filter false opensubtitles subtitle candidates

This commit is contained in:
dirtydishes 2026-05-25 12:22:58 -04:00
parent 6008272d0a
commit 11ed364094
6 changed files with 465 additions and 9 deletions

View file

@ -84,6 +84,11 @@ final class DreamioWebViewController: UIViewController {
const postedSubtitleURLs = new Set();
const subtitleURLPattern = /https?:\/\/[^\s"'<>]+(?:\.srt|\.vtt|\.ass|\.ssa|\.sub|opensubtitles|subtitle)[^\s"'<>]*/ig;
const subtitleSignalPattern = /subtitle|subtitles|opensubtitles|vtt|srt|ass|ssa/i;
const subtitleExtensions = new Set(["srt", "vtt", "ass", "ssa", "sub"]);
const nonSubtitleExtensions = new Set([
"aac", "avi", "bmp", "css", "gif", "heic", "ico", "jpeg", "jpg", "js", "json",
"m4a", "m4v", "mkv", "mov", "mp3", "mp4", "mpeg", "mpg", "png", "svg", "ts", "webm", "webp"
]);
const subtitleObjectKeys = [
"attributes",
"files",
@ -125,14 +130,51 @@ final class DreamioWebViewController: UIViewController {
}
};
const isDirectSubtitleFileURL = (url) => {
try {
const parsed = new URL(url, window.location.href);
const extension = parsed.pathname.split(".").pop().toLowerCase();
return subtitleExtensions.has(extension)
|| Array.from(subtitleExtensions).some((ext) => parsed.href.toLowerCase().includes(`.${ext}?`) || parsed.href.toLowerCase().includes(`.${ext}&`));
} catch (_) {
return false;
}
};
const isProbablyNonSubtitleAssetURL = (url) => {
try {
const extension = new URL(url, window.location.href).pathname.split(".").pop().toLowerCase();
return nonSubtitleExtensions.has(extension);
} catch (_) {
return false;
}
};
const isOpenSubtitlesDownloadURL = (url) => {
try {
const parsed = new URL(url, window.location.href);
const host = parsed.hostname.toLowerCase();
const path = parsed.pathname.toLowerCase();
if (!host.includes("opensubtitles")) {
return false;
}
if (/\/manifest\.json(?:_\d+)?$/i.test(path)) {
return false;
}
return /\/api\/v1\/download(?:\/|$)/i.test(path)
|| /\/download(?:\/|$)/i.test(path)
|| /\/subtitles?(?:\/|$)/i.test(path);
} catch (_) {
return false;
}
};
const isSubtitleURL = (url) => {
if (!url || isOpenSubtitlesManifestID(url)) {
return false;
}
subtitleURLPattern.lastIndex = 0;
const matches = subtitleURLPattern.test(url) || /api\.opensubtitles\.com\/api\/v1\/download/i.test(url);
subtitleURLPattern.lastIndex = 0;
return matches;
return !isProbablyNonSubtitleAssetURL(url)
&& (isDirectSubtitleFileURL(url) || isOpenSubtitlesDownloadURL(url));
};
const findResolverURL = () => {

View file

@ -132,6 +132,10 @@ struct StreamCandidate {
enum SubtitleCandidateParser {
private static let supportedExtensions = ["srt", "vtt", "ass", "ssa", "sub"]
private static let nonSubtitleExtensions = [
"aac", "avi", "bmp", "css", "gif", "heic", "ico", "jpeg", "jpg", "js", "json",
"m4a", "m4v", "mkv", "mov", "mp3", "mp4", "mpeg", "mpg", "png", "svg", "ts", "webm", "webp"
]
private static let urlFields = ["url", "href", "src", "link", "subtitles", "subtitle", "subtitleUrl", "subtitleURL", "file", "download", "fileUrl", "fileURL"]
private static let labelFields = ["label", "name", "title", "file_name", "filename", "lang", "language", "id"]
private struct CandidateContext {
@ -244,14 +248,14 @@ enum SubtitleCandidateParser {
return nil
}
let lowercased = url.absoluteString.lowercased()
if isOpenSubtitlesManifestIdentifier(url) {
return nil
}
guard supportedExtensions.contains(url.pathExtension.lowercased())
|| supportedExtensions.contains(where: { lowercased.contains(".\($0)?") || lowercased.contains(".\($0)&") })
|| lowercased.contains("subtitle")
|| lowercased.contains("opensubtitles")
guard !nonSubtitleExtensions.contains(url.pathExtension.lowercased()) else {
return nil
}
guard isDirectSubtitleFile(url)
|| isOpenSubtitlesDownloadURL(url)
else {
return nil
}
@ -259,6 +263,25 @@ enum SubtitleCandidateParser {
return url
}
private static func isDirectSubtitleFile(_ url: URL) -> Bool {
let lowercased = url.absoluteString.lowercased()
return supportedExtensions.contains(url.pathExtension.lowercased())
|| supportedExtensions.contains(where: { lowercased.contains(".\($0)?") || lowercased.contains(".\($0)&") })
}
private static func isOpenSubtitlesDownloadURL(_ url: URL) -> Bool {
guard url.host?.localizedCaseInsensitiveContains("opensubtitles") == true else {
return false
}
let path = url.path.lowercased()
guard !isOpenSubtitlesManifestIdentifier(url) else {
return false
}
return path.range(of: #"(^|/)api/v1/download(/|$)"#, options: .regularExpression) != nil
|| path.range(of: #"(^|/)download(/|$)"#, options: .regularExpression) != nil
|| path.range(of: #"(^|/)subtitles?(/|$)"#, options: .regularExpression) != nil
}
private static func isOpenSubtitlesManifestIdentifier(_ url: URL) -> Bool {
guard url.host?.localizedCaseInsensitiveContains("opensubtitles") == true else {
return false