@absolutejs/absolute 0.19.0-beta.551 → 0.19.0-beta.553

package/dist/ai/index.js

@@ -2551,6 +2551,7 @@ var resolveRAGHybridSearchOptions = (retrieval) => {
       fusionConstant: DEFAULT_FUSION_CONSTANT,
       lexicalTopK: undefined,
       lexicalWeight: 2,
+      sourceBalanceStrategy: "cap",
       mode: "vector",
       vectorWeight: 1
     };
@@ -2561,6 +2562,7 @@ var resolveRAGHybridSearchOptions = (retrieval) => {
       fusionConstant: DEFAULT_FUSION_CONSTANT,
       lexicalTopK: undefined,
       lexicalWeight: 2,
+      sourceBalanceStrategy: "cap",
       mode: retrieval,
       vectorWeight: 1
     };
@@ -2569,8 +2571,10 @@ var resolveRAGHybridSearchOptions = (retrieval) => {
     fusion: retrieval.fusion ?? "rrf",
     fusionConstant: Math.max(1, Math.floor(retrieval.fusionConstant ?? DEFAULT_FUSION_CONSTANT)),
     lexicalTopK: retrieval.lexicalTopK,
+    maxResultsPerSource: typeof retrieval.maxResultsPerSource === "number" ? Math.max(1, Math.floor(retrieval.maxResultsPerSource)) : undefined,
     lexicalWeight: Math.max(0, retrieval.lexicalWeight ?? 2),
     mode: retrieval.mode ?? "vector",
+    sourceBalanceStrategy: retrieval.sourceBalanceStrategy === "round_robin" ? "round_robin" : "cap",
     vectorWeight: Math.max(0, retrieval.vectorWeight ?? 1)
   };
 };
@@ -4314,6 +4318,64 @@ var mergeQueryResults = (results) => {
     return left.chunkId.localeCompare(right.chunkId);
   });
 };
+var getRAGSourceDiversityKey = (result) => {
+  const documentId = typeof result.metadata?.documentId === "string" ? result.metadata.documentId : undefined;
+  return result.source ?? documentId ?? result.title ?? result.chunkId;
+};
+var applyRAGSourceDiversity = (results, maxResultsPerSource, strategy = "cap") => {
+  if (typeof maxResultsPerSource !== "number") {
+    return results;
+  }
+  if (strategy === "round_robin") {
+    const grouped = new Map;
+    const order = [];
+    for (const result of results) {
+      const key = getRAGSourceDiversityKey(result);
+      const entries = grouped.get(key);
+      if (entries) {
+        entries.push(result);
+        continue;
+      }
+      grouped.set(key, [result]);
+      order.push(key);
+    }
+    const counts2 = new Map;
+    const balanced = [];
+    let remaining = true;
+    while (remaining) {
+      remaining = false;
+      for (const key of order) {
+        const entries = grouped.get(key);
+        if (!entries || entries.length === 0) {
+          continue;
+        }
+        if ((counts2.get(key) ?? 0) >= maxResultsPerSource) {
+          continue;
+        }
+        const next = entries.shift();
+        if (!next) {
+          continue;
+        }
+        counts2.set(key, (counts2.get(key) ?? 0) + 1);
+        balanced.push(next);
+        remaining = true;
+      }
+    }
+    return balanced;
+  }
+  const limited = [];
+  const counts = new Map;
+  for (const result of results) {
+    const key = getRAGSourceDiversityKey(result);
+    const count = counts.get(key) ?? 0;
+    if (count >= maxResultsPerSource) {
+      continue;
+    }
+    counts.set(key, count + 1);
+    limited.push(result);
+  }
+  return limited;
+};
 var weightQueryResults = (results, queryIndex) => {
   if (queryIndex === 0) {
     return results;
@@ -4369,9 +4431,11 @@ var createRAGCollection = (options) => {
           candidateTopK,
           hasQueryTransform,
           hasReranker,
+          maxResultsPerSource: retrieval.maxResultsPerSource ?? null,
           mode: retrieval.mode,
           runLexical,
           runVector,
+          sourceBalanceStrategy: retrieval.sourceBalanceStrategy,
           topK
         },
         stage: "input"
@@ -4486,7 +4550,19 @@ var createRAGCollection = (options) => {
       },
       stage: "rerank"
     });
-    const limited = reranked.slice(0, topK);
+    const diversified = applyRAGSourceDiversity(reranked, retrieval.maxResultsPerSource, retrieval.sourceBalanceStrategy);
+    if (typeof retrieval.maxResultsPerSource === "number") {
+      steps.push({
+        count: diversified.length,
+        label: "Balanced candidates across sources",
+        metadata: {
+          maxResultsPerSource: retrieval.maxResultsPerSource,
+          strategy: retrieval.sourceBalanceStrategy
+        },
+        stage: "source_balance"
+      });
+    }
+    const limited = diversified.slice(0, topK);
     if (typeof input.scoreThreshold !== "number") {
       steps.push({
         count: limited.length,
@@ -4501,17 +4577,19 @@ var createRAGCollection = (options) => {
         trace: {
           candidateTopK,
           lexicalTopK,
+          maxResultsPerSource: retrieval.maxResultsPerSource,
           mode: retrieval.mode,
           query: input.query,
           resultCounts: {
             final: limited.length,
             fused: results.length,
             lexical: lexicalResults.length,
-            reranked: reranked.length,
+            reranked: diversified.length,
             vector: vectorResults.length
           },
           runLexical,
           runVector,
+          sourceBalanceStrategy: retrieval.sourceBalanceStrategy,
           steps,
           topK,
           transformedQuery: transformed.query,
@@ -4542,18 +4620,20 @@ var createRAGCollection = (options) => {
       trace: {
         candidateTopK,
         lexicalTopK,
+        maxResultsPerSource: retrieval.maxResultsPerSource,
         mode: retrieval.mode,
         query: input.query,
         resultCounts: {
           final: filtered.length,
           fused: results.length,
           lexical: lexicalResults.length,
-          reranked: reranked.length,
+          reranked: diversified.length,
           vector: vectorResults.length
         },
         runLexical,
         runVector,
         scoreThreshold,
+        sourceBalanceStrategy: retrieval.sourceBalanceStrategy,
         steps,
         topK,
         transformedQuery: transformed.query,
@@ -12505,5 +12585,5 @@ export {
   aiChat
 };
 
-//# debugId=7DAB512A9E25D04564756E2164756E21
+//# debugId=BD4F3B903BF6366564756E2164756E21
 //# sourceMappingURL=index.js.map