* chore: update copyright year in license headers
* Revert "chore: update copyright year in license headers"
This reverts commit 3e58129c431b9a491089ce36b908f9bb6ba38ed3.
* chore: update copyright year in license headers
* fix: sort go imports
* fix: add missing license headers
* feat(filters): skip duplicates
* fix: add interface instead of any
* fix(filters): tonullint
* feat(filters): skip dupes check month day
* chore: cleanup
* feat(db): set autoincrement id
* feat(filters): add repack and proper to dupe profile
* feat(filters): add default dupe profiles
* feat(duplicates): check audio and website
* feat(duplicates): update tests
* feat(duplicates): add toggles on addform
* feat(duplicates): fix sqlite upgrade path and initialize duplicate profiles
* feat(duplicates): simplify sqlite upgrade
avoiding temp table and unwieldy select. Besides, FK constraints
are turned off anyway in #229.
* feat(duplicates): change CheckIsDuplicateRelease treatment of PROPER and REPACK
"Proper" and "Repack" are not parallel to the other conditions like "Title",
so they do not belong as dedup conditions. "PROPER" means there was an issue in
the previous release, and so a PROPER is never a duplicate, even if it replaces
another PROPER. Similarly, "REPACK" means there was an issue in the previous
release by that group, and so it is a duplicate only if we previously took a
release from a DIFFERENT group.
I have not removed Proper and Repack from the UI or the schema yet.
* feat(duplicates): update postgres schema to match sqlite
* feat(duplicates): fix web build errors
* feat(duplicates): fix postgres errors
* feat(filters): do leftjoin for duplicate profile
* fix(filters): partial update dupe profile
* go fmt `internal/domain/filter.go`
* feat(duplicates): restore straightforward logic for proper/repack
* feat(duplicates): remove mostly duplicate TV duplicate profiles
Having one profile seems the cleanest. If somebody wants multiple
resolutions then they can add Resolution to the duplicate profile.
Tested this profile with both weekly episodic releases and daily
show releases.
* feat(release): add db indexes and sub_title
* feat(release): add IsDuplicate tests
* feat(release): update action handler
* feat(release): add more tests for skip duplicates
* feat(duplicates): check audio
* feat(duplicates): add more tests
* feat(duplicates): match edition cut and more
* fix(duplicates): tests
* fix(duplicates): missing imports
* fix(duplicates): tests
* feat(duplicates): handle sub_title edition and language in ui
* fix(duplicates): tests
* feat(duplicates): check name against normalized hash
* fix(duplicates): tests
* chore: update .gitignore to ignore .pnpm-store
* fix: tests
* fix(filters): tests
* fix: bad conflict merge
* fix: update release type in test
* fix: use vendored hot-toast
* fix: release_test.go
* fix: rss_test.go
* feat(duplicates): improve title hashing for unique check
* feat(duplicates): further improve title hashing for unique check with lang
* feat(duplicates): fix tests
* feat(duplicates): add macros IsDuplicate and DuplicateProfile ID and name
* feat(duplicates): add normalized hash match option
* fix: headlessui-state prop warning
* fix(duplicates): add missing year in daily ep normalize
* fix(duplicates): check rejections len
---------
Co-authored-by: ze0s <ze0s@riseup.net>
* fix(filters): RED and OPS lossless parsing and filtering
* fix(filters): logscore and EP parsing
* fix(filters): tests
* fix(filters): tests
* feat(definitions): RED parse title variable
* feat(indexers): setup indexer to filter tests
* feat(indexers): tests and improve parsing
* feat(indexers): improve tests
* fix(http): flip to a shared transport and clients
* nice threads
* that is terrible
* fake uri for magnet
* lazy locking
* why bother with r's
* flip magic params to struct
* refactor(http-clients): use separate clients with shared transport
* refactor(http-clients): add missing license header
* refactor(http-clients): defer and fix errors
---------
Co-authored-by: ze0s <ze0s@riseup.net>
* feat(feeds): add force run
* fix: simplify ForceRun
* add confirmation modal
* handle errors by using the test func
* require user input to run
* make sure to reschedule next job after forcerun
* refactor modal centering with grid
* refactor: Simplify startJob and forceRun logic
- Refactor `startJob` to accept a `runImmediately` flag. This flag controls whether the job should be run immediately or scheduled for later. This change simplifies the `ForceRun` function by allowing it to call `startJob` with `runImmediately` set to `true`.
- Remove redundant checks in `ForceRun` related to feed type. These checks are handled in `startJob`.
BREAKING CHANGE: The `startJob` function now requires a second argument, `runImmediately`. This change affects all calls to `startJob`.
* fix(web) Invalidate queries after forceRun
* refactor(feeds): init and test run
---------
Co-authored-by: ze0s <43699394+zze0s@users.noreply.github.com>
* feat(feeds): improve caching
* fix(feeds): put cache if not empty
* fix(feeds): reassign loop var
* fix(feeds): enable busy_timeout again
* fix(feeds): enable busy_timeout again
* In rss feeds, sometimes the only place the size is mentioned is in the
description field. If the size has not already been determined from
another source try to read it from there.
* Accept sizes with no space between value and unit of measure
* feat(feeds): get size from description add test
* fix(feeds): tests
---------
Co-authored-by: ze0s <ze0s@riseup.net>
* feat(releases): support magnet links
* feat(feeds): support magnet links
* feat(actions): log messages
* fix: component warning
* fix: check hasprefix instead of hassuffix for magnet
* feat(release): resolve magnet uri from link
* fix(actions): deluge use magnet uri
* fix(macros): add `MagnetURI` var
* fix(actions): run magnet resolving before macros
* feat(feeds): set download type on creation
* Fixes issue 636 - fail to parse torrent size on some RSS feeds when enclosure is a fixed value
* Pick the biggest size
* minor comment
* Updates release size when new parsed string size is bigger
- Sets default size of `Release` to zero (unparsed)
- When there are parsing errors, then it doesn't RESET the size to 0 (assuming a tracker providing some size that can't be parsed)
- Updates the size to the new size only if the parsed size is bigger than the previous.
* fix(feeds): ignore default 39399 enclosure size
---------
Co-authored-by: ze0s <43699394+zze0s@users.noreply.github.com>
* feat(filters): add support for language
* feat(filters): add db migrations and repo
* feat(filters): fix failing tests
* feat(filters): fix failing tests
Feed dates are not always initialized, which defaults (if present) to epoch on some feed items. When this happens the results can be unpredictable, and not expected for what users are intending for. As such to have sane defaults, we ignore the date if it's before April 1st 1970, because nothing really happened before this date.
* feat(rss): implement relative URL support
* bp had some words
* feat(gh): uplift golang to 1.19.1
* feat(docker): uplift golang to 1.19
* feat(docker): uplift to go 1.19
* cleanup
* chore: update go version in go.mod
* feat: refactor and add test feed process item
* unescape query params after join
* break out logic into separate method
* add tests
* fix(torznab): use time for interval.
* flip Schedule to be duration based
* typefix
* Update service.go
* clear err
* pull err package
* casting to the moon
* enforce a minimum of 15m
* lower minimum based on feedback from Unit3d
