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

package/dist/ai/index.js

@@ -260,6 +260,12 @@ var formatMediaTimestamp = (value) => {
   const milliseconds = Math.floor(value % 1000);
   return `${String(minutes).padStart(2, "0")}:${String(seconds).padStart(2, "0")}.${String(milliseconds).padStart(3, "0")}`;
 };
+var formatMediaDurationLabel = (value) => {
+  if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
+    return;
+  }
+  return formatMediaTimestamp(value);
+};
 var buildLocatorLabel = (metadata, source, title) => {
   if (!metadata) {
     return;
@@ -322,6 +328,12 @@ var buildProvenanceLabel = (metadata) => {
   const sentAt = formatTimestampLabel(metadata.sentAt) ?? formatTimestampLabel(metadata.receivedAt);
   const speaker = getContextString(metadata.speaker);
   const mediaKind = getContextString(metadata.mediaKind);
+  const mediaSegmentCount = getContextNumber(metadata.mediaSegmentCount);
+  const mediaSegmentGroupSize = getContextNumber(metadata.mediaSegmentGroupSize);
+  const mediaSegmentGroupIndex = getContextNumber(metadata.mediaSegmentGroupIndex);
+  const mediaChannel = getContextString(metadata.mediaChannel);
+  const mediaSpeakerCount = getContextNumber(metadata.mediaSpeakerCount);
+  const mediaDurationLabel = formatMediaDurationLabel(metadata.mediaDurationMs);
   const transcriptSource = getContextString(metadata.transcriptSource);
   const pdfTextMode = getContextString(metadata.pdfTextMode);
   const ocrEngine = getContextString(metadata.ocrEngine);
@@ -331,6 +343,12 @@ var buildProvenanceLabel = (metadata) => {
     ocrEngine ? `OCR ${ocrEngine}` : "",
     typeof ocrConfidence === "number" ? `Confidence ${ocrConfidence.toFixed(2)}` : "",
     mediaKind ? `Media ${mediaKind}` : "",
+    mediaSegmentCount ? `${mediaSegmentCount} segments` : "",
+    mediaSegmentGroupSize ? `${mediaSegmentGroupSize} grouped segments` : "",
+    mediaSegmentGroupIndex !== undefined ? `Segment group ${mediaSegmentGroupIndex + 1}` : "",
+    mediaChannel ? `Channel ${mediaChannel}` : "",
+    mediaSpeakerCount ? `${mediaSpeakerCount} speakers` : "",
+    mediaDurationLabel ? `Duration ${mediaDurationLabel}` : "",
     transcriptSource ? `Transcript ${transcriptSource}` : "",
     threadTopic ? `Thread ${threadTopic}` : "",
     speaker ? `Speaker ${speaker}` : "",
@@ -4233,6 +4251,34 @@ var getAttachmentName2 = (source, title) => {
   }
   return;
 };
+var getSpreadsheetHeaders = (metadata) => Array.isArray(metadata?.spreadsheetHeaders) ? metadata.spreadsheetHeaders.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : [];
+var formatSpreadsheetRowRange = (rowStart, rowEnd) => {
+  if (typeof rowStart !== "number" || !Number.isFinite(rowStart)) {
+    return;
+  }
+  if (typeof rowEnd !== "number" && typeof rowStart === "number" && Number.isFinite(rowStart)) {
+    return `Rows ${rowStart}`;
+  }
+  if (rowStart === rowEnd) {
+    return `Rows ${rowStart}`;
+  }
+  return `Rows ${rowStart}-${rowEnd}`;
+};
+var formatSpreadsheetTableLabel = (tableIndex, tableCount) => {
+  if (typeof tableIndex !== "number" || !Number.isFinite(tableIndex) || tableIndex < 1) {
+    return;
+  }
+  if (typeof tableCount === "number" && Number.isFinite(tableCount) && tableCount >= tableIndex) {
+    return `Table ${tableIndex} of ${tableCount}`;
+  }
+  return `Table ${tableIndex}`;
+};
+var formatMediaDurationLabel2 = (value) => {
+  if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
+    return;
+  }
+  return formatMediaTimestamp2(value);
+};
 var buildContextLabel2 = (metadata) => {
   if (!metadata) {
     return;
@@ -4258,29 +4304,61 @@ var buildContextLabel2 = (metadata) => {
   }
   const emailKind = getContextString2(metadata.emailKind);
   if (emailKind === "attachment") {
-    return "Attachment evidence";
+    const attachmentName = getContextString2(metadata.attachmentName);
+    const threadTopic2 = getContextString2(metadata.threadTopic);
+    return attachmentName ? threadTopic2 ? `Attachment evidence ${attachmentName} in ${threadTopic2}` : `Attachment evidence ${attachmentName}` : "Attachment evidence";
   }
   if (emailKind === "message") {
+    const threadTopic2 = getContextString2(metadata.threadTopic);
     const from = getContextString2(metadata.from);
+    if (threadTopic2) {
+      return from ? `Message in ${threadTopic2} from ${from}` : `Message in ${threadTopic2}`;
+    }
     return from ? `Message from ${from}` : "Message evidence";
   }
   const page = getContextNumber2(metadata.page) ?? getContextNumber2(metadata.pageNumber) ?? (typeof metadata.pageIndex === "number" ? metadata.pageIndex + 1 : undefined);
   const region = getContextNumber2(metadata.regionNumber) ?? (typeof metadata.regionIndex === "number" ? metadata.regionIndex + 1 : undefined);
+  const hasOCRTrace = typeof metadata.ocrRegionConfidence === "number" || typeof metadata.ocrConfidence === "number" || getContextString2(metadata.pdfTextMode) === "ocr" || typeof metadata.ocrRegionCount === "number";
   if (page && region) {
+    if (hasOCRTrace) {
+      return `OCR page ${page} region ${region}`;
+    }
     return `Page ${page} region ${region}`;
   }
   if (page) {
+    if (hasOCRTrace) {
+      return `OCR page ${page}`;
+    }
     return `Page ${page}`;
   }
   const sheet = getContextString2(metadata.sheetName) ?? (Array.isArray(metadata.sheetNames) ? getContextString2(metadata.sheetNames[0]) : undefined);
   if (sheet) {
+    const tableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
+    const rowRange = formatSpreadsheetRowRange(getContextNumber2(metadata.spreadsheetRowStart), getContextNumber2(metadata.spreadsheetRowEnd));
+    const headers = getSpreadsheetHeaders(metadata);
+    if (tableLabel && rowRange) {
+      return `Sheet ${sheet} ${tableLabel} ${rowRange}`;
+    }
+    if (tableLabel) {
+      return `Sheet ${sheet} ${tableLabel}`;
+    }
+    if (rowRange) {
+      return `Sheet ${sheet} ${rowRange}`;
+    }
+    if (headers.length > 0) {
+      return `Sheet ${sheet} by ${headers.slice(0, 2).join(", ")}`;
+    }
     return `Sheet ${sheet}`;
   }
   const slide = getContextNumber2(metadata.slide) ?? getContextNumber2(metadata.slideNumber) ?? (typeof metadata.slideIndex === "number" ? metadata.slideIndex + 1 : undefined);
+  const slideTitle = getContextString2(metadata.slideTitle);
   if (slide) {
+    if (slideTitle) {
+      return `Slide ${slide} ${slideTitle}`;
+    }
     return `Slide ${slide}`;
   }
-  const archiveEntry = getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
+  const archiveEntry = getContextString2(metadata.archiveFullPath) ?? getContextString2(metadata.archivePath) ?? getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
   if (archiveEntry) {
     return `Archive entry ${archiveEntry}`;
   }
@@ -4305,6 +4383,9 @@ var buildLocatorLabel2 = (metadata, source, title) => {
   const officeBlockKind = getContextString2(metadata.officeBlockKind);
   const pdfBlockNumber = getContextNumber2(metadata.pdfBlockNumber);
   const officeBlockNumber = getContextNumber2(metadata.officeBlockNumber);
+  const spreadsheetRowStart = getContextNumber2(metadata.spreadsheetRowStart);
+  const spreadsheetRowEnd = getContextNumber2(metadata.spreadsheetRowEnd);
+  const slideTitle = getContextString2(metadata.slideTitle);
   const page = getContextNumber2(metadata.page) ?? getContextNumber2(metadata.pageNumber) ?? (typeof metadata.pageIndex === "number" ? metadata.pageIndex + 1 : undefined);
   const region = getContextNumber2(metadata.regionNumber) ?? (typeof metadata.regionIndex === "number" ? metadata.regionIndex + 1 : undefined);
   if (page && region) {
@@ -4321,19 +4402,31 @@ var buildLocatorLabel2 = (metadata, source, title) => {
   }
   const sheet = getContextString2(metadata.sheetName) ?? (Array.isArray(metadata.sheetNames) ? getContextString2(metadata.sheetNames[0]) : undefined);
   if (sheet) {
-    return `Sheet ${sheet}`;
+    const tableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
+    const rowRange = formatSpreadsheetRowRange(spreadsheetRowStart, spreadsheetRowEnd);
+    if (tableLabel && rowRange) {
+      return `Sheet ${sheet} \xB7 ${tableLabel} \xB7 ${rowRange}`;
+    }
+    if (tableLabel) {
+      return `Sheet ${sheet} \xB7 ${tableLabel}`;
+    }
+    return rowRange ? `Sheet ${sheet} \xB7 ${rowRange}` : `Sheet ${sheet}`;
   }
   const slide = getContextNumber2(metadata.slide) ?? getContextNumber2(metadata.slideNumber) ?? (typeof metadata.slideIndex === "number" ? metadata.slideIndex + 1 : undefined);
   if (slide) {
-    return `Slide ${slide}`;
+    return slideTitle ? `Slide ${slide} \xB7 ${slideTitle}` : `Slide ${slide}`;
   }
-  const archiveEntry = getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
+  const archiveEntry = getContextString2(metadata.archiveFullPath) ?? getContextString2(metadata.archivePath) ?? getContextString2(metadata.archiveEntryPath) ?? getContextString2(metadata.entryPath);
   if (archiveEntry) {
     return `Archive entry ${archiveEntry}`;
   }
   const emailKind = getContextString2(metadata.emailKind);
   if (emailKind === "attachment") {
     const attachmentName = getContextString2(metadata.attachmentName) ?? getAttachmentName2(source, title);
+    const replyDepth = getContextNumber2(metadata.replyDepth);
+    if (attachmentName && replyDepth && replyDepth > 0) {
+      return `Attachment ${attachmentName} \xB7 Reply depth ${replyDepth}`;
+    }
     return attachmentName ? `Attachment ${attachmentName}` : "Attachment";
   }
   const mediaStart = formatMediaTimestamp2(metadata.startMs);
@@ -4364,18 +4457,36 @@ var buildProvenanceLabel2 = (metadata) => {
     return;
   }
   const threadTopic = getContextString2(metadata.threadTopic);
+  const replyDepth = getContextNumber2(metadata.replyDepth);
+  const threadMessageCount = getContextNumber2(metadata.threadMessageCount);
+  const threadRootMessageId = getContextString2(metadata.threadRootMessageId);
   const from = getContextString2(metadata.from);
   const sentAt = formatTimestampLabel2(metadata.sentAt) ?? formatTimestampLabel2(metadata.receivedAt);
   const speaker = getContextString2(metadata.speaker);
   const mediaKind = getContextString2(metadata.mediaKind);
   const transcriptSource = getContextString2(metadata.transcriptSource);
+  const mediaSpeakerCount = getContextNumber2(metadata.mediaSpeakerCount);
+  const mediaSegmentCount = getContextNumber2(metadata.mediaSegmentCount);
+  const mediaSegmentGroupSize = getContextNumber2(metadata.mediaSegmentGroupSize);
+  const mediaSegmentGroupIndex = getContextNumber2(metadata.mediaSegmentGroupIndex);
+  const mediaChannel = getContextString2(metadata.mediaChannel);
+  const mediaDurationLabel = formatMediaDurationLabel2(metadata.mediaDurationMs);
+  const spreadsheetHeaders = getSpreadsheetHeaders(metadata);
+  const slideNotesText = getContextString2(metadata.slideNotesText);
   const pdfTextMode = getContextString2(metadata.pdfTextMode);
   const pdfTextKind = getContextString2(metadata.pdfTextKind);
   const officeBlockKind = getContextString2(metadata.officeBlockKind);
   const ocrEngine = getContextString2(metadata.ocrEngine);
   const extractorRegistryMatch = getContextString2(metadata.extractorRegistryMatch);
   const chunkingProfile = getContextString2(metadata.chunkingProfile);
+  const archiveDepth = getContextNumber2(metadata.archiveDepth);
+  const archiveNestedDepth = getContextNumber2(metadata.archiveNestedDepth);
+  const archiveContainerPath = getContextString2(metadata.archiveContainerPath);
+  const archiveRootName = getContextString2(metadata.archiveRootName);
+  const spreadsheetTableLabel = formatSpreadsheetTableLabel(getContextNumber2(metadata.spreadsheetTableIndex), getContextNumber2(metadata.spreadsheetTableCount));
   const ocrConfidence = getContextNumber2(metadata.ocrRegionConfidence) ?? getContextNumber2(metadata.ocrConfidence);
+  const ocrAverageConfidence = getContextNumber2(metadata.ocrPageAverageConfidence) ?? getContextNumber2(metadata.ocrAverageConfidence);
+  const ocrRegionCount = getContextNumber2(metadata.ocrRegionCount);
   const labels = [
     pdfTextMode ? `PDF ${pdfTextMode}` : "",
     pdfTextKind === "table_like" ? "PDF table block" : pdfTextKind === "paragraph" ? "PDF text block" : "",
@@ -4384,9 +4495,27 @@ var buildProvenanceLabel2 = (metadata) => {
     extractorRegistryMatch ? `Extractor ${extractorRegistryMatch}` : "",
     chunkingProfile ? `Chunking ${chunkingProfile}` : "",
     typeof ocrConfidence === "number" ? `Confidence ${ocrConfidence.toFixed(2)}` : "",
+    typeof ocrAverageConfidence === "number" && ocrAverageConfidence !== ocrConfidence ? `Average ${ocrAverageConfidence.toFixed(2)}` : "",
+    typeof ocrRegionCount === "number" ? `${ocrRegionCount} regions` : "",
+    spreadsheetHeaders.length > 0 ? `Spreadsheet ${spreadsheetHeaders.join(", ")}` : "",
+    spreadsheetTableLabel ? `Spreadsheet ${spreadsheetTableLabel}` : "",
     mediaKind ? `Media ${mediaKind}` : "",
+    mediaSegmentCount ? `${mediaSegmentCount} segments` : "",
+    mediaSegmentGroupSize ? `${mediaSegmentGroupSize} grouped segments` : "",
+    mediaSegmentGroupIndex !== undefined ? `Segment group ${mediaSegmentGroupIndex + 1}` : "",
+    mediaChannel ? `Channel ${mediaChannel}` : "",
+    mediaSpeakerCount ? `${mediaSpeakerCount} speakers` : "",
+    mediaDurationLabel ? `Duration ${mediaDurationLabel}` : "",
     transcriptSource ? `Transcript ${transcriptSource}` : "",
     threadTopic ? `Thread ${threadTopic}` : "",
+    threadRootMessageId ? `Thread root ${threadRootMessageId}` : "",
+    threadMessageCount ? `${threadMessageCount} thread messages` : "",
+    replyDepth ? `Reply depth ${replyDepth}` : "",
+    slideNotesText ? "Speaker notes" : "",
+    archiveDepth ? `Archive depth ${archiveDepth}` : "",
+    archiveNestedDepth ? `Archive nested depth ${archiveNestedDepth}` : "",
+    archiveContainerPath ? `Archive container ${archiveContainerPath}` : "",
+    archiveRootName ? `Archive root ${archiveRootName}` : "",
     speaker ? `Speaker ${speaker}` : "",
     from ? `Sender ${from}` : "",
     sentAt ? `Sent ${sentAt}` : ""
@@ -4763,7 +4892,7 @@ var getSectionPathFromSource = (source) => {
   const path = source.structure?.section?.path ?? (Array.isArray(source.metadata?.sectionPath) ? source.metadata.sectionPath.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : []);
   return path.length > 0 ? path : undefined;
 };
-var isBlockAwareContextLabel = (value) => typeof value === "string" && (value.startsWith("PDF ") || value.startsWith("Office "));
+var isBlockAwareContextLabel = (value) => typeof value === "string" && (value.startsWith("PDF ") || value.startsWith("Office ") || value.startsWith("Slide "));
 var getStructuredSectionScoreWeight = (metadata) => {
   if (!metadata) {
     return 1;
@@ -4771,6 +4900,8 @@ var getStructuredSectionScoreWeight = (metadata) => {
   const pdfTextKind = getContextString2(metadata.pdfTextKind);
   const officeBlockKind = getContextString2(metadata.officeBlockKind);
   const sectionKind = getContextString2(metadata.sectionKind);
+  const slideTitle = getContextString2(metadata.slideTitle);
+  const slideNotesText = getContextString2(metadata.slideNotesText);
   if (pdfTextKind === "table_like") {
     return 1.28;
   }
@@ -4780,6 +4911,12 @@ var getStructuredSectionScoreWeight = (metadata) => {
   if (sectionKind === "pdf_block" || sectionKind === "office_block" || officeBlockKind === "paragraph" || pdfTextKind === "paragraph") {
     return 1.12;
   }
+  if (sectionKind === "presentation_slide" && slideNotesText) {
+    return 1.2;
+  }
+  if (sectionKind === "presentation_slide" && slideTitle) {
+    return 1.14;
+  }
   return 1;
 };
 var getStructuredSourceLeadScore = (source) => source.score * getStructuredSectionScoreWeight(source.metadata);
@@ -8560,10 +8697,32 @@ var scoreLoosePhraseMatch2 = (query, text) => {
   return 0;
 };
 var queryHasAnyToken = (queryTokens, candidates) => candidates.some((candidate) => queryTokens.includes(candidate));
+var metadataString = (value) => typeof value === "string" && value.trim().length > 0 ? value.trim().toLowerCase() : undefined;
 var scoreStructuredEvidenceMatch = (queryTokens, result) => {
   const metadata = result.metadata ?? {};
   const pdfTextKind = typeof metadata.pdfTextKind === "string" ? metadata.pdfTextKind : undefined;
   const officeBlockKind = typeof metadata.officeBlockKind === "string" ? metadata.officeBlockKind : undefined;
+  const slideTitle = metadataString(metadata.slideTitle);
+  const slideNotesText = metadataString(metadata.slideNotesText);
+  const threadTopic = metadataString(metadata.threadTopic);
+  const threadRootMessageId = metadataString(metadata.threadRootMessageId);
+  const threadMessageCount = typeof metadata.threadMessageCount === "number" ? metadata.threadMessageCount : undefined;
+  const attachmentName = metadataString(metadata.attachmentName);
+  const archivePath = metadataString(metadata.archivePath);
+  const archiveFullPath = metadataString(metadata.archiveFullPath);
+  const archiveContainerPath = metadataString(metadata.archiveContainerPath);
+  const archiveNestedDepth = typeof metadata.archiveNestedDepth === "number" ? metadata.archiveNestedDepth : undefined;
+  const mediaSpeakerCount = typeof metadata.mediaSpeakerCount === "number" ? metadata.mediaSpeakerCount : undefined;
+  const mediaSegmentCount = typeof metadata.mediaSegmentCount === "number" ? metadata.mediaSegmentCount : undefined;
+  const mediaSegmentGroupSize = typeof metadata.mediaSegmentGroupSize === "number" ? metadata.mediaSegmentGroupSize : undefined;
+  const mediaChannel = metadataString(metadata.mediaChannel);
+  const speaker = metadataString(metadata.speaker);
+  const ocrConfidence = typeof metadata.ocrRegionConfidence === "number" ? metadata.ocrRegionConfidence : typeof metadata.ocrPageAverageConfidence === "number" ? metadata.ocrPageAverageConfidence : typeof metadata.ocrAverageConfidence === "number" ? metadata.ocrAverageConfidence : typeof metadata.ocrConfidence === "number" ? metadata.ocrConfidence : undefined;
+  const isOCREvidence = typeof ocrConfidence === "number" || metadataString(metadata.pdfTextMode) === "ocr";
+  const spreadsheetHeaders = Array.isArray(metadata.spreadsheetHeaders) ? metadata.spreadsheetHeaders.map((value) => metadataString(value)).filter((value) => typeof value === "string") : [];
+  const spreadsheetTableIndex = typeof metadata.spreadsheetTableIndex === "number" ? metadata.spreadsheetTableIndex : undefined;
+  const spreadsheetTableCount = typeof metadata.spreadsheetTableCount === "number" ? metadata.spreadsheetTableCount : undefined;
+  const hasSpreadsheetRows = typeof metadata.spreadsheetRowStart === "number" || typeof metadata.spreadsheetRowEnd === "number";
   const hasBlockMetadata = typeof metadata.pdfBlockNumber === "number" || typeof metadata.officeBlockNumber === "number";
   let score = 0;
   if (hasBlockMetadata) {
@@ -8606,6 +8765,162 @@ var scoreStructuredEvidenceMatch = (queryTokens, result) => {
   ])) {
     score += 0.55;
   }
+  if (spreadsheetHeaders.length > 0 && (queryHasAnyToken(queryTokens, [
+    "sheet",
+    "spreadsheet",
+    "workbook",
+    "column",
+    "columns",
+    "row",
+    "rows"
+  ]) || queryTokens.some((token) => spreadsheetHeaders.some((header) => header.includes(token))))) {
+    score += 0.45;
+  }
+  if (hasSpreadsheetRows && queryHasAnyToken(queryTokens, ["row", "rows", "sheet", "spreadsheet"])) {
+    score += 0.18;
+  }
+  if (typeof spreadsheetTableIndex === "number" && queryHasAnyToken(queryTokens, [
+    "table",
+    "tables",
+    "sheet",
+    "spreadsheet"
+  ])) {
+    score += 0.16;
+    if (typeof spreadsheetTableCount === "number" && spreadsheetTableCount > 1) {
+      score += 0.08;
+    }
+  }
+  if (slideTitle && (queryHasAnyToken(queryTokens, [
+    "slide",
+    "slides",
+    "deck",
+    "presentation"
+  ]) || queryTokens.some((token) => slideTitle.includes(token)))) {
+    score += 0.4;
+  }
+  if (slideNotesText && queryHasAnyToken(queryTokens, [
+    "notes",
+    "speaker",
+    "speakers",
+    "talking"
+  ])) {
+    score += 0.2;
+  }
+  if (speaker && queryHasAnyToken(queryTokens, ["speaker", "speakers", "said", "says"])) {
+    score += 0.22;
+  }
+  if (typeof mediaSpeakerCount === "number" && mediaSpeakerCount > 1 && queryHasAnyToken(queryTokens, [
+    "speaker",
+    "speakers",
+    "conversation",
+    "dialogue"
+  ])) {
+    score += 0.12;
+  }
+  if (typeof mediaSegmentCount === "number" && mediaSegmentCount > 1 && queryHasAnyToken(queryTokens, [
+    "timestamp",
+    "segment",
+    "segments",
+    "audio",
+    "video"
+  ])) {
+    score += 0.08;
+  }
+  if (typeof mediaSegmentGroupSize === "number" && mediaSegmentGroupSize > 1 && queryHasAnyToken(queryTokens, [
+    "segment",
+    "segments",
+    "timestamp",
+    "group",
+    "audio",
+    "video"
+  ])) {
+    score += 0.06;
+  }
+  if (mediaChannel && queryHasAnyToken(queryTokens, [
+    "channel",
+    "channels",
+    "left",
+    "right",
+    "audio",
+    "video"
+  ])) {
+    score += 0.12;
+  }
+  if (threadTopic && (queryHasAnyToken(queryTokens, [
+    "email",
+    "emails",
+    "thread",
+    "reply",
+    "replies",
+    "attachment"
+  ]) || queryTokens.some((token) => threadTopic.includes(token)))) {
+    score += 0.34;
+  }
+  if (threadRootMessageId && queryHasAnyToken(queryTokens, [
+    "thread",
+    "reply",
+    "replies",
+    "root",
+    "email"
+  ])) {
+    score += 0.12;
+  }
+  if (typeof threadMessageCount === "number" && threadMessageCount > 1 && queryHasAnyToken(queryTokens, [
+    "thread",
+    "reply",
+    "replies",
+    "attachment"
+  ])) {
+    score += 0.08;
+  }
+  if (attachmentName && queryHasAnyToken(queryTokens, [
+    "attachment",
+    "attachments",
+    "file",
+    "files"
+  ])) {
+    score += 0.18;
+  }
+  if ((archiveFullPath || archivePath) && (queryHasAnyToken(queryTokens, [
+    "archive",
+    "archives",
+    "entry",
+    "entries",
+    "bundle",
+    "zip"
+  ]) || queryTokens.some((token) => (archiveFullPath ?? archivePath ?? "").includes(token)))) {
+    score += 0.34;
+  }
+  if (archiveContainerPath && queryHasAnyToken(queryTokens, [
+    "nested",
+    "inner",
+    "container",
+    "archive"
+  ])) {
+    score += 0.12;
+  }
+  if (typeof archiveNestedDepth === "number" && archiveNestedDepth > 1 && queryHasAnyToken(queryTokens, ["nested", "inner", "archive"])) {
+    score += 0.08;
+  }
+  if (isOCREvidence && queryHasAnyToken(queryTokens, [
+    "ocr",
+    "scan",
+    "scanned",
+    "image",
+    "photo",
+    "region",
+    "regions",
+    "page",
+    "pages"
+  ])) {
+    if (typeof ocrConfidence === "number" && ocrConfidence >= 0.9) {
+      score += 0.12;
+    } else if (typeof ocrConfidence === "number" && ocrConfidence >= 0.75) {
+      score += 0.05;
+    } else if (typeof ocrConfidence === "number" && ocrConfidence < 0.55) {
+      score -= 0.05;
+    }
+  }
   return score;
 };
 var scoreHeuristicMatch = ({
@@ -8817,6 +9132,62 @@ var formatMediaTimestampForIngest = (value) => {
   const milliseconds = Math.floor(value % 1000);
   return `${String(minutes).padStart(2, "0")}:${String(seconds).padStart(2, "0")}.${String(milliseconds).padStart(3, "0")}`;
 };
+var normalizeMediaSpeaker = (value) => typeof value === "string" && value.trim().length > 0 ? value.trim() : undefined;
+var normalizeMediaChannel = (value) => typeof value === "string" && value.trim().length > 0 ? value.trim() : undefined;
+var buildMediaTimestampBoundary = (segments) => {
+  let startMs;
+  let endMs;
+  for (const segment of segments) {
+    if (typeof segment.startMs === "number" && Number.isFinite(segment.startMs)) {
+      startMs = segment.startMs;
+      break;
+    }
+  }
+  for (let index = segments.length - 1;index >= 0; index--) {
+    const segment = segments[index];
+    if (!segment) {
+      continue;
+    }
+    if (typeof segment.endMs === "number" && Number.isFinite(segment.endMs)) {
+      endMs = segment.endMs;
+      break;
+    }
+  }
+  return { endMs, startMs };
+};
+var groupTranscriptSegments = (segments) => {
+  const groups = [];
+  for (const segment of segments) {
+    if (!segment || typeof segment !== "object") {
+      continue;
+    }
+    const text = normalizeWhitespace(segment.text ?? "");
+    if (!text) {
+      continue;
+    }
+    const speaker = normalizeMediaSpeaker(segment.speaker);
+    const channel = normalizeMediaChannel(segment.channel);
+    const lastGroup = groups.at(-1);
+    if (!lastGroup || lastGroup.speaker !== speaker || lastGroup.channel !== channel) {
+      groups.push({
+        channel,
+        endMs: segment.endMs,
+        segments: [segment],
+        speaker,
+        startMs: segment.startMs
+      });
+      continue;
+    }
+    lastGroup.endMs = typeof segment.endMs === "number" && Number.isFinite(segment.endMs) ? segment.endMs : lastGroup.endMs;
+    lastGroup.segments.push(segment);
+    if (typeof segment.startMs === "number" && Number.isFinite(segment.startMs)) {
+      if (typeof lastGroup.startMs !== "number" || !Number.isFinite(lastGroup.startMs) || segment.startMs < lastGroup.startMs) {
+        lastGroup.startMs = segment.startMs;
+      }
+    }
+  }
+  return groups;
+};
 var decodeHtmlEntities = (value) => {
   let output = value;
   for (const [pattern, replacement] of HTML_ENTITY_REPLACEMENTS) {
@@ -9178,12 +9549,17 @@ var spreadsheetStructureUnits = (value, metadata) => {
     return [];
   }
   const sheetName = typeof metadata?.sheetName === "string" && metadata.sheetName || lines[0].replace(/^Sheet\s+/i, "");
+  const spreadsheetHeaders = Array.isArray(metadata?.sheetHeaders) ? metadata.sheetHeaders.filter((value2) => typeof value2 === "string" && value2.trim().length > 0) : [];
+  const repeatedHeaderRowNumbers = Array.isArray(metadata?.repeatedHeaderRowNumbers) ? metadata.repeatedHeaderRowNumbers.filter((value2) => typeof value2 === "number" && Number.isFinite(value2)) : [];
+  const spreadsheetTableCount = typeof metadata?.sheetTableCount === "number" && Number.isFinite(metadata.sheetTableCount) ? metadata.sheetTableCount : Math.max(repeatedHeaderRowNumbers.length + 1, 1);
   const rowLines = lines.filter((line) => /^Row \d+\./.test(line));
   if (rowLines.length === 0) {
     return [
       {
         sectionDepth: 1,
         sectionKind: "spreadsheet_rows",
+        ...spreadsheetHeaders.length > 0 ? { spreadsheetHeaders } : {},
+        ...spreadsheetTableCount > 1 ? { spreadsheetTableCount, spreadsheetTableIndex: 1 } : {},
         sectionPath: [sheetName],
         sectionTitle: sheetName,
         text: normalizeWhitespace(lines.join(`
@@ -9191,35 +9567,69 @@ var spreadsheetStructureUnits = (value, metadata) => {
       }
     ];
   }
-  const groups = [];
-  let current = [];
+  const tableSegments = [];
+  let currentTableRows = [];
+  let tableIndex = 1;
   for (const row of rowLines) {
-    const candidate = [...current, row].join(`
-`);
-    if (current.length > 0 && candidate.length > DEFAULT_MAX_CHUNK_LENGTH) {
-      groups.push(current);
-      current = [row];
+    const rowNumber = Number(row.match(/^Row (\d+)\./)?.[1] ?? NaN);
+    if (currentTableRows.length > 0 && Number.isFinite(rowNumber) && repeatedHeaderRowNumbers.includes(rowNumber)) {
+      tableSegments.push({ rows: currentTableRows, tableIndex });
+      currentTableRows = [row];
+      tableIndex += 1;
       continue;
     }
-    current.push(row);
+    currentTableRows.push(row);
+  }
+  if (currentTableRows.length > 0) {
+    tableSegments.push({ rows: currentTableRows, tableIndex });
   }
-  if (current.length > 0) {
-    groups.push(current);
+  const groups = [];
+  for (const segment of tableSegments) {
+    let current = [];
+    for (const row of segment.rows) {
+      const candidate = [...current, row].join(`
+`);
+      if (current.length > 0 && candidate.length > DEFAULT_MAX_CHUNK_LENGTH) {
+        groups.push({ rows: current, tableIndex: segment.tableIndex });
+        current = [row];
+        continue;
+      }
+      current.push(row);
+    }
+    if (current.length > 0) {
+      groups.push({ rows: current, tableIndex: segment.tableIndex });
+    }
   }
-  return groups.map((rows) => ({
-    preferredChunkUnits: rows,
-    sectionDepth: 1,
-    sectionKind: "spreadsheet_rows",
-    sectionPath: [sheetName],
-    sectionTitle: sheetName,
-    text: normalizeWhitespace([`Sheet ${sheetName}`, ...rows].join(`
+  return groups.map(({ rows, tableIndex: tableIndex2 }) => {
+    const rowNumbers = rows.map((row) => Number(row.match(/^Row (\d+)\./)?.[1] ?? NaN)).filter((value2) => Number.isFinite(value2));
+    return {
+      preferredChunkUnits: rows,
+      sectionDepth: 1,
+      sectionKind: "spreadsheet_rows",
+      sectionPath: [sheetName],
+      sectionTitle: sheetName,
+      ...spreadsheetHeaders.length > 0 ? { spreadsheetHeaders } : {},
+      ...spreadsheetTableCount > 1 ? { spreadsheetTableCount, spreadsheetTableIndex: tableIndex2 } : {},
+      ...rowNumbers.length > 0 ? {
+        spreadsheetRowEnd: rowNumbers[rowNumbers.length - 1],
+        spreadsheetRowStart: rowNumbers[0]
+      } : {},
+      text: normalizeWhitespace([`Sheet ${sheetName}`, ...rows].join(`
 `))
-  }));
+    };
+  });
 };
 var presentationStructureUnits = (value, metadata) => {
   const slideNumber = typeof metadata?.slideNumber === "number" ? metadata.slideNumber : typeof metadata?.slideIndex === "number" ? metadata.slideIndex + 1 : undefined;
-  const slideLabel = slideNumber ? `Slide ${slideNumber}` : "Slide";
-  const paragraphs = paragraphUnits(value);
+  const slideTitle = typeof metadata?.slideTitle === "string" && metadata.slideTitle.trim() ? metadata.slideTitle.trim() : undefined;
+  const slideBodyText = typeof metadata?.slideBodyText === "string" && metadata.slideBodyText.trim() ? metadata.slideBodyText.trim() : undefined;
+  const slideNotesText = typeof metadata?.slideNotesText === "string" && metadata.slideNotesText.trim() ? metadata.slideNotesText.trim() : undefined;
+  const slideLabel = slideTitle || (slideNumber ? `Slide ${slideNumber}` : "Slide");
+  const paragraphs = [
+    slideTitle,
+    slideBodyText,
+    slideNotesText ? `Speaker notes: ${slideNotesText}` : undefined
+  ].filter((entry) => Boolean(entry)).flatMap((entry) => paragraphUnits(entry));
   return [
     {
       preferredChunkUnits: paragraphs,
@@ -9674,6 +10084,8 @@ var spreadsheetRowText = (row, headers) => {
   });
   return normalizeWhitespace(entries.join(" | "));
 };
+var normalizeSpreadsheetHeaderValue = (value) => normalizeWhitespace(value).toLowerCase();
+var isSpreadsheetHeaderRow = (row, headers) => row.length === headers.length && row.every((cell, index) => normalizeSpreadsheetHeaderValue(cell.value) === normalizeSpreadsheetHeaderValue(headers[index] ?? ""));
 var spreadsheetSheetTexts = (entries) => {
   const sharedStrings = spreadsheetSharedStrings(entries);
   const sheetNames = spreadsheetSheetNames(entries);
@@ -9684,14 +10096,28 @@ var spreadsheetSheetTexts = (entries) => {
       return null;
     }
     const headers = rows[0].map((cell) => cell.value);
-    const rowTexts = rows.map((row, rowIndex) => normalizeWhitespace(`Row ${rowIndex + 1}. ${spreadsheetRowText(row, rowIndex === 0 ? [] : headers)}`));
+    const repeatedHeaderRowNumbers = [];
+    let tableCount = 1;
+    const rowTexts = rows.map((row, rowIndex) => {
+      const rowNumber = rowIndex + 1;
+      const isHeaderRow = rowIndex === 0 || isSpreadsheetHeaderRow(row, headers);
+      if (rowIndex > 0 && isHeaderRow) {
+        repeatedHeaderRowNumbers.push(rowNumber);
+        tableCount += 1;
+      }
+      return normalizeWhitespace(`Row ${rowNumber}. ${spreadsheetRowText(row, isHeaderRow ? [] : headers)}`);
+    });
     const text = normalizeWhitespace(rowTexts.join(`
 `));
     if (!text) {
       return null;
     }
     return {
+      headers,
       name: sheetNames[index] ?? `Sheet ${index + 1}`,
+      repeatedHeaderRowNumbers,
+      rowCount: rowTexts.length,
+      tableCount,
       text
     };
   }).filter((entry) => Boolean(entry));
@@ -9714,12 +10140,21 @@ var presentationNotesByIndex = (entries) => new Map(entries.filter((entry) => en
 var presentationSlides = (entries) => {
   const notesByIndex = presentationNotesByIndex(entries);
   return entries.filter((entry) => entry.path.startsWith("ppt/slides/") && entry.path.endsWith(".xml")).sort((left, right) => left.path.localeCompare(right.path)).map((entry, index) => {
-    const slideText = normalizeWhitespace(extractXmlText(decodeUtf8(entry.data)));
+    const textRuns = [
+      ...decodeUtf8(entry.data).matchAll(/<a:t[^>]*>([\s\S]*?)<\/a:t>/gi)
+    ].map((match) => normalizeWhitespace(decodeHtmlEntities(match[1] ?? ""))).filter(Boolean);
+    const slideTitle = textRuns[0];
+    const slideBodyText = normalizeWhitespace(textRuns.slice(1).join(`
+`));
+    const slideText = normalizeWhitespace([slideTitle, slideBodyText].filter(Boolean).join(`
+`));
     const notesText = notesByIndex.get(index);
     const text = normalizeWhitespace([slideText, notesText ? `Speaker notes: ${notesText}` : ""].filter(Boolean).join(`
 `));
     return {
       index,
+      slideBodyText,
+      slideTitle,
       notesText,
       text
     };
@@ -9864,6 +10299,15 @@ var parseEmailHeaders = (raw) => {
     to: getHeader("To")
   };
 };
+var normalizeEmailThreadKey = (value) => {
+  const normalized = normalizeWhitespace(value?.replace(/^(re|fw|fwd)\s*:\s*/gi, "")?.replace(/[<>]/g, "")?.toLowerCase() ?? "");
+  return normalized || undefined;
+};
+var normalizeEmailMessageId = (value) => {
+  const normalized = normalizeWhitespace(value ?? "");
+  return normalized || undefined;
+};
+var parseEmailReferenceChain = (references) => (references?.match(/<[^>]+>/g) ?? []).map((entry) => normalizeWhitespace(entry)).filter(Boolean);
 var stripRTF = (value) => {
   const withoutBinary = value.replace(/\\bin\d+ [\s\S]*?(?=[\\}])/g, " ");
   const withoutControls = withoutBinary.replace(/\\par[d]?/g, `
@@ -9881,6 +10325,9 @@ var extractPrintableStrings = (data) => {
 };
 var ocrMetadata = (result) => {
   const regions = result.regions?.filter((region) => normalizeWhitespace(region.text ?? "").length > 0);
+  const pageNumbers = [
+    ...new Set((regions ?? []).map((region) => typeof region.page === "number" && region.page > 0 ? region.page : undefined).filter((value) => value !== undefined))
+  ].sort((left, right) => left - right);
   const confidenceValues = [
     typeof result.confidence === "number" ? result.confidence : undefined,
     ...(regions ?? []).map((region) => typeof region.confidence === "number" ? region.confidence : undefined)
@@ -9889,6 +10336,8 @@ var ocrMetadata = (result) => {
   return {
     ...result.metadata ?? {},
     ocrConfidence: result.confidence,
+    ocrPageCount: pageNumbers.length,
+    ocrPageNumbers: pageNumbers,
     ocrRegionCount: regions?.length,
     ocrRegions: regions,
     ocrAverageConfidence: averageConfidence
@@ -9905,25 +10354,32 @@ var ocrPageDocuments = (result, input, baseMetadata) => {
     bucket.push({ ...region, text });
     grouped.set(region.page, bucket);
   }
-  return [...grouped.entries()].sort((left, right) => left[0] - right[0]).map(([pageNumber, regions]) => ({
-    chunking: input.chunking,
-    contentType: input.contentType,
-    format: "text",
-    metadata: {
-      ...input.metadata ?? {},
-      ...baseMetadata,
-      ocrRegionCount: regions.length,
-      ocrRegions: regions,
-      pageNumber,
-      pageIndex: pageNumber - 1,
-      sourceNativeKind: "pdf_page"
-    },
-    source: input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}.pdf`,
-    text: normalizeWhitespace(`PDF page ${pageNumber} from ${input.title ?? input.name ?? input.path ?? DEFAULT_BINARY_NAME}.
+  return [...grouped.entries()].sort((left, right) => left[0] - right[0]).map(([pageNumber, regions]) => {
+    const confidenceValues = regions.map((region) => typeof region.confidence === "number" ? region.confidence : undefined).filter((value) => value !== undefined);
+    const averageConfidence = confidenceValues.length > 0 ? confidenceValues.reduce((sum, value) => sum + value, 0) / confidenceValues.length : undefined;
+    return {
+      chunking: input.chunking,
+      contentType: input.contentType,
+      format: "text",
+      metadata: {
+        ...input.metadata ?? {},
+        ...baseMetadata,
+        ocrPageAverageConfidence: averageConfidence,
+        ocrPageConfidence: averageConfidence,
+        ocrRegionCount: regions.length,
+        ocrRegionNumbers: regions.map((_region, index) => index + 1),
+        ocrRegions: regions,
+        pageNumber,
+        pageIndex: pageNumber - 1,
+        sourceNativeKind: "pdf_page"
+      },
+      source: input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}.pdf`,
+      text: normalizeWhitespace(`PDF page ${pageNumber} from ${input.title ?? input.name ?? input.path ?? DEFAULT_BINARY_NAME}.
 ${regions.map((region) => region.text).join(`
 `)}`),
-    title: input.title ? `${input.title} \xB7 Page ${pageNumber}` : `Page ${pageNumber}`
-  }));
+      title: input.title ? `${input.title} \xB7 Page ${pageNumber}` : `Page ${pageNumber}`
+    };
+  });
 };
 var ocrRegionDocuments = (result, input, baseMetadata) => {
   const documents = [];
@@ -9941,6 +10397,8 @@ var ocrRegionDocuments = (result, input, baseMetadata) => {
       metadata: {
         ...input.metadata ?? {},
         ...baseMetadata,
+        ocrPageCount: 1,
+        ocrPageNumbers: [pageNumber],
         ocrRegionConfidence: region.confidence,
         ocrRegionHeight: region.height,
         ocrRegionWidth: region.width,
@@ -10051,15 +10509,34 @@ var createEmailExtractor = () => ({
     const { body } = splitEmailMessage(raw);
     const parsed = parseEmailMimeParts(body, headers.contentType);
     const source = input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}.eml`;
+    const referenceChain = parseEmailReferenceChain(headers.references);
+    const messageId = normalizeEmailMessageId(headers.messageId);
+    const inReplyTo = normalizeEmailMessageId(headers.inReplyTo);
+    const threadMessageIds = [
+      ...new Set([
+        ...referenceChain.map((entry) => normalizeEmailMessageId(entry)),
+        messageId
+      ].filter((value) => typeof value === "string"))
+    ];
+    const replyDepth = Math.max(referenceChain.length, headers.inReplyTo ? 1 : 0);
+    const threadTopic = headers.threadTopic ?? headers.subject;
+    const threadRootMessageId = normalizeEmailMessageId(referenceChain[0]) ?? inReplyTo ?? messageId;
+    const threadKey = normalizeEmailThreadKey(threadTopic) ?? normalizeEmailThreadKey(messageId) ?? normalizeEmailThreadKey(headers.subject);
     const messageMetadata = {
       ...input.metadata ?? {},
       emailKind: "message",
       fileKind: "email",
       from: headers.from,
-      inReplyTo: headers.inReplyTo,
-      messageId: headers.messageId,
+      inReplyTo,
+      messageId,
       references: headers.references,
-      threadTopic: headers.subject,
+      replyDepth,
+      replyReferenceCount: referenceChain.length,
+      threadMessageCount: threadMessageIds.length,
+      threadMessageIds,
+      threadKey,
+      threadRootMessageId,
+      threadTopic,
       to: headers.to,
       hasAttachments: parsed.attachments.length > 0
     };
@@ -10172,9 +10649,13 @@ var createOfficeDocumentExtractor = () => ({
           ...input.metadata ?? {},
           fileKind: "office",
           ...officeMetadata,
+          repeatedHeaderRowNumbers: sheet.repeatedHeaderRowNumbers,
+          sheetHeaders: sheet.headers,
           sourceNativeKind: "spreadsheet_sheet",
           sheetIndex: index,
-          sheetName: sheet.name
+          sheetName: sheet.name,
+          sheetRowCount: sheet.rowCount,
+          sheetTableCount: sheet.tableCount
         },
         source: input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}${extension || ".office"}`,
         text: normalizeWhitespace(`Spreadsheet workbook ${workbookLabel}. ` + `Worksheet ${index + 1}. ` + `Workbook sheet named ${sheet.name}. ` + `Sheet ${sheet.name} from spreadsheet workbook ${workbookLabel}.` + `
@@ -10195,6 +10676,9 @@ ${sheet.text}`),
           ...input.metadata ?? {},
           fileKind: "office",
           ...officeMetadata,
+          ...slide.slideBodyText ? { slideBodyText: slide.slideBodyText } : {},
+          ...slide.notesText ? { slideNotesText: slide.notesText } : {},
+          ...slide.slideTitle ? { slideTitle: slide.slideTitle } : {},
           sourceNativeKind: "presentation_slide",
           slideIndex: slide.index,
           slideNumber: slide.index + 1
@@ -10240,7 +10724,8 @@ var createRAGImageOCRExtractor = (provider) => ({
       metadata: {
         ...input.metadata ?? {},
         ...ocrMetadata(result),
-        fileKind: "image"
+        fileKind: "image",
+        sourceNativeKind: "image_ocr"
       },
       source: input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}.image.txt`,
       text: result.text,
@@ -10253,15 +10738,34 @@ var createRAGMediaFileExtractor = (transcriber) => ({
   supports: mediaExtractorSupports,
   extract: async (input) => {
     const result = await transcriber.transcribe(input);
+    const rawSegments = (result.segments ?? []).filter((segment) => {
+      if (!segment || typeof segment !== "object") {
+        return false;
+      }
+      return normalizeWhitespace(segment.text ?? "").length > 0;
+    });
+    const segmentGroups = groupTranscriptSegments(rawSegments);
     const source = input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}.media.txt`;
+    const segmentCount = rawSegments.length;
+    const mediaDurationMs = rawSegments.reduce((max, segment) => {
+      const endMs = typeof segment.endMs === "number" ? segment.endMs : undefined;
+      if (typeof endMs !== "number") {
+        return max;
+      }
+      return typeof max === "number" ? Math.max(max, endMs) : endMs;
+    }, undefined);
+    const mediaSpeakers = [
+      ...new Set(rawSegments.map((segment) => normalizeMediaSpeaker(segment.speaker)).filter((value) => typeof value === "string"))
+    ];
     const segmentDocuments = [];
-    for (const [index, segment] of (result.segments ?? []).entries()) {
-      const text = normalizeWhitespace(segment.text ?? "");
-      if (!text) {
+    for (const [index, segmentGroup] of segmentGroups.entries()) {
+      const { endMs, startMs } = buildMediaTimestampBoundary(segmentGroup.segments);
+      const groupText = normalizeWhitespace(segmentGroup.segments.map((segment) => normalizeWhitespace(segment.text ?? "")).filter((value) => value.length > 0).join(" "));
+      if (!groupText) {
         continue;
       }
-      const startMs = typeof segment.startMs === "number" ? segment.startMs : undefined;
-      const endMs = typeof segment.endMs === "number" ? segment.endMs : undefined;
+      const mediaSegmentStartMs = startMs;
+      const mediaSegmentEndMs = endMs;
       const startLabel = formatMediaTimestampForIngest(startMs);
       const endLabel = formatMediaTimestampForIngest(endMs);
       const mediaKind = typeof result.metadata?.mediaKind === "string" ? result.metadata.mediaKind : "media";
@@ -10274,15 +10778,27 @@ var createRAGMediaFileExtractor = (transcriber) => ({
           ...result.metadata ?? {},
           fileKind: "media",
           sourceNativeKind: "media_segment",
+          mediaDurationMs,
           mediaSegmentIndex: index,
-          mediaSegmentStartMs: startMs,
-          mediaSegmentEndMs: endMs,
-          mediaSegments: [segment],
-          speaker: typeof segment.speaker === "string" ? segment.speaker : undefined
+          mediaSegmentStartMs,
+          mediaSegmentEndMs,
+          mediaSegmentCount: segmentCount,
+          mediaSegmentGroupIndex: index,
+          mediaSegmentGroupSize: segmentGroup.segments.length,
+          mediaSegmentGroupSpeaker: segmentGroup.speaker,
+          mediaChannel: segmentGroup.channel,
+          mediaSegments: segmentGroup.segments,
+          startMs: mediaSegmentStartMs,
+          endMs: mediaSegmentEndMs,
+          ...mediaSpeakers.length > 0 ? {
+            mediaSpeakerCount: mediaSpeakers.length,
+            mediaSpeakers
+          } : {},
+          speaker: segmentGroup.speaker
         },
         source,
         text: normalizeWhitespace(`${mediaKind} transcript segment${startLabel ? ` at timestamp ${startLabel}${endLabel ? ` to ${endLabel}` : ""}` : ""} from ${input.title ?? input.name ?? input.path ?? DEFAULT_BINARY_NAME}. ` + `${mediaKind} timestamp evidence${startLabel ? ` ${startLabel}${endLabel ? ` to ${endLabel}` : ""}` : ""}.` + `
-${text}`),
+${groupText}`),
         title: input.title ? `${input.title} \xB7 ${mediaKind[0]?.toUpperCase() + mediaKind.slice(1)} segment ${index + 1}` : `${mediaKind[0]?.toUpperCase() + mediaKind.slice(1)} segment ${index + 1}`
       });
     }
@@ -10294,7 +10810,13 @@ ${text}`),
         ...input.metadata ?? {},
         ...result.metadata ?? {},
         fileKind: "media",
-        mediaSegments: result.segments
+        mediaDurationMs,
+        mediaSegmentCount: segmentCount,
+        mediaSegments: rawSegments,
+        ...mediaSpeakers.length > 0 ? {
+          mediaSpeakerCount: mediaSpeakers.length,
+          mediaSpeakers
+        } : {}
       },
       source,
       text: result.text,
@@ -10319,6 +10841,13 @@ var createTextFileExtractor = () => ({
   })
 });
 var expandArchiveEntry = async (entry, archiveInput, extractors, registry) => {
+  const parentArchiveLineage = Array.isArray(archiveInput.metadata?.archiveLineage) ? archiveInput.metadata.archiveLineage.filter((value) => typeof value === "string" && value.trim().length > 0) : [];
+  const entryArchiveLineage = entry.path.split(/[\\/]/).map((segment) => normalizeWhitespace(segment)).filter(Boolean);
+  const archiveLineage = [...parentArchiveLineage, ...entryArchiveLineage];
+  const parentArchivePath = typeof archiveInput.metadata?.archivePath === "string" && archiveInput.metadata.archivePath.trim().length > 0 ? archiveInput.metadata.archivePath.trim() : undefined;
+  const archiveFullPath = parentArchivePath ? `${parentArchivePath}!${entry.path}` : entry.path;
+  const archiveRootName = (typeof archiveInput.metadata?.archiveRootName === "string" && archiveInput.metadata.archiveRootName.trim().length > 0 ? archiveInput.metadata.archiveRootName.trim() : undefined) ?? archiveInput.name ?? archiveInput.path?.split(/[/\\]/).pop() ?? archiveInput.source;
+  const archiveRootSource = (typeof archiveInput.metadata?.archiveRootSource === "string" && archiveInput.metadata.archiveRootSource.trim().length > 0 ? archiveInput.metadata.archiveRootSource.trim() : undefined) ?? archiveInput.source ?? archiveInput.path ?? archiveInput.name;
   const documents = await extractRAGFileDocuments({
     chunking: archiveInput.chunking,
     contentType: entry.contentType,
@@ -10331,7 +10860,14 @@ var expandArchiveEntry = async (entry, archiveInput, extractors, registry) => {
       archiveEntryName: basename(entry.path),
       archiveParentName: archiveInput.name ?? archiveInput.path?.split(/[/\\]/).pop() ?? archiveInput.source,
       archiveParentSource: archiveInput.source ?? archiveInput.path ?? archiveInput.name,
+      archiveContainerPath: parentArchivePath,
+      archiveDepth: archiveLineage.length,
+      archiveFullPath,
+      archiveLineage,
       archivePath: entry.path,
+      archiveRootName,
+      archiveRootSource,
+      archiveNestedDepth: parentArchiveLineage.length + 1,
       fileKind: "archive_entry"
     },
     name: basename(entry.path),
@@ -10748,10 +11284,27 @@ ${text}`);
     }
     return text;
   };
-  return merged.map((text) => ({
-    ...unit,
-    text: decorateSourceAwareChunkText(text)
-  }));
+  const resolveSpreadsheetChunkRowRange = (text) => {
+    if (unit.sectionKind !== "spreadsheet_rows") {
+      return {};
+    }
+    const rowNumbers = [...text.matchAll(/^Row (\d+)\./gm)].map((match) => Number(match[1] ?? NaN)).filter((value) => Number.isFinite(value));
+    if (rowNumbers.length === 0) {
+      return {};
+    }
+    return {
+      spreadsheetRowEnd: rowNumbers[rowNumbers.length - 1],
+      spreadsheetRowStart: rowNumbers[0]
+    };
+  };
+  return merged.map((text) => {
+    const decoratedText = decorateSourceAwareChunkText(text);
+    return {
+      ...unit,
+      ...resolveSpreadsheetChunkRowRange(decoratedText),
+      text: decoratedText
+    };
+  });
 };
 var resolveChunkingUnits = (text, options) => {
   if (options.strategy === "fixed") {
@@ -10922,6 +11475,11 @@ var prepareRAGDocument = (document, defaultChunking, chunkingRegistry) => {
         ...sectionTitle ? { sectionTitle } : {},
         ...sectionPath && sectionPath.length > 0 ? { sectionPath } : {},
         ...typeof entry.sectionDepth === "number" ? { sectionDepth: entry.sectionDepth } : {},
+        ...Array.isArray(entry.spreadsheetHeaders) && entry.spreadsheetHeaders.length > 0 ? { spreadsheetHeaders: entry.spreadsheetHeaders } : {},
+        ...typeof entry.spreadsheetTableIndex === "number" ? { spreadsheetTableIndex: entry.spreadsheetTableIndex } : {},
+        ...typeof entry.spreadsheetTableCount === "number" ? { spreadsheetTableCount: entry.spreadsheetTableCount } : {},
+        ...typeof entry.spreadsheetRowStart === "number" ? { spreadsheetRowStart: entry.spreadsheetRowStart } : {},
+        ...typeof entry.spreadsheetRowEnd === "number" ? { spreadsheetRowEnd: entry.spreadsheetRowEnd } : {},
         ...typeof entry.pageNumber === "number" ? { pageNumber: entry.pageNumber } : {},
         ...typeof entry.officeBlockNumber === "number" ? { officeBlockNumber: entry.officeBlockNumber } : {},
         ...entry.officeBlockKind ? { officeBlockKind: entry.officeBlockKind } : {},
@@ -24278,5 +24836,5 @@ export {
   aiChat
 };
 
-//# debugId=23520EDE705830A964756E2164756E21
+//# debugId=AFAF0A5BC1AB4BC864756E2164756E21
 //# sourceMappingURL=index.js.map