@verbatra/sdk 0.4.4 → 0.5.0-next.1

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.
package/dist/index.cjs CHANGED
@@ -4,6 +4,8 @@ var fs = require('fs');
4
4
  var path = require('path');
5
5
  var cosmiconfig = require('cosmiconfig');
6
6
  var cosmiconfigTypescriptLoader = require('cosmiconfig-typescript-loader');
7
+ var crypto = require('crypto');
8
+ var promises = require('fs/promises');
7
9
  var zod = require('zod');
8
10
  var Anthropic = require('@anthropic-ai/sdk');
9
11
  var deepl = require('deepl-node');
@@ -11,8 +13,6 @@ var module$1 = require('module');
11
13
  var log = require('loglevel');
12
14
  var genai = require('@google/genai');
13
15
  var OpenAI = require('openai');
14
- var crypto = require('crypto');
15
- var promises = require('fs/promises');
16
16
  var icuMessageformatParser = require('@formatjs/icu-messageformat-parser');
17
17
  var xmldom = require('@xmldom/xmldom');
18
18
  var yaml = require('yaml');
@@ -68,6 +68,155 @@ var SdkError = class extends Error {
68
68
  this.code = code;
69
69
  }
70
70
  };
71
+ async function readBoundedUtf8(handle, size) {
72
+ const buffer = Buffer.allocUnsafe(size);
73
+ let offset = 0;
74
+ while (offset < size) {
75
+ const { bytesRead } = await handle.read(buffer, offset, size - offset, offset);
76
+ if (bytesRead === 0) {
77
+ break;
78
+ }
79
+ offset += bytesRead;
80
+ }
81
+ return buffer.toString("utf8", 0, offset);
82
+ }
83
+ async function readBounded(path, maxBytes) {
84
+ let handle;
85
+ try {
86
+ handle = await promises.open(path, "r");
87
+ } catch {
88
+ return { kind: "missing" };
89
+ }
90
+ try {
91
+ const info = await handle.stat();
92
+ if (!info.isFile()) {
93
+ return { kind: "missing" };
94
+ }
95
+ if (info.size > maxBytes) {
96
+ return { kind: "too-large" };
97
+ }
98
+ return { kind: "ok", content: await readBoundedUtf8(handle, info.size) };
99
+ } finally {
100
+ await handle.close();
101
+ }
102
+ }
103
+ async function readBoundedBytesInto(handle, size) {
104
+ const buffer = Buffer.allocUnsafeSlow(size);
105
+ let offset = 0;
106
+ while (offset < size) {
107
+ const { bytesRead } = await handle.read(buffer, offset, size - offset, offset);
108
+ if (bytesRead === 0) {
109
+ break;
110
+ }
111
+ offset += bytesRead;
112
+ }
113
+ return new Uint8Array(buffer.buffer, buffer.byteOffset, offset);
114
+ }
115
+ async function readBoundedBytes(path, maxBytes) {
116
+ let handle;
117
+ try {
118
+ handle = await promises.open(path, "r");
119
+ } catch {
120
+ return { kind: "missing" };
121
+ }
122
+ try {
123
+ const info = await handle.stat();
124
+ if (!info.isFile()) {
125
+ return { kind: "missing" };
126
+ }
127
+ if (info.size > maxBytes) {
128
+ return { kind: "too-large" };
129
+ }
130
+ return { kind: "ok", bytes: await readBoundedBytesInto(handle, info.size) };
131
+ } finally {
132
+ await handle.close();
133
+ }
134
+ }
135
+ function tempFileName(path$1) {
136
+ return path.join(path.dirname(path$1), `.${path.basename(path$1)}.tmp-${process.pid}-${Date.now()}-${crypto.randomUUID()}`);
137
+ }
138
+ async function atomicWrite(path, data) {
139
+ const tmp = tempFileName(path);
140
+ await (typeof data === "string" ? promises.writeFile(tmp, data, "utf8") : promises.writeFile(tmp, data));
141
+ try {
142
+ await promises.rename(tmp, path);
143
+ } catch (error) {
144
+ await promises.rm(tmp, { force: true });
145
+ throw error;
146
+ }
147
+ }
148
+ var defaultFs = {
149
+ async fileExists(path) {
150
+ try {
151
+ await promises.access(path);
152
+ return true;
153
+ } catch {
154
+ return false;
155
+ }
156
+ },
157
+ readFileBounded: (path, maxBytes) => readBounded(path, maxBytes),
158
+ readBytesBounded: (path, maxBytes) => readBoundedBytes(path, maxBytes),
159
+ writeFile: (path, data) => atomicWrite(path, data),
160
+ writeBytes: (path, data) => atomicWrite(path, data)
161
+ };
162
+ var MAX_GLOSSARY_FILE_BYTES = 1024 * 1024;
163
+ var glossaryRecordSchema = zod.z.record(zod.z.string(), zod.z.string());
164
+ var BOM = "\uFEFF";
165
+ var REPLACEMENT_CHARACTER = "\uFFFD";
166
+ var NUL = "\0";
167
+ function stripBom(content) {
168
+ return content.startsWith(BOM) ? content.slice(BOM.length) : content;
169
+ }
170
+ function looksLikeInvalidEncoding(content) {
171
+ return content.startsWith(REPLACEMENT_CHARACTER) || content.includes(NUL);
172
+ }
173
+ async function readGlossaryFile(resolvedPath, fs) {
174
+ const read = await fs.readFileBounded(resolvedPath, MAX_GLOSSARY_FILE_BYTES);
175
+ if (read.kind === "missing") {
176
+ throw new SdkError(
177
+ "CONFIG_INVALID",
178
+ `The glossary file at ${resolvedPath} was not found or could not be read.`
179
+ );
180
+ }
181
+ if (read.kind === "too-large") {
182
+ throw new SdkError(
183
+ "CONFIG_INVALID",
184
+ `The glossary file at ${resolvedPath} exceeds the maximum allowed size of ${MAX_GLOSSARY_FILE_BYTES} bytes.`
185
+ );
186
+ }
187
+ const content = stripBom(read.content);
188
+ if (looksLikeInvalidEncoding(content)) {
189
+ throw new SdkError(
190
+ "CONFIG_INVALID",
191
+ `The glossary file at ${resolvedPath} must be UTF-8 encoded.`
192
+ );
193
+ }
194
+ let parsed;
195
+ try {
196
+ parsed = JSON.parse(content);
197
+ } catch {
198
+ throw new SdkError("CONFIG_INVALID", `The glossary file at ${resolvedPath} is not valid JSON.`);
199
+ }
200
+ const result = glossaryRecordSchema.safeParse(parsed);
201
+ if (!result.success) {
202
+ throw new SdkError(
203
+ "CONFIG_INVALID",
204
+ `The glossary file at ${resolvedPath} must contain a flat object of string keys to string values.`
205
+ );
206
+ }
207
+ return result.data;
208
+ }
209
+ async function resolveGlossary(glossary, baseDir, fs) {
210
+ if (glossary === void 0) {
211
+ return { glossary: void 0, provenance: { source: "none" } };
212
+ }
213
+ if (typeof glossary === "string") {
214
+ const resolvedPath = path.resolve(baseDir, glossary);
215
+ const record = await readGlossaryFile(resolvedPath, fs);
216
+ return { glossary: record, provenance: { source: "file", path: resolvedPath } };
217
+ }
218
+ return { glossary, provenance: { source: "inline" } };
219
+ }
71
220
  var FNV_OFFSET_BASIS = 14695981039346656037n;
72
221
  var FNV_PRIME = 1099511628211n;
73
222
  var U64_MASK = (1n << 64n) - 1n;
@@ -192,6 +341,85 @@ var LOCALE_TOKEN = "{locale}";
192
341
  function localeFilePath(cwd, pattern, locale) {
193
342
  return path.resolve(cwd, pattern.replaceAll(LOCALE_TOKEN, locale));
194
343
  }
344
+ var RATE_LIMITED_STATUS = 429;
345
+ var TIMEOUT_STATUS = 408;
346
+ var AUTH_FAILED_STATUSES = /* @__PURE__ */ new Set([401, 403]);
347
+ var RATE_LIMITED_CLASS_NAMES = /* @__PURE__ */ new Set([
348
+ "RateLimitError",
349
+ // openai, @anthropic-ai/sdk
350
+ "TooManyRequestsError"
351
+ // deepl-node
352
+ ]);
353
+ var AUTH_FAILED_CLASS_NAMES = /* @__PURE__ */ new Set([
354
+ "AuthenticationError",
355
+ // openai, @anthropic-ai/sdk (401)
356
+ "PermissionDeniedError",
357
+ // openai, @anthropic-ai/sdk (403)
358
+ "AuthorizationError"
359
+ // deepl-node
360
+ ]);
361
+ var TIMEOUT_CLASS_NAMES = /* @__PURE__ */ new Set([
362
+ "APIConnectionTimeoutError",
363
+ // openai, @anthropic-ai/sdk
364
+ // deepl-node has no distinct timeout class; ConnectionError is its only network-failure
365
+ // class, so it is folded into TIMEOUT rather than the PROVIDER_ERROR fallback.
366
+ "ConnectionError"
367
+ ]);
368
+ var ABORT_ERROR_CLASS_NAMES = /* @__PURE__ */ new Set([
369
+ "APIUserAbortError"
370
+ // openai, @anthropic-ai/sdk
371
+ ]);
372
+ function getErrorStatus(error) {
373
+ if (typeof error !== "object" || error === null || !("status" in error)) {
374
+ return void 0;
375
+ }
376
+ const status = error.status;
377
+ return typeof status === "number" ? status : void 0;
378
+ }
379
+ function readClassName(error) {
380
+ if (typeof error !== "object" || error === null) {
381
+ return void 0;
382
+ }
383
+ return error.constructor?.name;
384
+ }
385
+ function classifyProviderError(error) {
386
+ const status = getErrorStatus(error);
387
+ if (status === RATE_LIMITED_STATUS) {
388
+ return "RATE_LIMITED";
389
+ }
390
+ if (status !== void 0 && AUTH_FAILED_STATUSES.has(status)) {
391
+ return "AUTH_FAILED";
392
+ }
393
+ if (status === TIMEOUT_STATUS) {
394
+ return "TIMEOUT";
395
+ }
396
+ const className = readClassName(error);
397
+ if (className !== void 0) {
398
+ if (RATE_LIMITED_CLASS_NAMES.has(className)) {
399
+ return "RATE_LIMITED";
400
+ }
401
+ if (AUTH_FAILED_CLASS_NAMES.has(className)) {
402
+ return "AUTH_FAILED";
403
+ }
404
+ if (TIMEOUT_CLASS_NAMES.has(className)) {
405
+ return "TIMEOUT";
406
+ }
407
+ }
408
+ return "PROVIDER_ERROR";
409
+ }
410
+ function isAbortShapedError(error) {
411
+ if (error instanceof Error && error.name === "AbortError") {
412
+ return true;
413
+ }
414
+ const className = readClassName(error);
415
+ return className !== void 0 && ABORT_ERROR_CLASS_NAMES.has(className);
416
+ }
417
+ function isAbortError(error, signal) {
418
+ if (signal !== void 0) {
419
+ return signal.aborted === true && isAbortShapedError(error);
420
+ }
421
+ return isAbortShapedError(error);
422
+ }
195
423
  var REDACTED = "[REDACTED]";
196
424
  var KEY_PATTERNS = [
197
425
  // The `\b` anchors `sk-` to a word start so hyphenated words like "risk-" or "task-" pass through.
@@ -226,17 +454,34 @@ var ProviderError = class extends Error {
226
454
  }
227
455
  };
228
456
  var PROVIDER_CALL_FAILED_MESSAGE = "The translation provider request failed.";
229
- async function guardProviderCall(call) {
457
+ var RATE_LIMITED_MESSAGE = "The translation provider rate-limited this request.";
458
+ var TIMEOUT_MESSAGE = "The translation provider request timed out.";
459
+ var AUTH_FAILED_MESSAGE = "The translation provider rejected the request credentials.";
460
+ var MESSAGE_BY_CODE = {
461
+ RATE_LIMITED: RATE_LIMITED_MESSAGE,
462
+ TIMEOUT: TIMEOUT_MESSAGE,
463
+ AUTH_FAILED: AUTH_FAILED_MESSAGE,
464
+ PROVIDER_ERROR: PROVIDER_CALL_FAILED_MESSAGE
465
+ };
466
+ async function guardProviderCall(call, signal) {
467
+ signal?.throwIfAborted();
230
468
  try {
231
469
  return await call();
232
- } catch {
233
- throw new ProviderError("PROVIDER_ERROR", PROVIDER_CALL_FAILED_MESSAGE);
470
+ } catch (error) {
471
+ if (isAbortError(error, signal)) {
472
+ throw error;
473
+ }
474
+ const code = classifyProviderError(error);
475
+ throw new ProviderError(code, MESSAGE_BY_CODE[code]);
234
476
  }
235
477
  }
236
- function checkBatchIntegrity(inputs, extract) {
478
+ function checkBatchIntegrity(inputs, extract, compare) {
237
479
  const integrity = /* @__PURE__ */ new Map();
238
- for (const { key, sourcePlaceholders, translatedValue } of inputs) {
239
- integrity.set(key, checkPlaceholders(sourcePlaceholders, extract(translatedValue)));
480
+ for (const { key, sourceValue, translatedValue } of inputs) {
481
+ integrity.set(
482
+ key,
483
+ compare !== void 0 ? compare(sourceValue, translatedValue) : checkPlaceholders(extract(sourceValue), extract(translatedValue))
484
+ );
240
485
  }
241
486
  return integrity;
242
487
  }
@@ -266,7 +511,7 @@ function toIntegrityInputs(entries, values) {
266
511
  "The provider response is missing one or more keys."
267
512
  );
268
513
  }
269
- return { key: entry.key, sourcePlaceholders: entry.placeholders, translatedValue };
514
+ return { key: entry.key, sourceValue: entry.value, translatedValue };
270
515
  });
271
516
  }
272
517
  function toItem(entry) {
@@ -333,11 +578,16 @@ async function runLlmTranslation(request, mechanism) {
333
578
  const data = validateRequest(request);
334
579
  const payloadJson = JSON.stringify(buildDataPayload(data));
335
580
  const requestedKeys = data.entries.map((entry) => entry.key);
336
- const completion = await mechanism.translate({ payloadJson, requestedKeys });
581
+ const completion = await mechanism.translate({
582
+ payloadJson,
583
+ requestedKeys,
584
+ ...request.signal !== void 0 ? { signal: request.signal } : {}
585
+ });
337
586
  const values = reconcileResult(completion.raw, requestedKeys);
338
587
  const integrity = checkBatchIntegrity(
339
588
  toIntegrityInputs(data.entries, values),
340
- request.extractPlaceholders
589
+ request.extractPlaceholders,
590
+ request.comparePlaceholders
341
591
  );
342
592
  return completion.usage === void 0 ? { values, integrity } : { values, integrity, usage: completion.usage };
343
593
  }
@@ -372,12 +622,29 @@ function requireGeminiKey() {
372
622
  function requireDeepLKey() {
373
623
  return readRequiredEnv(PROVIDER_ENV.deepl);
374
624
  }
625
+ var OPENAI_COMPATIBLE_ENV_VAR = "OPENAI_COMPATIBLE_API_KEY";
626
+ var OPENAI_COMPATIBLE_KEY_PLACEHOLDER = "local";
627
+ function resolveOpenAiCompatibleKey(customEnvVar) {
628
+ const varName = customEnvVar ?? OPENAI_COMPATIBLE_ENV_VAR;
629
+ const value = process.env[varName];
630
+ if (value !== void 0 && value.length > 0) {
631
+ return value;
632
+ }
633
+ if (customEnvVar !== void 0) {
634
+ throw new ProviderError(
635
+ "MISSING_API_KEY",
636
+ `The ${customEnvVar} environment variable is not set.`
637
+ );
638
+ }
639
+ return OPENAI_COMPATIBLE_KEY_PLACEHOLDER;
640
+ }
375
641
  function createDefaultClient() {
376
642
  const sdk = new Anthropic__default.default({ apiKey: requireAnthropicKey(), logLevel: "off" });
377
643
  return {
378
644
  messages: {
379
- create: async (body) => await sdk.messages.create(
380
- body
645
+ create: async (body, options) => await sdk.messages.create(
646
+ body,
647
+ options
381
648
  )
382
649
  }
383
650
  };
@@ -445,9 +712,9 @@ function createAnthropicProvider(config, deps = {}) {
445
712
  }
446
713
  function createMechanism(client, config) {
447
714
  return {
448
- translate: async ({ payloadJson }) => {
715
+ translate: async ({ payloadJson, signal }) => {
449
716
  const body = buildRequest(config, payloadJson);
450
- const message = await callClient(client, body);
717
+ const message = await callClient(client, body, signal);
451
718
  assertNotTruncated(message.stop_reason === "max_tokens");
452
719
  const raw = requireToolInput(message.content);
453
720
  const usage = toUsage(message.usage);
@@ -455,8 +722,11 @@ function createMechanism(client, config) {
455
722
  }
456
723
  };
457
724
  }
458
- function callClient(client, body) {
459
- return guardProviderCall(() => client.messages.create(body));
725
+ function callClient(client, body, signal) {
726
+ return guardProviderCall(
727
+ () => client.messages.create(body, signal !== void 0 ? { signal } : void 0),
728
+ signal
729
+ );
460
730
  }
461
731
  function toUsage(usage) {
462
732
  if (usage === void 0) {
@@ -551,7 +821,7 @@ function zipResults(entries, results) {
551
821
  values.set(entry.key, translatedValue);
552
822
  integrityInputs.push({
553
823
  key: entry.key,
554
- sourcePlaceholders: entry.placeholders,
824
+ sourceValue: entry.value,
555
825
  translatedValue
556
826
  });
557
827
  }
@@ -592,36 +862,97 @@ async function translate(bundle, config, request) {
592
862
  data,
593
863
  protectable,
594
864
  options,
595
- request.extractPlaceholders
865
+ request.extractPlaceholders,
866
+ request.comparePlaceholders,
867
+ request.signal
596
868
  );
597
869
  if (unprotectable.length > 0) {
598
870
  notices.push({ code: "PLACEHOLDER_UNSUPPORTED", message: PLACEHOLDER_UNSUPPORTED_MESSAGE });
599
871
  }
600
872
  return { values, integrity, notices };
601
873
  }
602
- async function translateProtectable(client, data, protectable, options, extract) {
874
+ async function translateProtectable(client, data, protectable, options, extract, compare, signal) {
603
875
  if (protectable.length === 0) {
604
876
  return { values: /* @__PURE__ */ new Map(), integrity: /* @__PURE__ */ new Map() };
605
877
  }
606
878
  const texts = protectable.map((entry) => entry.value);
607
- const results = await callClient2(client, texts, data.sourceLocale, data.targetLocale, options);
879
+ const results = await callClient2(
880
+ client,
881
+ texts,
882
+ data.sourceLocale,
883
+ data.targetLocale,
884
+ options,
885
+ signal
886
+ );
608
887
  const { values, integrityInputs } = zipResults(protectable, results);
609
- const integrity = checkBatchIntegrity(integrityInputs, extract);
888
+ const integrity = checkBatchIntegrity(integrityInputs, extract, compare);
610
889
  return { values, integrity };
611
890
  }
612
- function callClient2(client, texts, sourceLang, targetLang, options) {
613
- return guardProviderCall(() => client.translateText(texts, sourceLang, targetLang, options));
891
+ function callClient2(client, texts, sourceLang, targetLang, options, signal) {
892
+ return guardProviderCall(
893
+ () => client.translateText(texts, sourceLang, targetLang, options),
894
+ signal
895
+ );
614
896
  }
615
897
  var geminiConfigSchema = zod.z.object({
616
898
  model: zod.z.string().min(1),
617
899
  maxOutputTokens: zod.z.number().int().positive()
618
900
  });
901
+ var DEFAULT_GEMINI_RETRY = { attempts: 3, baseDelayMs: 250 };
902
+ function isRetryableStatus(error) {
903
+ const status = getErrorStatus(error);
904
+ return status === 429 || status !== void 0 && status >= 500 && status < 600;
905
+ }
906
+ function isAborted(signal) {
907
+ return signal?.aborted === true;
908
+ }
909
+ function throwAbort(signal) {
910
+ signal?.throwIfAborted();
911
+ throw new DOMException("This operation was aborted.", "AbortError");
912
+ }
913
+ function delay(ms, signal) {
914
+ return new Promise((resolve7) => {
915
+ const onAbort = () => {
916
+ clearTimeout(timer);
917
+ resolve7();
918
+ };
919
+ const timer = setTimeout(() => {
920
+ signal?.removeEventListener("abort", onAbort);
921
+ resolve7();
922
+ }, ms);
923
+ signal?.addEventListener("abort", onAbort, { once: true });
924
+ });
925
+ }
926
+ async function withGeminiRetry(call, signal, config = DEFAULT_GEMINI_RETRY) {
927
+ let attempt = 1;
928
+ while (true) {
929
+ try {
930
+ return await call();
931
+ } catch (error) {
932
+ if (isAborted(signal)) {
933
+ throwAbort(signal);
934
+ }
935
+ const exhausted = attempt >= config.attempts;
936
+ if (exhausted || !isRetryableStatus(error)) {
937
+ throw error;
938
+ }
939
+ await delay(config.baseDelayMs * 2 ** (attempt - 1), signal);
940
+ if (isAborted(signal)) {
941
+ throwAbort(signal);
942
+ }
943
+ attempt += 1;
944
+ }
945
+ }
946
+ }
619
947
  function createDefaultClient3() {
620
948
  const ai = new genai.GoogleGenAI({ apiKey: requireGeminiKey() });
621
949
  return {
622
950
  models: {
623
- generateContent: async (request) => await ai.models.generateContent(
624
- request
951
+ generateContent: (request) => withGeminiRetry(
952
+ async () => await ai.models.generateContent(
953
+ request
954
+ ),
955
+ request.config.abortSignal
625
956
  )
626
957
  }
627
958
  };
@@ -685,7 +1016,7 @@ var GEMINI_SYSTEM_RULES = [
685
1016
  "When a tone is provided, honor it.",
686
1017
  "Respond only with the structured object: exactly one entry per requested key, no commentary, no extra keys, and no key that was not requested."
687
1018
  ].join("\n");
688
- function buildGeminiRequest(config, payloadJson) {
1019
+ function buildGeminiRequest(config, payloadJson, signal) {
689
1020
  return {
690
1021
  model: config.model,
691
1022
  contents: [{ role: "user", parts: [{ text: payloadJson }] }],
@@ -693,7 +1024,8 @@ function buildGeminiRequest(config, payloadJson) {
693
1024
  systemInstruction: GEMINI_SYSTEM_RULES,
694
1025
  responseMimeType: "application/json",
695
1026
  responseSchema: toGeminiSchema(deriveJsonSchema(translationsResultSchema)),
696
- maxOutputTokens: config.maxOutputTokens
1027
+ maxOutputTokens: config.maxOutputTokens,
1028
+ ...signal !== void 0 ? { abortSignal: signal } : {}
697
1029
  }
698
1030
  };
699
1031
  }
@@ -757,15 +1089,15 @@ function createGeminiProvider(config, deps = {}) {
757
1089
  }
758
1090
  function createMechanism2(client, config) {
759
1091
  return {
760
- translate: async ({ payloadJson }) => {
761
- const request = buildGeminiRequest(config, payloadJson);
762
- const response = await callClient3(client, request);
1092
+ translate: async ({ payloadJson, signal }) => {
1093
+ const request = buildGeminiRequest(config, payloadJson, signal);
1094
+ const response = await callClient3(client, request, signal);
763
1095
  return extractGeminiResult(response);
764
1096
  }
765
1097
  };
766
1098
  }
767
- function callClient3(client, request) {
768
- return guardProviderCall(() => client.models.generateContent(request));
1099
+ function callClient3(client, request, signal) {
1100
+ return guardProviderCall(() => client.models.generateContent(request), signal);
769
1101
  }
770
1102
  var openAiConfigSchema = zod.z.object({
771
1103
  model: zod.z.string().min(1),
@@ -776,8 +1108,9 @@ function createDefaultClient4() {
776
1108
  return {
777
1109
  chat: {
778
1110
  completions: {
779
- create: async (body) => await sdk.chat.completions.create(
780
- body
1111
+ create: async (body, options) => await sdk.chat.completions.create(
1112
+ body,
1113
+ options
781
1114
  )
782
1115
  }
783
1116
  }
@@ -795,7 +1128,20 @@ var OPENAI_SYSTEM_RULES = [
795
1128
  "When a tone is provided, honor it.",
796
1129
  "Respond only with the structured object: exactly one entry per requested key, no commentary, no extra keys, and no key that was not requested."
797
1130
  ].join("\n");
798
- function buildOpenAiRequest(config, payloadJson) {
1131
+ function buildResponseFormat(mode) {
1132
+ if (mode === "json-object") {
1133
+ return { type: "json_object" };
1134
+ }
1135
+ return {
1136
+ type: "json_schema",
1137
+ json_schema: {
1138
+ name: RESULT_SCHEMA_NAME,
1139
+ strict: true,
1140
+ schema: deriveJsonSchema(translationsResultSchema)
1141
+ }
1142
+ };
1143
+ }
1144
+ function buildOpenAiRequest(config, payloadJson, mode = "strict-schema") {
799
1145
  return {
800
1146
  model: config.model,
801
1147
  max_completion_tokens: config.maxOutputTokens,
@@ -803,19 +1149,66 @@ function buildOpenAiRequest(config, payloadJson) {
803
1149
  { role: "system", content: OPENAI_SYSTEM_RULES },
804
1150
  { role: "user", content: payloadJson }
805
1151
  ],
806
- response_format: {
807
- type: "json_schema",
808
- json_schema: {
809
- name: RESULT_SCHEMA_NAME,
810
- strict: true,
811
- schema: deriveJsonSchema(translationsResultSchema)
1152
+ response_format: buildResponseFormat(mode)
1153
+ };
1154
+ }
1155
+ function advanceStringScan(char, escaped) {
1156
+ if (escaped) {
1157
+ return { inString: true, escaped: false };
1158
+ }
1159
+ if (char === "\\") {
1160
+ return { inString: true, escaped: true };
1161
+ }
1162
+ return { inString: char !== '"', escaped: false };
1163
+ }
1164
+ function scanBalancedObjectEnd(content, start) {
1165
+ let depth = 0;
1166
+ let inString = false;
1167
+ let escaped = false;
1168
+ for (let i = start; i < content.length; i += 1) {
1169
+ const char = content[i];
1170
+ if (inString) {
1171
+ ({ inString, escaped } = advanceStringScan(char, escaped));
1172
+ continue;
1173
+ }
1174
+ if (char === '"') {
1175
+ inString = true;
1176
+ } else if (char === "{") {
1177
+ depth += 1;
1178
+ } else if (char === "}") {
1179
+ depth -= 1;
1180
+ if (depth === 0) {
1181
+ return i;
812
1182
  }
813
1183
  }
814
- };
1184
+ }
1185
+ return null;
1186
+ }
1187
+ function extractBalancedJson(content) {
1188
+ let searchFrom = 0;
1189
+ while (searchFrom < content.length) {
1190
+ const start = content.indexOf("{", searchFrom);
1191
+ if (start === -1) {
1192
+ return null;
1193
+ }
1194
+ const end = scanBalancedObjectEnd(content, start);
1195
+ if (end === null) {
1196
+ return null;
1197
+ }
1198
+ const candidate = content.slice(start, end + 1);
1199
+ try {
1200
+ JSON.parse(candidate);
1201
+ return candidate;
1202
+ } catch {
1203
+ searchFrom = end + 1;
1204
+ }
1205
+ }
1206
+ return null;
815
1207
  }
816
- function parseContent2(content) {
1208
+ function parseContent2(content, tolerant) {
1209
+ const candidate = tolerant ? extractBalancedJson(content) ?? content : content;
817
1210
  try {
818
- return JSON.parse(content);
1211
+ return JSON.parse(candidate);
819
1212
  } catch {
820
1213
  throw new ProviderError("INVALID_RESPONSE", "The provider returned unparseable content.");
821
1214
  }
@@ -830,7 +1223,7 @@ function toUsage3(usage) {
830
1223
  }
831
1224
  return { inputTokens: prompt_tokens, outputTokens: completion_tokens };
832
1225
  }
833
- function extractOpenAiResult(completion) {
1226
+ function extractOpenAiResult(completion, tolerant = false) {
834
1227
  const choice = completion.choices[0];
835
1228
  if (choice === void 0) {
836
1229
  throw new ProviderError("INVALID_RESPONSE", "The provider returned no message.");
@@ -843,7 +1236,7 @@ function extractOpenAiResult(completion) {
843
1236
  if (message.content === void 0 || message.content === null) {
844
1237
  throw new ProviderError("INVALID_RESPONSE", "The provider returned no translation content.");
845
1238
  }
846
- const raw = parseContent2(message.content);
1239
+ const raw = parseContent2(message.content, tolerant);
847
1240
  const usage = toUsage3(completion.usage);
848
1241
  return usage === void 0 ? { raw } : { raw, usage };
849
1242
  }
@@ -861,15 +1254,84 @@ function createOpenAiProvider(config, deps = {}) {
861
1254
  }
862
1255
  function createMechanism3(client, config) {
863
1256
  return {
864
- translate: async ({ payloadJson }) => {
1257
+ translate: async ({ payloadJson, signal }) => {
865
1258
  const body = buildOpenAiRequest(config, payloadJson);
866
- const completion = await callClient4(client, body);
1259
+ const completion = await callClient4(client, body, signal);
867
1260
  return extractOpenAiResult(completion);
868
1261
  }
869
1262
  };
870
1263
  }
871
- function callClient4(client, body) {
872
- return guardProviderCall(() => client.chat.completions.create(body));
1264
+ function callClient4(client, body, signal) {
1265
+ return guardProviderCall(
1266
+ () => client.chat.completions.create(body, signal !== void 0 ? { signal } : void 0),
1267
+ signal
1268
+ );
1269
+ }
1270
+ var HOSTED_PROVIDER_ENV_VARS = new Set(
1271
+ Object.values(PROVIDER_ENV).map((name) => name.toUpperCase())
1272
+ );
1273
+ function isNotHostedProviderEnvVar(value) {
1274
+ return !HOSTED_PROVIDER_ENV_VARS.has(value.toUpperCase());
1275
+ }
1276
+ function isHttpOrHttpsUrl(value) {
1277
+ try {
1278
+ const { protocol } = new URL(value);
1279
+ return protocol === "http:" || protocol === "https:";
1280
+ } catch {
1281
+ return false;
1282
+ }
1283
+ }
1284
+ var openAiCompatibleConfigSchema = zod.z.object({
1285
+ baseUrl: zod.z.url({ message: "baseUrl must be a valid absolute URL." }).refine(isHttpOrHttpsUrl, { message: "baseUrl must use the http or https scheme." }),
1286
+ model: zod.z.string().min(1),
1287
+ maxOutputTokens: zod.z.number().int().positive(),
1288
+ apiKeyEnvVar: zod.z.string().min(1).refine(isNotHostedProviderEnvVar, {
1289
+ message: "apiKeyEnvVar must not name a hosted provider's environment variable."
1290
+ }).optional()
1291
+ });
1292
+ function createDefaultClient5(config) {
1293
+ const sdk = new OpenAI__default.default({
1294
+ apiKey: resolveOpenAiCompatibleKey(config.apiKeyEnvVar),
1295
+ baseURL: config.baseUrl,
1296
+ logLevel: "off"
1297
+ });
1298
+ return {
1299
+ chat: {
1300
+ completions: {
1301
+ create: async (body, options) => await sdk.chat.completions.create(
1302
+ body,
1303
+ options
1304
+ )
1305
+ }
1306
+ }
1307
+ };
1308
+ }
1309
+ var PROVIDER_ID5 = "openai-compatible";
1310
+ function createOpenAiCompatibleProvider(config, deps = {}) {
1311
+ const validConfig = openAiCompatibleConfigSchema.parse(config);
1312
+ const client = deps.client ?? createDefaultClient5(validConfig);
1313
+ const mechanism = createMechanism4(client, validConfig);
1314
+ return {
1315
+ id: PROVIDER_ID5,
1316
+ kind: "llm",
1317
+ supportsGlossary: true,
1318
+ translateBatch: (request) => runLlmTranslation(request, mechanism)
1319
+ };
1320
+ }
1321
+ function createMechanism4(client, config) {
1322
+ return {
1323
+ translate: async ({ payloadJson, signal }) => {
1324
+ const body = buildOpenAiRequest(config, payloadJson, "strict-schema");
1325
+ const completion = await callClient5(client, body, signal);
1326
+ return extractOpenAiResult(completion, true);
1327
+ }
1328
+ };
1329
+ }
1330
+ function callClient5(client, body, signal) {
1331
+ return guardProviderCall(
1332
+ () => client.chat.completions.create(body, signal !== void 0 ? { signal } : void 0),
1333
+ signal
1334
+ );
873
1335
  }
874
1336
  var SCAFFOLD_MODELS = {
875
1337
  anthropic: "claude-sonnet-4-6",
@@ -880,13 +1342,18 @@ var providerConfigSchema = zod.z.discriminatedUnion("id", [
880
1342
  zod.z.object({ id: zod.z.literal("anthropic"), options: anthropicConfigSchema.strict() }),
881
1343
  zod.z.object({ id: zod.z.literal("openai"), options: openAiConfigSchema.strict() }),
882
1344
  zod.z.object({ id: zod.z.literal("gemini"), options: geminiConfigSchema.strict() }),
883
- zod.z.object({ id: zod.z.literal("deepl"), options: deepLConfigSchema.strict() })
1345
+ zod.z.object({ id: zod.z.literal("deepl"), options: deepLConfigSchema.strict() }),
1346
+ zod.z.object({
1347
+ id: zod.z.literal("openai-compatible"),
1348
+ options: openAiCompatibleConfigSchema.strict()
1349
+ })
884
1350
  ]);
885
1351
  var providerFactories = {
886
1352
  anthropic: (options) => createAnthropicProvider(options),
887
1353
  openai: (options) => createOpenAiProvider(options),
888
1354
  gemini: (options) => createGeminiProvider(options),
889
- deepl: (options) => createDeepLProvider(options)
1355
+ deepl: (options) => createDeepLProvider(options),
1356
+ "openai-compatible": (options) => createOpenAiCompatibleProvider(options)
890
1357
  };
891
1358
  function buildProvider(config) {
892
1359
  const create = providerFactories[config.id];
@@ -903,7 +1370,13 @@ var verbatraConfigSchema = zod.z.strictObject({
903
1370
  pattern: zod.z.string().min(1)
904
1371
  }),
905
1372
  provider: providerConfigSchema,
906
- glossary: zod.z.record(zod.z.string(), zod.z.string()).optional(),
1373
+ /**
1374
+ * A glossary of source terms to preferred target terms, either inline or as a path to a JSON file
1375
+ * carrying the same shape. The two forms are mutually exclusive by type; there is no merge or
1376
+ * precedence between them. A path is resolved and validated by the loader, not by this schema; see
1377
+ * `resolve-glossary.ts`.
1378
+ */
1379
+ glossary: zod.z.union([zod.z.record(zod.z.string(), zod.z.string()), zod.z.string().min(1)]).optional(),
907
1380
  tone: zod.z.enum(["formal", "informal", "neutral"]).optional(),
908
1381
  /**
909
1382
  * Opt-in orphan pruning, off by default. When true, keys present in a target file but absent from
@@ -957,7 +1430,7 @@ function formatIssues(error) {
957
1430
  return issue.code === "unrecognized_keys" ? `${base} (API keys are read from the environment, not the config)` : base;
958
1431
  }).join("; ");
959
1432
  }
960
- function validate(input) {
1433
+ function parseConfig(input) {
961
1434
  const parsed = verbatraConfigSchema.safeParse(input);
962
1435
  if (!parsed.success) {
963
1436
  throw new SdkError(
@@ -967,7 +1440,16 @@ function validate(input) {
967
1440
  }
968
1441
  return parsed.data;
969
1442
  }
970
- async function loadExplicit(explorer, configPath, cwd) {
1443
+ async function finalizeConfig(parsed, baseDir, fs) {
1444
+ const { glossary: glossaryInput, ...rest } = parsed;
1445
+ const resolved = await resolveGlossary(glossaryInput, baseDir, fs);
1446
+ const config = {
1447
+ ...rest,
1448
+ ...resolved.glossary !== void 0 ? { glossary: resolved.glossary } : {}
1449
+ };
1450
+ return { config, glossary: resolved.provenance };
1451
+ }
1452
+ async function loadExplicitWithMeta(explorer, configPath, cwd, fs$1) {
971
1453
  const resolved = path.resolve(cwd ?? process.cwd(), configPath);
972
1454
  if (!fs.existsSync(resolved)) {
973
1455
  throw new SdkError("CONFIG_NOT_FOUND", `No verbatra configuration file at ${resolved}.`);
@@ -979,18 +1461,23 @@ async function loadExplicit(explorer, configPath, cwd) {
979
1461
  const detail = error instanceof Error ? error.message : String(error);
980
1462
  throw new SdkError("CONFIG_INVALID", `Failed to load the verbatra configuration: ${detail}`);
981
1463
  }
982
- return validate(result?.config);
1464
+ const parsed = parseConfig(result?.config);
1465
+ const { config, glossary } = await finalizeConfig(parsed, path.dirname(resolved), fs$1);
1466
+ return { config, source: { kind: "explicit", filepath: resolved }, glossary };
983
1467
  }
984
- async function loadConfig(options = {}) {
1468
+ async function loadConfigWithMeta(options = {}) {
1469
+ const fs = options.fs ?? defaultFs;
985
1470
  if (options.configOverride !== void 0) {
986
- return validate(options.configOverride);
1471
+ const parsed2 = parseConfig(options.configOverride);
1472
+ const { config: config2, glossary: glossary2 } = await finalizeConfig(parsed2, options.cwd ?? process.cwd(), fs);
1473
+ return { config: config2, source: { kind: "override" }, glossary: glossary2 };
987
1474
  }
988
1475
  const explorer = cosmiconfig.cosmiconfig(MODULE_NAME, {
989
1476
  searchPlaces: SEARCH_PLACES,
990
1477
  loaders: { ".ts": cosmiconfigTypescriptLoader.TypeScriptLoader() }
991
1478
  });
992
1479
  if (options.configPath !== void 0) {
993
- return loadExplicit(explorer, options.configPath, options.cwd);
1480
+ return loadExplicitWithMeta(explorer, options.configPath, options.cwd, fs);
994
1481
  }
995
1482
  let result;
996
1483
  try {
@@ -1005,99 +1492,14 @@ async function loadConfig(options = {}) {
1005
1492
  "No verbatra configuration found. Create a verbatra.config.ts, a .verbatrarc.json, or a 'verbatra' property in package.json."
1006
1493
  );
1007
1494
  }
1008
- return validate(result.config);
1009
- }
1010
- async function readBoundedUtf8(handle, size) {
1011
- const buffer = Buffer.allocUnsafe(size);
1012
- let offset = 0;
1013
- while (offset < size) {
1014
- const { bytesRead } = await handle.read(buffer, offset, size - offset, offset);
1015
- if (bytesRead === 0) {
1016
- break;
1017
- }
1018
- offset += bytesRead;
1019
- }
1020
- return buffer.toString("utf8", 0, offset);
1021
- }
1022
- async function readBounded(path, maxBytes) {
1023
- let handle;
1024
- try {
1025
- handle = await promises.open(path, "r");
1026
- } catch {
1027
- return { kind: "missing" };
1028
- }
1029
- try {
1030
- const info = await handle.stat();
1031
- if (!info.isFile()) {
1032
- return { kind: "missing" };
1033
- }
1034
- if (info.size > maxBytes) {
1035
- return { kind: "too-large" };
1036
- }
1037
- return { kind: "ok", content: await readBoundedUtf8(handle, info.size) };
1038
- } finally {
1039
- await handle.close();
1040
- }
1041
- }
1042
- async function readBoundedBytesInto(handle, size) {
1043
- const buffer = Buffer.allocUnsafeSlow(size);
1044
- let offset = 0;
1045
- while (offset < size) {
1046
- const { bytesRead } = await handle.read(buffer, offset, size - offset, offset);
1047
- if (bytesRead === 0) {
1048
- break;
1049
- }
1050
- offset += bytesRead;
1051
- }
1052
- return new Uint8Array(buffer.buffer, buffer.byteOffset, offset);
1053
- }
1054
- async function readBoundedBytes(path, maxBytes) {
1055
- let handle;
1056
- try {
1057
- handle = await promises.open(path, "r");
1058
- } catch {
1059
- return { kind: "missing" };
1060
- }
1061
- try {
1062
- const info = await handle.stat();
1063
- if (!info.isFile()) {
1064
- return { kind: "missing" };
1065
- }
1066
- if (info.size > maxBytes) {
1067
- return { kind: "too-large" };
1068
- }
1069
- return { kind: "ok", bytes: await readBoundedBytesInto(handle, info.size) };
1070
- } finally {
1071
- await handle.close();
1072
- }
1495
+ const parsed = parseConfig(result.config);
1496
+ const { config, glossary } = await finalizeConfig(parsed, path.dirname(result.filepath), fs);
1497
+ return { config, source: { kind: "search", filepath: result.filepath }, glossary };
1073
1498
  }
1074
- function tempFileName(path$1) {
1075
- return path.join(path.dirname(path$1), `.${path.basename(path$1)}.tmp-${process.pid}-${Date.now()}-${crypto.randomUUID()}`);
1076
- }
1077
- async function atomicWrite(path, data) {
1078
- const tmp = tempFileName(path);
1079
- await (typeof data === "string" ? promises.writeFile(tmp, data, "utf8") : promises.writeFile(tmp, data));
1080
- try {
1081
- await promises.rename(tmp, path);
1082
- } catch (error) {
1083
- await promises.rm(tmp, { force: true });
1084
- throw error;
1085
- }
1499
+ async function loadConfig(options = {}) {
1500
+ const { config } = await loadConfigWithMeta(options);
1501
+ return config;
1086
1502
  }
1087
- var defaultFs = {
1088
- async fileExists(path) {
1089
- try {
1090
- await promises.access(path);
1091
- return true;
1092
- } catch {
1093
- return false;
1094
- }
1095
- },
1096
- readFileBounded: (path, maxBytes) => readBounded(path, maxBytes),
1097
- readBytesBounded: (path, maxBytes) => readBoundedBytes(path, maxBytes),
1098
- writeFile: (path, data) => atomicWrite(path, data),
1099
- writeBytes: (path, data) => atomicWrite(path, data)
1100
- };
1101
1503
  var LOCK_FILE_NAME = "verbatra.lock.json";
1102
1504
  var CURRENT_VERSION = 1;
1103
1505
  var EMPTY_LOCK = { version: CURRENT_VERSION, locales: {} };
@@ -1130,6 +1532,12 @@ async function readLockFile(path, fs) {
1130
1532
  if (!result.success) {
1131
1533
  throw new SdkError("LOCK_FILE_INVALID", `The lock-file at ${path} has an unexpected shape.`);
1132
1534
  }
1535
+ if (result.data.version !== CURRENT_VERSION) {
1536
+ throw new SdkError(
1537
+ "LOCK_FILE_INVALID",
1538
+ `The lock-file at ${path} has version ${result.data.version}, but this version of verbatra supports version ${CURRENT_VERSION}.`
1539
+ );
1540
+ }
1133
1541
  return result.data;
1134
1542
  }
1135
1543
  function baselineFor(lock, locale) {
@@ -1262,6 +1670,150 @@ function icuInvalidKeys(entries) {
1262
1670
  }
1263
1671
  return invalid;
1264
1672
  }
1673
+ function tokenOf2(element) {
1674
+ switch (element.type) {
1675
+ case icuMessageformatParser.TYPE.argument:
1676
+ case icuMessageformatParser.TYPE.number:
1677
+ case icuMessageformatParser.TYPE.date:
1678
+ case icuMessageformatParser.TYPE.time:
1679
+ case icuMessageformatParser.TYPE.select:
1680
+ case icuMessageformatParser.TYPE.plural:
1681
+ return `{${element.value}}`;
1682
+ case icuMessageformatParser.TYPE.tag:
1683
+ return `<${element.value}>`;
1684
+ default:
1685
+ return void 0;
1686
+ }
1687
+ }
1688
+ function isBranching(element) {
1689
+ return element.type === icuMessageformatParser.TYPE.plural || element.type === icuMessageformatParser.TYPE.select;
1690
+ }
1691
+ function isTag(element) {
1692
+ return element.type === icuMessageformatParser.TYPE.tag;
1693
+ }
1694
+ function layerTokens(elements) {
1695
+ const tokens = [];
1696
+ for (const element of elements) {
1697
+ const token = tokenOf2(element);
1698
+ if (token !== void 0) {
1699
+ tokens.push(token);
1700
+ }
1701
+ }
1702
+ return tokens;
1703
+ }
1704
+ function deepTokens(elements) {
1705
+ const tokens = [];
1706
+ for (const element of elements) {
1707
+ const token = tokenOf2(element);
1708
+ if (token !== void 0) {
1709
+ tokens.push(token);
1710
+ }
1711
+ if (isTag(element)) {
1712
+ tokens.push(...deepTokens(element.children));
1713
+ } else if (isBranching(element)) {
1714
+ for (const option of Object.values(element.options)) {
1715
+ tokens.push(...deepTokens(option.value));
1716
+ }
1717
+ }
1718
+ }
1719
+ return tokens;
1720
+ }
1721
+ function branchesByCategory(element) {
1722
+ return new Map(
1723
+ Object.entries(element.options).map(([category, option]) => [category, option.value])
1724
+ );
1725
+ }
1726
+ function findUnconsumed(target, consumed, isMatch) {
1727
+ for (let index = 0; index < target.length; index += 1) {
1728
+ if (consumed.has(index)) {
1729
+ continue;
1730
+ }
1731
+ const candidate = target[index];
1732
+ if (candidate !== void 0 && isMatch(candidate)) {
1733
+ return { element: candidate, index };
1734
+ }
1735
+ }
1736
+ return void 0;
1737
+ }
1738
+ function findMatchingBranching(source, target, consumed) {
1739
+ return findUnconsumed(
1740
+ target,
1741
+ consumed,
1742
+ (candidate) => isBranching(candidate) && candidate.value === source.value
1743
+ );
1744
+ }
1745
+ function findMatchingTag(source, target, consumed) {
1746
+ return findUnconsumed(
1747
+ target,
1748
+ consumed,
1749
+ (candidate) => isTag(candidate) && candidate.value === source.value
1750
+ );
1751
+ }
1752
+ function combineResults(results) {
1753
+ const missing = results.flatMap((result) => result.missing).sort();
1754
+ const extra = results.flatMap((result) => result.extra).sort();
1755
+ return { matches: missing.length === 0 && extra.length === 0, missing, extra, reordered: false };
1756
+ }
1757
+ function compareAgainstSourceUnion(sourceBranches, targetBranch) {
1758
+ const union = /* @__PURE__ */ new Set();
1759
+ for (const branch of sourceBranches.values()) {
1760
+ for (const token of deepTokens(branch)) {
1761
+ union.add(token);
1762
+ }
1763
+ }
1764
+ const extra = deepTokens(targetBranch).filter((token) => !union.has(token)).sort();
1765
+ return { matches: extra.length === 0, missing: [], extra, reordered: false };
1766
+ }
1767
+ function compareBranching(source, target) {
1768
+ const sourceBranches = branchesByCategory(source);
1769
+ const targetBranches = branchesByCategory(target);
1770
+ const categories = /* @__PURE__ */ new Set([...sourceBranches.keys(), ...targetBranches.keys()]);
1771
+ const results = [];
1772
+ for (const category of categories) {
1773
+ const sourceBranch = sourceBranches.get(category);
1774
+ const targetBranch = targetBranches.get(category);
1775
+ if (sourceBranch !== void 0 && targetBranch !== void 0) {
1776
+ results.push(compareElements(sourceBranch, targetBranch));
1777
+ } else if (targetBranch !== void 0) {
1778
+ results.push(compareAgainstSourceUnion(sourceBranches, targetBranch));
1779
+ }
1780
+ }
1781
+ return combineResults(results);
1782
+ }
1783
+ function nestedResults(source, target) {
1784
+ const results = [];
1785
+ const consumed = /* @__PURE__ */ new Set();
1786
+ for (const element of source) {
1787
+ if (isBranching(element)) {
1788
+ const found = findMatchingBranching(element, target, consumed);
1789
+ if (found !== void 0) {
1790
+ consumed.add(found.index);
1791
+ results.push(compareBranching(element, found.element));
1792
+ }
1793
+ } else if (isTag(element)) {
1794
+ const found = findMatchingTag(element, target, consumed);
1795
+ if (found !== void 0) {
1796
+ consumed.add(found.index);
1797
+ results.push(compareElements(element.children, found.element.children));
1798
+ }
1799
+ }
1800
+ }
1801
+ return results;
1802
+ }
1803
+ function compareElements(source, target) {
1804
+ const layer = checkPlaceholders(layerTokens(source), layerTokens(target));
1805
+ const nested = nestedResults(source, target);
1806
+ return nested.length === 0 ? layer : combineResults([layer, ...nested]);
1807
+ }
1808
+ function compareIcuPlaceholders(sourceValue, targetValue) {
1809
+ try {
1810
+ const sourceAst = icuMessageformatParser.parse(sourceValue);
1811
+ const targetAst = icuMessageformatParser.parse(targetValue);
1812
+ return compareElements(sourceAst, targetAst);
1813
+ } catch {
1814
+ return checkPlaceholders(icuPlaceholders(sourceValue), icuPlaceholders(targetValue));
1815
+ }
1816
+ }
1265
1817
  var AdapterError = class extends Error {
1266
1818
  code;
1267
1819
  constructor(code, message) {
@@ -1563,9 +2115,9 @@ function unflattenEntries(entries) {
1563
2115
  }
1564
2116
  return root;
1565
2117
  }
1566
- function toEntries(content, namespace, parse2, deriveEntry, keyMode, validateTree) {
2118
+ function toEntries(content, namespace, parse3, deriveEntry, keyMode, validateTree) {
1567
2119
  try {
1568
- const tree = parse2(content);
2120
+ const tree = parse3(content);
1569
2121
  validateTree?.(tree);
1570
2122
  return flattenTree(tree, namespace, deriveEntry, keyMode);
1571
2123
  } catch (error) {
@@ -1577,7 +2129,7 @@ function createTreeFileAdapter(options) {
1577
2129
  format,
1578
2130
  extensions,
1579
2131
  sniff,
1580
- parse: parse2,
2132
+ parse: parse3,
1581
2133
  serialize,
1582
2134
  deriveEntry,
1583
2135
  extractPlaceholders,
@@ -1585,6 +2137,7 @@ function createTreeFileAdapter(options) {
1585
2137
  validateMessage,
1586
2138
  validateTree,
1587
2139
  buildWriteTree,
2140
+ comparePlaceholders,
1588
2141
  keyMode = "literal-leaf"
1589
2142
  } = options;
1590
2143
  return {
@@ -1592,10 +2145,11 @@ function createTreeFileAdapter(options) {
1592
2145
  canHandle: buildCanHandle(extensions, sniff),
1593
2146
  extractPlaceholders,
1594
2147
  validateMessage: validateMessage ?? (() => true),
2148
+ ...comparePlaceholders !== void 0 ? { comparePlaceholders } : {},
1595
2149
  async read(filePath, locale) {
1596
2150
  const content = await readFileContent(filePath);
1597
2151
  const namespace = namespaceOf(filePath);
1598
- const entries = toEntries(content, namespace, parse2, deriveEntry, keyMode, validateTree);
2152
+ const entries = toEntries(content, namespace, parse3, deriveEntry, keyMode, validateTree);
1599
2153
  const resource = { locale, namespace, format, entries };
1600
2154
  const invalidIcuKeys = computeIcu(entries, computeInvalidIcuKeys);
1601
2155
  return { resource, invalidIcuKeys };
@@ -1691,6 +2245,7 @@ function createArbAdapter() {
1691
2245
  deriveEntry: icuDeriveEntry,
1692
2246
  computeInvalidIcuKeys: icuInvalidKeys,
1693
2247
  validateMessage: icuIsValid,
2248
+ comparePlaceholders: compareIcuPlaceholders,
1694
2249
  buildWriteTree: buildArbWriteTree
1695
2250
  });
1696
2251
  }
@@ -1754,7 +2309,8 @@ function createNextIntlJsonAdapter() {
1754
2309
  extractPlaceholders: icuPlaceholders,
1755
2310
  deriveEntry: icuDeriveEntry,
1756
2311
  computeInvalidIcuKeys: icuInvalidKeys,
1757
- validateMessage: icuIsValid
2312
+ validateMessage: icuIsValid,
2313
+ comparePlaceholders: compareIcuPlaceholders
1758
2314
  });
1759
2315
  }
1760
2316
  function assertNoDottedNestedKey(tree) {
@@ -1956,6 +2512,7 @@ function extractXliffPlaceholders(value) {
1956
2512
  return result;
1957
2513
  }
1958
2514
  var ELEMENT_NODE = 1;
2515
+ var TEXT_NODE = 3;
1959
2516
  function isElement(node) {
1960
2517
  return node.nodeType === ELEMENT_NODE;
1961
2518
  }
@@ -2046,6 +2603,12 @@ function parseXliffEntries(content, namespace) {
2046
2603
  const serializer = new xmldom.XMLSerializer();
2047
2604
  const out = /* @__PURE__ */ new Map();
2048
2605
  for (const unit of walkUnits(root)) {
2606
+ if (out.has(unit.key)) {
2607
+ throw new AdapterError(
2608
+ "INVALID_STRUCTURE",
2609
+ "The XLIFF file has two trans-units with the same id."
2610
+ );
2611
+ }
2049
2612
  const value = unitValue(serializer, unit);
2050
2613
  out.set(unit.key, {
2051
2614
  key: unit.key,
@@ -2072,10 +2635,67 @@ async function readDestination(filePath) {
2072
2635
  }
2073
2636
  return outcome.content;
2074
2637
  }
2638
+ var XLIFF_INLINE_ELEMENTS = /* @__PURE__ */ new Set(["x", "g", "bx", "ex", "ph", "it", "mrk"]);
2639
+ var XLIFF_NAMESPACES = /* @__PURE__ */ new Set([
2640
+ "urn:oasis:names:tc:xliff:document:1.2",
2641
+ "urn:oasis:names:tc:xliff:document:2.0"
2642
+ ]);
2643
+ var INLINE_ELEMENT_ATTRIBUTES = {
2644
+ x: ["id", "ctype"],
2645
+ ph: ["id", "ctype"],
2646
+ g: ["id", "ctype"],
2647
+ bx: ["id", "rid", "ctype"],
2648
+ ex: ["id", "rid"],
2649
+ it: ["id", "pos", "ctype"],
2650
+ mrk: ["id", "mtype"]
2651
+ };
2652
+ function isAllowedInlineElement(element) {
2653
+ const namespace = element.namespaceURI;
2654
+ const hasAllowedNamespace = namespace === null || XLIFF_NAMESPACES.has(namespace);
2655
+ return hasAllowedNamespace && XLIFF_INLINE_ELEMENTS.has(element.localName ?? "");
2656
+ }
2657
+ function isAllowedFragmentNode(node) {
2658
+ if (node.nodeType === TEXT_NODE) {
2659
+ return true;
2660
+ }
2661
+ return isElement(node) && isAllowedInlineElement(node);
2662
+ }
2663
+ function allDescendantNodes(node) {
2664
+ const children = Array.from(node.childNodes);
2665
+ return children.flatMap((child) => [child, ...allDescendantNodes(child)]);
2666
+ }
2667
+ function hasDisallowedNode(root) {
2668
+ return allDescendantNodes(root).some((node) => !isAllowedFragmentNode(node));
2669
+ }
2670
+ function attributeNames(element) {
2671
+ const names = [];
2672
+ for (let i = 0; i < element.attributes.length; i += 1) {
2673
+ const attr = element.attributes.item(i);
2674
+ if (attr !== null) {
2675
+ names.push(attr.name);
2676
+ }
2677
+ }
2678
+ return names;
2679
+ }
2680
+ function sanitizeInlineAttributes(root) {
2681
+ for (const el of collectByTag(root, "*")) {
2682
+ const allowed = INLINE_ELEMENT_ATTRIBUTES[el.localName ?? ""] ?? [];
2683
+ for (const name of attributeNames(el)) {
2684
+ if (!allowed.includes(name)) {
2685
+ el.removeAttribute(name);
2686
+ }
2687
+ }
2688
+ }
2689
+ }
2075
2690
  function fragmentNodes(parser, value) {
2691
+ assertNoDoctype(value);
2076
2692
  try {
2077
2693
  const root = parser.parseFromString(`<wrapper>${value}</wrapper>`, "text/xml").documentElement;
2078
- return root === null ? null : Array.from(root.childNodes);
2694
+ if (root === null || hasDisallowedNode(root)) {
2695
+ return null;
2696
+ }
2697
+ sanitizeInlineAttributes(root);
2698
+ return Array.from(root.childNodes);
2079
2699
  } catch {
2080
2700
  return null;
2081
2701
  }
@@ -2254,6 +2874,40 @@ async function diff(input, deps = {}) {
2254
2874
  return { hasPendingChanges: locales.some((entry) => entry.hasPendingChanges), locales };
2255
2875
  }
2256
2876
 
2877
+ // src/flow/lock-state.ts
2878
+ function toLockLocaleState(locale, keyCount, diff2) {
2879
+ return {
2880
+ locale,
2881
+ keyCount,
2882
+ missing: diff2.missing.length,
2883
+ stale: diff2.changed.length,
2884
+ upToDate: diff2.unchanged.length
2885
+ };
2886
+ }
2887
+ async function lockState(input, deps = {}) {
2888
+ const config = input.config;
2889
+ const cwd = input.cwd ?? process.cwd();
2890
+ const fs = deps.fs ?? defaultFs;
2891
+ const locales = selectLocales(config, input.locales);
2892
+ const path = lockFilePath(cwd);
2893
+ const exists = await fs.fileExists(path);
2894
+ if (!exists) {
2895
+ return { exists: false };
2896
+ }
2897
+ const lock = await readLockFile(path, fs);
2898
+ const adapter = selectAdapter(config.format, deps.adapterRegistry);
2899
+ const source = await readSource(config, cwd, fs, adapter);
2900
+ const localeStates = await Promise.all(
2901
+ locales.map(async (locale) => {
2902
+ const target = await readTarget(cwd, config, adapter, fs, locale);
2903
+ const baseline = baselineFor(lock, locale);
2904
+ const diff2 = diffResources(source.resource, target, { baseline });
2905
+ return toLockLocaleState(locale, baseline.size, diff2);
2906
+ })
2907
+ );
2908
+ return { exists: true, version: lock.version, locales: localeStates };
2909
+ }
2910
+
2257
2911
  // src/selection/select-provider.ts
2258
2912
  function selectProvider(config, createProvider = buildProvider) {
2259
2913
  try {
@@ -2495,7 +3149,8 @@ function buildRequest2(context, entries) {
2495
3149
  entries,
2496
3150
  extractPlaceholders: context.adapter.extractPlaceholders,
2497
3151
  ...context.glossary !== void 0 ? { glossary: context.glossary } : {},
2498
- ...context.tone !== void 0 ? { tone: context.tone } : {}
3152
+ ...context.tone !== void 0 ? { tone: context.tone } : {},
3153
+ ...context.adapter.comparePlaceholders !== void 0 ? { comparePlaceholders: context.adapter.comparePlaceholders } : {}
2499
3154
  };
2500
3155
  }
2501
3156
  async function generatePluralForms(context) {
@@ -2565,7 +3220,8 @@ function buildRequest3(params, entries) {
2565
3220
  entries,
2566
3221
  extractPlaceholders: params.adapter.extractPlaceholders,
2567
3222
  ...params.glossary !== void 0 ? { glossary: params.glossary } : {},
2568
- ...params.tone !== void 0 ? { tone: params.tone } : {}
3223
+ ...params.tone !== void 0 ? { tone: params.tone } : {},
3224
+ ...params.adapter.comparePlaceholders !== void 0 ? { comparePlaceholders: params.adapter.comparePlaceholders } : {}
2569
3225
  };
2570
3226
  }
2571
3227
  async function runLocale(params) {
@@ -3028,7 +3684,7 @@ async function streamEntryBounded(file, remaining) {
3028
3684
  "The workbook decompresses to more than the permitted maximum bytes."
3029
3685
  );
3030
3686
  }
3031
- return Buffer.concat(chunks).toString("utf8");
3687
+ return { raw, content: Buffer.concat(chunks).toString("utf8") };
3032
3688
  }
3033
3689
  async function guardWorkbookBytes(bytes, limits) {
3034
3690
  let zip;
@@ -3060,14 +3716,8 @@ async function guardWorkbookBytes(bytes, limits) {
3060
3716
  let actualTotal = 0;
3061
3717
  for (const file of files) {
3062
3718
  const remaining = limits.maxDecompressedBytes - actualTotal;
3063
- const content = await streamEntryBounded(file, remaining);
3064
- actualTotal += Buffer.byteLength(content, "utf8");
3065
- if (actualTotal > limits.maxDecompressedBytes) {
3066
- throw new ExchangeError(
3067
- "WORKBOOK_INVALID",
3068
- `The workbook decompresses to more than the maximum of ${limits.maxDecompressedBytes} bytes.`
3069
- );
3070
- }
3719
+ const { raw, content } = await streamEntryBounded(file, remaining);
3720
+ actualTotal += raw;
3071
3721
  assertNoDoctype2(file.name, content);
3072
3722
  }
3073
3723
  }
@@ -3256,10 +3906,7 @@ function judge(row, sourceEntry, adapter) {
3256
3906
  if (contentHash(sourceEntry) !== row.sourceHash) {
3257
3907
  return "drift";
3258
3908
  }
3259
- const integrity = checkPlaceholders(
3260
- sourceEntry.placeholders,
3261
- adapter.extractPlaceholders(row.translation)
3262
- );
3909
+ const integrity = adapter.comparePlaceholders !== void 0 ? adapter.comparePlaceholders(sourceEntry.value, row.translation) : checkPlaceholders(sourceEntry.placeholders, adapter.extractPlaceholders(row.translation));
3263
3910
  if (!integrity.matches) {
3264
3911
  return "placeholder";
3265
3912
  }
@@ -3459,6 +4106,13 @@ async function importWorkbook(input, deps = {}) {
3459
4106
  return { dryRun, locales: summaries, succeeded, failed };
3460
4107
  }
3461
4108
 
4109
+ // src/lock/load-lock-file.ts
4110
+ async function loadLockFile(input = {}, deps = {}) {
4111
+ const cwd = input.cwd ?? process.cwd();
4112
+ const fs = deps.fs ?? defaultFs;
4113
+ return readLockFile(lockFilePath(cwd), fs);
4114
+ }
4115
+
3462
4116
  // src/scaffolding.ts
3463
4117
  var scaffoldingMetadata = {
3464
4118
  /** Provider id -> the environment variable its API key is read from. Owned by ai-providers. */
@@ -3575,6 +4229,7 @@ async function watch(input, deps = {}) {
3575
4229
  }
3576
4230
 
3577
4231
  exports.DEFAULT_WORKBOOK_PATH = DEFAULT_WORKBOOK_PATH;
4232
+ exports.LOCK_FILE_NAME = LOCK_FILE_NAME;
3578
4233
  exports.SdkError = SdkError;
3579
4234
  exports.check = check;
3580
4235
  exports.defineConfig = defineConfig;
@@ -3582,6 +4237,9 @@ exports.diff = diff;
3582
4237
  exports.exportWorkbook = exportWorkbook;
3583
4238
  exports.importWorkbook = importWorkbook;
3584
4239
  exports.loadConfig = loadConfig;
4240
+ exports.loadConfigWithMeta = loadConfigWithMeta;
4241
+ exports.loadLockFile = loadLockFile;
4242
+ exports.lockState = lockState;
3585
4243
  exports.scaffoldingMetadata = scaffoldingMetadata;
3586
4244
  exports.translate = translate2;
3587
4245
  exports.verbatraConfigSchema = verbatraConfigSchema;