This ommmit contains some security hardening measures for the Hugo build runtime.
There are some rarely used features in Hugo that would be good to have disabled by default. One example would be the "external helpers".
For `asciidoctor` and some others we use Go's `os/exec` package to start a new process.
These are a predefined set of binary names, all loaded from `PATH` and with a predefined set of arguments. Still, if you don't use `asciidoctor` in your project, you might as well have it turned off.
You can configure your own in the new `security` configuration section, but the defaults are configured to create a minimal amount of site breakage. And if that do happen, you will get clear instructions in the loa about what to do.
The default configuration is listed below. Note that almost all of these options are regular expression _whitelists_ (a string or a slice); the value `none` will block all.
```toml
[security]
enableInlineShortcodes = false
[security.exec]
allow = ['^dart-sass-embedded$', '^go$', '^npx$', '^postcss$']
osEnv = ['(?i)^(PATH|PATHEXT|APPDATA|TMP|TEMP|TERM)$']
[security.funcs]
getenv = ['^HUGO_']
[security.http]
methods = ['(?i)GET|POST']
urls = ['.*']
```
In Hugo 0.90.0 we introduced remote support in `resources.Get`.
But with remote resources comes with a higher chance of failing a build (network issues, remote server down etc.).
Before this commit we always failed the build on any unexpected error.
This commit allows the user to check for any error (and potentially fall back to a default local resource):
```htmlbars
{{ $result := resources.Get "https://gohugo.io/img/hugo-logo.png" }}
{{ with $result }}
{{ if .Err }}
{{/* log the error, insert a default image etc. *}}
{{ else }}
<img src="{{ .RelPermalink }}" width="{{ .Width }}" height="{{ .Height }}" alt="">
{{ end }}
{{ end }}
```
Note that the default behaviour is still to fail the build, but we will delay that error until you start using the `Resource`.
Fixes#9529
The existing endpoint will be retired and removed on November 23, 2021.
References:
- https://twittercommunity.com/t/consolidating-the-oembed-functionality/154690
- https://developer.twitter.com/en/docs/twitter-for-websites/oembed-api#Embedded
This is a backward compatible change.
The existing endpoint requires a single parameter: the id of the tweet.
The new endpoint requires two parameters: the id of the tweet, and the
user with whom it is associated. For the moment, if you supply the wrong
user, the request will be redirected (with a small delay) to the correct
user/id pair. This behavior is undocumented, but we will take advantage
of it as Hugo site authors transition to the new syntax.
{{< tweet 1453110110599868418 >}} --> works, throws warning, deprecate at some point
{{< tweet user="SanDiegoZoo" id="1453110110599868418" >}} --> new syntax
Fixes#8130
The old implementation had some issues, mostly related to the context (e.g. name, file paths) passed to the template.
This new implementation is using the exact same code path for evaluating the pages as in a regular build.
This also makes it more robust and easier to reason about in a multilingual setup.
Now, if you are explicit about the target path, Hugo will now always pick the correct mount and language:
```bash
hugo new content/en/posts/my-first-post.md
```
Fixes#9032Fixes#7589Fixes#9043Fixes#9046Fixes#9047
Previously, *minifyTransformation.Transform suppressed the
error returned by t.m.Minify. This meant that when minification
returned an error, the error would not reach the user. Instead,
minification would silently fail. For example, if a JavaScript
file included a call to the Date constructor with:
new Date(2020, 04, 02)
The package that the minification library uses to parse JS files,
github.com/tdewolff/parse would return an error, since "04" would
be parsed as a legacy octal. However, the JS file would remain
un-minified with no error.
Fixing this is not as simple as replacing "_" with an "err" in
*minifyTransformation.Transform, however (though this is
necessary). If we only returned this error from Transform,
then hugolib.TestResourceMinifyDisabled would fail. Instead of
being a no-op, as TestResourceMinifyDisabled expects, using the
"minify" template function with a "disableXML=true" config
setting instead returns the error, "minifier does not exist for
mimetype."
The "minifier does not exist" error is returned because of the
way minifiers.New works. If the user's config disables
minification for a particular MIME type, minifiers.New does
not add it to the resulting Client's *minify.M. However, this
also means that when the "minify" template function is executed,
a *resourceAdapter's transformations still add a minification.
When it comes time to call the minify.Minifier for a specific
MIME type via *M.MinifyMimetype, the github.com/tdewolff/minify
library throws the "does not exist" error for the missing MIME
type.
The solution was to change minifiers.New so, instead of skipping
a minifier for each disabled MIME type, it adds a NoOpMinifier,
which simply copies the source to the destination without
minification. This means that when the "minify" template
function is used for a particular resource, and that resource's
MIME type has minification disabled, minification is genuinely
skipped, and does not result in an error.
In order to add this, I've fixed a possibly unwanted interaction
between minifiers.TestConfigureMinify and
hugolib.TestResourceMinifyDisabled. The latter disables
minification and expects minification to be a no-op. The former
disables minification and expects it to result in an error. The
only reason hugolib.TestResourceMinifyDisabled passes in the
original code is that the "does not exist" error is suppressed.
However, we shouldn't suppress minification errors, since they
can leave users perplexed. I've changed the test assertion in
minifiers.TestConfigureMinify to expect no errors and a no-op
if minification is disabled for a particular MIME type.
Fixes#8954
Unless the merge strategy is set up to do so.
For `disableKinds` the current workaround is to make sure the project config has an entry, even if is empty:
```
disableKinds = []
```
Note that this issue only touches root, non-map config-values that either is not set in project config or in Hugo's defaults.
Fixes#8866
We have been using `go-toml` for language files only. This commit makes it the only TOML library.
It's spec compliant and very fast.
A benchark building a site with 200 pages with TOML front matter:
```bash
name old time/op new time/op delta
SiteNew/Regular_TOML_front_matter-16 48.5ms ± 1% 47.1ms ± 1% -2.85% (p=0.029 n=4+4)
name old alloc/op new alloc/op delta
SiteNew/Regular_TOML_front_matter-16 16.9MB ± 0% 16.7MB ± 0% -1.56% (p=0.029 n=4+4)
name old allocs/op new allocs/op delta
SiteNew/Regular_TOML_front_matter-16 302k ± 0% 296k ± 0% -2.20% (p=0.029 n=4+4)
```
Note that the front matter unmarshaling is only a small part of building a site, so the above is very good.
Fixes#8801
This commit started out investigating a `concurrent map read write` issue, ending by replacing the map with a struct.
This is easier to reason about, and it's more effective:
```
name old time/op new time/op delta
SiteNew/Regular_Deep_content_tree-16 71.5ms ± 3% 69.4ms ± 5% ~ (p=0.200 n=4+4)
name old alloc/op new alloc/op delta
SiteNew/Regular_Deep_content_tree-16 29.7MB ± 0% 27.9MB ± 0% -5.82% (p=0.029 n=4+4)
name old allocs/op new allocs/op delta
SiteNew/Regular_Deep_content_tree-16 313k ± 0% 303k ± 0% -3.35% (p=0.029 n=4+4)
```
See #8749
Currently the generated `<pre>` element isn't fully accessible as it can't be focused by keyboard users.
To make this fully accessible, the attribute `tabindex="0"` should be added to the `<pre>` tag.
Closes#7194
This commit adds support for using the `cascade` keyword in your configuration file(s), e.g. `config.toml`.
Note that
* Every feature of `cascade` is available, e.g. `_target` to target specific page sets.
* Pages, e.g. the home page, can overwrite the cascade defined in config.
Fixes#8741
We have some commands that will continue even if the config loading fails (e.g. because a module can not be found). In Hugo 0.84.0 we introduced a new `_merge` flag that we removed once the configuration was loaded. But we did not do that in error situations, leading to failures in some situations.
This commit fixes that by making sure the configuration is always cleaned before return, even in error situations.
Fixes#8697
The main motivation behind this is simplicity and correctnes, but the new small config library is also faster:
```
BenchmarkDefaultConfigProvider/Viper-16 252418 4546 ns/op 2720 B/op 30 allocs/op
BenchmarkDefaultConfigProvider/Custom-16 450756 2651 ns/op 1008 B/op 6 allocs/op
```
Fixes#8633Fixes#8618Fixes#8630
Updates #8591Closes#6680Closes#5192
Currently, no name shortcodes (`{{< >}}`) enter unexpected branch and
throw `BUG: template info not set`. This patch checks if shortcode has
name or not earlier and throws specific error.
Closes#6866
.Site.Menus is mutated when it is sorted for now and this causes concurrency problem (#7594)
In this patch, each related sort function copies Menu before sorting to prevent
race condition.
Pages already have such a sort and cache logic and this patch is identical to it.
Closes#7594
- Reorder code blocks
- Rename cssClassCollectorWriter to htmlElementCollectorWriter, as it just collect html element information
- Expand benchmark to test for minified and unminified content
Fixes#8396Fixes#8417
The recent fix for #8346 had an unfortunate side-effect, as it prevented overrides of config options used in the module collector, e.g. `HUGO_MODULE_REPLACEMENTS`.
This commit fixes that by applying the overrides twice.
So we can use it and output.Format as map key etc.
This commit also fixes the media.Type implementation so it does not need to mutate itself to handle different suffixes for the same MIME type, e.g. jpg vs. jpeg.
This means that there are no Suffix or FullSuffix on media.Type anymore.
Fixes#8317Fixes#8324
This commit fixes some issues where modules in /assets share the same name as in node_modules.
This was not intended, and with this commit the node_modules-components should be isolated. If you want to redefine something inside node_modules, use the `defines` option.
Fixes#7948
Updates #7777
Added support to allow SourceMap files to be external to the build.
In addition added more information when the js compilation has an error.
Correctly append sourceMappingURL to output file.
Fix merge conflict.
The TOML lib unmarshals slices of string maps to []map[string]interface{}
whereas YAML and JSON decode to []interface{}
The existing tests only check for TOML working correctly, and _target
with cascade did not work at all for frontmatter defined in other formats.
Add a function to normalize those slices
Fixes#7874
This change is mostly motivated to get a more stable CI build (we're building the Hugo site there, with Instagram and Twitter shortcodes sometimes failing).
Fixes#7866
This commit also introduces a convention where these common JS config files, including `package.hugo.json`, gets mounted into:
```
assets/_jsconfig
´``
These files mapped to their real filename will be added to the environment when running PostCSS, Babel etc., so you can do `process.env.HUGO_FILE_TAILWIND_CONFIG_JS` to resolve the real filename.
But do note that `assets` is a composite/union filesystem, so if your config file is not meant to be overridden, name them something specific.
This commit also adds adds `workDir/node_modules` to `NODE_PATH` and `HUGO_WORKDIR` to the env when running the JS tools above.
Fixes#7644Fixes#7656Fixes#7675
* Fix change detection when .GetPage/site.GetPage is used from shortcode
* Fix stale content for GetPage results with short name lookups on server reloads
Fixes#7623Fixes#7624Fixes#7625
This commit solves the relative path problem with asciidoctor tooling. An include will resolve relatively, so you can refer easily to files in the same folder.
Also `asciidoctor-diagram` and PlantUML rendering works now, because the created temporary files will be placed in the correct folder.
This patch covers just the Ruby version of asciidoctor. The old AsciiDoc CLI EOLs in Jan 2020, so this variant is removed from code.
The configuration is completely rewritten and now available in `config.toml` under the key `[markup.asciidocext]`:
```toml
[markup.asciidocext]
extensions = ["asciidoctor-html5s", "asciidoctor-diagram"]
workingFolderCurrent = true
trace = true
[markup.asciidocext.attributes]
my-base-url = "https://example.com/"
my-attribute-name = "my value"
```
- backends, safe-modes, and extensions are now whitelisted to the popular (ruby) extensions and valid values.
- the default for extensions is to not enable any, because they're all external dependencies so the build would break if the user didn't install them beforehand.
- the default backend is html5 because html5s is an external gem dependency.
- the default safe-mode is safe, explanations of the modes: https://asciidoctor.org/man/asciidoctor/
- the config is namespaced under asciidocext_config and the parser looks at asciidocext to allow a future native Go asciidoc.
- `uglyUrls=true` option and `--source` flag are supported
- `--destination` flag is required
Follow the updated documentation under `docs/content/en/content-management/formats.md`.
This patch would be a breaking change, because you need to correct all your absolute include pathes to relative paths, so using relative paths must be configured explicitly by setting `workingFolderCurrent = true`.
Before a robots.txt is created in every Site. So in public/robots.txt if there are no languages (was correct). But if there are multiple languages in every language directory, too (was wrong). If defaultContentLanguageInSubdir is true, no language is created into the root directory, so no robots.txt is in the root directory (was wrong). If multihosts are configured for each language, that is the only case where one robots.txt must be created in each language directory (was correct).
I've changed the behaviour, that only in the multihost case the robots.txt is created in the language directories. In any other case it is created in public/robots.txt. I've also added tests that files are not created in the wrong directories.
Fixes#5160
See also #4193
And we have taken great measures to limit potential site breakage:
* For `disableKinds` and `outputs` we try to map from old to new values if possible, if not we print an ERROR that can be toggled off if not relevant.
* The layout lookup is mostly compatible with more options for the new `term` kind.
That leaves:
* Where queries in site.Pages using taxonomy/taxonomyTerm Kind values as filter.
* Other places where these kind value are used in the templates (classes etc.)
Fixes#6911Fixes#7395
In the internal Radix we stored the directory based nodes without a traling slash, e.g. `/blog`.
The original motivation was probably to make it easy to do prefix searching: Give me all ancestors.
This, however have lead to some ambigouty with overlapping directory names.
This particular problem was, however, not possible to work around in an easy way, so from now we store these as `/blog/`.
Fixes#7301
The main use case for this is to use with resources.PostProcess and resources.PostCSS with purgecss.
You would normally set it up to extract keywords from your templates, doing it from the full /public takes forever for bigger sites.
Doing the template thing misses dynamically created class names etc., and it's hard/impossible to set up in when using themes.
You can enable this in your site config:
```toml
[build]
writeStats = true
```
It will then write a `hugo_stats.json` file to the project root as part of the build.
If you're only using this for the production build, you should consider putting it below `config/production`.
You can then set it up with PostCSS like this:
```js
const purgecss = require('@fullhuman/postcss-purgecss')({
content: [ './hugo_stats.json' ],
defaultExtractor: (content) => {
let els = JSON.parse(content).htmlElements;
return els.tags.concat(els.classes, els.ids);
}
});
module.exports = {
plugins: [
require('tailwindcss'),
require('autoprefixer'),
...(process.env.HUGO_ENVIRONMENT === 'production' ? [ purgecss ] : [])
]
};
```
Fixes#6999
This means that any HTML file inside /content will be treated as a regular file.
If you want it processes with shortcodes and a layout, add front matter.
The defintion of an HTML file here is:
* File with extension .htm or .html
* With first non-whitespace character "<" that isn't a HTML comment.
This is in line with the documentation.
Fixes#7030Fixes#7028
See #6789
Previously gordmark-based TOC renderes only `KindText` and `KindString`
This commit expands target node with Goldmark's renderer
I am not sure of what are expected results as TOC contents in some (rare) cases
but Blackfriday's behaviours are fundamentally respected.
For example,
- image `[image text](link)` is rendered as `<img>` tag
- GFM AutoLink `gohugo.io` is rendered as text
* Render AutoLink as <a> tag as Blackfriday does
Fixes#6736Fixes#6809
This was changed with good intentions in 0.63.0.
This behaviour was not documented, but it was of course in use.
This commit rolls back to how it behaved before:
For `Page.Type` you will get:
* `type` from front matter if set.
* `.Section`
* If none of the above returns anything, return "page"
Fixes#6805
This more or less completes the simplification of the template handling code in Hugo started in v0.62.
The main motivation was to fix a long lasting issue about a crash in HTML content files without front matter.
But this commit also comes with a big functional improvement.
As we now have moved the base template evaluation to the build stage we now use the same lookup rules for `baseof` as for `list` etc. type of templates.
This means that in this simple example you can have a `baseof` template for the `blog` section without having to duplicate the others:
```
layouts
├── _default
│ ├── baseof.html
│ ├── list.html
│ └── single.html
└── blog
└── baseof.html
```
Also, when simplifying code, you often get rid of some double work, as shown in the "site building" benchmarks below.
These benchmarks looks suspiciously good, but I have repeated the below with ca. the same result. Compared to master:
```
name old time/op new time/op delta
SiteNew/Bundle_with_image-16 13.1ms ± 1% 10.5ms ± 1% -19.34% (p=0.029 n=4+4)
SiteNew/Bundle_with_JSON_file-16 13.0ms ± 0% 10.7ms ± 1% -18.05% (p=0.029 n=4+4)
SiteNew/Tags_and_categories-16 46.4ms ± 2% 43.1ms ± 1% -7.15% (p=0.029 n=4+4)
SiteNew/Canonify_URLs-16 52.2ms ± 2% 47.8ms ± 1% -8.30% (p=0.029 n=4+4)
SiteNew/Deep_content_tree-16 77.9ms ± 1% 70.9ms ± 1% -9.01% (p=0.029 n=4+4)
SiteNew/Many_HTML_templates-16 43.0ms ± 0% 37.2ms ± 1% -13.54% (p=0.029 n=4+4)
SiteNew/Page_collections-16 58.2ms ± 1% 52.4ms ± 1% -9.95% (p=0.029 n=4+4)
name old alloc/op new alloc/op delta
SiteNew/Bundle_with_image-16 3.81MB ± 0% 2.22MB ± 0% -41.70% (p=0.029 n=4+4)
SiteNew/Bundle_with_JSON_file-16 3.60MB ± 0% 2.01MB ± 0% -44.20% (p=0.029 n=4+4)
SiteNew/Tags_and_categories-16 19.3MB ± 1% 14.1MB ± 0% -26.91% (p=0.029 n=4+4)
SiteNew/Canonify_URLs-16 70.7MB ± 0% 69.0MB ± 0% -2.40% (p=0.029 n=4+4)
SiteNew/Deep_content_tree-16 37.1MB ± 0% 31.2MB ± 0% -15.94% (p=0.029 n=4+4)
SiteNew/Many_HTML_templates-16 17.6MB ± 0% 10.6MB ± 0% -39.92% (p=0.029 n=4+4)
SiteNew/Page_collections-16 25.9MB ± 0% 21.2MB ± 0% -17.99% (p=0.029 n=4+4)
name old allocs/op new allocs/op delta
SiteNew/Bundle_with_image-16 52.3k ± 0% 26.1k ± 0% -50.18% (p=0.029 n=4+4)
SiteNew/Bundle_with_JSON_file-16 52.3k ± 0% 26.1k ± 0% -50.16% (p=0.029 n=4+4)
SiteNew/Tags_and_categories-16 336k ± 1% 269k ± 0% -19.90% (p=0.029 n=4+4)
SiteNew/Canonify_URLs-16 422k ± 0% 395k ± 0% -6.43% (p=0.029 n=4+4)
SiteNew/Deep_content_tree-16 401k ± 0% 313k ± 0% -21.79% (p=0.029 n=4+4)
SiteNew/Many_HTML_templates-16 247k ± 0% 143k ± 0% -42.17% (p=0.029 n=4+4)
SiteNew/Page_collections-16 282k ± 0% 207k ± 0% -26.55% (p=0.029 n=4+4)
```
Fixes#6716Fixes#6760Fixes#6768Fixes#6778
This is the first version of Goldmark that supports all the
Smartypants-style typographic punctuation transformations. Now, a
straight single quote in the middle of a word is translated into a curly
quote (e.g. "that's" becomes "that’s"). Earlier versions leave
them untouched. This brings Goldmark in line with Blackfriday.
Fixes#6571.
This commit also
* revises the change detection for templates used by content files in server mode.
* Adds a Page.RenderString method
Fixes#6545Fixes#4663Closes#6043
This is a big commit, but it deletes lots of code and simplifies a lot.
* Resolving the template funcs at execution time means we don't have to create template clones per site
* Having a custom map resolver means that we can remove the AST lower case transformation for the special lower case Params map
Not only is the above easier to reason about, it's also faster, especially if you have more than one language, as in the benchmark below:
```
name old time/op new time/op delta
SiteNew/Deep_content_tree-16 53.7ms ± 0% 48.1ms ± 2% -10.38% (p=0.029 n=4+4)
name old alloc/op new alloc/op delta
SiteNew/Deep_content_tree-16 41.0MB ± 0% 36.8MB ± 0% -10.26% (p=0.029 n=4+4)
name old allocs/op new allocs/op delta
SiteNew/Deep_content_tree-16 481k ± 0% 410k ± 0% -14.66% (p=0.029 n=4+4)
```
This should be even better if you also have lots of templates.
Closes#6594
