* improve ci
* improve ci
* use actions/setup-node@v2 everywhere
* run pwd for test
* debug gh action
* debug glob CI issue?
* Separate v1/v2 tests due to shell conflict
* cleanup
* test v1 change
* circleci fix
* fix test docusaurus v1 paths
* Refactor CI script names and use paths to filter actions from running unnecessary
* fix lighthouse url
* v1 tests runInBand
* try to fix v1 tests race conditions
* change rootDir for v1 tests
* minor CI improvements
* docs i18n initial poc
* docs i18n initial poc
* docs i18n initial poc
* docs i18n initial poc
* crowdin-v2 attempt
* fix source
* use crowdin env variable
* try to install crowdin on netlify
* try to install crowdin on netlify
* try to use crowdin jar directly
* try to curl the crowdin jar
* add java version cmd
* try to run crowdin jar in netlify
* fix translatedDocsDirPath
* fix loadContext issue due to site baseUrl not being modified in generted config file
* real validateLocalesFile
* add locale option to deploy command
* better LocalizationFile type
* create util getPluginI18nPath
* better core localization context loading code
* More explicit VersionMetadata type for localized docs folders
* Ability to translate blog posts with Crowdin!
* blog: refactor markdown loader + report broken links + try to get linkify working better
* upgrade crowdin config to upload all docs folder files except source code related files
* try to support translated pages
* make markdown pages translation work
* add write-translations cli command template
* fix site not reloaded with correct options
* refactor a bit the read/write of @generated/i18n.json file
* Add <Translate> + translate() API + use it on the docusaurus homepage
* watch locale translation dir
* early POC of adding babel parsing for translation extraction
* fs.stat => pathExists
* add install:fast script
* TSC: noUnusedLocals false as it's already checked by eslint
* POC of extracting translations from source code
* minor typo
* fix extracted key to code
* initial docs extracted translations
* stable plugin translations POC
* add crowdin commands
* quickfix for i18n deployment
* POC of themeConfig translation
* add ability to have localized site without path prefix
* sidebar typo
* refactor translation system to output multiple translation files
* translate properly the docs plugin
* improve theme classic translation
* rework translation extractor to handle new Chrome I18n JSON format (include id/description)
* writeTranslations: allow to pass locales cli arg
* fix ThemeConfig TS issues
* fix localizePath errors
* temporary add write-translations to netlify deploy preview
* complete example of french translated folder
* update fr folder
* remove all translations from repo
* minor translation refactors
* fix all docs-related tests
* fix blog feed tests
* fix last blog tests
* refactor i18n context a bit, extract codeTranslations in an extra generated file
* improve @generated/i18n type
* fix some i18n todos
* minor refactor
* fix logo typing issue after merge
* move i18n.json to siteConfig instead
* try to fix windows CI build
* fix config test
* attempt to fix windows non-posix path
* increase v1 minify css jest timeout due to flaky test
* proper support for localizePath on windows
* remove non-functional install:fast
* docs, fix docsDirPathLocalized
* fix Docs i18n / md linkify issues
* ensure theme-classic swizzling will use "nextjs" sources (transpiled less aggressively, to make them human readable)
* fix some snapshots
* improve themeConfig translation code
* refactor a bit getPluginI18nPath
* readTranslationFileContent => ensure files are valid, fail fast
* fix versions tests
* add extractSourceCodeAstTranslations comments/resource links
* ignore eslint: packages/docusaurus-theme-classic/lib-next/
* fix windows CI with cross-env
* crowdin ignore .DS_Store
* improve writeTranslations + add exhaustive tests for translations.ts
* remove typo
* Wire currentLocale to algolia search
* improve i18n locale error
* Add tests for translationsExtractor.ts
* better code translation extraction regarding statically evaluable code
* fix typo
* fix typo
* improve theme-classic transpilation
* refactor + add i18n tests
* typo
* test new utils
* add missing snapshots
* fix snapshot
* blog onBrokenMarkdownLink
* add sidebars tests
* theme-classic index should now use ES modules
* tests for theme-classic translations
* useless comment
* add more translation tests
* simplify/cleanup writeTranslations
* try to fix Netlify fr deployment
* blog: test translated md is used during feed generation
* blog: better i18n tests regarding editUrl + md translation application
* more i18n tests for docs plugin
* more i18n tests for docs plugin
* Add tests for pages i18n
* polish docusaurus build i18n logs
* fix(v1): Fix page title render issue when referred by search result
When Algolia DocSearch query finds a match for a page's title, it attempts to generate
a permalink. Because the page title element (`h1`) does not have an `id`, Algolia
generates uses the `id` from closes parent element. Because of this, the page title
scrolls to a position that is slightly overlayed by the fixed top navigation bar.
This fix sets an `id` for the page title so that the search result is able to generate
a more accurate permalink.
* Adjust css for handling post title to be on the top when referred by an anchor
The post title can sometimes be referred by an anchor using the "id" of that element.
In that case the title will be automatically be the first element within the viewport.
Since we have a header fixed at the top of the view port, the title becomes hidden or
not visible. That's why some css adjustments are needed so that if any user ends up
with a link to a page that is referring to the post title (i.e. auto generated anchor
link by algolia DocSearch).
The css code uses pseudo element `:before` to make the adjustments. Details on this
can be found in the following article:
https://css-tricks.com/hash-tag-links-padding/
* Adjust CSS so that different selectors are on separate lines
The approach here is to first strip the HTML from the heading's content, then rendered it with markdown to get the HTML content for the TOC entry, then to strip the HTML from the rendered content again, as to get the text for the TOC entry's link.
Adds an additional dependency of striptags (MIT licensed)
Example TOC Entry, given the heading of:
```markdown
```
```javascript
{
hashLink: 'foo',
rawContent: '<a name="foo"></a> _Foo_',
content: '<em>Foo</em>',
children: []
}
```
Previously this TOC entry would be:
```javascript
{
hashLink: 'a-name-foo-a-_foo_',
rawContent: '<a name="foo"></a> _Foo_',
content: '<a name="foo"></a> <em>Foo</em>',
children: []
}
```
closes issue #1703
The warning is produced on `cssnano` call. See below:
> Without `from` option PostCSS could generate wrong source map and will not find Browserslist config. Set it to CSS file path or to `undefined` to prevent this warning.
Reproduction:
```shell
mkdir work
cd work
npx docusaurus-init
cd website
yarn install
yarn run build
```
Environments:
- Node: 12.6.0
- Yarn: 1.17.3
Currently the items in table of contents are highlighted as bold based on items being
viewed on the screen or as the user scrolls up/down on a page. Added the `primaryColor`
of the site's configuration to make the current section being viewed standout more
in the table of content.
closes#1608
* Replace apostrophe with empty string in header
This change is to replicate the github behviour with respect to apostrophe in
headers. When there is an apostrophe in a header, github replaces it with empty string
when creating an anchor link to that header. Docusaurus should follow the same convention.
* Add `unlisted` header option for blog posts, fixes#1393.
Previously, the blog sidebar listed the most recent blog posts sorted by
their publish date. This commit adds the ability to hide a specific blog
post from the sidebar and the feed of all blog posts by adding the
`unlisted: true` header option.
```md
title: My Draft Post
unlisted: true # Hide from blog sidebar and main blog page feed.
---
```
* Rename "unlisted" into "draft".
In this case, the `docsNavContainer` does not pick up any styles or attributes from the `container` component it is wrapped in.
It does however pick up the pesky `.container` className, which when paired with bootstrap, messes up the styling of the top navigation in mobile.
This change is completely non-destructive, but helps enable the use of bootstrap on top of Docusaurus.
