This is in line with how it behaved before, but it was lifted a little for the project mount for Hugo Modules,
but that could create hard-to-detect loops.
This commit implements Hugo Modules.
This is a broad subject, but some keywords include:
* A new `module` configuration section where you can import almost anything. You can configure both your own file mounts nd the file mounts of the modules you import. This is the new recommended way of configuring what you earlier put in `configDir`, `staticDir` etc. And it also allows you to mount folders in non-Hugo-projects, e.g. the `SCSS` folder in the Bootstrap GitHub project.
* A module consists of a set of mounts to the standard 7 component types in Hugo: `static`, `content`, `layouts`, `data`, `assets`, `i18n`, and `archetypes`. Yes, Theme Components can now include content, which should be very useful, especially in bigger multilingual projects.
* Modules not in your local file cache will be downloaded automatically and even "hot replaced" while the server is running.
* Hugo Modules supports and encourages semver versioned modules, and uses the minimal version selection algorithm to resolve versions.
* A new set of CLI commands are provided to manage all of this: `hugo mod init`, `hugo mod get`, `hugo mod graph`, `hugo mod tidy`, and `hugo mod vendor`.
All of the above is backed by Go Modules.
Fixes#5973Fixes#5996Fixes#6010Fixes#5911Fixes#5940Fixes#6074Fixes#6082Fixes#6092
We occasionally see warnings when building our site:
```
WARN 2019/06/25 23:07:08 make non-relative ref/relref page reference(s) in page %q absolute, e.g. {{< ref "/blog/my-post.md" >}}
```
But the `%q` value is missing, making it difficult to track down the source of the warning.
This change addresses that, by including the source path in the warning:
```
WARN 2019/06/25 23:07:31 make non-relative ref/relref page reference(s) in page "blog/my-post/index.md" absolute, e.g. {{< ref "/blog/my-post.md" >}}
```
Hugo `0.55.0` introduced some new interface types for `Page` etc.
This worked great in general, but there were cases where this would fail in `where` and `sort`.
One such example would be sorting by `MenuItem.Page.Date` where `Page` on `MenuItem` was a small subset of the bigger `page.Page` interface.
This commit fixes that by unwrapping such interface values.
Fixes#5989
Sadly, goorgeous has not been updated in over a year and still has a lot of
open issues (e.g. no support for nested lists).
go-org fixes most of those issues and supports a larger subset of Org mode
syntax.
There have been reports about infrequent paginator crashes when running the Hugo server since 0.55.0.
The reason have been narrowed down to that of parallel rebuilds.
This isn't a new thing, but the changes in 0.55.0 made it extra important to serialize the page initialization.
This commit fixes that by protecting the `Build` method with a lock when running in server mode.
Fixes#5885Fixes#5968
In Hugo 0.55.0 we made AMP `permalinkable`. We also render the output formats in their natural sort order, meaning `AMP` will be rendered before `HTML`. References in the sitemap would then point to the AMP version, and this is normally not what you'd want.
This commit fixes that by making `HTML` by default sort before the others.
If this is not you want, you can set `weight` on the output format configuration.
Fixes#5910
In Hugo `0.55` we introduced the `permalinkable` config attribute on Output Format, default enabled for `AMP` and `HTML`.
This meant that a Page could have different `RelPermalink` and `Permalink` depending on the rendering format.
The menu `URL` did not reflect that fact.
Fixes#5849
The faulty logic published the bundled resources for the "first output" format.
This worked most of the time, but since the output formats list is sorted,
any output format only used for some of the pages (e.g. CSS) would not work properly.
Fixes#5858
In Hugo 0.55 we connected the taxonomy nodes with their owning Page.
This failed if you had, say, a content file for a author that did not author anything in the site:
```
content/authors/silent-persin/_index.md
```
Fixes#5847
We introduced a callback func() to get the owner Page in 0.55.0.
Sadly, funcs is not comparable type in Go.
This commit replaces the func with a struct pointer that wraps the Page.
Fixes#5850
This stopped working in Hugo 0.55:
```bash
---
menu: "main"
---
```
This was also the case for using a slice of menu entries.
This still worked:
---
menu:
main:
weight: 30
---
* Rewind paginator for server mode
* Add some more related tests.
* Replace the clumsy scratch constructs in internal paginator template with variables
See #5825
Add the ability to have a `summary` page variable that overrides
the auto-generated summary. Logic for obtaining summary becomes:
* if summary divider is present in content, use the text above it
* if summary variables is present in page metadata, use that
* auto-generate summary from first _x_ words of the content
Fixes#5800
And use that to calculate number of workers, if set, else fall back to number of logical CPUs.
Also tweak the relevant related settings to match the new setup.
Also remove the setting of `runtime.GOMAXPROCS` as this has been the default behaviour since Go 1.5.
Fixes#5814
Before this commit you would have to do this in multilingual setups:
```
---
title: "Custom!"
url: "/jp/custom/foo"
---
```
This commit allows for relative URLs, e.g:
```
---
title: "Custom!"
url: "custom/foo"
---
```
Which is obviously easier and more portable.
The meaning of relative may change to include more in the future (e.g. role based access).
Fixes#5704
This commit adds support for return values in partials.
This means that you can now do this and similar:
{{ $v := add . 42 }}
{{ return $v }}
Partials without a `return` statement will be rendered as before.
This works for both `partial` and `partialCached`.
Fixes#5783
The main motivation of this commit is to add a `page.Page` interface to replace the very file-oriented `hugolib.Page` struct.
This is all a preparation step for issue #5074, "pages from other data sources".
But this also fixes a set of annoying limitations, especially related to custom output formats, and shortcodes.
Most notable changes:
* The inner content of shortcodes using the `{{%` as the outer-most delimiter will now be sent to the content renderer, e.g. Blackfriday.
This means that any markdown will partake in the global ToC and footnote context etc.
* The Custom Output formats are now "fully virtualized". This removes many of the current limitations.
* The taxonomy list type now has a reference to the `Page` object.
This improves the taxonomy template `.Title` situation and make common template constructs much simpler.
See #5074Fixes#5763Fixes#5758Fixes#5090Fixes#5204Fixes#4695Fixes#5607Fixes#5707Fixes#5719Fixes#3113Fixes#5706Fixes#5767Fixes#5723Fixes#5769Fixes#5770Fixes#5771Fixes#5759Fixes#5776Fixes#5777Fixes#5778
With this commit ByParam takes into account a type of a value under a
key. If both values are numeric then they're coerced into float64 and
then get compared.
If any value isn't numeric, for example it's nil or string, then both
values coerced into string and get compared as strings
(lexicographicaly)
Nil values are always sent to the end.
Numeric values confirm to any type listed below:
uint8, uint16, uint32, uint64, int, int8, int16, int32, int64, float32, float64
Closes#5305
This commit solves an issue where hugo would ignore the cli -t flag
and only use a theme defined in config.toml.
Also allow -t flag to accept a string slice.
Closes#5569Closes#5061
Related #4868
That is, if only `.Content` is accessed.
This means that, for a transformed resource to be published to `/public`, you need to access either `.RelPermalink` or `Permalink`.
Fixes#4944
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
This commit adds support for a configuration directory (default `config`). The different pieces in this puzzle are:
* A new `--environment` (or `-e`) flag. This can also be set with the `HUGO_ENVIRONMENT` OS environment variable. The value for `environment` defaults to `production` when running `hugo` and `development` when running `hugo server`. You can set it to any value you want (e.g. `hugo server -e "Sensible Environment"`), but as it is used to load configuration from the file system, the letter case may be important. You can get this value in your templates with `{{ hugo.Environment }}`.
* A new `--configDir` flag (defaults to `config` below your project). This can also be set with `HUGO_CONFIGDIR` OS environment variable.
If the `configDir` exists, the configuration files will be read and merged on top of each other from left to right; the right-most value will win on duplicates.
Given the example tree below:
If `environment` is `production`, the left-most `config.toml` would be the one directly below the project (this can now be omitted if you want), and then `_default/config.toml` and finally `production/config.toml`. And since these will be merged, you can just provide the environment specific configuration setting in you production config, e.g. `enableGitInfo = true`. The order within the directories will be lexical (`config.toml` and then `params.toml`).
```bash
config
├── _default
│ ├── config.toml
│ ├── languages.toml
│ ├── menus
│ │ ├── menus.en.toml
│ │ └── menus.zh.toml
│ └── params.toml
├── development
│ └── params.toml
└── production
├── config.toml
└── params.toml
```
Some configuration maps support the language code in the filename (e.g. `menus.en.toml`): `menus` (`menu` also works) and `params`.
Also note that the only folders with "a meaning" in the above listing is the top level directories below `config`. The `menus` sub folder is just added for better organization.
We use `TOML` in the example above, but Hugo also supports `JSON` and `YAML` as configuration formats. These can be mixed.
Fixes#5422
This means that the current `.Site` and ´.Hugo` is available as a globals, so you can do `site.IsServer`, `hugo.Version` etc.
Fixes#5470Fixes#5467Fixes#5503
When the page parser was rewritten in 0.51, this was interpreted literally, but commented out front matter is used in the wild to "hide it from GitHub", e.g:
```
<!--
+++
title = "hello"
+++
-->
```
Fixes#5478
An inline shortcode's name must end with `.inline`, all lowercase.
E.g.:
```bash
{{< time.inline >}}{{ now }}{{< /time.inline >}}
```
The above will print the current date and time.
Note that an inline shortcode's inner content is parsed and executed as a Go text template with the same context as a regular shortcode template.
This means that the current page can be accessed via `.Page.Title` etc. This also means that there are no concept of "nested inline shortcodes".
The same inline shortcode can be reused later in the same content file, with different params if needed, using the self-closing syntax:
```
{{< time.inline />}}
```
Fixes#4011
In the newly consolidated file cache implementation, we forgot that we also look in the theme(s) for assets (SCSS transformations etc.), which is not good for Netlify and the demo sites.
Fixes#5460
This allows for "cache per Hugo project", making `hugo --gc` work as expected, even if you have several Hugo projects running on the same PC.
See #5439
This commits reworks how file caching is performed in Hugo. Now there is only one way, and it can be configured.
This is the default configuration:
```toml
[caches]
[caches.getjson]
dir = ":cacheDir"
maxAge = -1
[caches.getcsv]
dir = ":cacheDir"
maxAge = -1
[caches.images]
dir = ":resourceDir/_gen"
maxAge = -1
[caches.assets]
dir = ":resourceDir/_gen"
maxAge = -1
```
You can override any of these cache setting in your own `config.toml`.
The placeholders explained:
`:cacheDir`: This is the value of the `cacheDir` config option if set (can also be set via OS env variable `HUGO_CACHEDIR`). It will fall back to `/opt/build/cache/hugo_cache/` on Netlify, or a `hugo_cache` directory below the OS temp dir for the others.
`:resourceDir`: This is the value of the `resourceDir` config option.
`maxAge` is the time in seconds before a cache entry will be evicted, -1 means forever and 0 effectively turns that particular cache off.
This means that if you run your builds on Netlify, all caches configured with `:cacheDir` will be saved and restored on the next build. For other CI vendors, please read their documentation. For an CircleCI example, see 6c3960a8f4/.circleci/config.ymlFixes#5404
This commit also pulls down the log level for a set of WARN statements to INFO. There should be no ERRORs or WARNINGs in a regular Hugo build. That is the story about the Boy Who Cried Wolf.
Since the WARN log is now more visible, this commit also improves on some of them, most notable the "layout not found", which now would look something like this:
```bash
WARN 2018/11/02 09:02:18 Found no layout for "home", language "en", output format "CSS": create a template below /layouts with one of these filenames: index.en.css.css, home.en.css.css, list.en.css.css, index.css.css, home.css.css, list.css.css, index.en.css, home.en.css, list.en.css, index.css, home.css, list.css, _default/index.en.css.css, _default/home.en.css.css, _default/list.en.css.css, _default/index.css.css, _default/home.css.css, _default/list.css.css, _default/index.en.css, _default/home.en.css, _default/list.en.css, _default/index.css, _default/home.css, _default/list.css
```
Fixes#5203
Long identifiers will give errors on the format:
```bash
_default/single.html:5:14: executing "main" at <.ThisIsAVeryLongTitl...>: can't evaluate field ThisIsAVeryLongTitle
```
Hugo use this value to match the "base template or not", so we need to strip the "...".
Fixes#5346
`*json.UnmarshalTypeError` and `*json.SyntaxError` has a byte `Offset`, so use that.
This commit also reworks/simplifies the errror line matching logic. This also makes the file reading unbuffered, but that should be fine in this error case.
See #5324
We do that by re-render visited pages that is not already in the stack. This may potentially do some double work, but that small penalty should be well worth it.
Fixes#5281
The main item in this commit is showing of errors with a file context when running `hugo server`.
This can be turned off: `hugo server --disableBrowserError` (can also be set in `config.toml`).
But to get there, the error handling in Hugo needed a revision. There are some items left TODO for commits soon to follow, most notable errors in content and config files.
Fixes#5284Fixes#5290
See #5325
See #5324
In short:
* Avoid double tolower in MakeSegment
* Use MakePathSanitized for taxonomies in pageToPermalinkTitle; this matches what MakeSegment does.
* Move the "double hyphen and space" logic into UnicodeSanitize
The last bullet may be slightly breaking for some that now does not get the "--" in some URLs, but we need to reduce the amount of URL logic.
See #4926
When constructing permalinks, ensure that most inputs used as path
segments are normalized with PathSpec.MakeSegment instead of
PathSpec.URLize.
The primary exception to that rule is with taxonomy titles in
pageToPermalinkTitle(). The approach taken here is to use URLize for
taxonomy pages. Everything else will use MakeSegment. The reason for
this exception is that people use taxonomies such as "s1/p1" to generate
URLs precisely they way they wish (see #5223). Tests have been added to
check for this case.
Fixes#4926
Previously, calls to *Page.Eq(nil) would always return false because the
unwrapPage func didn't support the nil case. Add support for unwrapping
nil to a *Page.
Fixes#5043
In Hugo `0.49` we improved type support in `slice`. This has an unfortunate side effect in that `resources.Concat` now expects something that can resolve to `resource.Resources`.
This worked for most situations, but when you try to `slice` different `Resource` objects, you would be getting `[]interface {}` and not `resource.Resources`. And `concat` would fail:
```bash
error calling Concat: slice []interface {} not supported in concat.
```
This commit fixes that by simplifying the type checking logic in `Slice`:
* If the first item implements the `Slicer` interface, we try that
* If the above fails or the first item does not implement `Slicer`, we just return the `[]interface {}`
Fixes#5269
Introduce new page position variables in order to fix the ordering issue
of `.Next` and `.Prev` while also allowing an upgrade path via
deprecation.
`.NextInSection` becomes `.NextPageInSection`.
`.PrevInSection` becomes `.PrevPageInSection`.
`.Next` becomes a function returning `.PrevPage`.
`.Prev` becomes a function returning `.NextPage`.
Fixes#1061
Given this content:
```bash
archetypes
├── default.md
└── post-bundle
├── bio.md
├── images
│ └── featured.jpg
└── index.md
```
```bash
hugo new --kind post-bundle post/my-post
```
Will create a new folder in `/content/post/my-post` with the same set of files as in the `post-bundle` archetypes folder.
This commit also improves the archetype language detection, so, if you use template code in your content files, the `.Site` you get is for the correct language. This also means that it is now possible to translate strings defined in the `i18n` bundles, e.g. `{{ i18n "hello" }}`.
Fixes#4535
It is slightly slower, but correctnes is, of course, more important:
```bash
benchmark old ns/op new ns/op delta
BenchmarkSortByWeightAndReverse-4 367 645 +75.75%
benchmark old allocs new allocs delta
BenchmarkSortByWeightAndReverse-4 2 2 +0.00%
benchmark old bytes new bytes delta
BenchmarkSortByWeightAndReverse-4 64 64 +0.00%
```
Running the same benchmark without any cache (i.e. resorting the slice on every iteration) and then compare it to the current version shows that it still is plenty worth it:
```bash
▶ benchcmp 2.bench 1.bench
benchmark old ns/op new ns/op delta
BenchmarkSortByWeightAndReverse-4 1358757 645 -99.95%
benchmark old allocs new allocs delta
BenchmarkSortByWeightAndReverse-4 17159 2 -99.99%
benchmark old bytes new bytes delta
BenchmarkSortByWeightAndReverse-4 274573 64 -99.98%
```
Closes#5239
Before Hugo this commit we set the pseudo page kind RSS on the page when output to RSS. This had some unintended side effects, esp. when the only output format for that page was RSS.
For the page kinds that can have multiple output formats, the Kind should be one of the standard home, page etc.
Fixes#5138
Before this commit you would typically use `.Scratch.Add` to manually create slices in a loop.
With variable overwrite in Go 1.11, we can do better. This commit adds the `append` template func.
A made-up example:
```bash
{{ $p1 := index .Site.RegularPages 0 }}{{ $p2 := index .Site.RegularPages 1 }}
{{ $pages := slice }}
{{ if true }}
{{ $pages = $pages | append $p2 $p1 }}
{{ end }}
```
Note that with 2 slices as arguments, the two examples below will give the same result:
```bash
{{ $s1 := slice "a" "b" | append (slice "c" "d") }}
{{ $s2 := slice "a" "b" | append "c" "d" }}
```
Both of the above will give `[]string{a, b, c, d}`.
This commit also improves the type handling in the `slice` template function. Now `slice "a" "b"` will give a `[]string` slice. The old behaviour was to return a `[]interface{}`.
Fixes#5190
This extends the page grouping in Hugo with a template function that allows for ad-hoc grouping.
A made-up example:
```
{{ $cool := where .Site.RegularPages "Params.cool" true | group "cool" }}
{{ $blue := where .Site.RegularPages "Params.blue" true | group "blue" }}
{{ $paginator := .Paginate (slice $cool $blue) }}
```
Closes#4865
This also adjusts the pagination logic to allow for these new collections.
Note that we will follow up with a template function named `group` that will be the end user API. The `.Group` method on `Page` should be considered as internal.
Updates #4865
Doing so was probably a mistake. This may be a breaking change for some people,
but it's easy to restore the previous behaviour in the layouts.
Fixes#5172
This commit also removes the deprecated `Suffix` from MediaType. Now use `Suffixes` and put the MIME type suffix in the type, e.g. `application/svg+xml`.
Fixes#5093
Before this commit, the live reload logic in `hugo server` got confused when you dropped a new bundle into the project while the server was running. The workaround was to restart the server.
This commit fixes the "live reload bundle detection" in server mode, and also makes sure that the bundle headers are always processed first.
Fixes#5075
Currently it makes no practical difference, but this is more a protection if we in the future creates index from the content related fields. That will not work from a shortcode.
See #5071
Yesterday's commit was a little too agressive.
This commit makes sure that the duplication of resources to public/en etc. is only performed in multihost mode.
See #5058
In Hugo 0.46 we made the output of what you get from resources.Get and similar static, i.e. language agnostic. This makes total sense, as it is wasteful and time-consuming to do SASS/SCSS/PostCSS processing for lots of languages when the output is lots of duplicates with different filenames.
But since we now output the result once only, this had a negative side effect for multihost setups: We publish the resource once only to the root folder (i.e. not to the language "domain folder").
This commit removes the language code from the processed image keys. This creates less duplication in the file cache, but it means that you should do a `hugo --gc` to clean up stale files.
Fixes#5058
