@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.
@@ -260,6 +260,12 @@ var formatMediaTimestamp = (value) => {
260
260
  const milliseconds = Math.floor(value % 1000);
261
261
  return `${String(minutes).padStart(2, "0")}:${String(seconds).padStart(2, "0")}.${String(milliseconds).padStart(3, "0")}`;
262
262
  };
263
+ var formatMediaDurationLabel = (value) => {
264
+ if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
265
+ return;
266
+ }
267
+ return formatMediaTimestamp(value);
268
+ };
263
269
  var buildLocatorLabel = (metadata, source, title) => {
264
270
  if (!metadata) {
265
271
  return;
@@ -322,6 +328,12 @@ var buildProvenanceLabel = (metadata) => {
322
328
  const sentAt = formatTimestampLabel(metadata.sentAt) ?? formatTimestampLabel(metadata.receivedAt);
323
329
  const speaker = getContextString(metadata.speaker);
324
330
  const mediaKind = getContextString(metadata.mediaKind);
331
+ const mediaSegmentCount = getContextNumber(metadata.mediaSegmentCount);
332
+ const mediaSegmentGroupSize = getContextNumber(metadata.mediaSegmentGroupSize);
333
+ const mediaSegmentGroupIndex = getContextNumber(metadata.mediaSegmentGroupIndex);
334
+ const mediaChannel = getContextString(metadata.mediaChannel);
335
+ const mediaSpeakerCount = getContextNumber(metadata.mediaSpeakerCount);
336
+ const mediaDurationLabel = formatMediaDurationLabel(metadata.mediaDurationMs);
325
337
  const transcriptSource = getContextString(metadata.transcriptSource);
326
338
  const pdfTextMode = getContextString(metadata.pdfTextMode);
327
339
  const ocrEngine = getContextString(metadata.ocrEngine);
@@ -331,6 +343,12 @@ var buildProvenanceLabel = (metadata) => {
331
343
  ocrEngine ? `OCR ${ocrEngine}` : "",
332
344
  typeof ocrConfidence === "number" ? `Confidence ${ocrConfidence.toFixed(2)}` : "",
333
345
  mediaKind ? `Media ${mediaKind}` : "",
346
+ mediaSegmentCount ? `${mediaSegmentCount} segments` : "",
347
+ mediaSegmentGroupSize ? `${mediaSegmentGroupSize} grouped segments` : "",
348
+ mediaSegmentGroupIndex !== undefined ? `Segment group ${mediaSegmentGroupIndex + 1}` : "",
349
+ mediaChannel ? `Channel ${mediaChannel}` : "",
350
+ mediaSpeakerCount ? `${mediaSpeakerCount} speakers` : "",
351
+ mediaDurationLabel ? `Duration ${mediaDurationLabel}` : "",
334
352
  transcriptSource ? `Transcript ${transcriptSource}` : "",
335
353
  threadTopic ? `Thread ${threadTopic}` : "",
336
354
  speaker ? `Speaker ${speaker}` : "",
@@ -4233,6 +4251,34 @@ var getAttachmentName2 = (source, title) => {
4233
4251
  }
4234
4252
  return;
4235
4253
  };
4254
+ var getSpreadsheetHeaders = (metadata) => Array.isArray(metadata?.spreadsheetHeaders) ? metadata.spreadsheetHeaders.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : [];
4255
+ var formatSpreadsheetRowRange = (rowStart, rowEnd) => {
4256
+ if (typeof rowStart !== "number" || !Number.isFinite(rowStart)) {
4257
+ return;
4258
+ }
4259
+ if (typeof rowEnd !== "number" && typeof rowStart === "number" && Number.isFinite(rowStart)) {
4260
+ return `Rows ${rowStart}`;
4261
+ }
4262
+ if (rowStart === rowEnd) {
4263
+ return `Rows ${rowStart}`;
4264
+ }
4265
+ return `Rows ${rowStart}-${rowEnd}`;
4266
+ };
4267
+ var formatSpreadsheetTableLabel = (tableIndex, tableCount) => {
4268
+ if (typeof tableIndex !== "number" || !Number.isFinite(tableIndex) || tableIndex < 1) {
4269
+ return;
4270
+ }
4271
+ if (typeof tableCount === "number" && Number.isFinite(tableCount) && tableCount >= tableIndex) {
4272
+ return `Table ${tableIndex} of ${tableCount}`;
4273
+ }
4274
+ return `Table ${tableIndex}`;
4275
+ };
4276
+ var formatMediaDurationLabel2 = (value) => {
4277
+ if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
4278
+ return;
4279
+ }
4280
+ return formatMediaTimestamp2(value);
4281
+ };
4236
4282
  var buildContextLabel2 = (metadata) => {
4237
4283
  if (!metadata) {
4238
4284
  return;
@@ -4258,29 +4304,61 @@ var buildContextLabel2 = (metadata) => {
4258
4304
  }
4259
4305
  const emailKind = getContextString2(metadata.emailKind);
4260
4306
  if (emailKind === "attachment") {
4261
- return "Attachment evidence";
4307
+ const attachmentName = getContextString2(metadata.attachmentName);
4308
+ const threadTopic2 = getContextString2(metadata.threadTopic);
4309
+ return attachmentName ? threadTopic2 ? `Attachment evidence ${attachmentName} in ${threadTopic2}` : `Attachment evidence ${attachmentName}` : "Attachment evidence";
4262
4310
  }
4263
4311
  if (emailKind === "message") {
4312
+ const threadTopic2 = getContextString2(metadata.threadTopic);
4264
4313
  const from = getContextString2(metadata.from);
4314
+ if (threadTopic2) {
4315
+ return from ? `Message in ${threadTopic2} from ${from}` : `Message in ${threadTopic2}`;
4316
+ }
4265
4317
  return from ? `Message from ${from}` : "Message evidence";
4266
4318
  }
4267
4319
  const page = getContextNumber2(metadata.page) ?? getContextNumber2(metadata.pageNumber) ?? (typeof metadata.pageIndex === "number" ? metadata.pageIndex + 1 : undefined);
4268
4320
  const region = getContextNumber2(metadata.regionNumber) ?? (typeof metadata.regionIndex === "number" ? metadata.regionIndex + 1 : undefined);
4321
+ const hasOCRTrace = typeof metadata.ocrRegionConfidence === "number" || typeof metadata.ocrConfidence === "number" || getContextString2(metadata.pdfTextMode) === "ocr" || typeof metadata.ocrRegionCount === "number";
4269
4322
  if (page && region) {
4323
+ if (hasOCRTrace) {
4324
+ return `OCR page ${page} region ${region}`;
4325
+ }
4270
4326
  return `Page ${page} region ${region}`;
4271
4327
  }
4272
4328
  if (page) {
4329
+ if (hasOCRTrace) {
4330
+ return `OCR page ${page}`;
4331
+ }
4273
4332
  return `Page ${page}`;
4274
4333
  }
4275
4334
  const sheet = getContextString2(metadata.sheetName) ?? (Array.isArray(metadata.sheetNames) ? getContextString2(metadata.sheetNames[0]) : undefined);
4276
4335
  if (sheet) {
4336
+ const tableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
4337
+ const rowRange = formatSpreadsheetRowRange(getContextNumber2(metadata.spreadsheetRowStart), getContextNumber2(metadata.spreadsheetRowEnd));
4338
+ const headers = getSpreadsheetHeaders(metadata);
4339
+ if (tableLabel && rowRange) {
4340
+ return `Sheet ${sheet} ${tableLabel} ${rowRange}`;
4341
+ }
4342
+ if (tableLabel) {
4343
+ return `Sheet ${sheet} ${tableLabel}`;
4344
+ }
4345
+ if (rowRange) {
4346
+ return `Sheet ${sheet} ${rowRange}`;
4347
+ }
4348
+ if (headers.length > 0) {
4349
+ return `Sheet ${sheet} by ${headers.slice(0, 2).join(", ")}`;
4350
+ }
4277
4351
  return `Sheet ${sheet}`;
4278
4352
  }
4279
4353
  const slide = getContextNumber2(metadata.slide) ?? getContextNumber2(metadata.slideNumber) ?? (typeof metadata.slideIndex === "number" ? metadata.slideIndex + 1 : undefined);
4354
+ const slideTitle = getContextString2(metadata.slideTitle);
4280
4355
  if (slide) {
4356
+ if (slideTitle) {
4357
+ return `Slide ${slide} ${slideTitle}`;
4358
+ }
4281
4359
  return `Slide ${slide}`;
4282
4360
  }
4283
- const archiveEntry = getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
4361
+ const archiveEntry = getContextString2(metadata.archiveFullPath) ?? getContextString2(metadata.archivePath) ?? getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
4284
4362
  if (archiveEntry) {
4285
4363
  return `Archive entry ${archiveEntry}`;
4286
4364
  }
@@ -4305,6 +4383,9 @@ var buildLocatorLabel2 = (metadata, source, title) => {
4305
4383
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
4306
4384
  const pdfBlockNumber = getContextNumber2(metadata.pdfBlockNumber);
4307
4385
  const officeBlockNumber = getContextNumber2(metadata.officeBlockNumber);
4386
+ const spreadsheetRowStart = getContextNumber2(metadata.spreadsheetRowStart);
4387
+ const spreadsheetRowEnd = getContextNumber2(metadata.spreadsheetRowEnd);
4388
+ const slideTitle = getContextString2(metadata.slideTitle);
4308
4389
  const page = getContextNumber2(metadata.page) ?? getContextNumber2(metadata.pageNumber) ?? (typeof metadata.pageIndex === "number" ? metadata.pageIndex + 1 : undefined);
4309
4390
  const region = getContextNumber2(metadata.regionNumber) ?? (typeof metadata.regionIndex === "number" ? metadata.regionIndex + 1 : undefined);
4310
4391
  if (page && region) {
@@ -4321,19 +4402,31 @@ var buildLocatorLabel2 = (metadata, source, title) => {
4321
4402
  }
4322
4403
  const sheet = getContextString2(metadata.sheetName) ?? (Array.isArray(metadata.sheetNames) ? getContextString2(metadata.sheetNames[0]) : undefined);
4323
4404
  if (sheet) {
4324
- return `Sheet ${sheet}`;
4405
+ const tableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
4406
+ const rowRange = formatSpreadsheetRowRange(spreadsheetRowStart, spreadsheetRowEnd);
4407
+ if (tableLabel && rowRange) {
4408
+ return `Sheet ${sheet} \xB7 ${tableLabel} \xB7 ${rowRange}`;
4409
+ }
4410
+ if (tableLabel) {
4411
+ return `Sheet ${sheet} \xB7 ${tableLabel}`;
4412
+ }
4413
+ return rowRange ? `Sheet ${sheet} \xB7 ${rowRange}` : `Sheet ${sheet}`;
4325
4414
  }
4326
4415
  const slide = getContextNumber2(metadata.slide) ?? getContextNumber2(metadata.slideNumber) ?? (typeof metadata.slideIndex === "number" ? metadata.slideIndex + 1 : undefined);
4327
4416
  if (slide) {
4328
- return `Slide ${slide}`;
4417
+ return slideTitle ? `Slide ${slide} \xB7 ${slideTitle}` : `Slide ${slide}`;
4329
4418
  }
4330
- const archiveEntry = getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
4419
+ const archiveEntry = getContextString2(metadata.archiveFullPath) ?? getContextString2(metadata.archivePath) ?? getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
4331
4420
  if (archiveEntry) {
4332
4421
  return `Archive entry ${archiveEntry}`;
4333
4422
  }
4334
4423
  const emailKind = getContextString2(metadata.emailKind);
4335
4424
  if (emailKind === "attachment") {
4336
4425
  const attachmentName = getContextString2(metadata.attachmentName) ?? getAttachmentName2(source, title);
4426
+ const replyDepth = getContextNumber2(metadata.replyDepth);
4427
+ if (attachmentName && replyDepth && replyDepth > 0) {
4428
+ return `Attachment ${attachmentName} \xB7 Reply depth ${replyDepth}`;
4429
+ }
4337
4430
  return attachmentName ? `Attachment ${attachmentName}` : "Attachment";
4338
4431
  }
4339
4432
  const mediaStart = formatMediaTimestamp2(metadata.startMs);
@@ -4364,18 +4457,36 @@ var buildProvenanceLabel2 = (metadata) => {
4364
4457
  return;
4365
4458
  }
4366
4459
  const threadTopic = getContextString2(metadata.threadTopic);
4460
+ const replyDepth = getContextNumber2(metadata.replyDepth);
4461
+ const threadMessageCount = getContextNumber2(metadata.threadMessageCount);
4462
+ const threadRootMessageId = getContextString2(metadata.threadRootMessageId);
4367
4463
  const from = getContextString2(metadata.from);
4368
4464
  const sentAt = formatTimestampLabel2(metadata.sentAt) ?? formatTimestampLabel2(metadata.receivedAt);
4369
4465
  const speaker = getContextString2(metadata.speaker);
4370
4466
  const mediaKind = getContextString2(metadata.mediaKind);
4371
4467
  const transcriptSource = getContextString2(metadata.transcriptSource);
4468
+ const mediaSpeakerCount = getContextNumber2(metadata.mediaSpeakerCount);
4469
+ const mediaSegmentCount = getContextNumber2(metadata.mediaSegmentCount);
4470
+ const mediaSegmentGroupSize = getContextNumber2(metadata.mediaSegmentGroupSize);
4471
+ const mediaSegmentGroupIndex = getContextNumber2(metadata.mediaSegmentGroupIndex);
4472
+ const mediaChannel = getContextString2(metadata.mediaChannel);
4473
+ const mediaDurationLabel = formatMediaDurationLabel2(metadata.mediaDurationMs);
4474
+ const spreadsheetHeaders = getSpreadsheetHeaders(metadata);
4475
+ const slideNotesText = getContextString2(metadata.slideNotesText);
4372
4476
  const pdfTextMode = getContextString2(metadata.pdfTextMode);
4373
4477
  const pdfTextKind = getContextString2(metadata.pdfTextKind);
4374
4478
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
4375
4479
  const ocrEngine = getContextString2(metadata.ocrEngine);
4376
4480
  const extractorRegistryMatch = getContextString2(metadata.extractorRegistryMatch);
4377
4481
  const chunkingProfile = getContextString2(metadata.chunkingProfile);
4482
+ const archiveDepth = getContextNumber2(metadata.archiveDepth);
4483
+ const archiveNestedDepth = getContextNumber2(metadata.archiveNestedDepth);
4484
+ const archiveContainerPath = getContextString2(metadata.archiveContainerPath);
4485
+ const archiveRootName = getContextString2(metadata.archiveRootName);
4486
+ const spreadsheetTableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
4378
4487
  const ocrConfidence = getContextNumber2(metadata.ocrRegionConfidence) ?? getContextNumber2(metadata.ocrConfidence);
4488
+ const ocrAverageConfidence = getContextNumber2(metadata.ocrPageAverageConfidence) ?? getContextNumber2(metadata.ocrAverageConfidence);
4489
+ const ocrRegionCount = getContextNumber2(metadata.ocrRegionCount);
4379
4490
  const labels = [
4380
4491
  pdfTextMode ? `PDF ${pdfTextMode}` : "",
4381
4492
  pdfTextKind === "table_like" ? "PDF table block" : pdfTextKind === "paragraph" ? "PDF text block" : "",
@@ -4384,9 +4495,27 @@ var buildProvenanceLabel2 = (metadata) => {
4384
4495
  extractorRegistryMatch ? `Extractor ${extractorRegistryMatch}` : "",
4385
4496
  chunkingProfile ? `Chunking ${chunkingProfile}` : "",
4386
4497
  typeof ocrConfidence === "number" ? `Confidence ${ocrConfidence.toFixed(2)}` : "",
4498
+ typeof ocrAverageConfidence === "number" && ocrAverageConfidence !== ocrConfidence ? `Average ${ocrAverageConfidence.toFixed(2)}` : "",
4499
+ typeof ocrRegionCount === "number" ? `${ocrRegionCount} regions` : "",
4500
+ spreadsheetHeaders.length > 0 ? `Spreadsheet ${spreadsheetHeaders.join(", ")}` : "",
4501
+ spreadsheetTableLabel ? `Spreadsheet ${spreadsheetTableLabel}` : "",
4387
4502
  mediaKind ? `Media ${mediaKind}` : "",
4503
+ mediaSegmentCount ? `${mediaSegmentCount} segments` : "",
4504
+ mediaSegmentGroupSize ? `${mediaSegmentGroupSize} grouped segments` : "",
4505
+ mediaSegmentGroupIndex !== undefined ? `Segment group ${mediaSegmentGroupIndex + 1}` : "",
4506
+ mediaChannel ? `Channel ${mediaChannel}` : "",
4507
+ mediaSpeakerCount ? `${mediaSpeakerCount} speakers` : "",
4508
+ mediaDurationLabel ? `Duration ${mediaDurationLabel}` : "",
4388
4509
  transcriptSource ? `Transcript ${transcriptSource}` : "",
4389
4510
  threadTopic ? `Thread ${threadTopic}` : "",
4511
+ threadRootMessageId ? `Thread root ${threadRootMessageId}` : "",
4512
+ threadMessageCount ? `${threadMessageCount} thread messages` : "",
4513
+ replyDepth ? `Reply depth ${replyDepth}` : "",
4514
+ slideNotesText ? "Speaker notes" : "",
4515
+ archiveDepth ? `Archive depth ${archiveDepth}` : "",
4516
+ archiveNestedDepth ? `Archive nested depth ${archiveNestedDepth}` : "",
4517
+ archiveContainerPath ? `Archive container ${archiveContainerPath}` : "",
4518
+ archiveRootName ? `Archive root ${archiveRootName}` : "",
4390
4519
  speaker ? `Speaker ${speaker}` : "",
4391
4520
  from ? `Sender ${from}` : "",
4392
4521
  sentAt ? `Sent ${sentAt}` : ""
@@ -4763,7 +4892,7 @@ var getSectionPathFromSource = (source) => {
4763
4892
  const path = source.structure?.section?.path ?? (Array.isArray(source.metadata?.sectionPath) ? source.metadata.sectionPath.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : []);
4764
4893
  return path.length > 0 ? path : undefined;
4765
4894
  };
4766
- var isBlockAwareContextLabel = (value) => typeof value === "string" && (value.startsWith("PDF ") || value.startsWith("Office "));
4895
+ var isBlockAwareContextLabel = (value) => typeof value === "string" && (value.startsWith("PDF ") || value.startsWith("Office ") || value.startsWith("Slide "));
4767
4896
  var getStructuredSectionScoreWeight = (metadata) => {
4768
4897
  if (!metadata) {
4769
4898
  return 1;
@@ -4771,6 +4900,8 @@ var getStructuredSectionScoreWeight = (metadata) => {
4771
4900
  const pdfTextKind = getContextString2(metadata.pdfTextKind);
4772
4901
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
4773
4902
  const sectionKind = getContextString2(metadata.sectionKind);
4903
+ const slideTitle = getContextString2(metadata.slideTitle);
4904
+ const slideNotesText = getContextString2(metadata.slideNotesText);
4774
4905
  if (pdfTextKind === "table_like") {
4775
4906
  return 1.28;
4776
4907
  }
@@ -4780,6 +4911,12 @@ var getStructuredSectionScoreWeight = (metadata) => {
4780
4911
  if (sectionKind === "pdf_block" || sectionKind === "office_block" || officeBlockKind === "paragraph" || pdfTextKind === "paragraph") {
4781
4912
  return 1.12;
4782
4913
  }
4914
+ if (sectionKind === "presentation_slide" && slideNotesText) {
4915
+ return 1.2;
4916
+ }
4917
+ if (sectionKind === "presentation_slide" && slideTitle) {
4918
+ return 1.14;
4919
+ }
4783
4920
  return 1;
4784
4921
  };
4785
4922
  var getStructuredSourceLeadScore = (source) => source.score * getStructuredSectionScoreWeight(source.metadata);
@@ -8376,5 +8513,5 @@ export {
8376
8513
  createAIStream
8377
8514
  };
8378
8515
 
8379
- //# debugId=18F5304058E0B3D264756E2164756E21
8516
+ //# debugId=EBC8D4C0F871AFBA64756E2164756E21
8380
8517
  //# sourceMappingURL=index.js.map