aivee 0.1.0-alpha.0 → 0.1.0-dev.d34a6ba

package/src/transcript/corrections.mjs

@@ -1,21 +1,21 @@
-import { validateTranscript } from './validate.mjs';
-
-export function applyTranscriptCorrections(transcript, corrections) {
-  validateTranscript(transcript);
-  if (!Array.isArray(corrections)) throw new TypeError('Transcript corrections must be an array.');
-  const words = transcript.words.map((word) => ({ ...word }));
-  const updated = new Set();
-  for (const [index, correction] of corrections.entries()) {
-    if (!correction || !Number.isSafeInteger(correction.wordIndex) || correction.wordIndex < 0 || correction.wordIndex >= words.length) throw new TypeError(`Correction ${index + 1} has an invalid wordIndex.`);
-    if (updated.has(correction.wordIndex)) throw new TypeError(`Word ${correction.wordIndex} has duplicate corrections.`);
-    updated.add(correction.wordIndex);
-    const word = words[correction.wordIndex];
-    if (correction.text !== undefined) {
-      if (typeof correction.text !== 'string' || !correction.text.trim()) throw new TypeError(`Correction ${index + 1} text must be a non-empty string.`);
-      word.text = correction.text.trim();
-    }
-    if (correction.start !== undefined) word.start = correction.start;
-    if (correction.end !== undefined) word.end = correction.end;
-  }
-  return validateTranscript({ ...transcript, words });
-}
+import { validateTranscript } from './validate.mjs';
+
+export function applyTranscriptCorrections(transcript, corrections) {
+  validateTranscript(transcript);
+  if (!Array.isArray(corrections)) throw new TypeError('Transcript corrections must be an array.');
+  const words = transcript.words.map((word) => ({ ...word }));
+  const updated = new Set();
+  for (const [index, correction] of corrections.entries()) {
+    if (!correction || !Number.isSafeInteger(correction.wordIndex) || correction.wordIndex < 0 || correction.wordIndex >= words.length) throw new TypeError(`Correction ${index + 1} has an invalid wordIndex.`);
+    if (updated.has(correction.wordIndex)) throw new TypeError(`Word ${correction.wordIndex} has duplicate corrections.`);
+    updated.add(correction.wordIndex);
+    const word = words[correction.wordIndex];
+    if (correction.text !== undefined) {
+      if (typeof correction.text !== 'string' || !correction.text.trim()) throw new TypeError(`Correction ${index + 1} text must be a non-empty string.`);
+      word.text = correction.text.trim();
+    }
+    if (correction.start !== undefined) word.start = correction.start;
+    if (correction.end !== undefined) word.end = correction.end;
+  }
+  return validateTranscript({ ...transcript, words });
+}

package/src/transcript/normalize.mjs

@@ -1,17 +1,17 @@
-import { validateTranscript } from './validate.mjs';
-
-export function normalizeTranscript(input, durationSeconds, granularity = 'word') {
-  const sourceWords = Array.isArray(input) ? input : input?.words;
-  if (!Array.isArray(sourceWords)) throw new TypeError('Transcript input must contain a words array.');
-  const words = sourceWords
-    .filter((word) => word?.type === undefined || word.type === 'word')
-    .map((word) => ({
-      text: String(word.text ?? word.word ?? '').trim(),
-      start: Number(word.start),
-      end: Number(word.end),
-    }))
-    .filter((word) => word.text);
-  const transcript = { schemaVersion: 1, durationSeconds, granularity, words };
-  validateTranscript(transcript);
-  return transcript;
-}
+import { validateTranscript } from './validate.mjs';
+
+export function normalizeTranscript(input, durationSeconds, granularity = 'word') {
+  const sourceWords = Array.isArray(input) ? input : input?.words;
+  if (!Array.isArray(sourceWords)) throw new TypeError('Transcript input must contain a words array.');
+  const words = sourceWords
+    .filter((word) => word?.type === undefined || word.type === 'word')
+    .map((word) => ({
+      text: String(word.text ?? word.word ?? '').trim(),
+      start: Number(word.start),
+      end: Number(word.end),
+    }))
+    .filter((word) => word.text);
+  const transcript = { schemaVersion: 1, durationSeconds, granularity, words };
+  validateTranscript(transcript);
+  return transcript;
+}

package/src/transcript/provider.mjs

@@ -1,42 +1,42 @@
-import { existsSync, openSync, closeSync, unlinkSync, writeFileSync } from 'node:fs';
-import { join, resolve } from 'node:path';
-import { probeMedia } from '../media/probe.mjs';
-import { normalizeTranscript } from './normalize.mjs';
-
-/** Run an injected transcription provider and persist its normalized result. */
-export async function transcribeWithProvider(projectPath, mediaPath, provider, options = {}) {
-  const project = resolve(projectPath);
-  const media = resolve(mediaPath);
-  if (!existsSync(join(project, 'meta.json'))) throw new Error(`Not a video project: ${project}`);
-  if (!provider || typeof provider !== 'object'
-      || !/^[a-z][a-z0-9-]*$/.test(provider.id ?? '')
-      || typeof provider.transcribe !== 'function') {
-    throw new TypeError('Provider must define a kebab-case id and async transcribe function.');
-  }
-  const granularity = options.granularity ?? 'word';
-  if (!['word', 'cue'].includes(granularity)) throw new TypeError('Provider granularity must be word or cue.');
-
-  const mediaInfo = probeMedia(media, { requireVideo: false });
-  const raw = await provider.transcribe({
-    mediaPath: media,
-    durationSeconds: mediaInfo.durationSeconds,
-    options: structuredClone(options.providerOptions ?? {}),
-  });
-  const transcript = normalizeTranscript(raw, mediaInfo.durationSeconds, granularity);
-  const transcriptPath = join(project, 'transcripts', 'transcript.json');
-  const fd = openSync(transcriptPath, 'wx');
-  try {
-    writeFileSync(fd, `${JSON.stringify(transcript, null, 2)}\n`);
-  } catch (error) {
-    unlinkSync(transcriptPath);
-    throw error;
-  } finally {
-    closeSync(fd);
-  }
-  return {
-    providerId: provider.id,
-    transcriptPath,
-    wordCount: transcript.words.length,
-    durationSeconds: transcript.durationSeconds,
-  };
-}
+import { existsSync, openSync, closeSync, unlinkSync, writeFileSync } from 'node:fs';
+import { join, resolve } from 'node:path';
+import { probeMedia } from '../media/probe.mjs';
+import { normalizeTranscript } from './normalize.mjs';
+
+/** Run an injected transcription provider and persist its normalized result. */
+export async function transcribeWithProvider(projectPath, mediaPath, provider, options = {}) {
+  const project = resolve(projectPath);
+  const media = resolve(mediaPath);
+  if (!existsSync(join(project, 'meta.json'))) throw new Error(`Not a video project: ${project}`);
+  if (!provider || typeof provider !== 'object'
+      || !/^[a-z][a-z0-9-]*$/.test(provider.id ?? '')
+      || typeof provider.transcribe !== 'function') {
+    throw new TypeError('Provider must define a kebab-case id and async transcribe function.');
+  }
+  const granularity = options.granularity ?? 'word';
+  if (!['word', 'cue'].includes(granularity)) throw new TypeError('Provider granularity must be word or cue.');
+
+  const mediaInfo = probeMedia(media, { requireVideo: false });
+  const raw = await provider.transcribe({
+    mediaPath: media,
+    durationSeconds: mediaInfo.durationSeconds,
+    options: structuredClone(options.providerOptions ?? {}),
+  });
+  const transcript = normalizeTranscript(raw, mediaInfo.durationSeconds, granularity);
+  const transcriptPath = join(project, 'transcripts', 'transcript.json');
+  const fd = openSync(transcriptPath, 'wx');
+  try {
+    writeFileSync(fd, `${JSON.stringify(transcript, null, 2)}\n`);
+  } catch (error) {
+    unlinkSync(transcriptPath);
+    throw error;
+  } finally {
+    closeSync(fd);
+  }
+  return {
+    providerId: provider.id,
+    transcriptPath,
+    wordCount: transcript.words.length,
+    durationSeconds: transcript.durationSeconds,
+  };
+}

package/src/transcript/subtitles.mjs

@@ -1,57 +1,57 @@
-import { validateTranscript } from './validate.mjs';
-
-export function parseSubtitleText(contents, format) {
-  if (typeof contents !== 'string') throw new TypeError('Subtitle contents must be text.');
-  if (!['srt', 'vtt'].includes(format)) throw new TypeError('Subtitle format must be srt or vtt.');
-  const text = contents.replace(/^\uFEFF/, '').replaceAll('\r\n', '\n').replaceAll('\r', '\n');
-  const words = [];
-  for (const block of text.split(/\n\s*\n/)) {
-    const lines = block.split('\n');
-    const timeIndex = lines.findIndex((line) => line.includes('-->'));
-    if (timeIndex < 0) continue;
-    const match = lines[timeIndex].match(/^\s*((?:\d{2}:)?\d{2}:\d{2}[,.]\d{3})\s+-->\s+((?:\d{2}:)?\d{2}:\d{2}[,.]\d{3})(?:\s+.*)?$/);
-    if (!match) throw new TypeError(`Invalid ${format.toUpperCase()} timestamp: ${lines[timeIndex]}`);
-    const start = parseTimestamp(match[1]);
-    const end = parseTimestamp(match[2]);
-    const cueText = lines.slice(timeIndex + 1).join(' ')
-      .replace(/<[^>]*>/g, '')
-      .replace(/&amp;/gi, '&').replace(/&lt;/gi, '<').replace(/&gt;/gi, '>').replace(/&nbsp;/gi, ' ')
-      .replace(/\s+/g, ' ').trim();
-    if (cueText) words.push({ text: cueText, start, end });
-  }
-  if (!words.length) throw new TypeError(`No subtitle cues found in ${format.toUpperCase()} input.`);
-  return { words };
-}
-
-export function formatSubtitleText(transcript, format) {
-  validateTranscript(transcript);
-  if (!['srt', 'vtt'].includes(format)) throw new TypeError('Subtitle format must be srt or vtt.');
-  const cues = transcript.words.map((word, index) => [
-    ...(format === 'srt' ? [String(index + 1)] : []),
-    `${formatTimestamp(word.start, format)} --> ${formatTimestamp(word.end, format)}`,
-    word.text.replace(/\s*\n\s*/g, ' '),
-  ].join('\n'));
-  return `${format === 'vtt' ? 'WEBVTT\n\n' : ''}${cues.join('\n\n')}\n`;
-}
-
-function parseTimestamp(value) {
-  const normalized = value.replace(',', '.');
-  const parts = normalized.split(':');
-  const seconds = Number(parts.pop());
-  const minutes = Number(parts.pop());
-  const hours = parts.length ? Number(parts.pop()) : 0;
-  if (![hours, minutes, seconds].every(Number.isFinite) || minutes >= 60 || seconds >= 60) {
-    throw new TypeError(`Invalid subtitle time: ${value}`);
-  }
-  return hours * 3600 + minutes * 60 + seconds;
-}
-
-function formatTimestamp(seconds, format) {
-  const milliseconds = Math.round(seconds * 1000);
-  const hours = Math.floor(milliseconds / 3_600_000);
-  const minutes = Math.floor((milliseconds % 3_600_000) / 60_000);
-  const wholeSeconds = Math.floor((milliseconds % 60_000) / 1000);
-  const fraction = milliseconds % 1000;
-  const time = `${String(hours).padStart(2, '0')}:${String(minutes).padStart(2, '0')}:${String(wholeSeconds).padStart(2, '0')}`;
-  return `${time}${format === 'srt' ? ',' : '.'}${String(fraction).padStart(3, '0')}`;
-}
+import { validateTranscript } from './validate.mjs';
+
+export function parseSubtitleText(contents, format) {
+  if (typeof contents !== 'string') throw new TypeError('Subtitle contents must be text.');
+  if (!['srt', 'vtt'].includes(format)) throw new TypeError('Subtitle format must be srt or vtt.');
+  const text = contents.replace(/^\uFEFF/, '').replaceAll('\r\n', '\n').replaceAll('\r', '\n');
+  const words = [];
+  for (const block of text.split(/\n\s*\n/)) {
+    const lines = block.split('\n');
+    const timeIndex = lines.findIndex((line) => line.includes('-->'));
+    if (timeIndex < 0) continue;
+    const match = lines[timeIndex].match(/^\s*((?:\d{2}:)?\d{2}:\d{2}[,.]\d{3})\s+-->\s+((?:\d{2}:)?\d{2}:\d{2}[,.]\d{3})(?:\s+.*)?$/);
+    if (!match) throw new TypeError(`Invalid ${format.toUpperCase()} timestamp: ${lines[timeIndex]}`);
+    const start = parseTimestamp(match[1]);
+    const end = parseTimestamp(match[2]);
+    const cueText = lines.slice(timeIndex + 1).join(' ')
+      .replace(/<[^>]*>/g, '')
+      .replace(/&amp;/gi, '&').replace(/&lt;/gi, '<').replace(/&gt;/gi, '>').replace(/&nbsp;/gi, ' ')
+      .replace(/\s+/g, ' ').trim();
+    if (cueText) words.push({ text: cueText, start, end });
+  }
+  if (!words.length) throw new TypeError(`No subtitle cues found in ${format.toUpperCase()} input.`);
+  return { words };
+}
+
+export function formatSubtitleText(transcript, format) {
+  validateTranscript(transcript);
+  if (!['srt', 'vtt'].includes(format)) throw new TypeError('Subtitle format must be srt or vtt.');
+  const cues = transcript.words.map((word, index) => [
+    ...(format === 'srt' ? [String(index + 1)] : []),
+    `${formatTimestamp(word.start, format)} --> ${formatTimestamp(word.end, format)}`,
+    word.text.replace(/\s*\n\s*/g, ' '),
+  ].join('\n'));
+  return `${format === 'vtt' ? 'WEBVTT\n\n' : ''}${cues.join('\n\n')}\n`;
+}
+
+function parseTimestamp(value) {
+  const normalized = value.replace(',', '.');
+  const parts = normalized.split(':');
+  const seconds = Number(parts.pop());
+  const minutes = Number(parts.pop());
+  const hours = parts.length ? Number(parts.pop()) : 0;
+  if (![hours, minutes, seconds].every(Number.isFinite) || minutes >= 60 || seconds >= 60) {
+    throw new TypeError(`Invalid subtitle time: ${value}`);
+  }
+  return hours * 3600 + minutes * 60 + seconds;
+}
+
+function formatTimestamp(seconds, format) {
+  const milliseconds = Math.round(seconds * 1000);
+  const hours = Math.floor(milliseconds / 3_600_000);
+  const minutes = Math.floor((milliseconds % 3_600_000) / 60_000);
+  const wholeSeconds = Math.floor((milliseconds % 60_000) / 1000);
+  const fraction = milliseconds % 1000;
+  const time = `${String(hours).padStart(2, '0')}:${String(minutes).padStart(2, '0')}:${String(wholeSeconds).padStart(2, '0')}`;
+  return `${time}${format === 'srt' ? ',' : '.'}${String(fraction).padStart(3, '0')}`;
+}

package/src/transcript/validate.mjs

@@ -1,39 +1,39 @@
-export function validateTranscript(transcript) {
-  if (!transcript || typeof transcript !== 'object' || Array.isArray(transcript)) {
-    throw new TypeError('Transcript must be a JSON object.');
-  }
-  if (transcript.schemaVersion !== 1) {
-    throw new TypeError('Transcript schemaVersion must be 1.');
-  }
-  if (!Number.isFinite(transcript.durationSeconds) || transcript.durationSeconds <= 0) {
-    throw new TypeError('Transcript durationSeconds must be a positive number.');
-  }
-  if (!Array.isArray(transcript.words)) {
-    throw new TypeError('Transcript words must be an array.');
-  }
-  if (transcript.granularity !== undefined && !['word', 'cue'].includes(transcript.granularity)) {
-    throw new TypeError('Transcript granularity must be word or cue.');
-  }
-
-  let previousStart = -Infinity;
-  for (const [index, word] of transcript.words.entries()) {
-    const label = `Transcript word ${index + 1}`;
-    if (!word || typeof word !== 'object' || Array.isArray(word)) {
-      throw new TypeError(`${label} must be an object.`);
-    }
-    if (typeof word.text !== 'string' || !word.text.trim()) {
-      throw new TypeError(`${label} text must be a non-empty string.`);
-    }
-    if (!Number.isFinite(word.start) || !Number.isFinite(word.end) || word.start < 0 || word.end <= word.start) {
-      throw new TypeError(`${label} must have finite times with 0 <= start < end.`);
-    }
-    if (word.end > transcript.durationSeconds) {
-      throw new TypeError(`${label} ends after the source duration.`);
-    }
-    if (word.start < previousStart) {
-      throw new TypeError(`${label} starts before the previous word.`);
-    }
-    previousStart = word.start;
-  }
-  return transcript;
-}
+export function validateTranscript(transcript) {
+  if (!transcript || typeof transcript !== 'object' || Array.isArray(transcript)) {
+    throw new TypeError('Transcript must be a JSON object.');
+  }
+  if (transcript.schemaVersion !== 1) {
+    throw new TypeError('Transcript schemaVersion must be 1.');
+  }
+  if (!Number.isFinite(transcript.durationSeconds) || transcript.durationSeconds <= 0) {
+    throw new TypeError('Transcript durationSeconds must be a positive number.');
+  }
+  if (!Array.isArray(transcript.words)) {
+    throw new TypeError('Transcript words must be an array.');
+  }
+  if (transcript.granularity !== undefined && !['word', 'cue'].includes(transcript.granularity)) {
+    throw new TypeError('Transcript granularity must be word or cue.');
+  }
+
+  let previousStart = -Infinity;
+  for (const [index, word] of transcript.words.entries()) {
+    const label = `Transcript word ${index + 1}`;
+    if (!word || typeof word !== 'object' || Array.isArray(word)) {
+      throw new TypeError(`${label} must be an object.`);
+    }
+    if (typeof word.text !== 'string' || !word.text.trim()) {
+      throw new TypeError(`${label} text must be a non-empty string.`);
+    }
+    if (!Number.isFinite(word.start) || !Number.isFinite(word.end) || word.start < 0 || word.end <= word.start) {
+      throw new TypeError(`${label} must have finite times with 0 <= start < end.`);
+    }
+    if (word.end > transcript.durationSeconds) {
+      throw new TypeError(`${label} ends after the source duration.`);
+    }
+    if (word.start < previousStart) {
+      throw new TypeError(`${label} starts before the previous word.`);
+    }
+    previousStart = word.start;
+  }
+  return transcript;
+}