@librechat/agents 2.4.31 → 2.4.34

package/dist/cjs/tools/search/format.cjs

@@ -1,115 +1,202 @@
 'use strict';
 
-function formatResultsForLLM(results) {
-    let output = '';
+var utils = require('./utils.cjs');
+
+function addHighlightSection() {
+    return ['\n## Highlights', ''];
+}
+// Helper function to format a source (organic or top story)
+function formatSource(source, index, turn, sourceType, references) {
+    /** Array of all lines to include in the output */
+    const outputLines = [];
+    // Add the title
+    outputLines.push(`# ${sourceType.charAt(0).toUpperCase() + sourceType.slice(1)} ${index}: ${source.title != null && source.title ? `"${source.title}"` : '(no title)'}`);
+    outputLines.push(`\nAnchor: \\ue202turn${turn}${sourceType}${index}`);
+    outputLines.push(`URL: ${source.link}`);
+    // Add optional fields
+    if ('snippet' in source && source.snippet != null) {
+        outputLines.push(`Summary: ${source.snippet}`);
+    }
+    if (source.date != null) {
+        outputLines.push(`Date: ${source.date}`);
+    }
+    if (source.attribution != null) {
+        outputLines.push(`Source: ${source.attribution}`);
+    }
+    // Add highlight section or empty line
+    if ((source.highlights?.length ?? 0) > 0) {
+        outputLines.push(...addHighlightSection());
+    }
+    else {
+        outputLines.push('');
+    }
+    // Process highlights if they exist
+    (source.highlights ?? [])
+        .filter((h) => h.text.trim().length > 0)
+        .forEach((h, hIndex) => {
+        outputLines.push(`### Highlight ${hIndex + 1} [Relevance: ${h.score.toFixed(2)}]`);
+        outputLines.push('');
+        outputLines.push('```text');
+        outputLines.push(h.text.trim());
+        outputLines.push('```');
+        outputLines.push('');
+        if (h.references != null && h.references.length) {
+            let hasHeader = false;
+            const refLines = [];
+            for (let j = 0; j < h.references.length; j++) {
+                const ref = h.references[j];
+                if (ref.reference.originalUrl.includes('mailto:')) {
+                    continue;
+                }
+                references.push({
+                    type: ref.type,
+                    link: ref.reference.originalUrl,
+                    attribution: utils.getDomainName(ref.reference.originalUrl),
+                    title: (((ref.reference.title ?? '') || ref.reference.text) ??
+                        '').split('\n')[0],
+                });
+                if (ref.type !== 'link') {
+                    continue;
+                }
+                if (utils.fileExtRegex.test(ref.reference.originalUrl)) {
+                    continue;
+                }
+                if (!hasHeader) {
+                    refLines.push('Core References:');
+                    hasHeader = true;
+                }
+                refLines.push(`- ${ref.type}#${ref.originalIndex + 1}: ${ref.reference.originalUrl}`);
+                refLines.push(`\t- Anchor: \\ue202turn${turn}ref${references.length - 1}`);
+            }
+            if (hasHeader) {
+                outputLines.push(...refLines);
+                outputLines.push('');
+            }
+        }
+        if (hIndex < (source.highlights?.length ?? 0) - 1) {
+            outputLines.push('---');
+            outputLines.push('');
+        }
+    });
+    outputLines.push('');
+    return outputLines.join('\n');
+}
+function formatResultsForLLM(turn, results) {
+    /** Array to collect all output lines */
+    const outputLines = [];
     const addSection = (title) => {
-        output += `\n=== ${title} ===\n`;
+        outputLines.push('');
+        outputLines.push(`=== ${title} ===`);
+        outputLines.push('');
     };
+    const references = [];
     // Organic (web) results
-    const organic = results.organic ?? [];
-    if (organic.length) {
-        addSection('Web Results');
-        organic.forEach((r, i) => {
-            output += [
-                `Source ${i + 1}: ${r.title ?? '(no title)'}`,
-                `Citation Anchor: \\ue202turn0search${i + 1}`,
-                `URL: ${r.link}`,
-                r.snippet != null ? `Summary: ${r.snippet}` : '',
-                r.date != null ? `Date: ${r.date}` : '',
-                r.attribution != null ? `Source: ${r.attribution}` : '',
-                '',
-                '--- Content Highlights ---',
-                ...(r.highlights ?? [])
-                    .filter((h) => h.text.trim().length > 0)
-                    .map((h) => `[Relevance: ${h.score.toFixed(2)}]\n${h.text.trim()}`),
-                '',
-            ]
-                .filter(Boolean)
-                .join('\n');
-        });
+    if (results.organic?.length != null && results.organic.length > 0) {
+        addSection(`Web Results, Turn ${turn}`);
+        for (let i = 0; i < results.organic.length; i++) {
+            const r = results.organic[i];
+            outputLines.push(formatSource(r, i, turn, 'search', references));
+            delete results.organic[i].highlights;
+        }
+    }
+    // Top stories (news)
+    const topStories = results.topStories ?? [];
+    if (topStories.length) {
+        addSection('News Results');
+        for (let i = 0; i < topStories.length; i++) {
+            const r = topStories[i];
+            outputLines.push(formatSource(r, i, turn, 'news', references));
+            if (results.topStories?.[i]?.highlights) {
+                delete results.topStories[i].highlights;
+            }
+        }
     }
-    // Ignoring these sections for now
-    // // Top stories (news)
-    // const topStores = results.topStories ?? [];
-    // if (topStores.length) {
-    //   addSection('News Results');
-    //   topStores.forEach((r, i) => {
-    //     output += [
-    //       `Anchor: \ue202turn0news${i + 1}`,
-    //       `Title: ${r.title ?? '(no title)'}`,
-    //       `URL: ${r.link}`,
-    //       r.snippet != null ? `Snippet: ${r.snippet}` : '',
-    //       r.date != null ? `Date: ${r.date}` : '',
-    //       r.attribution != null ? `Source: ${r.attribution}` : '',
-    //       ''
-    //     ].filter(Boolean).join('\n');
-    //   });
-    // }
     // // Images
     // const images = results.images ?? [];
     // if (images.length) {
     //   addSection('Image Results');
-    //   images.forEach((img, i) => {
-    //     output += [
-    //       `Anchor: \ue202turn0image${i + 1}`,
-    //       `Title: ${img.title ?? '(no title)'}`,
-    //       `Image URL: ${img.imageUrl}`,
-    //       ''
-    //     ].join('\n');
-    //   });
+    //   const imageLines = images.map((img, i) => [
+    //     `Anchor: \ue202turn0image${i}`,
+    //     `Title: ${img.title ?? '(no title)'}`,
+    //     `Image URL: ${img.imageUrl}`,
+    //     ''
+    //   ].join('\n'));
+    //   outputLines.push(imageLines.join('\n'));
     // }
     // Knowledge Graph
     if (results.knowledgeGraph != null) {
         addSection('Knowledge Graph');
-        output += [
-            `Title: ${results.knowledgeGraph.title ?? '(no title)'}`,
+        const kgLines = [
+            `**Title:** ${results.knowledgeGraph.title ?? '(no title)'}`,
+            results.knowledgeGraph.type != null
+                ? `**Type:** ${results.knowledgeGraph.type}`
+                : '',
             results.knowledgeGraph.description != null
-                ? `Description: ${results.knowledgeGraph.description}`
+                ? `**Description:** ${results.knowledgeGraph.description}`
                 : '',
-            results.knowledgeGraph.type != null
-                ? `Type: ${results.knowledgeGraph.type}`
+            results.knowledgeGraph.descriptionSource != null
+                ? `**Description Source:** ${results.knowledgeGraph.descriptionSource}`
+                : '',
+            results.knowledgeGraph.descriptionLink != null
+                ? `**Description Link:** ${results.knowledgeGraph.descriptionLink}`
                 : '',
             results.knowledgeGraph.imageUrl != null
-                ? `Image URL: ${results.knowledgeGraph.imageUrl}`
+                ? `**Image URL:** ${results.knowledgeGraph.imageUrl}`
+                : '',
+            results.knowledgeGraph.website != null
+                ? `**Website:** ${results.knowledgeGraph.website}`
                 : '',
             results.knowledgeGraph.attributes != null
-                ? `Attributes: ${JSON.stringify(results.knowledgeGraph.attributes, null, 2)}`
+                ? `**Attributes:**\n\`\`\`json\n${JSON.stringify(results.knowledgeGraph.attributes, null, 2)}\n\`\`\``
                 : '',
             '',
-        ]
-            .filter(Boolean)
-            .join('\n');
+        ].filter(Boolean);
+        outputLines.push(kgLines.join('\n\n'));
     }
     // Answer Box
     if (results.answerBox != null) {
         addSection('Answer Box');
-        output += [
+        const abLines = [
             results.answerBox.title != null
-                ? `Title: ${results.answerBox.title}`
-                : '',
-            results.answerBox.answer != null
-                ? `Answer: ${results.answerBox.answer}`
+                ? `**Title:** ${results.answerBox.title}`
                 : '',
             results.answerBox.snippet != null
-                ? `Snippet: ${results.answerBox.snippet}`
+                ? `**Snippet:** ${results.answerBox.snippet}`
+                : '',
+            results.answerBox.snippetHighlighted != null
+                ? `**Snippet Highlighted:** ${results.answerBox.snippetHighlighted
+                    .map((s) => `\`${s}\``)
+                    .join(' ')}`
+                : '',
+            results.answerBox.link != null
+                ? `**Link:** ${results.answerBox.link}`
                 : '',
-            results.answerBox.date != null ? `Date: ${results.answerBox.date}` : '',
             '',
-        ]
-            .filter(Boolean)
-            .join('\n');
+        ].filter(Boolean);
+        outputLines.push(abLines.join('\n\n'));
     }
     // People also ask
     const peopleAlsoAsk = results.peopleAlsoAsk ?? [];
     if (peopleAlsoAsk.length) {
         addSection('People Also Ask');
-        peopleAlsoAsk.forEach((p, _i) => {
-            output += [`Q: ${p.question}`, `A: ${p.answer}`, '']
-                .filter(Boolean)
-                .join('\n');
+        const paaLines = [];
+        peopleAlsoAsk.forEach((p, i) => {
+            const questionLines = [
+                `### Question ${i + 1}:`,
+                `"${p.question}"`,
+                `${p.snippet != null && p.snippet ? `Snippet: ${p.snippet}` : ''}`,
+                `${p.title != null && p.title ? `Title: ${p.title}` : ''}`,
+                `${p.link != null && p.link ? `Link: ${p.link}` : ''}`,
+                '',
+            ].filter(Boolean);
+            paaLines.push(questionLines.join('\n\n'));
         });
+        outputLines.push(paaLines.join(''));
     }
-    return output.trim();
+    return {
+        output: outputLines.join('\n').trim(),
+        references,
+    };
 }
 
 exports.formatResultsForLLM = formatResultsForLLM;

package/dist/cjs/tools/search/format.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"format.cjs","sources":["../../../../src/tools/search/format.ts"],"sourcesContent":["import type * as t from './types';\n\nexport function formatResultsForLLM(results: t.SearchResultData): string {\n  let output = '';\n\n  const addSection = (title: string): void => {\n    output += `\\n=== ${title} ===\\n`;\n  };\n\n  // Organic (web) results\n  const organic = results.organic ?? [];\n  if (organic.length) {\n    addSection('Web Results');\n    organic.forEach((r, i) => {\n      output += [\n        `Source ${i + 1}: ${r.title ?? '(no title)'}`,\n        `Citation Anchor: \\\\ue202turn0search${i + 1}`,\n        `URL: ${r.link}`,\n        r.snippet != null ? `Summary: ${r.snippet}` : '',\n        r.date != null ? `Date: ${r.date}` : '',\n        r.attribution != null ? `Source: ${r.attribution}` : '',\n        '',\n        '--- Content Highlights ---',\n        ...(r.highlights ?? [])\n          .filter((h) => h.text.trim().length > 0)\n          .map((h) => `[Relevance: ${h.score.toFixed(2)}]\\n${h.text.trim()}`),\n        '',\n      ]\n        .filter(Boolean)\n        .join('\\n');\n    });\n  }\n\n  // Ignoring these sections for now\n  // // Top stories (news)\n  // const topStores = results.topStories ?? [];\n  // if (topStores.length) {\n  //   addSection('News Results');\n  //   topStores.forEach((r, i) => {\n  //     output += [\n  //       `Anchor: \\ue202turn0news${i + 1}`,\n  //       `Title: ${r.title ?? '(no title)'}`,\n  //       `URL: ${r.link}`,\n  //       r.snippet != null ? `Snippet: ${r.snippet}` : '',\n  //       r.date != null ? `Date: ${r.date}` : '',\n  //       r.attribution != null ? `Source: ${r.attribution}` : '',\n  //       ''\n  //     ].filter(Boolean).join('\\n');\n  //   });\n  // }\n\n  // // Images\n  // const images = results.images ?? [];\n  // if (images.length) {\n  //   addSection('Image Results');\n  //   images.forEach((img, i) => {\n  //     output += [\n  //       `Anchor: \\ue202turn0image${i + 1}`,\n  //       `Title: ${img.title ?? '(no title)'}`,\n  //       `Image URL: ${img.imageUrl}`,\n  //       ''\n  //     ].join('\\n');\n  //   });\n  // }\n\n  // Knowledge Graph\n  if (results.knowledgeGraph != null) {\n    addSection('Knowledge Graph');\n    output += [\n      `Title: ${results.knowledgeGraph.title ?? '(no title)'}`,\n      results.knowledgeGraph.description != null\n        ? `Description: ${results.knowledgeGraph.description}`\n        : '',\n      results.knowledgeGraph.type != null\n        ? `Type: ${results.knowledgeGraph.type}`\n        : '',\n      results.knowledgeGraph.imageUrl != null\n        ? `Image URL: ${results.knowledgeGraph.imageUrl}`\n        : '',\n      results.knowledgeGraph.attributes != null\n        ? `Attributes: ${JSON.stringify(results.knowledgeGraph.attributes, null, 2)}`\n        : '',\n      '',\n    ]\n      .filter(Boolean)\n      .join('\\n');\n  }\n\n  // Answer Box\n  if (results.answerBox != null) {\n    addSection('Answer Box');\n    output += [\n      results.answerBox.title != null\n        ? `Title: ${results.answerBox.title}`\n        : '',\n      results.answerBox.answer != null\n        ? `Answer: ${results.answerBox.answer}`\n        : '',\n      results.answerBox.snippet != null\n        ? `Snippet: ${results.answerBox.snippet}`\n        : '',\n      results.answerBox.date != null ? `Date: ${results.answerBox.date}` : '',\n      '',\n    ]\n      .filter(Boolean)\n      .join('\\n');\n  }\n\n  // People also ask\n  const peopleAlsoAsk = results.peopleAlsoAsk ?? [];\n  if (peopleAlsoAsk.length) {\n    addSection('People Also Ask');\n    peopleAlsoAsk.forEach((p, _i) => {\n      output += [`Q: ${p.question}`, `A: ${p.answer}`, '']\n        .filter(Boolean)\n        .join('\\n');\n    });\n  }\n\n  return output.trim();\n}\n"],"names":[],"mappings":";;AAEM,SAAU,mBAAmB,CAAC,OAA2B,EAAA;IAC7D,IAAI,MAAM,GAAG,EAAE;AAEf,IAAA,MAAM,UAAU,GAAG,CAAC,KAAa,KAAU;AACzC,QAAA,MAAM,IAAI,CAAA,MAAA,EAAS,KAAK,CAAA,MAAA,CAAQ;AAClC,KAAC;;AAGD,IAAA,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,IAAI,EAAE;AACrC,IAAA,IAAI,OAAO,CAAC,MAAM,EAAE;QAClB,UAAU,CAAC,aAAa,CAAC;QACzB,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,KAAI;AACvB,YAAA,MAAM,IAAI;gBACR,CAAU,OAAA,EAAA,CAAC,GAAG,CAAC,CAAA,EAAA,EAAK,CAAC,CAAC,KAAK,IAAI,YAAY,CAAE,CAAA;gBAC7C,CAAsC,mCAAA,EAAA,CAAC,GAAG,CAAC,CAAE,CAAA;gBAC7C,CAAQ,KAAA,EAAA,CAAC,CAAC,IAAI,CAAE,CAAA;AAChB,gBAAA,CAAC,CAAC,OAAO,IAAI,IAAI,GAAG,CAAY,SAAA,EAAA,CAAC,CAAC,OAAO,CAAA,CAAE,GAAG,EAAE;AAChD,gBAAA,CAAC,CAAC,IAAI,IAAI,IAAI,GAAG,CAAS,MAAA,EAAA,CAAC,CAAC,IAAI,CAAA,CAAE,GAAG,EAAE;AACvC,gBAAA,CAAC,CAAC,WAAW,IAAI,IAAI,GAAG,CAAW,QAAA,EAAA,CAAC,CAAC,WAAW,CAAA,CAAE,GAAG,EAAE;gBACvD,EAAE;gBACF,4BAA4B;AAC5B,gBAAA,GAAG,CAAC,CAAC,CAAC,UAAU,IAAI,EAAE;AACnB,qBAAA,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;qBACtC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAA,YAAA,EAAe,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAA,GAAA,EAAM,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAA,CAAE,CAAC;gBACrE,EAAE;AACH;iBACE,MAAM,CAAC,OAAO;iBACd,IAAI,CAAC,IAAI,CAAC;AACf,SAAC,CAAC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAoCJ,IAAA,IAAI,OAAO,CAAC,cAAc,IAAI,IAAI,EAAE;QAClC,UAAU,CAAC,iBAAiB,CAAC;AAC7B,QAAA,MAAM,IAAI;AACR,YAAA,CAAA,OAAA,EAAU,OAAO,CAAC,cAAc,CAAC,KAAK,IAAI,YAAY,CAAE,CAAA;AACxD,YAAA,OAAO,CAAC,cAAc,CAAC,WAAW,IAAI;AACpC,kBAAE,CAAgB,aAAA,EAAA,OAAO,CAAC,cAAc,CAAC,WAAW,CAAE;AACtD,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,IAAI,IAAI;AAC7B,kBAAE,CAAS,MAAA,EAAA,OAAO,CAAC,cAAc,CAAC,IAAI,CAAE;AACxC,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,QAAQ,IAAI;AACjC,kBAAE,CAAc,WAAA,EAAA,OAAO,CAAC,cAAc,CAAC,QAAQ,CAAE;AACjD,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,UAAU,IAAI;AACnC,kBAAE,CAAe,YAAA,EAAA,IAAI,CAAC,SAAS,CAAC,OAAO,CAAC,cAAc,CAAC,UAAU,EAAE,IAAI,EAAE,CAAC,CAAC,CAAE;AAC7E,kBAAE,EAAE;YACN,EAAE;AACH;aACE,MAAM,CAAC,OAAO;aACd,IAAI,CAAC,IAAI,CAAC;;;AAIf,IAAA,IAAI,OAAO,CAAC,SAAS,IAAI,IAAI,EAAE;QAC7B,UAAU,CAAC,YAAY,CAAC;AACxB,QAAA,MAAM,IAAI;AACR,YAAA,OAAO,CAAC,SAAS,CAAC,KAAK,IAAI;AACzB,kBAAE,CAAU,OAAA,EAAA,OAAO,CAAC,SAAS,CAAC,KAAK,CAAE;AACrC,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,MAAM,IAAI;AAC1B,kBAAE,CAAW,QAAA,EAAA,OAAO,CAAC,SAAS,CAAC,MAAM,CAAE;AACvC,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,OAAO,IAAI;AAC3B,kBAAE,CAAY,SAAA,EAAA,OAAO,CAAC,SAAS,CAAC,OAAO,CAAE;AACzC,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,IAAI,IAAI,IAAI,GAAG,SAAS,OAAO,CAAC,SAAS,CAAC,IAAI,EAAE,GAAG,EAAE;YACvE,EAAE;AACH;aACE,MAAM,CAAC,OAAO;aACd,IAAI,CAAC,IAAI,CAAC;;;AAIf,IAAA,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,EAAE;AACjD,IAAA,IAAI,aAAa,CAAC,MAAM,EAAE;QACxB,UAAU,CAAC,iBAAiB,CAAC;QAC7B,aAAa,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,KAAI;AAC9B,YAAA,MAAM,IAAI,CAAC,CAAM,GAAA,EAAA,CAAC,CAAC,QAAQ,CAAA,CAAE,EAAE,CAAA,GAAA,EAAM,CAAC,CAAC,MAAM,CAAE,CAAA,EAAE,EAAE;iBAChD,MAAM,CAAC,OAAO;iBACd,IAAI,CAAC,IAAI,CAAC;AACf,SAAC,CAAC;;AAGJ,IAAA,OAAO,MAAM,CAAC,IAAI,EAAE;AACtB;;;;"}
+{"version":3,"file":"format.cjs","sources":["../../../../src/tools/search/format.ts"],"sourcesContent":["import type * as t from './types';\nimport { getDomainName, fileExtRegex } from './utils';\n\nfunction addHighlightSection(): string[] {\n  return ['\\n## Highlights', ''];\n}\n\n// Helper function to format a source (organic or top story)\nfunction formatSource(\n  source: t.ValidSource,\n  index: number,\n  turn: number,\n  sourceType: 'search' | 'news',\n  references: t.ResultReference[]\n): string {\n  /** Array of all lines to include in the output */\n  const outputLines: string[] = [];\n\n  // Add the title\n  outputLines.push(\n    `# ${sourceType.charAt(0).toUpperCase() + sourceType.slice(1)} ${index}: ${source.title != null && source.title ? `\"${source.title}\"` : '(no title)'}`\n  );\n  outputLines.push(`\\nAnchor: \\\\ue202turn${turn}${sourceType}${index}`);\n  outputLines.push(`URL: ${source.link}`);\n\n  // Add optional fields\n  if ('snippet' in source && source.snippet != null) {\n    outputLines.push(`Summary: ${source.snippet}`);\n  }\n\n  if (source.date != null) {\n    outputLines.push(`Date: ${source.date}`);\n  }\n\n  if (source.attribution != null) {\n    outputLines.push(`Source: ${source.attribution}`);\n  }\n\n  // Add highlight section or empty line\n  if ((source.highlights?.length ?? 0) > 0) {\n    outputLines.push(...addHighlightSection());\n  } else {\n    outputLines.push('');\n  }\n\n  // Process highlights if they exist\n  (source.highlights ?? [])\n    .filter((h) => h.text.trim().length > 0)\n    .forEach((h, hIndex) => {\n      outputLines.push(\n        `### Highlight ${hIndex + 1} [Relevance: ${h.score.toFixed(2)}]`\n      );\n      outputLines.push('');\n      outputLines.push('```text');\n      outputLines.push(h.text.trim());\n      outputLines.push('```');\n      outputLines.push('');\n\n      if (h.references != null && h.references.length) {\n        let hasHeader = false;\n        const refLines: string[] = [];\n\n        for (let j = 0; j < h.references.length; j++) {\n          const ref = h.references[j];\n          if (ref.reference.originalUrl.includes('mailto:')) {\n            continue;\n          }\n          references.push({\n            type: ref.type,\n            link: ref.reference.originalUrl,\n            attribution: getDomainName(ref.reference.originalUrl),\n            title: (\n              ((ref.reference.title ?? '') || ref.reference.text) ??\n              ''\n            ).split('\\n')[0],\n          });\n\n          if (ref.type !== 'link') {\n            continue;\n          }\n\n          if (fileExtRegex.test(ref.reference.originalUrl)) {\n            continue;\n          }\n\n          if (!hasHeader) {\n            refLines.push('Core References:');\n            hasHeader = true;\n          }\n\n          refLines.push(\n            `- ${ref.type}#${ref.originalIndex + 1}: ${ref.reference.originalUrl}`\n          );\n          refLines.push(\n            `\\t- Anchor: \\\\ue202turn${turn}ref${references.length - 1}`\n          );\n        }\n\n        if (hasHeader) {\n          outputLines.push(...refLines);\n          outputLines.push('');\n        }\n      }\n\n      if (hIndex < (source.highlights?.length ?? 0) - 1) {\n        outputLines.push('---');\n        outputLines.push('');\n      }\n    });\n\n  outputLines.push('');\n  return outputLines.join('\\n');\n}\n\nexport function formatResultsForLLM(\n  turn: number,\n  results: t.SearchResultData\n): { output: string; references: t.ResultReference[] } {\n  /** Array to collect all output lines */\n  const outputLines: string[] = [];\n\n  const addSection = (title: string): void => {\n    outputLines.push('');\n    outputLines.push(`=== ${title} ===`);\n    outputLines.push('');\n  };\n\n  const references: t.ResultReference[] = [];\n\n  // Organic (web) results\n  if (results.organic?.length != null && results.organic.length > 0) {\n    addSection(`Web Results, Turn ${turn}`);\n    for (let i = 0; i < results.organic.length; i++) {\n      const r = results.organic[i];\n      outputLines.push(formatSource(r, i, turn, 'search', references));\n      delete results.organic[i].highlights;\n    }\n  }\n\n  // Top stories (news)\n  const topStories = results.topStories ?? [];\n  if (topStories.length) {\n    addSection('News Results');\n    for (let i = 0; i < topStories.length; i++) {\n      const r = topStories[i];\n      outputLines.push(formatSource(r, i, turn, 'news', references));\n      if (results.topStories?.[i]?.highlights) {\n        delete results.topStories[i].highlights;\n      }\n    }\n  }\n\n  // // Images\n  // const images = results.images ?? [];\n  // if (images.length) {\n  //   addSection('Image Results');\n  //   const imageLines = images.map((img, i) => [\n  //     `Anchor: \\ue202turn0image${i}`,\n  //     `Title: ${img.title ?? '(no title)'}`,\n  //     `Image URL: ${img.imageUrl}`,\n  //     ''\n  //   ].join('\\n'));\n  //   outputLines.push(imageLines.join('\\n'));\n  // }\n\n  // Knowledge Graph\n  if (results.knowledgeGraph != null) {\n    addSection('Knowledge Graph');\n    const kgLines = [\n      `**Title:** ${results.knowledgeGraph.title ?? '(no title)'}`,\n      results.knowledgeGraph.type != null\n        ? `**Type:** ${results.knowledgeGraph.type}`\n        : '',\n      results.knowledgeGraph.description != null\n        ? `**Description:** ${results.knowledgeGraph.description}`\n        : '',\n      results.knowledgeGraph.descriptionSource != null\n        ? `**Description Source:** ${results.knowledgeGraph.descriptionSource}`\n        : '',\n      results.knowledgeGraph.descriptionLink != null\n        ? `**Description Link:** ${results.knowledgeGraph.descriptionLink}`\n        : '',\n      results.knowledgeGraph.imageUrl != null\n        ? `**Image URL:** ${results.knowledgeGraph.imageUrl}`\n        : '',\n      results.knowledgeGraph.website != null\n        ? `**Website:** ${results.knowledgeGraph.website}`\n        : '',\n      results.knowledgeGraph.attributes != null\n        ? `**Attributes:**\\n\\`\\`\\`json\\n${JSON.stringify(\n          results.knowledgeGraph.attributes,\n          null,\n          2\n        )}\\n\\`\\`\\``\n        : '',\n      '',\n    ].filter(Boolean);\n\n    outputLines.push(kgLines.join('\\n\\n'));\n  }\n\n  // Answer Box\n  if (results.answerBox != null) {\n    addSection('Answer Box');\n    const abLines = [\n      results.answerBox.title != null\n        ? `**Title:** ${results.answerBox.title}`\n        : '',\n      results.answerBox.snippet != null\n        ? `**Snippet:** ${results.answerBox.snippet}`\n        : '',\n      results.answerBox.snippetHighlighted != null\n        ? `**Snippet Highlighted:** ${results.answerBox.snippetHighlighted\n          .map((s) => `\\`${s}\\``)\n          .join(' ')}`\n        : '',\n      results.answerBox.link != null\n        ? `**Link:** ${results.answerBox.link}`\n        : '',\n      '',\n    ].filter(Boolean);\n\n    outputLines.push(abLines.join('\\n\\n'));\n  }\n\n  // People also ask\n  const peopleAlsoAsk = results.peopleAlsoAsk ?? [];\n  if (peopleAlsoAsk.length) {\n    addSection('People Also Ask');\n\n    const paaLines: string[] = [];\n    peopleAlsoAsk.forEach((p, i) => {\n      const questionLines = [\n        `### Question ${i + 1}:`,\n        `\"${p.question}\"`,\n        `${p.snippet != null && p.snippet ? `Snippet: ${p.snippet}` : ''}`,\n        `${p.title != null && p.title ? `Title: ${p.title}` : ''}`,\n        `${p.link != null && p.link ? `Link: ${p.link}` : ''}`,\n        '',\n      ].filter(Boolean);\n\n      paaLines.push(questionLines.join('\\n\\n'));\n    });\n\n    outputLines.push(paaLines.join(''));\n  }\n\n  return {\n    output: outputLines.join('\\n').trim(),\n    references,\n  };\n}\n"],"names":["getDomainName","fileExtRegex"],"mappings":";;;;AAGA,SAAS,mBAAmB,GAAA;AAC1B,IAAA,OAAO,CAAC,iBAAiB,EAAE,EAAE,CAAC;AAChC;AAEA;AACA,SAAS,YAAY,CACnB,MAAqB,EACrB,KAAa,EACb,IAAY,EACZ,UAA6B,EAC7B,UAA+B,EAAA;;IAG/B,MAAM,WAAW,GAAa,EAAE;;IAGhC,WAAW,CAAC,IAAI,CACd,CAAA,EAAA,EAAK,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAA,CAAA,EAAI,KAAK,CAAA,EAAA,EAAK,MAAM,CAAC,KAAK,IAAI,IAAI,IAAI,MAAM,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,KAAK,CAAG,CAAA,CAAA,GAAG,YAAY,CAAE,CAAA,CACvJ;IACD,WAAW,CAAC,IAAI,CAAC,CAAwB,qBAAA,EAAA,IAAI,CAAG,EAAA,UAAU,CAAG,EAAA,KAAK,CAAE,CAAA,CAAC;IACrE,WAAW,CAAC,IAAI,CAAC,CAAA,KAAA,EAAQ,MAAM,CAAC,IAAI,CAAE,CAAA,CAAC;;IAGvC,IAAI,SAAS,IAAI,MAAM,IAAI,MAAM,CAAC,OAAO,IAAI,IAAI,EAAE;QACjD,WAAW,CAAC,IAAI,CAAC,CAAA,SAAA,EAAY,MAAM,CAAC,OAAO,CAAE,CAAA,CAAC;;AAGhD,IAAA,IAAI,MAAM,CAAC,IAAI,IAAI,IAAI,EAAE;QACvB,WAAW,CAAC,IAAI,CAAC,CAAA,MAAA,EAAS,MAAM,CAAC,IAAI,CAAE,CAAA,CAAC;;AAG1C,IAAA,IAAI,MAAM,CAAC,WAAW,IAAI,IAAI,EAAE;QAC9B,WAAW,CAAC,IAAI,CAAC,CAAA,QAAA,EAAW,MAAM,CAAC,WAAW,CAAE,CAAA,CAAC;;;AAInD,IAAA,IAAI,CAAC,MAAM,CAAC,UAAU,EAAE,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;AACxC,QAAA,WAAW,CAAC,IAAI,CAAC,GAAG,mBAAmB,EAAE,CAAC;;SACrC;AACL,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;;;AAItB,IAAA,CAAC,MAAM,CAAC,UAAU,IAAI,EAAE;AACrB,SAAA,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;AACtC,SAAA,OAAO,CAAC,CAAC,CAAC,EAAE,MAAM,KAAI;AACrB,QAAA,WAAW,CAAC,IAAI,CACd,iBAAiB,MAAM,GAAG,CAAC,CAAgB,aAAA,EAAA,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAA,CAAA,CAAG,CACjE;AACD,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACpB,QAAA,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC;QAC3B,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;AAC/B,QAAA,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC;AACvB,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AAEpB,QAAA,IAAI,CAAC,CAAC,UAAU,IAAI,IAAI,IAAI,CAAC,CAAC,UAAU,CAAC,MAAM,EAAE;YAC/C,IAAI,SAAS,GAAG,KAAK;YACrB,MAAM,QAAQ,GAAa,EAAE;AAE7B,YAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;gBAC5C,MAAM,GAAG,GAAG,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC;gBAC3B,IAAI,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE;oBACjD;;gBAEF,UAAU,CAAC,IAAI,CAAC;oBACd,IAAI,EAAE,GAAG,CAAC,IAAI;AACd,oBAAA,IAAI,EAAE,GAAG,CAAC,SAAS,CAAC,WAAW;oBAC/B,WAAW,EAAEA,mBAAa,CAAC,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC;AACrD,oBAAA,KAAK,EAAE,CACL,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,IAAI,EAAE,KAAK,GAAG,CAAC,SAAS,CAAC,IAAI;wBAClD,EAAE,EACF,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;AACjB,iBAAA,CAAC;AAEF,gBAAA,IAAI,GAAG,CAAC,IAAI,KAAK,MAAM,EAAE;oBACvB;;gBAGF,IAAIC,kBAAY,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC,EAAE;oBAChD;;gBAGF,IAAI,CAAC,SAAS,EAAE;AACd,oBAAA,QAAQ,CAAC,IAAI,CAAC,kBAAkB,CAAC;oBACjC,SAAS,GAAG,IAAI;;gBAGlB,QAAQ,CAAC,IAAI,CACX,CAAA,EAAA,EAAK,GAAG,CAAC,IAAI,IAAI,GAAG,CAAC,aAAa,GAAG,CAAC,KAAK,GAAG,CAAC,SAAS,CAAC,WAAW,CAAE,CAAA,CACvE;AACD,gBAAA,QAAQ,CAAC,IAAI,CACX,CAAA,uBAAA,EAA0B,IAAI,CAAA,GAAA,EAAM,UAAU,CAAC,MAAM,GAAG,CAAC,CAAA,CAAE,CAC5D;;YAGH,IAAI,SAAS,EAAE;AACb,gBAAA,WAAW,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC;AAC7B,gBAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;;;AAIxB,QAAA,IAAI,MAAM,GAAG,CAAC,MAAM,CAAC,UAAU,EAAE,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;AACjD,YAAA,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC;AACvB,YAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;;AAExB,KAAC,CAAC;AAEJ,IAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACpB,IAAA,OAAO,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC;AAC/B;AAEgB,SAAA,mBAAmB,CACjC,IAAY,EACZ,OAA2B,EAAA;;IAG3B,MAAM,WAAW,GAAa,EAAE;AAEhC,IAAA,MAAM,UAAU,GAAG,CAAC,KAAa,KAAU;AACzC,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACpB,QAAA,WAAW,CAAC,IAAI,CAAC,OAAO,KAAK,CAAA,IAAA,CAAM,CAAC;AACpC,QAAA,WAAW,CAAC,IAAI,CAAC,EAAE,CAAC;AACtB,KAAC;IAED,MAAM,UAAU,GAAwB,EAAE;;AAG1C,IAAA,IAAI,OAAO,CAAC,OAAO,EAAE,MAAM,IAAI,IAAI,IAAI,OAAO,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE;AACjE,QAAA,UAAU,CAAC,CAAA,kBAAA,EAAqB,IAAI,CAAA,CAAE,CAAC;AACvC,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YAC/C,MAAM,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC;AAC5B,YAAA,WAAW,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,UAAU,CAAC,CAAC;YAChE,OAAO,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,UAAU;;;;AAKxC,IAAA,MAAM,UAAU,GAAG,OAAO,CAAC,UAAU,IAAI,EAAE;AAC3C,IAAA,IAAI,UAAU,CAAC,MAAM,EAAE;QACrB,UAAU,CAAC,cAAc,CAAC;AAC1B,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;AAC1C,YAAA,MAAM,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC;AACvB,YAAA,WAAW,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,UAAU,CAAC,CAAC;YAC9D,IAAI,OAAO,CAAC,UAAU,GAAG,CAAC,CAAC,EAAE,UAAU,EAAE;gBACvC,OAAO,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,UAAU;;;;;;;;;;;;;;;;;AAmB7C,IAAA,IAAI,OAAO,CAAC,cAAc,IAAI,IAAI,EAAE;QAClC,UAAU,CAAC,iBAAiB,CAAC;AAC7B,QAAA,MAAM,OAAO,GAAG;AACd,YAAA,CAAA,WAAA,EAAc,OAAO,CAAC,cAAc,CAAC,KAAK,IAAI,YAAY,CAAE,CAAA;AAC5D,YAAA,OAAO,CAAC,cAAc,CAAC,IAAI,IAAI;AAC7B,kBAAE,CAAa,UAAA,EAAA,OAAO,CAAC,cAAc,CAAC,IAAI,CAAE;AAC5C,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,WAAW,IAAI;AACpC,kBAAE,CAAoB,iBAAA,EAAA,OAAO,CAAC,cAAc,CAAC,WAAW,CAAE;AAC1D,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,iBAAiB,IAAI;AAC1C,kBAAE,CAA2B,wBAAA,EAAA,OAAO,CAAC,cAAc,CAAC,iBAAiB,CAAE;AACvE,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,eAAe,IAAI;AACxC,kBAAE,CAAyB,sBAAA,EAAA,OAAO,CAAC,cAAc,CAAC,eAAe,CAAE;AACnE,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,QAAQ,IAAI;AACjC,kBAAE,CAAkB,eAAA,EAAA,OAAO,CAAC,cAAc,CAAC,QAAQ,CAAE;AACrD,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,OAAO,IAAI;AAChC,kBAAE,CAAgB,aAAA,EAAA,OAAO,CAAC,cAAc,CAAC,OAAO,CAAE;AAClD,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,cAAc,CAAC,UAAU,IAAI;AACnC,kBAAE,CAAgC,6BAAA,EAAA,IAAI,CAAC,SAAS,CAC9C,OAAO,CAAC,cAAc,CAAC,UAAU,EACjC,IAAI,EACJ,CAAC,CACF,CAAU,QAAA;AACX,kBAAE,EAAE;YACN,EAAE;AACH,SAAA,CAAC,MAAM,CAAC,OAAO,CAAC;QAEjB,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;;;AAIxC,IAAA,IAAI,OAAO,CAAC,SAAS,IAAI,IAAI,EAAE;QAC7B,UAAU,CAAC,YAAY,CAAC;AACxB,QAAA,MAAM,OAAO,GAAG;AACd,YAAA,OAAO,CAAC,SAAS,CAAC,KAAK,IAAI;AACzB,kBAAE,CAAc,WAAA,EAAA,OAAO,CAAC,SAAS,CAAC,KAAK,CAAE;AACzC,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,OAAO,IAAI;AAC3B,kBAAE,CAAgB,aAAA,EAAA,OAAO,CAAC,SAAS,CAAC,OAAO,CAAE;AAC7C,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,kBAAkB,IAAI;AACtC,kBAAE,CAA4B,yBAAA,EAAA,OAAO,CAAC,SAAS,CAAC;qBAC7C,GAAG,CAAC,CAAC,CAAC,KAAK,CAAA,EAAA,EAAK,CAAC,CAAA,EAAA,CAAI;qBACrB,IAAI,CAAC,GAAG,CAAC,CAAE;AACd,kBAAE,EAAE;AACN,YAAA,OAAO,CAAC,SAAS,CAAC,IAAI,IAAI;AACxB,kBAAE,CAAa,UAAA,EAAA,OAAO,CAAC,SAAS,CAAC,IAAI,CAAE;AACvC,kBAAE,EAAE;YACN,EAAE;AACH,SAAA,CAAC,MAAM,CAAC,OAAO,CAAC;QAEjB,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;;;AAIxC,IAAA,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,EAAE;AACjD,IAAA,IAAI,aAAa,CAAC,MAAM,EAAE;QACxB,UAAU,CAAC,iBAAiB,CAAC;QAE7B,MAAM,QAAQ,GAAa,EAAE;QAC7B,aAAa,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,KAAI;AAC7B,YAAA,MAAM,aAAa,GAAG;gBACpB,CAAgB,aAAA,EAAA,CAAC,GAAG,CAAC,CAAG,CAAA,CAAA;gBACxB,CAAI,CAAA,EAAA,CAAC,CAAC,QAAQ,CAAG,CAAA,CAAA;gBACjB,CAAG,EAAA,CAAC,CAAC,OAAO,IAAI,IAAI,IAAI,CAAC,CAAC,OAAO,GAAG,CAAA,SAAA,EAAY,CAAC,CAAC,OAAO,EAAE,GAAG,EAAE,CAAE,CAAA;gBAClE,CAAG,EAAA,CAAC,CAAC,KAAK,IAAI,IAAI,IAAI,CAAC,CAAC,KAAK,GAAG,CAAA,OAAA,EAAU,CAAC,CAAC,KAAK,EAAE,GAAG,EAAE,CAAE,CAAA;gBAC1D,CAAG,EAAA,CAAC,CAAC,IAAI,IAAI,IAAI,IAAI,CAAC,CAAC,IAAI,GAAG,CAAA,MAAA,EAAS,CAAC,CAAC,IAAI,EAAE,GAAG,EAAE,CAAE,CAAA;gBACtD,EAAE;AACH,aAAA,CAAC,MAAM,CAAC,OAAO,CAAC;YAEjB,QAAQ,CAAC,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AAC3C,SAAC,CAAC;QAEF,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;;IAGrC,OAAO;QACL,MAAM,EAAE,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE;QACrC,UAAU;KACX;AACH;;;;"}

package/dist/cjs/tools/search/highlights.cjs

@@ -103,6 +103,57 @@ function findBestBoundary(text, direction = 'backward') {
     // No match found, use character boundary
     return direction === 'backward' ? text.length : 0;
 }
+/**
+ * Tracks references used in a highlight without changing their numbers
+ */
+function trackReferencesInHighlight(text, sourceResult // Source containing the original references
+) {
+    // Track used references
+    const references = [];
+    if (!text || text.length === 0 || !text.includes('#')) {
+        return { references }; // Early return
+    }
+    // Quick check for reference markers
+    if (!text.includes('link#') &&
+        !text.includes('image#') &&
+        !text.includes('video#')) {
+        return { references };
+    }
+    // Get references from the source if available
+    const sourceRefs = sourceResult.references || {
+        links: [],
+        images: [],
+        videos: [],
+    };
+    // Find references but don't modify text
+    const refRegex = /\((link|image|video)#(\d+)(?:\s+"([^"]*)")?\)/g;
+    let match;
+    while ((match = refRegex.exec(text)) !== null) {
+        const [, type, indexStr] = match;
+        const originalIndex = parseInt(indexStr, 10) - 1; // Convert to 0-based
+        // Get the source array for this type
+        const refType = type;
+        const sourceArray = sourceRefs[`${refType}s`];
+        // Skip if invalid reference
+        if (!sourceArray ||
+            originalIndex < 0 ||
+            originalIndex >= sourceArray.length) {
+            continue; // Skip invalid references
+        }
+        // Get original reference
+        const reference = sourceArray[originalIndex];
+        // Track if not already tracked
+        const alreadyTracked = references.some((ref) => ref.type === refType && ref.originalIndex === originalIndex);
+        if (!alreadyTracked) {
+            references.push({
+                type: refType,
+                originalIndex,
+                reference,
+            });
+        }
+    }
+    return { references };
+}
 /**
  * Expand highlights in search results using smart boundary detection.
  *
@@ -112,19 +163,16 @@ function findBestBoundary(text, direction = 'backward') {
  * @param searchResults - Search results object
  * @param mainExpandBy - Primary expansion size on each side (default: 300)
  * @param separatorExpandBy - Additional range to look for separators (default: 150)
- * @returns Copy of search results with expanded highlights
+ * @returns Copy of search results with expanded highlights and tracked references
  */
 function expandHighlights(searchResults, mainExpandBy = 300, separatorExpandBy = 150) {
-    // 1. Avoid full deep copy - only copy what we modify
+    // Avoid deep copy - only copy what we modify
     const resultCopy = { ...searchResults };
-    // Only deep copy the relevant arrays
-    if (resultCopy.organic) {
+    if (resultCopy.organic)
         resultCopy.organic = [...resultCopy.organic];
-    }
-    if (resultCopy.topStories) {
+    if (resultCopy.topStories)
         resultCopy.topStories = [...resultCopy.topStories];
-    }
-    // 5. Process the results efficiently
+    // Process the results efficiently
     const processResultTypes = ['organic', 'topStories'];
     for (const resultType of processResultTypes) {
         if (!resultCopy[resultType])
@@ -143,17 +191,18 @@ function expandHighlights(searchResults, mainExpandBy = 300, separatorExpandBy =
             const highlights = [];
             // Process each highlight
             for (const highlight of result.highlights) {
-                const highlightText = highlight.text;
-                let startPos = content.indexOf(highlightText);
-                let highlightLen = highlightText.length;
+                const { references } = trackReferencesInHighlight(highlight.text, result);
+                let startPos = content.indexOf(highlight.text);
+                let highlightLen = highlight.text.length;
                 if (startPos === -1) {
                     // Try with stripped whitespace
-                    const strippedHighlight = highlightText.trim();
+                    const strippedHighlight = highlight.text.trim();
                     startPos = content.indexOf(strippedHighlight);
                     if (startPos === -1) {
                         highlights.push({
                             text: highlight.text,
                             score: highlight.score,
+                            references,
                         });
                         continue;
                     }
@@ -180,9 +229,12 @@ function expandHighlights(searchResults, mainExpandBy = 300, separatorExpandBy =
                 highlights.push({
                     text: expandedHighlightText,
                     score: highlight.score,
+                    references,
                 });
             }
             resultCopy.highlights = highlights;
+            delete resultCopy.content;
+            delete resultCopy.references;
             return resultCopy;
         });
     }

package/dist/cjs/tools/search/highlights.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"highlights.cjs","sources":["../../../../src/tools/search/highlights.ts"],"sourcesContent":["import type * as t from './types';\n\n// 2. Pre-compile all regular expressions (only do this once)\n// Group patterns by priority for early returns\nconst priorityPatterns = [\n  // High priority patterns (structural)\n  [\n    { regex: /\\n\\n/g }, // Double newline (paragraph break)\n    { regex: /\\n/g }, // Single newline\n    { regex: /={3,}\\s*\\n|-{3,}\\s*\\n/g }, // Section separators\n  ],\n  // Medium priority (semantic)\n  [\n    { regex: /[.!?][\")\\]]?\\s/g }, // End of sentence\n    { regex: /;\\s/g }, // Semicolon\n    { regex: /:\\s/g }, // Colon\n  ],\n  // Low priority (any breaks)\n  [\n    { regex: /,\\s/g }, // Comma\n    { regex: /\\s-\\s/g }, // Dash surrounded by spaces\n    { regex: /\\s/g }, // Any space\n  ],\n];\n\nfunction findFirstMatch(text: string, regex: RegExp): number {\n  // Reset regex\n  regex.lastIndex = 0;\n\n  // For very long texts, try chunking\n  if (text.length > 10000) {\n    const chunkSize = 2000;\n    let position = 0;\n\n    while (position < text.length) {\n      const chunk = text.substring(position, position + chunkSize);\n      regex.lastIndex = 0;\n\n      const match = regex.exec(chunk);\n      if (match) {\n        return position + match.index;\n      }\n\n      // Move to next chunk with some overlap\n      position += chunkSize - 100;\n      if (position >= text.length) break;\n    }\n    return -1;\n  }\n\n  // For shorter texts, normal regex search\n  const match = regex.exec(text);\n  return match ? match.index : -1;\n}\n\n// 3. Optimized boundary finding functions\nfunction findLastMatch(text: string, regex: RegExp): number {\n  // Reset regex state\n  regex.lastIndex = 0;\n\n  let lastIndex = -1;\n  let lastLength = 0;\n  let match;\n\n  // For very long texts, use a different approach to avoid regex engine slowdowns\n  if (text.length > 10000) {\n    // Try dividing the text into chunks for faster processing\n    const chunkSize = 2000;\n    let startPosition = Math.max(0, text.length - chunkSize);\n\n    while (startPosition >= 0) {\n      const chunk = text.substring(startPosition, startPosition + chunkSize);\n      regex.lastIndex = 0;\n\n      let chunkLastIndex = -1;\n      let chunkLastLength = 0;\n\n      while ((match = regex.exec(chunk)) !== null) {\n        chunkLastIndex = match.index;\n        chunkLastLength = match[0].length;\n      }\n\n      if (chunkLastIndex !== -1) {\n        return startPosition + chunkLastIndex + chunkLastLength;\n      }\n\n      // Move to previous chunk with some overlap\n      startPosition = Math.max(0, startPosition - chunkSize + 100) - 1;\n      if (startPosition <= 0) break;\n    }\n    return -1;\n  }\n\n  // For shorter texts, normal regex search\n  while ((match = regex.exec(text)) !== null) {\n    lastIndex = match.index;\n    lastLength = match[0].length;\n  }\n\n  return lastIndex === -1 ? -1 : lastIndex + lastLength;\n}\n\n// 4. Find the best boundary with priority groups\nfunction findBestBoundary(text: string, direction = 'backward'): number {\n  if (!text || text.length === 0) return 0;\n\n  // Try each priority group\n  for (const patternGroup of priorityPatterns) {\n    for (const pattern of patternGroup) {\n      const position =\n        direction === 'backward'\n          ? findLastMatch(text, pattern.regex)\n          : findFirstMatch(text, pattern.regex);\n\n      if (position !== -1) {\n        return position;\n      }\n    }\n  }\n\n  // No match found, use character boundary\n  return direction === 'backward' ? text.length : 0;\n}\n\n/**\n * Expand highlights in search results using smart boundary detection.\n *\n * This implementation finds natural text boundaries like paragraphs, sentences,\n * and phrases to provide context while maintaining readability.\n *\n * @param searchResults - Search results object\n * @param mainExpandBy - Primary expansion size on each side (default: 300)\n * @param separatorExpandBy - Additional range to look for separators (default: 150)\n * @returns Copy of search results with expanded highlights\n */\nexport function expandHighlights(\n  searchResults: t.SearchResultData,\n  mainExpandBy = 300,\n  separatorExpandBy = 150\n): t.SearchResultData {\n  // 1. Avoid full deep copy - only copy what we modify\n  const resultCopy = { ...searchResults };\n\n  // Only deep copy the relevant arrays\n  if (resultCopy.organic) {\n    resultCopy.organic = [...resultCopy.organic];\n  }\n  if (resultCopy.topStories) {\n    resultCopy.topStories = [...resultCopy.topStories];\n  }\n\n  // 5. Process the results efficiently\n  const processResultTypes = ['organic', 'topStories'] as const;\n\n  for (const resultType of processResultTypes) {\n    if (!resultCopy[resultType as 'organic' | 'topStories']) continue;\n\n    // Map results to new array with modified highlights\n    resultCopy[resultType] = resultCopy[resultType]?.map((result) => {\n      if (\n        result.content == null ||\n        result.content === '' ||\n        !result.highlights ||\n        result.highlights.length === 0\n      ) {\n        return result; // No modification needed\n      }\n\n      // Create a shallow copy with expanded highlights\n      const resultCopy = { ...result };\n      const content = result.content;\n      const highlights = [];\n\n      // Process each highlight\n      for (const highlight of result.highlights) {\n        const highlightText = highlight.text;\n\n        let startPos = content.indexOf(highlightText);\n        let highlightLen = highlightText.length;\n\n        if (startPos === -1) {\n          // Try with stripped whitespace\n          const strippedHighlight = highlightText.trim();\n          startPos = content.indexOf(strippedHighlight);\n\n          if (startPos === -1) {\n            highlights.push({\n              text: highlight.text,\n              score: highlight.score,\n            });\n            continue;\n          }\n          highlightLen = strippedHighlight.length;\n        }\n\n        // Calculate boundaries\n        const mainStart = Math.max(0, startPos - mainExpandBy);\n        const mainEnd = Math.min(\n          content.length,\n          startPos + highlightLen + mainExpandBy\n        );\n\n        const separatorStart = Math.max(0, mainStart - separatorExpandBy);\n        const separatorEnd = Math.min(\n          content.length,\n          mainEnd + separatorExpandBy\n        );\n\n        // Extract text segments\n        const headText = content.substring(separatorStart, mainStart);\n        const tailText = content.substring(mainEnd, separatorEnd);\n\n        // Find natural boundaries\n        const bestHeadBoundary = findBestBoundary(headText, 'backward');\n        const bestTailBoundary = findBestBoundary(tailText, 'forward');\n\n        // Calculate final positions\n        const finalStart = separatorStart + bestHeadBoundary;\n        const finalEnd = mainEnd + bestTailBoundary;\n\n        // Extract the expanded highlight\n        const expandedHighlightText = content\n          .substring(finalStart, finalEnd)\n          .trim();\n        highlights.push({\n          text: expandedHighlightText,\n          score: highlight.score,\n        });\n      }\n\n      resultCopy.highlights = highlights;\n      return resultCopy;\n    });\n  }\n\n  return resultCopy;\n}\n"],"names":[],"mappings":";;AAEA;AACA;AACA,MAAM,gBAAgB,GAAG;;AAEvB,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,OAAO,EAAE;AAClB,QAAA,EAAE,KAAK,EAAE,KAAK,EAAE;AAChB,QAAA,EAAE,KAAK,EAAE,wBAAwB,EAAE;AACpC,KAAA;;AAED,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,iBAAiB,EAAE;AAC5B,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AACjB,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AAClB,KAAA;;AAED,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AACjB,QAAA,EAAE,KAAK,EAAE,QAAQ,EAAE;AACnB,QAAA,EAAE,KAAK,EAAE,KAAK,EAAE;AACjB,KAAA;CACF;AAED,SAAS,cAAc,CAAC,IAAY,EAAE,KAAa,EAAA;;AAEjD,IAAA,KAAK,CAAC,SAAS,GAAG,CAAC;;AAGnB,IAAA,IAAI,IAAI,CAAC,MAAM,GAAG,KAAK,EAAE;QACvB,MAAM,SAAS,GAAG,IAAI;QACtB,IAAI,QAAQ,GAAG,CAAC;AAEhB,QAAA,OAAO,QAAQ,GAAG,IAAI,CAAC,MAAM,EAAE;AAC7B,YAAA,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,QAAQ,GAAG,SAAS,CAAC;AAC5D,YAAA,KAAK,CAAC,SAAS,GAAG,CAAC;YAEnB,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC;YAC/B,IAAI,KAAK,EAAE;AACT,gBAAA,OAAO,QAAQ,GAAG,KAAK,CAAC,KAAK;;;AAI/B,YAAA,QAAQ,IAAI,SAAS,GAAG,GAAG;AAC3B,YAAA,IAAI,QAAQ,IAAI,IAAI,CAAC,MAAM;gBAAE;;QAE/B,OAAO,EAAE;;;IAIX,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;AAC9B,IAAA,OAAO,KAAK,GAAG,KAAK,CAAC,KAAK,GAAG,EAAE;AACjC;AAEA;AACA,SAAS,aAAa,CAAC,IAAY,EAAE,KAAa,EAAA;;AAEhD,IAAA,KAAK,CAAC,SAAS,GAAG,CAAC;AAEnB,IAAA,IAAI,SAAS,GAAG,EAAE;IAClB,IAAI,UAAU,GAAG,CAAC;AAClB,IAAA,IAAI,KAAK;;AAGT,IAAA,IAAI,IAAI,CAAC,MAAM,GAAG,KAAK,EAAE;;QAEvB,MAAM,SAAS,GAAG,IAAI;AACtB,QAAA,IAAI,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,SAAS,CAAC;AAExD,QAAA,OAAO,aAAa,IAAI,CAAC,EAAE;AACzB,YAAA,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,aAAa,EAAE,aAAa,GAAG,SAAS,CAAC;AACtE,YAAA,KAAK,CAAC,SAAS,GAAG,CAAC;AAEnB,YAAA,IAAI,cAAc,GAAG,EAAE;YACvB,IAAI,eAAe,GAAG,CAAC;AAEvB,YAAA,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,EAAE;AAC3C,gBAAA,cAAc,GAAG,KAAK,CAAC,KAAK;AAC5B,gBAAA,eAAe,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM;;AAGnC,YAAA,IAAI,cAAc,KAAK,EAAE,EAAE;AACzB,gBAAA,OAAO,aAAa,GAAG,cAAc,GAAG,eAAe;;;AAIzD,YAAA,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,aAAa,GAAG,SAAS,GAAG,GAAG,CAAC,GAAG,CAAC;YAChE,IAAI,aAAa,IAAI,CAAC;gBAAE;;QAE1B,OAAO,EAAE;;;AAIX,IAAA,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,EAAE;AAC1C,QAAA,SAAS,GAAG,KAAK,CAAC,KAAK;AACvB,QAAA,UAAU,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM;;AAG9B,IAAA,OAAO,SAAS,KAAK,EAAE,GAAG,EAAE,GAAG,SAAS,GAAG,UAAU;AACvD;AAEA;AACA,SAAS,gBAAgB,CAAC,IAAY,EAAE,SAAS,GAAG,UAAU,EAAA;AAC5D,IAAA,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;AAAE,QAAA,OAAO,CAAC;;AAGxC,IAAA,KAAK,MAAM,YAAY,IAAI,gBAAgB,EAAE;AAC3C,QAAA,KAAK,MAAM,OAAO,IAAI,YAAY,EAAE;AAClC,YAAA,MAAM,QAAQ,GACZ,SAAS,KAAK;kBACV,aAAa,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK;kBACjC,cAAc,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK,CAAC;AAEzC,YAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;AACnB,gBAAA,OAAO,QAAQ;;;;;AAMrB,IAAA,OAAO,SAAS,KAAK,UAAU,GAAG,IAAI,CAAC,MAAM,GAAG,CAAC;AACnD;AAEA;;;;;;;;;;AAUG;AACG,SAAU,gBAAgB,CAC9B,aAAiC,EACjC,YAAY,GAAG,GAAG,EAClB,iBAAiB,GAAG,GAAG,EAAA;;AAGvB,IAAA,MAAM,UAAU,GAAG,EAAE,GAAG,aAAa,EAAE;;AAGvC,IAAA,IAAI,UAAU,CAAC,OAAO,EAAE;QACtB,UAAU,CAAC,OAAO,GAAG,CAAC,GAAG,UAAU,CAAC,OAAO,CAAC;;AAE9C,IAAA,IAAI,UAAU,CAAC,UAAU,EAAE;QACzB,UAAU,CAAC,UAAU,GAAG,CAAC,GAAG,UAAU,CAAC,UAAU,CAAC;;;AAIpD,IAAA,MAAM,kBAAkB,GAAG,CAAC,SAAS,EAAE,YAAY,CAAU;AAE7D,IAAA,KAAK,MAAM,UAAU,IAAI,kBAAkB,EAAE;AAC3C,QAAA,IAAI,CAAC,UAAU,CAAC,UAAsC,CAAC;YAAE;;AAGzD,QAAA,UAAU,CAAC,UAAU,CAAC,GAAG,UAAU,CAAC,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,MAAM,KAAI;AAC9D,YAAA,IACE,MAAM,CAAC,OAAO,IAAI,IAAI;gBACtB,MAAM,CAAC,OAAO,KAAK,EAAE;gBACrB,CAAC,MAAM,CAAC,UAAU;AAClB,gBAAA,MAAM,CAAC,UAAU,CAAC,MAAM,KAAK,CAAC,EAC9B;gBACA,OAAO,MAAM,CAAC;;;AAIhB,YAAA,MAAM,UAAU,GAAG,EAAE,GAAG,MAAM,EAAE;AAChC,YAAA,MAAM,OAAO,GAAG,MAAM,CAAC,OAAO;YAC9B,MAAM,UAAU,GAAG,EAAE;;AAGrB,YAAA,KAAK,MAAM,SAAS,IAAI,MAAM,CAAC,UAAU,EAAE;AACzC,gBAAA,MAAM,aAAa,GAAG,SAAS,CAAC,IAAI;gBAEpC,IAAI,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,aAAa,CAAC;AAC7C,gBAAA,IAAI,YAAY,GAAG,aAAa,CAAC,MAAM;AAEvC,gBAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;;AAEnB,oBAAA,MAAM,iBAAiB,GAAG,aAAa,CAAC,IAAI,EAAE;AAC9C,oBAAA,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,iBAAiB,CAAC;AAE7C,oBAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;wBACnB,UAAU,CAAC,IAAI,CAAC;4BACd,IAAI,EAAE,SAAS,CAAC,IAAI;4BACpB,KAAK,EAAE,SAAS,CAAC,KAAK;AACvB,yBAAA,CAAC;wBACF;;AAEF,oBAAA,YAAY,GAAG,iBAAiB,CAAC,MAAM;;;AAIzC,gBAAA,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,QAAQ,GAAG,YAAY,CAAC;AACtD,gBAAA,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CACtB,OAAO,CAAC,MAAM,EACd,QAAQ,GAAG,YAAY,GAAG,YAAY,CACvC;AAED,gBAAA,MAAM,cAAc,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,SAAS,GAAG,iBAAiB,CAAC;AACjE,gBAAA,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAC3B,OAAO,CAAC,MAAM,EACd,OAAO,GAAG,iBAAiB,CAC5B;;gBAGD,MAAM,QAAQ,GAAG,OAAO,CAAC,SAAS,CAAC,cAAc,EAAE,SAAS,CAAC;gBAC7D,MAAM,QAAQ,GAAG,OAAO,CAAC,SAAS,CAAC,OAAO,EAAE,YAAY,CAAC;;gBAGzD,MAAM,gBAAgB,GAAG,gBAAgB,CAAC,QAAQ,EAAE,UAAU,CAAC;gBAC/D,MAAM,gBAAgB,GAAG,gBAAgB,CAAC,QAAQ,EAAE,SAAS,CAAC;;AAG9D,gBAAA,MAAM,UAAU,GAAG,cAAc,GAAG,gBAAgB;AACpD,gBAAA,MAAM,QAAQ,GAAG,OAAO,GAAG,gBAAgB;;gBAG3C,MAAM,qBAAqB,GAAG;AAC3B,qBAAA,SAAS,CAAC,UAAU,EAAE,QAAQ;AAC9B,qBAAA,IAAI,EAAE;gBACT,UAAU,CAAC,IAAI,CAAC;AACd,oBAAA,IAAI,EAAE,qBAAqB;oBAC3B,KAAK,EAAE,SAAS,CAAC,KAAK;AACvB,iBAAA,CAAC;;AAGJ,YAAA,UAAU,CAAC,UAAU,GAAG,UAAU;AAClC,YAAA,OAAO,UAAU;AACnB,SAAC,CAAC;;AAGJ,IAAA,OAAO,UAAU;AACnB;;;;"}
+{"version":3,"file":"highlights.cjs","sources":["../../../../src/tools/search/highlights.ts"],"sourcesContent":["import type * as t from './types';\n\n// 2. Pre-compile all regular expressions (only do this once)\n// Group patterns by priority for early returns\nconst priorityPatterns = [\n  // High priority patterns (structural)\n  [\n    { regex: /\\n\\n/g }, // Double newline (paragraph break)\n    { regex: /\\n/g }, // Single newline\n    { regex: /={3,}\\s*\\n|-{3,}\\s*\\n/g }, // Section separators\n  ],\n  // Medium priority (semantic)\n  [\n    { regex: /[.!?][\")\\]]?\\s/g }, // End of sentence\n    { regex: /;\\s/g }, // Semicolon\n    { regex: /:\\s/g }, // Colon\n  ],\n  // Low priority (any breaks)\n  [\n    { regex: /,\\s/g }, // Comma\n    { regex: /\\s-\\s/g }, // Dash surrounded by spaces\n    { regex: /\\s/g }, // Any space\n  ],\n];\n\nfunction findFirstMatch(text: string, regex: RegExp): number {\n  // Reset regex\n  regex.lastIndex = 0;\n\n  // For very long texts, try chunking\n  if (text.length > 10000) {\n    const chunkSize = 2000;\n    let position = 0;\n\n    while (position < text.length) {\n      const chunk = text.substring(position, position + chunkSize);\n      regex.lastIndex = 0;\n\n      const match = regex.exec(chunk);\n      if (match) {\n        return position + match.index;\n      }\n\n      // Move to next chunk with some overlap\n      position += chunkSize - 100;\n      if (position >= text.length) break;\n    }\n    return -1;\n  }\n\n  // For shorter texts, normal regex search\n  const match = regex.exec(text);\n  return match ? match.index : -1;\n}\n\n// 3. Optimized boundary finding functions\nfunction findLastMatch(text: string, regex: RegExp): number {\n  // Reset regex state\n  regex.lastIndex = 0;\n\n  let lastIndex = -1;\n  let lastLength = 0;\n  let match;\n\n  // For very long texts, use a different approach to avoid regex engine slowdowns\n  if (text.length > 10000) {\n    // Try dividing the text into chunks for faster processing\n    const chunkSize = 2000;\n    let startPosition = Math.max(0, text.length - chunkSize);\n\n    while (startPosition >= 0) {\n      const chunk = text.substring(startPosition, startPosition + chunkSize);\n      regex.lastIndex = 0;\n\n      let chunkLastIndex = -1;\n      let chunkLastLength = 0;\n\n      while ((match = regex.exec(chunk)) !== null) {\n        chunkLastIndex = match.index;\n        chunkLastLength = match[0].length;\n      }\n\n      if (chunkLastIndex !== -1) {\n        return startPosition + chunkLastIndex + chunkLastLength;\n      }\n\n      // Move to previous chunk with some overlap\n      startPosition = Math.max(0, startPosition - chunkSize + 100) - 1;\n      if (startPosition <= 0) break;\n    }\n    return -1;\n  }\n\n  // For shorter texts, normal regex search\n  while ((match = regex.exec(text)) !== null) {\n    lastIndex = match.index;\n    lastLength = match[0].length;\n  }\n\n  return lastIndex === -1 ? -1 : lastIndex + lastLength;\n}\n\n// 4. Find the best boundary with priority groups\nfunction findBestBoundary(text: string, direction = 'backward'): number {\n  if (!text || text.length === 0) return 0;\n\n  // Try each priority group\n  for (const patternGroup of priorityPatterns) {\n    for (const pattern of patternGroup) {\n      const position =\n        direction === 'backward'\n          ? findLastMatch(text, pattern.regex)\n          : findFirstMatch(text, pattern.regex);\n\n      if (position !== -1) {\n        return position;\n      }\n    }\n  }\n\n  // No match found, use character boundary\n  return direction === 'backward' ? text.length : 0;\n}\n\n/**\n * Tracks references used in a highlight without changing their numbers\n */\nfunction trackReferencesInHighlight(\n  text: string,\n  sourceResult: t.ValidSource // Source containing the original references\n): {\n  references: {\n    type: 'link' | 'image' | 'video';\n    originalIndex: number;\n    reference: t.MediaReference; // Original reference object\n  }[];\n} {\n  // Track used references\n  const references: {\n    type: 'link' | 'image' | 'video';\n    originalIndex: number;\n    reference: t.MediaReference;\n  }[] = [];\n\n  if (!text || text.length === 0 || !text.includes('#')) {\n    return { references }; // Early return\n  }\n\n  // Quick check for reference markers\n  if (\n    !text.includes('link#') &&\n    !text.includes('image#') &&\n    !text.includes('video#')\n  ) {\n    return { references };\n  }\n\n  // Get references from the source if available\n  const sourceRefs = sourceResult.references || {\n    links: [],\n    images: [],\n    videos: [],\n  };\n\n  // Find references but don't modify text\n  const refRegex = /\\((link|image|video)#(\\d+)(?:\\s+\"([^\"]*)\")?\\)/g;\n  let match;\n\n  while ((match = refRegex.exec(text)) !== null) {\n    const [, type, indexStr] = match;\n    const originalIndex = parseInt(indexStr, 10) - 1; // Convert to 0-based\n\n    // Get the source array for this type\n    const refType = type as 'link' | 'image' | 'video';\n    const sourceArray = sourceRefs[`${refType}s`] as\n      | t.MediaReference[]\n      | undefined;\n\n    // Skip if invalid reference\n    if (\n      !sourceArray ||\n      originalIndex < 0 ||\n      originalIndex >= sourceArray.length\n    ) {\n      continue; // Skip invalid references\n    }\n\n    // Get original reference\n    const reference = sourceArray[originalIndex];\n\n    // Track if not already tracked\n    const alreadyTracked = references.some(\n      (ref) => ref.type === refType && ref.originalIndex === originalIndex\n    );\n\n    if (!alreadyTracked) {\n      references.push({\n        type: refType,\n        originalIndex,\n        reference,\n      });\n    }\n  }\n\n  return { references };\n}\n\n/**\n * Expand highlights in search results using smart boundary detection.\n *\n * This implementation finds natural text boundaries like paragraphs, sentences,\n * and phrases to provide context while maintaining readability.\n *\n * @param searchResults - Search results object\n * @param mainExpandBy - Primary expansion size on each side (default: 300)\n * @param separatorExpandBy - Additional range to look for separators (default: 150)\n * @returns Copy of search results with expanded highlights and tracked references\n */\nexport function expandHighlights(\n  searchResults: t.SearchResultData,\n  mainExpandBy = 300,\n  separatorExpandBy = 150\n): t.SearchResultData {\n  // Avoid deep copy - only copy what we modify\n  const resultCopy = { ...searchResults };\n  if (resultCopy.organic) resultCopy.organic = [...resultCopy.organic];\n  if (resultCopy.topStories) resultCopy.topStories = [...resultCopy.topStories];\n\n  // Process the results efficiently\n  const processResultTypes = ['organic', 'topStories'] as const;\n\n  for (const resultType of processResultTypes) {\n    if (!resultCopy[resultType as 'organic' | 'topStories']) continue;\n\n    // Map results to new array with modified highlights\n    resultCopy[resultType] = resultCopy[resultType]?.map((result) => {\n      if (\n        result.content == null ||\n        result.content === '' ||\n        !result.highlights ||\n        result.highlights.length === 0\n      ) {\n        return result; // No modification needed\n      }\n\n      // Create a shallow copy with expanded highlights\n      const resultCopy = { ...result };\n      const content = result.content;\n      const highlights = [];\n      // Process each highlight\n      for (const highlight of result.highlights) {\n        const { references } = trackReferencesInHighlight(\n          highlight.text,\n          result\n        );\n\n        let startPos = content.indexOf(highlight.text);\n        let highlightLen = highlight.text.length;\n\n        if (startPos === -1) {\n          // Try with stripped whitespace\n          const strippedHighlight = highlight.text.trim();\n          startPos = content.indexOf(strippedHighlight);\n\n          if (startPos === -1) {\n            highlights.push({\n              text: highlight.text,\n              score: highlight.score,\n              references,\n            });\n            continue;\n          }\n          highlightLen = strippedHighlight.length;\n        }\n\n        // Calculate boundaries\n        const mainStart = Math.max(0, startPos - mainExpandBy);\n        const mainEnd = Math.min(\n          content.length,\n          startPos + highlightLen + mainExpandBy\n        );\n\n        const separatorStart = Math.max(0, mainStart - separatorExpandBy);\n        const separatorEnd = Math.min(\n          content.length,\n          mainEnd + separatorExpandBy\n        );\n\n        // Extract text segments\n        const headText = content.substring(separatorStart, mainStart);\n        const tailText = content.substring(mainEnd, separatorEnd);\n\n        // Find natural boundaries\n        const bestHeadBoundary = findBestBoundary(headText, 'backward');\n        const bestTailBoundary = findBestBoundary(tailText, 'forward');\n\n        // Calculate final positions\n        const finalStart = separatorStart + bestHeadBoundary;\n        const finalEnd = mainEnd + bestTailBoundary;\n\n        // Extract the expanded highlight\n        const expandedHighlightText = content\n          .substring(finalStart, finalEnd)\n          .trim();\n        highlights.push({\n          text: expandedHighlightText,\n          score: highlight.score,\n          references,\n        });\n      }\n\n      resultCopy.highlights = highlights;\n      delete resultCopy.content;\n      delete resultCopy.references;\n      return resultCopy;\n    });\n  }\n\n  return resultCopy;\n}\n"],"names":[],"mappings":";;AAEA;AACA;AACA,MAAM,gBAAgB,GAAG;;AAEvB,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,OAAO,EAAE;AAClB,QAAA,EAAE,KAAK,EAAE,KAAK,EAAE;AAChB,QAAA,EAAE,KAAK,EAAE,wBAAwB,EAAE;AACpC,KAAA;;AAED,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,iBAAiB,EAAE;AAC5B,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AACjB,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AAClB,KAAA;;AAED,IAAA;AACE,QAAA,EAAE,KAAK,EAAE,MAAM,EAAE;AACjB,QAAA,EAAE,KAAK,EAAE,QAAQ,EAAE;AACnB,QAAA,EAAE,KAAK,EAAE,KAAK,EAAE;AACjB,KAAA;CACF;AAED,SAAS,cAAc,CAAC,IAAY,EAAE,KAAa,EAAA;;AAEjD,IAAA,KAAK,CAAC,SAAS,GAAG,CAAC;;AAGnB,IAAA,IAAI,IAAI,CAAC,MAAM,GAAG,KAAK,EAAE;QACvB,MAAM,SAAS,GAAG,IAAI;QACtB,IAAI,QAAQ,GAAG,CAAC;AAEhB,QAAA,OAAO,QAAQ,GAAG,IAAI,CAAC,MAAM,EAAE;AAC7B,YAAA,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,QAAQ,GAAG,SAAS,CAAC;AAC5D,YAAA,KAAK,CAAC,SAAS,GAAG,CAAC;YAEnB,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC;YAC/B,IAAI,KAAK,EAAE;AACT,gBAAA,OAAO,QAAQ,GAAG,KAAK,CAAC,KAAK;;;AAI/B,YAAA,QAAQ,IAAI,SAAS,GAAG,GAAG;AAC3B,YAAA,IAAI,QAAQ,IAAI,IAAI,CAAC,MAAM;gBAAE;;QAE/B,OAAO,EAAE;;;IAIX,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;AAC9B,IAAA,OAAO,KAAK,GAAG,KAAK,CAAC,KAAK,GAAG,EAAE;AACjC;AAEA;AACA,SAAS,aAAa,CAAC,IAAY,EAAE,KAAa,EAAA;;AAEhD,IAAA,KAAK,CAAC,SAAS,GAAG,CAAC;AAEnB,IAAA,IAAI,SAAS,GAAG,EAAE;IAClB,IAAI,UAAU,GAAG,CAAC;AAClB,IAAA,IAAI,KAAK;;AAGT,IAAA,IAAI,IAAI,CAAC,MAAM,GAAG,KAAK,EAAE;;QAEvB,MAAM,SAAS,GAAG,IAAI;AACtB,QAAA,IAAI,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,SAAS,CAAC;AAExD,QAAA,OAAO,aAAa,IAAI,CAAC,EAAE;AACzB,YAAA,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,aAAa,EAAE,aAAa,GAAG,SAAS,CAAC;AACtE,YAAA,KAAK,CAAC,SAAS,GAAG,CAAC;AAEnB,YAAA,IAAI,cAAc,GAAG,EAAE;YACvB,IAAI,eAAe,GAAG,CAAC;AAEvB,YAAA,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,EAAE;AAC3C,gBAAA,cAAc,GAAG,KAAK,CAAC,KAAK;AAC5B,gBAAA,eAAe,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM;;AAGnC,YAAA,IAAI,cAAc,KAAK,EAAE,EAAE;AACzB,gBAAA,OAAO,aAAa,GAAG,cAAc,GAAG,eAAe;;;AAIzD,YAAA,aAAa,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,aAAa,GAAG,SAAS,GAAG,GAAG,CAAC,GAAG,CAAC;YAChE,IAAI,aAAa,IAAI,CAAC;gBAAE;;QAE1B,OAAO,EAAE;;;AAIX,IAAA,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,EAAE;AAC1C,QAAA,SAAS,GAAG,KAAK,CAAC,KAAK;AACvB,QAAA,UAAU,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM;;AAG9B,IAAA,OAAO,SAAS,KAAK,EAAE,GAAG,EAAE,GAAG,SAAS,GAAG,UAAU;AACvD;AAEA;AACA,SAAS,gBAAgB,CAAC,IAAY,EAAE,SAAS,GAAG,UAAU,EAAA;AAC5D,IAAA,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;AAAE,QAAA,OAAO,CAAC;;AAGxC,IAAA,KAAK,MAAM,YAAY,IAAI,gBAAgB,EAAE;AAC3C,QAAA,KAAK,MAAM,OAAO,IAAI,YAAY,EAAE;AAClC,YAAA,MAAM,QAAQ,GACZ,SAAS,KAAK;kBACV,aAAa,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK;kBACjC,cAAc,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK,CAAC;AAEzC,YAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;AACnB,gBAAA,OAAO,QAAQ;;;;;AAMrB,IAAA,OAAO,SAAS,KAAK,UAAU,GAAG,IAAI,CAAC,MAAM,GAAG,CAAC;AACnD;AAEA;;AAEG;AACH,SAAS,0BAA0B,CACjC,IAAY,EACZ,YAA2B;;;IAS3B,MAAM,UAAU,GAIV,EAAE;AAER,IAAA,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE;AACrD,QAAA,OAAO,EAAE,UAAU,EAAE,CAAC;;;AAIxB,IAAA,IACE,CAAC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC;AACvB,QAAA,CAAC,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC;AACxB,QAAA,CAAC,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EACxB;QACA,OAAO,EAAE,UAAU,EAAE;;;AAIvB,IAAA,MAAM,UAAU,GAAG,YAAY,CAAC,UAAU,IAAI;AAC5C,QAAA,KAAK,EAAE,EAAE;AACT,QAAA,MAAM,EAAE,EAAE;AACV,QAAA,MAAM,EAAE,EAAE;KACX;;IAGD,MAAM,QAAQ,GAAG,gDAAgD;AACjE,IAAA,IAAI,KAAK;AAET,IAAA,OAAO,CAAC,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,EAAE;QAC7C,MAAM,GAAG,IAAI,EAAE,QAAQ,CAAC,GAAG,KAAK;AAChC,QAAA,MAAM,aAAa,GAAG,QAAQ,CAAC,QAAQ,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC;;QAGjD,MAAM,OAAO,GAAG,IAAkC;QAClD,MAAM,WAAW,GAAG,UAAU,CAAC,GAAG,OAAO,CAAA,CAAA,CAAG,CAE/B;;AAGb,QAAA,IACE,CAAC,WAAW;AACZ,YAAA,aAAa,GAAG,CAAC;AACjB,YAAA,aAAa,IAAI,WAAW,CAAC,MAAM,EACnC;AACA,YAAA,SAAS;;;AAIX,QAAA,MAAM,SAAS,GAAG,WAAW,CAAC,aAAa,CAAC;;QAG5C,MAAM,cAAc,GAAG,UAAU,CAAC,IAAI,CACpC,CAAC,GAAG,KAAK,GAAG,CAAC,IAAI,KAAK,OAAO,IAAI,GAAG,CAAC,aAAa,KAAK,aAAa,CACrE;QAED,IAAI,CAAC,cAAc,EAAE;YACnB,UAAU,CAAC,IAAI,CAAC;AACd,gBAAA,IAAI,EAAE,OAAO;gBACb,aAAa;gBACb,SAAS;AACV,aAAA,CAAC;;;IAIN,OAAO,EAAE,UAAU,EAAE;AACvB;AAEA;;;;;;;;;;AAUG;AACG,SAAU,gBAAgB,CAC9B,aAAiC,EACjC,YAAY,GAAG,GAAG,EAClB,iBAAiB,GAAG,GAAG,EAAA;;AAGvB,IAAA,MAAM,UAAU,GAAG,EAAE,GAAG,aAAa,EAAE;IACvC,IAAI,UAAU,CAAC,OAAO;QAAE,UAAU,CAAC,OAAO,GAAG,CAAC,GAAG,UAAU,CAAC,OAAO,CAAC;IACpE,IAAI,UAAU,CAAC,UAAU;QAAE,UAAU,CAAC,UAAU,GAAG,CAAC,GAAG,UAAU,CAAC,UAAU,CAAC;;AAG7E,IAAA,MAAM,kBAAkB,GAAG,CAAC,SAAS,EAAE,YAAY,CAAU;AAE7D,IAAA,KAAK,MAAM,UAAU,IAAI,kBAAkB,EAAE;AAC3C,QAAA,IAAI,CAAC,UAAU,CAAC,UAAsC,CAAC;YAAE;;AAGzD,QAAA,UAAU,CAAC,UAAU,CAAC,GAAG,UAAU,CAAC,UAAU,CAAC,EAAE,GAAG,CAAC,CAAC,MAAM,KAAI;AAC9D,YAAA,IACE,MAAM,CAAC,OAAO,IAAI,IAAI;gBACtB,MAAM,CAAC,OAAO,KAAK,EAAE;gBACrB,CAAC,MAAM,CAAC,UAAU;AAClB,gBAAA,MAAM,CAAC,UAAU,CAAC,MAAM,KAAK,CAAC,EAC9B;gBACA,OAAO,MAAM,CAAC;;;AAIhB,YAAA,MAAM,UAAU,GAAG,EAAE,GAAG,MAAM,EAAE;AAChC,YAAA,MAAM,OAAO,GAAG,MAAM,CAAC,OAAO;YAC9B,MAAM,UAAU,GAAG,EAAE;;AAErB,YAAA,KAAK,MAAM,SAAS,IAAI,MAAM,CAAC,UAAU,EAAE;AACzC,gBAAA,MAAM,EAAE,UAAU,EAAE,GAAG,0BAA0B,CAC/C,SAAS,CAAC,IAAI,EACd,MAAM,CACP;gBAED,IAAI,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,SAAS,CAAC,IAAI,CAAC;AAC9C,gBAAA,IAAI,YAAY,GAAG,SAAS,CAAC,IAAI,CAAC,MAAM;AAExC,gBAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;;oBAEnB,MAAM,iBAAiB,GAAG,SAAS,CAAC,IAAI,CAAC,IAAI,EAAE;AAC/C,oBAAA,QAAQ,GAAG,OAAO,CAAC,OAAO,CAAC,iBAAiB,CAAC;AAE7C,oBAAA,IAAI,QAAQ,KAAK,EAAE,EAAE;wBACnB,UAAU,CAAC,IAAI,CAAC;4BACd,IAAI,EAAE,SAAS,CAAC,IAAI;4BACpB,KAAK,EAAE,SAAS,CAAC,KAAK;4BACtB,UAAU;AACX,yBAAA,CAAC;wBACF;;AAEF,oBAAA,YAAY,GAAG,iBAAiB,CAAC,MAAM;;;AAIzC,gBAAA,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,QAAQ,GAAG,YAAY,CAAC;AACtD,gBAAA,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CACtB,OAAO,CAAC,MAAM,EACd,QAAQ,GAAG,YAAY,GAAG,YAAY,CACvC;AAED,gBAAA,MAAM,cAAc,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,SAAS,GAAG,iBAAiB,CAAC;AACjE,gBAAA,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAC3B,OAAO,CAAC,MAAM,EACd,OAAO,GAAG,iBAAiB,CAC5B;;gBAGD,MAAM,QAAQ,GAAG,OAAO,CAAC,SAAS,CAAC,cAAc,EAAE,SAAS,CAAC;gBAC7D,MAAM,QAAQ,GAAG,OAAO,CAAC,SAAS,CAAC,OAAO,EAAE,YAAY,CAAC;;gBAGzD,MAAM,gBAAgB,GAAG,gBAAgB,CAAC,QAAQ,EAAE,UAAU,CAAC;gBAC/D,MAAM,gBAAgB,GAAG,gBAAgB,CAAC,QAAQ,EAAE,SAAS,CAAC;;AAG9D,gBAAA,MAAM,UAAU,GAAG,cAAc,GAAG,gBAAgB;AACpD,gBAAA,MAAM,QAAQ,GAAG,OAAO,GAAG,gBAAgB;;gBAG3C,MAAM,qBAAqB,GAAG;AAC3B,qBAAA,SAAS,CAAC,UAAU,EAAE,QAAQ;AAC9B,qBAAA,IAAI,EAAE;gBACT,UAAU,CAAC,IAAI,CAAC;AACd,oBAAA,IAAI,EAAE,qBAAqB;oBAC3B,KAAK,EAAE,SAAS,CAAC,KAAK;oBACtB,UAAU;AACX,iBAAA,CAAC;;AAGJ,YAAA,UAAU,CAAC,UAAU,GAAG,UAAU;YAClC,OAAO,UAAU,CAAC,OAAO;YACzB,OAAO,UAAU,CAAC,UAAU;AAC5B,YAAA,OAAO,UAAU;AACnB,SAAC,CAAC;;AAGJ,IAAA,OAAO,UAAU;AACnB;;;;"}