4 426982864 424363832 -0.61%
```
Fixes #5534
2018-12-17 15:03:23 -05:00
|
|
|
func (s shortcode) innerString() string {
|
|
|
|
var sb strings.Builder
|
|
|
|
|
|
|
|
for _, inner := range s.inner {
|
|
|
|
sb.WriteString(inner.(string))
|
|
|
|
}
|
|
|
|
|
|
|
|
return sb.String()
|
|
|
|
}
|
|
|
|
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
func (sc shortcode) String() string {
|
|
|
|
// for testing (mostly), so any change here will break tests!
|
2022-03-17 17:03:27 -04:00
|
|
|
var params any
|
2014-11-18 04:20:52 -05:00
|
|
|
switch v := sc.params.(type) {
|
2022-03-17 17:03:27 -04:00
|
|
|
case map[string]any:
|
2014-11-18 04:20:52 -05:00
|
|
|
// sort the keys so test assertions won't fail
|
|
|
|
var keys []string
|
|
|
|
for k := range v {
|
|
|
|
keys = append(keys, k)
|
|
|
|
}
|
|
|
|
sort.Strings(keys)
|
2022-03-17 17:03:27 -04:00
|
|
|
tmp := make(map[string]any)
|
2014-11-18 04:20:52 -05:00
|
|
|
|
2019-09-29 08:51:51 -04:00
|
|
|
for _, k := range keys {
|
|
|
|
tmp[k] = v[k]
|
2014-11-18 04:20:52 -05:00
|
|
|
}
|
|
|
|
params = tmp
|
|
|
|
|
|
|
|
default:
|
|
|
|
// use it as is
|
|
|
|
params = sc.params
|
|
|
|
}
|
|
|
|
|
|
|
|
return fmt.Sprintf("%s(%q, %t){%s}", sc.name, params, sc.doMarkup, sc.inner)
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
}
|
2013-12-06 23:14:54 -05:00
|
|
|
|
2017-03-10 14:54:50 -05:00
|
|
|
type shortcodeHandler struct {
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
filename string
|
|
|
|
s *Site
|
2017-05-06 14:15:28 -04:00
|
|
|
|
2019-01-02 06:33:26 -05:00
|
|
|
// Ordered list of shortcodes for a page.
|
|
|
|
shortcodes []*shortcode
|
2017-03-10 14:54:50 -05:00
|
|
|
|
|
|
|
// All the shortcode names in this set.
|
2022-05-29 10:41:57 -04:00
|
|
|
nameSet map[string]bool
|
|
|
|
nameSetMu sync.RWMutex
|
2018-04-19 12:06:40 -04:00
|
|
|
|
2019-01-02 06:33:26 -05:00
|
|
|
// Configuration
|
2018-11-26 05:01:27 -05:00
|
|
|
enableInlineShortcodes bool
|
2018-04-19 12:06:40 -04:00
|
|
|
}
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
func newShortcodeHandler(filename string, s *Site) *shortcodeHandler {
|
2019-01-02 06:33:26 -05:00
|
|
|
sh := &shortcodeHandler{
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
filename: filename,
|
2019-01-02 06:33:26 -05:00
|
|
|
s: s,
|
2021-12-12 06:11:11 -05:00
|
|
|
enableInlineShortcodes: s.ExecHelper.Sec().EnableInlineShortcodes,
|
2019-01-02 06:33:26 -05:00
|
|
|
shortcodes: make([]*shortcode, 0, 4),
|
2018-11-26 05:01:27 -05:00
|
|
|
nameSet: make(map[string]bool),
|
2014-06-26 16:47:51 -04:00
|
|
|
}
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
|
2019-01-02 06:33:26 -05:00
|
|
|
return sh
|
2017-03-10 14:54:50 -05:00
|
|
|
}
|
|
|
|
|
2019-01-02 06:33:26 -05:00
|
|
|
const (
|
|
|
|
innerNewlineRegexp = "\n"
|
|
|
|
innerCleanupRegexp = `\A<p>(.*)</p>\n\z`
|
|
|
|
innerCleanupExpand = "$1"
|
|
|
|
)
|
2014-02-25 23:57:31 -05:00
|
|
|
|
2023-02-11 10:20:24 -05:00
|
|
|
func prepareShortcode(
|
|
|
|
ctx context.Context,
|
2019-01-02 06:33:26 -05:00
|
|
|
level int,
|
|
|
|
s *Site,
|
|
|
|
tplVariants tpl.TemplateVariants,
|
2018-04-22 08:07:29 -04:00
|
|
|
sc *shortcode,
|
2017-05-06 14:15:28 -04:00
|
|
|
parent *ShortcodeWithPage,
|
2023-10-25 05:31:01 -04:00
|
|
|
p *pageState,
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
isRenderString bool,
|
2023-10-25 05:31:01 -04:00
|
|
|
) (shortcodeRenderer, error) {
|
2023-02-11 10:20:24 -05:00
|
|
|
toParseErr := func(err error) error {
|
2024-01-29 04:02:24 -05:00
|
|
|
source := p.m.content.mustSource()
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
return p.parseError(fmt.Errorf("failed to render shortcode %q: %w", sc.name, err), source, sc.pos)
|
2023-02-11 10:20:24 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
// Allow the caller to delay the rendering of the shortcode if needed.
|
|
|
|
var fn shortcodeRenderFunc = func(ctx context.Context) ([]byte, bool, error) {
|
2024-03-17 06:12:33 -04:00
|
|
|
if p.m.pageConfig.ContentMediaType.IsMarkdown() && sc.doMarkup {
|
2024-04-11 11:46:18 -04:00
|
|
|
// Signal downwards that the content rendered will be
|
|
|
|
// parsed and rendered by Goldmark.
|
|
|
|
ctx = tpl.Context.IsInGoldmark.Set(ctx, true)
|
|
|
|
}
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
r, err := doRenderShortcode(ctx, level, s, tplVariants, sc, parent, p, isRenderString)
|
2023-02-11 10:20:24 -05:00
|
|
|
if err != nil {
|
|
|
|
return nil, false, toParseErr(err)
|
|
|
|
}
|
2024-04-11 11:46:18 -04:00
|
|
|
|
2023-02-11 10:20:24 -05:00
|
|
|
b, hasVariants, err := r.renderShortcode(ctx)
|
|
|
|
if err != nil {
|
|
|
|
return nil, false, toParseErr(err)
|
|
|
|
}
|
|
|
|
return b, hasVariants, nil
|
|
|
|
}
|
|
|
|
|
|
|
|
return fn, nil
|
|
|
|
}
|
|
|
|
|
|
|
|
func doRenderShortcode(
|
|
|
|
ctx context.Context,
|
|
|
|
level int,
|
|
|
|
s *Site,
|
|
|
|
tplVariants tpl.TemplateVariants,
|
|
|
|
sc *shortcode,
|
|
|
|
parent *ShortcodeWithPage,
|
2023-10-25 05:31:01 -04:00
|
|
|
p *pageState,
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
isRenderString bool,
|
2023-10-25 05:31:01 -04:00
|
|
|
) (shortcodeRenderer, error) {
|
2018-11-26 05:01:27 -05:00
|
|
|
var tmpl tpl.Template
|
|
|
|
|
2019-01-02 06:33:26 -05:00
|
|
|
// Tracks whether this shortcode or any of its children has template variations
|
|
|
|
// in other languages or output formats. We are currently only interested in
|
|
|
|
// the output formats, so we may get some false positives -- we
|
|
|
|
// should improve on that.
|
|
|
|
var hasVariants bool
|
|
|
|
|
2018-11-26 05:01:27 -05:00
|
|
|
if sc.isInline {
|
2021-12-12 06:11:11 -05:00
|
|
|
if !p.s.ExecHelper.Sec().EnableInlineShortcodes {
|
2023-02-11 10:20:24 -05:00
|
|
|
return zeroShortcode, nil
|
2019-01-31 05:53:21 -05:00
|
|
|
}
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
templName := path.Join("_inline_shortcode", p.Path(), sc.name)
|
2018-11-26 05:01:27 -05:00
|
|
|
if sc.isClosing {
|
Move the emoji parsing to pageparser
This avoids double parsing the page content when `enableEmoji=true`.
This commit also adds some general improvements to the parser, making it in general much faster:
```bash
benchmark old ns/op new ns/op delta
BenchmarkShortcodeLexer-4 90258 101730 +12.71%
BenchmarkParse-4 148940 15037 -89.90%
benchmark old allocs new allocs delta
BenchmarkShortcodeLexer-4 456 700 +53.51%
BenchmarkParse-4 28 33 +17.86%
benchmark old bytes new bytes delta
BenchmarkShortcodeLexer-4 69875 81014 +15.94%
BenchmarkParse-4 8128 8304 +2.17%
```
Running some site benchmarks with Emoji support turned on:
```bash
benchmark old ns/op new ns/op delta
BenchmarkSiteBuilding/TOML,num_langs=3,num_pages=5000,tags_per_page=5,shortcodes,render-4 924556797 818115620 -11.51%
benchmark old allocs new allocs delta
BenchmarkSiteBuilding/TOML,num_langs=3,num_pages=5000,tags_per_page=5,shortcodes,render-4 4112613 4133787 +0.51%
benchmark old bytes new bytes delta
BenchmarkSiteBuilding/TOML,num_langs=3,num_pages=5000,tags_per_page=5,shortcodes,render-4 426982864 424363832 -0.61%
```
Fixes #5534
2018-12-17 15:03:23 -05:00
|
|
|
templStr := sc.innerString()
|
2018-11-26 05:01:27 -05:00
|
|
|
|
|
|
|
var err error
|
2020-01-15 09:59:56 -05:00
|
|
|
tmpl, err = s.TextTmpl().Parse(templName, templStr)
|
2018-11-26 05:01:27 -05:00
|
|
|
if err != nil {
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
if isRenderString {
|
|
|
|
return zeroShortcode, p.wrapError(err)
|
|
|
|
}
|
2022-05-15 05:40:34 -04:00
|
|
|
fe := herrors.NewFileErrorFromName(err, p.File().Filename())
|
2022-05-02 10:07:52 -04:00
|
|
|
pos := fe.Position()
|
|
|
|
pos.LineNumber += p.posOffset(sc.pos).LineNumber
|
|
|
|
fe = fe.UpdatePosition(pos)
|
2023-02-11 10:20:24 -05:00
|
|
|
return zeroShortcode, p.wrapError(fe)
|
2018-11-26 05:01:27 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
} else {
|
|
|
|
// Re-use of shortcode defined earlier in the same page.
|
|
|
|
var found bool
|
2020-01-15 09:59:56 -05:00
|
|
|
tmpl, found = s.TextTmpl().Lookup(templName)
|
2018-11-26 05:01:27 -05:00
|
|
|
if !found {
|
2023-02-11 10:20:24 -05:00
|
|
|
return zeroShortcode, fmt.Errorf("no earlier definition of shortcode %q found", sc.name)
|
2018-11-26 05:01:27 -05:00
|
|
|
}
|
|
|
|
}
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
tmpl = tpl.AddIdentity(tmpl)
|
2018-11-26 05:01:27 -05:00
|
|
|
} else {
|
2019-01-02 06:33:26 -05:00
|
|
|
var found, more bool
|
2020-01-15 09:59:56 -05:00
|
|
|
tmpl, found, more = s.Tmpl().LookupVariant(sc.name, tplVariants)
|
2019-01-02 06:33:26 -05:00
|
|
|
if !found {
|
2020-10-21 05:17:48 -04:00
|
|
|
s.Log.Errorf("Unable to locate template for shortcode %q in page %q", sc.name, p.File().Path())
|
2023-02-11 10:20:24 -05:00
|
|
|
return zeroShortcode, nil
|
2019-01-02 06:33:26 -05:00
|
|
|
}
|
|
|
|
hasVariants = hasVariants || more
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
}
|
2014-02-25 23:57:31 -05:00
|
|
|
|
2024-11-13 05:07:32 -05:00
|
|
|
data := &ShortcodeWithPage{
|
|
|
|
Ordinal: sc.ordinal,
|
|
|
|
posOffset: sc.pos,
|
|
|
|
indentation: sc.indentation,
|
|
|
|
Params: sc.params,
|
|
|
|
Page: newPageForShortcode(p),
|
|
|
|
Parent: parent,
|
|
|
|
Name: sc.name,
|
|
|
|
}
|
|
|
|
|
2015-11-20 19:59:54 -05:00
|
|
|
if sc.params != nil {
|
|
|
|
data.IsNamedParams = reflect.TypeOf(sc.params).Kind() == reflect.Map
|
|
|
|
}
|
|
|
|
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
if len(sc.inner) > 0 {
|
|
|
|
var inner string
|
|
|
|
for _, innerData := range sc.inner {
|
2019-01-02 06:33:26 -05:00
|
|
|
switch innerData := innerData.(type) {
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
case string:
|
2019-01-02 06:33:26 -05:00
|
|
|
inner += innerData
|
2018-04-22 08:07:29 -04:00
|
|
|
case *shortcode:
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
s, err := prepareShortcode(ctx, level+1, s, tplVariants, innerData, data, p, isRenderString)
|
2018-10-03 08:58:09 -04:00
|
|
|
if err != nil {
|
2023-02-11 10:20:24 -05:00
|
|
|
return zeroShortcode, err
|
2018-10-03 08:58:09 -04:00
|
|
|
}
|
2023-02-11 10:20:24 -05:00
|
|
|
ss, more, err := s.renderShortcodeString(ctx)
|
2019-01-02 06:33:26 -05:00
|
|
|
hasVariants = hasVariants || more
|
2023-02-11 10:20:24 -05:00
|
|
|
if err != nil {
|
|
|
|
return zeroShortcode, err
|
|
|
|
}
|
|
|
|
inner += ss
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
default:
|
2020-10-21 05:17:48 -04:00
|
|
|
s.Log.Errorf("Illegal state on shortcode rendering of %q in page %q. Illegal type in inner data: %s ",
|
2019-01-02 06:33:26 -05:00
|
|
|
sc.name, p.File().Path(), reflect.TypeOf(innerData))
|
2023-02-11 10:20:24 -05:00
|
|
|
return zeroShortcode, nil
|
2014-02-25 23:57:31 -05:00
|
|
|
}
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
}
|
|
|
|
|
2024-02-18 06:16:30 -05:00
|
|
|
// Pre Hugo 0.55 this was the behavior even for the outer-most
|
2019-01-02 06:33:26 -05:00
|
|
|
// shortcode.
|
2019-12-28 06:07:23 -05:00
|
|
|
if sc.doMarkup && (level > 0 || sc.configVersion() == 1) {
|
2019-08-16 09:55:03 -04:00
|
|
|
var err error
|
2023-02-24 01:23:10 -05:00
|
|
|
b, err := p.pageOutput.contentRenderer.ParseAndRenderContent(ctx, []byte(inner), false)
|
2019-08-16 09:55:03 -04:00
|
|
|
if err != nil {
|
2023-02-11 10:20:24 -05:00
|
|
|
return zeroShortcode, err
|
2019-08-16 09:55:03 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
newInner := b.Bytes()
|
Provide (relative) reference funcs & shortcodes.
- `.Ref` and `.RelRef` take a reference (the logical filename for a
page, including extension and/or a document fragment ID) and return
a permalink (or relative permalink) to the referenced document.
- If the reference is a page name (such as `about.md`), the page
will be discovered and the permalink will be returned: `/about/`
- If the reference is a page name with a fragment (such as
`about.md#who`), the page will be discovered and used to add the
`page.UniqueID()` to the resulting fragment and permalink:
`/about/#who:deadbeef`.
- If the reference is a fragment and `.*Ref` has been called from
a `Node` or `SiteInfo`, it will be returned as is: `#who`.
- If the reference is a fragment and `.*Ref` has been called from
a `Page`, it will be returned with the page’s unique ID:
`#who:deadbeef`.
- `.*Ref` can be called from either `Node`, `SiteInfo` (e.g.,
`Node.Site`), `Page` objects, or `ShortcodeWithPage` objects in
templates.
- `.*Ref` cannot be used in content, so two shortcodes have been
created to provide the functionality to content: `ref` and `relref`.
These are intended to be used within markup, like `[Who]({{% ref
about.md#who %}})` or `<a href="{{% ref about.md#who %}}">Who</a>`.
- There are also `ref` and `relref` template functions (used to create
the shortcodes) that expect a `Page` or `Node` object and the
reference string (e.g., `{{ relref . "about.md" }}` or `{{
"about.md" | ref . }}`). It actually looks for `.*Ref` as defined on
`Node` or `Page` objects.
- Shortcode handling had to use a *differently unique* wrapper in
`createShortcodePlaceholder` because of the way that the `ref` and
`relref` are intended to be used in content.
2014-11-24 01:15:34 -05:00
|
|
|
|
2019-01-02 06:33:26 -05:00
|
|
|
// If the type is “” (unknown) or “markdown”, we assume the markdown
|
Provide (relative) reference funcs & shortcodes.
- `.Ref` and `.RelRef` take a reference (the logical filename for a
page, including extension and/or a document fragment ID) and return
a permalink (or relative permalink) to the referenced document.
- If the reference is a page name (such as `about.md`), the page
will be discovered and the permalink will be returned: `/about/`
- If the reference is a page name with a fragment (such as
`about.md#who`), the page will be discovered and used to add the
`page.UniqueID()` to the resulting fragment and permalink:
`/about/#who:deadbeef`.
- If the reference is a fragment and `.*Ref` has been called from
a `Node` or `SiteInfo`, it will be returned as is: `#who`.
- If the reference is a fragment and `.*Ref` has been called from
a `Page`, it will be returned with the page’s unique ID:
`#who:deadbeef`.
- `.*Ref` can be called from either `Node`, `SiteInfo` (e.g.,
`Node.Site`), `Page` objects, or `ShortcodeWithPage` objects in
templates.
- `.*Ref` cannot be used in content, so two shortcodes have been
created to provide the functionality to content: `ref` and `relref`.
These are intended to be used within markup, like `[Who]({{% ref
about.md#who %}})` or `<a href="{{% ref about.md#who %}}">Who</a>`.
- There are also `ref` and `relref` template functions (used to create
the shortcodes) that expect a `Page` or `Node` object and the
reference string (e.g., `{{ relref . "about.md" }}` or `{{
"about.md" | ref . }}`). It actually looks for `.*Ref` as defined on
`Node` or `Page` objects.
- Shortcode handling had to use a *differently unique* wrapper in
`createShortcodePlaceholder` because of the way that the `ref` and
`relref` are intended to be used in content.
2014-11-24 01:15:34 -05:00
|
|
|
// generation has been performed. Given the input: `a line`, markdown
|
|
|
|
// specifies the HTML `<p>a line</p>\n`. When dealing with documents as a
|
|
|
|
// whole, this is OK. When dealing with an `{{ .Inner }}` block in Hugo,
|
|
|
|
// this is not so good. This code does two things:
|
|
|
|
//
|
|
|
|
// 1. Check to see if inner has a newline in it. If so, the Inner data is
|
|
|
|
// unchanged.
|
|
|
|
// 2 If inner does not have a newline, strip the wrapping <p> block and
|
2019-01-02 06:33:26 -05:00
|
|
|
// the newline.
|
2024-03-17 06:12:33 -04:00
|
|
|
switch p.m.pageConfig.Content.Markup {
|
2019-01-02 06:33:26 -05:00
|
|
|
case "", "markdown":
|
Provide (relative) reference funcs & shortcodes.
- `.Ref` and `.RelRef` take a reference (the logical filename for a
page, including extension and/or a document fragment ID) and return
a permalink (or relative permalink) to the referenced document.
- If the reference is a page name (such as `about.md`), the page
will be discovered and the permalink will be returned: `/about/`
- If the reference is a page name with a fragment (such as
`about.md#who`), the page will be discovered and used to add the
`page.UniqueID()` to the resulting fragment and permalink:
`/about/#who:deadbeef`.
- If the reference is a fragment and `.*Ref` has been called from
a `Node` or `SiteInfo`, it will be returned as is: `#who`.
- If the reference is a fragment and `.*Ref` has been called from
a `Page`, it will be returned with the page’s unique ID:
`#who:deadbeef`.
- `.*Ref` can be called from either `Node`, `SiteInfo` (e.g.,
`Node.Site`), `Page` objects, or `ShortcodeWithPage` objects in
templates.
- `.*Ref` cannot be used in content, so two shortcodes have been
created to provide the functionality to content: `ref` and `relref`.
These are intended to be used within markup, like `[Who]({{% ref
about.md#who %}})` or `<a href="{{% ref about.md#who %}}">Who</a>`.
- There are also `ref` and `relref` template functions (used to create
the shortcodes) that expect a `Page` or `Node` object and the
reference string (e.g., `{{ relref . "about.md" }}` or `{{
"about.md" | ref . }}`). It actually looks for `.*Ref` as defined on
`Node` or `Page` objects.
- Shortcode handling had to use a *differently unique* wrapper in
`createShortcodePlaceholder` because of the way that the `ref` and
`relref` are intended to be used in content.
2014-11-24 01:15:34 -05:00
|
|
|
if match, _ := regexp.MatchString(innerNewlineRegexp, inner); !match {
|
|
|
|
cleaner, err := regexp.Compile(innerCleanupRegexp)
|
|
|
|
|
|
|
|
if err == nil {
|
|
|
|
newInner = cleaner.ReplaceAll(newInner, []byte(innerCleanupExpand))
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-05-06 14:15:28 -04:00
|
|
|
// TODO(bep) we may have plain text inner templates.
|
Provide (relative) reference funcs & shortcodes.
- `.Ref` and `.RelRef` take a reference (the logical filename for a
page, including extension and/or a document fragment ID) and return
a permalink (or relative permalink) to the referenced document.
- If the reference is a page name (such as `about.md`), the page
will be discovered and the permalink will be returned: `/about/`
- If the reference is a page name with a fragment (such as
`about.md#who`), the page will be discovered and used to add the
`page.UniqueID()` to the resulting fragment and permalink:
`/about/#who:deadbeef`.
- If the reference is a fragment and `.*Ref` has been called from
a `Node` or `SiteInfo`, it will be returned as is: `#who`.
- If the reference is a fragment and `.*Ref` has been called from
a `Page`, it will be returned with the page’s unique ID:
`#who:deadbeef`.
- `.*Ref` can be called from either `Node`, `SiteInfo` (e.g.,
`Node.Site`), `Page` objects, or `ShortcodeWithPage` objects in
templates.
- `.*Ref` cannot be used in content, so two shortcodes have been
created to provide the functionality to content: `ref` and `relref`.
These are intended to be used within markup, like `[Who]({{% ref
about.md#who %}})` or `<a href="{{% ref about.md#who %}}">Who</a>`.
- There are also `ref` and `relref` template functions (used to create
the shortcodes) that expect a `Page` or `Node` object and the
reference string (e.g., `{{ relref . "about.md" }}` or `{{
"about.md" | ref . }}`). It actually looks for `.*Ref` as defined on
`Node` or `Page` objects.
- Shortcode handling had to use a *differently unique* wrapper in
`createShortcodePlaceholder` because of the way that the `ref` and
`relref` are intended to be used in content.
2014-11-24 01:15:34 -05:00
|
|
|
data.Inner = template.HTML(newInner)
|
2014-02-25 23:57:31 -05:00
|
|
|
} else {
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
data.Inner = template.HTML(inner)
|
2014-02-25 23:57:31 -05:00
|
|
|
}
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
|
2014-02-25 23:57:31 -05:00
|
|
|
}
|
|
|
|
|
2023-02-11 10:20:24 -05:00
|
|
|
result, err := renderShortcodeWithPage(ctx, s.Tmpl(), tmpl, data)
|
2018-11-26 05:01:27 -05:00
|
|
|
|
|
|
|
if err != nil && sc.isInline {
|
2022-05-15 05:40:34 -04:00
|
|
|
fe := herrors.NewFileErrorFromName(err, p.File().Filename())
|
2022-05-02 10:07:52 -04:00
|
|
|
pos := fe.Position()
|
|
|
|
pos.LineNumber += p.posOffset(sc.pos).LineNumber
|
|
|
|
fe = fe.UpdatePosition(pos)
|
2023-02-11 10:20:24 -05:00
|
|
|
return zeroShortcode, fe
|
2018-04-23 23:57:33 -04:00
|
|
|
}
|
2017-05-06 14:15:28 -04:00
|
|
|
|
2022-05-28 07:18:50 -04:00
|
|
|
if len(sc.inner) == 0 && len(sc.indentation) > 0 {
|
|
|
|
b := bp.GetBuffer()
|
|
|
|
i := 0
|
|
|
|
text.VisitLinesAfter(result, func(line string) {
|
|
|
|
// The first line is correctly indented.
|
|
|
|
if i > 0 {
|
|
|
|
b.WriteString(sc.indentation)
|
|
|
|
}
|
|
|
|
i++
|
|
|
|
b.WriteString(line)
|
|
|
|
})
|
|
|
|
|
|
|
|
result = b.String()
|
|
|
|
bp.PutBuffer(b)
|
|
|
|
}
|
|
|
|
|
2023-02-11 10:20:24 -05:00
|
|
|
return prerenderedShortcode{s: result, hasVariants: hasVariants}, err
|
2016-03-17 10:51:52 -04:00
|
|
|
}
|
|
|
|
|
2022-05-29 10:41:57 -04:00
|
|
|
func (s *shortcodeHandler) addName(name string) {
|
|
|
|
s.nameSetMu.Lock()
|
|
|
|
defer s.nameSetMu.Unlock()
|
|
|
|
s.nameSet[name] = true
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *shortcodeHandler) transferNames(in *shortcodeHandler) {
|
|
|
|
s.nameSetMu.Lock()
|
|
|
|
defer s.nameSetMu.Unlock()
|
|
|
|
for k := range in.nameSet {
|
|
|
|
s.nameSet[k] = true
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *shortcodeHandler) hasName(name string) bool {
|
|
|
|
s.nameSetMu.RLock()
|
|
|
|
defer s.nameSetMu.RUnlock()
|
|
|
|
_, ok := s.nameSet[name]
|
|
|
|
return ok
|
|
|
|
}
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
func (s *shortcodeHandler) prepareShortcodesForPage(ctx context.Context, p *pageState, f output.Format, isRenderString bool) (map[string]shortcodeRenderer, error) {
|
2023-02-11 10:20:24 -05:00
|
|
|
rendered := make(map[string]shortcodeRenderer)
|
2017-05-06 14:15:28 -04:00
|
|
|
|
2019-01-02 06:33:26 -05:00
|
|
|
tplVariants := tpl.TemplateVariants{
|
|
|
|
Language: p.Language().Lang,
|
|
|
|
OutputFormat: f,
|
2017-05-06 14:15:28 -04:00
|
|
|
}
|
|
|
|
|
2019-01-02 06:33:26 -05:00
|
|
|
for _, v := range s.shortcodes {
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
s, err := prepareShortcode(ctx, 0, s.s, tplVariants, v, nil, p, isRenderString)
|
2016-08-01 17:04:44 -04:00
|
|
|
if err != nil {
|
2023-02-11 10:20:24 -05:00
|
|
|
return nil, err
|
2016-08-01 17:04:44 -04:00
|
|
|
}
|
2019-01-02 06:33:26 -05:00
|
|
|
rendered[v.placeholder] = s
|
2017-05-06 14:15:28 -04:00
|
|
|
|
2016-08-01 17:04:44 -04:00
|
|
|
}
|
|
|
|
|
2023-02-11 10:20:24 -05:00
|
|
|
return rendered, nil
|
2016-08-01 17:04:44 -04:00
|
|
|
}
|
|
|
|
|
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.
Closes #11455
Closes #11549
This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.
The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.
A list of the notable new features:
* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.
New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.
This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):
Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.
Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).
Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2023-12-24 13:11:05 -05:00
|
|
|
func posFromInput(filename string, input []byte, offset int) text.Position {
|
|
|
|
if offset < 0 {
|
|
|
|
return text.Position{
|
|
|
|
Filename: filename,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
lf := []byte("\n")
|
|
|
|
input = input[:offset]
|
|
|
|
lineNumber := bytes.Count(input, lf) + 1
|
|
|
|
endOfLastLine := bytes.LastIndex(input, lf)
|
|
|
|
|
|
|
|
return text.Position{
|
|
|
|
Filename: filename,
|
|
|
|
LineNumber: lineNumber,
|
|
|
|
ColumnNumber: offset - endOfLastLine,
|
|
|
|
Offset: offset,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
// pageTokens state:
|
|
|
|
// - before: positioned just before the shortcode start
|
|
|
|
// - after: shortcode(s) consumed (plural when they are nested)
|
2022-07-07 10:11:47 -04:00
|
|
|
func (s *shortcodeHandler) extractShortcode(ordinal, level int, source []byte, pt *pageparser.Iterator) (*shortcode, error) {
|
2019-01-02 06:33:26 -05:00
|
|
|
if s == nil {
|
|
|
|
panic("handler nil")
|
|
|
|
}
|
2018-04-23 02:09:56 -04:00
|
|
|
sc := &shortcode{ordinal: ordinal}
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
|
2022-05-28 07:18:50 -04:00
|
|
|
// Back up one to identify any indentation.
|
|
|
|
if pt.Pos() > 0 {
|
|
|
|
pt.Backup()
|
|
|
|
item := pt.Next()
|
|
|
|
if item.IsIndentation() {
|
2022-07-07 10:11:47 -04:00
|
|
|
sc.indentation = item.ValStr(source)
|
2022-05-28 07:18:50 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-02 07:23:25 -05:00
|
|
|
cnt := 0
|
|
|
|
nestedOrdinal := 0
|
|
|
|
nextLevel := level + 1
|
2023-02-23 02:08:17 -05:00
|
|
|
closed := false
|
2022-05-02 10:07:52 -04:00
|
|
|
const errorPrefix = "failed to extract shortcode"
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
|
|
|
|
Loop:
|
|
|
|
for {
|
2018-10-17 06:57:09 -04:00
|
|
|
currItem := pt.Next()
|
|
|
|
switch {
|
|
|
|
case currItem.IsLeftShortcodeDelim():
|
|
|
|
next := pt.Peek()
|
2021-05-24 08:59:02 -04:00
|
|
|
if next.IsRightShortcodeDelim() {
|
|
|
|
// no name: {{< >}} or {{% %}}
|
|
|
|
return sc, errors.New("shortcode has no name")
|
|
|
|
}
|
2018-10-17 06:57:09 -04:00
|
|
|
if next.IsShortcodeClose() {
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
continue
|
2014-01-29 17:50:31 -05:00
|
|
|
}
|
|
|
|
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
if cnt > 0 {
|
|
|
|
// nested shortcode; append it to inner content
|
2018-10-18 04:21:23 -04:00
|
|
|
pt.Backup()
|
2022-07-07 10:11:47 -04:00
|
|
|
nested, err := s.extractShortcode(nestedOrdinal, nextLevel, source, pt)
|
2018-04-23 02:09:56 -04:00
|
|
|
nestedOrdinal++
|
2019-08-19 10:26:53 -04:00
|
|
|
if nested != nil && nested.name != "" {
|
2022-05-29 10:41:57 -04:00
|
|
|
s.addName(nested.name)
|
2017-03-10 14:54:50 -05:00
|
|
|
}
|
2019-08-19 10:26:53 -04:00
|
|
|
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
if err == nil {
|
|
|
|
sc.inner = append(sc.inner, nested)
|
|
|
|
} else {
|
|
|
|
return sc, err
|
|
|
|
}
|
|
|
|
|
2014-01-29 17:50:31 -05:00
|
|
|
} else {
|
2018-10-17 06:57:09 -04:00
|
|
|
sc.doMarkup = currItem.IsShortcodeMarkupDelimiter()
|
2014-01-29 17:50:31 -05:00
|
|
|
}
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
|
|
|
|
cnt++
|
|
|
|
|
2018-10-17 06:57:09 -04:00
|
|
|
case currItem.IsRightShortcodeDelim():
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
// we trust the template on this:
|
|
|
|
// if there's no inner, we're done
|
2020-06-14 12:16:45 -04:00
|
|
|
if !sc.isInline {
|
|
|
|
if !sc.info.ParseInfo().IsInner {
|
|
|
|
return sc, nil
|
|
|
|
}
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
}
|
|
|
|
|
2018-10-17 06:57:09 -04:00
|
|
|
case currItem.IsShortcodeClose():
|
2023-02-23 02:08:17 -05:00
|
|
|
closed = true
|
2018-10-17 06:57:09 -04:00
|
|
|
next := pt.Peek()
|
2020-06-14 12:16:45 -04:00
|
|
|
if !sc.isInline {
|
2023-02-23 02:08:17 -05:00
|
|
|
if !sc.needsInner() {
|
2020-06-14 12:16:45 -04:00
|
|
|
if next.IsError() {
|
|
|
|
// return that error, more specific
|
|
|
|
continue
|
|
|
|
}
|
2023-02-23 02:38:51 -05:00
|
|
|
return nil, fmt.Errorf("%s: shortcode %q does not evaluate .Inner or .InnerDeindent, yet a closing tag was provided", errorPrefix, next.ValStr(source))
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
}
|
|
|
|
}
|
2018-10-17 06:57:09 -04:00
|
|
|
if next.IsRightShortcodeDelim() {
|
2015-03-02 15:23:16 -05:00
|
|
|
// self-closing
|
2018-10-17 06:57:09 -04:00
|
|
|
pt.Consume(1)
|
2015-03-02 15:23:16 -05:00
|
|
|
} else {
|
2018-11-26 05:01:27 -05:00
|
|
|
sc.isClosing = true
|
2018-10-17 06:57:09 -04:00
|
|
|
pt.Consume(2)
|
2015-03-02 15:23:16 -05:00
|
|
|
}
|
|
|
|
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
return sc, nil
|
2018-10-17 06:57:09 -04:00
|
|
|
case currItem.IsText():
|
2022-07-07 10:11:47 -04:00
|
|
|
sc.inner = append(sc.inner, currItem.ValStr(source))
|
2018-10-17 06:57:09 -04:00
|
|
|
case currItem.IsShortcodeName():
|
2019-01-02 06:33:26 -05:00
|
|
|
|
2022-07-07 10:11:47 -04:00
|
|
|
sc.name = currItem.ValStr(source)
|
2019-01-02 06:33:26 -05:00
|
|
|
|
2020-07-03 12:02:32 -04:00
|
|
|
// Used to check if the template expects inner content.
|
|
|
|
templs := s.s.Tmpl().LookupVariants(sc.name)
|
|
|
|
if templs == nil {
|
2022-05-02 10:07:52 -04:00
|
|
|
return nil, fmt.Errorf("%s: template for shortcode %q not found", errorPrefix, sc.name)
|
2015-11-15 14:53:12 -05:00
|
|
|
}
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
|
2020-07-03 12:02:32 -04:00
|
|
|
sc.info = templs[0].(tpl.Info)
|
|
|
|
sc.templs = templs
|
2018-11-26 05:01:27 -05:00
|
|
|
case currItem.IsInlineShortcodeName():
|
2022-07-07 10:11:47 -04:00
|
|
|
sc.name = currItem.ValStr(source)
|
2018-11-26 05:01:27 -05:00
|
|
|
sc.isInline = true
|
2018-10-17 06:57:09 -04:00
|
|
|
case currItem.IsShortcodeParam():
|
|
|
|
if !pt.IsValueNext() {
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
continue
|
2018-10-17 06:57:09 -04:00
|
|
|
} else if pt.Peek().IsShortcodeParamVal() {
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
// named params
|
|
|
|
if sc.params == nil {
|
2022-03-17 17:03:27 -04:00
|
|
|
params := make(map[string]any)
|
2022-07-07 10:11:47 -04:00
|
|
|
params[currItem.ValStr(source)] = pt.Next().ValTyped(source)
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
sc.params = params
|
2014-01-29 17:50:31 -05:00
|
|
|
} else {
|
2022-03-17 17:03:27 -04:00
|
|
|
if params, ok := sc.params.(map[string]any); ok {
|
2022-07-07 10:11:47 -04:00
|
|
|
params[currItem.ValStr(source)] = pt.Next().ValTyped(source)
|
2015-08-07 14:08:23 -04:00
|
|
|
} else {
|
2022-09-01 03:26:27 -04:00
|
|
|
return sc, fmt.Errorf("%s: invalid state: invalid param type %T for shortcode %q, expected a map", errorPrefix, params, sc.name)
|
2015-08-07 14:08:23 -04:00
|
|
|
}
|
2014-01-29 17:50:31 -05:00
|
|
|
}
|
|
|
|
} else {
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
// positional params
|
|
|
|
if sc.params == nil {
|
2022-03-17 17:03:27 -04:00
|
|
|
var params []any
|
2022-07-07 10:11:47 -04:00
|
|
|
params = append(params, currItem.ValTyped(source))
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
sc.params = params
|
2014-01-29 17:50:31 -05:00
|
|
|
} else {
|
2022-03-17 17:03:27 -04:00
|
|
|
if params, ok := sc.params.([]any); ok {
|
2022-07-07 10:11:47 -04:00
|
|
|
params = append(params, currItem.ValTyped(source))
|
2015-08-07 14:08:23 -04:00
|
|
|
sc.params = params
|
|
|
|
} else {
|
2022-09-01 03:26:27 -04:00
|
|
|
return sc, fmt.Errorf("%s: invalid state: invalid param type %T for shortcode %q, expected a slice", errorPrefix, params, sc.name)
|
2015-08-07 14:08:23 -04:00
|
|
|
}
|
2014-01-29 17:50:31 -05:00
|
|
|
}
|
|
|
|
}
|
2018-10-17 06:57:09 -04:00
|
|
|
case currItem.IsDone():
|
2023-02-23 02:08:17 -05:00
|
|
|
if !currItem.IsError() {
|
|
|
|
if !closed && sc.needsInner() {
|
2023-03-10 12:41:17 -05:00
|
|
|
return sc, fmt.Errorf("%s: shortcode %q must be closed or self-closed", errorPrefix, sc.name)
|
2023-02-23 02:08:17 -05:00
|
|
|
}
|
|
|
|
}
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
// handled by caller
|
2018-10-17 06:57:09 -04:00
|
|
|
pt.Backup()
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
break Loop
|
|
|
|
|
2014-01-29 17:50:31 -05:00
|
|
|
}
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
}
|
|
|
|
return sc, nil
|
|
|
|
}
|
|
|
|
|
2019-01-02 06:33:26 -05:00
|
|
|
// Replace prefixed shortcode tokens with the real content.
|
2015-10-20 14:35:12 -04:00
|
|
|
// Note: This function will rewrite the input slice.
|
2023-02-11 10:20:24 -05:00
|
|
|
func expandShortcodeTokens(
|
|
|
|
ctx context.Context,
|
|
|
|
source []byte,
|
|
|
|
tokenHandler func(ctx context.Context, token string) ([]byte, error),
|
|
|
|
) ([]byte, error) {
|
2015-06-21 07:08:30 -04:00
|
|
|
start := 0
|
|
|
|
|
2019-01-02 06:33:26 -05:00
|
|
|
pre := []byte(shortcodePlaceholderPrefix)
|
2016-09-08 15:23:01 -04:00
|
|
|
post := []byte("HBHB")
|
2015-06-21 07:08:30 -04:00
|
|
|
pStart := []byte("<p>")
|
|
|
|
pEnd := []byte("</p>")
|
|
|
|
|
|
|
|
k := bytes.Index(source[start:], pre)
|
|
|
|
|
|
|
|
for k != -1 {
|
|
|
|
j := start + k
|
|
|
|
postIdx := bytes.Index(source[j:], post)
|
|
|
|
if postIdx < 0 {
|
|
|
|
// this should never happen, but let the caller decide to panic or not
|
|
|
|
return nil, errors.New("illegal state in content; shortcode token missing end delim")
|
|
|
|
}
|
|
|
|
|
|
|
|
end := j + postIdx + 4
|
2023-02-11 10:20:24 -05:00
|
|
|
key := string(source[j:end])
|
|
|
|
newVal, err := tokenHandler(ctx, key)
|
|
|
|
if err != nil {
|
|
|
|
return nil, err
|
|
|
|
}
|
2015-06-21 07:08:30 -04:00
|
|
|
|
|
|
|
// Issue #1148: Check for wrapping p-tags <p>
|
|
|
|
if j >= 3 && bytes.Equal(source[j-3:j], pStart) {
|
2018-08-16 20:17:27 -04:00
|
|
|
if (k+4) < len(source) && bytes.Equal(source[end:end+4], pEnd) {
|
2015-06-21 07:08:30 -04:00
|
|
|
j -= 3
|
|
|
|
end += 4
|
2015-01-28 22:11:41 -05:00
|
|
|
}
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
}
|
Provide (relative) reference funcs & shortcodes.
- `.Ref` and `.RelRef` take a reference (the logical filename for a
page, including extension and/or a document fragment ID) and return
a permalink (or relative permalink) to the referenced document.
- If the reference is a page name (such as `about.md`), the page
will be discovered and the permalink will be returned: `/about/`
- If the reference is a page name with a fragment (such as
`about.md#who`), the page will be discovered and used to add the
`page.UniqueID()` to the resulting fragment and permalink:
`/about/#who:deadbeef`.
- If the reference is a fragment and `.*Ref` has been called from
a `Node` or `SiteInfo`, it will be returned as is: `#who`.
- If the reference is a fragment and `.*Ref` has been called from
a `Page`, it will be returned with the page’s unique ID:
`#who:deadbeef`.
- `.*Ref` can be called from either `Node`, `SiteInfo` (e.g.,
`Node.Site`), `Page` objects, or `ShortcodeWithPage` objects in
templates.
- `.*Ref` cannot be used in content, so two shortcodes have been
created to provide the functionality to content: `ref` and `relref`.
These are intended to be used within markup, like `[Who]({{% ref
about.md#who %}})` or `<a href="{{% ref about.md#who %}}">Who</a>`.
- There are also `ref` and `relref` template functions (used to create
the shortcodes) that expect a `Page` or `Node` object and the
reference string (e.g., `{{ relref . "about.md" }}` or `{{
"about.md" | ref . }}`). It actually looks for `.*Ref` as defined on
`Node` or `Page` objects.
- Shortcode handling had to use a *differently unique* wrapper in
`createShortcodePlaceholder` because of the way that the `ref` and
`relref` are intended to be used in content.
2014-11-24 01:15:34 -05:00
|
|
|
|
2015-10-20 14:35:12 -04:00
|
|
|
// This and other cool slice tricks: https://github.com/golang/go/wiki/SliceTricks
|
|
|
|
source = append(source[:j], append(newVal, source[end:]...)...)
|
2016-02-25 15:31:22 -05:00
|
|
|
start = j
|
2015-06-21 07:08:30 -04:00
|
|
|
k = bytes.Index(source[start:], pre)
|
2015-06-22 13:40:12 -04:00
|
|
|
|
2015-10-20 14:35:12 -04:00
|
|
|
}
|
2015-06-22 13:40:12 -04:00
|
|
|
|
2015-10-20 14:35:12 -04:00
|
|
|
return source, nil
|
Shortcode rewrite, take 2
This commit contains a restructuring and partial rewrite of the shortcode handling.
Prior to this commit rendering of the page content was mingled with handling of the shortcodes. This led to several oddities.
The new flow is:
1. Shortcodes are extracted from page and replaced with placeholders.
2. Shortcodes are processed and rendered
3. Page is processed
4. The placeholders are replaced with the rendered shortcodes
The handling of summaries is also made simpler by this.
This commit also introduces some other chenges:
1. distinction between shortcodes that need further processing and those who do not:
* `{{< >}}`: Typically raw HTML. Will not be processed.
* `{{% %}}`: Will be processed by the page's markup engine (Markdown or (infuture) Asciidoctor)
The above also involves a new shortcode-parser, with lexical scanning inspired by Rob Pike's talk called "Lexical Scanning in Go",
which should be easier to understand, give better error messages and perform better.
2. If you want to exclude a shortcode from being processed (for documentation etc.), the inner part of the shorcode must be commented out, i.e. `{{%/* movie 47238zzb */%}}`. See the updated shortcode section in the documentation for further examples.
The new parser supports nested shortcodes. This isn't new, but has two related design choices worth mentioning:
* The shortcodes will be rendered individually, so If both `{{< >}}` and `{{% %}}` are used in the nested hierarchy, one will be passed through the page's markdown processor, the other not.
* To avoid potential costly overhead of always looking far ahead for a possible closing tag, this implementation looks at the template itself, and is branded as a container with inner content if it contains a reference to `.Inner`
Fixes #565
Fixes #480
Fixes #461
And probably some others.
2014-10-27 16:48:30 -04:00
|
|
|
}
|
2013-07-04 11:32:55 -04:00
|
|
|
|
2023-02-11 10:20:24 -05:00
|
|
|
func renderShortcodeWithPage(ctx context.Context, h tpl.TemplateHandler, tmpl tpl.Template, data *ShortcodeWithPage) (string, error) {
|
2015-01-30 14:19:46 -05:00
|
|
|
buffer := bp.GetBuffer()
|
|
|
|
defer bp.PutBuffer(buffer)
|
|
|
|
|
2023-02-11 10:20:24 -05:00
|
|
|
err := h.ExecuteWithContext(ctx, tmpl, buffer, data)
|
2014-01-29 17:50:31 -05:00
|
|
|
if err != nil {
|
2022-05-02 10:07:52 -04:00
|
|
|
return "", fmt.Errorf("failed to process shortcode: %w", err)
|
2014-01-29 17:50:31 -05:00
|
|
|
}
|
2018-10-03 08:58:09 -04:00
|
|
|
return buffer.String(), nil
|
2013-07-04 11:32:55 -04:00
|
|
|
}
|