* git update auto change to edge branch
For existing installations using automatic git update, checkout *edge* branch if it was still using *master* or *dev*.
* One more prune
* Fix several small issues
* theirs does not work here
* Use migration mechanism
* Better handling of Migration errors
* Test details
* Fix tests
* Do not use new migration system for now
* Remove file data/do-install.txt
This file was painful during update because we had to remember to delete
it each time. It added a security issue by allowing an attacker to
reinstall FreshRSS during the update process.
The (more powerful) file data/applied_migrations.txt has been introduced
in 8619cf6fa to replace do-install.txt. We had to wait for at least one
release in order to make sure existing instances of FreshRSS created the
migration file. It should be ok now.
* Replace i18n install.not_deleted key
* Update documentation to update FreshRSS
When a feed returns an HTTP 410 Gone, mute the corresponding feed, i.e. stop refreshing it.
Example of such feed, Les Décodeurs (Libération) https://rss.liberation.fr/rss/100893/
#fix https://github.com/FreshRSS/FreshRSS/pull/3546
When `loadHTML()` is given a null or empty string.
```
PHP Fatal error: Uncaught ValueError: DOMDocument::loadHTML(): Argument #1 ($source) must not be empty in /var/www/freshrss/lib/SimplePie/SimplePie/Locator.php:83
```
Before, only standard shortcuts were supported. When other shortcuts were
configured, they were filtered out.
Now, those shortcuts are stored in the configuration and an alert message
is displayed to alert the user that he is using non-standard shortcuts.
See #3481
* Update Travis line length
* Also check whitespace in CSS files
* Fix line length ext.php
* More syntax, string templates
* Fix exclude-pattern
* Test JS files as well
* fix: handle big xml files which cause out of memory exceptions by working with chunks in cleanMd5 function (because of preg_replace) and parse (because of xml_parse)
* Review
* Fixes in error handling (case of the last call to xml_parse, case of
error during fopen, break in case of XML error...)
* Takes advantage of the chunking for computing the cache hash
* Larger chunks of 1MB
Co-authored-by: e <bokes74743@tjuln.com>
Co-authored-by: Alexandre Alapetite <alexandre@alapetite.fr>
Before, setting values did not refresh the configuration cache. Thus
generating some weird behavior when configuring extensions.
Now, the cache is updated with the most recent values when the
configuration is modified.
Fix regression from https://github.com/FreshRSS/FreshRSS/pull/3407
Identifier must not be indented
PHP 5.6.36 (cli) (built: Apr 28 2018 21:14:51)
Copyright (c) 1997-2016 The PHP Group
Zend Engine v2.6.0, Copyright (c) 1998-2016 Zend Technologies
with Zend OPcache v7.0.6-dev, Copyright (c) 1999-2016, by Zend
Technologies
PHP Parse error: syntax error, unexpected end of file in
lib/lib_rss.php on line 608
Errors parsing lib/lib_rss.php
* bump default logout from 30 to 365 days
* * Change cookie duration to constant
* Change cookie duration to three months
* use class
* use 90 days (otherwise login form says 91.3 days)
* change class
* also this works now
* Better error message
* inconsistent dot with the other message
* Better error message
* add errorMessage()
* fix style
* html escape the error title
* also html escape error message
* remove spaces before parentheses
* rework the error message
* Minz-friendly
* Update message
Do not advise running this script as wrong user
* Update lib/lib_rss.php
Co-authored-by: Martin <spleefer90@gmail.com>
Co-authored-by: Martin Rys <martin@rys.pw>
Co-authored-by: Frans de Jonge <fransdejonge@gmail.com>
Co-authored-by: Alexandre Alapetite <alexandre@alapetite.fr>
This will simplify extension code by removing a lot of logic from
the extension itself when it's not needed. I've tested it on one
of my extension with all the other recent extension modifications
and I could remove half of the code needed before.
Before, only one extension could be configured at a time. Thus we
were loosing the configuration for other extensions when saving.
Now, each extension can be saved without overriding data.
See #3397
Before, the extension configuration was handled by its author. There
was discrepancies between extensions on how the configuration was
stored.
Now, we could rely on a single way of storing configuration. This won't
invalidate how the extensions are storing their configuration but will
allow authors to focus on what is important.
* Add constant for PHP requirements
This new constant is used for PHP version check.
This way, we won't forget to modify some part of the code base.
* Remove PHP version checks
Some checks were obsolete because they were checking unsupported
PHP versions.
* Possiblity to autoinstall in Docker Compose
#fix https://github.com/FreshRSS/FreshRSS/issues/3349
It is simply calling our existing CLI: do-install.php and
create-user.php
https://github.com/FreshRSS/FreshRSS/tree/master/cli
FreshRSS will typically be ready a few seconds before the database, so
introduce a tolerance when the database is not available / up (yet) by
trying a few times to connect. Also useful to avoid service interruption
when DB service is restarted.
Example:
```yml
freshrss-app:
image: freshrss/freshrss
container_name:
freshrss-app
hostname: freshrss-app
restart: unless-stopped
ports:
- "8080:80"
depends_on:
- freshrss-db
volumes:
- data:/var/www/FreshRSS/data
-
extensions:/var/www/FreshRSS/extensions
environment:
CRON_MIN:
'*/20'
FRESHRSS_ENV: development
FRESHRSS_INSTALL: |-
--api_enabled
--base_url https://rss.example.net
--db-base freshrss
--db-host freshrss-db
--db-password
freshrss
--db-type pgsql
--db-user freshrss
--default_user admin
--language en
FRESHRSS_USER: |-
--api_password freshrss
--email user@example.net
--language en
--password freshrss
--user admin
TZ:
Europe/Paris
```
* Minor type f in find
* shellcheck
When an extension defines an `autoload` method, it will be registered
automatically before enabling the extension.
For the extension creator, it's easier because there is no need to
register it manually.
* add two new hooks
I develop a new extension and i need 2 new hooks for it
* update EN documentation
* Correct typing errors
* Update app/views/helpers/javascript_vars.phtml
Co-authored-by: Alexandre Alapetite <alexandre@alapetite.fr>
Before, there were some guidelines on how to use the extension class and how to extend it. Those guidelines were defined as comments.
Now, those guidelines are enforced by the code itself. There is no need for those comments anymore.
Extensions must override some parent methods. The rule is just a
guideline because only described in comments. The idea is to
deprecate that flexibility and add it to the code. To warn users
beforehand, we log error message regarding incomplete extensions.
See #3333
Before, when clicking on the author link, the search was done on the
main stream in the normal view. It's fine until the feed is not visible
in the main stream.
Now, the current context is used along with the search.
See #3314
* Avoid manual intialisations of system or user configuration
More consistent use of Context
* Simplify FreshRSS_Context::initUser
* Remove a few manual get_user_configuration
* A bit of debugging
* Fix context user init
* Fix install
* Fix concurrency
Concurrent requests could lead to bad race condition
* Fix actualize cron
Fix case when system i initialised several times
Before, the temp path was not check during install. With some configuration,
FRSS was not working because of a non-writable temp directory. It happened
with XAMPP on MacOS X but it might be the case for other platforms.
Now, the temp path is checked during install to make sure it is writable.
See #3310
As FRSS always comes with Minz, there is no need to validate that
it is there during install. If it turns out we need to validate that,
we might need to validate other libraries as well.
Before, we had 5 classes in the ModelPdo file. It was bad for 2 reasons.
The first reason is that it is considered bad practice to have multiple
class in one file. This is especially true when using autoloading. On top
of that it is less readable considering the size of the file. The second
reason is that so far we were lucky. Everytime we needed to access the
database, it was through the ModelPdo class which loads all the other
classes. If we want to access directly the connection, it wont be loaded.
On top of that, the system is configured to work on a single database,
but as we have every connection definition in a single file, all classes
were loaded at the same time. Thus using memory and processing time for
nothing.
Now, we have a file for each class. To work with autoloading, classes
were slightly renamed to match autoloading rules.
Before, we had 5 classes in the ModelPdo file. It was bad for 2 reasons.
The first reason is that it is considered bad practice to have multiple
class in one file. This is especially true when using autoloading. On top
of that it is less readable considering the size of the file. The second
reason is that so far we were lucky. Everytime we needed to access the
database, it was through the ModelPdo class which loads all the other
classes. If we want to access directly the connection, it wont be loaded.
On top of that, the system is configured to work on a single database,
but as we have every connection definition in a single file, all classes
were loaded at the same time. Thus using memory and processing time for
nothing.
Now, we have a file for each class. To work with autoloading, classes
were slightly renamed to match autoloading rules.
* Extract autoloading process
The process sits in its own file now to ease future improvements.
* Change the autoload process
Before, the autoload process was too restricted. It was really dependant on our
code tree. It was hard to add more classes to be loaded automatically. On top
of that, it did not support autoloading classes following the PSR-4 recommendation.
Now, the autoload process is more open. It supports partially the PSR-4 recommendation,
there is no specific code to load Minz classes or PHPMailer classes. This is the
starting point to reorganize the codebase to introduce long waiting changes as seen
in #789. It would be a nice to later rework the tree, rename classes, and add namespace
in a fashion that follows the PSR-4. Then specific FRSS workarounds in the autoload
could be dropped.
