@absolutejs/absolute 0.19.0-beta.619 → 0.19.0-beta.620

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.
@@ -1785,6 +1785,12 @@ var formatMediaTimestamp = (value) => {
1785
1785
  const milliseconds = Math.floor(value % 1000);
1786
1786
  return `${String(minutes).padStart(2, "0")}:${String(seconds).padStart(2, "0")}.${String(milliseconds).padStart(3, "0")}`;
1787
1787
  };
1788
+ var formatMediaDurationLabel = (value) => {
1789
+ if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
1790
+ return;
1791
+ }
1792
+ return formatMediaTimestamp(value);
1793
+ };
1788
1794
  var buildLocatorLabel = (metadata, source, title) => {
1789
1795
  if (!metadata) {
1790
1796
  return;
@@ -1847,6 +1853,12 @@ var buildProvenanceLabel = (metadata) => {
1847
1853
  const sentAt = formatTimestampLabel(metadata.sentAt) ?? formatTimestampLabel(metadata.receivedAt);
1848
1854
  const speaker = getContextString(metadata.speaker);
1849
1855
  const mediaKind = getContextString(metadata.mediaKind);
1856
+ const mediaSegmentCount = getContextNumber(metadata.mediaSegmentCount);
1857
+ const mediaSegmentGroupSize = getContextNumber(metadata.mediaSegmentGroupSize);
1858
+ const mediaSegmentGroupIndex = getContextNumber(metadata.mediaSegmentGroupIndex);
1859
+ const mediaChannel = getContextString(metadata.mediaChannel);
1860
+ const mediaSpeakerCount = getContextNumber(metadata.mediaSpeakerCount);
1861
+ const mediaDurationLabel = formatMediaDurationLabel(metadata.mediaDurationMs);
1850
1862
  const transcriptSource = getContextString(metadata.transcriptSource);
1851
1863
  const pdfTextMode = getContextString(metadata.pdfTextMode);
1852
1864
  const ocrEngine = getContextString(metadata.ocrEngine);
@@ -1856,6 +1868,12 @@ var buildProvenanceLabel = (metadata) => {
1856
1868
  ocrEngine ? `OCR ${ocrEngine}` : "",
1857
1869
  typeof ocrConfidence === "number" ? `Confidence ${ocrConfidence.toFixed(2)}` : "",
1858
1870
  mediaKind ? `Media ${mediaKind}` : "",
1871
+ mediaSegmentCount ? `${mediaSegmentCount} segments` : "",
1872
+ mediaSegmentGroupSize ? `${mediaSegmentGroupSize} grouped segments` : "",
1873
+ mediaSegmentGroupIndex !== undefined ? `Segment group ${mediaSegmentGroupIndex + 1}` : "",
1874
+ mediaChannel ? `Channel ${mediaChannel}` : "",
1875
+ mediaSpeakerCount ? `${mediaSpeakerCount} speakers` : "",
1876
+ mediaDurationLabel ? `Duration ${mediaDurationLabel}` : "",
1859
1877
  transcriptSource ? `Transcript ${transcriptSource}` : "",
1860
1878
  threadTopic ? `Thread ${threadTopic}` : "",
1861
1879
  speaker ? `Speaker ${speaker}` : "",
@@ -2211,6 +2229,34 @@ var getAttachmentName2 = (source, title) => {
2211
2229
  }
2212
2230
  return;
2213
2231
  };
2232
+ var getSpreadsheetHeaders = (metadata) => Array.isArray(metadata?.spreadsheetHeaders) ? metadata.spreadsheetHeaders.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : [];
2233
+ var formatSpreadsheetRowRange = (rowStart, rowEnd) => {
2234
+ if (typeof rowStart !== "number" || !Number.isFinite(rowStart)) {
2235
+ return;
2236
+ }
2237
+ if (typeof rowEnd !== "number" && typeof rowStart === "number" && Number.isFinite(rowStart)) {
2238
+ return `Rows ${rowStart}`;
2239
+ }
2240
+ if (rowStart === rowEnd) {
2241
+ return `Rows ${rowStart}`;
2242
+ }
2243
+ return `Rows ${rowStart}-${rowEnd}`;
2244
+ };
2245
+ var formatSpreadsheetTableLabel = (tableIndex, tableCount) => {
2246
+ if (typeof tableIndex !== "number" || !Number.isFinite(tableIndex) || tableIndex < 1) {
2247
+ return;
2248
+ }
2249
+ if (typeof tableCount === "number" && Number.isFinite(tableCount) && tableCount >= tableIndex) {
2250
+ return `Table ${tableIndex} of ${tableCount}`;
2251
+ }
2252
+ return `Table ${tableIndex}`;
2253
+ };
2254
+ var formatMediaDurationLabel2 = (value) => {
2255
+ if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
2256
+ return;
2257
+ }
2258
+ return formatMediaTimestamp2(value);
2259
+ };
2214
2260
  var buildContextLabel2 = (metadata) => {
2215
2261
  if (!metadata) {
2216
2262
  return;
@@ -2236,29 +2282,61 @@ var buildContextLabel2 = (metadata) => {
2236
2282
  }
2237
2283
  const emailKind = getContextString2(metadata.emailKind);
2238
2284
  if (emailKind === "attachment") {
2239
- return "Attachment evidence";
2285
+ const attachmentName = getContextString2(metadata.attachmentName);
2286
+ const threadTopic2 = getContextString2(metadata.threadTopic);
2287
+ return attachmentName ? threadTopic2 ? `Attachment evidence ${attachmentName} in ${threadTopic2}` : `Attachment evidence ${attachmentName}` : "Attachment evidence";
2240
2288
  }
2241
2289
  if (emailKind === "message") {
2290
+ const threadTopic2 = getContextString2(metadata.threadTopic);
2242
2291
  const from = getContextString2(metadata.from);
2292
+ if (threadTopic2) {
2293
+ return from ? `Message in ${threadTopic2} from ${from}` : `Message in ${threadTopic2}`;
2294
+ }
2243
2295
  return from ? `Message from ${from}` : "Message evidence";
2244
2296
  }
2245
2297
  const page = getContextNumber2(metadata.page) ?? getContextNumber2(metadata.pageNumber) ?? (typeof metadata.pageIndex === "number" ? metadata.pageIndex + 1 : undefined);
2246
2298
  const region = getContextNumber2(metadata.regionNumber) ?? (typeof metadata.regionIndex === "number" ? metadata.regionIndex + 1 : undefined);
2299
+ const hasOCRTrace = typeof metadata.ocrRegionConfidence === "number" || typeof metadata.ocrConfidence === "number" || getContextString2(metadata.pdfTextMode) === "ocr" || typeof metadata.ocrRegionCount === "number";
2247
2300
  if (page && region) {
2301
+ if (hasOCRTrace) {
2302
+ return `OCR page ${page} region ${region}`;
2303
+ }
2248
2304
  return `Page ${page} region ${region}`;
2249
2305
  }
2250
2306
  if (page) {
2307
+ if (hasOCRTrace) {
2308
+ return `OCR page ${page}`;
2309
+ }
2251
2310
  return `Page ${page}`;
2252
2311
  }
2253
2312
  const sheet = getContextString2(metadata.sheetName) ?? (Array.isArray(metadata.sheetNames) ? getContextString2(metadata.sheetNames[0]) : undefined);
2254
2313
  if (sheet) {
2314
+ const tableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
2315
+ const rowRange = formatSpreadsheetRowRange(getContextNumber2(metadata.spreadsheetRowStart), getContextNumber2(metadata.spreadsheetRowEnd));
2316
+ const headers = getSpreadsheetHeaders(metadata);
2317
+ if (tableLabel && rowRange) {
2318
+ return `Sheet ${sheet} ${tableLabel} ${rowRange}`;
2319
+ }
2320
+ if (tableLabel) {
2321
+ return `Sheet ${sheet} ${tableLabel}`;
2322
+ }
2323
+ if (rowRange) {
2324
+ return `Sheet ${sheet} ${rowRange}`;
2325
+ }
2326
+ if (headers.length > 0) {
2327
+ return `Sheet ${sheet} by ${headers.slice(0, 2).join(", ")}`;
2328
+ }
2255
2329
  return `Sheet ${sheet}`;
2256
2330
  }
2257
2331
  const slide = getContextNumber2(metadata.slide) ?? getContextNumber2(metadata.slideNumber) ?? (typeof metadata.slideIndex === "number" ? metadata.slideIndex + 1 : undefined);
2332
+ const slideTitle = getContextString2(metadata.slideTitle);
2258
2333
  if (slide) {
2334
+ if (slideTitle) {
2335
+ return `Slide ${slide} ${slideTitle}`;
2336
+ }
2259
2337
  return `Slide ${slide}`;
2260
2338
  }
2261
- const archiveEntry = getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
2339
+ const archiveEntry = getContextString2(metadata.archiveFullPath) ?? getContextString2(metadata.archivePath) ?? getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
2262
2340
  if (archiveEntry) {
2263
2341
  return `Archive entry ${archiveEntry}`;
2264
2342
  }
@@ -2283,6 +2361,9 @@ var buildLocatorLabel2 = (metadata, source, title) => {
2283
2361
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
2284
2362
  const pdfBlockNumber = getContextNumber2(metadata.pdfBlockNumber);
2285
2363
  const officeBlockNumber = getContextNumber2(metadata.officeBlockNumber);
2364
+ const spreadsheetRowStart = getContextNumber2(metadata.spreadsheetRowStart);
2365
+ const spreadsheetRowEnd = getContextNumber2(metadata.spreadsheetRowEnd);
2366
+ const slideTitle = getContextString2(metadata.slideTitle);
2286
2367
  const page = getContextNumber2(metadata.page) ?? getContextNumber2(metadata.pageNumber) ?? (typeof metadata.pageIndex === "number" ? metadata.pageIndex + 1 : undefined);
2287
2368
  const region = getContextNumber2(metadata.regionNumber) ?? (typeof metadata.regionIndex === "number" ? metadata.regionIndex + 1 : undefined);
2288
2369
  if (page && region) {
@@ -2299,19 +2380,31 @@ var buildLocatorLabel2 = (metadata, source, title) => {
2299
2380
  }
2300
2381
  const sheet = getContextString2(metadata.sheetName) ?? (Array.isArray(metadata.sheetNames) ? getContextString2(metadata.sheetNames[0]) : undefined);
2301
2382
  if (sheet) {
2302
- return `Sheet ${sheet}`;
2383
+ const tableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
2384
+ const rowRange = formatSpreadsheetRowRange(spreadsheetRowStart, spreadsheetRowEnd);
2385
+ if (tableLabel && rowRange) {
2386
+ return `Sheet ${sheet} · ${tableLabel} · ${rowRange}`;
2387
+ }
2388
+ if (tableLabel) {
2389
+ return `Sheet ${sheet} · ${tableLabel}`;
2390
+ }
2391
+ return rowRange ? `Sheet ${sheet} · ${rowRange}` : `Sheet ${sheet}`;
2303
2392
  }
2304
2393
  const slide = getContextNumber2(metadata.slide) ?? getContextNumber2(metadata.slideNumber) ?? (typeof metadata.slideIndex === "number" ? metadata.slideIndex + 1 : undefined);
2305
2394
  if (slide) {
2306
- return `Slide ${slide}`;
2395
+ return slideTitle ? `Slide ${slide} · ${slideTitle}` : `Slide ${slide}`;
2307
2396
  }
2308
- const archiveEntry = getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
2397
+ const archiveEntry = getContextString2(metadata.archiveFullPath) ?? getContextString2(metadata.archivePath) ?? getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
2309
2398
  if (archiveEntry) {
2310
2399
  return `Archive entry ${archiveEntry}`;
2311
2400
  }
2312
2401
  const emailKind = getContextString2(metadata.emailKind);
2313
2402
  if (emailKind === "attachment") {
2314
2403
  const attachmentName = getContextString2(metadata.attachmentName) ?? getAttachmentName2(source, title);
2404
+ const replyDepth = getContextNumber2(metadata.replyDepth);
2405
+ if (attachmentName && replyDepth && replyDepth > 0) {
2406
+ return `Attachment ${attachmentName} · Reply depth ${replyDepth}`;
2407
+ }
2315
2408
  return attachmentName ? `Attachment ${attachmentName}` : "Attachment";
2316
2409
  }
2317
2410
  const mediaStart = formatMediaTimestamp2(metadata.startMs);
@@ -2342,18 +2435,36 @@ var buildProvenanceLabel2 = (metadata) => {
2342
2435
  return;
2343
2436
  }
2344
2437
  const threadTopic = getContextString2(metadata.threadTopic);
2438
+ const replyDepth = getContextNumber2(metadata.replyDepth);
2439
+ const threadMessageCount = getContextNumber2(metadata.threadMessageCount);
2440
+ const threadRootMessageId = getContextString2(metadata.threadRootMessageId);
2345
2441
  const from = getContextString2(metadata.from);
2346
2442
  const sentAt = formatTimestampLabel2(metadata.sentAt) ?? formatTimestampLabel2(metadata.receivedAt);
2347
2443
  const speaker = getContextString2(metadata.speaker);
2348
2444
  const mediaKind = getContextString2(metadata.mediaKind);
2349
2445
  const transcriptSource = getContextString2(metadata.transcriptSource);
2446
+ const mediaSpeakerCount = getContextNumber2(metadata.mediaSpeakerCount);
2447
+ const mediaSegmentCount = getContextNumber2(metadata.mediaSegmentCount);
2448
+ const mediaSegmentGroupSize = getContextNumber2(metadata.mediaSegmentGroupSize);
2449
+ const mediaSegmentGroupIndex = getContextNumber2(metadata.mediaSegmentGroupIndex);
2450
+ const mediaChannel = getContextString2(metadata.mediaChannel);
2451
+ const mediaDurationLabel = formatMediaDurationLabel2(metadata.mediaDurationMs);
2452
+ const spreadsheetHeaders = getSpreadsheetHeaders(metadata);
2453
+ const slideNotesText = getContextString2(metadata.slideNotesText);
2350
2454
  const pdfTextMode = getContextString2(metadata.pdfTextMode);
2351
2455
  const pdfTextKind = getContextString2(metadata.pdfTextKind);
2352
2456
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
2353
2457
  const ocrEngine = getContextString2(metadata.ocrEngine);
2354
2458
  const extractorRegistryMatch = getContextString2(metadata.extractorRegistryMatch);
2355
2459
  const chunkingProfile = getContextString2(metadata.chunkingProfile);
2460
+ const archiveDepth = getContextNumber2(metadata.archiveDepth);
2461
+ const archiveNestedDepth = getContextNumber2(metadata.archiveNestedDepth);
2462
+ const archiveContainerPath = getContextString2(metadata.archiveContainerPath);
2463
+ const archiveRootName = getContextString2(metadata.archiveRootName);
2464
+ const spreadsheetTableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
2356
2465
  const ocrConfidence = getContextNumber2(metadata.ocrRegionConfidence) ?? getContextNumber2(metadata.ocrConfidence);
2466
+ const ocrAverageConfidence = getContextNumber2(metadata.ocrPageAverageConfidence) ?? getContextNumber2(metadata.ocrAverageConfidence);
2467
+ const ocrRegionCount = getContextNumber2(metadata.ocrRegionCount);
2357
2468
  const labels = [
2358
2469
  pdfTextMode ? `PDF ${pdfTextMode}` : "",
2359
2470
  pdfTextKind === "table_like" ? "PDF table block" : pdfTextKind === "paragraph" ? "PDF text block" : "",
@@ -2362,9 +2473,27 @@ var buildProvenanceLabel2 = (metadata) => {
2362
2473
  extractorRegistryMatch ? `Extractor ${extractorRegistryMatch}` : "",
2363
2474
  chunkingProfile ? `Chunking ${chunkingProfile}` : "",
2364
2475
  typeof ocrConfidence === "number" ? `Confidence ${ocrConfidence.toFixed(2)}` : "",
2476
+ typeof ocrAverageConfidence === "number" && ocrAverageConfidence !== ocrConfidence ? `Average ${ocrAverageConfidence.toFixed(2)}` : "",
2477
+ typeof ocrRegionCount === "number" ? `${ocrRegionCount} regions` : "",
2478
+ spreadsheetHeaders.length > 0 ? `Spreadsheet ${spreadsheetHeaders.join(", ")}` : "",
2479
+ spreadsheetTableLabel ? `Spreadsheet ${spreadsheetTableLabel}` : "",
2365
2480
  mediaKind ? `Media ${mediaKind}` : "",
2481
+ mediaSegmentCount ? `${mediaSegmentCount} segments` : "",
2482
+ mediaSegmentGroupSize ? `${mediaSegmentGroupSize} grouped segments` : "",
2483
+ mediaSegmentGroupIndex !== undefined ? `Segment group ${mediaSegmentGroupIndex + 1}` : "",
2484
+ mediaChannel ? `Channel ${mediaChannel}` : "",
2485
+ mediaSpeakerCount ? `${mediaSpeakerCount} speakers` : "",
2486
+ mediaDurationLabel ? `Duration ${mediaDurationLabel}` : "",
2366
2487
  transcriptSource ? `Transcript ${transcriptSource}` : "",
2367
2488
  threadTopic ? `Thread ${threadTopic}` : "",
2489
+ threadRootMessageId ? `Thread root ${threadRootMessageId}` : "",
2490
+ threadMessageCount ? `${threadMessageCount} thread messages` : "",
2491
+ replyDepth ? `Reply depth ${replyDepth}` : "",
2492
+ slideNotesText ? "Speaker notes" : "",
2493
+ archiveDepth ? `Archive depth ${archiveDepth}` : "",
2494
+ archiveNestedDepth ? `Archive nested depth ${archiveNestedDepth}` : "",
2495
+ archiveContainerPath ? `Archive container ${archiveContainerPath}` : "",
2496
+ archiveRootName ? `Archive root ${archiveRootName}` : "",
2368
2497
  speaker ? `Speaker ${speaker}` : "",
2369
2498
  from ? `Sender ${from}` : "",
2370
2499
  sentAt ? `Sent ${sentAt}` : ""
@@ -2741,7 +2870,7 @@ var getSectionPathFromSource = (source) => {
2741
2870
  const path = source.structure?.section?.path ?? (Array.isArray(source.metadata?.sectionPath) ? source.metadata.sectionPath.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : []);
2742
2871
  return path.length > 0 ? path : undefined;
2743
2872
  };
2744
- var isBlockAwareContextLabel = (value) => typeof value === "string" && (value.startsWith("PDF ") || value.startsWith("Office "));
2873
+ var isBlockAwareContextLabel = (value) => typeof value === "string" && (value.startsWith("PDF ") || value.startsWith("Office ") || value.startsWith("Slide "));
2745
2874
  var getStructuredSectionScoreWeight = (metadata) => {
2746
2875
  if (!metadata) {
2747
2876
  return 1;
@@ -2749,6 +2878,8 @@ var getStructuredSectionScoreWeight = (metadata) => {
2749
2878
  const pdfTextKind = getContextString2(metadata.pdfTextKind);
2750
2879
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
2751
2880
  const sectionKind = getContextString2(metadata.sectionKind);
2881
+ const slideTitle = getContextString2(metadata.slideTitle);
2882
+ const slideNotesText = getContextString2(metadata.slideNotesText);
2752
2883
  if (pdfTextKind === "table_like") {
2753
2884
  return 1.28;
2754
2885
  }
@@ -2758,6 +2889,12 @@ var getStructuredSectionScoreWeight = (metadata) => {
2758
2889
  if (sectionKind === "pdf_block" || sectionKind === "office_block" || officeBlockKind === "paragraph" || pdfTextKind === "paragraph") {
2759
2890
  return 1.12;
2760
2891
  }
2892
+ if (sectionKind === "presentation_slide" && slideNotesText) {
2893
+ return 1.2;
2894
+ }
2895
+ if (sectionKind === "presentation_slide" && slideTitle) {
2896
+ return 1.14;
2897
+ }
2761
2898
  return 1;
2762
2899
  };
2763
2900
  var getStructuredSourceLeadScore = (source) => source.score * getStructuredSectionScoreWeight(source.metadata);
@@ -927,6 +927,12 @@ var formatMediaTimestamp = (value) => {
927
927
  const milliseconds = Math.floor(value % 1000);
928
928
  return `${String(minutes).padStart(2, "0")}:${String(seconds).padStart(2, "0")}.${String(milliseconds).padStart(3, "0")}`;
929
929
  };
930
+ var formatMediaDurationLabel = (value) => {
931
+ if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
932
+ return;
933
+ }
934
+ return formatMediaTimestamp(value);
935
+ };
930
936
  var buildLocatorLabel = (metadata, source, title) => {
931
937
  if (!metadata) {
932
938
  return;
@@ -989,6 +995,12 @@ var buildProvenanceLabel = (metadata) => {
989
995
  const sentAt = formatTimestampLabel(metadata.sentAt) ?? formatTimestampLabel(metadata.receivedAt);
990
996
  const speaker = getContextString(metadata.speaker);
991
997
  const mediaKind = getContextString(metadata.mediaKind);
998
+ const mediaSegmentCount = getContextNumber(metadata.mediaSegmentCount);
999
+ const mediaSegmentGroupSize = getContextNumber(metadata.mediaSegmentGroupSize);
1000
+ const mediaSegmentGroupIndex = getContextNumber(metadata.mediaSegmentGroupIndex);
1001
+ const mediaChannel = getContextString(metadata.mediaChannel);
1002
+ const mediaSpeakerCount = getContextNumber(metadata.mediaSpeakerCount);
1003
+ const mediaDurationLabel = formatMediaDurationLabel(metadata.mediaDurationMs);
992
1004
  const transcriptSource = getContextString(metadata.transcriptSource);
993
1005
  const pdfTextMode = getContextString(metadata.pdfTextMode);
994
1006
  const ocrEngine = getContextString(metadata.ocrEngine);
@@ -998,6 +1010,12 @@ var buildProvenanceLabel = (metadata) => {
998
1010
  ocrEngine ? `OCR ${ocrEngine}` : "",
999
1011
  typeof ocrConfidence === "number" ? `Confidence ${ocrConfidence.toFixed(2)}` : "",
1000
1012
  mediaKind ? `Media ${mediaKind}` : "",
1013
+ mediaSegmentCount ? `${mediaSegmentCount} segments` : "",
1014
+ mediaSegmentGroupSize ? `${mediaSegmentGroupSize} grouped segments` : "",
1015
+ mediaSegmentGroupIndex !== undefined ? `Segment group ${mediaSegmentGroupIndex + 1}` : "",
1016
+ mediaChannel ? `Channel ${mediaChannel}` : "",
1017
+ mediaSpeakerCount ? `${mediaSpeakerCount} speakers` : "",
1018
+ mediaDurationLabel ? `Duration ${mediaDurationLabel}` : "",
1001
1019
  transcriptSource ? `Transcript ${transcriptSource}` : "",
1002
1020
  threadTopic ? `Thread ${threadTopic}` : "",
1003
1021
  speaker ? `Speaker ${speaker}` : "",
@@ -1783,6 +1801,34 @@ var getAttachmentName2 = (source, title) => {
1783
1801
  }
1784
1802
  return;
1785
1803
  };
1804
+ var getSpreadsheetHeaders = (metadata) => Array.isArray(metadata?.spreadsheetHeaders) ? metadata.spreadsheetHeaders.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : [];
1805
+ var formatSpreadsheetRowRange = (rowStart, rowEnd) => {
1806
+ if (typeof rowStart !== "number" || !Number.isFinite(rowStart)) {
1807
+ return;
1808
+ }
1809
+ if (typeof rowEnd !== "number" && typeof rowStart === "number" && Number.isFinite(rowStart)) {
1810
+ return `Rows ${rowStart}`;
1811
+ }
1812
+ if (rowStart === rowEnd) {
1813
+ return `Rows ${rowStart}`;
1814
+ }
1815
+ return `Rows ${rowStart}-${rowEnd}`;
1816
+ };
1817
+ var formatSpreadsheetTableLabel = (tableIndex, tableCount) => {
1818
+ if (typeof tableIndex !== "number" || !Number.isFinite(tableIndex) || tableIndex < 1) {
1819
+ return;
1820
+ }
1821
+ if (typeof tableCount === "number" && Number.isFinite(tableCount) && tableCount >= tableIndex) {
1822
+ return `Table ${tableIndex} of ${tableCount}`;
1823
+ }
1824
+ return `Table ${tableIndex}`;
1825
+ };
1826
+ var formatMediaDurationLabel2 = (value) => {
1827
+ if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
1828
+ return;
1829
+ }
1830
+ return formatMediaTimestamp2(value);
1831
+ };
1786
1832
  var buildContextLabel2 = (metadata) => {
1787
1833
  if (!metadata) {
1788
1834
  return;
@@ -1808,29 +1854,61 @@ var buildContextLabel2 = (metadata) => {
1808
1854
  }
1809
1855
  const emailKind = getContextString2(metadata.emailKind);
1810
1856
  if (emailKind === "attachment") {
1811
- return "Attachment evidence";
1857
+ const attachmentName = getContextString2(metadata.attachmentName);
1858
+ const threadTopic2 = getContextString2(metadata.threadTopic);
1859
+ return attachmentName ? threadTopic2 ? `Attachment evidence ${attachmentName} in ${threadTopic2}` : `Attachment evidence ${attachmentName}` : "Attachment evidence";
1812
1860
  }
1813
1861
  if (emailKind === "message") {
1862
+ const threadTopic2 = getContextString2(metadata.threadTopic);
1814
1863
  const from = getContextString2(metadata.from);
1864
+ if (threadTopic2) {
1865
+ return from ? `Message in ${threadTopic2} from ${from}` : `Message in ${threadTopic2}`;
1866
+ }
1815
1867
  return from ? `Message from ${from}` : "Message evidence";
1816
1868
  }
1817
1869
  const page = getContextNumber2(metadata.page) ?? getContextNumber2(metadata.pageNumber) ?? (typeof metadata.pageIndex === "number" ? metadata.pageIndex + 1 : undefined);
1818
1870
  const region = getContextNumber2(metadata.regionNumber) ?? (typeof metadata.regionIndex === "number" ? metadata.regionIndex + 1 : undefined);
1871
+ const hasOCRTrace = typeof metadata.ocrRegionConfidence === "number" || typeof metadata.ocrConfidence === "number" || getContextString2(metadata.pdfTextMode) === "ocr" || typeof metadata.ocrRegionCount === "number";
1819
1872
  if (page && region) {
1873
+ if (hasOCRTrace) {
1874
+ return `OCR page ${page} region ${region}`;
1875
+ }
1820
1876
  return `Page ${page} region ${region}`;
1821
1877
  }
1822
1878
  if (page) {
1879
+ if (hasOCRTrace) {
1880
+ return `OCR page ${page}`;
1881
+ }
1823
1882
  return `Page ${page}`;
1824
1883
  }
1825
1884
  const sheet = getContextString2(metadata.sheetName) ?? (Array.isArray(metadata.sheetNames) ? getContextString2(metadata.sheetNames[0]) : undefined);
1826
1885
  if (sheet) {
1886
+ const tableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
1887
+ const rowRange = formatSpreadsheetRowRange(getContextNumber2(metadata.spreadsheetRowStart), getContextNumber2(metadata.spreadsheetRowEnd));
1888
+ const headers = getSpreadsheetHeaders(metadata);
1889
+ if (tableLabel && rowRange) {
1890
+ return `Sheet ${sheet} ${tableLabel} ${rowRange}`;
1891
+ }
1892
+ if (tableLabel) {
1893
+ return `Sheet ${sheet} ${tableLabel}`;
1894
+ }
1895
+ if (rowRange) {
1896
+ return `Sheet ${sheet} ${rowRange}`;
1897
+ }
1898
+ if (headers.length > 0) {
1899
+ return `Sheet ${sheet} by ${headers.slice(0, 2).join(", ")}`;
1900
+ }
1827
1901
  return `Sheet ${sheet}`;
1828
1902
  }
1829
1903
  const slide = getContextNumber2(metadata.slide) ?? getContextNumber2(metadata.slideNumber) ?? (typeof metadata.slideIndex === "number" ? metadata.slideIndex + 1 : undefined);
1904
+ const slideTitle = getContextString2(metadata.slideTitle);
1830
1905
  if (slide) {
1906
+ if (slideTitle) {
1907
+ return `Slide ${slide} ${slideTitle}`;
1908
+ }
1831
1909
  return `Slide ${slide}`;
1832
1910
  }
1833
- const archiveEntry = getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
1911
+ const archiveEntry = getContextString2(metadata.archiveFullPath) ?? getContextString2(metadata.archivePath) ?? getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
1834
1912
  if (archiveEntry) {
1835
1913
  return `Archive entry ${archiveEntry}`;
1836
1914
  }
@@ -1855,6 +1933,9 @@ var buildLocatorLabel2 = (metadata, source, title) => {
1855
1933
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
1856
1934
  const pdfBlockNumber = getContextNumber2(metadata.pdfBlockNumber);
1857
1935
  const officeBlockNumber = getContextNumber2(metadata.officeBlockNumber);
1936
+ const spreadsheetRowStart = getContextNumber2(metadata.spreadsheetRowStart);
1937
+ const spreadsheetRowEnd = getContextNumber2(metadata.spreadsheetRowEnd);
1938
+ const slideTitle = getContextString2(metadata.slideTitle);
1858
1939
  const page = getContextNumber2(metadata.page) ?? getContextNumber2(metadata.pageNumber) ?? (typeof metadata.pageIndex === "number" ? metadata.pageIndex + 1 : undefined);
1859
1940
  const region = getContextNumber2(metadata.regionNumber) ?? (typeof metadata.regionIndex === "number" ? metadata.regionIndex + 1 : undefined);
1860
1941
  if (page && region) {
@@ -1871,19 +1952,31 @@ var buildLocatorLabel2 = (metadata, source, title) => {
1871
1952
  }
1872
1953
  const sheet = getContextString2(metadata.sheetName) ?? (Array.isArray(metadata.sheetNames) ? getContextString2(metadata.sheetNames[0]) : undefined);
1873
1954
  if (sheet) {
1874
- return `Sheet ${sheet}`;
1955
+ const tableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
1956
+ const rowRange = formatSpreadsheetRowRange(spreadsheetRowStart, spreadsheetRowEnd);
1957
+ if (tableLabel && rowRange) {
1958
+ return `Sheet ${sheet} \xB7 ${tableLabel} \xB7 ${rowRange}`;
1959
+ }
1960
+ if (tableLabel) {
1961
+ return `Sheet ${sheet} \xB7 ${tableLabel}`;
1962
+ }
1963
+ return rowRange ? `Sheet ${sheet} \xB7 ${rowRange}` : `Sheet ${sheet}`;
1875
1964
  }
1876
1965
  const slide = getContextNumber2(metadata.slide) ?? getContextNumber2(metadata.slideNumber) ?? (typeof metadata.slideIndex === "number" ? metadata.slideIndex + 1 : undefined);
1877
1966
  if (slide) {
1878
- return `Slide ${slide}`;
1967
+ return slideTitle ? `Slide ${slide} \xB7 ${slideTitle}` : `Slide ${slide}`;
1879
1968
  }
1880
- const archiveEntry = getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
1969
+ const archiveEntry = getContextString2(metadata.archiveFullPath) ?? getContextString2(metadata.archivePath) ?? getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
1881
1970
  if (archiveEntry) {
1882
1971
  return `Archive entry ${archiveEntry}`;
1883
1972
  }
1884
1973
  const emailKind = getContextString2(metadata.emailKind);
1885
1974
  if (emailKind === "attachment") {
1886
1975
  const attachmentName = getContextString2(metadata.attachmentName) ?? getAttachmentName2(source, title);
1976
+ const replyDepth = getContextNumber2(metadata.replyDepth);
1977
+ if (attachmentName && replyDepth && replyDepth > 0) {
1978
+ return `Attachment ${attachmentName} \xB7 Reply depth ${replyDepth}`;
1979
+ }
1887
1980
  return attachmentName ? `Attachment ${attachmentName}` : "Attachment";
1888
1981
  }
1889
1982
  const mediaStart = formatMediaTimestamp2(metadata.startMs);
@@ -1914,18 +2007,36 @@ var buildProvenanceLabel2 = (metadata) => {
1914
2007
  return;
1915
2008
  }
1916
2009
  const threadTopic = getContextString2(metadata.threadTopic);
2010
+ const replyDepth = getContextNumber2(metadata.replyDepth);
2011
+ const threadMessageCount = getContextNumber2(metadata.threadMessageCount);
2012
+ const threadRootMessageId = getContextString2(metadata.threadRootMessageId);
1917
2013
  const from = getContextString2(metadata.from);
1918
2014
  const sentAt = formatTimestampLabel2(metadata.sentAt) ?? formatTimestampLabel2(metadata.receivedAt);
1919
2015
  const speaker = getContextString2(metadata.speaker);
1920
2016
  const mediaKind = getContextString2(metadata.mediaKind);
1921
2017
  const transcriptSource = getContextString2(metadata.transcriptSource);
2018
+ const mediaSpeakerCount = getContextNumber2(metadata.mediaSpeakerCount);
2019
+ const mediaSegmentCount = getContextNumber2(metadata.mediaSegmentCount);
2020
+ const mediaSegmentGroupSize = getContextNumber2(metadata.mediaSegmentGroupSize);
2021
+ const mediaSegmentGroupIndex = getContextNumber2(metadata.mediaSegmentGroupIndex);
2022
+ const mediaChannel = getContextString2(metadata.mediaChannel);
2023
+ const mediaDurationLabel = formatMediaDurationLabel2(metadata.mediaDurationMs);
2024
+ const spreadsheetHeaders = getSpreadsheetHeaders(metadata);
2025
+ const slideNotesText = getContextString2(metadata.slideNotesText);
1922
2026
  const pdfTextMode = getContextString2(metadata.pdfTextMode);
1923
2027
  const pdfTextKind = getContextString2(metadata.pdfTextKind);
1924
2028
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
1925
2029
  const ocrEngine = getContextString2(metadata.ocrEngine);
1926
2030
  const extractorRegistryMatch = getContextString2(metadata.extractorRegistryMatch);
1927
2031
  const chunkingProfile = getContextString2(metadata.chunkingProfile);
2032
+ const archiveDepth = getContextNumber2(metadata.archiveDepth);
2033
+ const archiveNestedDepth = getContextNumber2(metadata.archiveNestedDepth);
2034
+ const archiveContainerPath = getContextString2(metadata.archiveContainerPath);
2035
+ const archiveRootName = getContextString2(metadata.archiveRootName);
2036
+ const spreadsheetTableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
1928
2037
  const ocrConfidence = getContextNumber2(metadata.ocrRegionConfidence) ?? getContextNumber2(metadata.ocrConfidence);
2038
+ const ocrAverageConfidence = getContextNumber2(metadata.ocrPageAverageConfidence) ?? getContextNumber2(metadata.ocrAverageConfidence);
2039
+ const ocrRegionCount = getContextNumber2(metadata.ocrRegionCount);
1929
2040
  const labels = [
1930
2041
  pdfTextMode ? `PDF ${pdfTextMode}` : "",
1931
2042
  pdfTextKind === "table_like" ? "PDF table block" : pdfTextKind === "paragraph" ? "PDF text block" : "",
@@ -1934,9 +2045,27 @@ var buildProvenanceLabel2 = (metadata) => {
1934
2045
  extractorRegistryMatch ? `Extractor ${extractorRegistryMatch}` : "",
1935
2046
  chunkingProfile ? `Chunking ${chunkingProfile}` : "",
1936
2047
  typeof ocrConfidence === "number" ? `Confidence ${ocrConfidence.toFixed(2)}` : "",
2048
+ typeof ocrAverageConfidence === "number" && ocrAverageConfidence !== ocrConfidence ? `Average ${ocrAverageConfidence.toFixed(2)}` : "",
2049
+ typeof ocrRegionCount === "number" ? `${ocrRegionCount} regions` : "",
2050
+ spreadsheetHeaders.length > 0 ? `Spreadsheet ${spreadsheetHeaders.join(", ")}` : "",
2051
+ spreadsheetTableLabel ? `Spreadsheet ${spreadsheetTableLabel}` : "",
1937
2052
  mediaKind ? `Media ${mediaKind}` : "",
2053
+ mediaSegmentCount ? `${mediaSegmentCount} segments` : "",
2054
+ mediaSegmentGroupSize ? `${mediaSegmentGroupSize} grouped segments` : "",
2055
+ mediaSegmentGroupIndex !== undefined ? `Segment group ${mediaSegmentGroupIndex + 1}` : "",
2056
+ mediaChannel ? `Channel ${mediaChannel}` : "",
2057
+ mediaSpeakerCount ? `${mediaSpeakerCount} speakers` : "",
2058
+ mediaDurationLabel ? `Duration ${mediaDurationLabel}` : "",
1938
2059
  transcriptSource ? `Transcript ${transcriptSource}` : "",
1939
2060
  threadTopic ? `Thread ${threadTopic}` : "",
2061
+ threadRootMessageId ? `Thread root ${threadRootMessageId}` : "",
2062
+ threadMessageCount ? `${threadMessageCount} thread messages` : "",
2063
+ replyDepth ? `Reply depth ${replyDepth}` : "",
2064
+ slideNotesText ? "Speaker notes" : "",
2065
+ archiveDepth ? `Archive depth ${archiveDepth}` : "",
2066
+ archiveNestedDepth ? `Archive nested depth ${archiveNestedDepth}` : "",
2067
+ archiveContainerPath ? `Archive container ${archiveContainerPath}` : "",
2068
+ archiveRootName ? `Archive root ${archiveRootName}` : "",
1940
2069
  speaker ? `Speaker ${speaker}` : "",
1941
2070
  from ? `Sender ${from}` : "",
1942
2071
  sentAt ? `Sent ${sentAt}` : ""
@@ -2313,7 +2442,7 @@ var getSectionPathFromSource = (source) => {
2313
2442
  const path = source.structure?.section?.path ?? (Array.isArray(source.metadata?.sectionPath) ? source.metadata.sectionPath.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : []);
2314
2443
  return path.length > 0 ? path : undefined;
2315
2444
  };
2316
- var isBlockAwareContextLabel = (value) => typeof value === "string" && (value.startsWith("PDF ") || value.startsWith("Office "));
2445
+ var isBlockAwareContextLabel = (value) => typeof value === "string" && (value.startsWith("PDF ") || value.startsWith("Office ") || value.startsWith("Slide "));
2317
2446
  var getStructuredSectionScoreWeight = (metadata) => {
2318
2447
  if (!metadata) {
2319
2448
  return 1;
@@ -2321,6 +2450,8 @@ var getStructuredSectionScoreWeight = (metadata) => {
2321
2450
  const pdfTextKind = getContextString2(metadata.pdfTextKind);
2322
2451
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
2323
2452
  const sectionKind = getContextString2(metadata.sectionKind);
2453
+ const slideTitle = getContextString2(metadata.slideTitle);
2454
+ const slideNotesText = getContextString2(metadata.slideNotesText);
2324
2455
  if (pdfTextKind === "table_like") {
2325
2456
  return 1.28;
2326
2457
  }
@@ -2330,6 +2461,12 @@ var getStructuredSectionScoreWeight = (metadata) => {
2330
2461
  if (sectionKind === "pdf_block" || sectionKind === "office_block" || officeBlockKind === "paragraph" || pdfTextKind === "paragraph") {
2331
2462
  return 1.12;
2332
2463
  }
2464
+ if (sectionKind === "presentation_slide" && slideNotesText) {
2465
+ return 1.2;
2466
+ }
2467
+ if (sectionKind === "presentation_slide" && slideTitle) {
2468
+ return 1.14;
2469
+ }
2333
2470
  return 1;
2334
2471
  };
2335
2472
  var getStructuredSourceLeadScore = (source) => source.score * getStructuredSectionScoreWeight(source.metadata);
@@ -4780,5 +4917,5 @@ export {
4780
4917
  AIStreamService
4781
4918
  };
4782
4919
 
4783
- //# debugId=367F6BF9D108824C64756E2164756E21
4920
+ //# debugId=9A2760460738B42C64756E2164756E21
4784
4921
  //# sourceMappingURL=index.js.map