@absolutejs/absolute 0.19.0-beta.618 → 0.19.0-beta.619

package/dist/ai/client/index.js

@@ -4834,6 +4834,25 @@ var buildContextLabel2 = (metadata) => {
   if (!metadata) {
     return;
   }
+  const pdfTextKind = getContextString2(metadata.pdfTextKind);
+  const officeBlockKind = getContextString2(metadata.officeBlockKind);
+  const sectionPath = Array.isArray(metadata.sectionPath) ? metadata.sectionPath.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : [];
+  const sectionTitle = getContextString2(metadata.sectionTitle) ?? sectionPath.at(-1);
+  if (pdfTextKind === "table_like" && sectionTitle) {
+    return `PDF table block ${sectionTitle}`;
+  }
+  if (pdfTextKind === "paragraph" && sectionTitle) {
+    return `PDF text block ${sectionTitle}`;
+  }
+  if (officeBlockKind === "table" && sectionTitle) {
+    return `Office table block ${sectionTitle}`;
+  }
+  if (officeBlockKind === "list" && sectionTitle) {
+    return `Office list block ${sectionTitle}`;
+  }
+  if (officeBlockKind === "paragraph" && sectionTitle) {
+    return `Office paragraph block ${sectionTitle}`;
+  }
   const emailKind = getContextString2(metadata.emailKind);
   if (emailKind === "attachment") {
     return "Attachment evidence";
@@ -4870,8 +4889,6 @@ var buildContextLabel2 = (metadata) => {
   if (speaker) {
     return `Speaker ${speaker}`;
   }
-  const sectionPath = Array.isArray(metadata.sectionPath) ? metadata.sectionPath.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : [];
-  const sectionTitle = getContextString2(metadata.sectionTitle) ?? sectionPath.at(-1);
   if (sectionTitle) {
     return `Section ${sectionTitle}`;
   }
@@ -4881,11 +4898,21 @@ var buildLocatorLabel2 = (metadata, source, title) => {
   if (!metadata) {
     return;
   }
+  const pdfTextKind = getContextString2(metadata.pdfTextKind);
+  const officeBlockKind = getContextString2(metadata.officeBlockKind);
+  const pdfBlockNumber = getContextNumber2(metadata.pdfBlockNumber);
+  const officeBlockNumber = getContextNumber2(metadata.officeBlockNumber);
   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) {
     return `Page ${page} \xB7 Region ${region}`;
   }
+  if (page && pdfBlockNumber && pdfTextKind === "table_like") {
+    return `Page ${page} \xB7 Table Block ${pdfBlockNumber}`;
+  }
+  if (page && pdfBlockNumber) {
+    return `Page ${page} \xB7 Text Block ${pdfBlockNumber}`;
+  }
   if (page) {
     return `Page ${page}`;
   }
@@ -4914,6 +4941,15 @@ var buildLocatorLabel2 = (metadata, source, title) => {
   if (mediaStart) {
     return `Timestamp ${mediaStart}`;
   }
+  if (officeBlockNumber && officeBlockKind === "table") {
+    return `Office table block ${officeBlockNumber}`;
+  }
+  if (officeBlockNumber && officeBlockKind === "list") {
+    return `Office list block ${officeBlockNumber}`;
+  }
+  if (officeBlockNumber && officeBlockKind === "paragraph") {
+    return `Office paragraph block ${officeBlockNumber}`;
+  }
   const sectionPath = Array.isArray(metadata.sectionPath) ? metadata.sectionPath.map((value) => getContextString2(value)).filter((value) => typeof value === "string") : [];
   if (sectionPath.length > 0) {
     return `Section ${sectionPath.join(" > ")}`;
@@ -4931,12 +4967,16 @@ var buildProvenanceLabel2 = (metadata) => {
   const mediaKind = getContextString2(metadata.mediaKind);
   const transcriptSource = getContextString2(metadata.transcriptSource);
   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 ocrConfidence = getContextNumber2(metadata.ocrRegionConfidence) ?? getContextNumber2(metadata.ocrConfidence);
   const labels = [
     pdfTextMode ? `PDF ${pdfTextMode}` : "",
+    pdfTextKind === "table_like" ? "PDF table block" : pdfTextKind === "paragraph" ? "PDF text block" : "",
+    officeBlockKind ? `Office ${officeBlockKind}` : "",
     ocrEngine ? `OCR ${ocrEngine}` : "",
     extractorRegistryMatch ? `Extractor ${extractorRegistryMatch}` : "",
     chunkingProfile ? `Chunking ${chunkingProfile}` : "",
@@ -4972,7 +5012,7 @@ var buildRAGChunkStructure = (metadata) => {
     return;
   }
   const sectionPath = Array.isArray(metadata.sectionPath) ? metadata.sectionPath.filter((value) => typeof value === "string" && value.trim().length > 0) : undefined;
-  const sectionKind = metadata.sectionKind === "markdown_heading" || metadata.sectionKind === "html_heading" || metadata.sectionKind === "office_heading" || metadata.sectionKind === "spreadsheet_rows" || metadata.sectionKind === "presentation_slide" ? metadata.sectionKind : undefined;
+  const sectionKind = metadata.sectionKind === "markdown_heading" || metadata.sectionKind === "html_heading" || metadata.sectionKind === "office_heading" || metadata.sectionKind === "office_block" || metadata.sectionKind === "pdf_block" || metadata.sectionKind === "spreadsheet_rows" || metadata.sectionKind === "presentation_slide" ? metadata.sectionKind : undefined;
   const section = {
     depth: getContextNumber2(metadata.sectionDepth),
     kind: sectionKind,
@@ -5292,7 +5332,7 @@ var buildRAGSourceSummaries = (sources) => {
   const citationReferenceMap = buildRAGCitationReferenceMap(citations);
   return sourceGroups.map((group) => {
     const groupCitations = citations.filter((citation) => group.chunks.some((chunk) => chunk.chunkId === citation.chunkId));
-    const leadChunk = group.chunks.slice().sort((left, right) => right.score - left.score)[0];
+    const leadChunk = getPreferredSourceLeadChunk(group.chunks);
     const excerpts = leadChunk ? buildRAGChunkExcerpts(group.chunks, leadChunk.chunkId) : undefined;
     const structure = leadChunk?.structure ?? buildRAGChunkStructure(leadChunk?.metadata);
     const excerptSelection = buildRAGExcerptSelection(excerpts, structure);
@@ -5320,13 +5360,45 @@ 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 getStructuredSectionScoreWeight = (metadata) => {
+  if (!metadata) {
+    return 1;
+  }
+  const pdfTextKind = getContextString2(metadata.pdfTextKind);
+  const officeBlockKind = getContextString2(metadata.officeBlockKind);
+  const sectionKind = getContextString2(metadata.sectionKind);
+  if (pdfTextKind === "table_like") {
+    return 1.28;
+  }
+  if (officeBlockKind === "table" || officeBlockKind === "list") {
+    return 1.24;
+  }
+  if (sectionKind === "pdf_block" || sectionKind === "office_block" || officeBlockKind === "paragraph" || pdfTextKind === "paragraph") {
+    return 1.12;
+  }
+  return 1;
+};
+var getStructuredSourceLeadScore = (source) => source.score * getStructuredSectionScoreWeight(source.metadata);
+var getPreferredSourceLeadChunk = (chunks) => chunks.slice().sort((left, right) => {
+  const leftWeightedScore = getStructuredSourceLeadScore(left);
+  const rightWeightedScore = getStructuredSourceLeadScore(right);
+  if (rightWeightedScore !== leftWeightedScore) {
+    return rightWeightedScore - leftWeightedScore;
+  }
+  if (right.score !== left.score) {
+    return right.score - left.score;
+  }
+  return left.chunkId.localeCompare(right.chunkId);
+})[0];
 var buildRAGSectionRetrievalDiagnostics = (sources, trace) => {
-  const totalScore = sources.reduce((sum, source) => sum + source.score, 0);
+  const totalScore = sources.reduce((sum, source) => sum + source.score * getStructuredSectionScoreWeight(source.metadata), 0);
   if (sources.length === 0 || totalScore <= 0) {
     return [];
   }
   const sections = new Map;
   for (const source of sources) {
+    const structuredScore = source.score * getStructuredSectionScoreWeight(source.metadata);
     const path = getSectionPathFromSource(source);
     if (!path) {
       continue;
@@ -5358,7 +5430,7 @@ var buildRAGSectionRetrievalDiagnostics = (sources, trace) => {
         sourceSet: new Set(source.source ? [source.source] : []),
         topChunkId: source.chunkId,
         topSource: source.source,
-        totalScore: source.score,
+        totalScore: structuredScore,
         transformedHits,
         variantHits,
         vectorHits
@@ -5366,7 +5438,7 @@ var buildRAGSectionRetrievalDiagnostics = (sources, trace) => {
       continue;
     }
     existing.count += 1;
-    existing.totalScore += source.score;
+    existing.totalScore += structuredScore;
     if (source.source) {
       existing.sourceSet.add(source.source);
     }
@@ -5394,6 +5466,8 @@ var buildRAGSectionRetrievalDiagnostics = (sources, trace) => {
     const parentTotal = siblingPool.reduce((sum, entry) => sum + entry.totalScore, 0);
     const scoreShare = section.totalScore / totalScore;
     const parentShare = parentTotal > 0 ? section.totalScore / parentTotal : undefined;
+    const topChunk = sources.find((source) => source.chunkId === section.topChunkId);
+    const topContextLabel = topChunk?.labels?.contextLabel ?? buildContextLabel2(topChunk?.metadata);
     const parentDistribution = parentTotal > 0 ? siblingPool.map((entry) => ({
       count: entry.count,
       isActive: entry.key === section.key,
@@ -5519,6 +5593,7 @@ var buildRAGSectionRetrievalDiagnostics = (sources, trace) => {
       reasons.push("concentrated_evidence");
     }
     const summaryParts = [
+      isBlockAwareContextLabel(topContextLabel) ? topContextLabel : "",
       `${section.count} hit${section.count === 1 ? "" : "s"}`,
       `${(scoreShare * 100).toFixed(0)}% score share`,
       `vector ${section.vectorHits} \xB7 lexical ${section.lexicalHits} \xB7 hybrid ${section.hybridHits}`,
@@ -5730,22 +5805,21 @@ var updateSourceGroup = (groups, source) => {
     groups.set(key, buildSourceGroup(source, key));
     return;
   }
-  if (source.score > existing.bestScore) {
-    existing.bestScore = source.score;
-    existing.label = buildSourceLabel2(source);
-    existing.labels = source.labels ?? buildRAGSourceLabels({
-      metadata: source.metadata,
-      source: source.source,
-      title: source.title
-    });
-    existing.structure = source.structure ?? buildRAGChunkStructure(source.metadata);
-    existing.source = source.source;
-    existing.title = source.title;
-  } else {
-    existing.bestScore = Math.max(existing.bestScore, source.score);
-  }
+  existing.bestScore = Math.max(existing.bestScore, source.score);
   existing.count += 1;
   existing.chunks.push(source);
+  const leadChunk = getPreferredSourceLeadChunk(existing.chunks);
+  if (leadChunk) {
+    existing.label = buildSourceLabel2(leadChunk);
+    existing.labels = leadChunk.labels ?? buildRAGSourceLabels({
+      metadata: leadChunk.metadata,
+      source: leadChunk.source,
+      title: leadChunk.title
+    });
+    existing.structure = leadChunk.structure ?? buildRAGChunkStructure(leadChunk.metadata);
+    existing.source = leadChunk.source;
+    existing.title = leadChunk.title;
+  }
 };
 var getLatestAssistantMessage = (messages) => {
   for (let index = messages.length - 1;index >= 0; index -= 1) {
@@ -7518,5 +7592,5 @@ export {
   buildRAGEvaluationLeaderboard
 };
 
-//# debugId=3CD9D6700B3CA41764756E2164756E21
+//# debugId=EFA1619967730BA064756E2164756E21
 //# sourceMappingURL=index.js.map