@tscaps/engine 0.1.1 → 0.3.0

package/dist/modules/transcription/BatchedMonoResampler.js

@@ -0,0 +1,116 @@
+/**
+ * Consumes mediabunny {@link AudioSample}s in source order, downmixes
+ * each to mono, and writes resampled frames into a caller-owned target
+ * buffer at monotonically increasing offsets.
+ *
+ * Batches source-rate frames in a fixed-size internal buffer so the
+ * whole track never sits in memory at once. Each full batch is
+ * resampled through a fresh `OfflineAudioContext` and copied straight
+ * into the target — a track of any length peaks at one batch worth of
+ * source-rate PCM plus the target buffer, instead of the full-track
+ * source-rate buffer the collect-then-resample approach needed. On a
+ * 60-minute 48 kHz mono source that is ~5.7 MB per batch instead of
+ * ~690 MB across the whole run.
+ *
+ * The source sample rate is discovered on the first appended sample —
+ * `AudioSample.sampleRate` is stable across a single mediabunny track.
+ * The target sample rate is fixed at construction. When the two match
+ * the batch bypasses `OfflineAudioContext` and is copied verbatim.
+ *
+ * The target buffer is written up to its own capacity; any tail beyond
+ * that is silently discarded, so the caller's pre-size determines the
+ * ceiling. `flush` returns the actual number of frames written and the
+ * caller can `subarray(0, written)` to trim any pre-sizing slack.
+ */
+export class BatchedMonoResampler {
+    target;
+    targetSampleRate;
+    static BATCH_SECONDS = 30;
+    batch = new Float32Array(0);
+    batchFrames = 0;
+    sourceSampleRate = 0;
+    channelScratch = new Float32Array(0);
+    targetOffset = 0;
+    constructor(target, targetSampleRate) {
+        this.target = target;
+        this.targetSampleRate = targetSampleRate;
+    }
+    async append(sample) {
+        if (this.sourceSampleRate === 0)
+            this.initFromFirstSample(sample);
+        const frames = sample.numberOfFrames;
+        if (frames === 0)
+            return;
+        this.ensureBatchCapacityFor(this.batchFrames + frames);
+        this.mixSampleIntoBatch(sample, this.batchFrames);
+        this.batchFrames += frames;
+        if (this.batchFrames >= this.batch.length)
+            await this.emitBatch();
+    }
+    async flush() {
+        if (this.batchFrames > 0)
+            await this.emitBatch();
+        return this.targetOffset;
+    }
+    initFromFirstSample(sample) {
+        this.sourceSampleRate = sample.sampleRate;
+        this.batch = new Float32Array(this.sourceSampleRate * BatchedMonoResampler.BATCH_SECONDS);
+    }
+    ensureBatchCapacityFor(needed) {
+        if (this.batch.length >= needed)
+            return;
+        const grown = new Float32Array(needed);
+        grown.set(this.batch.subarray(0, this.batchFrames));
+        this.batch = grown;
+    }
+    async emitBatch() {
+        const source = this.batch.subarray(0, this.batchFrames);
+        const resampled = this.sourceSampleRate === this.targetSampleRate
+            ? source
+            : await this.resampleBatch(source);
+        this.writeIntoTarget(resampled);
+        this.batchFrames = 0;
+    }
+    async resampleBatch(source) {
+        const targetFrames = Math.max(1, Math.ceil(source.length * this.targetSampleRate / this.sourceSampleRate));
+        const ctx = new OfflineAudioContext(1, targetFrames, this.targetSampleRate);
+        const buffer = ctx.createBuffer(1, source.length, this.sourceSampleRate);
+        buffer.getChannelData(0).set(source);
+        const node = ctx.createBufferSource();
+        node.buffer = buffer;
+        node.connect(ctx.destination);
+        node.start();
+        const rendered = await ctx.startRendering();
+        return rendered.getChannelData(0);
+    }
+    writeIntoTarget(source) {
+        const room = this.target.length - this.targetOffset;
+        if (room <= 0)
+            return;
+        const writeLen = Math.min(source.length, room);
+        this.target.set(source.subarray(0, writeLen), this.targetOffset);
+        this.targetOffset += writeLen;
+    }
+    mixSampleIntoBatch(sample, writeOffset) {
+        const frames = sample.numberOfFrames;
+        const channels = sample.numberOfChannels;
+        const target = this.batch.subarray(writeOffset, writeOffset + frames);
+        if (channels <= 1) {
+            sample.copyTo(target, { planeIndex: 0, format: 'f32-planar' });
+            return;
+        }
+        if (this.channelScratch.length < frames)
+            this.channelScratch = new Float32Array(frames);
+        const scratch = this.channelScratch.subarray(0, frames);
+        sample.copyTo(target, { planeIndex: 0, format: 'f32-planar' });
+        for (let channel = 1; channel < channels; channel += 1) {
+            sample.copyTo(scratch, { planeIndex: channel, format: 'f32-planar' });
+            for (let frame = 0; frame < frames; frame += 1)
+                target[frame] += scratch[frame];
+        }
+        const inverseChannels = 1 / channels;
+        for (let frame = 0; frame < frames; frame += 1)
+            target[frame] *= inverseChannels;
+    }
+}
+//# sourceMappingURL=BatchedMonoResampler.js.map

package/dist/modules/transcription/BatchedMonoResampler.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"BatchedMonoResampler.js","sourceRoot":"","sources":["../../../src/modules/transcription/BatchedMonoResampler.ts"],"names":[],"mappings":"AAEA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,MAAM,OAAO,oBAAoB;IAWZ;IACA;IAVX,MAAM,CAAU,aAAa,GAAG,EAAE,CAAC;IAEnC,KAAK,GAAiB,IAAI,YAAY,CAAC,CAAC,CAAC,CAAC;IAC1C,WAAW,GAAG,CAAC,CAAC;IAChB,gBAAgB,GAAG,CAAC,CAAC;IACrB,cAAc,GAAiB,IAAI,YAAY,CAAC,CAAC,CAAC,CAAC;IACnD,YAAY,GAAG,CAAC,CAAC;IAEzB,YACmB,MAAoB,EACpB,gBAAwB;QADxB,WAAM,GAAN,MAAM,CAAc;QACpB,qBAAgB,GAAhB,gBAAgB,CAAQ;IACxC,CAAC;IAEJ,KAAK,CAAC,MAAM,CAAC,MAAmB;QAC9B,IAAI,IAAI,CAAC,gBAAgB,KAAK,CAAC;YAAE,IAAI,CAAC,mBAAmB,CAAC,MAAM,CAAC,CAAC;QAClE,MAAM,MAAM,GAAG,MAAM,CAAC,cAAc,CAAC;QACrC,IAAI,MAAM,KAAK,CAAC;YAAE,OAAO;QACzB,IAAI,CAAC,sBAAsB,CAAC,IAAI,CAAC,WAAW,GAAG,MAAM,CAAC,CAAC;QACvD,IAAI,CAAC,kBAAkB,CAAC,MAAM,EAAE,IAAI,CAAC,WAAW,CAAC,CAAC;QAClD,IAAI,CAAC,WAAW,IAAI,MAAM,CAAC;QAC3B,IAAI,IAAI,CAAC,WAAW,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM;YAAE,MAAM,IAAI,CAAC,SAAS,EAAE,CAAC;IACpE,CAAC;IAED,KAAK,CAAC,KAAK;QACT,IAAI,IAAI,CAAC,WAAW,GAAG,CAAC;YAAE,MAAM,IAAI,CAAC,SAAS,EAAE,CAAC;QACjD,OAAO,IAAI,CAAC,YAAY,CAAC;IAC3B,CAAC;IAEO,mBAAmB,CAAC,MAAmB;QAC7C,IAAI,CAAC,gBAAgB,GAAG,MAAM,CAAC,UAAU,CAAC;QAC1C,IAAI,CAAC,KAAK,GAAG,IAAI,YAAY,CAAC,IAAI,CAAC,gBAAgB,GAAG,oBAAoB,CAAC,aAAa,CAAC,CAAC;IAC5F,CAAC;IAEO,sBAAsB,CAAC,MAAc;QAC3C,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,MAAM;YAAE,OAAO;QACxC,MAAM,KAAK,GAAG,IAAI,YAAY,CAAC,MAAM,CAAC,CAAC;QACvC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,EAAE,IAAI,CAAC,WAAW,CAAC,CAAC,CAAC;QACpD,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;IACrB,CAAC;IAEO,KAAK,CAAC,SAAS;QACrB,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,EAAE,IAAI,CAAC,WAAW,CAAC,CAAC;QACxD,MAAM,SAAS,GAAG,IAAI,CAAC,gBAAgB,KAAK,IAAI,CAAC,gBAAgB;YAC/D,CAAC,CAAC,MAAM;YACR,CAAC,CAAC,MAAM,IAAI,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC;QACrC,IAAI,CAAC,eAAe,CAAC,SAAS,CAAC,CAAC;QAChC,IAAI,CAAC,WAAW,GAAG,CAAC,CAAC;IACvB,CAAC;IAEO,KAAK,CAAC,aAAa,CAAC,MAAoB;QAC9C,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,IAAI,CAAC,gBAAgB,GAAG,IAAI,CAAC,gBAAgB,CAAC,CAAC,CAAC;QAC3G,MAAM,GAAG,GAAG,IAAI,mBAAmB,CAAC,CAAC,EAAE,YAAY,EAAE,IAAI,CAAC,gBAAgB,CAAC,CAAC;QAC5E,MAAM,MAAM,GAAG,GAAG,CAAC,YAAY,CAAC,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,IAAI,CAAC,gBAAgB,CAAC,CAAC;QACzE,MAAM,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;QACrC,MAAM,IAAI,GAAG,GAAG,CAAC,kBAAkB,EAAE,CAAC;QACtC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,WAAW,CAAC,CAAC;QAC9B,IAAI,CAAC,KAAK,EAAE,CAAC;QACb,MAAM,QAAQ,GAAG,MAAM,GAAG,CAAC,cAAc,EAAE,CAAC;QAC5C,OAAO,QAAQ,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC;IACpC,CAAC;IAEO,eAAe,CAAC,MAAoB;QAC1C,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,IAAI,CAAC,YAAY,CAAC;QACpD,IAAI,IAAI,IAAI,CAAC;YAAE,OAAO;QACtB,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;QAC/C,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,QAAQ,CAAC,EAAE,IAAI,CAAC,YAAY,CAAC,CAAC;QACjE,IAAI,CAAC,YAAY,IAAI,QAAQ,CAAC;IAChC,CAAC;IAEO,kBAAkB,CAAC,MAAmB,EAAE,WAAmB;QACjE,MAAM,MAAM,GAAG,MAAM,CAAC,cAAc,CAAC;QACrC,MAAM,QAAQ,GAAG,MAAM,CAAC,gBAAgB,CAAC;QACzC,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,WAAW,EAAE,WAAW,GAAG,MAAM,CAAC,CAAC;QAEtE,IAAI,QAAQ,IAAI,CAAC,EAAE,CAAC;YAClB,MAAM,CAAC,MAAM,CAAC,MAAM,EAAE,EAAE,UAAU,EAAE,CAAC,EAAE,MAAM,EAAE,YAAY,EAAE,CAAC,CAAC;YAC/D,OAAO;QACT,CAAC;QAED,IAAI,IAAI,CAAC,cAAc,CAAC,MAAM,GAAG,MAAM;YAAE,IAAI,CAAC,cAAc,GAAG,IAAI,YAAY,CAAC,MAAM,CAAC,CAAC;QACxF,MAAM,OAAO,GAAG,IAAI,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC,EAAE,MAAM,CAAC,CAAC;QACxD,MAAM,CAAC,MAAM,CAAC,MAAM,EAAE,EAAE,UAAU,EAAE,CAAC,EAAE,MAAM,EAAE,YAAY,EAAE,CAAC,CAAC;QAC/D,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,OAAO,GAAG,QAAQ,EAAE,OAAO,IAAI,CAAC,EAAE,CAAC;YACvD,MAAM,CAAC,MAAM,CAAC,OAAO,EAAE,EAAE,UAAU,EAAE,OAAO,EAAE,MAAM,EAAE,YAAY,EAAE,CAAC,CAAC;YACtE,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,MAAM,EAAE,KAAK,IAAI,CAAC;gBAAE,MAAM,CAAC,KAAK,CAAE,IAAI,OAAO,CAAC,KAAK,CAAE,CAAC;QACpF,CAAC;QACD,MAAM,eAAe,GAAG,CAAC,GAAG,QAAQ,CAAC;QACrC,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,MAAM,EAAE,KAAK,IAAI,CAAC;YAAE,MAAM,CAAC,KAAK,CAAE,IAAI,eAAe,CAAC;IACpF,CAAC"}

package/dist/modules/transcription/CacheStorageModelFileCache.d.ts

@@ -0,0 +1,38 @@
+import type { ModelFileCache } from '../transcription/ModelFileCache';
+/**
+ * Keeps model files in the browser's Cache Storage.
+ *
+ * Browsers refuse this write far more often than one would expect —
+ * an origin over its quota, a private window, a page outside a secure
+ * context — and every refusal costs the next run a full re-download of
+ * hundreds of megabytes with nothing on screen to explain it. The
+ * refusal is therefore handed to `onKeepFailure` instead of being
+ * swallowed, while the run in flight carries on regardless: the files
+ * are already in memory and the transcription does not need them kept.
+ *
+ * Only the first refusal of an instance's lifetime is announced. A
+ * model is several files and the condition behind a refusal applies to
+ * all of them, so the rest would say the same thing again.
+ */
+export declare class CacheStorageModelFileCache implements ModelFileCache {
+    private readonly onKeepFailure;
+    private static readonly CACHE_NAME;
+    private opening;
+    private failureAnnounced;
+    /**
+     * @param onKeepFailure Called at most once per instance, with the
+     * refusal that stopped a file from being kept.
+     */
+    constructor(onKeepFailure: (error: unknown) => void);
+    match(key: string): Promise<Response | undefined>;
+    put(key: string, response: Response): Promise<void>;
+    private announce;
+    /**
+     * Resolves to `null` when this runtime has no Cache Storage at all,
+     * which is a standing condition rather than a transient one — hence
+     * the single attempt, reused by every later call.
+     */
+    private open;
+    private openOnce;
+}
+//# sourceMappingURL=CacheStorageModelFileCache.d.ts.map

package/dist/modules/transcription/CacheStorageModelFileCache.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CacheStorageModelFileCache.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/CacheStorageModelFileCache.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,uCAAuC,CAAC;AAG5E;;;;;;;;;;;;;;GAcG;AACH,qBAAa,0BAA2B,YAAW,cAAc;IAcnD,OAAO,CAAC,QAAQ,CAAC,aAAa;IAT1C,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,UAAU,CAAwB;IAE1D,OAAO,CAAC,OAAO,CAAsC;IACrD,OAAO,CAAC,gBAAgB,CAAS;IAEjC;;;OAGG;gBAC0B,aAAa,EAAE,CAAC,KAAK,EAAE,OAAO,KAAK,IAAI;IAE9D,KAAK,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC,QAAQ,GAAG,SAAS,CAAC;IAYjD,GAAG,CAAC,GAAG,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,GAAG,OAAO,CAAC,IAAI,CAAC;IAazD,OAAO,CAAC,QAAQ;IAMhB;;;;OAIG;IACH,OAAO,CAAC,IAAI;YAKE,QAAQ;CAUvB"}

package/dist/modules/transcription/CacheStorageModelFileCache.js

@@ -0,0 +1,86 @@
+import { ModelFileCacheUnavailableError } from '../transcription/ModelFileCacheUnavailableError';
+/**
+ * Keeps model files in the browser's Cache Storage.
+ *
+ * Browsers refuse this write far more often than one would expect —
+ * an origin over its quota, a private window, a page outside a secure
+ * context — and every refusal costs the next run a full re-download of
+ * hundreds of megabytes with nothing on screen to explain it. The
+ * refusal is therefore handed to `onKeepFailure` instead of being
+ * swallowed, while the run in flight carries on regardless: the files
+ * are already in memory and the transcription does not need them kept.
+ *
+ * Only the first refusal of an instance's lifetime is announced. A
+ * model is several files and the condition behind a refusal applies to
+ * all of them, so the rest would say the same thing again.
+ */
+export class CacheStorageModelFileCache {
+    onKeepFailure;
+    // The name transformers.js opens by default. Matching it keeps the
+    // files downloaded by earlier versions, which used the library's own
+    // caching, readable by this one.
+    static CACHE_NAME = 'transformers-cache';
+    opening = null;
+    failureAnnounced = false;
+    /**
+     * @param onKeepFailure Called at most once per instance, with the
+     * refusal that stopped a file from being kept.
+     */
+    constructor(onKeepFailure) {
+        this.onKeepFailure = onKeepFailure;
+    }
+    async match(key) {
+        const cache = await this.open();
+        if (!cache)
+            return undefined;
+        try {
+            return await cache.match(key);
+        }
+        catch {
+            // A cache that cannot be read is indistinguishable from one that
+            // never held the file: both mean it has to be downloaded.
+            return undefined;
+        }
+    }
+    async put(key, response) {
+        const cache = await this.open();
+        if (!cache) {
+            this.announce(new ModelFileCacheUnavailableError('Cache Storage is not exposed in this context'));
+            return;
+        }
+        try {
+            await cache.put(key, response);
+        }
+        catch (error) {
+            this.announce(error);
+        }
+    }
+    announce(error) {
+        if (this.failureAnnounced)
+            return;
+        this.failureAnnounced = true;
+        this.onKeepFailure(error);
+    }
+    /**
+     * Resolves to `null` when this runtime has no Cache Storage at all,
+     * which is a standing condition rather than a transient one — hence
+     * the single attempt, reused by every later call.
+     */
+    open() {
+        this.opening ??= this.openOnce();
+        return this.opening;
+    }
+    async openOnce() {
+        if (typeof caches === 'undefined')
+            return null;
+        try {
+            return await caches.open(CacheStorageModelFileCache.CACHE_NAME);
+        }
+        catch {
+            // Opening is refused outright in some privacy modes. There is
+            // nothing to fall back to, and `put` reports the consequence.
+            return null;
+        }
+    }
+}
+//# sourceMappingURL=CacheStorageModelFileCache.js.map

package/dist/modules/transcription/CacheStorageModelFileCache.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CacheStorageModelFileCache.js","sourceRoot":"","sources":["../../../src/modules/transcription/CacheStorageModelFileCache.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,8BAA8B,EAAE,MAAM,uDAAuD,CAAC;AAEvG;;;;;;;;;;;;;;GAcG;AACH,MAAM,OAAO,0BAA0B;IAcR;IAZ7B,mEAAmE;IACnE,qEAAqE;IACrE,iCAAiC;IACzB,MAAM,CAAU,UAAU,GAAG,oBAAoB,CAAC;IAElD,OAAO,GAAiC,IAAI,CAAC;IAC7C,gBAAgB,GAAG,KAAK,CAAC;IAEjC;;;OAGG;IACH,YAA6B,aAAuC;QAAvC,kBAAa,GAAb,aAAa,CAA0B;IAAG,CAAC;IAExE,KAAK,CAAC,KAAK,CAAC,GAAW;QACrB,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,IAAI,EAAE,CAAC;QAChC,IAAI,CAAC,KAAK;YAAE,OAAO,SAAS,CAAC;QAC7B,IAAI,CAAC;YACH,OAAO,MAAM,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAChC,CAAC;QAAC,MAAM,CAAC;YACP,iEAAiE;YACjE,0DAA0D;YAC1D,OAAO,SAAS,CAAC;QACnB,CAAC;IACH,CAAC;IAED,KAAK,CAAC,GAAG,CAAC,GAAW,EAAE,QAAkB;QACvC,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,IAAI,EAAE,CAAC;QAChC,IAAI,CAAC,KAAK,EAAE,CAAC;YACX,IAAI,CAAC,QAAQ,CAAC,IAAI,8BAA8B,CAAC,8CAA8C,CAAC,CAAC,CAAC;YAClG,OAAO;QACT,CAAC;QACD,IAAI,CAAC;YACH,MAAM,KAAK,CAAC,GAAG,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC;QACjC,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;QACvB,CAAC;IACH,CAAC;IAEO,QAAQ,CAAC,KAAc;QAC7B,IAAI,IAAI,CAAC,gBAAgB;YAAE,OAAO;QAClC,IAAI,CAAC,gBAAgB,GAAG,IAAI,CAAC;QAC7B,IAAI,CAAC,aAAa,CAAC,KAAK,CAAC,CAAC;IAC5B,CAAC;IAED;;;;OAIG;IACK,IAAI;QACV,IAAI,CAAC,OAAO,KAAK,IAAI,CAAC,QAAQ,EAAE,CAAC;QACjC,OAAO,IAAI,CAAC,OAAO,CAAC;IACtB,CAAC;IAEO,KAAK,CAAC,QAAQ;QACpB,IAAI,OAAO,MAAM,KAAK,WAAW;YAAE,OAAO,IAAI,CAAC;QAC/C,IAAI,CAAC;YACH,OAAO,MAAM,MAAM,CAAC,IAAI,CAAC,0BAA0B,CAAC,UAAU,CAAC,CAAC;QAClE,CAAC;QAAC,MAAM,CAAC;YACP,8DAA8D;YAC9D,8DAA8D;YAC9D,OAAO,IAAI,CAAC;QACd,CAAC;IACH,CAAC"}

package/dist/modules/transcription/CueDocumentBuilder.d.ts

@@ -0,0 +1,40 @@
+import type { CueParseResult } from '../transcription/CueParseResult';
+import type { CueTimecodeReader } from '../transcription/CueTimecodeReader';
+import type { CueWordReader } from '../transcription/CueWordReader';
+/**
+ * Builds a Document from the cue blocks of a caption file: blank lines
+ * separate cues, a `-->` line states the range, and everything below it
+ * is the text. Each cue becomes one Segment holding one Line.
+ *
+ * A counter or identifier above the range line is ignored, since
+ * neither says anything the Document keeps. A cue whose text is empty
+ * is dropped. How the text becomes timed words is the reader's business
+ * and the only thing that differs between formats.
+ *
+ * A block carrying no range line is skipped and handed back in the
+ * result rather than ending the read. This is what the WebVTT parser
+ * algorithm prescribes, and it is the only behaviour that serves
+ * someone holding a file they did not write: one stray blank line
+ * inside a cue, or one paragraph of prose at the end, would otherwise
+ * cost them every cue in the file. Deciding whether the remainder is
+ * worth using belongs to the caller, which is why the count of what
+ * was lost travels with it.
+ *
+ * A file where every block was skipped is refused outright: what is
+ * left is not a thin transcript, it is no transcript, and returning
+ * one would say the file was silent rather than unreadable. A file
+ * with no blocks at all is not refused — there is nothing in it to
+ * misread.
+ *
+ * A malformed *timecode* still throws. That block announced its
+ * intent to state a time and got it wrong, so placing its text
+ * anywhere would be a guess.
+ */
+export declare class CueDocumentBuilder {
+    private readonly timecodeReader;
+    constructor(timecodeReader: CueTimecodeReader);
+    build(source: string, wordReader: CueWordReader): CueParseResult;
+    private splitIntoBlocks;
+    private buildSegment;
+}
+//# sourceMappingURL=CueDocumentBuilder.d.ts.map

package/dist/modules/transcription/CueDocumentBuilder.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CueDocumentBuilder.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/CueDocumentBuilder.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,uCAAuC,CAAC;AAC5E,OAAO,KAAK,EAAE,iBAAiB,EAAE,MAAM,0CAA0C,CAAC;AAClF,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,sCAAsC,CAAC;AAK1E;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AACH,qBAAa,kBAAkB;IACjB,OAAO,CAAC,QAAQ,CAAC,cAAc;gBAAd,cAAc,EAAE,iBAAiB;IAE9D,KAAK,CAAC,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,aAAa,GAAG,cAAc;IAgBhE,OAAO,CAAC,eAAe;IAMvB,OAAO,CAAC,YAAY;CAkBrB"}

package/dist/modules/transcription/CueDocumentBuilder.js

@@ -0,0 +1,77 @@
+import { Document } from '../document/Document';
+import { Line } from '../document/Line';
+import { Section } from '../document/Section';
+import { Segment } from '../document/Segment';
+import { SubtitleFileUnreadableError } from '../transcription/SubtitleFileUnreadableError';
+const BOM_RE = /^\uFEFF/;
+/**
+ * Builds a Document from the cue blocks of a caption file: blank lines
+ * separate cues, a `-->` line states the range, and everything below it
+ * is the text. Each cue becomes one Segment holding one Line.
+ *
+ * A counter or identifier above the range line is ignored, since
+ * neither says anything the Document keeps. A cue whose text is empty
+ * is dropped. How the text becomes timed words is the reader's business
+ * and the only thing that differs between formats.
+ *
+ * A block carrying no range line is skipped and handed back in the
+ * result rather than ending the read. This is what the WebVTT parser
+ * algorithm prescribes, and it is the only behaviour that serves
+ * someone holding a file they did not write: one stray blank line
+ * inside a cue, or one paragraph of prose at the end, would otherwise
+ * cost them every cue in the file. Deciding whether the remainder is
+ * worth using belongs to the caller, which is why the count of what
+ * was lost travels with it.
+ *
+ * A file where every block was skipped is refused outright: what is
+ * left is not a thin transcript, it is no transcript, and returning
+ * one would say the file was silent rather than unreadable. A file
+ * with no blocks at all is not refused — there is nothing in it to
+ * misread.
+ *
+ * A malformed *timecode* still throws. That block announced its
+ * intent to state a time and got it wrong, so placing its text
+ * anywhere would be a guess.
+ */
+export class CueDocumentBuilder {
+    timecodeReader;
+    constructor(timecodeReader) {
+        this.timecodeReader = timecodeReader;
+    }
+    build(source, wordReader) {
+        const skippedBlocks = [];
+        const segments = this.splitIntoBlocks(source)
+            .map((block) => this.buildSegment(block, wordReader, skippedBlocks))
+            .filter((segment) => segment !== null);
+        if (segments.length === 0 && skippedBlocks.length > 0) {
+            throw new SubtitleFileUnreadableError(`None of the ${skippedBlocks.length} block(s) in the file carried a timecode line`);
+        }
+        return {
+            document: new Document({ sections: [new Section({ segments, kind: '' })] }),
+            skippedBlocks,
+        };
+    }
+    splitIntoBlocks(source) {
+        const normalized = source.replace(BOM_RE, '').replace(/\r\n?/g, '\n').trim();
+        if (normalized.length === 0)
+            return [];
+        return normalized.split(/\n{2,}/);
+    }
+    buildSegment(block, wordReader, skippedBlocks) {
+        const lines = block.split('\n').map((line) => line.trim()).filter((line) => line.length > 0);
+        if (lines.length === 0)
+            return null;
+        const rangeAt = lines.findIndex((line) => this.timecodeReader.isRangeLine(line));
+        if (rangeAt === -1) {
+            skippedBlocks.push(block);
+            return null;
+        }
+        const range = this.timecodeReader.readRange(lines[rangeAt]);
+        const text = lines.slice(rangeAt + 1).join(' ');
+        const words = wordReader.read(text, range.startSeconds, range.endSeconds);
+        if (words.length === 0)
+            return null;
+        return new Segment({ lines: [new Line({ words })] });
+    }
+}
+//# sourceMappingURL=CueDocumentBuilder.js.map

package/dist/modules/transcription/CueDocumentBuilder.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CueDocumentBuilder.js","sourceRoot":"","sources":["../../../src/modules/transcription/CueDocumentBuilder.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,4BAA4B,CAAC;AACtD,OAAO,EAAE,IAAI,EAAE,MAAM,wBAAwB,CAAC;AAC9C,OAAO,EAAE,OAAO,EAAE,MAAM,2BAA2B,CAAC;AACpD,OAAO,EAAE,OAAO,EAAE,MAAM,2BAA2B,CAAC;AAIpD,OAAO,EAAE,2BAA2B,EAAE,MAAM,oDAAoD,CAAC;AAEjG,MAAM,MAAM,GAAG,SAAS,CAAC;AAEzB;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AACH,MAAM,OAAO,kBAAkB;IACA;IAA7B,YAA6B,cAAiC;QAAjC,mBAAc,GAAd,cAAc,CAAmB;IAAG,CAAC;IAElE,KAAK,CAAC,MAAc,EAAE,UAAyB;QAC7C,MAAM,aAAa,GAAa,EAAE,CAAC;QACnC,MAAM,QAAQ,GAAG,IAAI,CAAC,eAAe,CAAC,MAAM,CAAC;aAC1C,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,IAAI,CAAC,YAAY,CAAC,KAAK,EAAE,UAAU,EAAE,aAAa,CAAC,CAAC;aACnE,MAAM,CAAC,CAAC,OAAO,EAAsB,EAAE,CAAC,OAAO,KAAK,IAAI,CAAC,CAAC;QAC7D,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,IAAI,aAAa,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACtD,MAAM,IAAI,2BAA2B,CACnC,eAAe,aAAa,CAAC,MAAM,+CAA+C,CACnF,CAAC;QACJ,CAAC;QACD,OAAO;YACL,QAAQ,EAAE,IAAI,QAAQ,CAAC,EAAE,QAAQ,EAAE,CAAC,IAAI,OAAO,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC;YAC3E,aAAa;SACd,CAAC;IACJ,CAAC;IAEO,eAAe,CAAC,MAAc;QACpC,MAAM,UAAU,GAAG,MAAM,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;QAC7E,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,EAAE,CAAC;QACvC,OAAO,UAAU,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC;IACpC,CAAC;IAEO,YAAY,CAClB,KAAa,EACb,UAAyB,EACzB,aAAuB;QAEvB,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QAC7F,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,IAAI,CAAC;QACpC,MAAM,OAAO,GAAG,KAAK,CAAC,SAAS,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,cAAc,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC;QACjF,IAAI,OAAO,KAAK,CAAC,CAAC,EAAE,CAAC;YACnB,aAAa,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YAC1B,OAAO,IAAI,CAAC;QACd,CAAC;QACD,MAAM,KAAK,GAAG,IAAI,CAAC,cAAc,CAAC,SAAS,CAAC,KAAK,CAAC,OAAO,CAAE,CAAC,CAAC;QAC7D,MAAM,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC,OAAO,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QAChD,MAAM,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,IAAI,EAAE,KAAK,CAAC,YAAY,EAAE,KAAK,CAAC,UAAU,CAAC,CAAC;QAC1E,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,IAAI,CAAC;QACpC,OAAO,IAAI,OAAO,CAAC,EAAE,KAAK,EAAE,CAAC,IAAI,IAAI,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC;IACvD,CAAC;CACF"}

package/dist/modules/transcription/CueParseResult.d.ts

@@ -0,0 +1,16 @@
+import type { Document } from '../document/Document';
+/**
+ * What reading a caption file produced: the cues that could be read,
+ * and the blocks that could not.
+ *
+ * The two travel together because a caption file is rarely all right
+ * or all wrong. A reader that returned only the document would leave
+ * its caller unable to tell a file that was fully understood from one
+ * that was mostly understood.
+ */
+export interface CueParseResult {
+    readonly document: Document;
+    /** Verbatim text of every block that carried no timecode. */
+    readonly skippedBlocks: readonly string[];
+}
+//# sourceMappingURL=CueParseResult.d.ts.map

package/dist/modules/transcription/CueParseResult.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CueParseResult.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/CueParseResult.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,4BAA4B,CAAC;AAE3D;;;;;;;;GAQG;AACH,MAAM,WAAW,cAAc;IAC7B,QAAQ,CAAC,QAAQ,EAAE,QAAQ,CAAC;IAE5B,6DAA6D;IAC7D,QAAQ,CAAC,aAAa,EAAE,SAAS,MAAM,EAAE,CAAC;CAC3C"}

package/dist/modules/transcription/CueParseResult.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=CueParseResult.js.map

package/dist/modules/transcription/CueParseResult.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CueParseResult.js","sourceRoot":"","sources":["../../../src/modules/transcription/CueParseResult.ts"],"names":[],"mappings":""}

package/dist/modules/transcription/CueTextTokenizer.d.ts

@@ -0,0 +1,10 @@
+/**
+ * Reduces the text of one cue to the words it says.
+ *
+ * Markup is dropped, runs of whitespace collapse, and a cue that turns
+ * out to hold nothing but markup yields no words at all.
+ */
+export declare class CueTextTokenizer {
+    tokenize(text: string): ReadonlyArray<string>;
+}
+//# sourceMappingURL=CueTextTokenizer.d.ts.map

package/dist/modules/transcription/CueTextTokenizer.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CueTextTokenizer.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/CueTextTokenizer.ts"],"names":[],"mappings":"AAOA;;;;;GAKG;AACH,qBAAa,gBAAgB;IAE3B,QAAQ,CAAC,IAAI,EAAE,MAAM,GAAG,aAAa,CAAC,MAAM,CAAC;CAM9C"}

package/dist/modules/transcription/CueTextTokenizer.js

@@ -0,0 +1,21 @@
+/**
+ * Markup a caption file may wrap its text in: WebVTT's angle-bracket
+ * tags and the brace-delimited overrides SubRip files sometimes carry
+ * from SubStation. None of it survives into a word.
+ */
+const MARKUP = /<[^>]+>|\{[^}]+\}/g;
+/**
+ * Reduces the text of one cue to the words it says.
+ *
+ * Markup is dropped, runs of whitespace collapse, and a cue that turns
+ * out to hold nothing but markup yields no words at all.
+ */
+export class CueTextTokenizer {
+    tokenize(text) {
+        return text
+            .replace(MARKUP, '')
+            .split(/\s+/)
+            .filter((token) => token.length > 0);
+    }
+}
+//# sourceMappingURL=CueTextTokenizer.js.map

package/dist/modules/transcription/CueTextTokenizer.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CueTextTokenizer.js","sourceRoot":"","sources":["../../../src/modules/transcription/CueTextTokenizer.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AACH,MAAM,MAAM,GAAG,oBAAoB,CAAC;AAEpC;;;;;GAKG;AACH,MAAM,OAAO,gBAAgB;IAE3B,QAAQ,CAAC,IAAY;QACnB,OAAO,IAAI;aACR,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC;aACnB,KAAK,CAAC,KAAK,CAAC;aACZ,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IACzC,CAAC;CACF"}

package/dist/modules/transcription/CueTimecodeReader.d.ts

@@ -0,0 +1,19 @@
+/** Half-open range a cue occupies, in seconds. */
+export interface CueTimeRange {
+    readonly startSeconds: number;
+    readonly endSeconds: number;
+}
+/**
+ * Reads the timecodes a caption file states its timings with.
+ *
+ * Positions are returned in seconds. A malformed timecode throws rather
+ * than resolving to zero, so a file that cannot be trusted is refused
+ * instead of silently landing every caption at the start.
+ */
+export declare class CueTimecodeReader {
+    isRangeLine(line: string): boolean;
+    readRange(line: string): CueTimeRange;
+    readPosition(timecode: string): number;
+    private toSeconds;
+}
+//# sourceMappingURL=CueTimecodeReader.d.ts.map

package/dist/modules/transcription/CueTimecodeReader.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CueTimecodeReader.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/CueTimecodeReader.ts"],"names":[],"mappings":"AAAA,kDAAkD;AAClD,MAAM,WAAW,YAAY;IAC3B,QAAQ,CAAC,YAAY,EAAE,MAAM,CAAC;IAC9B,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;CAC7B;AAYD;;;;;;GAMG;AACH,qBAAa,iBAAiB;IAE5B,WAAW,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO;IAIlC,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,YAAY;IAWrC,YAAY,CAAC,QAAQ,EAAE,MAAM,GAAG,MAAM;IAMtC,OAAO,CAAC,SAAS;CAalB"}

package/dist/modules/transcription/CueTimecodeReader.js

@@ -0,0 +1,45 @@
+/**
+ * A position written `HH:MM:SS,mmm` or `HH:MM:SS.mmm`. SubRip settled
+ * on the comma and WebVTT on the dot, and both are accepted either way
+ * because files in the wild mix them. WebVTT also allows the hour to be
+ * left out, so it is optional.
+ */
+const POSITION = String.raw `(?:(\d{1,3}):)?(\d{2}):(\d{2})[,.](\d{1,3})`;
+const RANGE_LINE = new RegExp(`^${POSITION}\\s*-->\\s*${POSITION}`);
+const POSITION_ONLY = new RegExp(`^${POSITION}$`);
+/**
+ * Reads the timecodes a caption file states its timings with.
+ *
+ * Positions are returned in seconds. A malformed timecode throws rather
+ * than resolving to zero, so a file that cannot be trusted is refused
+ * instead of silently landing every caption at the start.
+ */
+export class CueTimecodeReader {
+    isRangeLine(line) {
+        return RANGE_LINE.test(line);
+    }
+    readRange(line) {
+        const match = RANGE_LINE.exec(line);
+        if (match === null)
+            throw new Error(`Malformed cue timecode: ${JSON.stringify(line)}`);
+        const startSeconds = this.toSeconds(match[1], match[2], match[3], match[4]);
+        const endSeconds = this.toSeconds(match[5], match[6], match[7], match[8]);
+        if (endSeconds < startSeconds) {
+            throw new Error(`Cue timecode ends before it starts: ${JSON.stringify(line)}`);
+        }
+        return { startSeconds, endSeconds };
+    }
+    readPosition(timecode) {
+        const match = POSITION_ONLY.exec(timecode);
+        if (match === null)
+            throw new Error(`Malformed timecode: ${JSON.stringify(timecode)}`);
+        return this.toSeconds(match[1], match[2], match[3], match[4]);
+    }
+    toSeconds(hours, minutes, seconds, milliseconds) {
+        return (Number(hours ?? 0) * 3600 +
+            Number(minutes) * 60 +
+            Number(seconds) +
+            Number(milliseconds.padEnd(3, '0')) / 1000);
+    }
+}
+//# sourceMappingURL=CueTimecodeReader.js.map

package/dist/modules/transcription/CueTimecodeReader.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CueTimecodeReader.js","sourceRoot":"","sources":["../../../src/modules/transcription/CueTimecodeReader.ts"],"names":[],"mappings":"AAMA;;;;;GAKG;AACH,MAAM,QAAQ,GAAG,MAAM,CAAC,GAAG,CAAA,6CAA6C,CAAC;AACzE,MAAM,UAAU,GAAG,IAAI,MAAM,CAAC,IAAI,QAAQ,cAAc,QAAQ,EAAE,CAAC,CAAC;AACpE,MAAM,aAAa,GAAG,IAAI,MAAM,CAAC,IAAI,QAAQ,GAAG,CAAC,CAAC;AAElD;;;;;;GAMG;AACH,MAAM,OAAO,iBAAiB;IAE5B,WAAW,CAAC,IAAY;QACtB,OAAO,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC/B,CAAC;IAED,SAAS,CAAC,IAAY;QACpB,MAAM,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACpC,IAAI,KAAK,KAAK,IAAI;YAAE,MAAM,IAAI,KAAK,CAAC,2BAA2B,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACvF,MAAM,YAAY,GAAG,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,CAAC,CAAC;QAC/E,MAAM,UAAU,GAAG,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,CAAC,CAAC;QAC7E,IAAI,UAAU,GAAG,YAAY,EAAE,CAAC;YAC9B,MAAM,IAAI,KAAK,CAAC,uCAAuC,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACjF,CAAC;QACD,OAAO,EAAE,YAAY,EAAE,UAAU,EAAE,CAAC;IACtC,CAAC;IAED,YAAY,CAAC,QAAgB;QAC3B,MAAM,KAAK,GAAG,aAAa,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QAC3C,IAAI,KAAK,KAAK,IAAI;YAAE,MAAM,IAAI,KAAK,CAAC,uBAAuB,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC;QACvF,OAAO,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,EAAE,KAAK,CAAC,CAAC,CAAE,CAAC,CAAC;IACnE,CAAC;IAEO,SAAS,CACf,KAAyB,EACzB,OAAe,EACf,OAAe,EACf,YAAoB;QAEpB,OAAO,CACL,MAAM,CAAC,KAAK,IAAI,CAAC,CAAC,GAAG,IAAI;YACzB,MAAM,CAAC,OAAO,CAAC,GAAG,EAAE;YACpB,MAAM,CAAC,OAAO,CAAC;YACf,MAAM,CAAC,YAAY,CAAC,MAAM,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,GAAG,IAAI,CAC3C,CAAC;IACJ,CAAC;CACF"}

package/dist/modules/transcription/CueWordReader.d.ts

@@ -0,0 +1,12 @@
+import type { Word } from '../document/Word';
+/**
+ * Turns the text of one cue into the words it holds, timed.
+ *
+ * Implementations differ in how much the file told them: some formats
+ * time nothing below the cue and leave the words to be guessed at,
+ * others mark when each word is spoken and can be taken at their word.
+ */
+export interface CueWordReader {
+    read(text: string, startSeconds: number, endSeconds: number): ReadonlyArray<Word>;
+}
+//# sourceMappingURL=CueWordReader.d.ts.map

package/dist/modules/transcription/CueWordReader.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CueWordReader.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/CueWordReader.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,wBAAwB,CAAC;AAEnD;;;;;;GAMG;AACH,MAAM,WAAW,aAAa;IAC5B,IAAI,CAAC,IAAI,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,GAAG,aAAa,CAAC,IAAI,CAAC,CAAC;CACnF"}

package/dist/modules/transcription/CueWordReader.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=CueWordReader.js.map

package/dist/modules/transcription/CueWordReader.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CueWordReader.js","sourceRoot":"","sources":["../../../src/modules/transcription/CueWordReader.ts"],"names":[],"mappings":""}

package/dist/modules/transcription/EstimatedCueWordReader.d.ts

@@ -0,0 +1,16 @@
+import type { Word } from '../document/Word';
+import type { CueTextTokenizer } from '../transcription/CueTextTokenizer';
+import type { CueWordReader } from '../transcription/CueWordReader';
+import type { WordTimingEstimator } from '../transcription/WordTimingEstimator';
+/**
+ * Reads a cue from a format that times nothing below the cue itself.
+ * The words are known, when each is spoken is not, so the cue's own
+ * window is shared out among them.
+ */
+export declare class EstimatedCueWordReader implements CueWordReader {
+    private readonly tokenizer;
+    private readonly estimator;
+    constructor(tokenizer: CueTextTokenizer, estimator: WordTimingEstimator);
+    read(text: string, startSeconds: number, endSeconds: number): ReadonlyArray<Word>;
+}
+//# sourceMappingURL=EstimatedCueWordReader.d.ts.map

package/dist/modules/transcription/EstimatedCueWordReader.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"EstimatedCueWordReader.d.ts","sourceRoot":"","sources":["../../../src/modules/transcription/EstimatedCueWordReader.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,wBAAwB,CAAC;AACnD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,yCAAyC,CAAC;AAChF,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,sCAAsC,CAAC;AAC1E,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,4CAA4C,CAAC;AAEtF;;;;GAIG;AACH,qBAAa,sBAAuB,YAAW,aAAa;IAExD,OAAO,CAAC,QAAQ,CAAC,SAAS;IAC1B,OAAO,CAAC,QAAQ,CAAC,SAAS;gBADT,SAAS,EAAE,gBAAgB,EAC3B,SAAS,EAAE,mBAAmB;IAGjD,IAAI,CAAC,IAAI,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,GAAG,aAAa,CAAC,IAAI,CAAC;CAGlF"}

package/dist/modules/transcription/EstimatedCueWordReader.js

@@ -0,0 +1,17 @@
+/**
+ * Reads a cue from a format that times nothing below the cue itself.
+ * The words are known, when each is spoken is not, so the cue's own
+ * window is shared out among them.
+ */
+export class EstimatedCueWordReader {
+    tokenizer;
+    estimator;
+    constructor(tokenizer, estimator) {
+        this.tokenizer = tokenizer;
+        this.estimator = estimator;
+    }
+    read(text, startSeconds, endSeconds) {
+        return this.estimator.spread(this.tokenizer.tokenize(text), startSeconds, endSeconds);
+    }
+}
+//# sourceMappingURL=EstimatedCueWordReader.js.map

package/dist/modules/transcription/EstimatedCueWordReader.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"EstimatedCueWordReader.js","sourceRoot":"","sources":["../../../src/modules/transcription/EstimatedCueWordReader.ts"],"names":[],"mappings":"AAKA;;;;GAIG;AACH,MAAM,OAAO,sBAAsB;IAEd;IACA;IAFnB,YACmB,SAA2B,EAC3B,SAA8B;QAD9B,cAAS,GAAT,SAAS,CAAkB;QAC3B,cAAS,GAAT,SAAS,CAAqB;IAC9C,CAAC;IAEJ,IAAI,CAAC,IAAY,EAAE,YAAoB,EAAE,UAAkB;QACzD,OAAO,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,YAAY,EAAE,UAAU,CAAC,CAAC;IACxF,CAAC;CACF"}