scribe-cms 0.0.12 → 0.0.14

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (55) hide show
  1. package/README.md +21 -6
  2. package/dist/cli/index.cjs +1207 -204
  3. package/dist/cli/index.cjs.map +1 -1
  4. package/dist/cli/index.js +1208 -205
  5. package/dist/cli/index.js.map +1 -1
  6. package/dist/cli/prompt-translate.d.ts +3 -0
  7. package/dist/cli/prompt-translate.d.ts.map +1 -1
  8. package/dist/cli/translate-progress.d.ts.map +1 -1
  9. package/dist/index.cjs +969 -166
  10. package/dist/index.cjs.map +1 -1
  11. package/dist/index.js +970 -168
  12. package/dist/index.js.map +1 -1
  13. package/dist/runtime.cjs +136 -62
  14. package/dist/runtime.cjs.map +1 -1
  15. package/dist/runtime.js +136 -62
  16. package/dist/runtime.js.map +1 -1
  17. package/dist/src/config/resolve-config.d.ts.map +1 -1
  18. package/dist/src/core/introspect-schema.d.ts +7 -1
  19. package/dist/src/core/introspect-schema.d.ts.map +1 -1
  20. package/dist/src/core/types.d.ts +6 -0
  21. package/dist/src/core/types.d.ts.map +1 -1
  22. package/dist/src/create-project.d.ts.map +1 -1
  23. package/dist/src/i18n/resolve-document.d.ts +1 -1
  24. package/dist/src/i18n/resolve-document.d.ts.map +1 -1
  25. package/dist/src/index.d.ts +3 -3
  26. package/dist/src/index.d.ts.map +1 -1
  27. package/dist/src/storage/batch-jobs.d.ts +53 -0
  28. package/dist/src/storage/batch-jobs.d.ts.map +1 -0
  29. package/dist/src/storage/sqlite.d.ts.map +1 -1
  30. package/dist/src/translate/batch-worklist.d.ts +87 -0
  31. package/dist/src/translate/batch-worklist.d.ts.map +1 -0
  32. package/dist/src/translate/gemini-batch.d.ts +26 -0
  33. package/dist/src/translate/gemini-batch.d.ts.map +1 -0
  34. package/dist/src/translate/gemini-client.d.ts +18 -0
  35. package/dist/src/translate/gemini-client.d.ts.map +1 -1
  36. package/dist/src/translate/gemini-models.d.ts +9 -0
  37. package/dist/src/translate/gemini-models.d.ts.map +1 -1
  38. package/dist/src/translate/gemini-pricing.d.ts +2 -1
  39. package/dist/src/translate/gemini-pricing.d.ts.map +1 -1
  40. package/dist/src/translate/page-translator.d.ts +25 -52
  41. package/dist/src/translate/page-translator.d.ts.map +1 -1
  42. package/dist/src/translate/prompts/translation-prompt.d.ts +6 -0
  43. package/dist/src/translate/prompts/translation-prompt.d.ts.map +1 -1
  44. package/dist/src/translate/response-schema.d.ts +8 -1
  45. package/dist/src/translate/response-schema.d.ts.map +1 -1
  46. package/dist/src/translate/retry.d.ts +16 -0
  47. package/dist/src/translate/retry.d.ts.map +1 -0
  48. package/dist/src/translate/sanitize-mdx-jsx.d.ts.map +1 -1
  49. package/dist/src/translate/translate-core.d.ts +155 -0
  50. package/dist/src/translate/translate-core.d.ts.map +1 -0
  51. package/dist/studio/server.cjs +84 -47
  52. package/dist/studio/server.cjs.map +1 -1
  53. package/dist/studio/server.js +84 -47
  54. package/dist/studio/server.js.map +1 -1
  55. package/package.json +1 -1
package/dist/cli/index.js CHANGED
@@ -10,7 +10,7 @@ import remarkParse from 'remark-parse';
10
10
  import { unified } from 'unified';
11
11
  import { createJiti } from 'jiti';
12
12
  import { createHash } from 'crypto';
13
- import { GoogleGenAI } from '@google/genai';
13
+ import { GoogleGenAI, ThinkingLevel, ApiError } from '@google/genai';
14
14
  import dotenv from 'dotenv';
15
15
  import { serve } from '@hono/node-server';
16
16
  import { Hono } from 'hono';
@@ -59,6 +59,8 @@ function addColumnIfMissing(db, table, column, ddlType) {
59
59
  }
60
60
  }
61
61
  function readSchemaVersion(db) {
62
+ const metaTable = db.prepare(`SELECT name FROM sqlite_master WHERE type = 'table' AND name = 'meta'`).get();
63
+ if (!metaTable) return 0;
62
64
  const row = db.prepare(`SELECT value FROM meta WHERE key = 'schema_version'`).get();
63
65
  return row ? Number.parseInt(row.value, 10) || 0 : 0;
64
66
  }
@@ -83,7 +85,7 @@ var SCHEMA_VERSION, MIGRATIONS;
83
85
  var init_sqlite = __esm({
84
86
  "src/storage/sqlite.ts"() {
85
87
  init_esm_shims();
86
- SCHEMA_VERSION = 4;
88
+ SCHEMA_VERSION = 5;
87
89
  MIGRATIONS = [
88
90
  `CREATE TABLE IF NOT EXISTS meta (
89
91
  key TEXT PRIMARY KEY,
@@ -131,7 +133,30 @@ var init_sqlite = __esm({
131
133
  UNIQUE (content_type, en_slug, en_hash)
132
134
  )`,
133
135
  `CREATE INDEX IF NOT EXISTS idx_en_snapshots_lookup
134
- ON en_snapshots(content_type, en_slug, created_at DESC)`
136
+ ON en_snapshots(content_type, en_slug, created_at DESC)`,
137
+ `CREATE TABLE IF NOT EXISTS translation_batch_jobs (
138
+ id INTEGER PRIMARY KEY AUTOINCREMENT,
139
+ job_name TEXT NOT NULL UNIQUE,
140
+ model TEXT NOT NULL,
141
+ display_model TEXT NOT NULL,
142
+ created_at TEXT NOT NULL,
143
+ state TEXT NOT NULL,
144
+ completed_at TEXT
145
+ )`,
146
+ `CREATE TABLE IF NOT EXISTS translation_batch_items (
147
+ job_id INTEGER NOT NULL,
148
+ request_index INTEGER NOT NULL,
149
+ content_type TEXT NOT NULL,
150
+ en_slug TEXT NOT NULL,
151
+ locale TEXT NOT NULL,
152
+ en_hash TEXT NOT NULL,
153
+ snapshot_id INTEGER NOT NULL,
154
+ status TEXT NOT NULL,
155
+ error TEXT,
156
+ PRIMARY KEY (job_id, request_index)
157
+ )`,
158
+ `CREATE INDEX IF NOT EXISTS idx_batch_items_status
159
+ ON translation_batch_items(status)`
135
160
  ];
136
161
  }
137
162
  });
@@ -173,26 +198,6 @@ function getRelationTarget(schema) {
173
198
  }
174
199
  return null;
175
200
  }
176
- function unwrapSchema(schema) {
177
- let current = schema;
178
- for (let i = 0; i < 8; i++) {
179
- const anySchema = current;
180
- if (typeof anySchema.unwrap === "function") {
181
- current = anySchema.unwrap();
182
- continue;
183
- }
184
- if (typeof anySchema.removeDefault === "function") {
185
- current = anySchema.removeDefault();
186
- continue;
187
- }
188
- if (anySchema._def?.innerType) {
189
- current = anySchema._def.innerType;
190
- continue;
191
- }
192
- break;
193
- }
194
- return current;
195
- }
196
201
  function peelOptionalWrappers(schema) {
197
202
  let current = schema;
198
203
  for (let i = 0; i < 8; i++) {
@@ -338,6 +343,7 @@ function resolveConfig(input, baseDir) {
338
343
  if (localeRouting.strategy === "search-param" && !localeRouting.param) {
339
344
  throw new Error('scribe config: localeRouting search-param requires a "param" name');
340
345
  }
346
+ const localeFallbacks = raw.localeFallbacks === false ? {} : deriveLocaleFallbacks(raw.locales, defaultLocale);
341
347
  const projectRoot = path4.resolve(baseDir ?? process.cwd(), raw.rootDir);
342
348
  const contentRoot = path4.resolve(projectRoot, raw.contentDir ?? "content");
343
349
  const storePath = path4.resolve(projectRoot, raw.store ?? ".scribe/store.sqlite");
@@ -367,6 +373,7 @@ function resolveConfig(input, baseDir) {
367
373
  defaultLocale,
368
374
  localeRouting,
369
375
  localePresets: raw.localePresets,
376
+ localeFallbacks,
370
377
  translate: raw.translate,
371
378
  types
372
379
  };
@@ -376,12 +383,37 @@ function resolveConfig(input, baseDir) {
376
383
  });
377
384
  return config;
378
385
  }
386
+ function deriveLocaleFallbacks(locales, defaultLocale) {
387
+ const byLowercase = new Map(locales.map((l) => [l.toLowerCase(), l]));
388
+ const fallbacks = {};
389
+ for (const locale of locales) {
390
+ const subtags = locale.split("-");
391
+ if (subtags.length < 2) continue;
392
+ const chain = [];
393
+ for (let end = subtags.length - 1; end >= 1; end--) {
394
+ const prefix = byLowercase.get(subtags.slice(0, end).join("-").toLowerCase());
395
+ if (prefix && prefix !== locale && prefix !== defaultLocale) {
396
+ chain.push(prefix);
397
+ }
398
+ }
399
+ if (chain.length > 0) {
400
+ fallbacks[locale] = chain;
401
+ }
402
+ }
403
+ return fallbacks;
404
+ }
379
405
 
380
406
  // src/create-project.ts
381
407
  init_esm_shims();
382
408
 
383
409
  // src/core/introspect-schema.ts
384
410
  init_esm_shims();
411
+ function getArrayElement(schema) {
412
+ if (schema instanceof Object && "element" in schema && schema._def?.type === "array") {
413
+ return schema.element;
414
+ }
415
+ return null;
416
+ }
385
417
  function introspectSchema(schema, prefix = []) {
386
418
  const relation = getRelationTarget(schema);
387
419
  if (relation && prefix.length > 0) {
@@ -395,7 +427,10 @@ function introspectSchema(schema, prefix = []) {
395
427
  }
396
428
  ];
397
429
  }
398
- const base = unwrapSchema(schema);
430
+ if (prefix.length > 0 && getFieldKind(schema) === "translatable") {
431
+ return [{ path: prefix, kind: "translatable" }];
432
+ }
433
+ const base = peelOptionalWrappers(schema);
399
434
  if (base instanceof Object && "shape" in base) {
400
435
  const shape = base.shape;
401
436
  const fields = [];
@@ -404,9 +439,9 @@ function introspectSchema(schema, prefix = []) {
404
439
  }
405
440
  return fields;
406
441
  }
407
- if (base instanceof Object && "element" in base) {
408
- const element = base.element;
409
- const elementBase = unwrapSchema(element);
442
+ const element = getArrayElement(base);
443
+ if (element) {
444
+ const elementBase = peelOptionalWrappers(element);
410
445
  if (elementBase instanceof Object && "shape" in elementBase) {
411
446
  const shape = elementBase.shape;
412
447
  const fields = [];
@@ -418,15 +453,48 @@ function introspectSchema(schema, prefix = []) {
418
453
  }
419
454
  return [{ path: prefix, kind: getFieldKind(schema) }];
420
455
  }
421
- function extractByPaths(data, paths) {
422
- const out = {};
456
+ function buildPathTrie(paths) {
457
+ const root = { leaf: false, children: /* @__PURE__ */ new Map() };
423
458
  for (const path16 of paths) {
424
- const value = getAtPath(data, path16);
459
+ let node = root;
460
+ for (const segment of path16) {
461
+ let child = node.children.get(segment);
462
+ if (!child) {
463
+ child = { leaf: false, children: /* @__PURE__ */ new Map() };
464
+ node.children.set(segment, child);
465
+ }
466
+ node = child;
467
+ }
468
+ node.leaf = true;
469
+ }
470
+ return root;
471
+ }
472
+ function extractNode(data, trie) {
473
+ if (trie.leaf) return data;
474
+ const star = trie.children.get("*");
475
+ if (star) {
476
+ if (!Array.isArray(data)) return void 0;
477
+ return data.map(
478
+ (item) => typeof item === "object" && item !== null ? extractNode(item, star) ?? {} : {}
479
+ );
480
+ }
481
+ if (typeof data !== "object" || data === null || Array.isArray(data)) return void 0;
482
+ const record = data;
483
+ const out = {};
484
+ let any = false;
485
+ for (const [key, child] of trie.children) {
486
+ const value = extractNode(record[key], child);
425
487
  if (value !== void 0) {
426
- setAtPath(out, path16.filter((p) => p !== "*"), value);
488
+ out[key] = value;
489
+ any = true;
427
490
  }
428
491
  }
429
- return out;
492
+ return any ? out : void 0;
493
+ }
494
+ function extractByPaths(data, paths) {
495
+ if (paths.length === 0) return {};
496
+ const extracted = extractNode(data, buildPathTrie(paths));
497
+ return extracted && typeof extracted === "object" && !Array.isArray(extracted) ? extracted : {};
430
498
  }
431
499
  function pickTranslatable(data, schema) {
432
500
  const translatablePaths = introspectSchema(schema).filter((f) => f.kind === "translatable").map((f) => f.path);
@@ -441,32 +509,6 @@ function mergeStructuralOntoLocale(localeData, enData, schema) {
441
509
  const translatable = pickTranslatable(localeData, schema);
442
510
  return deepMerge(structural, translatable);
443
511
  }
444
- function getAtPath(obj, path16) {
445
- let current = obj;
446
- for (const segment of path16) {
447
- if (segment === "*") {
448
- if (!Array.isArray(current)) return void 0;
449
- return current.map(
450
- (item) => typeof item === "object" && item !== null ? getAtPath(item, path16.slice(path16.indexOf("*") + 1)) : void 0
451
- );
452
- }
453
- if (typeof current !== "object" || current === null || Array.isArray(current)) return void 0;
454
- current = current[segment];
455
- }
456
- return current;
457
- }
458
- function setAtPath(obj, path16, value) {
459
- if (path16.length === 0) return;
460
- if (path16.length === 1) {
461
- obj[path16[0]] = value;
462
- return;
463
- }
464
- const [head, ...rest] = path16;
465
- if (!(head in obj) || typeof obj[head] !== "object" || obj[head] === null) {
466
- obj[head] = {};
467
- }
468
- setAtPath(obj[head], rest, value);
469
- }
470
512
  function deepMerge(base, overlay) {
471
513
  const out = { ...base };
472
514
  for (const [key, value] of Object.entries(overlay)) {
@@ -757,7 +799,8 @@ function sanitizeMdxJsxAttributeQuotes(body) {
757
799
  if (i >= body.length || body[i] === ">" || body[i] === "/") break;
758
800
  const attrStart = i;
759
801
  while (i < body.length && /[\w:.-]/.test(body[i] ?? "")) i += 1;
760
- body.slice(attrStart, i);
802
+ const attrName = body.slice(attrStart, i);
803
+ if (!attrName) break;
761
804
  while (i < body.length && /\s/.test(body[i] ?? "")) i += 1;
762
805
  if (body[i] !== "=") {
763
806
  tagOut += body.slice(attrStart, i);
@@ -766,10 +809,12 @@ function sanitizeMdxJsxAttributeQuotes(body) {
766
809
  tagOut += body.slice(attrStart, i);
767
810
  tagOut += "=";
768
811
  i += 1;
812
+ const wsStart = i;
769
813
  while (i < body.length && /\s/.test(body[i] ?? "")) i += 1;
814
+ tagOut += body.slice(wsStart, i);
770
815
  const quote = body[i];
771
816
  if (quote !== '"' && quote !== "'") {
772
- continue;
817
+ break;
773
818
  }
774
819
  if (quote === "'") {
775
820
  const valStart2 = i + 1;
@@ -1091,7 +1136,7 @@ function getTranslatablePayload(doc, type) {
1091
1136
 
1092
1137
  // src/i18n/resolve-document.ts
1093
1138
  init_esm_shims();
1094
- function resolveLocalizedDocument(slug, locale, defaultLocale, allDocs, type, localeRouting) {
1139
+ function resolveLocalizedDocument(slug, locale, defaultLocale, allDocs, type, localeRouting, fallbackLocales = []) {
1095
1140
  const urlBuilder = createUrlBuilder({
1096
1141
  locales: [defaultLocale, locale],
1097
1142
  defaultLocale,
@@ -1105,15 +1150,28 @@ function resolveLocalizedDocument(slug, locale, defaultLocale, allDocs, type, lo
1105
1150
  for (const [docLocale, docIdx] of allDocs) {
1106
1151
  const found = docIdx.bySlug.get(slug);
1107
1152
  if (!found) continue;
1108
- const correctSlug = getSlugForLocale(found, docLocale, locale, allDocs, defaultLocale);
1109
- if (correctSlug && correctSlug !== slug) {
1153
+ const enSlug = docLocale === defaultLocale ? found.slug : found.enSlug;
1154
+ let target;
1155
+ let targetLocale = locale;
1156
+ for (const candidateLocale of [locale, ...fallbackLocales]) {
1157
+ const cand = candidateLocale === defaultLocale ? allDocs.get(candidateLocale)?.bySlug.get(enSlug) : allDocs.get(candidateLocale)?.byEnSlug.get(enSlug);
1158
+ if (cand) {
1159
+ target = cand;
1160
+ targetLocale = candidateLocale;
1161
+ break;
1162
+ }
1163
+ }
1164
+ if (target) {
1165
+ if (target.slug === slug) {
1166
+ return { document: target, actualLocale: targetLocale };
1167
+ }
1110
1168
  if (!type.path) {
1111
1169
  return { document: null, actualLocale: locale };
1112
1170
  }
1113
1171
  return {
1114
1172
  document: null,
1115
1173
  actualLocale: locale,
1116
- shouldRedirectTo: urlBuilder.resolvePath(type.path, correctSlug, locale)
1174
+ shouldRedirectTo: urlBuilder.resolvePath(type.path, target.slug, locale)
1117
1175
  };
1118
1176
  }
1119
1177
  if (docLocale === defaultLocale) {
@@ -1144,13 +1202,6 @@ function resolveLocalizedDocument(slug, locale, defaultLocale, allDocs, type, lo
1144
1202
  }
1145
1203
  return { document: null, actualLocale: locale };
1146
1204
  }
1147
- function getSlugForLocale(document, sourceLocale, targetLocale, allDocs, defaultLocale) {
1148
- if (sourceLocale === targetLocale) return document.slug;
1149
- const englishSlug = sourceLocale === defaultLocale ? document.slug : document.enSlug;
1150
- if (!englishSlug) return null;
1151
- if (targetLocale === defaultLocale) return englishSlug;
1152
- return allDocs.get(targetLocale)?.byEnSlug.get(englishSlug)?.slug ?? null;
1153
- }
1154
1205
 
1155
1206
  // src/create-project.ts
1156
1207
  function comparatorFor(orderBy) {
@@ -1203,7 +1254,8 @@ function buildRuntime(config, type, getRuntime) {
1203
1254
  config.defaultLocale,
1204
1255
  load(),
1205
1256
  type,
1206
- config.localeRouting
1257
+ config.localeRouting,
1258
+ config.localeFallbacks[locale] ?? []
1207
1259
  );
1208
1260
  if (result.document && type.path) {
1209
1261
  return {
@@ -1225,8 +1277,14 @@ function buildRuntime(config, type, getRuntime) {
1225
1277
  const params = [];
1226
1278
  for (const locale of options.locales ?? config.locales) {
1227
1279
  const localeIdx = all.get(locale);
1280
+ const fallbacks = config.localeFallbacks[locale] ?? [];
1228
1281
  for (const doc of enIdx.bySlug.values()) {
1229
- const slug = locale === config.defaultLocale ? doc.slug : localeIdx?.byEnSlug.get(doc.slug)?.slug ?? doc.slug;
1282
+ let slug;
1283
+ if (locale === config.defaultLocale) {
1284
+ slug = doc.slug;
1285
+ } else {
1286
+ slug = localeIdx?.byEnSlug.get(doc.slug)?.slug ?? fallbacks.map((fb) => all.get(fb)?.byEnSlug.get(doc.slug)?.slug).find((s) => s !== void 0) ?? doc.slug;
1287
+ }
1230
1288
  params.push({ locale, slug });
1231
1289
  }
1232
1290
  }
@@ -1688,7 +1746,7 @@ init_sqlite();
1688
1746
 
1689
1747
  // src/validate/validate-relations.ts
1690
1748
  init_esm_shims();
1691
- function getAtPath2(obj, fieldPath) {
1749
+ function getAtPath(obj, fieldPath) {
1692
1750
  let current = obj;
1693
1751
  for (const segment of fieldPath) {
1694
1752
  if (typeof current !== "object" || current === null || Array.isArray(current)) return void 0;
@@ -1733,7 +1791,7 @@ function validateRelations(project) {
1733
1791
  });
1734
1792
  continue;
1735
1793
  }
1736
- const value = getAtPath2(doc.frontmatter, fieldMeta.path);
1794
+ const value = getAtPath(doc.frontmatter, fieldMeta.path);
1737
1795
  if (value === void 0 || value === null || value === "") continue;
1738
1796
  const targetSlugs = slugIndex.get(targetTypeId) ?? /* @__PURE__ */ new Set();
1739
1797
  const fieldLabel = fieldMeta.path.join(".");
@@ -2117,6 +2175,66 @@ function resolveLocalesFromPreset(config, preset, explicitLocales) {
2117
2175
  // src/translate/page-translator.ts
2118
2176
  init_esm_shims();
2119
2177
 
2178
+ // src/storage/batch-jobs.ts
2179
+ init_esm_shims();
2180
+ function insertBatchJob(db, input) {
2181
+ const info = db.prepare(
2182
+ `INSERT INTO translation_batch_jobs (job_name, model, display_model, created_at, state)
2183
+ VALUES (?, ?, ?, ?, ?)`
2184
+ ).run(input.jobName, input.model, input.displayModel, input.createdAt, input.state);
2185
+ return Number(info.lastInsertRowid);
2186
+ }
2187
+ function insertBatchItems(db, jobId, items) {
2188
+ const stmt = db.prepare(
2189
+ `INSERT INTO translation_batch_items (
2190
+ job_id, request_index, content_type, en_slug, locale, en_hash, snapshot_id, status
2191
+ ) VALUES (?, ?, ?, ?, ?, ?, ?, 'pending')`
2192
+ );
2193
+ for (const item of items) {
2194
+ stmt.run(
2195
+ jobId,
2196
+ item.requestIndex,
2197
+ item.contentType,
2198
+ item.enSlug,
2199
+ item.locale,
2200
+ item.enHash,
2201
+ item.snapshotId
2202
+ );
2203
+ }
2204
+ }
2205
+ function listPendingBatchJobs(db) {
2206
+ return db.prepare(`SELECT * FROM translation_batch_jobs WHERE completed_at IS NULL ORDER BY id`).all();
2207
+ }
2208
+ function listBatchItems(db, jobId) {
2209
+ return db.prepare(`SELECT * FROM translation_batch_items WHERE job_id = ? ORDER BY request_index`).all(jobId);
2210
+ }
2211
+ function listPendingBatchItems(db) {
2212
+ return db.prepare(
2213
+ `SELECT i.* FROM translation_batch_items i
2214
+ JOIN translation_batch_jobs j ON j.id = i.job_id
2215
+ WHERE j.completed_at IS NULL AND i.status = 'pending'
2216
+ ORDER BY i.job_id, i.request_index`
2217
+ ).all();
2218
+ }
2219
+ function claimBatchJobCompletion(db, jobId, state, completedAt) {
2220
+ const info = db.prepare(
2221
+ `UPDATE translation_batch_jobs SET state = ?, completed_at = ?
2222
+ WHERE id = ? AND completed_at IS NULL`
2223
+ ).run(state, completedAt, jobId);
2224
+ return info.changes > 0;
2225
+ }
2226
+ function updateBatchItemStatus(db, jobId, requestIndex, status, error) {
2227
+ db.prepare(
2228
+ `UPDATE translation_batch_items SET status = ?, error = ? WHERE job_id = ? AND request_index = ?`
2229
+ ).run(status, error ?? null, jobId, requestIndex);
2230
+ }
2231
+
2232
+ // src/translate/page-translator.ts
2233
+ init_sqlite();
2234
+
2235
+ // src/translate/batch-worklist.ts
2236
+ init_esm_shims();
2237
+
2120
2238
  // src/history/record-snapshot.ts
2121
2239
  init_esm_shims();
2122
2240
  init_sqlite();
@@ -2134,9 +2252,111 @@ function recordEnSnapshot(config, input, db) {
2134
2252
  return id;
2135
2253
  }
2136
2254
 
2137
- // src/translate/page-translator.ts
2255
+ // src/translate/batch-worklist.ts
2138
2256
  init_sqlite();
2139
2257
 
2258
+ // src/translate/gemini-batch.ts
2259
+ init_esm_shims();
2260
+
2261
+ // src/translate/retry.ts
2262
+ init_esm_shims();
2263
+ var TRANSIENT_STATUS = /* @__PURE__ */ new Set([429, 500, 502, 503, 504]);
2264
+ function statusFromError(error) {
2265
+ if (error instanceof ApiError && typeof error.status === "number") {
2266
+ return error.status;
2267
+ }
2268
+ if (error && typeof error === "object" && "status" in error) {
2269
+ const status = error.status;
2270
+ if (typeof status === "number") return status;
2271
+ }
2272
+ const message = error instanceof Error ? error.message : String(error);
2273
+ const match = message.match(/\b(429|500|502|503|504)\b/);
2274
+ if (match) return Number(match[1]);
2275
+ return void 0;
2276
+ }
2277
+ function isNetworkError(error) {
2278
+ const err = error;
2279
+ const code = typeof err?.code === "string" ? err.code : "";
2280
+ if (code === "ECONNRESET" || code === "ETIMEDOUT" || code === "ECONNREFUSED" || code === "ENOTFOUND" || code === "EAI_AGAIN" || code === "EPIPE" || code === "UND_ERR_SOCKET") {
2281
+ return true;
2282
+ }
2283
+ const message = error instanceof Error ? error.message.toLowerCase() : "";
2284
+ return message.includes("network") || message.includes("fetch failed") || message.includes("socket hang up") || message.includes("terminated");
2285
+ }
2286
+ function isTransientError(error) {
2287
+ const status = statusFromError(error);
2288
+ if (status !== void 0) return TRANSIENT_STATUS.has(status);
2289
+ return isNetworkError(error);
2290
+ }
2291
+ var defaultSleep = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
2292
+ async function withRetry(fn, options = {}) {
2293
+ const attempts = Math.max(1, options.attempts ?? 3);
2294
+ const baseDelayMs = options.baseDelayMs ?? 1e3;
2295
+ const sleep2 = options.sleep ?? defaultSleep;
2296
+ const random = options.random ?? Math.random;
2297
+ let lastError;
2298
+ for (let attempt = 1; attempt <= attempts; attempt++) {
2299
+ try {
2300
+ return await fn();
2301
+ } catch (error) {
2302
+ lastError = error;
2303
+ if (attempt >= attempts || !isTransientError(error)) throw error;
2304
+ const backoff = baseDelayMs * 2 ** (attempt - 1);
2305
+ const delay = backoff + random() * backoff;
2306
+ await sleep2(delay);
2307
+ }
2308
+ }
2309
+ throw lastError;
2310
+ }
2311
+
2312
+ // src/translate/gemini-batch.ts
2313
+ var STATE_FAMILY_PREFIX = /^(JOB_STATE_|BATCH_STATE_)/;
2314
+ var TERMINAL_STATES = /* @__PURE__ */ new Set([
2315
+ "SUCCEEDED",
2316
+ "FAILED",
2317
+ "CANCELLED",
2318
+ "EXPIRED",
2319
+ "PARTIALLY_SUCCEEDED"
2320
+ ]);
2321
+ var SUCCESSFUL_STATES = /* @__PURE__ */ new Set(["SUCCEEDED", "PARTIALLY_SUCCEEDED"]);
2322
+ function normalizeBatchState(state) {
2323
+ return String(state ?? "UNKNOWN").replace(STATE_FAMILY_PREFIX, "");
2324
+ }
2325
+ function isTerminalBatchState(state) {
2326
+ return TERMINAL_STATES.has(normalizeBatchState(state));
2327
+ }
2328
+ function isSuccessfulBatchState(state) {
2329
+ return SUCCESSFUL_STATES.has(normalizeBatchState(state));
2330
+ }
2331
+ function makeClient(apiKey) {
2332
+ const key = apiKey ?? process.env.GEMINI_API_KEY;
2333
+ if (!key) {
2334
+ throw new Error("GEMINI_API_KEY is required for scribe translate");
2335
+ }
2336
+ return new GoogleGenAI({ apiKey: key });
2337
+ }
2338
+ function textFromBatchResponse(response) {
2339
+ if (typeof response.text === "string") return response.text;
2340
+ const parts = response.candidates?.[0]?.content?.parts ?? [];
2341
+ return parts.filter((part) => !part.thought && typeof part.text === "string").map((part) => part.text).join("");
2342
+ }
2343
+ async function createGeminiBatchJob(input) {
2344
+ const ai = makeClient(input.apiKey);
2345
+ const job = await withRetry(
2346
+ () => ai.batches.create({
2347
+ model: input.model,
2348
+ src: input.requests,
2349
+ config: { displayName: input.displayName ?? `scribe-translate-${Date.now()}` }
2350
+ })
2351
+ );
2352
+ if (!job.name) throw new Error("Gemini batch job was created without a name");
2353
+ return job;
2354
+ }
2355
+ async function getGeminiBatchJob(input) {
2356
+ const ai = makeClient(input.apiKey);
2357
+ return withRetry(() => ai.batches.get({ name: input.name }));
2358
+ }
2359
+
2140
2360
  // src/translate/gemini-client.ts
2141
2361
  init_esm_shims();
2142
2362
 
@@ -2161,6 +2381,12 @@ function normalizeGeminiDisplayName(model) {
2161
2381
  if (alias) return alias[0];
2162
2382
  return name;
2163
2383
  }
2384
+ function resolveThinkingConfig(apiModelId) {
2385
+ const id = stripModelsPrefix(apiModelId).toLowerCase();
2386
+ if (id.startsWith("gemini-3")) return { thinkingLevel: ThinkingLevel.LOW };
2387
+ if (id.includes("2.5")) return { thinkingBudget: 128 };
2388
+ return void 0;
2389
+ }
2164
2390
 
2165
2391
  // src/translate/gemini-client.ts
2166
2392
  var DEFAULT_MODEL = DEFAULT_GEMINI_MODEL;
@@ -2174,11 +2400,36 @@ function extractJson(text) {
2174
2400
  return trimmed;
2175
2401
  }
2176
2402
  function parseGeminiResponse(text) {
2403
+ let parsed;
2177
2404
  try {
2178
- return JSON.parse(text.trim());
2405
+ parsed = JSON.parse(text.trim());
2179
2406
  } catch {
2180
- return JSON.parse(extractJson(text));
2407
+ parsed = JSON.parse(extractJson(text));
2408
+ }
2409
+ if (parsed !== null && typeof parsed === "object" && parsed.frontmatter === void 0) {
2410
+ parsed.frontmatter = {};
2181
2411
  }
2412
+ return parsed;
2413
+ }
2414
+ function buildGeminiRequestConfig(input) {
2415
+ const thinkingConfig = resolveThinkingConfig(input.apiModelId);
2416
+ return {
2417
+ responseMimeType: "application/json",
2418
+ ...input.responseSchema ? { responseSchema: input.responseSchema } : {},
2419
+ ...thinkingConfig ? { thinkingConfig } : {}
2420
+ };
2421
+ }
2422
+ function usageFromResponse(response) {
2423
+ const meta = response.usageMetadata ?? response.usage_metadata ?? {};
2424
+ const thoughtsTokens = meta.thoughtsTokenCount ?? meta.thoughts_token_count ?? 0;
2425
+ return {
2426
+ inputTokens: meta.promptTokenCount ?? meta.prompt_token_count ?? 0,
2427
+ // candidatesTokenCount does not include thoughts, but thoughts are billed as
2428
+ // output tokens, so fold them in here for accurate cost accounting.
2429
+ outputTokens: (meta.candidatesTokenCount ?? meta.candidates_token_count ?? 0) + thoughtsTokens,
2430
+ thoughtsTokens,
2431
+ totalTokens: meta.totalTokenCount ?? meta.total_token_count ?? 0
2432
+ };
2182
2433
  }
2183
2434
  async function translatePageWithGemini(input) {
2184
2435
  const apiKey = input.apiKey ?? process.env.GEMINI_API_KEY;
@@ -2190,28 +2441,28 @@ async function translatePageWithGemini(input) {
2190
2441
  );
2191
2442
  const apiModel = resolveGeminiModelId(displayModel);
2192
2443
  const ai = new GoogleGenAI({ apiKey });
2193
- const response = await ai.models.generateContent({
2194
- model: apiModel,
2195
- contents: input.prompt,
2196
- config: {
2197
- responseMimeType: "application/json",
2198
- ...input.responseSchema ? { responseSchema: input.responseSchema } : {}
2199
- }
2200
- });
2444
+ const response = await withRetry(
2445
+ () => ai.models.generateContent({
2446
+ model: apiModel,
2447
+ contents: input.prompt,
2448
+ config: buildGeminiRequestConfig({
2449
+ apiModelId: apiModel,
2450
+ responseSchema: input.responseSchema
2451
+ })
2452
+ })
2453
+ );
2201
2454
  const raw = response.text ?? "";
2202
2455
  const parsed = parseGeminiResponse(raw);
2203
2456
  if (!parsed.frontmatter || typeof parsed.body !== "string") {
2204
2457
  throw new Error("Gemini response missing frontmatter/body");
2205
2458
  }
2206
- const usageMetadata = response.usageMetadata;
2207
- const usage = {
2208
- inputTokens: usageMetadata?.promptTokenCount ?? 0,
2209
- outputTokens: usageMetadata?.candidatesTokenCount ?? 0,
2210
- totalTokens: usageMetadata?.totalTokenCount ?? 0
2211
- };
2212
- return { model: displayModel, raw, parsed, usage };
2459
+ return { model: displayModel, raw, parsed, usage: usageFromResponse(response) };
2213
2460
  }
2214
2461
 
2462
+ // src/translate/translate-core.ts
2463
+ init_esm_shims();
2464
+ init_sqlite();
2465
+
2215
2466
  // src/translate/gemini-pricing.ts
2216
2467
  init_esm_shims();
2217
2468
  var CONTEXT_TIER_TOKENS = 2e5;
@@ -2230,12 +2481,14 @@ function resolveModelPricing(model) {
2230
2481
  const match = Object.entries(GEMINI_PRICING_USD).find(([key]) => id.includes(key));
2231
2482
  return match?.[1];
2232
2483
  }
2233
- function estimateTranslationCostUsd(model, inputTokens, outputTokens) {
2484
+ var BATCH_DISCOUNT = 0.5;
2485
+ function estimateTranslationCostUsd(model, inputTokens, outputTokens, mode = "interactive") {
2234
2486
  const pricing = resolveModelPricing(model);
2235
2487
  if (!pricing) return void 0;
2236
2488
  const inputRate = tierRate(inputTokens, pricing.input);
2237
2489
  const outputRate = tierRate(outputTokens, pricing.output);
2238
- return inputTokens / 1e6 * inputRate + outputTokens / 1e6 * outputRate;
2490
+ const multiplier = mode === "batch" ? BATCH_DISCOUNT : 1;
2491
+ return (inputTokens / 1e6 * inputRate + outputTokens / 1e6 * outputRate) * multiplier;
2239
2492
  }
2240
2493
  function formatTokenCount(tokens) {
2241
2494
  if (tokens >= 1e6) return `${(tokens / 1e6).toFixed(2)}M`;
@@ -2263,6 +2516,7 @@ var LOCALE_NAMES = {
2263
2516
  "zh-CN": "Simplified Chinese",
2264
2517
  "zh-TW": "Traditional Chinese",
2265
2518
  pt: "Portuguese",
2519
+ "pt-BR": "Brazilian Portuguese",
2266
2520
  ja: "Japanese",
2267
2521
  ru: "Russian",
2268
2522
  it: "Italian",
@@ -2270,29 +2524,29 @@ var LOCALE_NAMES = {
2270
2524
  };
2271
2525
  function defaultLocalizationPrompt(localeName, locale) {
2272
2526
  return [
2273
- `Localize the content below into natural ${localeName} (${locale}).`,
2527
+ `Localize the content above into natural ${localeName} (${locale}).`,
2274
2528
  "Do not translate word-for-word.",
2275
2529
  "Preserve the source tone and brand voice.",
2276
2530
  "Write as if a native speaker authored it for the target market."
2277
2531
  ].join(" ");
2278
2532
  }
2533
+ var TASK_FRAMING = "You are localizing the content below into a target language specified at the end of this prompt.";
2534
+ function buildOutputFormatLine(hasFrontmatter, slugStrategy) {
2535
+ if (!hasFrontmatter) {
2536
+ return slugStrategy === "localized" ? "`body` (string, full MDX body), `slug` (string)." : "`body` (string, full MDX body).";
2537
+ }
2538
+ return slugStrategy === "localized" ? "`frontmatter` (object with translated frontmatter fields), `body` (string, full MDX body), `slug` (string)." : "`frontmatter` (object), `body` (string).";
2539
+ }
2279
2540
  function buildPageTranslationPrompt(input) {
2280
2541
  const localeName = LOCALE_NAMES[input.targetLocale] ?? input.targetLocale;
2281
- const prompt = input.resolved.promptOverride ?? defaultLocalizationPrompt(localeName, input.targetLocale);
2282
- const lines = [
2283
- prompt,
2542
+ const localizationPrompt = input.resolved.promptOverride ?? defaultLocalizationPrompt(localeName, input.targetLocale);
2543
+ const prefix = [
2544
+ TASK_FRAMING,
2284
2545
  "",
2285
2546
  ...input.resolved.context ? ["## Context", input.resolved.context, ""] : [],
2286
2547
  ...input.contextLabel ? [`Document: ${input.contextLabel}`, ""] : [],
2287
2548
  "## Rules",
2288
2549
  ...input.resolved.rules.map((rule) => `- ${rule}`),
2289
- ...input.slugStrategy === "localized" ? [
2290
- `- The slug MUST be written in ${localeName}, derived from the ${localeName} title and its meaning \u2014 never the English slug. Transliterate non-Latin ${localeName} into ASCII Latin.`
2291
- ] : [],
2292
- "",
2293
- "## Output format",
2294
- "Return ONLY valid JSON with keys:",
2295
- input.slugStrategy === "localized" ? "`frontmatter` (object with translated frontmatter fields), `body` (string, full MDX body), `slug` (string)." : "`frontmatter` (object), `body` (string).",
2296
2550
  "",
2297
2551
  "## EN translatable frontmatter (JSON)",
2298
2552
  JSON.stringify(input.translatableFrontmatter, null, 2),
@@ -2300,7 +2554,22 @@ function buildPageTranslationPrompt(input) {
2300
2554
  "## EN body (MDX)",
2301
2555
  input.enBody
2302
2556
  ];
2303
- return lines.join("\n");
2557
+ const suffix = [
2558
+ "",
2559
+ "## Target language",
2560
+ localizationPrompt,
2561
+ ...input.slugStrategy === "localized" ? [
2562
+ `The slug MUST be written in ${localeName}, derived from the ${localeName} title and its meaning \u2014 never the English slug.`
2563
+ ] : [],
2564
+ "",
2565
+ "## Output format",
2566
+ "Return ONLY valid JSON with keys:",
2567
+ buildOutputFormatLine(
2568
+ Object.keys(input.translatableFrontmatter).length > 0,
2569
+ input.slugStrategy
2570
+ )
2571
+ ];
2572
+ return [...prefix, ...suffix].join("\n");
2304
2573
  }
2305
2574
 
2306
2575
  // src/translate/response-schema.ts
@@ -2351,9 +2620,8 @@ function buildTranslatableSubschema(schema) {
2351
2620
  function buildGeminiResponseSchema(schema, slugStrategy) {
2352
2621
  try {
2353
2622
  const translatable = buildTranslatableSubschema(schema);
2354
- if (!translatable) return null;
2355
2623
  const responseShape = {
2356
- frontmatter: translatable,
2624
+ ...translatable ? { frontmatter: translatable } : {},
2357
2625
  body: z.string()
2358
2626
  };
2359
2627
  if (slugStrategy === "localized") {
@@ -2434,7 +2702,12 @@ function validateTranslatedFrontmatter(enDoc, localeFrontmatter, typeSchema) {
2434
2702
  return { ok: true, frontmatter };
2435
2703
  }
2436
2704
 
2437
- // src/translate/page-translator.ts
2705
+ // src/translate/translate-core.ts
2706
+ function translationItemKey(item) {
2707
+ const contentType = item.contentType ?? item.content_type ?? "";
2708
+ const enSlug = item.enSlug ?? item.en_slug ?? "";
2709
+ return `${contentType}\0${enSlug}\0${item.locale}`;
2710
+ }
2438
2711
  function summarizeResults(results, durationMs) {
2439
2712
  return results.reduce(
2440
2713
  (totals, result) => {
@@ -2470,17 +2743,6 @@ function formatTranslateError(error) {
2470
2743
  }
2471
2744
  return message.length > 160 ? `${message.slice(0, 157)}\u2026` : message;
2472
2745
  }
2473
- async function runPool(items, concurrency, worker) {
2474
- if (items.length === 0) return;
2475
- let nextIndex = 0;
2476
- const runners = Array.from({ length: Math.min(concurrency, items.length) }, async () => {
2477
- while (nextIndex < items.length) {
2478
- const index = nextIndex++;
2479
- await worker(items[index], index);
2480
- }
2481
- });
2482
- await Promise.all(runners);
2483
- }
2484
2746
  function resolveContextLabel(enDoc, enSlug) {
2485
2747
  const fm = enDoc.frontmatter;
2486
2748
  for (const key of ["title", "name", "h1"]) {
@@ -2489,73 +2751,82 @@ function resolveContextLabel(enDoc, enSlug) {
2489
2751
  }
2490
2752
  return enSlug;
2491
2753
  }
2492
- async function translatePage(config, item, options = {}) {
2493
- const startedAt = Date.now();
2494
- const base = {
2754
+ function baseForItem(item) {
2755
+ return {
2495
2756
  contentType: item.contentType,
2496
2757
  enSlug: item.enSlug,
2497
2758
  locale: item.locale
2498
2759
  };
2499
- try {
2500
- const type = config.types.find((t) => t.id === item.contentType);
2501
- if (!type) throw new Error(`Unknown content type ${item.contentType}`);
2502
- const enDoc = readEnDocument(config, type, item.enSlug);
2503
- if (!enDoc) throw new Error(`EN document not found: ${item.enSlug}`);
2504
- const payload = getTranslatablePayload(enDoc, type);
2505
- const currentEnHash = computePageEnHash(payload.frontmatter, payload.body);
2506
- const db = openStore(config, "readonly");
2507
- const existing = getTranslation(db, type.id, item.enSlug, item.locale);
2508
- db.close();
2509
- if (!options.force && existing && existing.en_hash === currentEnHash) {
2510
- return {
2511
- ...base,
2760
+ }
2761
+ function prepareTranslation(config, item, options, startedAt) {
2762
+ const type = config.types.find((t) => t.id === item.contentType);
2763
+ if (!type) throw new Error(`Unknown content type ${item.contentType}`);
2764
+ const enDoc = readEnDocument(config, type, item.enSlug);
2765
+ if (!enDoc) throw new Error(`EN document not found: ${item.enSlug}`);
2766
+ const payload = getTranslatablePayload(enDoc, type);
2767
+ const currentEnHash = computePageEnHash(payload.frontmatter, payload.body);
2768
+ const db = openStore(config, "readonly");
2769
+ const existing = getTranslation(db, type.id, item.enSlug, item.locale);
2770
+ db.close();
2771
+ if (!options.force && existing && existing.en_hash === currentEnHash) {
2772
+ return {
2773
+ status: "done",
2774
+ result: {
2775
+ ...baseForItem(item),
2512
2776
  skipped: true,
2513
2777
  reason: "fresh",
2514
2778
  durationMs: Date.now() - startedAt
2515
- };
2516
- }
2517
- const resolvedTranslate = resolveTranslateConfig(config, type);
2518
- const model = options.model ?? resolvedTranslate.model;
2519
- const prompt = buildPageTranslationPrompt({
2520
- resolved: resolvedTranslate,
2521
- targetLocale: item.locale,
2522
- contextLabel: resolveContextLabel(enDoc, item.enSlug),
2523
- translatableFrontmatter: payload.frontmatter,
2524
- enBody: payload.body,
2525
- slugStrategy: type.slugStrategy
2526
- });
2527
- if (options.dryRun) {
2528
- return {
2529
- ...base,
2530
- skipped: false,
2531
- model,
2532
- durationMs: Date.now() - startedAt
2533
- };
2534
- }
2535
- const responseSchema = buildGeminiResponseSchema(type.schema, type.slugStrategy);
2536
- const result = await translatePageWithGemini({
2537
- prompt,
2779
+ }
2780
+ };
2781
+ }
2782
+ const resolvedTranslate = resolveTranslateConfig(config, type);
2783
+ const model = options.model ?? resolvedTranslate.model;
2784
+ const prompt = buildPageTranslationPrompt({
2785
+ resolved: resolvedTranslate,
2786
+ targetLocale: item.locale,
2787
+ contextLabel: resolveContextLabel(enDoc, item.enSlug),
2788
+ translatableFrontmatter: payload.frontmatter,
2789
+ enBody: payload.body,
2790
+ slugStrategy: type.slugStrategy
2791
+ });
2792
+ const responseSchema = buildGeminiResponseSchema(type.schema, type.slugStrategy);
2793
+ return {
2794
+ status: "ready",
2795
+ prepared: {
2796
+ item,
2797
+ type,
2798
+ enDoc,
2799
+ payload,
2800
+ currentEnHash,
2801
+ existingSlug: existing?.slug,
2538
2802
  model,
2803
+ prompt,
2539
2804
  responseSchema: responseSchema ?? void 0
2540
- });
2541
- const rawSlug = type.slugStrategy === "localized" ? result.parsed.slug ?? existing?.slug ?? item.enSlug : item.enSlug;
2805
+ }
2806
+ };
2807
+ }
2808
+ function finalizeTranslation(config, prepared, output, options) {
2809
+ const { item, type, enDoc, payload } = prepared;
2810
+ const base = baseForItem(item);
2811
+ try {
2812
+ const rawSlug = type.slugStrategy === "localized" ? output.parsed.slug ?? prepared.existingSlug ?? item.enSlug : item.enSlug;
2542
2813
  const localeCodes = config.locales.filter((l) => l !== config.defaultLocale);
2543
2814
  const { slug, stripped, matchedCode } = stripLocaleSuffixFromSlug(rawSlug, localeCodes);
2544
2815
  const slugAdjusted = stripped && matchedCode ? { from: rawSlug, to: slug, matchedCode } : void 0;
2545
- const validated = validateTranslatedFrontmatter(enDoc, result.parsed.frontmatter, type.schema);
2816
+ const validated = validateTranslatedFrontmatter(enDoc, output.parsed.frontmatter, type.schema);
2546
2817
  if (!validated.ok) {
2547
2818
  throw new Error(`Translation validation failed: ${validated.error}`);
2548
2819
  }
2549
2820
  const { body: translatedBody, adjusted: mdxAdjusted } = assertValidTranslatedMdxBody(
2550
- result.parsed.body
2821
+ output.parsed.body
2551
2822
  );
2552
2823
  const writeDb = openStore(config, "readwrite");
2553
- const snapshotId = recordEnSnapshot(
2824
+ const snapshotId = options.snapshotId ?? recordEnSnapshot(
2554
2825
  config,
2555
2826
  {
2556
2827
  contentType: type.id,
2557
2828
  enSlug: item.enSlug,
2558
- enHash: currentEnHash,
2829
+ enHash: prepared.currentEnHash,
2559
2830
  frontmatter: payload.frontmatter,
2560
2831
  body: payload.body
2561
2832
  },
@@ -2568,24 +2839,25 @@ async function translatePage(config, item, options = {}) {
2568
2839
  slug,
2569
2840
  frontmatter: validated.frontmatter,
2570
2841
  body: translatedBody,
2571
- enHash: currentEnHash,
2842
+ enHash: prepared.currentEnHash,
2572
2843
  translatedAt: (/* @__PURE__ */ new Date()).toISOString(),
2573
- model: result.model,
2844
+ model: output.model,
2574
2845
  snapshotId
2575
2846
  });
2576
2847
  writeDb.close();
2577
2848
  const estimatedCostUsd = estimateTranslationCostUsd(
2578
- normalizeGeminiDisplayName(result.model),
2579
- result.usage.inputTokens,
2580
- result.usage.outputTokens
2849
+ normalizeGeminiDisplayName(output.model),
2850
+ output.usage.inputTokens,
2851
+ output.usage.outputTokens,
2852
+ options.costMode
2581
2853
  );
2582
2854
  return {
2583
2855
  ...base,
2584
2856
  skipped: false,
2585
- model: result.model,
2586
- usage: result.usage,
2857
+ model: output.model,
2858
+ usage: output.usage,
2587
2859
  estimatedCostUsd,
2588
- durationMs: Date.now() - startedAt,
2860
+ durationMs: Date.now() - options.startedAt,
2589
2861
  slugAdjusted,
2590
2862
  mdxAdjusted: mdxAdjusted || void 0
2591
2863
  };
@@ -2595,33 +2867,559 @@ async function translatePage(config, item, options = {}) {
2595
2867
  skipped: false,
2596
2868
  failed: true,
2597
2869
  error: formatTranslateError(error),
2870
+ durationMs: Date.now() - options.startedAt
2871
+ };
2872
+ }
2873
+ }
2874
+ function displayModelFor(prepared) {
2875
+ return normalizeGeminiDisplayName(
2876
+ prepared.model ?? process.env.PROSE_GEMINI_MODEL ?? DEFAULT_GEMINI_MODEL
2877
+ );
2878
+ }
2879
+
2880
+ // src/translate/batch-worklist.ts
2881
+ var MAX_REQUESTS_PER_JOB = 100;
2882
+ var MAX_PROMPT_BYTES_PER_JOB = 15 * 1024 * 1024;
2883
+ var INITIAL_POLL_MS = 5e3;
2884
+ var MAX_POLL_MS = 3e4;
2885
+ var POLL_BACKOFF_FACTOR = 1.5;
2886
+ function sleep(ms) {
2887
+ return new Promise((resolve) => setTimeout(resolve, ms));
2888
+ }
2889
+ function planBatchJobs(entries) {
2890
+ const byModel = /* @__PURE__ */ new Map();
2891
+ for (const entry of entries) {
2892
+ const group = byModel.get(entry.apiModel);
2893
+ if (group) group.push(entry);
2894
+ else byModel.set(entry.apiModel, [entry]);
2895
+ }
2896
+ const plans = [];
2897
+ for (const [apiModel, group] of byModel) {
2898
+ let current = [];
2899
+ let currentBytes = 0;
2900
+ for (const entry of group) {
2901
+ const size = Buffer.byteLength(entry.prompt, "utf8");
2902
+ if (current.length > 0 && (current.length >= MAX_REQUESTS_PER_JOB || currentBytes + size > MAX_PROMPT_BYTES_PER_JOB)) {
2903
+ plans.push({ apiModel, entries: current });
2904
+ current = [];
2905
+ currentBytes = 0;
2906
+ }
2907
+ current.push(entry);
2908
+ currentBytes += size;
2909
+ }
2910
+ if (current.length > 0) plans.push({ apiModel, entries: current });
2911
+ }
2912
+ return plans;
2913
+ }
2914
+ function readPendingBatchWork(config) {
2915
+ const db = openStore(config, "readwrite");
2916
+ const jobs = listPendingBatchJobs(db);
2917
+ const pendingItems = listPendingBatchItems(db);
2918
+ db.close();
2919
+ const inFlightKeys = new Set(pendingItems.map((item) => translationItemKey(item)));
2920
+ return { jobs, inFlightKeys, pendingItems };
2921
+ }
2922
+ async function submitBatchJobPlan(config, input) {
2923
+ const { plan, displayModel, jobIndex, jobCount } = input;
2924
+ const job = await createGeminiBatchJob({
2925
+ model: plan.apiModel,
2926
+ requests: plan.entries.map(({ prepared }) => ({
2927
+ contents: prepared.prompt,
2928
+ config: buildGeminiRequestConfig({
2929
+ apiModelId: plan.apiModel,
2930
+ responseSchema: prepared.responseSchema
2931
+ })
2932
+ })),
2933
+ displayName: `scribe-translate-${(/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-")}-${jobIndex + 1}`
2934
+ });
2935
+ const createdAt = (/* @__PURE__ */ new Date()).toISOString();
2936
+ const db = openStore(config, "readwrite");
2937
+ const jobId = insertBatchJob(db, {
2938
+ jobName: job.name,
2939
+ model: plan.apiModel,
2940
+ displayModel,
2941
+ state: String(job.state ?? "JOB_STATE_PENDING"),
2942
+ createdAt
2943
+ });
2944
+ insertBatchItems(
2945
+ db,
2946
+ jobId,
2947
+ plan.entries.map(({ prepared }, requestIndex) => ({
2948
+ requestIndex,
2949
+ contentType: prepared.item.contentType,
2950
+ enSlug: prepared.item.enSlug,
2951
+ locale: prepared.item.locale,
2952
+ enHash: prepared.currentEnHash,
2953
+ // Snapshot the EN source now so ingestion after a resume does not depend
2954
+ // on the EN files still matching (or existing) on disk.
2955
+ snapshotId: recordEnSnapshot(
2956
+ config,
2957
+ {
2958
+ contentType: prepared.item.contentType,
2959
+ enSlug: prepared.item.enSlug,
2960
+ enHash: prepared.currentEnHash,
2961
+ frontmatter: prepared.payload.frontmatter,
2962
+ body: prepared.payload.body
2963
+ },
2964
+ db
2965
+ )
2966
+ }))
2967
+ );
2968
+ const row = {
2969
+ id: jobId,
2970
+ job_name: job.name,
2971
+ model: plan.apiModel,
2972
+ display_model: displayModel,
2973
+ created_at: createdAt,
2974
+ state: String(job.state ?? "JOB_STATE_PENDING"),
2975
+ completed_at: null
2976
+ };
2977
+ db.close();
2978
+ input.onProgress?.({
2979
+ type: "batch-submitted",
2980
+ name: job.name,
2981
+ count: plan.entries.length,
2982
+ model: displayModel,
2983
+ jobIndex,
2984
+ jobCount,
2985
+ createdAt
2986
+ });
2987
+ return row;
2988
+ }
2989
+ function buildIngestContext(config, db, jobRow, itemRow) {
2990
+ const type = config.types.find((t) => t.id === itemRow.content_type);
2991
+ if (!type) throw new Error(`Unknown content type ${itemRow.content_type}`);
2992
+ let enDoc = readEnDocument(config, type, itemRow.en_slug);
2993
+ if (!enDoc) {
2994
+ const snapshot = getEnSnapshot(db, itemRow.snapshot_id);
2995
+ if (!snapshot) {
2996
+ throw new Error(
2997
+ `EN document and snapshot #${itemRow.snapshot_id} not found for ${itemRow.en_slug}`
2998
+ );
2999
+ }
3000
+ enDoc = {
3001
+ slug: itemRow.en_slug,
3002
+ enSlug: itemRow.en_slug,
3003
+ locale: config.defaultLocale,
3004
+ noindex: false,
3005
+ frontmatter: JSON.parse(snapshot.frontmatter_json),
3006
+ content: snapshot.body
3007
+ };
3008
+ }
3009
+ const item = {
3010
+ contentType: itemRow.content_type,
3011
+ enSlug: itemRow.en_slug,
3012
+ locale: itemRow.locale,
3013
+ reason: "missing",
3014
+ currentEnHash: itemRow.en_hash
3015
+ };
3016
+ return {
3017
+ item,
3018
+ type,
3019
+ enDoc,
3020
+ // Unused on ingest: finalizeTranslation receives the pre-recorded snapshotId.
3021
+ payload: { frontmatter: enDoc.frontmatter, body: enDoc.content },
3022
+ currentEnHash: itemRow.en_hash,
3023
+ existingSlug: getTranslation(db, type.id, itemRow.en_slug, itemRow.locale)?.slug,
3024
+ model: jobRow.display_model,
3025
+ prompt: "",
3026
+ responseSchema: void 0
3027
+ };
3028
+ }
3029
+ function ingestBatchJob(config, jobRow, batchJob, onResult) {
3030
+ const state = String(batchJob.state ?? "UNKNOWN");
3031
+ const db = openStore(config, "readwrite");
3032
+ if (!claimBatchJobCompletion(db, jobRow.id, state, (/* @__PURE__ */ new Date()).toISOString())) {
3033
+ db.close();
3034
+ return null;
3035
+ }
3036
+ const pendingItems = listBatchItems(db, jobRow.id).filter((item) => item.status === "pending");
3037
+ const results = [];
3038
+ const failItem = (itemRow, message, startedAt) => {
3039
+ const result = {
3040
+ contentType: itemRow.content_type,
3041
+ enSlug: itemRow.en_slug,
3042
+ locale: itemRow.locale,
3043
+ skipped: false,
3044
+ failed: true,
3045
+ error: formatTranslateError(new Error(message)),
2598
3046
  durationMs: Date.now() - startedAt
2599
3047
  };
3048
+ updateBatchItemStatus(db, jobRow.id, itemRow.request_index, "failed", result.error);
3049
+ results.push(result);
3050
+ onResult?.(result);
3051
+ };
3052
+ if (isSuccessfulBatchState(state)) {
3053
+ const responses = batchJob.dest?.inlinedResponses ?? [];
3054
+ for (const itemRow of pendingItems) {
3055
+ const startedAt = Date.now();
3056
+ const inlined = responses[itemRow.request_index];
3057
+ if (!inlined || inlined.error || !inlined.response) {
3058
+ failItem(
3059
+ itemRow,
3060
+ inlined?.error?.message ?? (inlined ? "Batch response missing content" : "Batch response missing"),
3061
+ startedAt
3062
+ );
3063
+ continue;
3064
+ }
3065
+ let result;
3066
+ try {
3067
+ const response = inlined.response;
3068
+ const raw = textFromBatchResponse(response);
3069
+ const parsed = parseGeminiResponse(raw);
3070
+ if (!parsed.frontmatter || typeof parsed.body !== "string") {
3071
+ throw new Error("Gemini response missing frontmatter/body");
3072
+ }
3073
+ const prepared = buildIngestContext(config, db, jobRow, itemRow);
3074
+ result = finalizeTranslation(
3075
+ config,
3076
+ prepared,
3077
+ { model: jobRow.display_model, parsed, usage: usageFromResponse(response) },
3078
+ { costMode: "batch", startedAt, snapshotId: itemRow.snapshot_id }
3079
+ );
3080
+ } catch (error) {
3081
+ failItem(itemRow, error instanceof Error ? error.message : String(error), startedAt);
3082
+ continue;
3083
+ }
3084
+ updateBatchItemStatus(
3085
+ db,
3086
+ jobRow.id,
3087
+ itemRow.request_index,
3088
+ result.failed ? "failed" : "done",
3089
+ result.error
3090
+ );
3091
+ results.push(result);
3092
+ onResult?.(result);
3093
+ }
3094
+ } else {
3095
+ const message = batchJob.error?.message ?? `Batch job ended in state ${normalizeBatchState(state)}`;
3096
+ const startedAt = Date.now();
3097
+ for (const itemRow of pendingItems) {
3098
+ failItem(itemRow, message, startedAt);
3099
+ }
3100
+ }
3101
+ db.close();
3102
+ return results;
3103
+ }
3104
+ async function pollAndIngestBatchJobs(config, jobs, options = {}) {
3105
+ const jobCount = options.jobCount ?? jobs.length;
3106
+ const active = jobs.map((row, jobIndex) => ({ row, jobIndex }));
3107
+ const loopStartedAt = Date.now();
3108
+ let delay = INITIAL_POLL_MS;
3109
+ let firstRound = true;
3110
+ const elapsedFor = (row) => {
3111
+ const createdAtMs = Date.parse(row.created_at);
3112
+ return Number.isNaN(createdAtMs) ? Date.now() - loopStartedAt : Date.now() - createdAtMs;
3113
+ };
3114
+ while (active.length > 0) {
3115
+ if (!firstRound) {
3116
+ await sleep(delay);
3117
+ delay = Math.min(delay * POLL_BACKOFF_FACTOR, MAX_POLL_MS);
3118
+ }
3119
+ firstRound = false;
3120
+ for (const tracked of [...active]) {
3121
+ const batchJob = await getGeminiBatchJob({ name: tracked.row.job_name });
3122
+ const state = normalizeBatchState(batchJob.state);
3123
+ options.onProgress?.({
3124
+ type: "batch-polling",
3125
+ name: tracked.row.job_name,
3126
+ state,
3127
+ elapsedMs: elapsedFor(tracked.row),
3128
+ jobIndex: tracked.jobIndex,
3129
+ jobCount
3130
+ });
3131
+ if (isTerminalBatchState(batchJob.state)) {
3132
+ active.splice(active.indexOf(tracked), 1);
3133
+ const results = ingestBatchJob(
3134
+ config,
3135
+ tracked.row,
3136
+ batchJob,
3137
+ options.onResult
3138
+ );
3139
+ if (results === null) {
3140
+ options.onProgress?.({
3141
+ type: "batch-done",
3142
+ name: tracked.row.job_name,
3143
+ state,
3144
+ model: tracked.row.display_model,
3145
+ count: 0,
3146
+ jobIndex: tracked.jobIndex,
3147
+ jobCount,
3148
+ translated: 0,
3149
+ failed: 0,
3150
+ inputTokens: 0,
3151
+ outputTokens: 0,
3152
+ estimatedCostUsd: 0,
3153
+ elapsedMs: elapsedFor(tracked.row),
3154
+ alreadyIngested: true
3155
+ });
3156
+ continue;
3157
+ }
3158
+ options.onProgress?.({
3159
+ type: "batch-done",
3160
+ name: tracked.row.job_name,
3161
+ state,
3162
+ model: tracked.row.display_model,
3163
+ count: results.length,
3164
+ jobIndex: tracked.jobIndex,
3165
+ jobCount,
3166
+ translated: results.filter((r) => !r.failed && !r.skipped).length,
3167
+ failed: results.filter((r) => r.failed).length,
3168
+ inputTokens: results.reduce((sum, r) => sum + (r.usage?.inputTokens ?? 0), 0),
3169
+ outputTokens: results.reduce((sum, r) => sum + (r.usage?.outputTokens ?? 0), 0),
3170
+ estimatedCostUsd: results.reduce((sum, r) => sum + (r.estimatedCostUsd ?? 0), 0),
3171
+ elapsedMs: elapsedFor(tracked.row)
3172
+ });
3173
+ }
3174
+ }
2600
3175
  }
2601
3176
  }
3177
+ function failedResultForItem(item, error, startedAt) {
3178
+ return {
3179
+ ...baseForItem(item),
3180
+ skipped: false,
3181
+ failed: true,
3182
+ error: formatTranslateError(error),
3183
+ durationMs: Date.now() - startedAt
3184
+ };
3185
+ }
3186
+
3187
+ // src/translate/page-translator.ts
3188
+ async function runPool(items, concurrency, worker) {
3189
+ if (items.length === 0) return;
3190
+ let nextIndex = 0;
3191
+ const runners = Array.from({ length: Math.min(concurrency, items.length) }, async () => {
3192
+ while (nextIndex < items.length) {
3193
+ const index = nextIndex++;
3194
+ await worker(items[index], index);
3195
+ }
3196
+ });
3197
+ await Promise.all(runners);
3198
+ }
2602
3199
  function labelForItem(item) {
2603
3200
  return `${item.contentType}/${item.enSlug}@${item.locale}`;
2604
3201
  }
3202
+ async function translatePage(config, item, options = {}) {
3203
+ const startedAt = Date.now();
3204
+ let outcome;
3205
+ try {
3206
+ outcome = prepareTranslation(config, item, options, startedAt);
3207
+ } catch (error) {
3208
+ return failedResultForItem(item, error, startedAt);
3209
+ }
3210
+ if (outcome.status === "done") return outcome.result;
3211
+ const prepared = outcome.prepared;
3212
+ if (options.dryRun) {
3213
+ return {
3214
+ ...baseForItem(item),
3215
+ skipped: false,
3216
+ model: prepared.model,
3217
+ durationMs: Date.now() - startedAt
3218
+ };
3219
+ }
3220
+ try {
3221
+ const result = await translatePageWithGemini({
3222
+ prompt: prepared.prompt,
3223
+ model: prepared.model,
3224
+ responseSchema: prepared.responseSchema
3225
+ });
3226
+ return finalizeTranslation(
3227
+ config,
3228
+ prepared,
3229
+ { model: result.model, parsed: result.parsed, usage: result.usage },
3230
+ { costMode: "interactive", startedAt }
3231
+ );
3232
+ } catch (error) {
3233
+ return failedResultForItem(item, error, startedAt);
3234
+ }
3235
+ }
3236
+ function createResultCollector(items, onProgress) {
3237
+ const slotByKey = /* @__PURE__ */ new Map();
3238
+ items.forEach((item, index) => slotByKey.set(translationItemKey(item), index));
3239
+ const slots = new Array(items.length);
3240
+ const extras = [];
3241
+ return {
3242
+ emit(result) {
3243
+ const slot = slotByKey.get(translationItemKey(result));
3244
+ if (slot !== void 0 && slots[slot] === void 0) slots[slot] = result;
3245
+ else extras.push(result);
3246
+ onProgress?.({ type: "item-done", result });
3247
+ },
3248
+ assemble() {
3249
+ return [
3250
+ ...slots.filter((result) => result !== void 0),
3251
+ ...extras
3252
+ ];
3253
+ }
3254
+ };
3255
+ }
2605
3256
  async function translateWorklist(config, items, options = {}) {
3257
+ const mode = options.mode ?? "batch";
2606
3258
  const concurrency = Math.max(1, options.concurrency ?? 3);
2607
3259
  const startedAt = Date.now();
2608
- const results = new Array(items.length);
2609
- const active = /* @__PURE__ */ new Set();
3260
+ if (options.dryRun) {
3261
+ options.onProgress?.({
3262
+ type: "start",
3263
+ total: items.length,
3264
+ concurrency,
3265
+ dryRun: true,
3266
+ model: options.model,
3267
+ mode
3268
+ });
3269
+ const results2 = items.map((item) => {
3270
+ const itemStartedAt = Date.now();
3271
+ try {
3272
+ const outcome = prepareTranslation(config, item, options, itemStartedAt);
3273
+ return outcome.status === "done" ? outcome.result : {
3274
+ ...baseForItem(item),
3275
+ skipped: false,
3276
+ model: outcome.prepared.model,
3277
+ durationMs: Date.now() - itemStartedAt
3278
+ };
3279
+ } catch (error) {
3280
+ return failedResultForItem(item, error, itemStartedAt);
3281
+ }
3282
+ });
3283
+ for (const result of results2) options.onProgress?.({ type: "item-done", result });
3284
+ const totals2 = summarizeResults(results2, Date.now() - startedAt);
3285
+ options.onProgress?.({ type: "done", results: results2, totals: totals2 });
3286
+ return results2;
3287
+ }
3288
+ const pending = readPendingBatchWork(config);
3289
+ const inputKeys = new Set(items.map((item) => translationItemKey(item)));
3290
+ const resumedExtraCount = pending.pendingItems.filter(
3291
+ (item) => !inputKeys.has(translationItemKey(item))
3292
+ ).length;
2610
3293
  options.onProgress?.({
2611
3294
  type: "start",
2612
- total: items.length,
3295
+ total: items.length + resumedExtraCount,
2613
3296
  concurrency,
2614
- dryRun: Boolean(options.dryRun),
2615
- model: options.model
3297
+ dryRun: false,
3298
+ model: options.model,
3299
+ mode
2616
3300
  });
2617
- await runPool(items, concurrency, async (item, index) => {
2618
- const label = labelForItem(item);
2619
- active.add(label);
2620
- options.onProgress?.({ type: "item-start", item, active: [...active] });
2621
- const result = await translatePage(config, item, options);
2622
- results[index] = result;
2623
- active.delete(label);
2624
- options.onProgress?.({ type: "item-done", result });
3301
+ const collector = createResultCollector(items, options.onProgress);
3302
+ const workItems = items.filter((item) => !pending.inFlightKeys.has(translationItemKey(item)));
3303
+ if (mode === "direct") {
3304
+ const pendingCounts = countPendingItems(config, pending.jobs);
3305
+ emitResumedJobs(pending.jobs, pendingCounts, pending.jobs.length, options.onProgress);
3306
+ const active = /* @__PURE__ */ new Set();
3307
+ await Promise.all([
3308
+ runPool(workItems, concurrency, async (item) => {
3309
+ const label = labelForItem(item);
3310
+ active.add(label);
3311
+ options.onProgress?.({ type: "item-start", item, active: [...active] });
3312
+ const result = await translatePage(config, item, options);
3313
+ active.delete(label);
3314
+ collector.emit(result);
3315
+ }),
3316
+ pending.jobs.length > 0 ? pollAndIngestBatchJobs(config, pending.jobs, {
3317
+ onProgress: options.onProgress,
3318
+ onResult: collector.emit
3319
+ }) : Promise.resolve()
3320
+ ]);
3321
+ } else {
3322
+ const entries = [];
3323
+ for (const item of workItems) {
3324
+ const itemStartedAt = Date.now();
3325
+ try {
3326
+ const outcome = prepareTranslation(config, item, options, itemStartedAt);
3327
+ if (outcome.status === "done") {
3328
+ collector.emit(outcome.result);
3329
+ } else {
3330
+ entries.push({
3331
+ apiModel: resolveGeminiModelId(displayModelFor(outcome.prepared)),
3332
+ prompt: outcome.prepared.prompt,
3333
+ prepared: outcome.prepared
3334
+ });
3335
+ }
3336
+ } catch (error) {
3337
+ collector.emit(failedResultForItem(item, error, itemStartedAt));
3338
+ }
3339
+ }
3340
+ const plans = planBatchJobs(entries);
3341
+ const jobCount = pending.jobs.length + plans.length;
3342
+ const pendingCounts = countPendingItems(config, pending.jobs);
3343
+ emitResumedJobs(pending.jobs, pendingCounts, jobCount, options.onProgress);
3344
+ const submitted = await Promise.all(
3345
+ plans.map(async (plan, planIndex) => {
3346
+ try {
3347
+ return await submitBatchJobPlan(config, {
3348
+ plan,
3349
+ displayModel: normalizeGeminiDisplayName(plan.apiModel),
3350
+ jobIndex: pending.jobs.length + planIndex,
3351
+ jobCount,
3352
+ onProgress: options.onProgress
3353
+ });
3354
+ } catch (error) {
3355
+ for (const entry of plan.entries) {
3356
+ collector.emit(failedResultForItem(entry.prepared.item, error, startedAt));
3357
+ }
3358
+ return void 0;
3359
+ }
3360
+ })
3361
+ );
3362
+ const jobsToPoll = [
3363
+ ...pending.jobs,
3364
+ ...submitted.filter((row) => row !== void 0)
3365
+ ];
3366
+ if (jobsToPoll.length > 0) {
3367
+ await pollAndIngestBatchJobs(config, jobsToPoll, {
3368
+ jobCount,
3369
+ onProgress: options.onProgress,
3370
+ onResult: collector.emit
3371
+ });
3372
+ }
3373
+ }
3374
+ const results = collector.assemble();
3375
+ const totals = summarizeResults(results, Date.now() - startedAt);
3376
+ options.onProgress?.({ type: "done", results, totals });
3377
+ return results;
3378
+ }
3379
+ function countPendingItems(config, jobs) {
3380
+ if (jobs.length === 0) return /* @__PURE__ */ new Map();
3381
+ const db = openStore(config, "readonly");
3382
+ const counts = /* @__PURE__ */ new Map();
3383
+ for (const job of jobs) {
3384
+ counts.set(job.id, listBatchItems(db, job.id).filter((i) => i.status === "pending").length);
3385
+ }
3386
+ db.close();
3387
+ return counts;
3388
+ }
3389
+ function emitResumedJobs(jobs, counts, jobCount, onProgress) {
3390
+ jobs.forEach((job, jobIndex) => {
3391
+ onProgress?.({
3392
+ type: "batch-submitted",
3393
+ name: job.job_name,
3394
+ count: counts.get(job.id) ?? 0,
3395
+ model: job.display_model,
3396
+ jobIndex,
3397
+ jobCount,
3398
+ resumed: true,
3399
+ createdAt: job.created_at
3400
+ });
3401
+ });
3402
+ }
3403
+ async function resumeTranslationJobs(config, options = {}) {
3404
+ const startedAt = Date.now();
3405
+ const pending = readPendingBatchWork(config);
3406
+ if (pending.jobs.length === 0) return null;
3407
+ options.onProgress?.({
3408
+ type: "start",
3409
+ total: pending.pendingItems.length,
3410
+ concurrency: 1,
3411
+ dryRun: false,
3412
+ mode: "batch"
3413
+ });
3414
+ const counts = countPendingItems(config, pending.jobs);
3415
+ emitResumedJobs(pending.jobs, counts, pending.jobs.length, options.onProgress);
3416
+ const results = [];
3417
+ await pollAndIngestBatchJobs(config, pending.jobs, {
3418
+ onProgress: options.onProgress,
3419
+ onResult: (result) => {
3420
+ results.push(result);
3421
+ options.onProgress?.({ type: "item-done", result });
3422
+ }
2625
3423
  });
2626
3424
  const totals = summarizeResults(results, Date.now() - startedAt);
2627
3425
  options.onProgress?.({ type: "done", results, totals });
@@ -3239,35 +4037,43 @@ async function promptContentType(config) {
3239
4037
  })
3240
4038
  );
3241
4039
  }
4040
+ var ALL_LOCALES = /* @__PURE__ */ Symbol("all-locales");
4041
+ var PICK_LOCALES = /* @__PURE__ */ Symbol("pick-locales");
4042
+ async function promptManualLocales(config) {
4043
+ const locales = nonDefaultLocales(config);
4044
+ if (locales.length <= 1) return {};
4045
+ const picked = await runPrompt(
4046
+ checkbox({
4047
+ message: "Locales to translate",
4048
+ choices: locales.map((locale) => ({ name: locale, value: locale, checked: true })),
4049
+ validate: (values) => values.length > 0 || "Pick at least one locale"
4050
+ })
4051
+ );
4052
+ return picked.length === locales.length ? {} : { locale: picked };
4053
+ }
3242
4054
  async function promptLocalePreset(config) {
3243
4055
  const presets = Object.entries(config.localePresets ?? {}).filter(
3244
4056
  (entry) => Array.isArray(entry[1])
3245
4057
  );
3246
4058
  if (presets.length === 0) {
3247
- const locales = nonDefaultLocales(config);
3248
- if (locales.length <= 1) return {};
3249
- const picked = await runPrompt(
3250
- checkbox({
3251
- message: "Locales to translate",
3252
- choices: locales.map((locale) => ({ name: locale, value: locale, checked: true })),
3253
- validate: (values) => values.length > 0 || "Pick at least one locale"
3254
- })
3255
- );
3256
- return picked.length === locales.length ? {} : { locale: picked };
4059
+ return promptManualLocales(config);
3257
4060
  }
3258
4061
  const choice = await runPrompt(
3259
4062
  select({
3260
4063
  message: "Locale preset",
3261
4064
  choices: [
3262
- { name: "All locales", value: void 0 },
4065
+ { name: "All locales", value: ALL_LOCALES },
3263
4066
  ...presets.map(([name, locales]) => ({
3264
4067
  name: `${name} (${locales.join(", ")})`,
3265
4068
  value: name
3266
- }))
4069
+ })),
4070
+ { name: "Choose locales manually\u2026", value: PICK_LOCALES }
3267
4071
  ]
3268
4072
  })
3269
4073
  );
3270
- return choice ? { preset: choice } : {};
4074
+ if (choice === PICK_LOCALES) return promptManualLocales(config);
4075
+ if (choice === ALL_LOCALES) return {};
4076
+ return { preset: choice };
3271
4077
  }
3272
4078
  async function promptStrategy() {
3273
4079
  return runPrompt(
@@ -3281,12 +4087,25 @@ async function promptStrategy() {
3281
4087
  })
3282
4088
  );
3283
4089
  }
4090
+ async function promptMode() {
4091
+ return runPrompt(
4092
+ select({
4093
+ message: "Translation mode",
4094
+ choices: [
4095
+ { name: "Batch \u2014 50% cheaper, async, resumable (recommended)", value: "batch" },
4096
+ { name: "Direct \u2014 immediate results, full price", value: "direct" }
4097
+ ],
4098
+ default: "batch"
4099
+ })
4100
+ );
4101
+ }
3284
4102
  async function promptTranslateSelection(config, flags) {
3285
4103
  const selection = {
3286
4104
  contentType: flags.type,
3287
4105
  preset: flags.preset,
3288
4106
  locale: flags.locale,
3289
- strategy: flags.strategy
4107
+ strategy: flags.strategy,
4108
+ mode: flags.mode
3290
4109
  };
3291
4110
  if (!isInteractive()) return selection;
3292
4111
  if (!selection.contentType) {
@@ -3298,6 +4117,9 @@ async function promptTranslateSelection(config, flags) {
3298
4117
  if (!selection.strategy) {
3299
4118
  selection.strategy = await promptStrategy();
3300
4119
  }
4120
+ if (!selection.mode) {
4121
+ selection.mode = await promptMode();
4122
+ }
3301
4123
  return selection;
3302
4124
  }
3303
4125
 
@@ -3319,6 +4141,34 @@ function progressBar(ratio, width = 28) {
3319
4141
  function labelForResult(result) {
3320
4142
  return `${result.contentType}/${result.enSlug}@${result.locale}`;
3321
4143
  }
4144
+ function shortJobId(name) {
4145
+ const segments = name.split("/");
4146
+ return segments[segments.length - 1] || name;
4147
+ }
4148
+ function formatElapsed(elapsedMs) {
4149
+ return `${Math.round(elapsedMs / 1e3)}s`;
4150
+ }
4151
+ function batchDoneParts(event) {
4152
+ if (event.alreadyIngested) {
4153
+ return [
4154
+ `job ${shortJobId(event.name)}`,
4155
+ event.model ?? "?",
4156
+ "already ingested by another scribe run (results saved there)",
4157
+ formatElapsed(event.elapsedMs)
4158
+ ].join(" \xB7 ");
4159
+ }
4160
+ const parts = [
4161
+ `job ${shortJobId(event.name)}`,
4162
+ event.model ?? "?",
4163
+ `${event.count} request${event.count === 1 ? "" : "s"}`,
4164
+ `${event.translated} translated`
4165
+ ];
4166
+ if (event.failed > 0) parts.push(`${event.failed} failed`);
4167
+ parts.push(`${formatTokenCount(event.inputTokens)} in / ${formatTokenCount(event.outputTokens)} out`);
4168
+ parts.push(`${formatUsd(event.estimatedCostUsd)} est.`);
4169
+ parts.push(formatElapsed(event.elapsedMs));
4170
+ return parts.join(" \xB7 ");
4171
+ }
3322
4172
  function statusForResult(result, dryRun) {
3323
4173
  if (result.failed) return red("failed");
3324
4174
  if (result.skipped) return dim("skipped");
@@ -3364,6 +4214,25 @@ function createTranslateProgressReporter(options = {}) {
3364
4214
  if (!enabled) {
3365
4215
  return {
3366
4216
  onEvent(event) {
4217
+ if (event.type === "batch-submitted") {
4218
+ const verb = event.resumed ? "Resuming" : "Submitted";
4219
+ console.log(
4220
+ `${verb} batch job ${event.jobIndex + 1}/${event.jobCount} with ${event.count} request${event.count === 1 ? "" : "s"}${event.model ? ` (${event.model})` : ""}: ${event.name}`
4221
+ );
4222
+ return;
4223
+ }
4224
+ if (event.type === "batch-polling") {
4225
+ console.log(
4226
+ `batch ${event.jobIndex + 1}/${event.jobCount} ${shortJobId(event.name)}: ${event.state.replace(/^JOB_STATE_/, "")} (${formatElapsed(event.elapsedMs)} elapsed)`
4227
+ );
4228
+ return;
4229
+ }
4230
+ if (event.type === "batch-done") {
4231
+ const line = `batch ${event.jobIndex + 1}/${event.jobCount} ${event.state} \xB7 ${batchDoneParts(event)}`;
4232
+ if (event.failed > 0 && event.translated === 0) console.error(line);
4233
+ else console.log(line);
4234
+ return;
4235
+ }
3367
4236
  if (event.type === "item-done") {
3368
4237
  const label = labelForResult(event.result);
3369
4238
  const status = statusForResult(event.result, dryRun);
@@ -3394,6 +4263,9 @@ function createTranslateProgressReporter(options = {}) {
3394
4263
  let outputTokens = 0;
3395
4264
  let estimatedCostUsd = 0;
3396
4265
  let active = [];
4266
+ let mode;
4267
+ const batchJobs = /* @__PURE__ */ new Map();
4268
+ let batchElapsedMs = 0;
3397
4269
  const recent = [];
3398
4270
  let renderedLines = 0;
3399
4271
  let cursorHidden = false;
@@ -3409,6 +4281,15 @@ function createTranslateProgressReporter(options = {}) {
3409
4281
  cursorHidden = false;
3410
4282
  }
3411
4283
  }
4284
+ function printPersistent(lines) {
4285
+ if (renderedLines > 0) {
4286
+ process.stdout.write(`\x1B[${renderedLines}A`);
4287
+ }
4288
+ for (const line of lines) {
4289
+ process.stdout.write("\x1B[2K" + line + "\n");
4290
+ }
4291
+ renderedLines = 0;
4292
+ }
3412
4293
  function render() {
3413
4294
  hideCursor();
3414
4295
  if (renderedLines > 0) {
@@ -3416,12 +4297,34 @@ function createTranslateProgressReporter(options = {}) {
3416
4297
  }
3417
4298
  const lines = [];
3418
4299
  const title = dryRun ? "Dry run" : "Translating";
3419
- lines.push(cyan(`${title} ${done}/${total}`) + dim(` \xB7 ${concurrency} parallel`) + (model ? dim(` \xB7 ${model}`) : ""));
4300
+ const isBatch = mode === "batch" || batchJobs.size > 0;
4301
+ lines.push(
4302
+ cyan(`${title} ${done}/${total}`) + dim(isBatch ? " \xB7 batch" : ` \xB7 ${concurrency} parallel`) + (model ? dim(` \xB7 ${model}`) : "")
4303
+ );
3420
4304
  lines.push(progressBar(total === 0 ? 0 : done / total) + dim(` ${Math.round(total === 0 ? 0 : done / total * 100)}%`));
3421
4305
  lines.push(
3422
4306
  dim("Tokens ") + `${formatTokenCount(inputTokens)} in \xB7 ${formatTokenCount(outputTokens)} out` + dim(" \xB7 Cost ") + formatUsd(estimatedCostUsd) + dim(" est.")
3423
4307
  );
3424
- if (active.length > 0) {
4308
+ if (batchJobs.size > 0 && done < total) {
4309
+ const terminal = /^(SUCCEEDED|FAILED|CANCELLED|EXPIRED|PARTIALLY_SUCCEEDED)$/;
4310
+ let running = 0;
4311
+ let finished = 0;
4312
+ let requests = 0;
4313
+ for (const job of batchJobs.values()) {
4314
+ if (terminal.test(job.state)) finished += 1;
4315
+ else running += 1;
4316
+ requests += job.count;
4317
+ }
4318
+ lines.push(
4319
+ dim("Batches ") + `${running} running \xB7 ${finished} done` + dim(` \xB7 ${requests} request${requests === 1 ? "" : "s"} \xB7 ${Math.round(batchElapsedMs / 1e3)}s elapsed`)
4320
+ );
4321
+ for (const [name, job] of batchJobs) {
4322
+ if (terminal.test(job.state)) continue;
4323
+ lines.push(
4324
+ dim(" ") + shortJobId(name) + dim(` \xB7 ${job.model ?? "?"} \xB7 ${job.count} req \xB7 `) + job.state + dim(` \xB7 ${formatElapsed(job.elapsedMs)}`)
4325
+ );
4326
+ }
4327
+ } else if (active.length > 0) {
3425
4328
  lines.push(dim("Active ") + active.slice(0, 3).join(", ") + (active.length > 3 ? dim(` +${active.length - 3}`) : ""));
3426
4329
  } else if (done < total) {
3427
4330
  lines.push(dim("Active ") + "starting\u2026");
@@ -3449,12 +4352,58 @@ function createTranslateProgressReporter(options = {}) {
3449
4352
  total = event.total;
3450
4353
  concurrency = event.concurrency;
3451
4354
  model = event.model;
4355
+ mode = event.mode;
3452
4356
  render();
3453
4357
  break;
3454
4358
  case "item-start":
3455
4359
  active = event.active;
3456
4360
  render();
3457
4361
  break;
4362
+ case "batch-submitted": {
4363
+ const createdAtMs = event.createdAt ? Date.parse(event.createdAt) : void 0;
4364
+ const elapsedMs = createdAtMs && !Number.isNaN(createdAtMs) ? Date.now() - createdAtMs : 0;
4365
+ batchJobs.set(event.name, {
4366
+ state: event.resumed ? "RESUMED" : "SUBMITTED",
4367
+ count: event.count,
4368
+ model: event.model,
4369
+ createdAtMs: Number.isNaN(createdAtMs) ? void 0 : createdAtMs,
4370
+ elapsedMs
4371
+ });
4372
+ const requests = `${event.count} ${event.resumed ? "pending " : ""}request${event.count === 1 ? "" : "s"}`;
4373
+ const origin = event.resumed ? `resumed job (submitted ${formatElapsed(elapsedMs)} ago)` : "submitted job";
4374
+ printPersistent([
4375
+ dim(`${event.resumed ? "\u21BB" : "\u2192"} ${origin} ${shortJobId(event.name)} \xB7 ${event.model ?? "?"} \xB7 ${requests}`)
4376
+ ]);
4377
+ render();
4378
+ break;
4379
+ }
4380
+ case "batch-polling": {
4381
+ const job = batchJobs.get(event.name);
4382
+ batchJobs.set(event.name, {
4383
+ state: event.state.replace(/^JOB_STATE_/, ""),
4384
+ count: job?.count ?? 0,
4385
+ model: job?.model,
4386
+ createdAtMs: job?.createdAtMs,
4387
+ elapsedMs: event.elapsedMs
4388
+ });
4389
+ batchElapsedMs = Math.max(batchElapsedMs, event.elapsedMs);
4390
+ render();
4391
+ break;
4392
+ }
4393
+ case "batch-done": {
4394
+ const job = batchJobs.get(event.name);
4395
+ batchJobs.set(event.name, {
4396
+ state: event.state.replace(/^JOB_STATE_/, ""),
4397
+ count: job?.count ?? event.count,
4398
+ model: job?.model ?? event.model,
4399
+ createdAtMs: job?.createdAtMs,
4400
+ elapsedMs: event.elapsedMs
4401
+ });
4402
+ const marker = event.failed > 0 && event.translated === 0 ? red("\u2717") : green("\u2713");
4403
+ printPersistent([`${marker} ${dim(batchDoneParts(event))}`]);
4404
+ render();
4405
+ break;
4406
+ }
3458
4407
  case "item-done": {
3459
4408
  done += 1;
3460
4409
  active = active.filter((label) => label !== labelForResult(event.result));
@@ -3577,6 +4526,21 @@ function parseArgs(argv) {
3577
4526
  i++;
3578
4527
  continue;
3579
4528
  }
4529
+ if (arg === "--batch") {
4530
+ options.batch = true;
4531
+ i++;
4532
+ continue;
4533
+ }
4534
+ if (arg === "--direct") {
4535
+ options.direct = true;
4536
+ i++;
4537
+ continue;
4538
+ }
4539
+ if (arg === "--resume") {
4540
+ options.resume = true;
4541
+ i++;
4542
+ continue;
4543
+ }
3580
4544
  if (arg === "--strategy") {
3581
4545
  const value = argv[++i];
3582
4546
  if (value !== "all" && value !== "missing-only") {
@@ -3645,11 +4609,19 @@ Translate flags:
3645
4609
  --locale <code>... Target locale(s); overrides --preset
3646
4610
  --slug <en-slug> Single English document
3647
4611
  --model <id> Gemini model override
3648
- --concurrency <n> Parallel translations (default: 3)
4612
+ --batch Force Gemini Batch API mode (50% token cost; default)
4613
+ --direct Force direct interactive calls (immediate, full price)
4614
+ --resume Only poll/ingest pending batch jobs; submit nothing new
4615
+ --concurrency <n> Parallel translations in --direct mode (default: 3)
3649
4616
  --dry-run List work without writing
3650
4617
  --force Re-translate even when hashes match
3651
4618
  --strategy <mode> all (default) or missing-only
3652
4619
  --no-progress Plain line logging instead of live progress
4620
+
4621
+ Batch mode submits jobs upfront and persists them in the store, so Ctrl+C
4622
+ during polling is safe: the next run (or --resume) picks the jobs back up and
4623
+ ingests their results. In a TTY without --batch/--direct, a prompt asks which
4624
+ mode to use.
3653
4625
  `);
3654
4626
  return;
3655
4627
  }
@@ -3695,11 +4667,29 @@ Translate flags:
3695
4667
  break;
3696
4668
  }
3697
4669
  case "translate": {
4670
+ if (options.batch && options.direct) {
4671
+ throw new Error("--batch and --direct are mutually exclusive");
4672
+ }
4673
+ if (options.resume) {
4674
+ const reporter2 = createTranslateProgressReporter({
4675
+ enabled: !options.noProgress,
4676
+ dryRun: false
4677
+ });
4678
+ const results2 = await resumeTranslationJobs(config, { onProgress: reporter2.onEvent });
4679
+ reporter2.finish();
4680
+ if (results2 === null) {
4681
+ console.log("No pending translation batch jobs.");
4682
+ break;
4683
+ }
4684
+ if (results2.some((result) => result.failed)) process.exitCode = 1;
4685
+ break;
4686
+ }
3698
4687
  const selection = await promptTranslateSelection(config, {
3699
4688
  type: options.type,
3700
4689
  preset: options.preset,
3701
4690
  locale: options.locale,
3702
- strategy: options.strategy
4691
+ strategy: options.strategy,
4692
+ mode: options.batch ? "batch" : options.direct ? "direct" : void 0
3703
4693
  });
3704
4694
  const locales = resolveLocalesFromPreset(config, selection.preset, selection.locale);
3705
4695
  const worklist = buildWorklist(config, {
@@ -3709,6 +4699,18 @@ Translate flags:
3709
4699
  strategy: selection.strategy ?? "all"
3710
4700
  });
3711
4701
  if (worklist.length === 0) {
4702
+ if (!options.dryRun) {
4703
+ const reporter2 = createTranslateProgressReporter({
4704
+ enabled: !options.noProgress,
4705
+ dryRun: false
4706
+ });
4707
+ const resumed = await resumeTranslationJobs(config, { onProgress: reporter2.onEvent });
4708
+ reporter2.finish();
4709
+ if (resumed !== null) {
4710
+ if (resumed.some((result) => result.failed)) process.exitCode = 1;
4711
+ break;
4712
+ }
4713
+ }
3712
4714
  console.log("Nothing to translate.");
3713
4715
  break;
3714
4716
  }
@@ -3721,6 +4723,7 @@ Translate flags:
3721
4723
  dryRun: options.dryRun,
3722
4724
  force: options.force,
3723
4725
  concurrency: options.concurrency,
4726
+ mode: selection.mode,
3724
4727
  onProgress: reporter.onEvent
3725
4728
  });
3726
4729
  reporter.finish();