@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}` : "",
@@ -4830,6 +4848,34 @@ var getAttachmentName2 = (source, title) => {
4830
4848
  }
4831
4849
  return;
4832
4850
  };
4851
+ var getSpreadsheetHeaders = (metadata) => Array.isArray(metadata?.spreadsheetHeaders) ? metadata.spreadsheetHeaders.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : [];
4852
+ var formatSpreadsheetRowRange = (rowStart, rowEnd) => {
4853
+ if (typeof rowStart !== "number" || !Number.isFinite(rowStart)) {
4854
+ return;
4855
+ }
4856
+ if (typeof rowEnd !== "number" && typeof rowStart === "number" && Number.isFinite(rowStart)) {
4857
+ return `Rows ${rowStart}`;
4858
+ }
4859
+ if (rowStart === rowEnd) {
4860
+ return `Rows ${rowStart}`;
4861
+ }
4862
+ return `Rows ${rowStart}-${rowEnd}`;
4863
+ };
4864
+ var formatSpreadsheetTableLabel = (tableIndex, tableCount) => {
4865
+ if (typeof tableIndex !== "number" || !Number.isFinite(tableIndex) || tableIndex < 1) {
4866
+ return;
4867
+ }
4868
+ if (typeof tableCount === "number" && Number.isFinite(tableCount) && tableCount >= tableIndex) {
4869
+ return `Table ${tableIndex} of ${tableCount}`;
4870
+ }
4871
+ return `Table ${tableIndex}`;
4872
+ };
4873
+ var formatMediaDurationLabel2 = (value) => {
4874
+ if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
4875
+ return;
4876
+ }
4877
+ return formatMediaTimestamp2(value);
4878
+ };
4833
4879
  var buildContextLabel2 = (metadata) => {
4834
4880
  if (!metadata) {
4835
4881
  return;
@@ -4855,29 +4901,61 @@ var buildContextLabel2 = (metadata) => {
4855
4901
  }
4856
4902
  const emailKind = getContextString2(metadata.emailKind);
4857
4903
  if (emailKind === "attachment") {
4858
- return "Attachment evidence";
4904
+ const attachmentName = getContextString2(metadata.attachmentName);
4905
+ const threadTopic2 = getContextString2(metadata.threadTopic);
4906
+ return attachmentName ? threadTopic2 ? `Attachment evidence ${attachmentName} in ${threadTopic2}` : `Attachment evidence ${attachmentName}` : "Attachment evidence";
4859
4907
  }
4860
4908
  if (emailKind === "message") {
4909
+ const threadTopic2 = getContextString2(metadata.threadTopic);
4861
4910
  const from = getContextString2(metadata.from);
4911
+ if (threadTopic2) {
4912
+ return from ? `Message in ${threadTopic2} from ${from}` : `Message in ${threadTopic2}`;
4913
+ }
4862
4914
  return from ? `Message from ${from}` : "Message evidence";
4863
4915
  }
4864
4916
  const page = getContextNumber2(metadata.page) ?? getContextNumber2(metadata.pageNumber) ?? (typeof metadata.pageIndex === "number" ? metadata.pageIndex + 1 : undefined);
4865
4917
  const region = getContextNumber2(metadata.regionNumber) ?? (typeof metadata.regionIndex === "number" ? metadata.regionIndex + 1 : undefined);
4918
+ const hasOCRTrace = typeof metadata.ocrRegionConfidence === "number" || typeof metadata.ocrConfidence === "number" || getContextString2(metadata.pdfTextMode) === "ocr" || typeof metadata.ocrRegionCount === "number";
4866
4919
  if (page && region) {
4920
+ if (hasOCRTrace) {
4921
+ return `OCR page ${page} region ${region}`;
4922
+ }
4867
4923
  return `Page ${page} region ${region}`;
4868
4924
  }
4869
4925
  if (page) {
4926
+ if (hasOCRTrace) {
4927
+ return `OCR page ${page}`;
4928
+ }
4870
4929
  return `Page ${page}`;
4871
4930
  }
4872
4931
  const sheet = getContextString2(metadata.sheetName) ?? (Array.isArray(metadata.sheetNames) ? getContextString2(metadata.sheetNames[0]) : undefined);
4873
4932
  if (sheet) {
4933
+ const tableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
4934
+ const rowRange = formatSpreadsheetRowRange(getContextNumber2(metadata.spreadsheetRowStart), getContextNumber2(metadata.spreadsheetRowEnd));
4935
+ const headers = getSpreadsheetHeaders(metadata);
4936
+ if (tableLabel && rowRange) {
4937
+ return `Sheet ${sheet} ${tableLabel} ${rowRange}`;
4938
+ }
4939
+ if (tableLabel) {
4940
+ return `Sheet ${sheet} ${tableLabel}`;
4941
+ }
4942
+ if (rowRange) {
4943
+ return `Sheet ${sheet} ${rowRange}`;
4944
+ }
4945
+ if (headers.length > 0) {
4946
+ return `Sheet ${sheet} by ${headers.slice(0, 2).join(", ")}`;
4947
+ }
4874
4948
  return `Sheet ${sheet}`;
4875
4949
  }
4876
4950
  const slide = getContextNumber2(metadata.slide) ?? getContextNumber2(metadata.slideNumber) ?? (typeof metadata.slideIndex === "number" ? metadata.slideIndex + 1 : undefined);
4951
+ const slideTitle = getContextString2(metadata.slideTitle);
4877
4952
  if (slide) {
4953
+ if (slideTitle) {
4954
+ return `Slide ${slide} ${slideTitle}`;
4955
+ }
4878
4956
  return `Slide ${slide}`;
4879
4957
  }
4880
- const archiveEntry = getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
4958
+ const archiveEntry = getContextString2(metadata.archiveFullPath) ?? getContextString2(metadata.archivePath) ?? getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
4881
4959
  if (archiveEntry) {
4882
4960
  return `Archive entry ${archiveEntry}`;
4883
4961
  }
@@ -4902,6 +4980,9 @@ var buildLocatorLabel2 = (metadata, source, title) => {
4902
4980
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
4903
4981
  const pdfBlockNumber = getContextNumber2(metadata.pdfBlockNumber);
4904
4982
  const officeBlockNumber = getContextNumber2(metadata.officeBlockNumber);
4983
+ const spreadsheetRowStart = getContextNumber2(metadata.spreadsheetRowStart);
4984
+ const spreadsheetRowEnd = getContextNumber2(metadata.spreadsheetRowEnd);
4985
+ const slideTitle = getContextString2(metadata.slideTitle);
4905
4986
  const page = getContextNumber2(metadata.page) ?? getContextNumber2(metadata.pageNumber) ?? (typeof metadata.pageIndex === "number" ? metadata.pageIndex + 1 : undefined);
4906
4987
  const region = getContextNumber2(metadata.regionNumber) ?? (typeof metadata.regionIndex === "number" ? metadata.regionIndex + 1 : undefined);
4907
4988
  if (page && region) {
@@ -4918,19 +4999,31 @@ var buildLocatorLabel2 = (metadata, source, title) => {
4918
4999
  }
4919
5000
  const sheet = getContextString2(metadata.sheetName) ?? (Array.isArray(metadata.sheetNames) ? getContextString2(metadata.sheetNames[0]) : undefined);
4920
5001
  if (sheet) {
4921
- return `Sheet ${sheet}`;
5002
+ const tableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
5003
+ const rowRange = formatSpreadsheetRowRange(spreadsheetRowStart, spreadsheetRowEnd);
5004
+ if (tableLabel && rowRange) {
5005
+ return `Sheet ${sheet} \xB7 ${tableLabel} \xB7 ${rowRange}`;
5006
+ }
5007
+ if (tableLabel) {
5008
+ return `Sheet ${sheet} \xB7 ${tableLabel}`;
5009
+ }
5010
+ return rowRange ? `Sheet ${sheet} \xB7 ${rowRange}` : `Sheet ${sheet}`;
4922
5011
  }
4923
5012
  const slide = getContextNumber2(metadata.slide) ?? getContextNumber2(metadata.slideNumber) ?? (typeof metadata.slideIndex === "number" ? metadata.slideIndex + 1 : undefined);
4924
5013
  if (slide) {
4925
- return `Slide ${slide}`;
5014
+ return slideTitle ? `Slide ${slide} \xB7 ${slideTitle}` : `Slide ${slide}`;
4926
5015
  }
4927
- const archiveEntry = getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
5016
+ const archiveEntry = getContextString2(metadata.archiveFullPath) ?? getContextString2(metadata.archivePath) ?? getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
4928
5017
  if (archiveEntry) {
4929
5018
  return `Archive entry ${archiveEntry}`;
4930
5019
  }
4931
5020
  const emailKind = getContextString2(metadata.emailKind);
4932
5021
  if (emailKind === "attachment") {
4933
5022
  const attachmentName = getContextString2(metadata.attachmentName) ?? getAttachmentName2(source, title);
5023
+ const replyDepth = getContextNumber2(metadata.replyDepth);
5024
+ if (attachmentName && replyDepth && replyDepth > 0) {
5025
+ return `Attachment ${attachmentName} \xB7 Reply depth ${replyDepth}`;
5026
+ }
4934
5027
  return attachmentName ? `Attachment ${attachmentName}` : "Attachment";
4935
5028
  }
4936
5029
  const mediaStart = formatMediaTimestamp2(metadata.startMs);
@@ -4961,18 +5054,36 @@ var buildProvenanceLabel2 = (metadata) => {
4961
5054
  return;
4962
5055
  }
4963
5056
  const threadTopic = getContextString2(metadata.threadTopic);
5057
+ const replyDepth = getContextNumber2(metadata.replyDepth);
5058
+ const threadMessageCount = getContextNumber2(metadata.threadMessageCount);
5059
+ const threadRootMessageId = getContextString2(metadata.threadRootMessageId);
4964
5060
  const from = getContextString2(metadata.from);
4965
5061
  const sentAt = formatTimestampLabel2(metadata.sentAt) ?? formatTimestampLabel2(metadata.receivedAt);
4966
5062
  const speaker = getContextString2(metadata.speaker);
4967
5063
  const mediaKind = getContextString2(metadata.mediaKind);
4968
5064
  const transcriptSource = getContextString2(metadata.transcriptSource);
5065
+ const mediaSpeakerCount = getContextNumber2(metadata.mediaSpeakerCount);
5066
+ const mediaSegmentCount = getContextNumber2(metadata.mediaSegmentCount);
5067
+ const mediaSegmentGroupSize = getContextNumber2(metadata.mediaSegmentGroupSize);
5068
+ const mediaSegmentGroupIndex = getContextNumber2(metadata.mediaSegmentGroupIndex);
5069
+ const mediaChannel = getContextString2(metadata.mediaChannel);
5070
+ const mediaDurationLabel = formatMediaDurationLabel2(metadata.mediaDurationMs);
5071
+ const spreadsheetHeaders = getSpreadsheetHeaders(metadata);
5072
+ const slideNotesText = getContextString2(metadata.slideNotesText);
4969
5073
  const pdfTextMode = getContextString2(metadata.pdfTextMode);
4970
5074
  const pdfTextKind = getContextString2(metadata.pdfTextKind);
4971
5075
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
4972
5076
  const ocrEngine = getContextString2(metadata.ocrEngine);
4973
5077
  const extractorRegistryMatch = getContextString2(metadata.extractorRegistryMatch);
4974
5078
  const chunkingProfile = getContextString2(metadata.chunkingProfile);
5079
+ const archiveDepth = getContextNumber2(metadata.archiveDepth);
5080
+ const archiveNestedDepth = getContextNumber2(metadata.archiveNestedDepth);
5081
+ const archiveContainerPath = getContextString2(metadata.archiveContainerPath);
5082
+ const archiveRootName = getContextString2(metadata.archiveRootName);
5083
+ const spreadsheetTableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
4975
5084
  const ocrConfidence = getContextNumber2(metadata.ocrRegionConfidence) ?? getContextNumber2(metadata.ocrConfidence);
5085
+ const ocrAverageConfidence = getContextNumber2(metadata.ocrPageAverageConfidence) ?? getContextNumber2(metadata.ocrAverageConfidence);
5086
+ const ocrRegionCount = getContextNumber2(metadata.ocrRegionCount);
4976
5087
  const labels = [
4977
5088
  pdfTextMode ? `PDF ${pdfTextMode}` : "",
4978
5089
  pdfTextKind === "table_like" ? "PDF table block" : pdfTextKind === "paragraph" ? "PDF text block" : "",
@@ -4981,9 +5092,27 @@ var buildProvenanceLabel2 = (metadata) => {
4981
5092
  extractorRegistryMatch ? `Extractor ${extractorRegistryMatch}` : "",
4982
5093
  chunkingProfile ? `Chunking ${chunkingProfile}` : "",
4983
5094
  typeof ocrConfidence === "number" ? `Confidence ${ocrConfidence.toFixed(2)}` : "",
5095
+ typeof ocrAverageConfidence === "number" && ocrAverageConfidence !== ocrConfidence ? `Average ${ocrAverageConfidence.toFixed(2)}` : "",
5096
+ typeof ocrRegionCount === "number" ? `${ocrRegionCount} regions` : "",
5097
+ spreadsheetHeaders.length > 0 ? `Spreadsheet ${spreadsheetHeaders.join(", ")}` : "",
5098
+ spreadsheetTableLabel ? `Spreadsheet ${spreadsheetTableLabel}` : "",
4984
5099
  mediaKind ? `Media ${mediaKind}` : "",
5100
+ mediaSegmentCount ? `${mediaSegmentCount} segments` : "",
5101
+ mediaSegmentGroupSize ? `${mediaSegmentGroupSize} grouped segments` : "",
5102
+ mediaSegmentGroupIndex !== undefined ? `Segment group ${mediaSegmentGroupIndex + 1}` : "",
5103
+ mediaChannel ? `Channel ${mediaChannel}` : "",
5104
+ mediaSpeakerCount ? `${mediaSpeakerCount} speakers` : "",
5105
+ mediaDurationLabel ? `Duration ${mediaDurationLabel}` : "",
4985
5106
  transcriptSource ? `Transcript ${transcriptSource}` : "",
4986
5107
  threadTopic ? `Thread ${threadTopic}` : "",
5108
+ threadRootMessageId ? `Thread root ${threadRootMessageId}` : "",
5109
+ threadMessageCount ? `${threadMessageCount} thread messages` : "",
5110
+ replyDepth ? `Reply depth ${replyDepth}` : "",
5111
+ slideNotesText ? "Speaker notes" : "",
5112
+ archiveDepth ? `Archive depth ${archiveDepth}` : "",
5113
+ archiveNestedDepth ? `Archive nested depth ${archiveNestedDepth}` : "",
5114
+ archiveContainerPath ? `Archive container ${archiveContainerPath}` : "",
5115
+ archiveRootName ? `Archive root ${archiveRootName}` : "",
4987
5116
  speaker ? `Speaker ${speaker}` : "",
4988
5117
  from ? `Sender ${from}` : "",
4989
5118
  sentAt ? `Sent ${sentAt}` : ""
@@ -5360,7 +5489,7 @@ var getSectionPathFromSource = (source) => {
5360
5489
  const path = source.structure?.section?.path ?? (Array.isArray(source.metadata?.sectionPath) ? source.metadata.sectionPath.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : []);
5361
5490
  return path.length > 0 ? path : undefined;
5362
5491
  };
5363
- var isBlockAwareContextLabel = (value) => typeof value === "string" && (value.startsWith("PDF ") || value.startsWith("Office "));
5492
+ var isBlockAwareContextLabel = (value) => typeof value === "string" && (value.startsWith("PDF ") || value.startsWith("Office ") || value.startsWith("Slide "));
5364
5493
  var getStructuredSectionScoreWeight = (metadata) => {
5365
5494
  if (!metadata) {
5366
5495
  return 1;
@@ -5368,6 +5497,8 @@ var getStructuredSectionScoreWeight = (metadata) => {
5368
5497
  const pdfTextKind = getContextString2(metadata.pdfTextKind);
5369
5498
  const officeBlockKind = getContextString2(metadata.officeBlockKind);
5370
5499
  const sectionKind = getContextString2(metadata.sectionKind);
5500
+ const slideTitle = getContextString2(metadata.slideTitle);
5501
+ const slideNotesText = getContextString2(metadata.slideNotesText);
5371
5502
  if (pdfTextKind === "table_like") {
5372
5503
  return 1.28;
5373
5504
  }
@@ -5377,6 +5508,12 @@ var getStructuredSectionScoreWeight = (metadata) => {
5377
5508
  if (sectionKind === "pdf_block" || sectionKind === "office_block" || officeBlockKind === "paragraph" || pdfTextKind === "paragraph") {
5378
5509
  return 1.12;
5379
5510
  }
5511
+ if (sectionKind === "presentation_slide" && slideNotesText) {
5512
+ return 1.2;
5513
+ }
5514
+ if (sectionKind === "presentation_slide" && slideTitle) {
5515
+ return 1.14;
5516
+ }
5380
5517
  return 1;
5381
5518
  };
5382
5519
  var getStructuredSourceLeadScore = (source) => source.score * getStructuredSectionScoreWeight(source.metadata);
@@ -7592,5 +7729,5 @@ export {
7592
7729
  buildRAGEvaluationLeaderboard
7593
7730
  };
7594
7731
 
7595
- //# debugId=EFA1619967730BA064756E2164756E21
7732
+ //# debugId=69E44CDEAA9B989C64756E2164756E21
7596
7733
  //# sourceMappingURL=index.js.map