@media-engine/providers 0.1.1 → 1.0.0

package/dist/imdb-dataset/memory-storage.js

@@ -0,0 +1,168 @@
+import { normalizeProviderSearchText } from "../shared/mapping.js";
+// Builds the legacy in-memory adapter used for small fixtures and compatibility.
+// Создаёт прежний in-memory adapter для малых fixtures и совместимости.
+export function createImdbDatasetMemoryStorage(options) {
+    const recordsById = new Map();
+    const ratingsById = parseRatings(options.titleRatingsTsv);
+    for (const row of parseTsv(options.titleBasicsTsv)) {
+        const record = mapTitleBasicsRow(row, options.includeAdult ?? false);
+        if (record) {
+            recordsById.set(record.imdbId, record);
+        }
+    }
+    return {
+        getTitleById(imdbId, lookupOptions) {
+            lookupOptions?.signal?.throwIfAborted();
+            const record = recordsById.get(imdbId);
+            return record ? createStorageRecord(record, ratingsById.get(imdbId)) : undefined;
+        },
+        searchTitles(query) {
+            return searchMemoryStorage(recordsById, ratingsById, query);
+        },
+    };
+}
+function searchMemoryStorage(recordsById, ratingsById, query) {
+    query.signal?.throwIfAborted();
+    const matches = [];
+    let visited = 0;
+    for (const record of recordsById.values()) {
+        if (visited % 4096 === 0) {
+            query.signal?.throwIfAborted();
+        }
+        visited += 1;
+        if ((query.type !== undefined && record.type !== query.type) ||
+            (query.year !== undefined && record.startYear !== query.year)) {
+            continue;
+        }
+        const confidence = scoreTitle(record, query.normalizedTitle);
+        if (confidence <= 0) {
+            continue;
+        }
+        const candidate = { record, confidence };
+        const worstMatch = matches.at(-1);
+        if (matches.length >= query.limit &&
+            worstMatch &&
+            compareSearchMatches(candidate, worstMatch) >= 0) {
+            continue;
+        }
+        matches.push(candidate);
+        matches.sort(compareSearchMatches);
+        if (matches.length > query.limit) {
+            matches.pop();
+        }
+    }
+    query.signal?.throwIfAborted();
+    return matches.map(({ record, confidence }) => ({
+        record: createStorageRecord(record, ratingsById.get(record.imdbId)),
+        confidence,
+    }));
+}
+function mapTitleBasicsRow(row, includeAdult) {
+    const type = mapTitleType(row.titleType);
+    if (!type || (!includeAdult && row.isAdult === "1")) {
+        return undefined;
+    }
+    const originalTitle = emptyToUndefined(row.originalTitle);
+    return {
+        imdbId: row.tconst,
+        type,
+        primaryTitle: row.primaryTitle,
+        originalTitle,
+        startYear: parseNumber(row.startYear),
+        endYear: parseNumber(row.endYear),
+        runtimeMinutes: parseNumber(row.runtimeMinutes),
+        genres: parseList(row.genres),
+        normalizedPrimaryTitle: normalizeProviderSearchText(row.primaryTitle),
+        normalizedOriginalTitle: normalizeProviderSearchText(originalTitle ?? ""),
+    };
+}
+function createStorageRecord(record, rating) {
+    return {
+        imdbId: record.imdbId,
+        type: record.type,
+        primaryTitle: record.primaryTitle,
+        originalTitle: record.originalTitle,
+        startYear: record.startYear,
+        endYear: record.endYear,
+        runtimeMinutes: record.runtimeMinutes,
+        genres: record.genres,
+        rating,
+    };
+}
+function parseRatings(tsv) {
+    const ratings = new Map();
+    if (!tsv) {
+        return ratings;
+    }
+    for (const row of parseTsv(tsv)) {
+        const rating = parseNumber(row.averageRating);
+        if (row.tconst && rating !== undefined) {
+            ratings.set(row.tconst, {
+                averageRating: rating,
+                numVotes: parseNumber(row.numVotes),
+            });
+        }
+    }
+    return ratings;
+}
+function* parseTsv(content) {
+    const firstLineEnd = content.indexOf("\n");
+    const headerLine = (firstLineEnd < 0 ? content : content.slice(0, firstLineEnd)).replace(/\r$/, "");
+    const headers = headerLine.split("\t");
+    let offset = firstLineEnd < 0 ? content.length : firstLineEnd + 1;
+    while (offset < content.length) {
+        const nextLineEnd = content.indexOf("\n", offset);
+        const lineEnd = nextLineEnd < 0 ? content.length : nextLineEnd;
+        const line = content.slice(offset, lineEnd).replace(/\r$/, "");
+        offset = nextLineEnd < 0 ? content.length : nextLineEnd + 1;
+        if (!line) {
+            continue;
+        }
+        const values = line.split("\t");
+        const row = {};
+        headers.forEach((header, index) => {
+            row[header] = values[index] ?? "";
+        });
+        yield row;
+    }
+}
+function scoreTitle(record, normalizedQuery) {
+    const primary = record.normalizedPrimaryTitle;
+    const original = record.normalizedOriginalTitle;
+    if (primary === normalizedQuery || original === normalizedQuery) {
+        return 1;
+    }
+    if (primary.includes(normalizedQuery) || original.includes(normalizedQuery)) {
+        return 0.75;
+    }
+    return 0;
+}
+function mapTitleType(titleType) {
+    if (titleType === "movie") {
+        return "movie";
+    }
+    if (titleType === "tvSeries") {
+        return "series";
+    }
+    return undefined;
+}
+function parseList(value) {
+    const normalized = emptyToUndefined(value);
+    return normalized ? normalized.split(",").filter(Boolean) : undefined;
+}
+function parseNumber(value) {
+    const normalized = emptyToUndefined(value);
+    if (!normalized) {
+        return undefined;
+    }
+    const parsed = Number(normalized);
+    return Number.isFinite(parsed) ? parsed : undefined;
+}
+function emptyToUndefined(value) {
+    return value && value !== "\\N" ? value : undefined;
+}
+function compareSearchMatches(left, right) {
+    return (right.confidence - left.confidence ||
+        (right.record.startYear ?? 0) - (left.record.startYear ?? 0));
+}
+//# sourceMappingURL=memory-storage.js.map

package/dist/imdb-dataset/memory-storage.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"memory-storage.js","sourceRoot":"","sources":["../../src/imdb-dataset/memory-storage.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,2BAA2B,EAAE,MAAM,sBAAsB,CAAC;AAsBnE,iFAAiF;AACjF,wEAAwE;AACxE,MAAM,UAAU,8BAA8B,CAC5C,OAAwC;IAExC,MAAM,WAAW,GAAG,IAAI,GAAG,EAA8B,CAAC;IAC1D,MAAM,WAAW,GAAG,YAAY,CAAC,OAAO,CAAC,eAAe,CAAC,CAAC;IAE1D,KAAK,MAAM,GAAG,IAAI,QAAQ,CAAC,OAAO,CAAC,cAAc,CAAC,EAAE,CAAC;QACnD,MAAM,MAAM,GAAG,iBAAiB,CAAC,GAAG,EAAE,OAAO,CAAC,YAAY,IAAI,KAAK,CAAC,CAAC;QAErE,IAAI,MAAM,EAAE,CAAC;YACX,WAAW,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;QACzC,CAAC;IACH,CAAC;IAED,OAAO;QACL,YAAY,CAAC,MAAM,EAAE,aAAa;YAChC,aAAa,EAAE,MAAM,EAAE,cAAc,EAAE,CAAC;YACxC,MAAM,MAAM,GAAG,WAAW,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;YACvC,OAAO,MAAM,CAAC,CAAC,CAAC,mBAAmB,CAAC,MAAM,EAAE,WAAW,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;QACnF,CAAC;QACD,YAAY,CAAC,KAAK;YAChB,OAAO,mBAAmB,CAAC,WAAW,EAAE,WAAW,EAAE,KAAK,CAAC,CAAC;QAC9D,CAAC;KACF,CAAC;AACJ,CAAC;AAED,SAAS,mBAAmB,CAC1B,WAAoD,EACpD,WAAyD,EACzD,KAAoC;IAEpC,KAAK,CAAC,MAAM,EAAE,cAAc,EAAE,CAAC;IAC/B,MAAM,OAAO,GAA8D,EAAE,CAAC;IAC9E,IAAI,OAAO,GAAG,CAAC,CAAC;IAEhB,KAAK,MAAM,MAAM,IAAI,WAAW,CAAC,MAAM,EAAE,EAAE,CAAC;QAC1C,IAAI,OAAO,GAAG,IAAI,KAAK,CAAC,EAAE,CAAC;YACzB,KAAK,CAAC,MAAM,EAAE,cAAc,EAAE,CAAC;QACjC,CAAC;QACD,OAAO,IAAI,CAAC,CAAC;QAEb,IACE,CAAC,KAAK,CAAC,IAAI,KAAK,SAAS,IAAI,MAAM,CAAC,IAAI,KAAK,KAAK,CAAC,IAAI,CAAC;YACxD,CAAC,KAAK,CAAC,IAAI,KAAK,SAAS,IAAI,MAAM,CAAC,SAAS,KAAK,KAAK,CAAC,IAAI,CAAC,EAC7D,CAAC;YACD,SAAS;QACX,CAAC;QAED,MAAM,UAAU,GAAG,UAAU,CAAC,MAAM,EAAE,KAAK,CAAC,eAAe,CAAC,CAAC;QAE7D,IAAI,UAAU,IAAI,CAAC,EAAE,CAAC;YACpB,SAAS;QACX,CAAC;QAED,MAAM,SAAS,GAAG,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC;QACzC,MAAM,UAAU,GAAG,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QAElC,IACE,OAAO,CAAC,MAAM,IAAI,KAAK,CAAC,KAAK;YAC7B,UAAU;YACV,oBAAoB,CAAC,SAAS,EAAE,UAAU,CAAC,IAAI,CAAC,EAChD,CAAC;YACD,SAAS;QACX,CAAC;QAED,OAAO,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;QACxB,OAAO,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAC;QAEnC,IAAI,OAAO,CAAC,MAAM,GAAG,KAAK,CAAC,KAAK,EAAE,CAAC;YACjC,OAAO,CAAC,GAAG,EAAE,CAAC;QAChB,CAAC;IACH,CAAC;IAED,KAAK,CAAC,MAAM,EAAE,cAAc,EAAE,CAAC;IAC/B,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,EAAE,UAAU,EAAE,EAAE,EAAE,CAAC,CAAC;QAC9C,MAAM,EAAE,mBAAmB,CAAC,MAAM,EAAE,WAAW,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;QACnE,UAAU;KACX,CAAC,CAAC,CAAC;AACN,CAAC;AAED,SAAS,iBAAiB,CACxB,GAA2B,EAC3B,YAAqB;IAErB,MAAM,IAAI,GAAG,YAAY,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;IAEzC,IAAI,CAAC,IAAI,IAAI,CAAC,CAAC,YAAY,IAAI,GAAG,CAAC,OAAO,KAAK,GAAG,CAAC,EAAE,CAAC;QACpD,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,aAAa,GAAG,gBAAgB,CAAC,GAAG,CAAC,aAAa,CAAC,CAAC;IAE1D,OAAO;QACL,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,IAAI;QACJ,YAAY,EAAE,GAAG,CAAC,YAAY;QAC9B,aAAa;QACb,SAAS,EAAE,WAAW,CAAC,GAAG,CAAC,SAAS,CAAC;QACrC,OAAO,EAAE,WAAW,CAAC,GAAG,CAAC,OAAO,CAAC;QACjC,cAAc,EAAE,WAAW,CAAC,GAAG,CAAC,cAAc,CAAC;QAC/C,MAAM,EAAE,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC;QAC7B,sBAAsB,EAAE,2BAA2B,CAAC,GAAG,CAAC,YAAY,CAAC;QACrE,uBAAuB,EAAE,2BAA2B,CAAC,aAAa,IAAI,EAAE,CAAC;KAC1E,CAAC;AACJ,CAAC;AAED,SAAS,mBAAmB,CAC1B,MAA0B,EAC1B,MAA2C;IAE3C,OAAO;QACL,MAAM,EAAE,MAAM,CAAC,MAAM;QACrB,IAAI,EAAE,MAAM,CAAC,IAAI;QACjB,YAAY,EAAE,MAAM,CAAC,YAAY;QACjC,aAAa,EAAE,MAAM,CAAC,aAAa;QACnC,SAAS,EAAE,MAAM,CAAC,SAAS;QAC3B,OAAO,EAAE,MAAM,CAAC,OAAO;QACvB,cAAc,EAAE,MAAM,CAAC,cAAc;QACrC,MAAM,EAAE,MAAM,CAAC,MAAM;QACrB,MAAM;KACP,CAAC;AACJ,CAAC;AAED,SAAS,YAAY,CAAC,GAAuB;IAC3C,MAAM,OAAO,GAAG,IAAI,GAAG,EAAmC,CAAC;IAE3D,IAAI,CAAC,GAAG,EAAE,CAAC;QACT,OAAO,OAAO,CAAC;IACjB,CAAC;IAED,KAAK,MAAM,GAAG,IAAI,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC;QAChC,MAAM,MAAM,GAAG,WAAW,CAAC,GAAG,CAAC,aAAa,CAAC,CAAC;QAE9C,IAAI,GAAG,CAAC,MAAM,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;YACvC,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,MAAM,EAAE;gBACtB,aAAa,EAAE,MAAM;gBACrB,QAAQ,EAAE,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC;aACpC,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,QAAQ,CAAC,CAAC,QAAQ,CAAC,OAAe;IAChC,MAAM,YAAY,GAAG,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;IAC3C,MAAM,UAAU,GAAG,CAAC,YAAY,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,YAAY,CAAC,CAAC,CAAC,OAAO,CACtF,KAAK,EACL,EAAE,CACH,CAAC;IACF,MAAM,OAAO,GAAG,UAAU,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACvC,IAAI,MAAM,GAAG,YAAY,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,GAAG,CAAC,CAAC;IAElE,OAAO,MAAM,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC;QAC/B,MAAM,WAAW,GAAG,OAAO,CAAC,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;QAClD,MAAM,OAAO,GAAG,WAAW,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,WAAW,CAAC;QAC/D,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;QAC/D,MAAM,GAAG,WAAW,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,WAAW,GAAG,CAAC,CAAC;QAE5D,IAAI,CAAC,IAAI,EAAE,CAAC;YACV,SAAS;QACX,CAAC;QAED,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAChC,MAAM,GAAG,GAA2B,EAAE,CAAC;QAEvC,OAAO,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE;YAChC,GAAG,CAAC,MAAM,CAAC,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;QACpC,CAAC,CAAC,CAAC;QAEH,MAAM,GAAG,CAAC;IACZ,CAAC;AACH,CAAC;AAED,SAAS,UAAU,CAAC,MAA0B,EAAE,eAAuB;IACrE,MAAM,OAAO,GAAG,MAAM,CAAC,sBAAsB,CAAC;IAC9C,MAAM,QAAQ,GAAG,MAAM,CAAC,uBAAuB,CAAC;IAEhD,IAAI,OAAO,KAAK,eAAe,IAAI,QAAQ,KAAK,eAAe,EAAE,CAAC;QAChE,OAAO,CAAC,CAAC;IACX,CAAC;IAED,IAAI,OAAO,CAAC,QAAQ,CAAC,eAAe,CAAC,IAAI,QAAQ,CAAC,QAAQ,CAAC,eAAe,CAAC,EAAE,CAAC;QAC5E,OAAO,IAAI,CAAC;IACd,CAAC;IAED,OAAO,CAAC,CAAC;AACX,CAAC;AAED,SAAS,YAAY,CAAC,SAA6B;IACjD,IAAI,SAAS,KAAK,OAAO,EAAE,CAAC;QAC1B,OAAO,OAAO,CAAC;IACjB,CAAC;IAED,IAAI,SAAS,KAAK,UAAU,EAAE,CAAC;QAC7B,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,SAAS,CAAC,KAAyB;IAC1C,MAAM,UAAU,GAAG,gBAAgB,CAAC,KAAK,CAAC,CAAC;IAC3C,OAAO,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;AACxE,CAAC;AAED,SAAS,WAAW,CAAC,KAAyB;IAC5C,MAAM,UAAU,GAAG,gBAAgB,CAAC,KAAK,CAAC,CAAC;IAE3C,IAAI,CAAC,UAAU,EAAE,CAAC;QAChB,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,MAAM,GAAG,MAAM,CAAC,UAAU,CAAC,CAAC;IAClC,OAAO,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,CAAC;AACtD,CAAC;AAED,SAAS,gBAAgB,CAAC,KAAyB;IACjD,OAAO,KAAK,IAAI,KAAK,KAAK,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAAC;AACtD,CAAC;AAED,SAAS,oBAAoB,CAC3B,IAAwD,EACxD,KAAyD;IAEzD,OAAO,CACL,KAAK,CAAC,UAAU,GAAG,IAAI,CAAC,UAAU;QAClC,CAAC,KAAK,CAAC,MAAM,CAAC,SAAS,IAAI,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,IAAI,CAAC,CAAC,CAC7D,CAAC;AACJ,CAAC"}

package/dist/imdb-dataset/sqlite-import.d.ts

@@ -0,0 +1,24 @@
+export type ImdbDatasetSqliteBuildPhase = "titles" | "ratings" | "finalize";
+export interface ImdbDatasetSqliteBuildProgress {
+    phase: ImdbDatasetSqliteBuildPhase;
+    processedRows: number;
+    importedRows: number;
+}
+export interface ImdbDatasetSqliteBuildOptions {
+    titleBasicsPath: string;
+    titleRatingsPath?: string;
+    outputPath: string;
+    includeAdult?: boolean;
+    signal?: AbortSignal;
+    onProgress?: (progress: ImdbDatasetSqliteBuildProgress) => void;
+}
+export interface ImdbDatasetSqliteBuildResult {
+    outputPath: string;
+    schemaVersion: number;
+    titleRows: number;
+    ratingRows: number;
+    skippedTitleRows: number;
+    indexBytes: number;
+    durationMs: number;
+}
+export declare function buildImdbDatasetSqliteIndex(options: ImdbDatasetSqliteBuildOptions): Promise<ImdbDatasetSqliteBuildResult>;

package/dist/imdb-dataset/sqlite-import.js

@@ -0,0 +1,298 @@
+import { createReadStream } from "node:fs";
+import { mkdir, open, rename, stat, unlink } from "node:fs/promises";
+import { basename, dirname, resolve } from "node:path";
+import { createInterface } from "node:readline";
+import { randomUUID } from "node:crypto";
+import { createGunzip } from "node:zlib";
+import { addAbortSignal } from "node:stream";
+import { normalizeProviderSearchText } from "../shared/mapping.js";
+import { createImdbSqliteDatabase } from "./sqlite-runtime.js";
+import { createImdbDatasetSqliteSchema, IMDB_DATASET_SQLITE_SCHEMA_VERSION, validateImdbDatasetSqliteIntegrity, validateImdbDatasetSqliteSchema, } from "./sqlite-schema.js";
+const PROGRESS_INTERVAL_ROWS = 50_000;
+const REQUIRED_BASICS_COLUMNS = [
+    "tconst",
+    "titleType",
+    "primaryTitle",
+    "originalTitle",
+    "isAdult",
+    "startYear",
+    "endYear",
+    "runtimeMinutes",
+    "genres",
+];
+const REQUIRED_RATINGS_COLUMNS = ["tconst", "averageRating", "numVotes"];
+// Streams official plain or gzip TSV files into a versioned index and atomically publishes it.
+// Потоково импортирует plain или gzip TSV в versioned index и атомарно публикует его.
+export async function buildImdbDatasetSqliteIndex(options) {
+    const startedAt = performance.now();
+    const titleBasicsPath = resolveRequiredPath(options.titleBasicsPath, "titleBasicsPath");
+    const titleRatingsPath = options.titleRatingsPath
+        ? resolveRequiredPath(options.titleRatingsPath, "titleRatingsPath")
+        : undefined;
+    const outputPath = resolveRequiredPath(options.outputPath, "outputPath");
+    if (outputPath === titleBasicsPath || outputPath === titleRatingsPath) {
+        throw new TypeError("IMDb SQLite outputPath must differ from every input path");
+    }
+    options.signal?.throwIfAborted();
+    await mkdir(dirname(outputPath), { recursive: true });
+    const temporaryPath = resolve(dirname(outputPath), `.${basename(outputPath)}.${process.pid}.${randomUUID()}.tmp`);
+    let database;
+    try {
+        database = await createImdbSqliteDatabase(temporaryPath, {
+            enableForeignKeyConstraints: true,
+            enableDoubleQuotedStringLiterals: false,
+            allowExtension: false,
+        });
+        createImdbDatasetSqliteSchema(database);
+        database.exec("BEGIN IMMEDIATE");
+        let titles;
+        let ratings = { processed: 0, imported: 0 };
+        try {
+            titles = await importTitles(database, titleBasicsPath, options);
+            if (titleRatingsPath) {
+                ratings = await importRatings(database, titleRatingsPath, options);
+            }
+            database.prepare("UPDATE metadata SET value = 'ready' WHERE key = 'state'").run();
+            database
+                .prepare("INSERT INTO metadata(key, value) VALUES ('title_rows', ?)")
+                .run(String(titles.imported));
+            database
+                .prepare("INSERT INTO metadata(key, value) VALUES ('rating_rows', ?)")
+                .run(String(ratings.imported));
+            database.exec("COMMIT");
+        }
+        catch (error) {
+            database.exec("ROLLBACK");
+            throw error;
+        }
+        options.signal?.throwIfAborted();
+        emitProgress(options, "finalize", titles.processed, titles.imported);
+        database.exec("INSERT INTO title_search(title_search) VALUES ('optimize'); ANALYZE; PRAGMA optimize; VACUUM;");
+        validateImdbDatasetSqliteSchema(database);
+        validateImdbDatasetSqliteIntegrity(database);
+        database.close();
+        database = undefined;
+        await syncFile(temporaryPath);
+        options.signal?.throwIfAborted();
+        await rename(temporaryPath, outputPath);
+        await syncDirectory(dirname(outputPath));
+        const index = await stat(outputPath);
+        return {
+            outputPath,
+            schemaVersion: IMDB_DATASET_SQLITE_SCHEMA_VERSION,
+            titleRows: titles.imported,
+            ratingRows: ratings.imported,
+            skippedTitleRows: titles.processed - titles.imported,
+            indexBytes: index.size,
+            durationMs: performance.now() - startedAt,
+        };
+    }
+    catch (error) {
+        if (database) {
+            try {
+                database.close();
+            }
+            catch {
+                // Preserve the original import error.
+            }
+        }
+        await cleanupTemporaryDatabase(temporaryPath);
+        throw error;
+    }
+}
+async function importTitles(database, path, options) {
+    const insertTitle = database.prepare(`
+    INSERT OR IGNORE INTO titles (
+      imdb_id,
+      media_type,
+      primary_title,
+      original_title,
+      start_year,
+      end_year,
+      runtime_minutes,
+      genres,
+      normalized_primary_title,
+      normalized_original_title
+    ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+  `);
+    const insertSearch = database.prepare(`
+    INSERT INTO title_search(rowid, normalized_primary_title, normalized_original_title)
+    VALUES (?, ?, ?)
+  `);
+    const counts = { processed: 0, imported: 0 };
+    await forEachTsvRow(path, REQUIRED_BASICS_COLUMNS, options.signal, (values, columns) => {
+        counts.processed += 1;
+        checkProgress(options, "titles", counts);
+        const imdbId = readValue(values, columns.tconst);
+        const mediaType = mapMediaType(readValue(values, columns.titleType));
+        const primaryTitle = emptyToUndefined(readValue(values, columns.primaryTitle));
+        if (!/^tt\d{7,12}$/.test(imdbId) ||
+            !mediaType ||
+            !primaryTitle ||
+            (!options.includeAdult && readValue(values, columns.isAdult) === "1")) {
+            return;
+        }
+        const originalTitle = emptyToUndefined(readValue(values, columns.originalTitle));
+        const normalizedPrimaryTitle = normalizeProviderSearchText(primaryTitle);
+        const normalizedOriginalTitle = normalizeProviderSearchText(originalTitle ?? "");
+        if (!normalizedPrimaryTitle) {
+            return;
+        }
+        const insertion = insertTitle.run(imdbId, mediaType, primaryTitle, originalTitle ?? null, parseNumber(readValue(values, columns.startYear)) ?? null, parseNumber(readValue(values, columns.endYear)) ?? null, parseNumber(readValue(values, columns.runtimeMinutes)) ?? null, emptyToUndefined(readValue(values, columns.genres)) ?? null, normalizedPrimaryTitle, normalizedOriginalTitle);
+        if (Number(insertion.changes) === 0) {
+            return;
+        }
+        insertSearch.run(insertion.lastInsertRowid, normalizedPrimaryTitle, normalizedOriginalTitle === normalizedPrimaryTitle ? "" : normalizedOriginalTitle);
+        counts.imported += 1;
+    });
+    emitProgress(options, "titles", counts.processed, counts.imported);
+    return counts;
+}
+async function importRatings(database, path, options) {
+    const insertRating = database.prepare(`
+    INSERT OR REPLACE INTO ratings(imdb_id, average_rating, num_votes)
+    SELECT ?, ?, ?
+    WHERE EXISTS (SELECT 1 FROM titles WHERE imdb_id = ?)
+  `);
+    const counts = { processed: 0, imported: 0 };
+    await forEachTsvRow(path, REQUIRED_RATINGS_COLUMNS, options.signal, (values, columns) => {
+        counts.processed += 1;
+        checkProgress(options, "ratings", counts);
+        const imdbId = readValue(values, columns.tconst);
+        const averageRating = parseNumber(readValue(values, columns.averageRating));
+        if (!/^tt\d{7,12}$/.test(imdbId) || averageRating === undefined) {
+            return;
+        }
+        const insertion = insertRating.run(imdbId, averageRating, parseNumber(readValue(values, columns.numVotes)) ?? null, imdbId);
+        counts.imported += Number(insertion.changes);
+    });
+    emitProgress(options, "ratings", counts.processed, counts.imported);
+    return counts;
+}
+async function forEachTsvRow(path, requiredColumns, signal, visit) {
+    signal?.throwIfAborted();
+    const file = createReadStream(path, { signal });
+    let input = file;
+    let forwardFileError;
+    if (path.toLowerCase().endsWith(".gz")) {
+        const gunzip = createGunzip();
+        forwardFileError = (error) => gunzip.destroy(error);
+        file.once("error", forwardFileError);
+        input = file.pipe(gunzip);
+        if (signal) {
+            addAbortSignal(signal, input);
+        }
+    }
+    const lines = createInterface({ input, crlfDelay: Number.POSITIVE_INFINITY });
+    let columns;
+    try {
+        for await (const rawLine of lines) {
+            signal?.throwIfAborted();
+            const line = rawLine.replace(/\r$/, "");
+            if (!columns) {
+                columns = resolveColumns(line.replace(/^\uFEFF/, ""), requiredColumns, path);
+                continue;
+            }
+            if (line) {
+                visit(line.split("\t"), columns);
+            }
+        }
+    }
+    finally {
+        lines.close();
+        input.destroy();
+        file.destroy();
+        if (forwardFileError) {
+            file.off("error", forwardFileError);
+        }
+    }
+    if (!columns) {
+        throw new Error(`IMDb TSV file is empty: ${path}`);
+    }
+}
+function resolveColumns(headerLine, requiredColumns, path) {
+    const headers = headerLine.split("\t");
+    const columns = {};
+    for (const name of requiredColumns) {
+        const index = headers.indexOf(name);
+        if (index < 0) {
+            throw new Error(`IMDb TSV file ${path} is missing required column ${name}`);
+        }
+        columns[name] = index;
+    }
+    return columns;
+}
+function checkProgress(options, phase, counts) {
+    if (counts.processed % 4096 === 0) {
+        options.signal?.throwIfAborted();
+    }
+    if (counts.processed % PROGRESS_INTERVAL_ROWS === 0) {
+        emitProgress(options, phase, counts.processed, counts.imported);
+    }
+}
+function emitProgress(options, phase, processedRows, importedRows) {
+    options.signal?.throwIfAborted();
+    options.onProgress?.({ phase, processedRows, importedRows });
+}
+function readValue(values, index) {
+    return values[index] ?? "";
+}
+function mapMediaType(value) {
+    if (value === "movie") {
+        return "movie";
+    }
+    if (value === "tvSeries") {
+        return "series";
+    }
+    return undefined;
+}
+function parseNumber(value) {
+    const normalized = emptyToUndefined(value);
+    if (!normalized) {
+        return undefined;
+    }
+    const parsed = Number(normalized);
+    return Number.isFinite(parsed) ? parsed : undefined;
+}
+function emptyToUndefined(value) {
+    return value && value !== "\\N" ? value : undefined;
+}
+function resolveRequiredPath(value, name) {
+    if (typeof value !== "string" || !value.trim()) {
+        throw new TypeError(`${name} must be a non-empty string`);
+    }
+    return resolve(value);
+}
+async function syncFile(path) {
+    const file = await open(path, "r");
+    try {
+        await file.sync();
+    }
+    finally {
+        await file.close();
+    }
+}
+async function syncDirectory(path) {
+    try {
+        await syncFile(path);
+    }
+    catch (error) {
+        const code = error.code;
+        if (code !== "EINVAL" && code !== "EPERM" && code !== "EISDIR") {
+            throw error;
+        }
+    }
+}
+async function cleanupTemporaryDatabase(path) {
+    await Promise.all([path, `${path}-journal`, `${path}-wal`, `${path}-shm`].map(async (candidate) => {
+        try {
+            await unlink(candidate);
+        }
+        catch (error) {
+            if (error.code !== "ENOENT") {
+                throw error;
+            }
+        }
+    }));
+}
+//# sourceMappingURL=sqlite-import.js.map

package/dist/imdb-dataset/sqlite-import.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"sqlite-import.js","sourceRoot":"","sources":["../../src/imdb-dataset/sqlite-import.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,MAAM,SAAS,CAAC;AAC3C,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,kBAAkB,CAAC;AACrE,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AACvD,OAAO,EAAE,eAAe,EAAE,MAAM,eAAe,CAAC;AAChD,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,YAAY,EAAE,MAAM,WAAW,CAAC;AACzC,OAAO,EAAE,cAAc,EAAiB,MAAM,aAAa,CAAC;AAE5D,OAAO,EAAE,2BAA2B,EAAE,MAAM,sBAAsB,CAAC;AACnE,OAAO,EAAE,wBAAwB,EAA2B,MAAM,qBAAqB,CAAC;AACxF,OAAO,EACL,6BAA6B,EAC7B,kCAAkC,EAClC,kCAAkC,EAClC,+BAA+B,GAChC,MAAM,oBAAoB,CAAC;AAE5B,MAAM,sBAAsB,GAAG,MAAM,CAAC;AAEtC,MAAM,uBAAuB,GAAG;IAC9B,QAAQ;IACR,WAAW;IACX,cAAc;IACd,eAAe;IACf,SAAS;IACT,WAAW;IACX,SAAS;IACT,gBAAgB;IAChB,QAAQ;CACA,CAAC;AAEX,MAAM,wBAAwB,GAAG,CAAC,QAAQ,EAAE,eAAe,EAAE,UAAU,CAAU,CAAC;AAkClF,+FAA+F;AAC/F,sFAAsF;AACtF,MAAM,CAAC,KAAK,UAAU,2BAA2B,CAC/C,OAAsC;IAEtC,MAAM,SAAS,GAAG,WAAW,CAAC,GAAG,EAAE,CAAC;IACpC,MAAM,eAAe,GAAG,mBAAmB,CAAC,OAAO,CAAC,eAAe,EAAE,iBAAiB,CAAC,CAAC;IACxF,MAAM,gBAAgB,GAAG,OAAO,CAAC,gBAAgB;QAC/C,CAAC,CAAC,mBAAmB,CAAC,OAAO,CAAC,gBAAgB,EAAE,kBAAkB,CAAC;QACnE,CAAC,CAAC,SAAS,CAAC;IACd,MAAM,UAAU,GAAG,mBAAmB,CAAC,OAAO,CAAC,UAAU,EAAE,YAAY,CAAC,CAAC;IAEzE,IAAI,UAAU,KAAK,eAAe,IAAI,UAAU,KAAK,gBAAgB,EAAE,CAAC;QACtE,MAAM,IAAI,SAAS,CAAC,0DAA0D,CAAC,CAAC;IAClF,CAAC;IAED,OAAO,CAAC,MAAM,EAAE,cAAc,EAAE,CAAC;IACjC,MAAM,KAAK,CAAC,OAAO,CAAC,UAAU,CAAC,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;IACtD,MAAM,aAAa,GAAG,OAAO,CAC3B,OAAO,CAAC,UAAU,CAAC,EACnB,IAAI,QAAQ,CAAC,UAAU,CAAC,IAAI,OAAO,CAAC,GAAG,IAAI,UAAU,EAAE,MAAM,CAC9D,CAAC;IACF,IAAI,QAAwC,CAAC;IAE7C,IAAI,CAAC;QACH,QAAQ,GAAG,MAAM,wBAAwB,CAAC,aAAa,EAAE;YACvD,2BAA2B,EAAE,IAAI;YACjC,gCAAgC,EAAE,KAAK;YACvC,cAAc,EAAE,KAAK;SACtB,CAAC,CAAC;QACH,6BAA6B,CAAC,QAAQ,CAAC,CAAC;QACxC,QAAQ,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC;QAEjC,IAAI,MAAoB,CAAC;QACzB,IAAI,OAAO,GAAiB,EAAE,SAAS,EAAE,CAAC,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;QAE1D,IAAI,CAAC;YACH,MAAM,GAAG,MAAM,YAAY,CAAC,QAAQ,EAAE,eAAe,EAAE,OAAO,CAAC,CAAC;YAEhE,IAAI,gBAAgB,EAAE,CAAC;gBACrB,OAAO,GAAG,MAAM,aAAa,CAAC,QAAQ,EAAE,gBAAgB,EAAE,OAAO,CAAC,CAAC;YACrE,CAAC;YAED,QAAQ,CAAC,OAAO,CAAC,yDAAyD,CAAC,CAAC,GAAG,EAAE,CAAC;YAClF,QAAQ;iBACL,OAAO,CAAC,2DAA2D,CAAC;iBACpE,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC;YAChC,QAAQ;iBACL,OAAO,CAAC,4DAA4D,CAAC;iBACrE,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC;YACjC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QAC1B,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,QAAQ,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;YAC1B,MAAM,KAAK,CAAC;QACd,CAAC;QAED,OAAO,CAAC,MAAM,EAAE,cAAc,EAAE,CAAC;QACjC,YAAY,CAAC,OAAO,EAAE,UAAU,EAAE,MAAM,CAAC,SAAS,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAC;QACrE,QAAQ,CAAC,IAAI,CACX,+FAA+F,CAChG,CAAC;QACF,+BAA+B,CAAC,QAAQ,CAAC,CAAC;QAC1C,kCAAkC,CAAC,QAAQ,CAAC,CAAC;QAC7C,QAAQ,CAAC,KAAK,EAAE,CAAC;QACjB,QAAQ,GAAG,SAAS,CAAC;QAErB,MAAM,QAAQ,CAAC,aAAa,CAAC,CAAC;QAC9B,OAAO,CAAC,MAAM,EAAE,cAAc,EAAE,CAAC;QACjC,MAAM,MAAM,CAAC,aAAa,EAAE,UAAU,CAAC,CAAC;QACxC,MAAM,aAAa,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC;QACzC,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,UAAU,CAAC,CAAC;QAErC,OAAO;YACL,UAAU;YACV,aAAa,EAAE,kCAAkC;YACjD,SAAS,EAAE,MAAM,CAAC,QAAQ;YAC1B,UAAU,EAAE,OAAO,CAAC,QAAQ;YAC5B,gBAAgB,EAAE,MAAM,CAAC,SAAS,GAAG,MAAM,CAAC,QAAQ;YACpD,UAAU,EAAE,KAAK,CAAC,IAAI;YACtB,UAAU,EAAE,WAAW,CAAC,GAAG,EAAE,GAAG,SAAS;SAC1C,CAAC;IACJ,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAI,QAAQ,EAAE,CAAC;YACb,IAAI,CAAC;gBACH,QAAQ,CAAC,KAAK,EAAE,CAAC;YACnB,CAAC;YAAC,MAAM,CAAC;gBACP,sCAAsC;YACxC,CAAC;QACH,CAAC;QAED,MAAM,wBAAwB,CAAC,aAAa,CAAC,CAAC;QAC9C,MAAM,KAAK,CAAC;IACd,CAAC;AACH,CAAC;AAED,KAAK,UAAU,YAAY,CACzB,QAA4B,EAC5B,IAAY,EACZ,OAAsC;IAEtC,MAAM,WAAW,GAAG,QAAQ,CAAC,OAAO,CAAC;;;;;;;;;;;;;GAapC,CAAC,CAAC;IACH,MAAM,YAAY,GAAG,QAAQ,CAAC,OAAO,CAAC;;;GAGrC,CAAC,CAAC;IACH,MAAM,MAAM,GAAiB,EAAE,SAAS,EAAE,CAAC,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;IAE3D,MAAM,aAAa,CAAC,IAAI,EAAE,uBAAuB,EAAE,OAAO,CAAC,MAAM,EAAE,CAAC,MAAM,EAAE,OAAO,EAAE,EAAE;QACrF,MAAM,CAAC,SAAS,IAAI,CAAC,CAAC;QACtB,aAAa,CAAC,OAAO,EAAE,QAAQ,EAAE,MAAM,CAAC,CAAC;QACzC,MAAM,MAAM,GAAG,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC;QACjD,MAAM,SAAS,GAAG,YAAY,CAAC,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,SAAS,CAAC,CAAC,CAAC;QACrE,MAAM,YAAY,GAAG,gBAAgB,CAAC,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,YAAY,CAAC,CAAC,CAAC;QAE/E,IACE,CAAC,cAAc,CAAC,IAAI,CAAC,MAAM,CAAC;YAC5B,CAAC,SAAS;YACV,CAAC,YAAY;YACb,CAAC,CAAC,OAAO,CAAC,YAAY,IAAI,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,OAAO,CAAC,KAAK,GAAG,CAAC,EACrE,CAAC;YACD,OAAO;QACT,CAAC;QAED,MAAM,aAAa,GAAG,gBAAgB,CAAC,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,aAAa,CAAC,CAAC,CAAC;QACjF,MAAM,sBAAsB,GAAG,2BAA2B,CAAC,YAAY,CAAC,CAAC;QACzE,MAAM,uBAAuB,GAAG,2BAA2B,CAAC,aAAa,IAAI,EAAE,CAAC,CAAC;QAEjF,IAAI,CAAC,sBAAsB,EAAE,CAAC;YAC5B,OAAO;QACT,CAAC;QAED,MAAM,SAAS,GAAG,WAAW,CAAC,GAAG,CAC/B,MAAM,EACN,SAAS,EACT,YAAY,EACZ,aAAa,IAAI,IAAI,EACrB,WAAW,CAAC,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,SAAS,CAAC,CAAC,IAAI,IAAI,EACzD,WAAW,CAAC,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,IAAI,EACvD,WAAW,CAAC,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,cAAc,CAAC,CAAC,IAAI,IAAI,EAC9D,gBAAgB,CAAC,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,IAAI,IAAI,EAC3D,sBAAsB,EACtB,uBAAuB,CACxB,CAAC;QAEF,IAAI,MAAM,CAAC,SAAS,CAAC,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;YACpC,OAAO;QACT,CAAC;QAED,YAAY,CAAC,GAAG,CACd,SAAS,CAAC,eAAe,EACzB,sBAAsB,EACtB,uBAAuB,KAAK,sBAAsB,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,uBAAuB,CAClF,CAAC;QACF,MAAM,CAAC,QAAQ,IAAI,CAAC,CAAC;IACvB,CAAC,CAAC,CAAC;IAEH,YAAY,CAAC,OAAO,EAAE,QAAQ,EAAE,MAAM,CAAC,SAAS,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAC;IACnE,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,KAAK,UAAU,aAAa,CAC1B,QAA4B,EAC5B,IAAY,EACZ,OAAsC;IAEtC,MAAM,YAAY,GAAG,QAAQ,CAAC,OAAO,CAAC;;;;GAIrC,CAAC,CAAC;IACH,MAAM,MAAM,GAAiB,EAAE,SAAS,EAAE,CAAC,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;IAE3D,MAAM,aAAa,CAAC,IAAI,EAAE,wBAAwB,EAAE,OAAO,CAAC,MAAM,EAAE,CAAC,MAAM,EAAE,OAAO,EAAE,EAAE;QACtF,MAAM,CAAC,SAAS,IAAI,CAAC,CAAC;QACtB,aAAa,CAAC,OAAO,EAAE,SAAS,EAAE,MAAM,CAAC,CAAC;QAC1C,MAAM,MAAM,GAAG,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC;QACjD,MAAM,aAAa,GAAG,WAAW,CAAC,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,aAAa,CAAC,CAAC,CAAC;QAE5E,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,aAAa,KAAK,SAAS,EAAE,CAAC;YAChE,OAAO;QACT,CAAC;QAED,MAAM,SAAS,GAAG,YAAY,CAAC,GAAG,CAChC,MAAM,EACN,aAAa,EACb,WAAW,CAAC,SAAS,CAAC,MAAM,EAAE,OAAO,CAAC,QAAQ,CAAC,CAAC,IAAI,IAAI,EACxD,MAAM,CACP,CAAC;QACF,MAAM,CAAC,QAAQ,IAAI,MAAM,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC;IAC/C,CAAC,CAAC,CAAC;IAEH,YAAY,CAAC,OAAO,EAAE,SAAS,EAAE,MAAM,CAAC,SAAS,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAC;IACpE,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,KAAK,UAAU,aAAa,CAC1B,IAAY,EACZ,eAAkB,EAClB,MAA+B,EAC/B,KAAqE;IAErE,MAAM,EAAE,cAAc,EAAE,CAAC;IACzB,MAAM,IAAI,GAAG,gBAAgB,CAAC,IAAI,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;IAChD,IAAI,KAAK,GAAa,IAAI,CAAC;IAC3B,IAAI,gBAAsD,CAAC;IAE3D,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QACvC,MAAM,MAAM,GAAG,YAAY,EAAE,CAAC;QAC9B,gBAAgB,GAAG,CAAC,KAAK,EAAE,EAAE,CAAC,MAAM,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;QACpD,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,gBAAgB,CAAC,CAAC;QACrC,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAE1B,IAAI,MAAM,EAAE,CAAC;YACX,cAAc,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC;QAChC,CAAC;IACH,CAAC;IAED,MAAM,KAAK,GAAG,eAAe,CAAC,EAAE,KAAK,EAAE,SAAS,EAAE,MAAM,CAAC,iBAAiB,EAAE,CAAC,CAAC;IAC9E,IAAI,OAA8C,CAAC;IAEnD,IAAI,CAAC;QACH,IAAI,KAAK,EAAE,MAAM,OAAO,IAAI,KAAK,EAAE,CAAC;YAClC,MAAM,EAAE,cAAc,EAAE,CAAC;YACzB,MAAM,IAAI,GAAG,OAAO,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;YAExC,IAAI,CAAC,OAAO,EAAE,CAAC;gBACb,OAAO,GAAG,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,EAAE,eAAe,EAAE,IAAI,CAAC,CAAC;gBAC7E,SAAS;YACX,CAAC;YAED,IAAI,IAAI,EAAE,CAAC;gBACT,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,OAAO,CAAC,CAAC;YACnC,CAAC;QACH,CAAC;IACH,CAAC;YAAS,CAAC;QACT,KAAK,CAAC,KAAK,EAAE,CAAC;QACd,KAAK,CAAC,OAAO,EAAE,CAAC;QAChB,IAAI,CAAC,OAAO,EAAE,CAAC;QAEf,IAAI,gBAAgB,EAAE,CAAC;YACrB,IAAI,CAAC,GAAG,CAAC,OAAO,EAAE,gBAAgB,CAAC,CAAC;QACtC,CAAC;IACH,CAAC;IAED,IAAI,CAAC,OAAO,EAAE,CAAC;QACb,MAAM,IAAI,KAAK,CAAC,2BAA2B,IAAI,EAAE,CAAC,CAAC;IACrD,CAAC;AACH,CAAC;AAED,SAAS,cAAc,CACrB,UAAkB,EAClB,eAAkB,EAClB,IAAY;IAEZ,MAAM,OAAO,GAAG,UAAU,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACvC,MAAM,OAAO,GAAG,EAA+B,CAAC;IAEhD,KAAK,MAAM,IAAI,IAAI,eAAe,EAAE,CAAC;QACnC,MAAM,KAAK,GAAG,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;QAEpC,IAAI,KAAK,GAAG,CAAC,EAAE,CAAC;YACd,MAAM,IAAI,KAAK,CAAC,iBAAiB,IAAI,+BAA+B,IAAI,EAAE,CAAC,CAAC;QAC9E,CAAC;QAED,OAAO,CAAC,IAAiB,CAAC,GAAG,KAAK,CAAC;IACrC,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,SAAS,aAAa,CACpB,OAAsC,EACtC,KAAkC,EAClC,MAAoB;IAEpB,IAAI,MAAM,CAAC,SAAS,GAAG,IAAI,KAAK,CAAC,EAAE,CAAC;QAClC,OAAO,CAAC,MAAM,EAAE,cAAc,EAAE,CAAC;IACnC,CAAC;IAED,IAAI,MAAM,CAAC,SAAS,GAAG,sBAAsB,KAAK,CAAC,EAAE,CAAC;QACpD,YAAY,CAAC,OAAO,EAAE,KAAK,EAAE,MAAM,CAAC,SAAS,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAC;IAClE,CAAC;AACH,CAAC;AAED,SAAS,YAAY,CACnB,OAAsC,EACtC,KAAkC,EAClC,aAAqB,EACrB,YAAoB;IAEpB,OAAO,CAAC,MAAM,EAAE,cAAc,EAAE,CAAC;IACjC,OAAO,CAAC,UAAU,EAAE,CAAC,EAAE,KAAK,EAAE,aAAa,EAAE,YAAY,EAAE,CAAC,CAAC;AAC/D,CAAC;AAED,SAAS,SAAS,CAAC,MAAgB,EAAE,KAAa;IAChD,OAAO,MAAM,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;AAC7B,CAAC;AAED,SAAS,YAAY,CAAC,KAAa;IACjC,IAAI,KAAK,KAAK,OAAO,EAAE,CAAC;QACtB,OAAO,OAAO,CAAC;IACjB,CAAC;IAED,IAAI,KAAK,KAAK,UAAU,EAAE,CAAC;QACzB,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,WAAW,CAAC,KAAa;IAChC,MAAM,UAAU,GAAG,gBAAgB,CAAC,KAAK,CAAC,CAAC;IAE3C,IAAI,CAAC,UAAU,EAAE,CAAC;QAChB,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,MAAM,GAAG,MAAM,CAAC,UAAU,CAAC,CAAC;IAClC,OAAO,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,CAAC;AACtD,CAAC;AAED,SAAS,gBAAgB,CAAC,KAAa;IACrC,OAAO,KAAK,IAAI,KAAK,KAAK,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAAC;AACtD,CAAC;AAED,SAAS,mBAAmB,CAAC,KAAa,EAAE,IAAY;IACtD,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC;QAC/C,MAAM,IAAI,SAAS,CAAC,GAAG,IAAI,6BAA6B,CAAC,CAAC;IAC5D,CAAC;IAED,OAAO,OAAO,CAAC,KAAK,CAAC,CAAC;AACxB,CAAC;AAED,KAAK,UAAU,QAAQ,CAAC,IAAY;IAClC,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;IAEnC,IAAI,CAAC;QACH,MAAM,IAAI,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;YAAS,CAAC;QACT,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;IACrB,CAAC;AACH,CAAC;AAED,KAAK,UAAU,aAAa,CAAC,IAAY;IACvC,IAAI,CAAC;QACH,MAAM,QAAQ,CAAC,IAAI,CAAC,CAAC;IACvB,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,MAAM,IAAI,GAAI,KAA+B,CAAC,IAAI,CAAC;QAEnD,IAAI,IAAI,KAAK,QAAQ,IAAI,IAAI,KAAK,OAAO,IAAI,IAAI,KAAK,QAAQ,EAAE,CAAC;YAC/D,MAAM,KAAK,CAAC;QACd,CAAC;IACH,CAAC;AACH,CAAC;AAED,KAAK,UAAU,wBAAwB,CAAC,IAAY;IAClD,MAAM,OAAO,CAAC,GAAG,CACf,CAAC,IAAI,EAAE,GAAG,IAAI,UAAU,EAAE,GAAG,IAAI,MAAM,EAAE,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,KAAK,EAAE,SAAS,EAAE,EAAE;QAC9E,IAAI,CAAC;YACH,MAAM,MAAM,CAAC,SAAS,CAAC,CAAC;QAC1B,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,IAAK,KAA+B,CAAC,IAAI,KAAK,QAAQ,EAAE,CAAC;gBACvD,MAAM,KAAK,CAAC;YACd,CAAC;QACH,CAAC;IACH,CAAC,CAAC,CACH,CAAC;AACJ,CAAC"}

package/dist/imdb-dataset/sqlite-runtime.d.ts

@@ -0,0 +1,3 @@
+import type { DatabaseSync } from "node:sqlite";
+export type ImdbSqliteDatabase = DatabaseSync;
+export declare function createImdbSqliteDatabase(path: string, options?: ConstructorParameters<typeof DatabaseSync>[1]): Promise<DatabaseSync>;

package/dist/imdb-dataset/sqlite-runtime.js

@@ -0,0 +1,11 @@
+export async function createImdbSqliteDatabase(path, options) {
+    let DatabaseSyncConstructor;
+    try {
+        ({ DatabaseSync: DatabaseSyncConstructor } = await import("node:sqlite"));
+    }
+    catch (error) {
+        throw new Error("The persisted IMDb SQLite backend requires Node.js 22.13 or newer with node:sqlite support", { cause: error });
+    }
+    return new DatabaseSyncConstructor(path, options);
+}
+//# sourceMappingURL=sqlite-runtime.js.map

package/dist/imdb-dataset/sqlite-runtime.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"sqlite-runtime.js","sourceRoot":"","sources":["../../src/imdb-dataset/sqlite-runtime.ts"],"names":[],"mappings":"AAIA,MAAM,CAAC,KAAK,UAAU,wBAAwB,CAC5C,IAAY,EACZ,OAAuD;IAEvD,IAAI,uBAA4C,CAAC;IAEjD,IAAI,CAAC;QACH,CAAC,EAAE,YAAY,EAAE,uBAAuB,EAAE,GAAG,MAAM,MAAM,CAAC,aAAa,CAAC,CAAC,CAAC;IAC5E,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,MAAM,IAAI,KAAK,CACb,4FAA4F,EAC5F,EAAE,KAAK,EAAE,KAAK,EAAE,CACjB,CAAC;IACJ,CAAC;IAED,OAAO,IAAI,uBAAuB,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;AACpD,CAAC"}

package/dist/imdb-dataset/sqlite-schema.d.ts

@@ -0,0 +1,6 @@
+import type { DatabaseSync } from "node:sqlite";
+export declare const IMDB_DATASET_SQLITE_SCHEMA_VERSION = 1;
+export declare const IMDB_DATASET_SQLITE_APPLICATION_ID = 1296385097;
+export declare function createImdbDatasetSqliteSchema(database: DatabaseSync): void;
+export declare function validateImdbDatasetSqliteSchema(database: DatabaseSync): void;
+export declare function validateImdbDatasetSqliteIntegrity(database: DatabaseSync): void;

package/dist/imdb-dataset/sqlite-schema.js

@@ -0,0 +1,88 @@
+export const IMDB_DATASET_SQLITE_SCHEMA_VERSION = 1;
+export const IMDB_DATASET_SQLITE_APPLICATION_ID = 0x4d454449;
+const SCHEMA_SQL = `
+  PRAGMA page_size = 4096;
+  PRAGMA journal_mode = OFF;
+  PRAGMA synchronous = OFF;
+  PRAGMA temp_store = FILE;
+  PRAGMA cache_size = -16384;
+  PRAGMA locking_mode = EXCLUSIVE;
+  PRAGMA application_id = ${IMDB_DATASET_SQLITE_APPLICATION_ID};
+  PRAGMA user_version = ${IMDB_DATASET_SQLITE_SCHEMA_VERSION};
+
+  CREATE TABLE metadata (
+    key TEXT PRIMARY KEY NOT NULL,
+    value TEXT NOT NULL
+  ) WITHOUT ROWID;
+
+  CREATE TABLE titles (
+    id INTEGER PRIMARY KEY,
+    imdb_id TEXT NOT NULL UNIQUE,
+    media_type TEXT NOT NULL CHECK (media_type IN ('movie', 'series')),
+    primary_title TEXT NOT NULL,
+    original_title TEXT,
+    start_year INTEGER,
+    end_year INTEGER,
+    runtime_minutes INTEGER,
+    genres TEXT,
+    normalized_primary_title TEXT NOT NULL,
+    normalized_original_title TEXT NOT NULL
+  );
+
+  CREATE INDEX titles_primary_title_idx
+    ON titles(normalized_primary_title, media_type, start_year DESC);
+  CREATE INDEX titles_original_title_idx
+    ON titles(normalized_original_title, media_type, start_year DESC);
+
+  CREATE TABLE ratings (
+    imdb_id TEXT PRIMARY KEY NOT NULL REFERENCES titles(imdb_id) ON DELETE CASCADE,
+    average_rating REAL NOT NULL,
+    num_votes INTEGER
+  ) WITHOUT ROWID;
+
+  CREATE VIRTUAL TABLE title_search USING fts5(
+    normalized_primary_title,
+    normalized_original_title,
+    content='',
+    tokenize='trigram remove_diacritics 1'
+  );
+
+  INSERT INTO metadata(key, value) VALUES
+    ('state', 'building'),
+    ('schema_version', '${IMDB_DATASET_SQLITE_SCHEMA_VERSION}');
+`;
+export function createImdbDatasetSqliteSchema(database) {
+    try {
+        database.exec(SCHEMA_SQL);
+    }
+    catch (error) {
+        throw new Error("Unable to create the IMDb SQLite index; Node.js SQLite must include FTS5 trigram support", { cause: error });
+    }
+}
+export function validateImdbDatasetSqliteSchema(database) {
+    const applicationId = readPragmaNumber(database, "application_id");
+    const schemaVersion = readPragmaNumber(database, "user_version");
+    const state = database.prepare("SELECT value FROM metadata WHERE key = 'state'").get()?.value;
+    if (applicationId !== IMDB_DATASET_SQLITE_APPLICATION_ID) {
+        throw new Error("File is not a Media Engine IMDb SQLite index");
+    }
+    if (schemaVersion !== IMDB_DATASET_SQLITE_SCHEMA_VERSION) {
+        throw new Error(`Unsupported IMDb SQLite schema version ${schemaVersion}; expected ${IMDB_DATASET_SQLITE_SCHEMA_VERSION}`);
+    }
+    if (state !== "ready") {
+        throw new Error("IMDb SQLite index is incomplete");
+    }
+}
+export function validateImdbDatasetSqliteIntegrity(database) {
+    const result = database.prepare("PRAGMA quick_check(1)").get();
+    const status = result ? Object.values(result)[0] : undefined;
+    if (status !== "ok") {
+        throw new Error(`IMDb SQLite index integrity check failed: ${String(status)}`);
+    }
+}
+function readPragmaNumber(database, name) {
+    const result = database.prepare(`PRAGMA ${name}`).get();
+    const value = result?.[name];
+    return typeof value === "number" ? value : Number.NaN;
+}
+//# sourceMappingURL=sqlite-schema.js.map

package/dist/imdb-dataset/sqlite-schema.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"sqlite-schema.js","sourceRoot":"","sources":["../../src/imdb-dataset/sqlite-schema.ts"],"names":[],"mappings":"AAEA,MAAM,CAAC,MAAM,kCAAkC,GAAG,CAAC,CAAC;AACpD,MAAM,CAAC,MAAM,kCAAkC,GAAG,UAAU,CAAC;AAE7D,MAAM,UAAU,GAAG;;;;;;;4BAOS,kCAAkC;0BACpC,kCAAkC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;0BAyClC,kCAAkC;CAC3D,CAAC;AAEF,MAAM,UAAU,6BAA6B,CAAC,QAAsB;IAClE,IAAI,CAAC;QACH,QAAQ,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;IAC5B,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,MAAM,IAAI,KAAK,CACb,0FAA0F,EAC1F,EAAE,KAAK,EAAE,KAAK,EAAE,CACjB,CAAC;IACJ,CAAC;AACH,CAAC;AAED,MAAM,UAAU,+BAA+B,CAAC,QAAsB;IACpE,MAAM,aAAa,GAAG,gBAAgB,CAAC,QAAQ,EAAE,gBAAgB,CAAC,CAAC;IACnE,MAAM,aAAa,GAAG,gBAAgB,CAAC,QAAQ,EAAE,cAAc,CAAC,CAAC;IACjE,MAAM,KAAK,GAAG,QAAQ,CAAC,OAAO,CAAC,gDAAgD,CAAC,CAAC,GAAG,EAAE,EAAE,KAAK,CAAC;IAE9F,IAAI,aAAa,KAAK,kCAAkC,EAAE,CAAC;QACzD,MAAM,IAAI,KAAK,CAAC,8CAA8C,CAAC,CAAC;IAClE,CAAC;IAED,IAAI,aAAa,KAAK,kCAAkC,EAAE,CAAC;QACzD,MAAM,IAAI,KAAK,CACb,0CAA0C,aAAa,cAAc,kCAAkC,EAAE,CAC1G,CAAC;IACJ,CAAC;IAED,IAAI,KAAK,KAAK,OAAO,EAAE,CAAC;QACtB,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAC;IACrD,CAAC;AACH,CAAC;AAED,MAAM,UAAU,kCAAkC,CAAC,QAAsB;IACvE,MAAM,MAAM,GAAG,QAAQ,CAAC,OAAO,CAAC,uBAAuB,CAAC,CAAC,GAAG,EAAE,CAAC;IAC/D,MAAM,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;IAE7D,IAAI,MAAM,KAAK,IAAI,EAAE,CAAC;QACpB,MAAM,IAAI,KAAK,CAAC,6CAA6C,MAAM,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;IACjF,CAAC;AACH,CAAC;AAED,SAAS,gBAAgB,CAAC,QAAsB,EAAE,IAAuC;IACvF,MAAM,MAAM,GAAG,QAAQ,CAAC,OAAO,CAAC,UAAU,IAAI,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC;IACxD,MAAM,KAAK,GAAG,MAAM,EAAE,CAAC,IAAI,CAAC,CAAC;IAE7B,OAAO,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC;AACxD,CAAC"}

package/dist/imdb-dataset/sqlite-storage.d.ts

@@ -0,0 +1,12 @@
+import type { ImdbDatasetStorage } from "./storage.js";
+export declare const IMDB_DATASET_SQLITE_SCHEMA_VERSION = 1;
+export interface ImdbDatasetSqliteStorageOptions {
+    path: string;
+    busyTimeoutMs?: number;
+}
+export interface ImdbDatasetSqliteStorage extends ImdbDatasetStorage {
+    readonly path: string;
+    readonly schemaVersion: number;
+    close(): void;
+}
+export declare function openImdbDatasetSqliteStorage(options: ImdbDatasetSqliteStorageOptions): Promise<ImdbDatasetSqliteStorage>;