* Increase max feed URL length and drop unicity
#fix https://github.com/FreshRSS/FreshRSS/issues/4338
Drop the unicity constraint on our feed URL. In practice, this did not add much value as identical feeds could have different URLs. And it generated several problems, for instance during renaming or automatic redirections such as from HTTP to HTTPS, with collisions for which we dot not have any elegant handling.
I have kept a high limit of 32768 because cURL does not seem to have any limit except memory, which might lead to memory problems. This is the highest server-side value reported by https://www.sistrix.com/ask-sistrix/technical-seo/site-structure/url-length-how-long-can-a-url-be
* Same for Web site URL
For consistency
* fix: Fix undefined GLOB_BRACE on Alpine
The manual states that:
> Note: The GLOB_BRACE flag is not available on some non GNU systems,
> like Solaris or Alpine Linux.
This generated an error on Alpine.
Reference: https://www.php.net/manual/function.glob.php
* fix: List details of feeds for OPML exportation
The details are necessary to export the XPath information, the CSS full
content path and read actions filters.
* Update LibOpml to 0.4.0
* Refactor OPML importation to be more robust
First, it fixes two regressions introduced by the update of lib_opml:
- title attribute is used when text attribute is missing;
- the OPML category attribute is used as a fallback for feeds categories.
In a related way, if also fixes a problem when a feed had both a parent
category outline and a category attribute. Before, it only considered the
attribute as its category, but now it considers the parent outline.
Then, it counts category limit correctly by not increasing
`$nb_categories` if the category already exists.
* Exclude lib_opml from the CodeSniffer
* Fix variable names when logging some errors
* Fix catch of LibOpml Exception
* Make sure to declare the category
* Exclude lib_opml from PHPStan analyze
* Disable markdownlint for lib_opml
* Fix typos
* Use auto-loading and allow updates via Composer
* Fix broken links to lib_opml
* Bring back the ability to import the OPML frss:opmlUrl attribute
* Refactor the logs of OPML errors
* Update lib_opml to the version 0.5.0
Co-authored-by: Alexandre Alapetite <alexandre@alapetite.fr>
* Better enclosures
#fix https://github.com/FreshRSS/FreshRSS/issues/4702
Improvement of https://github.com/FreshRSS/FreshRSS/pull/2898
* A few fixes
* Better enclosure titles
* Improve thumbnails
* Implement thumbnail for HTML+XPath
* Avoid duplicate enclosures
#fix https://github.com/FreshRSS/FreshRSS/issues/1668
* Fix regex
* Add basic support for media:credit
And use <figure> for enclosures
* Fix link encoding + simplify code
* Fix some SimplePie bugs
Encoding errors in enclosure links
* Remove debugging syslog
* Remove debugging syslog
* SimplePie fix multiple RSS2 enclosures
#fix https://github.com/FreshRSS/FreshRSS/issues/4974
* Improve thumbnails
* Performance with yield
Avoid generating all enclosures if not used
* API keep providing enclosures inside content
Clients are typically not showing the enclosures to the users (tested with News+, FeedMe, Readrops, Fluent Reader Lite)
* Lint
* Fix API output enclosure
* Fix API content strcut
* API tolerate enclosures without a type
Before, the feeds were not ordered every time there was a change in the category
feed list. This behavior was causing discrepancies in the displayed list.
Now, the feeds are ordered every time there is a change in the category feed list.
See #4790
Fix regressions from https://github.com/FreshRSS/FreshRSS/pull/4561
Example:
```
PHP Fatal error: Uncaught TypeError: Argument 1 passed to checkToken() must be an instance of FreshRSS_UserConfiguration, instance of Minz_Configuration given, called in /var/www/FreshRSS/p/api/greader.php on line 1091 and defined in /var/www/FreshRSS/p/api/greader.php:223
Stack trace:
#0 /var/www/FreshRSS/p/api/greader.php(1091): checkToken()
#1 {main}
thrown in /var/www/FreshRSS/p/api/greader.php on line 223
```
Improvement of https://github.com/FreshRSS/FreshRSS/pull/4110
#fix https://github.com/FreshRSS/FreshRSS/issues/4600
We first need to commit the SQL transaction before being able to auto-update.
Avoid error such as:
> 25P02: 7 ERROR: current transaction is aborted, commands ignored until end of transaction block
* API protection limit content size
We had no limit to the length of an article sent through our GReader API to clients, which was a problem:
The article https://devblogs.microsoft.com/dotnet/performance_improvements_in_net_7/
from the feed https://devblogs.microsoft.com/dotnet/feed/
is about 700KB long, which makes some clients (tested with News+) fail.
We should try other clients to verify that 500KB is a safe enough limit.
* Light optimisation
* Use class constant
Regression from https://github.com/FreshRSS/FreshRSS/pull/4202
> Warning: Declaration of FreshRSS_EntryDAOSQLite::markRead($ids, $is_read = true) should be compatible with FreshRSS_EntryDAO::markRead($ids, bool $is_read = true) in /var/www/FreshRSS/app/Models/EntryDAOSQLite.php on line 3
* Easier full-text search possibility
Contributes to https://github.com/FreshRSS/FreshRSS/issues/1331
Avoid concats in searches to make text indexes easier to build
* Fix tests
* Documentation
* Basic support for negative searches with parentheses
* `!((author:Alice intitle:hello) OR (author:Bob intitle:world))`
* `(author:Alice intitle:hello) !(author:Bob intitle:world)`
* `!(S:1 OR S:2)`
* Minor documentation / comment
* Remove syslog debug line
* Refactor entry to GReader API format
Some code was copied in two locations and not completely uniform.
Cleaning of related variables and functions (e.g. better types for entries and categories as objects vs. as IDs).
Usecase: I need to call the same GReader-compatible serialization from an extension
* Fixed some edge cases
* Keep summary instead of content
`summary` and `content` seems to be used interchangeably in the Google Reader API. We have been using `summary` for our client API and `content` in our export/import, so stick to that.
* Add database field attributes for entries
Just like we already have for categories, feeds, etc.
No core use yet, but allows in particular extensions to save per-entry data
* Fix PHPStand
* Fix wrong variable
* New option to automatically mark as read gone articles
Option to automatically and immediately mark as read entries / articles that are no longer provided in their upstream RSS / ATOM / XPath feed
* Reduce SQL queries
Optimisation: Perform cache update only once
Set feed error state to true if the *list* of items cannot be find by XPath, but do not set the error state to true if that list happens to be empty (the resulting feed will be with an *empty* state instead of *error* state)
Follow-up of https://github.com/FreshRSS/FreshRSS/pull/4220
