Compare commits

..

3 Commits

Author SHA1 Message Date
Gauvino
d2a0fd875a fix(detect-duplicate): sanitize reposted issue titles
Security audit: the bot echoes other issues' titles back into a comment, so a
maliciously-named issue could ping (@everyone) or inject markdown/HTML. Break
@-mentions with a zero-width space and strip markdown/HTML control chars before
posting.
2026-06-01 20:32:39 +02:00
Gauvino
d2c48de60d ci(issues): flag likely-duplicate issues on open
Adds .github/workflows/detect-duplicate.yml + scripts/detect-duplicate-issue.mjs
(Bun, dep-free, no API key): on a new issue, compares its title/body to open
issues via Jaccard similarity (with light stemming and stop-words), and if the
top matches pass a threshold, posts one comment listing them and adds a
'possible duplicate' label. Inspired by seerr's detect-duplicate, minus the
embedding/Groq dependency.
2026-06-01 17:38:19 +02:00
lance chant
54ee507209 fix: fixing the time variable (#1638)
Signed-off-by: Lance Chant <13349722+lancechant@users.noreply.github.com>
2026-06-01 15:22:39 +02:00
6 changed files with 257 additions and 24 deletions

38
.github/workflows/detect-duplicate.yml vendored Normal file
View File

@@ -0,0 +1,38 @@
name: 🔁 Detect Duplicate Issues
on:
issues:
types: [opened]
permissions:
contents: read
concurrency:
group: detect-duplicate-${{ github.event.issue.number }}
cancel-in-progress: true
jobs:
detect:
name: 🔍 Find similar issues
if: github.actor != 'github-actions[bot]'
runs-on: ubuntu-24.04
permissions:
issues: write
contents: read
steps:
- name: 📥 Checkout repository
uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
- name: 🍞 Setup Bun
uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2.2.0
with:
bun-version: latest
- name: 🔍 Detect duplicate issues
run: bun scripts/detect-duplicate-issue.mjs
env:
GH_TOKEN: ${{ github.token }}
GITHUB_REPOSITORY: ${{ github.repository }}
ISSUE_NUMBER: ${{ github.event.issue.number }}
ISSUE_TITLE: ${{ github.event.issue.title }}
ISSUE_BODY: ${{ github.event.issue.body }}

View File

@@ -401,6 +401,10 @@ export const TVJellyseerrSearchResults: React.FC<
}) => {
const { t } = useTranslation();
const hasMovies = movieResults && movieResults.length > 0;
const hasTv = tvResults && tvResults.length > 0;
const hasPersons = personResults && personResults.length > 0;
if (loading) {
return null;
}
@@ -427,26 +431,22 @@ export const TVJellyseerrSearchResults: React.FC<
return (
<View>
{/* No section requests `hasTVPreferredFocus`: the native search field
keeps focus while typing, otherwise the first result would re-grab
focus on every keystroke as results re-render. The user navigates
down to the grid manually. */}
<TVJellyseerrMovieSection
title={t("search.request_movies")}
items={movieResults}
isFirstSection={false}
isFirstSection={hasMovies}
onItemPress={onMoviePress}
/>
<TVJellyseerrTvSection
title={t("search.request_series")}
items={tvResults}
isFirstSection={false}
isFirstSection={!hasMovies && hasTv}
onItemPress={onTvPress}
/>
<TVJellyseerrPersonSection
title={t("search.actors")}
items={personResults}
isFirstSection={false}
isFirstSection={!hasMovies && !hasTv && hasPersons}
onItemPress={onPersonPress}
/>
</View>

View File

@@ -235,13 +235,10 @@ export const TVSearchPage: React.FC<TVSearchPageProps> = ({
module). It renders the native search bar + grid keyboard and
forwards typed text into the existing query pipeline via setSearch;
our own results grid renders below. */}
{/* No horizontal margin here: the native tvOS search bar centers itself
and renders a trailing "Hold to Dictate in <Language>" hint. Extra
margins squeeze the bar's width and clip that trailing hint, so let
the native view span the full width and own its own insets. */}
<View
style={{
marginBottom: 24,
marginHorizontal: HORIZONTAL_PADDING,
height: SEARCH_AREA_HEIGHT,
}}
>
@@ -283,17 +280,13 @@ export const TVSearchPage: React.FC<TVSearchPageProps> = ({
{/* Library Search Results */}
{isLibraryMode && !loading && (
<View style={{ gap: SECTION_GAP }}>
{sections.map((section) => (
{sections.map((section, index) => (
<TVSearchSection
key={section.key}
title={section.title}
items={section.items!}
orientation={section.orientation || "vertical"}
// Never auto-focus a result. The native search field owns focus
// while typing; `hasTVPreferredFocus` here would re-grab focus on
// every keystroke as results re-render. User navigates down to the
// grid manually.
isFirstSection={false}
isFirstSection={index === 0}
onItemPress={onItemPress}
onItemLongPress={onItemLongPress}
imageUrlGetter={

View File

@@ -297,12 +297,12 @@ export const TVSearchSection: React.FC<TVSearchSectionProps> = ({
removeClippedSubviews={false}
getItemLayout={getItemLayout}
style={{ overflow: "visible" }}
// Edge padding via contentContainerStyle, NOT contentInset+contentOffset.
// contentOffset only applies on initial mount; since this FlatList is
// reused across searches (stable key), a second search left the inset
// without the offset and the grid snapped flush to the left edge.
contentInset={{
left: edgePadding,
right: edgePadding,
}}
contentOffset={{ x: -edgePadding, y: 0 }}
contentContainerStyle={{
paddingHorizontal: edgePadding,
paddingVertical: SCALE_PADDING,
}}
/>

View File

@@ -1254,7 +1254,7 @@ export const Controls: FC<Props> = ({
<Text
style={[styles.endsAtText, { fontSize: typography.callout }]}
>
{t("player.ends_at")} {getFinishTime()}
{t("player.ends_at", { time: getFinishTime() })}
</Text>
</View>
)}
@@ -1448,7 +1448,7 @@ export const Controls: FC<Props> = ({
<Text
style={[styles.endsAtText, { fontSize: typography.callout }]}
>
{t("player.ends_at")} {getFinishTime()}
{t("player.ends_at", { time: getFinishTime() })}
</Text>
</View>
)}

View File

@@ -0,0 +1,202 @@
#!/usr/bin/env bun
/**
* Flags likely-duplicate issues when a new issue is opened, using lexical similarity
* (Jaccard over word sets of the title and body) — no API key, no embeddings.
*
* On a match it posts ONE comment listing the closest open issues and adds the
* "possible duplicate" label. If nothing is similar enough, it does nothing.
*
* Env:
* GITHUB_REPOSITORY owner/repo
* ISSUE_NUMBER the new issue number
* ISSUE_TITLE the new issue title
* ISSUE_BODY the new issue body
* GH_TOKEN/GITHUB_TOKEN for gh (provided in CI)
* DUP_THRESHOLD similarity threshold 0..1 (default 0.3)
* DUP_MAX max matches to report (default 5)
* DUP_FIXTURE optional path to a JSON array of {number,title,body} (local testing)
* DRY_RUN if set, print results instead of commenting/labelling
*/
import { execFileSync } from "node:child_process";
import { readFileSync } from "node:fs";
const REPO = process.env.GITHUB_REPOSITORY || "streamyfin/streamyfin";
const NUMBER = Number(process.env.ISSUE_NUMBER);
const TITLE = process.env.ISSUE_TITLE || "";
const BODY = process.env.ISSUE_BODY || "";
const THRESHOLD = Number(process.env.DUP_THRESHOLD) || 0.3;
const MAX = Number(process.env.DUP_MAX) || 5;
const DRY = !!process.env.DRY_RUN;
const LABEL = "possible duplicate";
// Generic stop words only — keep domain/feature/platform words (android, downloads,
// subtitles…) since those are exactly what makes two reports the same or different.
const STOP = new Set(
(
"a an the and or but if then of to in on at by for with from as is are was were be been being do does did " +
"it its this that these those i you we they me my your our their he she him her " +
"when while where what which who how why so just then than too very can could would should will " +
"not no nor only own same s t don dont im ive please thanks hi hello also still get got use used using " +
"app application streamyfin issue bug"
).split(/\s+/),
);
const stem = (w) => w.replace(/(ing|ed|es|s)$/, "");
const tokens = (s) =>
(s || "")
.toLowerCase()
.replace(/```[\s\S]*?```/g, " ") // drop code blocks
.replace(/<!--[\s\S]*?-->/g, " ") // drop html comments
.replace(/https?:\/\/\S+/g, " ") // drop urls
.replace(/[^a-z0-9\s]/g, " ")
.split(/\s+/)
.filter((w) => w.length > 2 && !STOP.has(w))
.map(stem)
.filter((w) => w.length > 2);
const jaccard = (a, b) => {
const A = new Set(a);
const B = new Set(b);
if (!A.size || !B.size) return 0;
let inter = 0;
for (const x of A) if (B.has(x)) inter++;
return inter / (A.size + B.size - inter);
};
const newTitle = tokens(TITLE);
const newBody = tokens(BODY);
const score = (o) =>
0.6 * jaccard(newTitle, tokens(o.title)) +
0.4 * jaccard(newBody, tokens(o.body));
// fetch open issues (excluding PRs and the new issue itself)
let issues;
if (process.env.DUP_FIXTURE) {
issues = JSON.parse(readFileSync(process.env.DUP_FIXTURE, "utf8"));
} else {
const raw = execFileSync(
"gh",
[
"api",
`repos/${REPO}/issues`,
"--paginate",
"-X",
"GET",
"-f",
"state=open",
"-f",
"per_page=100",
"--jq",
".[] | select(.pull_request | not) | {number, title, body}",
],
{ encoding: "utf8", maxBuffer: 1e8 },
);
issues = raw
.split("\n")
.filter(Boolean)
.map((l) => JSON.parse(l));
}
const matches = issues
.filter((o) => o.number !== NUMBER)
.map((o) => ({ ...o, s: score(o) }))
.filter((o) => o.s >= THRESHOLD)
.sort((a, b) => b.s - a.s)
.slice(0, MAX);
if (!matches.length) {
console.log("No likely duplicates found.");
process.exit(0);
}
// Neutralise other issues' titles before echoing them back: break @mentions and
// strip markdown/HTML control chars so a maliciously-named issue can't ping people
// or inject formatting into our comment. GitHub linkifies "#123" on its own.
const safeTitle = (t) =>
(t || "")
.replace(/@/g, "@")
.replace(/[`<>|*_~[\]]/g, " ")
.replace(/\s+/g, " ")
.trim()
.slice(0, 140);
const list = matches
.map(
(m) =>
`- #${m.number}${safeTitle(m.title)} (≈ ${Math.round(m.s * 100)}% similar)`,
)
.join("\n");
const comment = [
"<!-- duplicate-detector -->",
"🔍 **This looks like it might be a duplicate.** Possibly related open issues:",
"",
list,
"",
"If yours is different, ignore this — a maintainer will confirm. Otherwise, please 👍 the existing issue and add any extra details there.",
].join("\n");
console.log(`Found ${matches.length} possible duplicate(s):\n${list}`);
if (DRY) {
console.log("\nDRY_RUN: not commenting/labelling.");
process.exit(0);
}
execFileSync(
"gh",
[
"api",
"-X",
"POST",
`repos/${REPO}/issues/${NUMBER}/comments`,
"-f",
`body=${comment}`,
],
{ stdio: "ignore" },
);
try {
execFileSync(
"gh",
[
"api",
"-X",
"POST",
`repos/${REPO}/issues/${NUMBER}/labels`,
"-f",
`labels[]=${LABEL}`,
],
{ stdio: "ignore" },
);
} catch {
// label may not exist yet — create then add
execFileSync(
"gh",
[
"api",
"-X",
"POST",
`repos/${REPO}/labels`,
"-f",
`name=${LABEL}`,
"-f",
"color=fbca04",
"-f",
"description=Automatically flagged as a possible duplicate",
],
{ stdio: "ignore" },
);
execFileSync(
"gh",
[
"api",
"-X",
"POST",
`repos/${REPO}/issues/${NUMBER}/labels`,
"-f",
`labels[]=${LABEL}`,
],
{ stdio: "ignore" },
);
}
console.log("Commented and labelled.");