Three indexer form fields save and reload but are never read: Enable RSS, Minimum Age, Maximum Size
メンテナーはふだん 1 日以内に返信
まだ誰も着手していません。
評価
- 難易度
- 5/5
- 見積もり時間
- 1週間以上
- 初心者へのやさしさ
- 35/100
調査の方向性
Start with listenarr.domain/Search/Indexer.cs, the form in fe/src/components/settings/IndexerFormModal.vue, and the round-trip in listenarr.api/Features/Indexers/IndexersController.cs; then inspect listenarr.application/Search/Scoring/SearchResultScorer.cs and the cited search/grab paths. Decide separately whether to implement Minimum Age and Maximum Size and where their checks belong, and whether Enable RSS should be implemented or its control and badge removed. Done means the chosen behavior is enforced or the unused controls are gone, with relevant tests updated or added.
索引モデルが issue の本文から書いたものです。
説明
Checked against canary d25b3e11 (1.3.3). Every line number below is from that commit.
Summary
Three fields on the indexer edit form save and reload correctly and are not read anywhere in the backend: Enable RSS, Minimum Age (minutes), and Maximum Size (MB).
Enable RSS is the one I would look at first, because it also drives a badge. Turning it on puts a green RSS pill on the indexer's card in the Indexers list, in a row labelled Features, next to two pills that do correspond to working behaviour. There is no RSS mechanism anywhere in the backend for it to describe.
Six other fields on the same class are consumed, including the two declared immediately after EnableRss in the same file, which are read in a single line of the repository. So the problem is these three fields specifically, and not a layer that ignores its configuration generally.
I have not run any of this. The citations are the evidence.
Enable RSS
Declared at listenarr.domain/Search/Indexer.cs:69-71:
/// <summary>
/// Whether to enable RSS sync
/// </summary>
public bool EnableRss { get; set; } = true;
Rendered as a CheckboxCard at fe/src/components/settings/IndexerFormModal.vue:208-213, titled "Enable RSS" and described as "Use RSS feeds to monitor for new releases". Defaults to true at :348. Prowlarr-imported indexers get EnableRss = true at listenarr.api/Features/Prowlarr/ProwlarrIndexerImportWorkflow.cs:200.
The badge is at fe/src/views/settings/IndexersTab.vue:130:
<div class="feature-badges">
<Pill variant="success" v-if="indexer.enableRss">RSS</Pill>
<Pill variant="primary" v-if="indexer.enableAutomaticSearch">Automatic Search</Pill>
<Pill variant="info" v-if="indexer.enableInteractiveSearch">Interactive Search</Pill>
</div>
The row is labelled "Features:" (:128), and the badges sit at :129-135. The other two badges in it are backed by real behaviour, at listenarr.infrastructure/Persistence/Repositories/EfIndexerRepository.cs:50:
.Where(i => i.IsEnabled && (isAutomaticSearch ? i.EnableAutomaticSearch : i.EnableInteractiveSearch))
I want to be precise about what the badge does and does not claim, because I think the distinction matters for what you do about it. It is a feature badge and not a live status indicator. Pill's own component documentation says so explicitly ("Different from StatusBadge which is for status indicators"), and the card has separate controls for enabled state and last test result elsewhere. So it does not say RSS sync is running right now. It says this indexer has the RSS feature switched on, in a list where the two neighbouring entries mean exactly that and are true.
There is no RSS mechanism to be switched on. Searching the backend for rss case-insensitively returns the EnableRss declaration, its doc comment, the controller assignment, and the Prowlarr import default. Everything else that matches does so incidentally: IndexersSearch, duplicatedAdaptersString, _rootFoldersSync. No hosted service, no background job, no scheduled task, no feed parser. git ls-files | grep -i rss returns two files, both RootFolders.
Nor is it something that was lost. git log -S "EnableRss" over the four backend projects with migrations excluded returns three commits: the one that introduced the property with its controller round-trip (48636573), the one that added the Prowlarr default (8353e80a), and one that moved files between projects (f5e6ad7c). No commit has ever added or removed a consumer. This was built as a control ahead of a feature that has not been started.
I also could not find an issue or PR tracking RSS support, open or closed, so I am filing on the assumption it is not already queued somewhere. If it is, say so and I will drop this part.
Minimum Age
Indexer.cs:99-101:
/// <summary>
/// Minimum age in minutes before NZBs are grabbed
/// </summary>
public int MinimumAge { get; set; } = 0;
Input at IndexerFormModal.vue:245-252, labelled "Minimum Age (minutes)" with help "Wait time before grabbing new releases (0 = disabled)", hidden for the Internet Archive implementation.
The only backend reference is the round-trip assignment at listenarr.api/Features/Indexers/IndexersController.cs:294. Nothing in the scorer, nothing in any grab-submission path. git log -S shows the same two-commit history as EnableRss: introduced with its round trip, moved once, never consumed.
This is the usenet anti-race-condition delay that Sonarr and Radarr both ship, where a very fresh NZB is left alone for a while so the poster can finish and the repair blocks can propagate. Setting it here has no effect, and the effect it would have is one people specifically go looking for after a run of failed unpacks.
The natural comparison is Indexer.Retention, declared five lines below it at :106 and read at listenarr.application/Search/Scoring/SearchResultScorer.cs:128. Retention and Minimum Age are the two ends of the same usenet window. One end is enforced.
Maximum Size
Indexer.cs:108-111:
/// <summary>
/// Maximum size in MB (0 = unlimited)
/// </summary>
public int MaximumSize { get; set; } = 0;
Input at IndexerFormModal.vue:265-271, labelled "Maximum Size (MB)" with help "Maximum allowed size in megabytes (0 = unlimited)". Round-tripped at IndexersController.cs:296, read nowhere.
This one is the easiest to miss, because QualityProfile.MaximumSize has the same name and does work, at SearchResultScorer.cs:101-105:
if (profile.MaximumSize > 0 && searchResult.Size > profile.MaximumSize * 1024 * 1024)
{
score.RejectionReasons.Add($"File too large (> {profile.MaximumSize} MB)");
score.TotalScore = -1;
return score;
}
A grep for MaximumSize finds that line and looks like a hit. git log -S "indexer.MaximumSize" over the whole history returns only the commit that introduced the round trip and the one that moved the file, so the indexer-level field has never had a reader of its own.
Control
Six fields on the same entity are consumed:
| Field | Read at |
|---|---|
IsEnabled |
EfIndexerRepository.cs:50 |
EnableAutomaticSearch |
EfIndexerRepository.cs:50 |
EnableInteractiveSearch |
EfIndexerRepository.cs:50 |
Retention |
SearchResultScorer.cs:128 |
Categories |
TorznabNewznabRequestBuilder.cs:55-57 |
AdditionalSettings |
42 lines across 16 backend files, including IndexerSearchWorkflow.cs:118,191 and MyAnonamouseSearchProvider.cs:54 |
EnableRss is declared at :71, between Tags at :66 and EnableAutomaticSearch at :76. MinimumAge is at :101 and Retention at :106. Every one of these travels through the same controller block at IndexersController.cs:285-297 and comes back out of the same repository. Some are read afterwards and three are not.
I have deliberately left Indexer.Tags out of this. It has no reader either, but it also has no input on the form, so nobody can set it and then be surprised.
Reproducing
Run at the repo root on d25b3e11:
# the three fields across the backend: one line each, the controller round trip
git grep -n "EnableRss\|MinimumAge" -- \
'listenarr.api/**/*.cs' 'listenarr.application/**/*.cs' \
'listenarr.domain/**/*.cs' 'listenarr.infrastructure/**/*.cs' | grep -v '/Migrations/'
git grep -n "indexer.MaximumSize\|existing.MaximumSize" -- 'listenarr.api/**/*.cs'
# the same query for six siblings on the same entity
git grep -n "EnableAutomaticSearch\|EnableInteractiveSearch\|\.Retention\|indexer.Categories\|AdditionalSettings" -- \
'listenarr.api/**/*.cs' 'listenarr.application/**/*.cs' \
'listenarr.domain/**/*.cs' 'listenarr.infrastructure/**/*.cs' | grep -v '/Migrations/'
# any RSS mechanism at all
git grep -in "rss" -- 'listenarr.api/**/*.cs' 'listenarr.application/**/*.cs' \
'listenarr.domain/**/*.cs' 'listenarr.infrastructure/**/*.cs' | grep -v '/Migrations/'
git ls-files | grep -i rss
# no commit has ever added or removed a consumer for any of the three
git log -S "EnableRss" --oneline canary -- \
'listenarr.domain' 'listenarr.application' 'listenarr.infrastructure' 'listenarr.api' ':!**/Migrations/**'
git log -S "indexer.MaximumSize" --oneline canary
Use git grep rather than grep -rn: on a built tree grep -rn also walks bin/ and obj/, where generated XML doc files carry the same symbols. Exclude fe/eslint-report.json from any frontend grep, since it is checked in and repeats every symbol. git log -S needs Migrations excluded or the regenerated EF model snapshot buries the answer.
Options
These three are not the same shape of problem and I do not think one answer covers them.
Minimum Age is a small piece of real work with an obvious home. SearchResultScorer already loads the indexer to read Retention at :121-128, so a minimum-age rejection could sit next to the existing age checks at :150-176 without a new lookup. Sonarr does this at grab time rather than at score time, so there is a design question about where it belongs, but the plumbing is already in place either way.
Maximum Size is the same, and smaller still: one condition alongside the profile-level check at :101-105. The question there is whether a per-indexer size cap is wanted at all given the profile-level one works, and if not, removing the input is a fair answer.
Enable RSS is the one where I think removal deserves serious consideration rather than being the fallback. Building RSS sync to justify a checkbox would be a large feature driven by a UI control, which is backwards. Taking the checkbox and the badge out costs almost nothing and stops the Features row from listing something the application does not do. If RSS sync is on the roadmap then the control is a placeholder rather than a mistake and the right move is probably to hide it until then.
I have not opened a PR for any of this. The two wiring jobs need a design decision about where the check belongs, and removing controls from your settings pages is not a call I should be making.
- 主要言語
- C#
- スター
- 901
- フォーク
- 70
- 平均マージ
- 16時間 27分
- マージ済み PR(30日)
- 60
環境構築
- Dockerfile または Docker Compose ファイルあり
- プルリクエストのテンプレートあり
- コントリビューションガイドを読む
はじめの一歩
- issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
- 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
- リポジトリをフォークし、ブランチを切って変更します。
- issue 番号を参照したプルリクエストを送ります。
Listenarrs/Listenarr のほかの issue
-
Search button on a book does not retry without the author when the full query finds nothing対応中かも このイシューにリンクされたプルリクエストがオープン中、またはマージ済みです。 オープン
難易度 2/5 1〜3時間 初心者へのやさしさ 75/100
Listenarrs/Listenarr#1173 ·
メンテナーはふだん 1 日以内に返信
-
難易度 2/5 1〜3時間 初心者へのやさしさ 82/100
Listenarrs/Listenarr#1133 ·
メンテナーはふだん 1 日以内に返信
-
難易度 2/5 1〜3時間 初心者へのやさしさ 76/100
Listenarrs/Listenarr#1128 ·
メンテナーはふだん 1 日以内に返信
-
Add New's search splits a title that contains "Series:", "Author:"… as if they were search keywordsオープン
難易度 2/5 1〜3時間 初心者へのやさしさ 88/100
Listenarrs/Listenarr#1127 ·
メンテナーはふだん 1 日以内に返信
-
p3
難易度 2/5 1〜3時間 初心者へのやさしさ 77/100
Listenarrs/Listenarr#971 ·
メンテナーはふだん 1 日以内に返信
Listenarrs/Listenarr の issue をすべて見る
似ている issue
-
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
conductor-oss/csharp-sdk#183 ·
-
CreateNewLangProj throws when the default analysis writing system's ICU locale differs from its tagオープン
難易度 2/5 1〜3時間 初心者へのやさしさ 76/100
-
kind/bug
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
unoplatform/uno#26555 ·
メンテナーはふだん 1 日以内に返信
-
難易度 2/5 1〜3時間 初心者へのやさしさ 77/100
CactuseSecurity/firewall-orchestrator#5739 ·
メンテナーはふだん 1 日以内に返信
-
Bug
難易度 2/5 1〜3時間 初心者へのやさしさ 65/100
メンテナーはふだん 1 日以内に返信