Dev/test ci #1
Open
Dev/test ci #1
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Until now, to change the log-level for a subcommand, the following was to be executed: recitale --log-level DEBUG build This patch moves the parameter to subcommands so that the following can be executed instead: recitale build --log-level DEBUG which feels more natural. Signed-off-by: Quentin Schulz <[email protected]>
width and height variables are strings when split from the output of subprocess.check_output. However, in ratio property, width is divided by height which only works for numbers which isn't the case unless they are casted to a number before being put in self.size tuple. Signed-off-by: Quentin Schulz <[email protected]>
BaseVideo.size is only set by ratio property since the reencode() function which calls ffprobe sets the size attribute only to the ReencodedVideo and not the BaseVideo. Thereofre, when BaseVideo.ratio is called, the size attribute does not exist yet and will raise an AttributeError. So instead of checking if the size attribute is set, let's check if the object has the size attribute. Signed-off-by: Quentin Schulz <[email protected]>
This prepares for getting more info about a video in a later commit in one go instead of calling ffprobe multiple times. Signed-off-by: Quentin Schulz <[email protected]>
To prepare for getting more infos from ffprobe and make it easier to parse, let's just switch to json instead of doing manual parsing of csv. Signed-off-by: Quentin Schulz <[email protected]>
This allows themes and rendering functions to access the duration of a video. This will be useful to display during rendering how much of a video has already been reencoded. Signed-off-by: Quentin Schulz <[email protected]>
This allows to display a progress bar displaying the amount of the video that has already been reencoded making the wait a bit more user-friendly. Signed-off-by: Quentin Schulz <[email protected]>
This allows themes and rendering functions to access the duration of an audio file. This will be useful to display during rendering how much of an audio file has already been reencoded. Signed-off-by: Quentin Schulz <[email protected]>
This allows to display a progress bar displaying the amount of the audio that has already been reencoded making the wait a bit more user-friendly. Signed-off-by: Quentin Schulz <[email protected]>
This adds untested line numbers to the report so it's easier to know what kind of tests should be written next. Signed-off-by: Quentin Schulz <[email protected]>
== operator is different from the "is" operator. Since the factories should return the same object if the same parameters are given, "is" operator should be used and not the equal operator, which can return two different objects with the same content. Signed-off-by: Quentin Schulz <[email protected]>
This adds missing tests for BaseAudio so that 100% code coverage is reached. Signed-off-by: Quentin Schulz <[email protected]>
This adds missing unit tests for BaseImage to improve test coverage. Signed-off-by: Quentin Schulz <[email protected]>
This adds missing unit tests for BaseVideo to improve test coverage. Signed-off-by: Quentin Schulz <[email protected]>
…ryptodomex To limit the number of external, non-Python dependencies and subprocess calls, let's migrate the HTML page encryption from subprocess'ed OpenSSL encryption to pycryptodomex. However, the key and IV derivation implementation supported by CryptoJS (the client side decryption tool) is OpenSSL specific and therefore needed to be re-implemented in Python. Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
…g name There's a variable available to get the branch/tag name that can be used insteaf of using a regular expression for the GitHub refspec so let's use it. Signed-off-by: Quentin Schulz <[email protected]>
This adds a Dockerfile which fetches recitale pip package and make it available through a Debian-based container image. Signed-off-by: Quentin Schulz <[email protected]>
…ghcr.io This builds a container image based on the recitale PYPI package that was pushed in the pypi-package job, does a small check by building the example gallery and then pushes it to ghcr.io registry. Signed-off-by: Quentin Schulz <[email protected]>
pathlib is a Python standard library so no additional package is needed. Also, it's been reported for prosopopee that MacOS users cannot use the software because of some path.py function being broken so using pathlib instead seems like a good idea. See: Psycojoker/prosopopee#124 Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
Wild guess turned out to be wrong, let's fix the URL with a working one. Signed-off-by: Quentin Schulz <[email protected]>
For Jpeg, MPO and TIFF formats, dpi is of type IFDRational and is thus not serializable which crashes the cache saving. This casts the dpi tuple into a float tuple to make it serializable. Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
With the with_ extension being built-in since 2.9 and removed in 3.1, in order to be compatible with 3.1 and later, the min version needs to be 2.9. Signed-off-by: Quentin Schulz <[email protected]>
From Jinja2 docs: """ Changed in version 2.9. This extension is now built-in and no longer does anything. """ https://jinja.palletsprojects.com/en/2.9.x/extensions/#with-statement This was not an issue until 3.1 version was released, which removed this extension and thus fails recitale. c.f. https://jinja.palletsprojects.com/en/3.1.x/changes/#version-3-1-0 Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
flake apparently now errors out with this inline comment, so let's move the comment before the line to make flake happy. Signed-off-by: Quentin Schulz <[email protected]>
The video is actually slightly bigger and starts vertically a bit lower than the thumbnail, leaving a few pixels from the thumbnail above the video. There's no reason for this so let's fix that by having the video match the thumbnail. Signed-off-by: Quentin Schulz <[email protected]>
While the thumbnail is properly rendered, the video was smaller than the thumbnail, making it a bit funny looking. Let's have the video fill the thumbnail entirely instead. Signed-off-by: Quentin Schulz <[email protected]>
This allows to not provide a 'date' setting for autogenerated galleries. The 'date' will automatically be retrieved from the oldest file in the gallery, either from one of its EXIF date tags or from its mtime. Suggested-by: crypto512 <[email protected]> Signed-off-by: Quentin Schulz <[email protected]>
show_date is supported since commit dc51723 ("[enh] add show_date setting, to be able to order galleries, without showing the date") but it wasn't documented, so let's fix this oversight. Signed-off-by: Quentin Schulz <[email protected]>
ffmpeg_audio encoder settings are supported since commit 2f657c9 ("start audio support") so let's document those settings. Signed-off-by: Quentin Schulz <[email protected]>
It's ffmpeg and not ffmeg so let's fix the typo. While at it, remove a spurious whitespace. Signed-off-by: Quentin Schulz <[email protected]>
On systems where neither LC_TIME, LANGUAGE, LC_ALL, LC_CTYPE nor LANG
environment variables are set, babel.default_locale('LC_TIME') will return
an object of NoneType type which will then fail the babel.format_date()
later in the code path.
In order to account for those systems, default to the POSIX locale as is
supposed to be the case on POSIX-compliant systems[1].
[1] https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap07.html section 7.2
Reported-by: Sebastien Stormacq <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
…x.html for the video section
Typo in href of the link. Signed-off-by: Quentin Schulz <[email protected]>
It's 2023, time to use HTTPS now. Signed-off-by: Quentin Schulz <[email protected]>
This domain is for sale (and has been for a while since it was removed from README.md three years ago already). Signed-off-by: Quentin Schulz <[email protected]>
Google pulled the plug on Google+ more than 4 years ago, so we can remove those share links now. Signed-off-by: Quentin Schulz <[email protected]>
The second background URL was missing a separator between the gallery link and the thumbnail filename, let's add it to have a more proper path. Signed-off-by: Quentin Schulz <[email protected]>
0.18.0 has removed the safe_load function we use. Let's just force an older version instead until we fix it (or switch for something else). Signed-off-by: Quentin Schulz <[email protected]>
The caption is part of a div being 100% of the media to show (be it a video or a picture), therefore adding a right and left margin makes it only take the centered 75% of the media instead of the full width. The 11.5% are actually applied to the parent parent container. It's a mistake to duplicate it, so let's remove it. Signed-off-by: Quentin Schulz <[email protected]>
black 24.1a1 suggests changes that other versions do not. black used to have only pre-releases before 22.1.0 and we had a hard dependency on black being older than 19.10b0. Because of how pip installs dependencies[1], providing 19.10b0 as minimum version asked pip to install the latest version, even if it is a pre-release, therefore installing 24.1a1. Considering that black now uses "proper" releases, let's use the first release version as minimum dependency to avoid the changes suggested in 24.1a1. [1] https://pip.pypa.io/en/stable/cli/pip_install/#pre-release-versions Signed-off-by: Quentin Schulz <[email protected]>
Python v3.12 and later raise an exception for this line with the
following message:
"""
AttributeError: 'called_once_with' is not a valid assertion. Use a spec for the mock if 'called_once_with' is meant to be an attribute.. Did you mean: 'assert_called_once_with'?
"""
A bit of research returned commit 1d4d677d1c90 ("gh-100690: Raise an
AttributeError when the assert_ prefix is forgotten when using Mock
(#100691)") from CPython git history, which in turns points at
python/cpython#100690.
This GitHub issue mentions that called_once_with returns a Mock object
instead of doing a proper assertion. This test is basically a no-op.
Therefore, let's fix the test by doing a proper assertion.
Fixes: 94d64c8 ("test: image: improve code coverage by testing BaseImage")
Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
The caption is part of a div being 100% of the media to show (be it a video or a picture), therefore adding a right and left margin makes it only take the centered 75% of the media instead of the full width. The 11.5% are actually applied to the parent parent container. It's a mistake to duplicate it, so let's remove it. Signed-off-by: Quentin Schulz <[email protected]>
In case there's a subgallery with a video as cover, the link is improper as gallery.link provides the absolute path of the subgallery. Considering that all paths here are expected to be relative, this ends up being wrong. Instead, let's use the name of the subgallery instead of its link so that the path relative to the subgallery as returned by vid and video.thumbnail can simply be concatenated with the name to provide the path relative to the current gallery. Signed-off-by: Quentin Schulz <[email protected]>
TIME_FORMAT is used both for encoding and decoding a date. If a media file doesn't contain an exif field for the date, it'll be provided one from the mtime, itself encoded via TIME_FORMAT. However if a media file does contain an exif field for the date, then it'll be used without going through the encoding part. Therefore, the decoding should take into account that the seconds part of the date can be something else than 00. This fixes autogen for the oldest file, in a gallery without a predefined cover, with an exif date field where the seconds part was non-zero. Fixes: 5e525a5 ("autogen: autofill date gallery setting if missing") Signed-off-by: Quentin Schulz <[email protected]>
To cover potential regressions, let's make sure that non-zero numbers for seconds in exif fields are used to test autogen. This is a follow up for the previous commit fixing a non-zero number of seconds in exif fields when generating a gallery with no predefined cover. Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.