@semiont/core 0.5.24 → 0.5.25

package/dist/index.js

@@ -1,6 +1,6 @@
 export { createTomlConfigLoader, loadTomlConfig } from './chunk-XQTWEBJ5.js';
-import { BUS_OPERATIONS } from './chunk-4JTZMWZB.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-4JTZMWZB.js';
+import { BUS_OPERATIONS } from './chunk-FMFOBVTE.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-FMFOBVTE.js';
 import { Observable, merge, TimeoutError, throwError, firstValueFrom } from 'rxjs';
 import { filter, map, take, timeout, catchError, defaultIfEmpty } from 'rxjs/operators';
 
@@ -179,6 +179,10 @@ var CHANNEL_SCHEMAS = {
   "browse:resource-result": "BrowseResourceResult",
   "browse:resource-failed": null,
   // { correlationId } & CommandError
+  "browse:anchored-text-requested": "BrowseAnchoredTextRequest",
+  "browse:anchored-text-result": "BrowseAnchoredTextResult",
+  "browse:anchored-text-failed": null,
+  // { correlationId } & CommandError
   "browse:resources-requested": "BrowseResourcesRequest",
   "browse:resources-result": "BrowseResourcesResult",
   "browse:resources-failed": null,
@@ -271,12 +275,17 @@ var CHANNEL_SCHEMAS = {
   "weave:applied": null,
   // { resourceId; sequenceNumber }
   "smelt:settled": null,
-  // { resourceId; contentChecksum; outcome }
+  // { resourceId; contentChecksum; outcome; reason? }
   "weave:rebuild": "WeaveRebuildCommand",
   "weave:rebuild-ok": null,
   // { correlationId }
   "weave:rebuild-failed": null,
   // { correlationId; message }
+  "smelt:rebuild-anchors": "SmeltRebuildAnchorsCommand",
+  "smelt:rebuild-anchors-ok": null,
+  // { correlationId }
+  "smelt:rebuild-anchors-failed": null,
+  // { correlationId; message }
   // ── SSE infrastructure ──────────────────────────────────────────
   "stream-connected": null,
   // Record<string, never>
@@ -745,6 +754,136 @@ function getPageFromFragment(fragment) {
   return match ? parseInt(match[1], 10) : null;
 }
 
+// src/pdf-anchoring.ts
+function isTextRun(item) {
+  return typeof item === "object" && item !== null && "str" in item;
+}
+function anchorRuns(runs, page) {
+  const items = [];
+  let text = "";
+  for (const run of runs) {
+    if (run.str.trim()) {
+      const start = text.length;
+      text += run.str;
+      const end = text.length;
+      const [, , , , x, y] = run.transform;
+      items.push({ start, end, page, x, y, width: run.width, height: run.height });
+      text += run.hasEOL ? "\n" : " ";
+    } else if (run.hasEOL) {
+      text += "\n";
+    }
+  }
+  return { text, items };
+}
+var SAME_LINE_THRESHOLD_PT = 2;
+function locate(anchored, start, end) {
+  const overlap = anchored.items.filter(
+    (item) => item.start < end && item.end > start
+  );
+  if (overlap.length === 0) return { rects: [], overlap };
+  const pages = groupItemsByPage(overlap);
+  const rects = [];
+  for (const [page, pageItems] of pages) {
+    const lines = groupItemsByLine(pageItems, SAME_LINE_THRESHOLD_PT);
+    for (const lineItems of lines) {
+      const edges = lineItems.map((i) => {
+        const chars = i.end - i.start;
+        const left = i.start < start && chars > 0 ? i.x + i.width * ((start - i.start) / chars) : i.x;
+        const right2 = i.end > end && chars > 0 ? i.x + i.width * ((end - i.start) / chars) : i.x + i.width;
+        return { left, right: right2 };
+      });
+      const x = Math.min(...edges.map((e) => e.left));
+      const right = Math.max(...edges.map((e) => e.right));
+      const y = Math.min(...lineItems.map((i) => i.y));
+      const top = Math.max(...lineItems.map((i) => i.y + i.height));
+      rects.push({ page, x, y, width: right - x, height: top - y });
+    }
+  }
+  return { rects, overlap };
+}
+function textUnder(anchored, rect) {
+  const covered = anchored.items.filter((item) => item.page === rect.page && covers(item, rect)).sort((a, b) => a.start - b.start);
+  if (covered.length === 0) return "";
+  let quoted = slice(anchored, covered[0]);
+  for (let i = 1; i < covered.length; i++) {
+    const gap = anchored.text.slice(covered[i - 1].end, covered[i].start);
+    quoted += (gap.trim() === "" ? gap : " ") + slice(anchored, covered[i]);
+  }
+  return quoted.trim();
+}
+var slice = (anchored, item) => anchored.text.slice(item.start, item.end);
+var RUN_COVERAGE_THRESHOLD = 0.5;
+function covers(item, rect) {
+  const overlapX = Math.min(item.x + item.width, rect.x + rect.width) - Math.max(item.x, rect.x);
+  const overlapY = Math.min(item.y + item.height, rect.y + rect.height) - Math.max(item.y, rect.y);
+  if (overlapX <= 0 || overlapY <= 0) return false;
+  return overlapX * overlapY >= RUN_COVERAGE_THRESHOLD * item.width * item.height;
+}
+function groupItemsByPage(items) {
+  const map2 = /* @__PURE__ */ new Map();
+  for (const item of items) {
+    const existing = map2.get(item.page);
+    if (existing) {
+      existing.push(item);
+    } else {
+      map2.set(item.page, [item]);
+    }
+  }
+  return map2;
+}
+function groupItemsByLine(items, sameLineThreshold) {
+  const sorted = [...items].sort((a, b) => b.y - a.y || a.x - b.x);
+  const lines = [];
+  let currentLine = [];
+  for (const item of sorted) {
+    if (currentLine.length === 0 || Math.abs(item.y - currentLine[0].y) <= sameLineThreshold) {
+      currentLine.push(item);
+    } else {
+      lines.push(currentLine);
+      currentLine = [item];
+    }
+  }
+  if (currentLine.length > 0) lines.push(currentLine);
+  return lines;
+}
+
+// src/pdf-citation-search.ts
+var BREAK_MARKER = "
";
+var BREAK_GAP = `(?: ?${BREAK_MARKER} ?)?`;
+var escapeRegExp = (ch) => ch.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
+function findClaimSpan(anchored, exact) {
+  const needle = exact.replace(/\s+/g, " ").trim();
+  if (needle.length === 0) return null;
+  let norm = "";
+  const map2 = [];
+  let pendingWsAt = -1;
+  for (let i = 0; i < anchored.text.length; i++) {
+    const ch = anchored.text[i];
+    if (/\s/.test(ch)) {
+      if (norm.length > 0 && pendingWsAt < 0) pendingWsAt = i;
+      continue;
+    }
+    if (pendingWsAt >= 0) {
+      norm += " ";
+      map2.push(pendingWsAt);
+      pendingWsAt = -1;
+    }
+    norm += ch;
+    map2.push(i);
+  }
+  const idx = norm.indexOf(needle);
+  if (idx >= 0) {
+    return { start: map2[idx], end: map2[idx + needle.length - 1] + 1 };
+  }
+  const marker = anchored.text.replace(/\n/g, BREAK_MARKER);
+  const pattern = [...needle].map(escapeRegExp).join(BREAK_GAP);
+  const match = new RegExp(pattern).exec(marker);
+  if (match) {
+    return { start: match.index, end: match.index + match[0].length };
+  }
+  return null;
+}
+
 // src/resource-utils.ts
 function getResourceId(resource) {
   if (!resource) return void 0;
@@ -1579,7 +1718,8 @@ var storedBinary = (extension, label) => ({
   anchoring: "none",
   extractText: "none",
   authorable: false,
-  uploadable: true
+  uploadable: true,
+  generatable: false
 });
 var storedText = (extension, label) => ({
   ...storedBinary(extension, label),
@@ -1587,13 +1727,15 @@ var storedText = (extension, label) => ({
 });
 var MEDIA_TYPES = {
   // Full-capability tier
-  "text/markdown": { extension: ".md", label: "Markdown", render: "text", anchoring: "text-selector", extractText: "decode", authorable: true, uploadable: true },
-  "text/plain": { extension: ".txt", label: "Plain Text", render: "text", anchoring: "text-selector", extractText: "decode", authorable: true, uploadable: true },
-  "text/html": { extension: ".html", label: "HTML", render: "text", anchoring: "text-selector", extractText: "decode", authorable: true, uploadable: true },
-  "application/json": { extension: ".json", label: "JSON", render: "text", anchoring: "text-selector", extractText: "decode", authorable: false, uploadable: true },
-  "image/png": { extension: ".png", label: "PNG image", render: "image", anchoring: "spatial", extractText: "none", authorable: false, uploadable: true },
-  "image/jpeg": { extension: ".jpg", label: "JPEG image", render: "image", anchoring: "spatial", extractText: "none", authorable: false, uploadable: true },
-  "application/pdf": { extension: ".pdf", label: "PDF", render: "pdf", anchoring: "spatial", extractText: "pdf-text-layer", authorable: false, uploadable: true },
+  // `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 },
   // Storage tier — the big tent. Every row is a deliberate admission,
   // promotable by editing its row. Text-flavored rows embed (decode).
   // Text
@@ -1707,6 +1849,53 @@ var AUTHORABLE_MEDIA_TYPES = REGISTRY_KEYS.filter(
 var EMBEDDABLE_MEDIA_TYPES = REGISTRY_KEYS.filter(
   (type) => MEDIA_TYPES[type].extractText !== "none"
 );
+var GENERATABLE_MEDIA_TYPES = REGISTRY_KEYS.filter(
+  (type) => MEDIA_TYPES[type].generatable
+);
+
+// src/chunking.ts
+var DEFAULT_CHUNKING_CONFIG = {
+  chunkSize: 512,
+  overlap: 64
+};
+function estimateTokens(text) {
+  return Math.ceil(text.length / 4);
+}
+function chunkText(text, config = DEFAULT_CHUNKING_CONFIG) {
+  if (text.length === 0) return [];
+  const totalTokens = estimateTokens(text);
+  if (totalTokens <= config.chunkSize) {
+    return [text];
+  }
+  const chunkChars = config.chunkSize * 4;
+  const overlapChars = config.overlap * 4;
+  const chunks = [];
+  let start = 0;
+  while (start < text.length) {
+    let end = Math.min(start + chunkChars, text.length);
+    if (end < text.length) {
+      const paraBreak = text.lastIndexOf("\n\n", end);
+      if (paraBreak > start + chunkChars / 2) {
+        end = paraBreak;
+      } else {
+        const sentenceBreak = text.lastIndexOf(". ", end);
+        if (sentenceBreak > start + chunkChars / 2) {
+          end = sentenceBreak + 1;
+        } else {
+          const wordBreak = text.lastIndexOf(" ", end);
+          if (wordBreak > start + chunkChars / 2) {
+            end = wordBreak;
+          }
+        }
+      }
+    }
+    chunks.push(text.slice(start, end).trim());
+    const nextStart = end - overlapChars;
+    start = nextStart > start ? nextStart : end;
+    if (start >= text.length) break;
+  }
+  return chunks.filter((c) => c.length > 0);
+}
 
 // src/type-guards.ts
 function isString(value) {
@@ -1922,9 +2111,36 @@ async function retryWithBackoff(fn, isRetryable, policy, onRetry) {
   }
 }
 
+// src/shard-utils.ts
+function jumpConsistentHash(key, numBuckets = 65536) {
+  const hash = hashToUint32(key);
+  return hash % numBuckets;
+}
+function hashToUint32(str) {
+  let hash = 0;
+  for (let i = 0; i < str.length; i++) {
+    hash = (hash << 5) - hash + str.charCodeAt(i);
+    hash = hash & 4294967295;
+  }
+  return Math.abs(hash);
+}
+function shardIdToPath(shardId) {
+  if (shardId < 0 || shardId >= 65536) {
+    throw new Error(`Invalid shard ID: ${shardId}. Must be 0-65535 for 4-hex sharding.`);
+  }
+  const shardHex = shardId.toString(16).padStart(4, "0");
+  const ab = shardHex.substring(0, 2);
+  const cd = shardHex.substring(2, 4);
+  return [ab, cd];
+}
+function getShardPath(key, numBuckets = 65536) {
+  const shardId = jumpConsistentHash(key, numBuckets);
+  return shardIdToPath(shardId);
+}
+
 // 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, DISCOVERY_URL_PATH, EMBEDDABLE_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, annotationId, applyBodyOperations, assembleAnnotation, baseMediaType, buildContentCache, burstBuffer, busRequest, capabilitiesOf, createCircleSvg, createFragmentSelector, createPolygonSvg, createRectangleSvg, decodeRepresentation, decodeWithCharset, deriveViews, didToAgent, errField, extensionForMediaType, extractBoundingBox, extractCharset, extractContext, findBestTextMatch, findBodyItem, 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, getStorageUri, getSvgSelector, getTargetSelector, getTargetSource, getTextPositionSelector, getTextQuoteSelector, hasTargetSelector, isAnnotationId, isArchived, isArray, isAssessment, isBodyResolved, isBoolean, isComment, isDefined, isDraft, isEventRelatedToAnnotation, isFunction, isHighlight, isNull, isNullish, isNumber, isObject, isReference, isResolvedReference, isResourceId, isStoredEvent, isString, isStubReference, isSupportedMediaType, isTag, isTransientFetchError, isUndefined, isValidEmail, isValidPlatformType, kbDid, mediaTypeForExtension, normalizeCoordinates, normalizeText, parseEnvironment, parseFragmentSelector, parseSvgSelector, reconcileSelector, resourceId, retryWithBackoff, scaleSvgToNative, serializePerKey, softwareToAgent, textExtractionOf, 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, createCircleSvg, createFragmentSelector, createPolygonSvg, createRectangleSvg, decodeRepresentation, decodeWithCharset, deriveViews, didToAgent, errField, estimateTokens, extensionForMediaType, extractBoundingBox, extractCharset, extractContext, findBestTextMatch, findBodyItem, findClaimSpan, 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, isAnnotationId, isArchived, isArray, isAssessment, isBodyResolved, isBoolean, isComment, isDefined, isDraft, isEventRelatedToAnnotation, isFunction, 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, reconcileSelector, resourceId, retryWithBackoff, scaleSvgToNative, serializePerKey, softwareToAgent, textExtractionOf, textUnder, userId, userToAgent, userToDid, uuidV4, validateData, validateEnvironment, validateSvgMarkup, verifyPosition };
 //# sourceMappingURL=index.js.map
 //# sourceMappingURL=index.js.map