This feature is based general.enable_async_metadata setting and works with existing metadata.js file.
The script is compatible with any template:
- the thumbnail div bloc must have attribute
- the bookmark bloc must have attribute with the bookmark ID as value
Fixes#1564
How it works:
1. when a fulltext search is made, Shaarli looks for the first
occurence position of every term matching the search. No change here,
but we store these positions in an array, in Bookmark's additionalContent.
2. when formatting bookmarks (through BookmarkFormatter
implementation):
1. first we insert specific tokens at every search result positions
2. we format the content (escape HTML, apply markdown, etc.)
3. as a last step, we replace our token with displayable span
elements
Cons: this tightens coupling between search filters and formatters
Pros: it was absolutely necessary not to perform the
search twice. this solution has close to no impact on performances.
Fixes#205
- Shaarli can be fully used by prefixing any URL with /index.php/
- {$base_path} used in templates already works with this configuration
- Assets path (outside of theme's assets) must be prefixed with {$root_url}/
- Documentation section in « Server configuration »
Fixes#1590
- There is a new standalone script (metadata.js) which requests
a new controller to get bookmark metadata and fill the form async
- This feature is enabled with the new setting: general.enable_async_metadata
(enabled by default)
- general.retrieve_description is now enabled by default
- A small rotating loader animation has a been added to bookmark inputs
when metadata is being retrieved (default template)
- Custom JS htmlentities has been removed and mathiasbynens/he
library is used instead
Fixes#1563
Parameters typing and using strict types overall increase the codebase
quality by enforcing the a given parameter will have the expected type.
It also removes the need to unnecessary unit tests checking methods
behavior with invalid input.
XSS vulnerabilities fixed in editlink, linklist, tag.cloud and tag.list.
Also fixed tag search with special characters: urlencode function needs to be applied on raw data, before espaping, otherwise the rendered URL is wrong.