netfelix-audio-fix/server/services/path-parser.ts

import path from "node:path";

const VIDEO_EXTENSIONS = new Set(["mkv", "mp4", "avi", "m4v", "ts", "wmv"]);

export interface ParsedPath {
	type: "Movie" | "Episode";
	name: string;
	year: number | null;
	seriesName: string | null;
	seasonNumber: number | null;
	episodeNumber: number | null;
	imdbId: string | null;
	tmdbId: string | null;
	tvdbId: string | null;
	container: string;
}

function extractYear(str: string): number | null {
	const m = str.match(/\((\d{4})\)/);
	return m ? parseInt(m[1], 10) : null;
}

function extractId(str: string, prefix: string): string | null {
	const re = new RegExp(`\\[${prefix}-([^\\]]+)\\]`, "i");
	const m = str.match(re);
	return m ? m[1] : null;
}

/** Strip trailing `(YYYY)` and any bracketed provider-id tokens from a folder name, return bare title. */
function titleFromFolder(folderName: string): string {
	return folderName.replace(/\s*\(\d{4}\).*$/, "").trim();
}

export function parsePath(filePath: string, moviesRoot: string, tvRoot: string): ParsedPath | null {
	const ext = path.extname(filePath).replace(/^\./, "").toLowerCase();
	if (!VIDEO_EXTENSIONS.has(ext)) return null;

	// Normalise roots so comparisons work regardless of trailing slash.
	const moviesPrefix = moviesRoot.replace(/\/+$/, "") + "/";
	const tvPrefix = tvRoot.replace(/\/+$/, "") + "/";

	const fileName = path.basename(filePath, `.${ext}`);

	if (filePath.startsWith(moviesPrefix)) {
		return parseMovie(filePath, moviesPrefix, fileName, ext);
	}

	if (filePath.startsWith(tvPrefix)) {
		return parseEpisode(filePath, tvPrefix, fileName, ext);
	}

	return null;
}

function parseMovie(filePath: string, moviesPrefix: string, fileName: string, ext: string): ParsedPath | null {
	// Path: /movies/{Folder}/{filename}.ext  — folder is the immediate child of moviesRoot.
	const relative = filePath.slice(moviesPrefix.length); // e.g. "Hot Fuzz (2007)/Hot Fuzz (2007) [imdbid-...].mkv"
	const folderName = relative.split("/")[0];

	const name = titleFromFolder(folderName);
	const year = extractYear(folderName);

	// IDs can appear in filename or folder name.
	const searchStr = `${folderName} ${fileName}`;
	const imdbId = extractId(searchStr, "imdbid");
	const tmdbId = extractId(searchStr, "tmdbid");

	return {
		type: "Movie",
		name,
		year,
		seriesName: null,
		seasonNumber: null,
		episodeNumber: null,
		imdbId,
		tmdbId,
		tvdbId: null,
		container: ext,
	};
}

function parseEpisode(filePath: string, tvPrefix: string, fileName: string, ext: string): ParsedPath | null {
	// Path: /tv/{Series Folder}/Season NN/{filename}.ext
	const relative = filePath.slice(tvPrefix.length);
	const parts = relative.split("/");
	if (parts.length < 2) return null;

	const seriesFolder = parts[0]; // e.g. "Arrow (2012) [tvdbid-257655]"

	const seriesName = titleFromFolder(seriesFolder);
	const year = extractYear(seriesFolder);
	const tvdbId = extractId(seriesFolder, "tvdbid");

	// Season/episode from S01E01 (or S02E01-E13 for multi-episode).
	const seMatch = fileName.match(/S(\d{2})E(\d{2})/i);
	if (!seMatch) return null;

	const seasonNumber = parseInt(seMatch[1], 10);
	const episodeNumber = parseInt(seMatch[2], 10);

	// Episode title: everything after "- S01E01... -" up to the first "[".
	// e.g. "Breaking Bad (2008) - S05E03 - Hazard Pay [WEBDL-...]"
	// Some files have no episode title: "Lost (2004) - S02E21 - [DSNP WEBDL-1080p]..."
	const epTitleMatch = fileName.match(/S\d{2}E\d{2}(?:-E\d{2})?\s*-\s*([^\[]+?)(?:\s*\[|$)/i);
	const episodeTitle = epTitleMatch ? epTitleMatch[1].trim() || `S${seMatch[1]}E${seMatch[2]}` : `S${seMatch[1]}E${seMatch[2]}`;

	return {
		type: "Episode",
		name: episodeTitle,
		year,
		seriesName,
		seasonNumber,
		episodeNumber,
		imdbId: null,
		tmdbId: null,
		tvdbId,
		container: ext,
	};
}