scribe-cms 0.0.11 → 0.0.13

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 (56) hide show
  1. package/README.md +21 -6
  2. package/dist/cli/index.cjs +1258 -219
  3. package/dist/cli/index.cjs.map +1 -1
  4. package/dist/cli/index.js +1259 -220
  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 +1020 -181
  10. package/dist/index.cjs.map +1 -1
  11. package/dist/index.js +1021 -183
  12. package/dist/index.js.map +1 -1
  13. package/dist/runtime.cjs +149 -62
  14. package/dist/runtime.cjs.map +1 -1
  15. package/dist/runtime.js +149 -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 +23 -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/src/validate/validate-project.d.ts.map +1 -1
  52. package/dist/studio/server.cjs +84 -47
  53. package/dist/studio/server.cjs.map +1 -1
  54. package/dist/studio/server.js +84 -47
  55. package/dist/studio/server.js.map +1 -1
  56. 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 +343,39 @@ 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 = {};
347
+ for (const [locale, chain] of Object.entries(raw.localeFallbacks ?? {})) {
348
+ if (!raw.locales.includes(locale)) {
349
+ throw new Error(
350
+ `scribe config: localeFallbacks key "${locale}" is not in locales [${raw.locales.join(", ")}]`
351
+ );
352
+ }
353
+ const seen = /* @__PURE__ */ new Set();
354
+ for (const fallback of chain) {
355
+ if (!raw.locales.includes(fallback)) {
356
+ throw new Error(
357
+ `scribe config: localeFallbacks["${locale}"] entry "${fallback}" is not in locales [${raw.locales.join(", ")}]`
358
+ );
359
+ }
360
+ if (fallback === locale) {
361
+ throw new Error(
362
+ `scribe config: localeFallbacks["${locale}"] must not contain its own key "${locale}"`
363
+ );
364
+ }
365
+ if (fallback === defaultLocale) {
366
+ throw new Error(
367
+ `scribe config: localeFallbacks["${locale}"] must not contain the defaultLocale "${defaultLocale}" \u2014 it is always the final fallback; remove it`
368
+ );
369
+ }
370
+ if (seen.has(fallback)) {
371
+ throw new Error(
372
+ `scribe config: localeFallbacks["${locale}"] contains duplicate entry "${fallback}"`
373
+ );
374
+ }
375
+ seen.add(fallback);
376
+ }
377
+ localeFallbacks[locale] = [...chain];
378
+ }
341
379
  const projectRoot = path4.resolve(baseDir ?? process.cwd(), raw.rootDir);
342
380
  const contentRoot = path4.resolve(projectRoot, raw.contentDir ?? "content");
343
381
  const storePath = path4.resolve(projectRoot, raw.store ?? ".scribe/store.sqlite");
@@ -367,6 +405,7 @@ function resolveConfig(input, baseDir) {
367
405
  defaultLocale,
368
406
  localeRouting,
369
407
  localePresets: raw.localePresets,
408
+ localeFallbacks,
370
409
  translate: raw.translate,
371
410
  types
372
411
  };
@@ -382,6 +421,12 @@ init_esm_shims();
382
421
 
383
422
  // src/core/introspect-schema.ts
384
423
  init_esm_shims();
424
+ function getArrayElement(schema) {
425
+ if (schema instanceof Object && "element" in schema && schema._def?.type === "array") {
426
+ return schema.element;
427
+ }
428
+ return null;
429
+ }
385
430
  function introspectSchema(schema, prefix = []) {
386
431
  const relation = getRelationTarget(schema);
387
432
  if (relation && prefix.length > 0) {
@@ -395,7 +440,10 @@ function introspectSchema(schema, prefix = []) {
395
440
  }
396
441
  ];
397
442
  }
398
- const base = unwrapSchema(schema);
443
+ if (prefix.length > 0 && getFieldKind(schema) === "translatable") {
444
+ return [{ path: prefix, kind: "translatable" }];
445
+ }
446
+ const base = peelOptionalWrappers(schema);
399
447
  if (base instanceof Object && "shape" in base) {
400
448
  const shape = base.shape;
401
449
  const fields = [];
@@ -404,9 +452,9 @@ function introspectSchema(schema, prefix = []) {
404
452
  }
405
453
  return fields;
406
454
  }
407
- if (base instanceof Object && "element" in base) {
408
- const element = base.element;
409
- const elementBase = unwrapSchema(element);
455
+ const element = getArrayElement(base);
456
+ if (element) {
457
+ const elementBase = peelOptionalWrappers(element);
410
458
  if (elementBase instanceof Object && "shape" in elementBase) {
411
459
  const shape = elementBase.shape;
412
460
  const fields = [];
@@ -418,15 +466,48 @@ function introspectSchema(schema, prefix = []) {
418
466
  }
419
467
  return [{ path: prefix, kind: getFieldKind(schema) }];
420
468
  }
421
- function extractByPaths(data, paths) {
422
- const out = {};
469
+ function buildPathTrie(paths) {
470
+ const root = { leaf: false, children: /* @__PURE__ */ new Map() };
423
471
  for (const path16 of paths) {
424
- const value = getAtPath(data, path16);
472
+ let node = root;
473
+ for (const segment of path16) {
474
+ let child = node.children.get(segment);
475
+ if (!child) {
476
+ child = { leaf: false, children: /* @__PURE__ */ new Map() };
477
+ node.children.set(segment, child);
478
+ }
479
+ node = child;
480
+ }
481
+ node.leaf = true;
482
+ }
483
+ return root;
484
+ }
485
+ function extractNode(data, trie) {
486
+ if (trie.leaf) return data;
487
+ const star = trie.children.get("*");
488
+ if (star) {
489
+ if (!Array.isArray(data)) return void 0;
490
+ return data.map(
491
+ (item) => typeof item === "object" && item !== null ? extractNode(item, star) ?? {} : {}
492
+ );
493
+ }
494
+ if (typeof data !== "object" || data === null || Array.isArray(data)) return void 0;
495
+ const record = data;
496
+ const out = {};
497
+ let any = false;
498
+ for (const [key, child] of trie.children) {
499
+ const value = extractNode(record[key], child);
425
500
  if (value !== void 0) {
426
- setAtPath(out, path16.filter((p) => p !== "*"), value);
501
+ out[key] = value;
502
+ any = true;
427
503
  }
428
504
  }
429
- return out;
505
+ return any ? out : void 0;
506
+ }
507
+ function extractByPaths(data, paths) {
508
+ if (paths.length === 0) return {};
509
+ const extracted = extractNode(data, buildPathTrie(paths));
510
+ return extracted && typeof extracted === "object" && !Array.isArray(extracted) ? extracted : {};
430
511
  }
431
512
  function pickTranslatable(data, schema) {
432
513
  const translatablePaths = introspectSchema(schema).filter((f) => f.kind === "translatable").map((f) => f.path);
@@ -441,32 +522,6 @@ function mergeStructuralOntoLocale(localeData, enData, schema) {
441
522
  const translatable = pickTranslatable(localeData, schema);
442
523
  return deepMerge(structural, translatable);
443
524
  }
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
525
  function deepMerge(base, overlay) {
471
526
  const out = { ...base };
472
527
  for (const [key, value] of Object.entries(overlay)) {
@@ -757,7 +812,8 @@ function sanitizeMdxJsxAttributeQuotes(body) {
757
812
  if (i >= body.length || body[i] === ">" || body[i] === "/") break;
758
813
  const attrStart = i;
759
814
  while (i < body.length && /[\w:.-]/.test(body[i] ?? "")) i += 1;
760
- body.slice(attrStart, i);
815
+ const attrName = body.slice(attrStart, i);
816
+ if (!attrName) break;
761
817
  while (i < body.length && /\s/.test(body[i] ?? "")) i += 1;
762
818
  if (body[i] !== "=") {
763
819
  tagOut += body.slice(attrStart, i);
@@ -766,10 +822,12 @@ function sanitizeMdxJsxAttributeQuotes(body) {
766
822
  tagOut += body.slice(attrStart, i);
767
823
  tagOut += "=";
768
824
  i += 1;
825
+ const wsStart = i;
769
826
  while (i < body.length && /\s/.test(body[i] ?? "")) i += 1;
827
+ tagOut += body.slice(wsStart, i);
770
828
  const quote = body[i];
771
829
  if (quote !== '"' && quote !== "'") {
772
- continue;
830
+ break;
773
831
  }
774
832
  if (quote === "'") {
775
833
  const valStart2 = i + 1;
@@ -1091,7 +1149,7 @@ function getTranslatablePayload(doc, type) {
1091
1149
 
1092
1150
  // src/i18n/resolve-document.ts
1093
1151
  init_esm_shims();
1094
- function resolveLocalizedDocument(slug, locale, defaultLocale, allDocs, type, localeRouting) {
1152
+ function resolveLocalizedDocument(slug, locale, defaultLocale, allDocs, type, localeRouting, fallbackLocales = []) {
1095
1153
  const urlBuilder = createUrlBuilder({
1096
1154
  locales: [defaultLocale, locale],
1097
1155
  defaultLocale,
@@ -1105,15 +1163,28 @@ function resolveLocalizedDocument(slug, locale, defaultLocale, allDocs, type, lo
1105
1163
  for (const [docLocale, docIdx] of allDocs) {
1106
1164
  const found = docIdx.bySlug.get(slug);
1107
1165
  if (!found) continue;
1108
- const correctSlug = getSlugForLocale(found, docLocale, locale, allDocs, defaultLocale);
1109
- if (correctSlug && correctSlug !== slug) {
1166
+ const enSlug = docLocale === defaultLocale ? found.slug : found.enSlug;
1167
+ let target;
1168
+ let targetLocale = locale;
1169
+ for (const candidateLocale of [locale, ...fallbackLocales]) {
1170
+ const cand = candidateLocale === defaultLocale ? allDocs.get(candidateLocale)?.bySlug.get(enSlug) : allDocs.get(candidateLocale)?.byEnSlug.get(enSlug);
1171
+ if (cand) {
1172
+ target = cand;
1173
+ targetLocale = candidateLocale;
1174
+ break;
1175
+ }
1176
+ }
1177
+ if (target) {
1178
+ if (target.slug === slug) {
1179
+ return { document: target, actualLocale: targetLocale };
1180
+ }
1110
1181
  if (!type.path) {
1111
1182
  return { document: null, actualLocale: locale };
1112
1183
  }
1113
1184
  return {
1114
1185
  document: null,
1115
1186
  actualLocale: locale,
1116
- shouldRedirectTo: urlBuilder.resolvePath(type.path, correctSlug, locale)
1187
+ shouldRedirectTo: urlBuilder.resolvePath(type.path, target.slug, locale)
1117
1188
  };
1118
1189
  }
1119
1190
  if (docLocale === defaultLocale) {
@@ -1144,13 +1215,6 @@ function resolveLocalizedDocument(slug, locale, defaultLocale, allDocs, type, lo
1144
1215
  }
1145
1216
  return { document: null, actualLocale: locale };
1146
1217
  }
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
1218
 
1155
1219
  // src/create-project.ts
1156
1220
  function comparatorFor(orderBy) {
@@ -1203,7 +1267,8 @@ function buildRuntime(config, type, getRuntime) {
1203
1267
  config.defaultLocale,
1204
1268
  load(),
1205
1269
  type,
1206
- config.localeRouting
1270
+ config.localeRouting,
1271
+ config.localeFallbacks[locale] ?? []
1207
1272
  );
1208
1273
  if (result.document && type.path) {
1209
1274
  return {
@@ -1225,8 +1290,14 @@ function buildRuntime(config, type, getRuntime) {
1225
1290
  const params = [];
1226
1291
  for (const locale of options.locales ?? config.locales) {
1227
1292
  const localeIdx = all.get(locale);
1293
+ const fallbacks = config.localeFallbacks[locale] ?? [];
1228
1294
  for (const doc of enIdx.bySlug.values()) {
1229
- const slug = locale === config.defaultLocale ? doc.slug : localeIdx?.byEnSlug.get(doc.slug)?.slug ?? doc.slug;
1295
+ let slug;
1296
+ if (locale === config.defaultLocale) {
1297
+ slug = doc.slug;
1298
+ } else {
1299
+ 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;
1300
+ }
1230
1301
  params.push({ locale, slug });
1231
1302
  }
1232
1303
  }
@@ -1688,7 +1759,7 @@ init_sqlite();
1688
1759
 
1689
1760
  // src/validate/validate-relations.ts
1690
1761
  init_esm_shims();
1691
- function getAtPath2(obj, fieldPath) {
1762
+ function getAtPath(obj, fieldPath) {
1692
1763
  let current = obj;
1693
1764
  for (const segment of fieldPath) {
1694
1765
  if (typeof current !== "object" || current === null || Array.isArray(current)) return void 0;
@@ -1733,7 +1804,7 @@ function validateRelations(project) {
1733
1804
  });
1734
1805
  continue;
1735
1806
  }
1736
- const value = getAtPath2(doc.frontmatter, fieldMeta.path);
1807
+ const value = getAtPath(doc.frontmatter, fieldMeta.path);
1737
1808
  if (value === void 0 || value === null || value === "") continue;
1738
1809
  const targetSlugs = slugIndex.get(targetTypeId) ?? /* @__PURE__ */ new Set();
1739
1810
  const fieldLabel = fieldMeta.path.join(".");
@@ -1889,6 +1960,20 @@ function listEnSlugs3(rootDir, contentDir) {
1889
1960
  if (!fs2.existsSync(dir)) return [];
1890
1961
  return fs2.readdirSync(dir).filter(isPublishableContentFile).map((f) => f.replace(/\.(md|mdx)$/, ""));
1891
1962
  }
1963
+ function validateDocumentMdxBody(issues, input) {
1964
+ const preparedBody = prepareTranslatedMdxBody(input.body).body;
1965
+ const mdxValidation = validateMdxBody(preparedBody);
1966
+ if (!mdxValidation.ok) {
1967
+ issues.push({
1968
+ level: "error",
1969
+ contentType: input.contentType,
1970
+ enSlug: input.enSlug,
1971
+ locale: input.locale,
1972
+ field: "body",
1973
+ message: `Invalid MDX: ${mdxValidation.error}`
1974
+ });
1975
+ }
1976
+ }
1892
1977
  function validateProject(config) {
1893
1978
  const issues = [];
1894
1979
  const project = createProject(config);
@@ -1947,6 +2032,12 @@ function validateProject(config) {
1947
2032
  })) {
1948
2033
  issues.push(issue);
1949
2034
  }
2035
+ validateDocumentMdxBody(issues, {
2036
+ contentType: type.id,
2037
+ enSlug,
2038
+ locale: config.defaultLocale,
2039
+ body: enDoc.content
2040
+ });
1950
2041
  for (const locale of config.locales) {
1951
2042
  if (locale === config.defaultLocale) continue;
1952
2043
  const row = getTranslation(db, type.id, enSlug, locale);
@@ -1972,17 +2063,12 @@ function validateProject(config) {
1972
2063
  })) {
1973
2064
  issues.push(issue);
1974
2065
  }
1975
- const mdxValidation = validateMdxBody(preparedBody);
1976
- if (!mdxValidation.ok) {
1977
- issues.push({
1978
- level: "error",
1979
- contentType: type.id,
1980
- enSlug,
1981
- locale,
1982
- field: "body",
1983
- message: `Invalid MDX: ${mdxValidation.error}`
1984
- });
1985
- }
2066
+ validateDocumentMdxBody(issues, {
2067
+ contentType: type.id,
2068
+ enSlug,
2069
+ locale,
2070
+ body: row.body
2071
+ });
1986
2072
  }
1987
2073
  }
1988
2074
  try {
@@ -2102,6 +2188,66 @@ function resolveLocalesFromPreset(config, preset, explicitLocales) {
2102
2188
  // src/translate/page-translator.ts
2103
2189
  init_esm_shims();
2104
2190
 
2191
+ // src/storage/batch-jobs.ts
2192
+ init_esm_shims();
2193
+ function insertBatchJob(db, input) {
2194
+ const info = db.prepare(
2195
+ `INSERT INTO translation_batch_jobs (job_name, model, display_model, created_at, state)
2196
+ VALUES (?, ?, ?, ?, ?)`
2197
+ ).run(input.jobName, input.model, input.displayModel, input.createdAt, input.state);
2198
+ return Number(info.lastInsertRowid);
2199
+ }
2200
+ function insertBatchItems(db, jobId, items) {
2201
+ const stmt = db.prepare(
2202
+ `INSERT INTO translation_batch_items (
2203
+ job_id, request_index, content_type, en_slug, locale, en_hash, snapshot_id, status
2204
+ ) VALUES (?, ?, ?, ?, ?, ?, ?, 'pending')`
2205
+ );
2206
+ for (const item of items) {
2207
+ stmt.run(
2208
+ jobId,
2209
+ item.requestIndex,
2210
+ item.contentType,
2211
+ item.enSlug,
2212
+ item.locale,
2213
+ item.enHash,
2214
+ item.snapshotId
2215
+ );
2216
+ }
2217
+ }
2218
+ function listPendingBatchJobs(db) {
2219
+ return db.prepare(`SELECT * FROM translation_batch_jobs WHERE completed_at IS NULL ORDER BY id`).all();
2220
+ }
2221
+ function listBatchItems(db, jobId) {
2222
+ return db.prepare(`SELECT * FROM translation_batch_items WHERE job_id = ? ORDER BY request_index`).all(jobId);
2223
+ }
2224
+ function listPendingBatchItems(db) {
2225
+ return db.prepare(
2226
+ `SELECT i.* FROM translation_batch_items i
2227
+ JOIN translation_batch_jobs j ON j.id = i.job_id
2228
+ WHERE j.completed_at IS NULL AND i.status = 'pending'
2229
+ ORDER BY i.job_id, i.request_index`
2230
+ ).all();
2231
+ }
2232
+ function claimBatchJobCompletion(db, jobId, state, completedAt) {
2233
+ const info = db.prepare(
2234
+ `UPDATE translation_batch_jobs SET state = ?, completed_at = ?
2235
+ WHERE id = ? AND completed_at IS NULL`
2236
+ ).run(state, completedAt, jobId);
2237
+ return info.changes > 0;
2238
+ }
2239
+ function updateBatchItemStatus(db, jobId, requestIndex, status, error) {
2240
+ db.prepare(
2241
+ `UPDATE translation_batch_items SET status = ?, error = ? WHERE job_id = ? AND request_index = ?`
2242
+ ).run(status, error ?? null, jobId, requestIndex);
2243
+ }
2244
+
2245
+ // src/translate/page-translator.ts
2246
+ init_sqlite();
2247
+
2248
+ // src/translate/batch-worklist.ts
2249
+ init_esm_shims();
2250
+
2105
2251
  // src/history/record-snapshot.ts
2106
2252
  init_esm_shims();
2107
2253
  init_sqlite();
@@ -2119,9 +2265,111 @@ function recordEnSnapshot(config, input, db) {
2119
2265
  return id;
2120
2266
  }
2121
2267
 
2122
- // src/translate/page-translator.ts
2268
+ // src/translate/batch-worklist.ts
2123
2269
  init_sqlite();
2124
2270
 
2271
+ // src/translate/gemini-batch.ts
2272
+ init_esm_shims();
2273
+
2274
+ // src/translate/retry.ts
2275
+ init_esm_shims();
2276
+ var TRANSIENT_STATUS = /* @__PURE__ */ new Set([429, 500, 502, 503, 504]);
2277
+ function statusFromError(error) {
2278
+ if (error instanceof ApiError && typeof error.status === "number") {
2279
+ return error.status;
2280
+ }
2281
+ if (error && typeof error === "object" && "status" in error) {
2282
+ const status = error.status;
2283
+ if (typeof status === "number") return status;
2284
+ }
2285
+ const message = error instanceof Error ? error.message : String(error);
2286
+ const match = message.match(/\b(429|500|502|503|504)\b/);
2287
+ if (match) return Number(match[1]);
2288
+ return void 0;
2289
+ }
2290
+ function isNetworkError(error) {
2291
+ const err = error;
2292
+ const code = typeof err?.code === "string" ? err.code : "";
2293
+ if (code === "ECONNRESET" || code === "ETIMEDOUT" || code === "ECONNREFUSED" || code === "ENOTFOUND" || code === "EAI_AGAIN" || code === "EPIPE" || code === "UND_ERR_SOCKET") {
2294
+ return true;
2295
+ }
2296
+ const message = error instanceof Error ? error.message.toLowerCase() : "";
2297
+ return message.includes("network") || message.includes("fetch failed") || message.includes("socket hang up") || message.includes("terminated");
2298
+ }
2299
+ function isTransientError(error) {
2300
+ const status = statusFromError(error);
2301
+ if (status !== void 0) return TRANSIENT_STATUS.has(status);
2302
+ return isNetworkError(error);
2303
+ }
2304
+ var defaultSleep = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
2305
+ async function withRetry(fn, options = {}) {
2306
+ const attempts = Math.max(1, options.attempts ?? 3);
2307
+ const baseDelayMs = options.baseDelayMs ?? 1e3;
2308
+ const sleep2 = options.sleep ?? defaultSleep;
2309
+ const random = options.random ?? Math.random;
2310
+ let lastError;
2311
+ for (let attempt = 1; attempt <= attempts; attempt++) {
2312
+ try {
2313
+ return await fn();
2314
+ } catch (error) {
2315
+ lastError = error;
2316
+ if (attempt >= attempts || !isTransientError(error)) throw error;
2317
+ const backoff = baseDelayMs * 2 ** (attempt - 1);
2318
+ const delay = backoff + random() * backoff;
2319
+ await sleep2(delay);
2320
+ }
2321
+ }
2322
+ throw lastError;
2323
+ }
2324
+
2325
+ // src/translate/gemini-batch.ts
2326
+ var STATE_FAMILY_PREFIX = /^(JOB_STATE_|BATCH_STATE_)/;
2327
+ var TERMINAL_STATES = /* @__PURE__ */ new Set([
2328
+ "SUCCEEDED",
2329
+ "FAILED",
2330
+ "CANCELLED",
2331
+ "EXPIRED",
2332
+ "PARTIALLY_SUCCEEDED"
2333
+ ]);
2334
+ var SUCCESSFUL_STATES = /* @__PURE__ */ new Set(["SUCCEEDED", "PARTIALLY_SUCCEEDED"]);
2335
+ function normalizeBatchState(state) {
2336
+ return String(state ?? "UNKNOWN").replace(STATE_FAMILY_PREFIX, "");
2337
+ }
2338
+ function isTerminalBatchState(state) {
2339
+ return TERMINAL_STATES.has(normalizeBatchState(state));
2340
+ }
2341
+ function isSuccessfulBatchState(state) {
2342
+ return SUCCESSFUL_STATES.has(normalizeBatchState(state));
2343
+ }
2344
+ function makeClient(apiKey) {
2345
+ const key = apiKey ?? process.env.GEMINI_API_KEY;
2346
+ if (!key) {
2347
+ throw new Error("GEMINI_API_KEY is required for scribe translate");
2348
+ }
2349
+ return new GoogleGenAI({ apiKey: key });
2350
+ }
2351
+ function textFromBatchResponse(response) {
2352
+ if (typeof response.text === "string") return response.text;
2353
+ const parts = response.candidates?.[0]?.content?.parts ?? [];
2354
+ return parts.filter((part) => !part.thought && typeof part.text === "string").map((part) => part.text).join("");
2355
+ }
2356
+ async function createGeminiBatchJob(input) {
2357
+ const ai = makeClient(input.apiKey);
2358
+ const job = await withRetry(
2359
+ () => ai.batches.create({
2360
+ model: input.model,
2361
+ src: input.requests,
2362
+ config: { displayName: input.displayName ?? `scribe-translate-${Date.now()}` }
2363
+ })
2364
+ );
2365
+ if (!job.name) throw new Error("Gemini batch job was created without a name");
2366
+ return job;
2367
+ }
2368
+ async function getGeminiBatchJob(input) {
2369
+ const ai = makeClient(input.apiKey);
2370
+ return withRetry(() => ai.batches.get({ name: input.name }));
2371
+ }
2372
+
2125
2373
  // src/translate/gemini-client.ts
2126
2374
  init_esm_shims();
2127
2375
 
@@ -2146,16 +2394,55 @@ function normalizeGeminiDisplayName(model) {
2146
2394
  if (alias) return alias[0];
2147
2395
  return name;
2148
2396
  }
2397
+ function resolveThinkingConfig(apiModelId) {
2398
+ const id = stripModelsPrefix(apiModelId).toLowerCase();
2399
+ if (id.startsWith("gemini-3")) return { thinkingLevel: ThinkingLevel.LOW };
2400
+ if (id.includes("2.5")) return { thinkingBudget: 128 };
2401
+ return void 0;
2402
+ }
2149
2403
 
2150
2404
  // src/translate/gemini-client.ts
2151
2405
  var DEFAULT_MODEL = DEFAULT_GEMINI_MODEL;
2152
2406
  function extractJson(text) {
2153
- const fenced = text.match(/```(?:json)?\s*([\s\S]*?)```/);
2407
+ const trimmed = text.trim();
2408
+ const fenced = trimmed.match(/^```(?:json)?\s*\n?([\s\S]*?)\n?```$/);
2154
2409
  if (fenced?.[1]) return fenced[1].trim();
2155
- const start = text.indexOf("{");
2156
- const end = text.lastIndexOf("}");
2157
- if (start >= 0 && end > start) return text.slice(start, end + 1);
2158
- return text.trim();
2410
+ const start = trimmed.indexOf("{");
2411
+ const end = trimmed.lastIndexOf("}");
2412
+ if (start >= 0 && end > start) return trimmed.slice(start, end + 1);
2413
+ return trimmed;
2414
+ }
2415
+ function parseGeminiResponse(text) {
2416
+ let parsed;
2417
+ try {
2418
+ parsed = JSON.parse(text.trim());
2419
+ } catch {
2420
+ parsed = JSON.parse(extractJson(text));
2421
+ }
2422
+ if (parsed !== null && typeof parsed === "object" && parsed.frontmatter === void 0) {
2423
+ parsed.frontmatter = {};
2424
+ }
2425
+ return parsed;
2426
+ }
2427
+ function buildGeminiRequestConfig(input) {
2428
+ const thinkingConfig = resolveThinkingConfig(input.apiModelId);
2429
+ return {
2430
+ responseMimeType: "application/json",
2431
+ ...input.responseSchema ? { responseSchema: input.responseSchema } : {},
2432
+ ...thinkingConfig ? { thinkingConfig } : {}
2433
+ };
2434
+ }
2435
+ function usageFromResponse(response) {
2436
+ const meta = response.usageMetadata ?? response.usage_metadata ?? {};
2437
+ const thoughtsTokens = meta.thoughtsTokenCount ?? meta.thoughts_token_count ?? 0;
2438
+ return {
2439
+ inputTokens: meta.promptTokenCount ?? meta.prompt_token_count ?? 0,
2440
+ // candidatesTokenCount does not include thoughts, but thoughts are billed as
2441
+ // output tokens, so fold them in here for accurate cost accounting.
2442
+ outputTokens: (meta.candidatesTokenCount ?? meta.candidates_token_count ?? 0) + thoughtsTokens,
2443
+ thoughtsTokens,
2444
+ totalTokens: meta.totalTokenCount ?? meta.total_token_count ?? 0
2445
+ };
2159
2446
  }
2160
2447
  async function translatePageWithGemini(input) {
2161
2448
  const apiKey = input.apiKey ?? process.env.GEMINI_API_KEY;
@@ -2167,28 +2454,28 @@ async function translatePageWithGemini(input) {
2167
2454
  );
2168
2455
  const apiModel = resolveGeminiModelId(displayModel);
2169
2456
  const ai = new GoogleGenAI({ apiKey });
2170
- const response = await ai.models.generateContent({
2171
- model: apiModel,
2172
- contents: input.prompt,
2173
- config: {
2174
- responseMimeType: "application/json",
2175
- ...input.responseSchema ? { responseSchema: input.responseSchema } : {}
2176
- }
2177
- });
2457
+ const response = await withRetry(
2458
+ () => ai.models.generateContent({
2459
+ model: apiModel,
2460
+ contents: input.prompt,
2461
+ config: buildGeminiRequestConfig({
2462
+ apiModelId: apiModel,
2463
+ responseSchema: input.responseSchema
2464
+ })
2465
+ })
2466
+ );
2178
2467
  const raw = response.text ?? "";
2179
- const parsed = JSON.parse(extractJson(raw));
2468
+ const parsed = parseGeminiResponse(raw);
2180
2469
  if (!parsed.frontmatter || typeof parsed.body !== "string") {
2181
2470
  throw new Error("Gemini response missing frontmatter/body");
2182
2471
  }
2183
- const usageMetadata = response.usageMetadata;
2184
- const usage = {
2185
- inputTokens: usageMetadata?.promptTokenCount ?? 0,
2186
- outputTokens: usageMetadata?.candidatesTokenCount ?? 0,
2187
- totalTokens: usageMetadata?.totalTokenCount ?? 0
2188
- };
2189
- return { model: displayModel, raw, parsed, usage };
2472
+ return { model: displayModel, raw, parsed, usage: usageFromResponse(response) };
2190
2473
  }
2191
2474
 
2475
+ // src/translate/translate-core.ts
2476
+ init_esm_shims();
2477
+ init_sqlite();
2478
+
2192
2479
  // src/translate/gemini-pricing.ts
2193
2480
  init_esm_shims();
2194
2481
  var CONTEXT_TIER_TOKENS = 2e5;
@@ -2207,12 +2494,14 @@ function resolveModelPricing(model) {
2207
2494
  const match = Object.entries(GEMINI_PRICING_USD).find(([key]) => id.includes(key));
2208
2495
  return match?.[1];
2209
2496
  }
2210
- function estimateTranslationCostUsd(model, inputTokens, outputTokens) {
2497
+ var BATCH_DISCOUNT = 0.5;
2498
+ function estimateTranslationCostUsd(model, inputTokens, outputTokens, mode = "interactive") {
2211
2499
  const pricing = resolveModelPricing(model);
2212
2500
  if (!pricing) return void 0;
2213
2501
  const inputRate = tierRate(inputTokens, pricing.input);
2214
2502
  const outputRate = tierRate(outputTokens, pricing.output);
2215
- return inputTokens / 1e6 * inputRate + outputTokens / 1e6 * outputRate;
2503
+ const multiplier = mode === "batch" ? BATCH_DISCOUNT : 1;
2504
+ return (inputTokens / 1e6 * inputRate + outputTokens / 1e6 * outputRate) * multiplier;
2216
2505
  }
2217
2506
  function formatTokenCount(tokens) {
2218
2507
  if (tokens >= 1e6) return `${(tokens / 1e6).toFixed(2)}M`;
@@ -2240,6 +2529,7 @@ var LOCALE_NAMES = {
2240
2529
  "zh-CN": "Simplified Chinese",
2241
2530
  "zh-TW": "Traditional Chinese",
2242
2531
  pt: "Portuguese",
2532
+ "pt-BR": "Brazilian Portuguese",
2243
2533
  ja: "Japanese",
2244
2534
  ru: "Russian",
2245
2535
  it: "Italian",
@@ -2247,29 +2537,29 @@ var LOCALE_NAMES = {
2247
2537
  };
2248
2538
  function defaultLocalizationPrompt(localeName, locale) {
2249
2539
  return [
2250
- `Localize the content below into natural ${localeName} (${locale}).`,
2540
+ `Localize the content above into natural ${localeName} (${locale}).`,
2251
2541
  "Do not translate word-for-word.",
2252
2542
  "Preserve the source tone and brand voice.",
2253
2543
  "Write as if a native speaker authored it for the target market."
2254
2544
  ].join(" ");
2255
2545
  }
2546
+ var TASK_FRAMING = "You are localizing the content below into a target language specified at the end of this prompt.";
2547
+ function buildOutputFormatLine(hasFrontmatter, slugStrategy) {
2548
+ if (!hasFrontmatter) {
2549
+ return slugStrategy === "localized" ? "`body` (string, full MDX body), `slug` (string)." : "`body` (string, full MDX body).";
2550
+ }
2551
+ return slugStrategy === "localized" ? "`frontmatter` (object with translated frontmatter fields), `body` (string, full MDX body), `slug` (string)." : "`frontmatter` (object), `body` (string).";
2552
+ }
2256
2553
  function buildPageTranslationPrompt(input) {
2257
2554
  const localeName = LOCALE_NAMES[input.targetLocale] ?? input.targetLocale;
2258
- const prompt = input.resolved.promptOverride ?? defaultLocalizationPrompt(localeName, input.targetLocale);
2259
- const lines = [
2260
- prompt,
2555
+ const localizationPrompt = input.resolved.promptOverride ?? defaultLocalizationPrompt(localeName, input.targetLocale);
2556
+ const prefix = [
2557
+ TASK_FRAMING,
2261
2558
  "",
2262
2559
  ...input.resolved.context ? ["## Context", input.resolved.context, ""] : [],
2263
2560
  ...input.contextLabel ? [`Document: ${input.contextLabel}`, ""] : [],
2264
2561
  "## Rules",
2265
2562
  ...input.resolved.rules.map((rule) => `- ${rule}`),
2266
- ...input.slugStrategy === "localized" ? [
2267
- `- 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.`
2268
- ] : [],
2269
- "",
2270
- "## Output format",
2271
- "Return ONLY valid JSON with keys:",
2272
- input.slugStrategy === "localized" ? "`frontmatter` (object with translated frontmatter fields), `body` (string, full MDX body), `slug` (string)." : "`frontmatter` (object), `body` (string).",
2273
2563
  "",
2274
2564
  "## EN translatable frontmatter (JSON)",
2275
2565
  JSON.stringify(input.translatableFrontmatter, null, 2),
@@ -2277,7 +2567,22 @@ function buildPageTranslationPrompt(input) {
2277
2567
  "## EN body (MDX)",
2278
2568
  input.enBody
2279
2569
  ];
2280
- return lines.join("\n");
2570
+ const suffix = [
2571
+ "",
2572
+ "## Target language",
2573
+ localizationPrompt,
2574
+ ...input.slugStrategy === "localized" ? [
2575
+ `The slug MUST be written in ${localeName}, derived from the ${localeName} title and its meaning \u2014 never the English slug.`
2576
+ ] : [],
2577
+ "",
2578
+ "## Output format",
2579
+ "Return ONLY valid JSON with keys:",
2580
+ buildOutputFormatLine(
2581
+ Object.keys(input.translatableFrontmatter).length > 0,
2582
+ input.slugStrategy
2583
+ )
2584
+ ];
2585
+ return [...prefix, ...suffix].join("\n");
2281
2586
  }
2282
2587
 
2283
2588
  // src/translate/response-schema.ts
@@ -2328,9 +2633,8 @@ function buildTranslatableSubschema(schema) {
2328
2633
  function buildGeminiResponseSchema(schema, slugStrategy) {
2329
2634
  try {
2330
2635
  const translatable = buildTranslatableSubschema(schema);
2331
- if (!translatable) return null;
2332
2636
  const responseShape = {
2333
- frontmatter: translatable,
2637
+ ...translatable ? { frontmatter: translatable } : {},
2334
2638
  body: z.string()
2335
2639
  };
2336
2640
  if (slugStrategy === "localized") {
@@ -2411,7 +2715,12 @@ function validateTranslatedFrontmatter(enDoc, localeFrontmatter, typeSchema) {
2411
2715
  return { ok: true, frontmatter };
2412
2716
  }
2413
2717
 
2414
- // src/translate/page-translator.ts
2718
+ // src/translate/translate-core.ts
2719
+ function translationItemKey(item) {
2720
+ const contentType = item.contentType ?? item.content_type ?? "";
2721
+ const enSlug = item.enSlug ?? item.en_slug ?? "";
2722
+ return `${contentType}\0${enSlug}\0${item.locale}`;
2723
+ }
2415
2724
  function summarizeResults(results, durationMs) {
2416
2725
  return results.reduce(
2417
2726
  (totals, result) => {
@@ -2447,17 +2756,6 @@ function formatTranslateError(error) {
2447
2756
  }
2448
2757
  return message.length > 160 ? `${message.slice(0, 157)}\u2026` : message;
2449
2758
  }
2450
- async function runPool(items, concurrency, worker) {
2451
- if (items.length === 0) return;
2452
- let nextIndex = 0;
2453
- const runners = Array.from({ length: Math.min(concurrency, items.length) }, async () => {
2454
- while (nextIndex < items.length) {
2455
- const index = nextIndex++;
2456
- await worker(items[index], index);
2457
- }
2458
- });
2459
- await Promise.all(runners);
2460
- }
2461
2759
  function resolveContextLabel(enDoc, enSlug) {
2462
2760
  const fm = enDoc.frontmatter;
2463
2761
  for (const key of ["title", "name", "h1"]) {
@@ -2466,73 +2764,82 @@ function resolveContextLabel(enDoc, enSlug) {
2466
2764
  }
2467
2765
  return enSlug;
2468
2766
  }
2469
- async function translatePage(config, item, options = {}) {
2470
- const startedAt = Date.now();
2471
- const base = {
2767
+ function baseForItem(item) {
2768
+ return {
2472
2769
  contentType: item.contentType,
2473
2770
  enSlug: item.enSlug,
2474
2771
  locale: item.locale
2475
2772
  };
2476
- try {
2477
- const type = config.types.find((t) => t.id === item.contentType);
2478
- if (!type) throw new Error(`Unknown content type ${item.contentType}`);
2479
- const enDoc = readEnDocument(config, type, item.enSlug);
2480
- if (!enDoc) throw new Error(`EN document not found: ${item.enSlug}`);
2481
- const payload = getTranslatablePayload(enDoc, type);
2482
- const currentEnHash = computePageEnHash(payload.frontmatter, payload.body);
2483
- const db = openStore(config, "readonly");
2484
- const existing = getTranslation(db, type.id, item.enSlug, item.locale);
2485
- db.close();
2486
- if (!options.force && existing && existing.en_hash === currentEnHash) {
2487
- return {
2488
- ...base,
2773
+ }
2774
+ function prepareTranslation(config, item, options, startedAt) {
2775
+ const type = config.types.find((t) => t.id === item.contentType);
2776
+ if (!type) throw new Error(`Unknown content type ${item.contentType}`);
2777
+ const enDoc = readEnDocument(config, type, item.enSlug);
2778
+ if (!enDoc) throw new Error(`EN document not found: ${item.enSlug}`);
2779
+ const payload = getTranslatablePayload(enDoc, type);
2780
+ const currentEnHash = computePageEnHash(payload.frontmatter, payload.body);
2781
+ const db = openStore(config, "readonly");
2782
+ const existing = getTranslation(db, type.id, item.enSlug, item.locale);
2783
+ db.close();
2784
+ if (!options.force && existing && existing.en_hash === currentEnHash) {
2785
+ return {
2786
+ status: "done",
2787
+ result: {
2788
+ ...baseForItem(item),
2489
2789
  skipped: true,
2490
2790
  reason: "fresh",
2491
2791
  durationMs: Date.now() - startedAt
2492
- };
2493
- }
2494
- const resolvedTranslate = resolveTranslateConfig(config, type);
2495
- const model = options.model ?? resolvedTranslate.model;
2496
- const prompt = buildPageTranslationPrompt({
2497
- resolved: resolvedTranslate,
2498
- targetLocale: item.locale,
2499
- contextLabel: resolveContextLabel(enDoc, item.enSlug),
2500
- translatableFrontmatter: payload.frontmatter,
2501
- enBody: payload.body,
2502
- slugStrategy: type.slugStrategy
2503
- });
2504
- if (options.dryRun) {
2505
- return {
2506
- ...base,
2507
- skipped: false,
2508
- model,
2509
- durationMs: Date.now() - startedAt
2510
- };
2511
- }
2512
- const responseSchema = buildGeminiResponseSchema(type.schema, type.slugStrategy);
2513
- const result = await translatePageWithGemini({
2514
- prompt,
2792
+ }
2793
+ };
2794
+ }
2795
+ const resolvedTranslate = resolveTranslateConfig(config, type);
2796
+ const model = options.model ?? resolvedTranslate.model;
2797
+ const prompt = buildPageTranslationPrompt({
2798
+ resolved: resolvedTranslate,
2799
+ targetLocale: item.locale,
2800
+ contextLabel: resolveContextLabel(enDoc, item.enSlug),
2801
+ translatableFrontmatter: payload.frontmatter,
2802
+ enBody: payload.body,
2803
+ slugStrategy: type.slugStrategy
2804
+ });
2805
+ const responseSchema = buildGeminiResponseSchema(type.schema, type.slugStrategy);
2806
+ return {
2807
+ status: "ready",
2808
+ prepared: {
2809
+ item,
2810
+ type,
2811
+ enDoc,
2812
+ payload,
2813
+ currentEnHash,
2814
+ existingSlug: existing?.slug,
2515
2815
  model,
2816
+ prompt,
2516
2817
  responseSchema: responseSchema ?? void 0
2517
- });
2518
- const rawSlug = type.slugStrategy === "localized" ? result.parsed.slug ?? existing?.slug ?? item.enSlug : item.enSlug;
2818
+ }
2819
+ };
2820
+ }
2821
+ function finalizeTranslation(config, prepared, output, options) {
2822
+ const { item, type, enDoc, payload } = prepared;
2823
+ const base = baseForItem(item);
2824
+ try {
2825
+ const rawSlug = type.slugStrategy === "localized" ? output.parsed.slug ?? prepared.existingSlug ?? item.enSlug : item.enSlug;
2519
2826
  const localeCodes = config.locales.filter((l) => l !== config.defaultLocale);
2520
2827
  const { slug, stripped, matchedCode } = stripLocaleSuffixFromSlug(rawSlug, localeCodes);
2521
2828
  const slugAdjusted = stripped && matchedCode ? { from: rawSlug, to: slug, matchedCode } : void 0;
2522
- const validated = validateTranslatedFrontmatter(enDoc, result.parsed.frontmatter, type.schema);
2829
+ const validated = validateTranslatedFrontmatter(enDoc, output.parsed.frontmatter, type.schema);
2523
2830
  if (!validated.ok) {
2524
2831
  throw new Error(`Translation validation failed: ${validated.error}`);
2525
2832
  }
2526
2833
  const { body: translatedBody, adjusted: mdxAdjusted } = assertValidTranslatedMdxBody(
2527
- result.parsed.body
2834
+ output.parsed.body
2528
2835
  );
2529
2836
  const writeDb = openStore(config, "readwrite");
2530
- const snapshotId = recordEnSnapshot(
2837
+ const snapshotId = options.snapshotId ?? recordEnSnapshot(
2531
2838
  config,
2532
2839
  {
2533
2840
  contentType: type.id,
2534
2841
  enSlug: item.enSlug,
2535
- enHash: currentEnHash,
2842
+ enHash: prepared.currentEnHash,
2536
2843
  frontmatter: payload.frontmatter,
2537
2844
  body: payload.body
2538
2845
  },
@@ -2545,24 +2852,25 @@ async function translatePage(config, item, options = {}) {
2545
2852
  slug,
2546
2853
  frontmatter: validated.frontmatter,
2547
2854
  body: translatedBody,
2548
- enHash: currentEnHash,
2855
+ enHash: prepared.currentEnHash,
2549
2856
  translatedAt: (/* @__PURE__ */ new Date()).toISOString(),
2550
- model: result.model,
2857
+ model: output.model,
2551
2858
  snapshotId
2552
2859
  });
2553
2860
  writeDb.close();
2554
2861
  const estimatedCostUsd = estimateTranslationCostUsd(
2555
- normalizeGeminiDisplayName(result.model),
2556
- result.usage.inputTokens,
2557
- result.usage.outputTokens
2862
+ normalizeGeminiDisplayName(output.model),
2863
+ output.usage.inputTokens,
2864
+ output.usage.outputTokens,
2865
+ options.costMode
2558
2866
  );
2559
2867
  return {
2560
2868
  ...base,
2561
2869
  skipped: false,
2562
- model: result.model,
2563
- usage: result.usage,
2870
+ model: output.model,
2871
+ usage: output.usage,
2564
2872
  estimatedCostUsd,
2565
- durationMs: Date.now() - startedAt,
2873
+ durationMs: Date.now() - options.startedAt,
2566
2874
  slugAdjusted,
2567
2875
  mdxAdjusted: mdxAdjusted || void 0
2568
2876
  };
@@ -2572,33 +2880,559 @@ async function translatePage(config, item, options = {}) {
2572
2880
  skipped: false,
2573
2881
  failed: true,
2574
2882
  error: formatTranslateError(error),
2883
+ durationMs: Date.now() - options.startedAt
2884
+ };
2885
+ }
2886
+ }
2887
+ function displayModelFor(prepared) {
2888
+ return normalizeGeminiDisplayName(
2889
+ prepared.model ?? process.env.PROSE_GEMINI_MODEL ?? DEFAULT_GEMINI_MODEL
2890
+ );
2891
+ }
2892
+
2893
+ // src/translate/batch-worklist.ts
2894
+ var MAX_REQUESTS_PER_JOB = 100;
2895
+ var MAX_PROMPT_BYTES_PER_JOB = 15 * 1024 * 1024;
2896
+ var INITIAL_POLL_MS = 5e3;
2897
+ var MAX_POLL_MS = 3e4;
2898
+ var POLL_BACKOFF_FACTOR = 1.5;
2899
+ function sleep(ms) {
2900
+ return new Promise((resolve) => setTimeout(resolve, ms));
2901
+ }
2902
+ function planBatchJobs(entries) {
2903
+ const byModel = /* @__PURE__ */ new Map();
2904
+ for (const entry of entries) {
2905
+ const group = byModel.get(entry.apiModel);
2906
+ if (group) group.push(entry);
2907
+ else byModel.set(entry.apiModel, [entry]);
2908
+ }
2909
+ const plans = [];
2910
+ for (const [apiModel, group] of byModel) {
2911
+ let current = [];
2912
+ let currentBytes = 0;
2913
+ for (const entry of group) {
2914
+ const size = Buffer.byteLength(entry.prompt, "utf8");
2915
+ if (current.length > 0 && (current.length >= MAX_REQUESTS_PER_JOB || currentBytes + size > MAX_PROMPT_BYTES_PER_JOB)) {
2916
+ plans.push({ apiModel, entries: current });
2917
+ current = [];
2918
+ currentBytes = 0;
2919
+ }
2920
+ current.push(entry);
2921
+ currentBytes += size;
2922
+ }
2923
+ if (current.length > 0) plans.push({ apiModel, entries: current });
2924
+ }
2925
+ return plans;
2926
+ }
2927
+ function readPendingBatchWork(config) {
2928
+ const db = openStore(config, "readwrite");
2929
+ const jobs = listPendingBatchJobs(db);
2930
+ const pendingItems = listPendingBatchItems(db);
2931
+ db.close();
2932
+ const inFlightKeys = new Set(pendingItems.map((item) => translationItemKey(item)));
2933
+ return { jobs, inFlightKeys, pendingItems };
2934
+ }
2935
+ async function submitBatchJobPlan(config, input) {
2936
+ const { plan, displayModel, jobIndex, jobCount } = input;
2937
+ const job = await createGeminiBatchJob({
2938
+ model: plan.apiModel,
2939
+ requests: plan.entries.map(({ prepared }) => ({
2940
+ contents: prepared.prompt,
2941
+ config: buildGeminiRequestConfig({
2942
+ apiModelId: plan.apiModel,
2943
+ responseSchema: prepared.responseSchema
2944
+ })
2945
+ })),
2946
+ displayName: `scribe-translate-${(/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-")}-${jobIndex + 1}`
2947
+ });
2948
+ const createdAt = (/* @__PURE__ */ new Date()).toISOString();
2949
+ const db = openStore(config, "readwrite");
2950
+ const jobId = insertBatchJob(db, {
2951
+ jobName: job.name,
2952
+ model: plan.apiModel,
2953
+ displayModel,
2954
+ state: String(job.state ?? "JOB_STATE_PENDING"),
2955
+ createdAt
2956
+ });
2957
+ insertBatchItems(
2958
+ db,
2959
+ jobId,
2960
+ plan.entries.map(({ prepared }, requestIndex) => ({
2961
+ requestIndex,
2962
+ contentType: prepared.item.contentType,
2963
+ enSlug: prepared.item.enSlug,
2964
+ locale: prepared.item.locale,
2965
+ enHash: prepared.currentEnHash,
2966
+ // Snapshot the EN source now so ingestion after a resume does not depend
2967
+ // on the EN files still matching (or existing) on disk.
2968
+ snapshotId: recordEnSnapshot(
2969
+ config,
2970
+ {
2971
+ contentType: prepared.item.contentType,
2972
+ enSlug: prepared.item.enSlug,
2973
+ enHash: prepared.currentEnHash,
2974
+ frontmatter: prepared.payload.frontmatter,
2975
+ body: prepared.payload.body
2976
+ },
2977
+ db
2978
+ )
2979
+ }))
2980
+ );
2981
+ const row = {
2982
+ id: jobId,
2983
+ job_name: job.name,
2984
+ model: plan.apiModel,
2985
+ display_model: displayModel,
2986
+ created_at: createdAt,
2987
+ state: String(job.state ?? "JOB_STATE_PENDING"),
2988
+ completed_at: null
2989
+ };
2990
+ db.close();
2991
+ input.onProgress?.({
2992
+ type: "batch-submitted",
2993
+ name: job.name,
2994
+ count: plan.entries.length,
2995
+ model: displayModel,
2996
+ jobIndex,
2997
+ jobCount,
2998
+ createdAt
2999
+ });
3000
+ return row;
3001
+ }
3002
+ function buildIngestContext(config, db, jobRow, itemRow) {
3003
+ const type = config.types.find((t) => t.id === itemRow.content_type);
3004
+ if (!type) throw new Error(`Unknown content type ${itemRow.content_type}`);
3005
+ let enDoc = readEnDocument(config, type, itemRow.en_slug);
3006
+ if (!enDoc) {
3007
+ const snapshot = getEnSnapshot(db, itemRow.snapshot_id);
3008
+ if (!snapshot) {
3009
+ throw new Error(
3010
+ `EN document and snapshot #${itemRow.snapshot_id} not found for ${itemRow.en_slug}`
3011
+ );
3012
+ }
3013
+ enDoc = {
3014
+ slug: itemRow.en_slug,
3015
+ enSlug: itemRow.en_slug,
3016
+ locale: config.defaultLocale,
3017
+ noindex: false,
3018
+ frontmatter: JSON.parse(snapshot.frontmatter_json),
3019
+ content: snapshot.body
3020
+ };
3021
+ }
3022
+ const item = {
3023
+ contentType: itemRow.content_type,
3024
+ enSlug: itemRow.en_slug,
3025
+ locale: itemRow.locale,
3026
+ reason: "missing",
3027
+ currentEnHash: itemRow.en_hash
3028
+ };
3029
+ return {
3030
+ item,
3031
+ type,
3032
+ enDoc,
3033
+ // Unused on ingest: finalizeTranslation receives the pre-recorded snapshotId.
3034
+ payload: { frontmatter: enDoc.frontmatter, body: enDoc.content },
3035
+ currentEnHash: itemRow.en_hash,
3036
+ existingSlug: getTranslation(db, type.id, itemRow.en_slug, itemRow.locale)?.slug,
3037
+ model: jobRow.display_model,
3038
+ prompt: "",
3039
+ responseSchema: void 0
3040
+ };
3041
+ }
3042
+ function ingestBatchJob(config, jobRow, batchJob, onResult) {
3043
+ const state = String(batchJob.state ?? "UNKNOWN");
3044
+ const db = openStore(config, "readwrite");
3045
+ if (!claimBatchJobCompletion(db, jobRow.id, state, (/* @__PURE__ */ new Date()).toISOString())) {
3046
+ db.close();
3047
+ return null;
3048
+ }
3049
+ const pendingItems = listBatchItems(db, jobRow.id).filter((item) => item.status === "pending");
3050
+ const results = [];
3051
+ const failItem = (itemRow, message, startedAt) => {
3052
+ const result = {
3053
+ contentType: itemRow.content_type,
3054
+ enSlug: itemRow.en_slug,
3055
+ locale: itemRow.locale,
3056
+ skipped: false,
3057
+ failed: true,
3058
+ error: formatTranslateError(new Error(message)),
2575
3059
  durationMs: Date.now() - startedAt
2576
3060
  };
3061
+ updateBatchItemStatus(db, jobRow.id, itemRow.request_index, "failed", result.error);
3062
+ results.push(result);
3063
+ onResult?.(result);
3064
+ };
3065
+ if (isSuccessfulBatchState(state)) {
3066
+ const responses = batchJob.dest?.inlinedResponses ?? [];
3067
+ for (const itemRow of pendingItems) {
3068
+ const startedAt = Date.now();
3069
+ const inlined = responses[itemRow.request_index];
3070
+ if (!inlined || inlined.error || !inlined.response) {
3071
+ failItem(
3072
+ itemRow,
3073
+ inlined?.error?.message ?? (inlined ? "Batch response missing content" : "Batch response missing"),
3074
+ startedAt
3075
+ );
3076
+ continue;
3077
+ }
3078
+ let result;
3079
+ try {
3080
+ const response = inlined.response;
3081
+ const raw = textFromBatchResponse(response);
3082
+ const parsed = parseGeminiResponse(raw);
3083
+ if (!parsed.frontmatter || typeof parsed.body !== "string") {
3084
+ throw new Error("Gemini response missing frontmatter/body");
3085
+ }
3086
+ const prepared = buildIngestContext(config, db, jobRow, itemRow);
3087
+ result = finalizeTranslation(
3088
+ config,
3089
+ prepared,
3090
+ { model: jobRow.display_model, parsed, usage: usageFromResponse(response) },
3091
+ { costMode: "batch", startedAt, snapshotId: itemRow.snapshot_id }
3092
+ );
3093
+ } catch (error) {
3094
+ failItem(itemRow, error instanceof Error ? error.message : String(error), startedAt);
3095
+ continue;
3096
+ }
3097
+ updateBatchItemStatus(
3098
+ db,
3099
+ jobRow.id,
3100
+ itemRow.request_index,
3101
+ result.failed ? "failed" : "done",
3102
+ result.error
3103
+ );
3104
+ results.push(result);
3105
+ onResult?.(result);
3106
+ }
3107
+ } else {
3108
+ const message = batchJob.error?.message ?? `Batch job ended in state ${normalizeBatchState(state)}`;
3109
+ const startedAt = Date.now();
3110
+ for (const itemRow of pendingItems) {
3111
+ failItem(itemRow, message, startedAt);
3112
+ }
2577
3113
  }
3114
+ db.close();
3115
+ return results;
3116
+ }
3117
+ async function pollAndIngestBatchJobs(config, jobs, options = {}) {
3118
+ const jobCount = options.jobCount ?? jobs.length;
3119
+ const active = jobs.map((row, jobIndex) => ({ row, jobIndex }));
3120
+ const loopStartedAt = Date.now();
3121
+ let delay = INITIAL_POLL_MS;
3122
+ let firstRound = true;
3123
+ const elapsedFor = (row) => {
3124
+ const createdAtMs = Date.parse(row.created_at);
3125
+ return Number.isNaN(createdAtMs) ? Date.now() - loopStartedAt : Date.now() - createdAtMs;
3126
+ };
3127
+ while (active.length > 0) {
3128
+ if (!firstRound) {
3129
+ await sleep(delay);
3130
+ delay = Math.min(delay * POLL_BACKOFF_FACTOR, MAX_POLL_MS);
3131
+ }
3132
+ firstRound = false;
3133
+ for (const tracked of [...active]) {
3134
+ const batchJob = await getGeminiBatchJob({ name: tracked.row.job_name });
3135
+ const state = normalizeBatchState(batchJob.state);
3136
+ options.onProgress?.({
3137
+ type: "batch-polling",
3138
+ name: tracked.row.job_name,
3139
+ state,
3140
+ elapsedMs: elapsedFor(tracked.row),
3141
+ jobIndex: tracked.jobIndex,
3142
+ jobCount
3143
+ });
3144
+ if (isTerminalBatchState(batchJob.state)) {
3145
+ active.splice(active.indexOf(tracked), 1);
3146
+ const results = ingestBatchJob(
3147
+ config,
3148
+ tracked.row,
3149
+ batchJob,
3150
+ options.onResult
3151
+ );
3152
+ if (results === null) {
3153
+ options.onProgress?.({
3154
+ type: "batch-done",
3155
+ name: tracked.row.job_name,
3156
+ state,
3157
+ model: tracked.row.display_model,
3158
+ count: 0,
3159
+ jobIndex: tracked.jobIndex,
3160
+ jobCount,
3161
+ translated: 0,
3162
+ failed: 0,
3163
+ inputTokens: 0,
3164
+ outputTokens: 0,
3165
+ estimatedCostUsd: 0,
3166
+ elapsedMs: elapsedFor(tracked.row),
3167
+ alreadyIngested: true
3168
+ });
3169
+ continue;
3170
+ }
3171
+ options.onProgress?.({
3172
+ type: "batch-done",
3173
+ name: tracked.row.job_name,
3174
+ state,
3175
+ model: tracked.row.display_model,
3176
+ count: results.length,
3177
+ jobIndex: tracked.jobIndex,
3178
+ jobCount,
3179
+ translated: results.filter((r) => !r.failed && !r.skipped).length,
3180
+ failed: results.filter((r) => r.failed).length,
3181
+ inputTokens: results.reduce((sum, r) => sum + (r.usage?.inputTokens ?? 0), 0),
3182
+ outputTokens: results.reduce((sum, r) => sum + (r.usage?.outputTokens ?? 0), 0),
3183
+ estimatedCostUsd: results.reduce((sum, r) => sum + (r.estimatedCostUsd ?? 0), 0),
3184
+ elapsedMs: elapsedFor(tracked.row)
3185
+ });
3186
+ }
3187
+ }
3188
+ }
3189
+ }
3190
+ function failedResultForItem(item, error, startedAt) {
3191
+ return {
3192
+ ...baseForItem(item),
3193
+ skipped: false,
3194
+ failed: true,
3195
+ error: formatTranslateError(error),
3196
+ durationMs: Date.now() - startedAt
3197
+ };
3198
+ }
3199
+
3200
+ // src/translate/page-translator.ts
3201
+ async function runPool(items, concurrency, worker) {
3202
+ if (items.length === 0) return;
3203
+ let nextIndex = 0;
3204
+ const runners = Array.from({ length: Math.min(concurrency, items.length) }, async () => {
3205
+ while (nextIndex < items.length) {
3206
+ const index = nextIndex++;
3207
+ await worker(items[index], index);
3208
+ }
3209
+ });
3210
+ await Promise.all(runners);
2578
3211
  }
2579
3212
  function labelForItem(item) {
2580
3213
  return `${item.contentType}/${item.enSlug}@${item.locale}`;
2581
3214
  }
3215
+ async function translatePage(config, item, options = {}) {
3216
+ const startedAt = Date.now();
3217
+ let outcome;
3218
+ try {
3219
+ outcome = prepareTranslation(config, item, options, startedAt);
3220
+ } catch (error) {
3221
+ return failedResultForItem(item, error, startedAt);
3222
+ }
3223
+ if (outcome.status === "done") return outcome.result;
3224
+ const prepared = outcome.prepared;
3225
+ if (options.dryRun) {
3226
+ return {
3227
+ ...baseForItem(item),
3228
+ skipped: false,
3229
+ model: prepared.model,
3230
+ durationMs: Date.now() - startedAt
3231
+ };
3232
+ }
3233
+ try {
3234
+ const result = await translatePageWithGemini({
3235
+ prompt: prepared.prompt,
3236
+ model: prepared.model,
3237
+ responseSchema: prepared.responseSchema
3238
+ });
3239
+ return finalizeTranslation(
3240
+ config,
3241
+ prepared,
3242
+ { model: result.model, parsed: result.parsed, usage: result.usage },
3243
+ { costMode: "interactive", startedAt }
3244
+ );
3245
+ } catch (error) {
3246
+ return failedResultForItem(item, error, startedAt);
3247
+ }
3248
+ }
3249
+ function createResultCollector(items, onProgress) {
3250
+ const slotByKey = /* @__PURE__ */ new Map();
3251
+ items.forEach((item, index) => slotByKey.set(translationItemKey(item), index));
3252
+ const slots = new Array(items.length);
3253
+ const extras = [];
3254
+ return {
3255
+ emit(result) {
3256
+ const slot = slotByKey.get(translationItemKey(result));
3257
+ if (slot !== void 0 && slots[slot] === void 0) slots[slot] = result;
3258
+ else extras.push(result);
3259
+ onProgress?.({ type: "item-done", result });
3260
+ },
3261
+ assemble() {
3262
+ return [
3263
+ ...slots.filter((result) => result !== void 0),
3264
+ ...extras
3265
+ ];
3266
+ }
3267
+ };
3268
+ }
2582
3269
  async function translateWorklist(config, items, options = {}) {
3270
+ const mode = options.mode ?? "batch";
2583
3271
  const concurrency = Math.max(1, options.concurrency ?? 3);
2584
3272
  const startedAt = Date.now();
2585
- const results = new Array(items.length);
2586
- const active = /* @__PURE__ */ new Set();
3273
+ if (options.dryRun) {
3274
+ options.onProgress?.({
3275
+ type: "start",
3276
+ total: items.length,
3277
+ concurrency,
3278
+ dryRun: true,
3279
+ model: options.model,
3280
+ mode
3281
+ });
3282
+ const results2 = items.map((item) => {
3283
+ const itemStartedAt = Date.now();
3284
+ try {
3285
+ const outcome = prepareTranslation(config, item, options, itemStartedAt);
3286
+ return outcome.status === "done" ? outcome.result : {
3287
+ ...baseForItem(item),
3288
+ skipped: false,
3289
+ model: outcome.prepared.model,
3290
+ durationMs: Date.now() - itemStartedAt
3291
+ };
3292
+ } catch (error) {
3293
+ return failedResultForItem(item, error, itemStartedAt);
3294
+ }
3295
+ });
3296
+ for (const result of results2) options.onProgress?.({ type: "item-done", result });
3297
+ const totals2 = summarizeResults(results2, Date.now() - startedAt);
3298
+ options.onProgress?.({ type: "done", results: results2, totals: totals2 });
3299
+ return results2;
3300
+ }
3301
+ const pending = readPendingBatchWork(config);
3302
+ const inputKeys = new Set(items.map((item) => translationItemKey(item)));
3303
+ const resumedExtraCount = pending.pendingItems.filter(
3304
+ (item) => !inputKeys.has(translationItemKey(item))
3305
+ ).length;
2587
3306
  options.onProgress?.({
2588
3307
  type: "start",
2589
- total: items.length,
3308
+ total: items.length + resumedExtraCount,
2590
3309
  concurrency,
2591
- dryRun: Boolean(options.dryRun),
2592
- model: options.model
3310
+ dryRun: false,
3311
+ model: options.model,
3312
+ mode
3313
+ });
3314
+ const collector = createResultCollector(items, options.onProgress);
3315
+ const workItems = items.filter((item) => !pending.inFlightKeys.has(translationItemKey(item)));
3316
+ if (mode === "direct") {
3317
+ const pendingCounts = countPendingItems(config, pending.jobs);
3318
+ emitResumedJobs(pending.jobs, pendingCounts, pending.jobs.length, options.onProgress);
3319
+ const active = /* @__PURE__ */ new Set();
3320
+ await Promise.all([
3321
+ runPool(workItems, concurrency, async (item) => {
3322
+ const label = labelForItem(item);
3323
+ active.add(label);
3324
+ options.onProgress?.({ type: "item-start", item, active: [...active] });
3325
+ const result = await translatePage(config, item, options);
3326
+ active.delete(label);
3327
+ collector.emit(result);
3328
+ }),
3329
+ pending.jobs.length > 0 ? pollAndIngestBatchJobs(config, pending.jobs, {
3330
+ onProgress: options.onProgress,
3331
+ onResult: collector.emit
3332
+ }) : Promise.resolve()
3333
+ ]);
3334
+ } else {
3335
+ const entries = [];
3336
+ for (const item of workItems) {
3337
+ const itemStartedAt = Date.now();
3338
+ try {
3339
+ const outcome = prepareTranslation(config, item, options, itemStartedAt);
3340
+ if (outcome.status === "done") {
3341
+ collector.emit(outcome.result);
3342
+ } else {
3343
+ entries.push({
3344
+ apiModel: resolveGeminiModelId(displayModelFor(outcome.prepared)),
3345
+ prompt: outcome.prepared.prompt,
3346
+ prepared: outcome.prepared
3347
+ });
3348
+ }
3349
+ } catch (error) {
3350
+ collector.emit(failedResultForItem(item, error, itemStartedAt));
3351
+ }
3352
+ }
3353
+ const plans = planBatchJobs(entries);
3354
+ const jobCount = pending.jobs.length + plans.length;
3355
+ const pendingCounts = countPendingItems(config, pending.jobs);
3356
+ emitResumedJobs(pending.jobs, pendingCounts, jobCount, options.onProgress);
3357
+ const submitted = await Promise.all(
3358
+ plans.map(async (plan, planIndex) => {
3359
+ try {
3360
+ return await submitBatchJobPlan(config, {
3361
+ plan,
3362
+ displayModel: normalizeGeminiDisplayName(plan.apiModel),
3363
+ jobIndex: pending.jobs.length + planIndex,
3364
+ jobCount,
3365
+ onProgress: options.onProgress
3366
+ });
3367
+ } catch (error) {
3368
+ for (const entry of plan.entries) {
3369
+ collector.emit(failedResultForItem(entry.prepared.item, error, startedAt));
3370
+ }
3371
+ return void 0;
3372
+ }
3373
+ })
3374
+ );
3375
+ const jobsToPoll = [
3376
+ ...pending.jobs,
3377
+ ...submitted.filter((row) => row !== void 0)
3378
+ ];
3379
+ if (jobsToPoll.length > 0) {
3380
+ await pollAndIngestBatchJobs(config, jobsToPoll, {
3381
+ jobCount,
3382
+ onProgress: options.onProgress,
3383
+ onResult: collector.emit
3384
+ });
3385
+ }
3386
+ }
3387
+ const results = collector.assemble();
3388
+ const totals = summarizeResults(results, Date.now() - startedAt);
3389
+ options.onProgress?.({ type: "done", results, totals });
3390
+ return results;
3391
+ }
3392
+ function countPendingItems(config, jobs) {
3393
+ if (jobs.length === 0) return /* @__PURE__ */ new Map();
3394
+ const db = openStore(config, "readonly");
3395
+ const counts = /* @__PURE__ */ new Map();
3396
+ for (const job of jobs) {
3397
+ counts.set(job.id, listBatchItems(db, job.id).filter((i) => i.status === "pending").length);
3398
+ }
3399
+ db.close();
3400
+ return counts;
3401
+ }
3402
+ function emitResumedJobs(jobs, counts, jobCount, onProgress) {
3403
+ jobs.forEach((job, jobIndex) => {
3404
+ onProgress?.({
3405
+ type: "batch-submitted",
3406
+ name: job.job_name,
3407
+ count: counts.get(job.id) ?? 0,
3408
+ model: job.display_model,
3409
+ jobIndex,
3410
+ jobCount,
3411
+ resumed: true,
3412
+ createdAt: job.created_at
3413
+ });
3414
+ });
3415
+ }
3416
+ async function resumeTranslationJobs(config, options = {}) {
3417
+ const startedAt = Date.now();
3418
+ const pending = readPendingBatchWork(config);
3419
+ if (pending.jobs.length === 0) return null;
3420
+ options.onProgress?.({
3421
+ type: "start",
3422
+ total: pending.pendingItems.length,
3423
+ concurrency: 1,
3424
+ dryRun: false,
3425
+ mode: "batch"
2593
3426
  });
2594
- await runPool(items, concurrency, async (item, index) => {
2595
- const label = labelForItem(item);
2596
- active.add(label);
2597
- options.onProgress?.({ type: "item-start", item, active: [...active] });
2598
- const result = await translatePage(config, item, options);
2599
- results[index] = result;
2600
- active.delete(label);
2601
- options.onProgress?.({ type: "item-done", result });
3427
+ const counts = countPendingItems(config, pending.jobs);
3428
+ emitResumedJobs(pending.jobs, counts, pending.jobs.length, options.onProgress);
3429
+ const results = [];
3430
+ await pollAndIngestBatchJobs(config, pending.jobs, {
3431
+ onProgress: options.onProgress,
3432
+ onResult: (result) => {
3433
+ results.push(result);
3434
+ options.onProgress?.({ type: "item-done", result });
3435
+ }
2602
3436
  });
2603
3437
  const totals = summarizeResults(results, Date.now() - startedAt);
2604
3438
  options.onProgress?.({ type: "done", results, totals });
@@ -3216,35 +4050,43 @@ async function promptContentType(config) {
3216
4050
  })
3217
4051
  );
3218
4052
  }
4053
+ var ALL_LOCALES = /* @__PURE__ */ Symbol("all-locales");
4054
+ var PICK_LOCALES = /* @__PURE__ */ Symbol("pick-locales");
4055
+ async function promptManualLocales(config) {
4056
+ const locales = nonDefaultLocales(config);
4057
+ if (locales.length <= 1) return {};
4058
+ const picked = await runPrompt(
4059
+ checkbox({
4060
+ message: "Locales to translate",
4061
+ choices: locales.map((locale) => ({ name: locale, value: locale, checked: true })),
4062
+ validate: (values) => values.length > 0 || "Pick at least one locale"
4063
+ })
4064
+ );
4065
+ return picked.length === locales.length ? {} : { locale: picked };
4066
+ }
3219
4067
  async function promptLocalePreset(config) {
3220
4068
  const presets = Object.entries(config.localePresets ?? {}).filter(
3221
4069
  (entry) => Array.isArray(entry[1])
3222
4070
  );
3223
4071
  if (presets.length === 0) {
3224
- const locales = nonDefaultLocales(config);
3225
- if (locales.length <= 1) return {};
3226
- const picked = await runPrompt(
3227
- checkbox({
3228
- message: "Locales to translate",
3229
- choices: locales.map((locale) => ({ name: locale, value: locale, checked: true })),
3230
- validate: (values) => values.length > 0 || "Pick at least one locale"
3231
- })
3232
- );
3233
- return picked.length === locales.length ? {} : { locale: picked };
4072
+ return promptManualLocales(config);
3234
4073
  }
3235
4074
  const choice = await runPrompt(
3236
4075
  select({
3237
4076
  message: "Locale preset",
3238
4077
  choices: [
3239
- { name: "All locales", value: void 0 },
4078
+ { name: "All locales", value: ALL_LOCALES },
3240
4079
  ...presets.map(([name, locales]) => ({
3241
4080
  name: `${name} (${locales.join(", ")})`,
3242
4081
  value: name
3243
- }))
4082
+ })),
4083
+ { name: "Choose locales manually\u2026", value: PICK_LOCALES }
3244
4084
  ]
3245
4085
  })
3246
4086
  );
3247
- return choice ? { preset: choice } : {};
4087
+ if (choice === PICK_LOCALES) return promptManualLocales(config);
4088
+ if (choice === ALL_LOCALES) return {};
4089
+ return { preset: choice };
3248
4090
  }
3249
4091
  async function promptStrategy() {
3250
4092
  return runPrompt(
@@ -3258,12 +4100,25 @@ async function promptStrategy() {
3258
4100
  })
3259
4101
  );
3260
4102
  }
4103
+ async function promptMode() {
4104
+ return runPrompt(
4105
+ select({
4106
+ message: "Translation mode",
4107
+ choices: [
4108
+ { name: "Batch \u2014 50% cheaper, async, resumable (recommended)", value: "batch" },
4109
+ { name: "Direct \u2014 immediate results, full price", value: "direct" }
4110
+ ],
4111
+ default: "batch"
4112
+ })
4113
+ );
4114
+ }
3261
4115
  async function promptTranslateSelection(config, flags) {
3262
4116
  const selection = {
3263
4117
  contentType: flags.type,
3264
4118
  preset: flags.preset,
3265
4119
  locale: flags.locale,
3266
- strategy: flags.strategy
4120
+ strategy: flags.strategy,
4121
+ mode: flags.mode
3267
4122
  };
3268
4123
  if (!isInteractive()) return selection;
3269
4124
  if (!selection.contentType) {
@@ -3275,6 +4130,9 @@ async function promptTranslateSelection(config, flags) {
3275
4130
  if (!selection.strategy) {
3276
4131
  selection.strategy = await promptStrategy();
3277
4132
  }
4133
+ if (!selection.mode) {
4134
+ selection.mode = await promptMode();
4135
+ }
3278
4136
  return selection;
3279
4137
  }
3280
4138
 
@@ -3296,6 +4154,34 @@ function progressBar(ratio, width = 28) {
3296
4154
  function labelForResult(result) {
3297
4155
  return `${result.contentType}/${result.enSlug}@${result.locale}`;
3298
4156
  }
4157
+ function shortJobId(name) {
4158
+ const segments = name.split("/");
4159
+ return segments[segments.length - 1] || name;
4160
+ }
4161
+ function formatElapsed(elapsedMs) {
4162
+ return `${Math.round(elapsedMs / 1e3)}s`;
4163
+ }
4164
+ function batchDoneParts(event) {
4165
+ if (event.alreadyIngested) {
4166
+ return [
4167
+ `job ${shortJobId(event.name)}`,
4168
+ event.model ?? "?",
4169
+ "already ingested by another scribe run (results saved there)",
4170
+ formatElapsed(event.elapsedMs)
4171
+ ].join(" \xB7 ");
4172
+ }
4173
+ const parts = [
4174
+ `job ${shortJobId(event.name)}`,
4175
+ event.model ?? "?",
4176
+ `${event.count} request${event.count === 1 ? "" : "s"}`,
4177
+ `${event.translated} translated`
4178
+ ];
4179
+ if (event.failed > 0) parts.push(`${event.failed} failed`);
4180
+ parts.push(`${formatTokenCount(event.inputTokens)} in / ${formatTokenCount(event.outputTokens)} out`);
4181
+ parts.push(`${formatUsd(event.estimatedCostUsd)} est.`);
4182
+ parts.push(formatElapsed(event.elapsedMs));
4183
+ return parts.join(" \xB7 ");
4184
+ }
3299
4185
  function statusForResult(result, dryRun) {
3300
4186
  if (result.failed) return red("failed");
3301
4187
  if (result.skipped) return dim("skipped");
@@ -3341,6 +4227,25 @@ function createTranslateProgressReporter(options = {}) {
3341
4227
  if (!enabled) {
3342
4228
  return {
3343
4229
  onEvent(event) {
4230
+ if (event.type === "batch-submitted") {
4231
+ const verb = event.resumed ? "Resuming" : "Submitted";
4232
+ console.log(
4233
+ `${verb} batch job ${event.jobIndex + 1}/${event.jobCount} with ${event.count} request${event.count === 1 ? "" : "s"}${event.model ? ` (${event.model})` : ""}: ${event.name}`
4234
+ );
4235
+ return;
4236
+ }
4237
+ if (event.type === "batch-polling") {
4238
+ console.log(
4239
+ `batch ${event.jobIndex + 1}/${event.jobCount} ${shortJobId(event.name)}: ${event.state.replace(/^JOB_STATE_/, "")} (${formatElapsed(event.elapsedMs)} elapsed)`
4240
+ );
4241
+ return;
4242
+ }
4243
+ if (event.type === "batch-done") {
4244
+ const line = `batch ${event.jobIndex + 1}/${event.jobCount} ${event.state} \xB7 ${batchDoneParts(event)}`;
4245
+ if (event.failed > 0 && event.translated === 0) console.error(line);
4246
+ else console.log(line);
4247
+ return;
4248
+ }
3344
4249
  if (event.type === "item-done") {
3345
4250
  const label = labelForResult(event.result);
3346
4251
  const status = statusForResult(event.result, dryRun);
@@ -3371,6 +4276,9 @@ function createTranslateProgressReporter(options = {}) {
3371
4276
  let outputTokens = 0;
3372
4277
  let estimatedCostUsd = 0;
3373
4278
  let active = [];
4279
+ let mode;
4280
+ const batchJobs = /* @__PURE__ */ new Map();
4281
+ let batchElapsedMs = 0;
3374
4282
  const recent = [];
3375
4283
  let renderedLines = 0;
3376
4284
  let cursorHidden = false;
@@ -3386,6 +4294,15 @@ function createTranslateProgressReporter(options = {}) {
3386
4294
  cursorHidden = false;
3387
4295
  }
3388
4296
  }
4297
+ function printPersistent(lines) {
4298
+ if (renderedLines > 0) {
4299
+ process.stdout.write(`\x1B[${renderedLines}A`);
4300
+ }
4301
+ for (const line of lines) {
4302
+ process.stdout.write("\x1B[2K" + line + "\n");
4303
+ }
4304
+ renderedLines = 0;
4305
+ }
3389
4306
  function render() {
3390
4307
  hideCursor();
3391
4308
  if (renderedLines > 0) {
@@ -3393,12 +4310,34 @@ function createTranslateProgressReporter(options = {}) {
3393
4310
  }
3394
4311
  const lines = [];
3395
4312
  const title = dryRun ? "Dry run" : "Translating";
3396
- lines.push(cyan(`${title} ${done}/${total}`) + dim(` \xB7 ${concurrency} parallel`) + (model ? dim(` \xB7 ${model}`) : ""));
4313
+ const isBatch = mode === "batch" || batchJobs.size > 0;
4314
+ lines.push(
4315
+ cyan(`${title} ${done}/${total}`) + dim(isBatch ? " \xB7 batch" : ` \xB7 ${concurrency} parallel`) + (model ? dim(` \xB7 ${model}`) : "")
4316
+ );
3397
4317
  lines.push(progressBar(total === 0 ? 0 : done / total) + dim(` ${Math.round(total === 0 ? 0 : done / total * 100)}%`));
3398
4318
  lines.push(
3399
4319
  dim("Tokens ") + `${formatTokenCount(inputTokens)} in \xB7 ${formatTokenCount(outputTokens)} out` + dim(" \xB7 Cost ") + formatUsd(estimatedCostUsd) + dim(" est.")
3400
4320
  );
3401
- if (active.length > 0) {
4321
+ if (batchJobs.size > 0 && done < total) {
4322
+ const terminal = /^(SUCCEEDED|FAILED|CANCELLED|EXPIRED|PARTIALLY_SUCCEEDED)$/;
4323
+ let running = 0;
4324
+ let finished = 0;
4325
+ let requests = 0;
4326
+ for (const job of batchJobs.values()) {
4327
+ if (terminal.test(job.state)) finished += 1;
4328
+ else running += 1;
4329
+ requests += job.count;
4330
+ }
4331
+ lines.push(
4332
+ dim("Batches ") + `${running} running \xB7 ${finished} done` + dim(` \xB7 ${requests} request${requests === 1 ? "" : "s"} \xB7 ${Math.round(batchElapsedMs / 1e3)}s elapsed`)
4333
+ );
4334
+ for (const [name, job] of batchJobs) {
4335
+ if (terminal.test(job.state)) continue;
4336
+ lines.push(
4337
+ dim(" ") + shortJobId(name) + dim(` \xB7 ${job.model ?? "?"} \xB7 ${job.count} req \xB7 `) + job.state + dim(` \xB7 ${formatElapsed(job.elapsedMs)}`)
4338
+ );
4339
+ }
4340
+ } else if (active.length > 0) {
3402
4341
  lines.push(dim("Active ") + active.slice(0, 3).join(", ") + (active.length > 3 ? dim(` +${active.length - 3}`) : ""));
3403
4342
  } else if (done < total) {
3404
4343
  lines.push(dim("Active ") + "starting\u2026");
@@ -3426,12 +4365,58 @@ function createTranslateProgressReporter(options = {}) {
3426
4365
  total = event.total;
3427
4366
  concurrency = event.concurrency;
3428
4367
  model = event.model;
4368
+ mode = event.mode;
3429
4369
  render();
3430
4370
  break;
3431
4371
  case "item-start":
3432
4372
  active = event.active;
3433
4373
  render();
3434
4374
  break;
4375
+ case "batch-submitted": {
4376
+ const createdAtMs = event.createdAt ? Date.parse(event.createdAt) : void 0;
4377
+ const elapsedMs = createdAtMs && !Number.isNaN(createdAtMs) ? Date.now() - createdAtMs : 0;
4378
+ batchJobs.set(event.name, {
4379
+ state: event.resumed ? "RESUMED" : "SUBMITTED",
4380
+ count: event.count,
4381
+ model: event.model,
4382
+ createdAtMs: Number.isNaN(createdAtMs) ? void 0 : createdAtMs,
4383
+ elapsedMs
4384
+ });
4385
+ const requests = `${event.count} ${event.resumed ? "pending " : ""}request${event.count === 1 ? "" : "s"}`;
4386
+ const origin = event.resumed ? `resumed job (submitted ${formatElapsed(elapsedMs)} ago)` : "submitted job";
4387
+ printPersistent([
4388
+ dim(`${event.resumed ? "\u21BB" : "\u2192"} ${origin} ${shortJobId(event.name)} \xB7 ${event.model ?? "?"} \xB7 ${requests}`)
4389
+ ]);
4390
+ render();
4391
+ break;
4392
+ }
4393
+ case "batch-polling": {
4394
+ const job = batchJobs.get(event.name);
4395
+ batchJobs.set(event.name, {
4396
+ state: event.state.replace(/^JOB_STATE_/, ""),
4397
+ count: job?.count ?? 0,
4398
+ model: job?.model,
4399
+ createdAtMs: job?.createdAtMs,
4400
+ elapsedMs: event.elapsedMs
4401
+ });
4402
+ batchElapsedMs = Math.max(batchElapsedMs, event.elapsedMs);
4403
+ render();
4404
+ break;
4405
+ }
4406
+ case "batch-done": {
4407
+ const job = batchJobs.get(event.name);
4408
+ batchJobs.set(event.name, {
4409
+ state: event.state.replace(/^JOB_STATE_/, ""),
4410
+ count: job?.count ?? event.count,
4411
+ model: job?.model ?? event.model,
4412
+ createdAtMs: job?.createdAtMs,
4413
+ elapsedMs: event.elapsedMs
4414
+ });
4415
+ const marker = event.failed > 0 && event.translated === 0 ? red("\u2717") : green("\u2713");
4416
+ printPersistent([`${marker} ${dim(batchDoneParts(event))}`]);
4417
+ render();
4418
+ break;
4419
+ }
3435
4420
  case "item-done": {
3436
4421
  done += 1;
3437
4422
  active = active.filter((label) => label !== labelForResult(event.result));
@@ -3554,6 +4539,21 @@ function parseArgs(argv) {
3554
4539
  i++;
3555
4540
  continue;
3556
4541
  }
4542
+ if (arg === "--batch") {
4543
+ options.batch = true;
4544
+ i++;
4545
+ continue;
4546
+ }
4547
+ if (arg === "--direct") {
4548
+ options.direct = true;
4549
+ i++;
4550
+ continue;
4551
+ }
4552
+ if (arg === "--resume") {
4553
+ options.resume = true;
4554
+ i++;
4555
+ continue;
4556
+ }
3557
4557
  if (arg === "--strategy") {
3558
4558
  const value = argv[++i];
3559
4559
  if (value !== "all" && value !== "missing-only") {
@@ -3622,11 +4622,19 @@ Translate flags:
3622
4622
  --locale <code>... Target locale(s); overrides --preset
3623
4623
  --slug <en-slug> Single English document
3624
4624
  --model <id> Gemini model override
3625
- --concurrency <n> Parallel translations (default: 3)
4625
+ --batch Force Gemini Batch API mode (50% token cost; default)
4626
+ --direct Force direct interactive calls (immediate, full price)
4627
+ --resume Only poll/ingest pending batch jobs; submit nothing new
4628
+ --concurrency <n> Parallel translations in --direct mode (default: 3)
3626
4629
  --dry-run List work without writing
3627
4630
  --force Re-translate even when hashes match
3628
4631
  --strategy <mode> all (default) or missing-only
3629
4632
  --no-progress Plain line logging instead of live progress
4633
+
4634
+ Batch mode submits jobs upfront and persists them in the store, so Ctrl+C
4635
+ during polling is safe: the next run (or --resume) picks the jobs back up and
4636
+ ingests their results. In a TTY without --batch/--direct, a prompt asks which
4637
+ mode to use.
3630
4638
  `);
3631
4639
  return;
3632
4640
  }
@@ -3672,11 +4680,29 @@ Translate flags:
3672
4680
  break;
3673
4681
  }
3674
4682
  case "translate": {
4683
+ if (options.batch && options.direct) {
4684
+ throw new Error("--batch and --direct are mutually exclusive");
4685
+ }
4686
+ if (options.resume) {
4687
+ const reporter2 = createTranslateProgressReporter({
4688
+ enabled: !options.noProgress,
4689
+ dryRun: false
4690
+ });
4691
+ const results2 = await resumeTranslationJobs(config, { onProgress: reporter2.onEvent });
4692
+ reporter2.finish();
4693
+ if (results2 === null) {
4694
+ console.log("No pending translation batch jobs.");
4695
+ break;
4696
+ }
4697
+ if (results2.some((result) => result.failed)) process.exitCode = 1;
4698
+ break;
4699
+ }
3675
4700
  const selection = await promptTranslateSelection(config, {
3676
4701
  type: options.type,
3677
4702
  preset: options.preset,
3678
4703
  locale: options.locale,
3679
- strategy: options.strategy
4704
+ strategy: options.strategy,
4705
+ mode: options.batch ? "batch" : options.direct ? "direct" : void 0
3680
4706
  });
3681
4707
  const locales = resolveLocalesFromPreset(config, selection.preset, selection.locale);
3682
4708
  const worklist = buildWorklist(config, {
@@ -3686,6 +4712,18 @@ Translate flags:
3686
4712
  strategy: selection.strategy ?? "all"
3687
4713
  });
3688
4714
  if (worklist.length === 0) {
4715
+ if (!options.dryRun) {
4716
+ const reporter2 = createTranslateProgressReporter({
4717
+ enabled: !options.noProgress,
4718
+ dryRun: false
4719
+ });
4720
+ const resumed = await resumeTranslationJobs(config, { onProgress: reporter2.onEvent });
4721
+ reporter2.finish();
4722
+ if (resumed !== null) {
4723
+ if (resumed.some((result) => result.failed)) process.exitCode = 1;
4724
+ break;
4725
+ }
4726
+ }
3689
4727
  console.log("Nothing to translate.");
3690
4728
  break;
3691
4729
  }
@@ -3698,6 +4736,7 @@ Translate flags:
3698
4736
  dryRun: options.dryRun,
3699
4737
  force: options.force,
3700
4738
  concurrency: options.concurrency,
4739
+ mode: selection.mode,
3701
4740
  onProgress: reporter.onEvent
3702
4741
  });
3703
4742
  reporter.finish();