@semiont/core 0.5.29 → 0.5.31

package/dist/index.js

@@ -1,6 +1,6 @@
-export { createTomlConfigLoader, loadTomlConfig } from './chunk-VBMXNYPL.js';
-import { BUS_OPERATIONS } from './chunk-MM4HBHMM.js';
-export { BRIDGED_CHANNELS, BUS_OPERATIONS, EventBus, ScopedEventBus, accessToken, annotationUri, authCode, baseUrl, busLog, busLogEnabled, cloneToken, email, entityType, googleCredential, jobId, mcpToken, refreshToken, resourceAnnotationUri, resourceUri, searchQuery, setBusLogTraceIdProvider, userDID } from './chunk-MM4HBHMM.js';
+export { createTomlConfigLoader, loadTomlConfig } from './chunk-UGI6IOOX.js';
+import { BUS_OPERATIONS } from './chunk-W6N56PWO.js';
+export { BRIDGED_CHANNELS, BUS_OPERATIONS, EventBus, ScopedEventBus, accessToken, annotationUri, authCode, baseUrl, busLog, busLogEnabled, cloneToken, email, entityType, googleCredential, jobId, mcpToken, refreshToken, resourceAnnotationUri, resourceUri, searchQuery, setBusLogTraceIdProvider, userDID } from './chunk-W6N56PWO.js';
 import './chunk-YLJ4XMA6.js';
 import { Observable, merge, TimeoutError, throwError, firstValueFrom } from 'rxjs';
 import { filter, map, take, timeout, catchError, defaultIfEmpty } from 'rxjs/operators';
@@ -122,6 +122,9 @@ var CHANNEL_SCHEMAS = {
   "frame:add-tag-schema": "FrameAddTagSchemaCommand",
   "mark:create-ok": "MarkCreateOk",
   "mark:create-failed": "CommandError",
+  "mark:commit": "MarkCommitCommand",
+  "mark:commit-ok": "MarkCommitOk",
+  "mark:commit-failed": "CommandError",
   "mark:delete-ok": "MarkDeleteOk",
   "mark:delete-failed": "CommandError",
   "mark:archive-ok": null,
@@ -187,10 +190,6 @@ var CHANNEL_SCHEMAS = {
   "browse:anchored-text-result": "BrowseAnchoredTextResult",
   "browse:anchored-text-failed": null,
   // { correlationId } & CommandError
-  "browse:anchored-text-by-checksum-requested": "BrowseAnchoredTextByChecksumRequest",
-  "browse:anchored-text-by-checksum-result": "BrowseAnchoredTextResult",
-  "browse:anchored-text-by-checksum-failed": null,
-  // { correlationId } & CommandError
   "browse:resources-requested": "BrowseResourcesRequest",
   "browse:resources-result": "BrowseResourcesResult",
   "browse:resources-failed": null,
@@ -258,8 +257,10 @@ var CHANNEL_SCHEMAS = {
   "job:report-progress": "JobReportProgressCommand",
   "job:complete": "JobCompleteCommand",
   "job:fail": "JobFailCommand",
+  "job:checkpoint": "JobCheckpointCommand",
   "job:queued": "JobQueuedEvent",
   "job:cancel-requested": "JobCancelRequest",
+  "job:cancel": "JobCancelCommand",
   "job:status-requested": "JobStatusRequest",
   "job:create": "JobCreateCommand",
   "job:claim": "JobClaimCommand",
@@ -1008,10 +1009,32 @@ var BusRequestError = class extends SemiontError {
     this.name = "BusRequestError";
   }
 };
+function replyChannelsFor(channels) {
+  const out = /* @__PURE__ */ new Set();
+  for (const ch of channels) {
+    const op = BUS_OPERATIONS[ch];
+    if (!op) continue;
+    out.add(op.result);
+    out.add(op.failure);
+    if ("progress" in op && op.progress) out.add(op.progress);
+  }
+  return [...out];
+}
 async function busRequest(bus, operation, payload, timeoutMs = 3e4) {
   const correlationId = uuidV4();
   const fullPayload = { ...payload, correlationId };
   const { result: resultChannel, failure: failureChannel } = BUS_OPERATIONS[operation];
+  if (bus.isSubscribed) {
+    for (const replyChannel of [resultChannel, failureChannel]) {
+      if (!bus.isSubscribed(replyChannel)) {
+        throw new BusRequestError(
+          `Transport is not subscribed to reply channel ${replyChannel} \u2014 a reply to ${operation} can never arrive. Add this operation's reply channels to the transport's channel set.`,
+          "bus.unsubscribed",
+          { channel: operation, resultChannel, failureChannel }
+        );
+      }
+    }
+  }
   const result$ = merge(
     bus.stream(resultChannel).pipe(
       filter((e) => e.correlationId === correlationId),
@@ -1726,26 +1749,26 @@ var storedBinary = (extension, label) => ({
   label,
   render: "none",
   anchoring: "none",
-  extractText: "none",
+  textSource: "none",
   authorable: false,
   uploadable: true,
   generatable: false
 });
 var storedText = (extension, label) => ({
   ...storedBinary(extension, label),
-  extractText: "decode"
+  textSource: "decode"
 });
 var MEDIA_TYPES = {
   // Full-capability tier
   // `generatable: application/pdf` — the Typst renderer (PDF-GENERATION P3):
   // the model writes Typst, the worker's pinned binary compiles it.
-  "text/markdown": { extension: ".md", label: "Markdown", render: "text", anchoring: "text-selector", extractText: "decode", authorable: true, uploadable: true, generatable: true },
-  "text/plain": { extension: ".txt", label: "Plain Text", render: "text", anchoring: "text-selector", extractText: "decode", authorable: true, uploadable: true, generatable: true },
-  "text/html": { extension: ".html", label: "HTML", render: "text", anchoring: "text-selector", extractText: "decode", authorable: true, uploadable: true, generatable: false },
-  "application/json": { extension: ".json", label: "JSON", render: "text", anchoring: "text-selector", extractText: "decode", authorable: false, uploadable: true, generatable: false },
-  "image/png": { extension: ".png", label: "PNG image", render: "image", anchoring: "spatial", extractText: "none", authorable: false, uploadable: true, generatable: false },
-  "image/jpeg": { extension: ".jpg", label: "JPEG image", render: "image", anchoring: "spatial", extractText: "none", authorable: false, uploadable: true, generatable: false },
-  "application/pdf": { extension: ".pdf", label: "PDF", render: "pdf", anchoring: "spatial", extractText: "pdf-text-layer", authorable: false, uploadable: true, generatable: true },
+  "text/markdown": { extension: ".md", label: "Markdown", render: "text", anchoring: "text-selector", textSource: "decode", authorable: true, uploadable: true, generatable: true },
+  "text/plain": { extension: ".txt", label: "Plain Text", render: "text", anchoring: "text-selector", textSource: "decode", authorable: true, uploadable: true, generatable: true },
+  "text/html": { extension: ".html", label: "HTML", render: "text", anchoring: "text-selector", textSource: "decode", authorable: true, uploadable: true, generatable: false },
+  "application/json": { extension: ".json", label: "JSON", render: "text", anchoring: "text-selector", textSource: "decode", authorable: false, uploadable: true, generatable: false },
+  "image/png": { extension: ".png", label: "PNG image", render: "image", anchoring: "spatial", textSource: "none", authorable: false, uploadable: true, generatable: false },
+  "image/jpeg": { extension: ".jpg", label: "JPEG image", render: "image", anchoring: "spatial", textSource: "none", authorable: false, uploadable: true, generatable: false },
+  "application/pdf": { extension: ".pdf", label: "PDF", render: "pdf", anchoring: "spatial", textSource: "pdf-text-layer", authorable: false, uploadable: true, generatable: true },
   // Storage tier — the big tent. Every row is a deliberate admission,
   // promotable by editing its row. Text-flavored rows embed (decode).
   // Text
@@ -1851,11 +1874,19 @@ function mediaTypeForExtension(ext) {
   const dotted = lower.startsWith(".") ? lower : `.${lower}`;
   return EXTENSION_TO_MEDIA_TYPE.get(EXTENSION_ALIASES[dotted] ?? dotted);
 }
-function textExtractionOf(format) {
+function textSourceOf(format) {
   const caps = capabilitiesOf(format);
-  if (caps) return caps.extractText;
+  if (caps) return caps.textSource;
   return baseMediaType(format).startsWith("text/") ? "decode" : "none";
 }
+var GEOMETRY_BY_STRATEGY = {
+  "decode": false,
+  "pdf-text-layer": true,
+  "none": false
+};
+function yieldsGeometryOf(format) {
+  return GEOMETRY_BY_STRATEGY[textSourceOf(format)];
+}
 function isAnnotatable(format) {
   const caps = capabilitiesOf(format);
   return caps !== void 0 && caps.anchoring !== "none";
@@ -1865,7 +1896,7 @@ var AUTHORABLE_MEDIA_TYPES = REGISTRY_KEYS.filter(
   (type) => MEDIA_TYPES[type].authorable
 );
 var EMBEDDABLE_MEDIA_TYPES = REGISTRY_KEYS.filter(
-  (type) => MEDIA_TYPES[type].extractText !== "none"
+  (type) => MEDIA_TYPES[type].textSource !== "none"
 );
 var GENERATABLE_MEDIA_TYPES = REGISTRY_KEYS.filter(
   (type) => MEDIA_TYPES[type].generatable
@@ -2140,6 +2171,9 @@ function deriveViews(graph, mainResourceId, focalAnnotationId) {
 }
 
 // src/retry.ts
+function equalJitter(cap) {
+  return cap / 2 + Math.random() * (cap / 2);
+}
 var STARTUP_FETCH_RETRY = {
   attempts: 8,
   initialDelayMs: 1e3,
@@ -2151,16 +2185,25 @@ function isTransientFetchError(error) {
   const code = error.cause?.code;
   return typeof code === "string" && code.length > 0;
 }
+var RETRYABLE_STATUSES = /* @__PURE__ */ new Set([429, 503, 504]);
+function isRetryableRequestError(error) {
+  if (isTransientFetchError(error)) return true;
+  if (typeof error !== "object" || error === null) return false;
+  if (error.name === "TimeoutError") return true;
+  const status = error.status;
+  return typeof status === "number" && RETRYABLE_STATUSES.has(status);
+}
 async function retryWithBackoff(fn, isRetryable, policy, onRetry) {
-  let delayMs = policy.initialDelayMs;
+  let cap = policy.initialDelayMs;
   for (let attempt = 1; ; attempt++) {
     try {
       return await fn();
     } catch (error) {
       if (attempt >= policy.attempts || !isRetryable(error)) throw error;
+      const delayMs = equalJitter(cap);
       onRetry?.({ attempt, attempts: policy.attempts, delayMs, error });
       await new Promise((resolve) => setTimeout(resolve, delayMs));
-      delayMs = Math.min(delayMs * 2, policy.maxDelayMs);
+      cap = Math.min(cap * 2, policy.maxDelayMs);
     }
   }
 }
@@ -2195,6 +2238,6 @@ function getShardPath(key, numBuckets = 65536) {
 // src/discovery.ts
 var DISCOVERY_URL_PATH = "/discovery/kbs.json";
 
-export { AUTHORABLE_MEDIA_TYPES, BusRequestError, CHANNEL_SCHEMAS, CONTEXT_FULL_WEIGHT, CONTEXT_PARTIAL_WEIGHT, ConfigurationError, ConflictError, DEFAULT_CHUNKING_CONFIG, DISCOVERY_URL_PATH, EMBEDDABLE_MEDIA_TYPES, GENERATABLE_MEDIA_TYPES, JWTTokenSchema, LOCALES, MEDIA_TYPES, NotFoundError, PERSISTED_EVENT_TYPES, POSITION_WEIGHT_MAX, POSITION_WINDOW, RESOURCE_BROADCAST_TYPES, STARTUP_FETCH_RETRY, ScriptError, SemiontError, UnauthorizedError, ValidationError, agentToDid, anchorAnnotation, anchorRuns, annotationId, applyBodyOperations, assembleAnnotation, baseMediaType, buildContentCache, burstBuffer, busRequest, capabilitiesOf, chunkText, cloneFormat, createCircleSvg, createFragmentSelector, createPolygonSvg, createRectangleSvg, decodeRepresentation, decodeWithCharset, deriveStorageUri, deriveViews, didToAgent, errField, estimateTokens, extensionForMediaType, extractBoundingBox, extractCharset, extractContext, findBestTextMatch, findBodyItem, findClaimSpan, folderOf, formatLocaleDisplay, generateUuid, getAllLocaleCodes, getAllPlatformTypes, getAnnotationExactText, getAnnotationUriFromEvent, getBodySource, getBodyType, getChecksum, getCommentText, getCreator, getDerivedFrom, getExactText, getFragmentSelector, getLanguage, getLocaleEnglishName, getLocaleInfo, getLocaleNativeName, getNodeEncoding, getPageFromFragment, getPrimaryMediaType, getPrimaryRepresentation, getPrimarySelector, getResourceEntityTypes, getResourceId, getShardPath, getStorageUri, getSvgSelector, getTargetSelector, getTargetSource, getTextPositionSelector, getTextQuoteSelector, hasTargetSelector, isAnnotatable, isAnnotationId, isArchived, isArray, isAssessment, isBodyResolved, isBoolean, isComment, isDefined, isDraft, isEventRelatedToAnnotation, isFunction, isGatheredContext, isGenerationJobParams, isHighlight, isNull, isNullish, isNumber, isObject, isReference, isResolvedReference, isResourceId, isStoredEvent, isString, isStubReference, isSupportedMediaType, isTag, isTextRun, isTransientFetchError, isUndefined, isValidEmail, isValidPlatformType, jumpConsistentHash, kbDid, locate, mediaTypeForExtension, normalizeCoordinates, normalizeText, parseEnvironment, parseFragmentSelector, parseSvgSelector, proposeStoragePath, reconcileSelector, resourceId, retryWithBackoff, scaleSvgToNative, serializePerKey, softwareToAgent, storageFileName, textExtractionOf, textUnder, userId, userToAgent, userToDid, uuidV4, validateData, validateEnvironment, validateSvgMarkup, verifyPosition };
+export { AUTHORABLE_MEDIA_TYPES, BusRequestError, CHANNEL_SCHEMAS, CONTEXT_FULL_WEIGHT, CONTEXT_PARTIAL_WEIGHT, ConfigurationError, ConflictError, DEFAULT_CHUNKING_CONFIG, DISCOVERY_URL_PATH, EMBEDDABLE_MEDIA_TYPES, GENERATABLE_MEDIA_TYPES, JWTTokenSchema, LOCALES, MEDIA_TYPES, NotFoundError, PERSISTED_EVENT_TYPES, POSITION_WEIGHT_MAX, POSITION_WINDOW, RESOURCE_BROADCAST_TYPES, STARTUP_FETCH_RETRY, ScriptError, SemiontError, UnauthorizedError, ValidationError, agentToDid, anchorAnnotation, anchorRuns, annotationId, applyBodyOperations, assembleAnnotation, baseMediaType, buildContentCache, burstBuffer, busRequest, capabilitiesOf, chunkText, cloneFormat, createCircleSvg, createFragmentSelector, createPolygonSvg, createRectangleSvg, decodeRepresentation, decodeWithCharset, deriveStorageUri, deriveViews, didToAgent, errField, estimateTokens, extensionForMediaType, extractBoundingBox, extractCharset, extractContext, findBestTextMatch, findBodyItem, findClaimSpan, folderOf, formatLocaleDisplay, generateUuid, getAllLocaleCodes, getAllPlatformTypes, getAnnotationExactText, getAnnotationUriFromEvent, getBodySource, getBodyType, getChecksum, getCommentText, getCreator, getDerivedFrom, getExactText, getFragmentSelector, getLanguage, getLocaleEnglishName, getLocaleInfo, getLocaleNativeName, getNodeEncoding, getPageFromFragment, getPrimaryMediaType, getPrimaryRepresentation, getPrimarySelector, getResourceEntityTypes, getResourceId, getShardPath, getStorageUri, getSvgSelector, getTargetSelector, getTargetSource, getTextPositionSelector, getTextQuoteSelector, hasTargetSelector, isAnnotatable, isAnnotationId, isArchived, isArray, isAssessment, isBodyResolved, isBoolean, isComment, isDefined, isDraft, isEventRelatedToAnnotation, isFunction, isGatheredContext, isGenerationJobParams, isHighlight, isNull, isNullish, isNumber, isObject, isReference, isResolvedReference, isResourceId, isRetryableRequestError, isStoredEvent, isString, isStubReference, isSupportedMediaType, isTag, isTextRun, isTransientFetchError, isUndefined, isValidEmail, isValidPlatformType, jumpConsistentHash, kbDid, locate, mediaTypeForExtension, normalizeCoordinates, normalizeText, parseEnvironment, parseFragmentSelector, parseSvgSelector, proposeStoragePath, reconcileSelector, replyChannelsFor, resourceId, retryWithBackoff, scaleSvgToNative, serializePerKey, softwareToAgent, storageFileName, textSourceOf, textUnder, userId, userToAgent, userToDid, uuidV4, validateData, validateEnvironment, validateSvgMarkup, verifyPosition, yieldsGeometryOf };
 //# sourceMappingURL=index.js.map
 //# sourceMappingURL=index.js.map