@librechat/agents 2.4.31 → 2.4.33
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/cjs/events.cjs +3 -3
- package/dist/cjs/events.cjs.map +1 -1
- package/dist/cjs/graphs/Graph.cjs +2 -1
- package/dist/cjs/graphs/Graph.cjs.map +1 -1
- package/dist/cjs/main.cjs +5 -2
- package/dist/cjs/main.cjs.map +1 -1
- package/dist/cjs/messages/ids.cjs +23 -0
- package/dist/cjs/messages/ids.cjs.map +1 -0
- package/dist/cjs/splitStream.cjs +2 -1
- package/dist/cjs/splitStream.cjs.map +1 -1
- package/dist/cjs/stream.cjs +87 -154
- package/dist/cjs/stream.cjs.map +1 -1
- package/dist/cjs/tools/ToolNode.cjs +14 -3
- package/dist/cjs/tools/ToolNode.cjs.map +1 -1
- package/dist/cjs/tools/handlers.cjs +144 -0
- package/dist/cjs/tools/handlers.cjs.map +1 -0
- package/dist/cjs/tools/search/content.cjs +140 -0
- package/dist/cjs/tools/search/content.cjs.map +1 -0
- package/dist/cjs/tools/search/firecrawl.cjs +23 -41
- package/dist/cjs/tools/search/firecrawl.cjs.map +1 -1
- package/dist/cjs/tools/search/format.cjs +161 -74
- package/dist/cjs/tools/search/format.cjs.map +1 -1
- package/dist/cjs/tools/search/highlights.cjs +64 -12
- package/dist/cjs/tools/search/highlights.cjs.map +1 -1
- package/dist/cjs/tools/search/rerankers.cjs +43 -36
- package/dist/cjs/tools/search/rerankers.cjs.map +1 -1
- package/dist/cjs/tools/search/schema.cjs +70 -0
- package/dist/cjs/tools/search/schema.cjs.map +1 -0
- package/dist/cjs/tools/search/search.cjs +150 -69
- package/dist/cjs/tools/search/search.cjs.map +1 -1
- package/dist/cjs/tools/search/tool.cjs +247 -58
- package/dist/cjs/tools/search/tool.cjs.map +1 -1
- package/dist/cjs/tools/search/utils.cjs +66 -0
- package/dist/cjs/tools/search/utils.cjs.map +1 -0
- package/dist/esm/events.mjs +1 -1
- package/dist/esm/events.mjs.map +1 -1
- package/dist/esm/graphs/Graph.mjs +2 -1
- package/dist/esm/graphs/Graph.mjs.map +1 -1
- package/dist/esm/main.mjs +3 -1
- package/dist/esm/main.mjs.map +1 -1
- package/dist/esm/messages/ids.mjs +21 -0
- package/dist/esm/messages/ids.mjs.map +1 -0
- package/dist/esm/splitStream.mjs +2 -1
- package/dist/esm/splitStream.mjs.map +1 -1
- package/dist/esm/stream.mjs +87 -152
- package/dist/esm/stream.mjs.map +1 -1
- package/dist/esm/tools/ToolNode.mjs +14 -3
- package/dist/esm/tools/ToolNode.mjs.map +1 -1
- package/dist/esm/tools/handlers.mjs +141 -0
- package/dist/esm/tools/handlers.mjs.map +1 -0
- package/dist/esm/tools/search/content.mjs +119 -0
- package/dist/esm/tools/search/content.mjs.map +1 -0
- package/dist/esm/tools/search/firecrawl.mjs +24 -41
- package/dist/esm/tools/search/firecrawl.mjs.map +1 -1
- package/dist/esm/tools/search/format.mjs +161 -74
- package/dist/esm/tools/search/format.mjs.map +1 -1
- package/dist/esm/tools/search/highlights.mjs +64 -12
- package/dist/esm/tools/search/highlights.mjs.map +1 -1
- package/dist/esm/tools/search/rerankers.mjs +43 -36
- package/dist/esm/tools/search/rerankers.mjs.map +1 -1
- package/dist/esm/tools/search/schema.mjs +61 -0
- package/dist/esm/tools/search/schema.mjs.map +1 -0
- package/dist/esm/tools/search/search.mjs +150 -69
- package/dist/esm/tools/search/search.mjs.map +1 -1
- package/dist/esm/tools/search/tool.mjs +246 -57
- package/dist/esm/tools/search/tool.mjs.map +1 -1
- package/dist/esm/tools/search/utils.mjs +61 -0
- package/dist/esm/tools/search/utils.mjs.map +1 -0
- package/dist/types/graphs/Graph.d.ts +1 -1
- package/dist/types/index.d.ts +1 -0
- package/dist/types/messages/ids.d.ts +3 -0
- package/dist/types/messages/index.d.ts +1 -0
- package/dist/types/stream.d.ts +0 -8
- package/dist/types/tools/ToolNode.d.ts +6 -0
- package/dist/types/tools/example.d.ts +23 -3
- package/dist/types/tools/handlers.d.ts +8 -0
- package/dist/types/tools/search/content.d.ts +4 -0
- package/dist/types/tools/search/firecrawl.d.ts +7 -86
- package/dist/types/tools/search/format.d.ts +4 -1
- package/dist/types/tools/search/highlights.d.ts +1 -1
- package/dist/types/tools/search/rerankers.d.ts +8 -4
- package/dist/types/tools/search/schema.d.ts +16 -0
- package/dist/types/tools/search/search.d.ts +2 -2
- package/dist/types/tools/search/test.d.ts +1 -0
- package/dist/types/tools/search/tool.d.ts +25 -4
- package/dist/types/tools/search/types.d.ts +443 -53
- package/dist/types/tools/search/utils.d.ts +10 -0
- package/package.json +9 -7
- package/src/events.ts +49 -15
- package/src/graphs/Graph.ts +6 -2
- package/src/index.ts +1 -0
- package/src/messages/ids.ts +26 -0
- package/src/messages/index.ts +1 -0
- package/src/scripts/search.ts +8 -3
- package/src/splitStream.test.ts +132 -71
- package/src/splitStream.ts +2 -1
- package/src/stream.ts +94 -183
- package/src/tools/ToolNode.ts +37 -14
- package/src/tools/handlers.ts +167 -0
- package/src/tools/search/content.test.ts +173 -0
- package/src/tools/search/content.ts +147 -0
- package/src/tools/search/firecrawl.ts +36 -148
- package/src/tools/search/format.ts +205 -74
- package/src/tools/search/highlights.ts +99 -16
- package/src/tools/search/output.md +2775 -0
- package/src/tools/search/rerankers.ts +57 -36
- package/src/tools/search/schema.ts +63 -0
- package/src/tools/search/search.ts +230 -117
- package/src/tools/search/test.html +884 -0
- package/src/tools/search/test.md +643 -0
- package/src/tools/search/test.ts +159 -0
- package/src/tools/search/tool.ts +363 -87
- package/src/tools/search/types.ts +503 -61
- package/src/tools/search/utils.ts +79 -0
- package/src/utils/llmConfig.ts +1 -1
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"highlights.mjs","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;;;;"}
|
|
1
|
+
{"version":3,"file":"highlights.mjs","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;;;;"}
|
|
@@ -1,10 +1,12 @@
|
|
|
1
1
|
import axios from 'axios';
|
|
2
|
+
import { createDefaultLogger } from './utils.mjs';
|
|
2
3
|
|
|
3
|
-
/* eslint-disable no-console */
|
|
4
4
|
class BaseReranker {
|
|
5
5
|
apiKey;
|
|
6
|
-
|
|
6
|
+
logger;
|
|
7
|
+
constructor(logger) {
|
|
7
8
|
// Each specific reranker will set its API key
|
|
9
|
+
this.logger = logger || createDefaultLogger();
|
|
8
10
|
}
|
|
9
11
|
getDefaultRanking(documents, topK) {
|
|
10
12
|
return documents
|
|
@@ -12,22 +14,22 @@ class BaseReranker {
|
|
|
12
14
|
.map((doc) => ({ text: doc, score: 0 }));
|
|
13
15
|
}
|
|
14
16
|
logDocumentSamples(documents) {
|
|
15
|
-
|
|
17
|
+
this.logger.debug('Sample documents being sent to API:');
|
|
16
18
|
for (let i = 0; i < Math.min(3, documents.length); i++) {
|
|
17
|
-
|
|
19
|
+
this.logger.debug(`Document ${i}: ${documents[i].substring(0, 100)}...`);
|
|
18
20
|
}
|
|
19
21
|
}
|
|
20
22
|
}
|
|
21
23
|
class JinaReranker extends BaseReranker {
|
|
22
|
-
constructor({ apiKey = process.env.JINA_API_KEY }) {
|
|
23
|
-
super();
|
|
24
|
+
constructor({ apiKey = process.env.JINA_API_KEY, logger, }) {
|
|
25
|
+
super(logger);
|
|
24
26
|
this.apiKey = apiKey;
|
|
25
27
|
}
|
|
26
28
|
async rerank(query, documents, topK = 5) {
|
|
27
|
-
|
|
29
|
+
this.logger.debug(`Reranking ${documents.length} documents with Jina`);
|
|
28
30
|
try {
|
|
29
31
|
if (this.apiKey == null || this.apiKey === '') {
|
|
30
|
-
|
|
32
|
+
this.logger.warn('JINA_API_KEY is not set. Using default ranking.');
|
|
31
33
|
return this.getDefaultRanking(documents, topK);
|
|
32
34
|
}
|
|
33
35
|
this.logDocumentSamples(documents);
|
|
@@ -45,13 +47,13 @@ class JinaReranker extends BaseReranker {
|
|
|
45
47
|
},
|
|
46
48
|
});
|
|
47
49
|
// Log the response data structure
|
|
48
|
-
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
|
|
50
|
+
this.logger.debug('Jina API response structure:');
|
|
51
|
+
this.logger.debug('Model:', response.data?.model);
|
|
52
|
+
this.logger.debug('Usage:', response.data?.usage);
|
|
53
|
+
this.logger.debug('Results count:', response.data?.results.length);
|
|
52
54
|
// Log a sample of the results
|
|
53
55
|
if ((response.data?.results.length ?? 0) > 0) {
|
|
54
|
-
|
|
56
|
+
this.logger.debug('Sample result:', JSON.stringify(response.data?.results[0], null, 2));
|
|
55
57
|
}
|
|
56
58
|
if (response.data && response.data.results.length) {
|
|
57
59
|
return response.data.results.map((result) => {
|
|
@@ -76,27 +78,27 @@ class JinaReranker extends BaseReranker {
|
|
|
76
78
|
});
|
|
77
79
|
}
|
|
78
80
|
else {
|
|
79
|
-
|
|
81
|
+
this.logger.warn('Unexpected response format from Jina API. Using default ranking.');
|
|
80
82
|
return this.getDefaultRanking(documents, topK);
|
|
81
83
|
}
|
|
82
84
|
}
|
|
83
85
|
catch (error) {
|
|
84
|
-
|
|
86
|
+
this.logger.error('Error using Jina reranker:', error);
|
|
85
87
|
// Fallback to default ranking on error
|
|
86
88
|
return this.getDefaultRanking(documents, topK);
|
|
87
89
|
}
|
|
88
90
|
}
|
|
89
91
|
}
|
|
90
92
|
class CohereReranker extends BaseReranker {
|
|
91
|
-
constructor({ apiKey = process.env.COHERE_API_KEY }) {
|
|
92
|
-
super();
|
|
93
|
+
constructor({ apiKey = process.env.COHERE_API_KEY, logger, }) {
|
|
94
|
+
super(logger);
|
|
93
95
|
this.apiKey = apiKey;
|
|
94
96
|
}
|
|
95
97
|
async rerank(query, documents, topK = 5) {
|
|
96
|
-
|
|
98
|
+
this.logger.debug(`Reranking ${documents.length} documents with Cohere`);
|
|
97
99
|
try {
|
|
98
100
|
if (this.apiKey == null || this.apiKey === '') {
|
|
99
|
-
|
|
101
|
+
this.logger.warn('COHERE_API_KEY is not set. Using default ranking.');
|
|
100
102
|
return this.getDefaultRanking(documents, topK);
|
|
101
103
|
}
|
|
102
104
|
this.logDocumentSamples(documents);
|
|
@@ -113,13 +115,13 @@ class CohereReranker extends BaseReranker {
|
|
|
113
115
|
},
|
|
114
116
|
});
|
|
115
117
|
// Log the response data structure
|
|
116
|
-
|
|
117
|
-
|
|
118
|
-
|
|
119
|
-
|
|
118
|
+
this.logger.debug('Cohere API response structure:');
|
|
119
|
+
this.logger.debug('ID:', response.data?.id);
|
|
120
|
+
this.logger.debug('Meta:', response.data?.meta);
|
|
121
|
+
this.logger.debug('Results count:', response.data?.results.length);
|
|
120
122
|
// Log a sample of the results
|
|
121
123
|
if ((response.data?.results.length ?? 0) > 0) {
|
|
122
|
-
|
|
124
|
+
this.logger.debug('Sample result:', JSON.stringify(response.data?.results[0], null, 2));
|
|
123
125
|
}
|
|
124
126
|
if (response.data && response.data.results.length) {
|
|
125
127
|
return response.data.results.map((result) => {
|
|
@@ -130,24 +132,24 @@ class CohereReranker extends BaseReranker {
|
|
|
130
132
|
});
|
|
131
133
|
}
|
|
132
134
|
else {
|
|
133
|
-
|
|
135
|
+
this.logger.warn('Unexpected response format from Cohere API. Using default ranking.');
|
|
134
136
|
return this.getDefaultRanking(documents, topK);
|
|
135
137
|
}
|
|
136
138
|
}
|
|
137
139
|
catch (error) {
|
|
138
|
-
|
|
140
|
+
this.logger.error('Error using Cohere reranker:', error);
|
|
139
141
|
// Fallback to default ranking on error
|
|
140
142
|
return this.getDefaultRanking(documents, topK);
|
|
141
143
|
}
|
|
142
144
|
}
|
|
143
145
|
}
|
|
144
146
|
class InfinityReranker extends BaseReranker {
|
|
145
|
-
constructor() {
|
|
146
|
-
super();
|
|
147
|
+
constructor(logger) {
|
|
148
|
+
super(logger);
|
|
147
149
|
// No API key needed for the placeholder implementation
|
|
148
150
|
}
|
|
149
151
|
async rerank(query, documents, topK = 5) {
|
|
150
|
-
|
|
152
|
+
this.logger.debug(`Reranking ${documents.length} documents with Infinity (placeholder)`);
|
|
151
153
|
// This would be replaced with actual Infinity reranker implementation
|
|
152
154
|
return this.getDefaultRanking(documents, topK);
|
|
153
155
|
}
|
|
@@ -156,20 +158,25 @@ class InfinityReranker extends BaseReranker {
|
|
|
156
158
|
* Creates the appropriate reranker based on type and configuration
|
|
157
159
|
*/
|
|
158
160
|
const createReranker = (config) => {
|
|
159
|
-
const { rerankerType, jinaApiKey, cohereApiKey } = config;
|
|
161
|
+
const { rerankerType, jinaApiKey, cohereApiKey, logger } = config;
|
|
162
|
+
// Create a default logger if none is provided
|
|
163
|
+
const defaultLogger = logger || createDefaultLogger();
|
|
160
164
|
switch (rerankerType.toLowerCase()) {
|
|
161
165
|
case 'jina':
|
|
162
|
-
return new JinaReranker({ apiKey: jinaApiKey });
|
|
166
|
+
return new JinaReranker({ apiKey: jinaApiKey, logger: defaultLogger });
|
|
163
167
|
case 'cohere':
|
|
164
|
-
return new CohereReranker({
|
|
168
|
+
return new CohereReranker({
|
|
169
|
+
apiKey: cohereApiKey,
|
|
170
|
+
logger: defaultLogger,
|
|
171
|
+
});
|
|
165
172
|
case 'infinity':
|
|
166
|
-
return new InfinityReranker();
|
|
173
|
+
return new InfinityReranker(defaultLogger);
|
|
167
174
|
case 'none':
|
|
168
|
-
|
|
175
|
+
defaultLogger.debug('Skipping reranking as reranker is set to "none"');
|
|
169
176
|
return undefined;
|
|
170
177
|
default:
|
|
171
|
-
|
|
172
|
-
return new JinaReranker({ apiKey: jinaApiKey });
|
|
178
|
+
defaultLogger.warn(`Unknown reranker type: ${rerankerType}. Defaulting to InfinityReranker.`);
|
|
179
|
+
return new JinaReranker({ apiKey: jinaApiKey, logger: defaultLogger });
|
|
173
180
|
}
|
|
174
181
|
};
|
|
175
182
|
// Example usage:
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"rerankers.mjs","sources":["../../../../src/tools/search/rerankers.ts"],"sourcesContent":["/* eslint-disable no-console */\nimport axios from 'axios';\nimport type * as t from './types';\n\nexport abstract class BaseReranker {\n protected apiKey: string | undefined;\n\n constructor() {\n // Each specific reranker will set its API key\n }\n\n abstract rerank(\n query: string,\n documents: string[],\n topK?: number\n ): Promise<t.Highlight[]>;\n\n protected getDefaultRanking(\n documents: string[],\n topK: number\n ): t.Highlight[] {\n return documents\n .slice(0, Math.min(topK, documents.length))\n .map((doc) => ({ text: doc, score: 0 }));\n }\n\n protected logDocumentSamples(documents: string[]): void {\n console.log('Sample documents being sent to API:');\n for (let i = 0; i < Math.min(3, documents.length); i++) {\n console.log(`Document ${i}: ${documents[i].substring(0, 100)}...`);\n }\n }\n}\n\nexport class JinaReranker extends BaseReranker {\n constructor({ apiKey = process.env.JINA_API_KEY }: { apiKey?: string }) {\n super();\n this.apiKey = apiKey;\n }\n\n async rerank(\n query: string,\n documents: string[],\n topK: number = 5\n ): Promise<t.Highlight[]> {\n console.log(`Reranking ${documents.length} documents with Jina`);\n\n try {\n if (this.apiKey == null || this.apiKey === '') {\n console.warn('JINA_API_KEY is not set. Using default ranking.');\n return this.getDefaultRanking(documents, topK);\n }\n\n this.logDocumentSamples(documents);\n\n const requestData = {\n model: 'jina-reranker-v2-base-multilingual',\n query: query,\n top_n: topK,\n documents: documents,\n return_documents: true,\n };\n\n const response = await axios.post<t.JinaRerankerResponse | undefined>(\n 'https://api.jina.ai/v1/rerank',\n requestData,\n {\n headers: {\n 'Content-Type': 'application/json',\n Authorization: `Bearer ${this.apiKey}`,\n },\n }\n );\n\n // Log the response data structure\n console.log('Jina API response structure:');\n console.log('Model:', response.data?.model);\n console.log('Usage:', response.data?.usage);\n console.log('Results count:', response.data?.results.length);\n\n // Log a sample of the results\n if ((response.data?.results.length ?? 0) > 0) {\n console.log(\n 'Sample result:',\n JSON.stringify(response.data?.results[0], null, 2)\n );\n }\n\n if (response.data && response.data.results.length) {\n return response.data.results.map((result) => {\n const docIndex = result.index;\n const score = result.relevance_score;\n let text = '';\n\n // If return_documents is true, the document field will be present\n if (result.document != null) {\n const doc = result.document;\n if (typeof doc === 'object' && 'text' in doc) {\n text = doc.text;\n } else if (typeof doc === 'string') {\n text = doc;\n }\n } else {\n // Otherwise, use the index to get the document\n text = documents[docIndex];\n }\n\n return { text, score };\n });\n } else {\n console.warn(\n 'Unexpected response format from Jina API. Using default ranking.'\n );\n return this.getDefaultRanking(documents, topK);\n }\n } catch (error) {\n console.error('Error using Jina reranker:', error);\n // Fallback to default ranking on error\n return this.getDefaultRanking(documents, topK);\n }\n }\n}\n\nexport class CohereReranker extends BaseReranker {\n constructor({ apiKey = process.env.COHERE_API_KEY }: { apiKey?: string }) {\n super();\n this.apiKey = apiKey;\n }\n\n async rerank(\n query: string,\n documents: string[],\n topK: number = 5\n ): Promise<t.Highlight[]> {\n console.log(`Reranking ${documents.length} documents with Cohere`);\n\n try {\n if (this.apiKey == null || this.apiKey === '') {\n console.warn('COHERE_API_KEY is not set. Using default ranking.');\n return this.getDefaultRanking(documents, topK);\n }\n\n this.logDocumentSamples(documents);\n\n const requestData = {\n model: 'rerank-v3.5',\n query: query,\n top_n: topK,\n documents: documents,\n };\n\n const response = await axios.post<t.CohereRerankerResponse | undefined>(\n 'https://api.cohere.com/v2/rerank',\n requestData,\n {\n headers: {\n 'Content-Type': 'application/json',\n Authorization: `Bearer ${this.apiKey}`,\n },\n }\n );\n\n // Log the response data structure\n console.log('Cohere API response structure:');\n console.log('ID:', response.data?.id);\n console.log('Meta:', response.data?.meta);\n console.log('Results count:', response.data?.results.length);\n\n // Log a sample of the results\n if ((response.data?.results.length ?? 0) > 0) {\n console.log(\n 'Sample result:',\n JSON.stringify(response.data?.results[0], null, 2)\n );\n }\n\n if (response.data && response.data.results.length) {\n return response.data.results.map((result) => {\n const docIndex = result.index;\n const score = result.relevance_score;\n const text = documents[docIndex];\n return { text, score };\n });\n } else {\n console.warn(\n 'Unexpected response format from Cohere API. Using default ranking.'\n );\n return this.getDefaultRanking(documents, topK);\n }\n } catch (error) {\n console.error('Error using Cohere reranker:', error);\n // Fallback to default ranking on error\n return this.getDefaultRanking(documents, topK);\n }\n }\n}\n\nexport class InfinityReranker extends BaseReranker {\n constructor() {\n super();\n // No API key needed for the placeholder implementation\n }\n\n async rerank(\n query: string,\n documents: string[],\n topK: number = 5\n ): Promise<t.Highlight[]> {\n console.log(\n `Reranking ${documents.length} documents with Infinity (placeholder)`\n );\n // This would be replaced with actual Infinity reranker implementation\n return this.getDefaultRanking(documents, topK);\n }\n}\n\n/**\n * Creates the appropriate reranker based on type and configuration\n */\nexport const createReranker = (config: {\n rerankerType: t.RerankerType;\n jinaApiKey?: string;\n cohereApiKey?: string;\n}): BaseReranker | undefined => {\n const { rerankerType, jinaApiKey, cohereApiKey } = config;\n\n switch (rerankerType.toLowerCase()) {\n case 'jina':\n return new JinaReranker({ apiKey: jinaApiKey });\n case 'cohere':\n return new CohereReranker({ apiKey: cohereApiKey });\n case 'infinity':\n return new InfinityReranker();\n case 'none':\n console.log('Skipping reranking as reranker is set to \"none\"');\n return undefined;\n default:\n console.warn(\n `Unknown reranker type: ${rerankerType}. Defaulting to InfinityReranker.`\n );\n return new JinaReranker({ apiKey: jinaApiKey });\n }\n};\n\n// Example usage:\n// const jinaReranker = new JinaReranker();\n// const cohereReranker = new CohereReranker();\n// const infinityReranker = new InfinityReranker();\n"],"names":[],"mappings":";;AAAA;MAIsB,YAAY,CAAA;AACtB,IAAA,MAAM;AAEhB,IAAA,WAAA,GAAA;;;IAUU,iBAAiB,CACzB,SAAmB,EACnB,IAAY,EAAA;AAEZ,QAAA,OAAO;AACJ,aAAA,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,SAAS,CAAC,MAAM,CAAC;AACzC,aAAA,GAAG,CAAC,CAAC,GAAG,MAAM,EAAE,IAAI,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;;AAGlC,IAAA,kBAAkB,CAAC,SAAmB,EAAA;AAC9C,QAAA,OAAO,CAAC,GAAG,CAAC,qCAAqC,CAAC;QAClD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,SAAS,CAAC,MAAM,CAAC,EAAE,CAAC,EAAE,EAAE;AACtD,YAAA,OAAO,CAAC,GAAG,CAAC,YAAY,CAAC,CAAA,EAAA,EAAK,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC,CAAA,GAAA,CAAK,CAAC;;;AAGvE;AAEK,MAAO,YAAa,SAAQ,YAAY,CAAA;IAC5C,WAAY,CAAA,EAAE,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,YAAY,EAAuB,EAAA;AACpE,QAAA,KAAK,EAAE;AACP,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM;;IAGtB,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;QAEhB,OAAO,CAAC,GAAG,CAAC,CAAA,UAAA,EAAa,SAAS,CAAC,MAAM,CAAsB,oBAAA,CAAA,CAAC;AAEhE,QAAA,IAAI;AACF,YAAA,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,EAAE,EAAE;AAC7C,gBAAA,OAAO,CAAC,IAAI,CAAC,iDAAiD,CAAC;gBAC/D,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAGhD,YAAA,IAAI,CAAC,kBAAkB,CAAC,SAAS,CAAC;AAElC,YAAA,MAAM,WAAW,GAAG;AAClB,gBAAA,KAAK,EAAE,oCAAoC;AAC3C,gBAAA,KAAK,EAAE,KAAK;AACZ,gBAAA,KAAK,EAAE,IAAI;AACX,gBAAA,SAAS,EAAE,SAAS;AACpB,gBAAA,gBAAgB,EAAE,IAAI;aACvB;YAED,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,+BAA+B,EAC/B,WAAW,EACX;AACE,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;AACF,aAAA,CACF;;AAGD,YAAA,OAAO,CAAC,GAAG,CAAC,8BAA8B,CAAC;YAC3C,OAAO,CAAC,GAAG,CAAC,QAAQ,EAAE,QAAQ,CAAC,IAAI,EAAE,KAAK,CAAC;YAC3C,OAAO,CAAC,GAAG,CAAC,QAAQ,EAAE,QAAQ,CAAC,IAAI,EAAE,KAAK,CAAC;AAC3C,YAAA,OAAO,CAAC,GAAG,CAAC,gBAAgB,EAAE,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,CAAC;;AAG5D,YAAA,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;gBAC5C,OAAO,CAAC,GAAG,CACT,gBAAgB,EAChB,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,CACnD;;AAGH,YAAA,IAAI,QAAQ,CAAC,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE;gBACjD,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,KAAI;AAC1C,oBAAA,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK;AAC7B,oBAAA,MAAM,KAAK,GAAG,MAAM,CAAC,eAAe;oBACpC,IAAI,IAAI,GAAG,EAAE;;AAGb,oBAAA,IAAI,MAAM,CAAC,QAAQ,IAAI,IAAI,EAAE;AAC3B,wBAAA,MAAM,GAAG,GAAG,MAAM,CAAC,QAAQ;wBAC3B,IAAI,OAAO,GAAG,KAAK,QAAQ,IAAI,MAAM,IAAI,GAAG,EAAE;AAC5C,4BAAA,IAAI,GAAG,GAAG,CAAC,IAAI;;AACV,6BAAA,IAAI,OAAO,GAAG,KAAK,QAAQ,EAAE;4BAClC,IAAI,GAAG,GAAG;;;yBAEP;;AAEL,wBAAA,IAAI,GAAG,SAAS,CAAC,QAAQ,CAAC;;AAG5B,oBAAA,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE;AACxB,iBAAC,CAAC;;iBACG;AACL,gBAAA,OAAO,CAAC,IAAI,CACV,kEAAkE,CACnE;gBACD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;QAEhD,OAAO,KAAK,EAAE;AACd,YAAA,OAAO,CAAC,KAAK,CAAC,4BAA4B,EAAE,KAAK,CAAC;;YAElD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;AAGnD;AAEK,MAAO,cAAe,SAAQ,YAAY,CAAA;IAC9C,WAAY,CAAA,EAAE,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,cAAc,EAAuB,EAAA;AACtE,QAAA,KAAK,EAAE;AACP,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM;;IAGtB,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;QAEhB,OAAO,CAAC,GAAG,CAAC,CAAA,UAAA,EAAa,SAAS,CAAC,MAAM,CAAwB,sBAAA,CAAA,CAAC;AAElE,QAAA,IAAI;AACF,YAAA,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,EAAE,EAAE;AAC7C,gBAAA,OAAO,CAAC,IAAI,CAAC,mDAAmD,CAAC;gBACjE,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAGhD,YAAA,IAAI,CAAC,kBAAkB,CAAC,SAAS,CAAC;AAElC,YAAA,MAAM,WAAW,GAAG;AAClB,gBAAA,KAAK,EAAE,aAAa;AACpB,gBAAA,KAAK,EAAE,KAAK;AACZ,gBAAA,KAAK,EAAE,IAAI;AACX,gBAAA,SAAS,EAAE,SAAS;aACrB;YAED,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,kCAAkC,EAClC,WAAW,EACX;AACE,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;AACF,aAAA,CACF;;AAGD,YAAA,OAAO,CAAC,GAAG,CAAC,gCAAgC,CAAC;YAC7C,OAAO,CAAC,GAAG,CAAC,KAAK,EAAE,QAAQ,CAAC,IAAI,EAAE,EAAE,CAAC;YACrC,OAAO,CAAC,GAAG,CAAC,OAAO,EAAE,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC;AACzC,YAAA,OAAO,CAAC,GAAG,CAAC,gBAAgB,EAAE,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,CAAC;;AAG5D,YAAA,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;gBAC5C,OAAO,CAAC,GAAG,CACT,gBAAgB,EAChB,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,CACnD;;AAGH,YAAA,IAAI,QAAQ,CAAC,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE;gBACjD,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,KAAI;AAC1C,oBAAA,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK;AAC7B,oBAAA,MAAM,KAAK,GAAG,MAAM,CAAC,eAAe;AACpC,oBAAA,MAAM,IAAI,GAAG,SAAS,CAAC,QAAQ,CAAC;AAChC,oBAAA,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE;AACxB,iBAAC,CAAC;;iBACG;AACL,gBAAA,OAAO,CAAC,IAAI,CACV,oEAAoE,CACrE;gBACD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;QAEhD,OAAO,KAAK,EAAE;AACd,YAAA,OAAO,CAAC,KAAK,CAAC,8BAA8B,EAAE,KAAK,CAAC;;YAEpD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;AAGnD;AAEK,MAAO,gBAAiB,SAAQ,YAAY,CAAA;AAChD,IAAA,WAAA,GAAA;AACE,QAAA,KAAK,EAAE;;;IAIT,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;QAEhB,OAAO,CAAC,GAAG,CACT,CAAA,UAAA,EAAa,SAAS,CAAC,MAAM,CAAwC,sCAAA,CAAA,CACtE;;QAED,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAEjD;AAED;;AAEG;AACU,MAAA,cAAc,GAAG,CAAC,MAI9B,KAA8B;IAC7B,MAAM,EAAE,YAAY,EAAE,UAAU,EAAE,YAAY,EAAE,GAAG,MAAM;AAEzD,IAAA,QAAQ,YAAY,CAAC,WAAW,EAAE;AAClC,QAAA,KAAK,MAAM;YACT,OAAO,IAAI,YAAY,CAAC,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC;AACjD,QAAA,KAAK,QAAQ;YACX,OAAO,IAAI,cAAc,CAAC,EAAE,MAAM,EAAE,YAAY,EAAE,CAAC;AACrD,QAAA,KAAK,UAAU;YACb,OAAO,IAAI,gBAAgB,EAAE;AAC/B,QAAA,KAAK,MAAM;AACT,YAAA,OAAO,CAAC,GAAG,CAAC,iDAAiD,CAAC;AAC9D,YAAA,OAAO,SAAS;AAClB,QAAA;AACE,YAAA,OAAO,CAAC,IAAI,CACV,0BAA0B,YAAY,CAAA,iCAAA,CAAmC,CAC1E;YACD,OAAO,IAAI,YAAY,CAAC,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC;;AAEnD;AAEA;AACA;AACA;AACA;;;;"}
|
|
1
|
+
{"version":3,"file":"rerankers.mjs","sources":["../../../../src/tools/search/rerankers.ts"],"sourcesContent":["import axios from 'axios';\nimport type * as t from './types';\nimport { createDefaultLogger } from './utils';\n\nexport abstract class BaseReranker {\n protected apiKey: string | undefined;\n protected logger: t.Logger;\n\n constructor(logger?: t.Logger) {\n // Each specific reranker will set its API key\n this.logger = logger || createDefaultLogger();\n }\n\n abstract rerank(\n query: string,\n documents: string[],\n topK?: number\n ): Promise<t.Highlight[]>;\n\n protected getDefaultRanking(\n documents: string[],\n topK: number\n ): t.Highlight[] {\n return documents\n .slice(0, Math.min(topK, documents.length))\n .map((doc) => ({ text: doc, score: 0 }));\n }\n\n protected logDocumentSamples(documents: string[]): void {\n this.logger.debug('Sample documents being sent to API:');\n for (let i = 0; i < Math.min(3, documents.length); i++) {\n this.logger.debug(`Document ${i}: ${documents[i].substring(0, 100)}...`);\n }\n }\n}\n\nexport class JinaReranker extends BaseReranker {\n constructor({\n apiKey = process.env.JINA_API_KEY,\n logger,\n }: {\n apiKey?: string;\n logger?: t.Logger;\n }) {\n super(logger);\n this.apiKey = apiKey;\n }\n\n async rerank(\n query: string,\n documents: string[],\n topK: number = 5\n ): Promise<t.Highlight[]> {\n this.logger.debug(`Reranking ${documents.length} documents with Jina`);\n\n try {\n if (this.apiKey == null || this.apiKey === '') {\n this.logger.warn('JINA_API_KEY is not set. Using default ranking.');\n return this.getDefaultRanking(documents, topK);\n }\n\n this.logDocumentSamples(documents);\n\n const requestData = {\n model: 'jina-reranker-v2-base-multilingual',\n query: query,\n top_n: topK,\n documents: documents,\n return_documents: true,\n };\n\n const response = await axios.post<t.JinaRerankerResponse | undefined>(\n 'https://api.jina.ai/v1/rerank',\n requestData,\n {\n headers: {\n 'Content-Type': 'application/json',\n Authorization: `Bearer ${this.apiKey}`,\n },\n }\n );\n\n // Log the response data structure\n this.logger.debug('Jina API response structure:');\n this.logger.debug('Model:', response.data?.model);\n this.logger.debug('Usage:', response.data?.usage);\n this.logger.debug('Results count:', response.data?.results.length);\n\n // Log a sample of the results\n if ((response.data?.results.length ?? 0) > 0) {\n this.logger.debug(\n 'Sample result:',\n JSON.stringify(response.data?.results[0], null, 2)\n );\n }\n\n if (response.data && response.data.results.length) {\n return response.data.results.map((result) => {\n const docIndex = result.index;\n const score = result.relevance_score;\n let text = '';\n\n // If return_documents is true, the document field will be present\n if (result.document != null) {\n const doc = result.document;\n if (typeof doc === 'object' && 'text' in doc) {\n text = doc.text;\n } else if (typeof doc === 'string') {\n text = doc;\n }\n } else {\n // Otherwise, use the index to get the document\n text = documents[docIndex];\n }\n\n return { text, score };\n });\n } else {\n this.logger.warn(\n 'Unexpected response format from Jina API. Using default ranking.'\n );\n return this.getDefaultRanking(documents, topK);\n }\n } catch (error) {\n this.logger.error('Error using Jina reranker:', error);\n // Fallback to default ranking on error\n return this.getDefaultRanking(documents, topK);\n }\n }\n}\n\nexport class CohereReranker extends BaseReranker {\n constructor({\n apiKey = process.env.COHERE_API_KEY,\n logger,\n }: {\n apiKey?: string;\n logger?: t.Logger;\n }) {\n super(logger);\n this.apiKey = apiKey;\n }\n\n async rerank(\n query: string,\n documents: string[],\n topK: number = 5\n ): Promise<t.Highlight[]> {\n this.logger.debug(`Reranking ${documents.length} documents with Cohere`);\n\n try {\n if (this.apiKey == null || this.apiKey === '') {\n this.logger.warn('COHERE_API_KEY is not set. Using default ranking.');\n return this.getDefaultRanking(documents, topK);\n }\n\n this.logDocumentSamples(documents);\n\n const requestData = {\n model: 'rerank-v3.5',\n query: query,\n top_n: topK,\n documents: documents,\n };\n\n const response = await axios.post<t.CohereRerankerResponse | undefined>(\n 'https://api.cohere.com/v2/rerank',\n requestData,\n {\n headers: {\n 'Content-Type': 'application/json',\n Authorization: `Bearer ${this.apiKey}`,\n },\n }\n );\n\n // Log the response data structure\n this.logger.debug('Cohere API response structure:');\n this.logger.debug('ID:', response.data?.id);\n this.logger.debug('Meta:', response.data?.meta);\n this.logger.debug('Results count:', response.data?.results.length);\n\n // Log a sample of the results\n if ((response.data?.results.length ?? 0) > 0) {\n this.logger.debug(\n 'Sample result:',\n JSON.stringify(response.data?.results[0], null, 2)\n );\n }\n\n if (response.data && response.data.results.length) {\n return response.data.results.map((result) => {\n const docIndex = result.index;\n const score = result.relevance_score;\n const text = documents[docIndex];\n return { text, score };\n });\n } else {\n this.logger.warn(\n 'Unexpected response format from Cohere API. Using default ranking.'\n );\n return this.getDefaultRanking(documents, topK);\n }\n } catch (error) {\n this.logger.error('Error using Cohere reranker:', error);\n // Fallback to default ranking on error\n return this.getDefaultRanking(documents, topK);\n }\n }\n}\n\nexport class InfinityReranker extends BaseReranker {\n constructor(logger?: t.Logger) {\n super(logger);\n // No API key needed for the placeholder implementation\n }\n\n async rerank(\n query: string,\n documents: string[],\n topK: number = 5\n ): Promise<t.Highlight[]> {\n this.logger.debug(\n `Reranking ${documents.length} documents with Infinity (placeholder)`\n );\n // This would be replaced with actual Infinity reranker implementation\n return this.getDefaultRanking(documents, topK);\n }\n}\n\n/**\n * Creates the appropriate reranker based on type and configuration\n */\nexport const createReranker = (config: {\n rerankerType: t.RerankerType;\n jinaApiKey?: string;\n cohereApiKey?: string;\n logger?: t.Logger;\n}): BaseReranker | undefined => {\n const { rerankerType, jinaApiKey, cohereApiKey, logger } = config;\n\n // Create a default logger if none is provided\n const defaultLogger = logger || createDefaultLogger();\n\n switch (rerankerType.toLowerCase()) {\n case 'jina':\n return new JinaReranker({ apiKey: jinaApiKey, logger: defaultLogger });\n case 'cohere':\n return new CohereReranker({\n apiKey: cohereApiKey,\n logger: defaultLogger,\n });\n case 'infinity':\n return new InfinityReranker(defaultLogger);\n case 'none':\n defaultLogger.debug('Skipping reranking as reranker is set to \"none\"');\n return undefined;\n default:\n defaultLogger.warn(\n `Unknown reranker type: ${rerankerType}. Defaulting to InfinityReranker.`\n );\n return new JinaReranker({ apiKey: jinaApiKey, logger: defaultLogger });\n }\n};\n\n// Example usage:\n// const jinaReranker = new JinaReranker();\n// const cohereReranker = new CohereReranker();\n// const infinityReranker = new InfinityReranker();\n"],"names":[],"mappings":";;;MAIsB,YAAY,CAAA;AACtB,IAAA,MAAM;AACN,IAAA,MAAM;AAEhB,IAAA,WAAA,CAAY,MAAiB,EAAA;;AAE3B,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM,IAAI,mBAAmB,EAAE;;IASrC,iBAAiB,CACzB,SAAmB,EACnB,IAAY,EAAA;AAEZ,QAAA,OAAO;AACJ,aAAA,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,SAAS,CAAC,MAAM,CAAC;AACzC,aAAA,GAAG,CAAC,CAAC,GAAG,MAAM,EAAE,IAAI,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;;AAGlC,IAAA,kBAAkB,CAAC,SAAmB,EAAA;AAC9C,QAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,qCAAqC,CAAC;QACxD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,SAAS,CAAC,MAAM,CAAC,EAAE,CAAC,EAAE,EAAE;YACtD,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAY,SAAA,EAAA,CAAC,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC,CAAK,GAAA,CAAA,CAAC;;;AAG7E;AAEK,MAAO,YAAa,SAAQ,YAAY,CAAA;IAC5C,WAAY,CAAA,EACV,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,YAAY,EACjC,MAAM,GAIP,EAAA;QACC,KAAK,CAAC,MAAM,CAAC;AACb,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM;;IAGtB,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;QAEhB,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAa,UAAA,EAAA,SAAS,CAAC,MAAM,CAAsB,oBAAA,CAAA,CAAC;AAEtE,QAAA,IAAI;AACF,YAAA,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,EAAE,EAAE;AAC7C,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,iDAAiD,CAAC;gBACnE,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAGhD,YAAA,IAAI,CAAC,kBAAkB,CAAC,SAAS,CAAC;AAElC,YAAA,MAAM,WAAW,GAAG;AAClB,gBAAA,KAAK,EAAE,oCAAoC;AAC3C,gBAAA,KAAK,EAAE,KAAK;AACZ,gBAAA,KAAK,EAAE,IAAI;AACX,gBAAA,SAAS,EAAE,SAAS;AACpB,gBAAA,gBAAgB,EAAE,IAAI;aACvB;YAED,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,+BAA+B,EAC/B,WAAW,EACX;AACE,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;AACF,aAAA,CACF;;AAGD,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,8BAA8B,CAAC;AACjD,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,QAAQ,EAAE,QAAQ,CAAC,IAAI,EAAE,KAAK,CAAC;AACjD,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,QAAQ,EAAE,QAAQ,CAAC,IAAI,EAAE,KAAK,CAAC;AACjD,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,gBAAgB,EAAE,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,CAAC;;AAGlE,YAAA,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;gBAC5C,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,gBAAgB,EAChB,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,CACnD;;AAGH,YAAA,IAAI,QAAQ,CAAC,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE;gBACjD,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,KAAI;AAC1C,oBAAA,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK;AAC7B,oBAAA,MAAM,KAAK,GAAG,MAAM,CAAC,eAAe;oBACpC,IAAI,IAAI,GAAG,EAAE;;AAGb,oBAAA,IAAI,MAAM,CAAC,QAAQ,IAAI,IAAI,EAAE;AAC3B,wBAAA,MAAM,GAAG,GAAG,MAAM,CAAC,QAAQ;wBAC3B,IAAI,OAAO,GAAG,KAAK,QAAQ,IAAI,MAAM,IAAI,GAAG,EAAE;AAC5C,4BAAA,IAAI,GAAG,GAAG,CAAC,IAAI;;AACV,6BAAA,IAAI,OAAO,GAAG,KAAK,QAAQ,EAAE;4BAClC,IAAI,GAAG,GAAG;;;yBAEP;;AAEL,wBAAA,IAAI,GAAG,SAAS,CAAC,QAAQ,CAAC;;AAG5B,oBAAA,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE;AACxB,iBAAC,CAAC;;iBACG;AACL,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,kEAAkE,CACnE;gBACD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;QAEhD,OAAO,KAAK,EAAE;YACd,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,4BAA4B,EAAE,KAAK,CAAC;;YAEtD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;AAGnD;AAEK,MAAO,cAAe,SAAQ,YAAY,CAAA;IAC9C,WAAY,CAAA,EACV,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,cAAc,EACnC,MAAM,GAIP,EAAA;QACC,KAAK,CAAC,MAAM,CAAC;AACb,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM;;IAGtB,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;QAEhB,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAa,UAAA,EAAA,SAAS,CAAC,MAAM,CAAwB,sBAAA,CAAA,CAAC;AAExE,QAAA,IAAI;AACF,YAAA,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,EAAE,EAAE;AAC7C,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,mDAAmD,CAAC;gBACrE,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAGhD,YAAA,IAAI,CAAC,kBAAkB,CAAC,SAAS,CAAC;AAElC,YAAA,MAAM,WAAW,GAAG;AAClB,gBAAA,KAAK,EAAE,aAAa;AACpB,gBAAA,KAAK,EAAE,KAAK;AACZ,gBAAA,KAAK,EAAE,IAAI;AACX,gBAAA,SAAS,EAAE,SAAS;aACrB;YAED,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,kCAAkC,EAClC,WAAW,EACX;AACE,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;AACF,aAAA,CACF;;AAGD,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,gCAAgC,CAAC;AACnD,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,KAAK,EAAE,QAAQ,CAAC,IAAI,EAAE,EAAE,CAAC;AAC3C,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC;AAC/C,YAAA,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,gBAAgB,EAAE,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,CAAC;;AAGlE,YAAA,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;gBAC5C,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,gBAAgB,EAChB,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,CACnD;;AAGH,YAAA,IAAI,QAAQ,CAAC,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE;gBACjD,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,MAAM,KAAI;AAC1C,oBAAA,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK;AAC7B,oBAAA,MAAM,KAAK,GAAG,MAAM,CAAC,eAAe;AACpC,oBAAA,MAAM,IAAI,GAAG,SAAS,CAAC,QAAQ,CAAC;AAChC,oBAAA,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE;AACxB,iBAAC,CAAC;;iBACG;AACL,gBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,oEAAoE,CACrE;gBACD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;QAEhD,OAAO,KAAK,EAAE;YACd,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,8BAA8B,EAAE,KAAK,CAAC;;YAExD,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;;AAGnD;AAEK,MAAO,gBAAiB,SAAQ,YAAY,CAAA;AAChD,IAAA,WAAA,CAAY,MAAiB,EAAA;QAC3B,KAAK,CAAC,MAAM,CAAC;;;IAIf,MAAM,MAAM,CACV,KAAa,EACb,SAAmB,EACnB,OAAe,CAAC,EAAA;QAEhB,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,CAAa,UAAA,EAAA,SAAS,CAAC,MAAM,CAAwC,sCAAA,CAAA,CACtE;;QAED,OAAO,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,IAAI,CAAC;;AAEjD;AAED;;AAEG;AACU,MAAA,cAAc,GAAG,CAAC,MAK9B,KAA8B;IAC7B,MAAM,EAAE,YAAY,EAAE,UAAU,EAAE,YAAY,EAAE,MAAM,EAAE,GAAG,MAAM;;AAGjE,IAAA,MAAM,aAAa,GAAG,MAAM,IAAI,mBAAmB,EAAE;AAErD,IAAA,QAAQ,YAAY,CAAC,WAAW,EAAE;AAClC,QAAA,KAAK,MAAM;AACT,YAAA,OAAO,IAAI,YAAY,CAAC,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,aAAa,EAAE,CAAC;AACxE,QAAA,KAAK,QAAQ;YACX,OAAO,IAAI,cAAc,CAAC;AACxB,gBAAA,MAAM,EAAE,YAAY;AACpB,gBAAA,MAAM,EAAE,aAAa;AACtB,aAAA,CAAC;AACJ,QAAA,KAAK,UAAU;AACb,YAAA,OAAO,IAAI,gBAAgB,CAAC,aAAa,CAAC;AAC5C,QAAA,KAAK,MAAM;AACT,YAAA,aAAa,CAAC,KAAK,CAAC,iDAAiD,CAAC;AACtE,YAAA,OAAO,SAAS;AAClB,QAAA;AACE,YAAA,aAAa,CAAC,IAAI,CAChB,0BAA0B,YAAY,CAAA,iCAAA,CAAmC,CAC1E;AACD,YAAA,OAAO,IAAI,YAAY,CAAC,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,aAAa,EAAE,CAAC;;AAE1E;AAEA;AACA;AACA;AACA;;;;"}
|
|
@@ -0,0 +1,61 @@
|
|
|
1
|
+
import { z } from 'zod';
|
|
2
|
+
|
|
3
|
+
var DATE_RANGE;
|
|
4
|
+
(function (DATE_RANGE) {
|
|
5
|
+
DATE_RANGE["PAST_HOUR"] = "h";
|
|
6
|
+
DATE_RANGE["PAST_24_HOURS"] = "d";
|
|
7
|
+
DATE_RANGE["PAST_WEEK"] = "w";
|
|
8
|
+
DATE_RANGE["PAST_MONTH"] = "m";
|
|
9
|
+
DATE_RANGE["PAST_YEAR"] = "y";
|
|
10
|
+
})(DATE_RANGE || (DATE_RANGE = {}));
|
|
11
|
+
const DEFAULT_QUERY_DESCRIPTION = `
|
|
12
|
+
GUIDELINES:
|
|
13
|
+
- Start broad, then narrow: Begin with key concepts, then refine with specifics
|
|
14
|
+
- Think like sources: Use terminology experts would use in the field
|
|
15
|
+
- Consider perspective: Frame queries from different viewpoints for better results
|
|
16
|
+
- Quality over quantity: A precise 3-4 word query often beats lengthy sentences
|
|
17
|
+
|
|
18
|
+
TECHNIQUES (combine for power searches):
|
|
19
|
+
- EXACT PHRASES: Use quotes ("climate change report")
|
|
20
|
+
- EXCLUDE TERMS: Use minus to remove unwanted results (-wikipedia)
|
|
21
|
+
- SITE-SPECIFIC: Restrict to websites (site:edu research)
|
|
22
|
+
- FILETYPE: Find specific documents (filetype:pdf study)
|
|
23
|
+
- OR OPERATOR: Find alternatives (electric OR hybrid cars)
|
|
24
|
+
- DATE RANGE: Recent information (data after:2020)
|
|
25
|
+
- WILDCARDS: Use * for unknown terms (how to * bread)
|
|
26
|
+
- SPECIFIC QUESTIONS: Use who/what/when/where/why/how
|
|
27
|
+
- DOMAIN TERMS: Include technical terminology for specialized topics
|
|
28
|
+
- CONCISE TERMS: Prioritize keywords over sentences
|
|
29
|
+
`.trim();
|
|
30
|
+
const DEFAULT_COUNTRY_DESCRIPTION = `Country code to localize search results.
|
|
31
|
+
Use standard 2-letter country codes: "us", "uk", "ca", "de", "fr", "jp", "br", etc.
|
|
32
|
+
Provide this when the search should return results specific to a particular country.
|
|
33
|
+
Examples:
|
|
34
|
+
- "us" for United States (default)
|
|
35
|
+
- "de" for Germany
|
|
36
|
+
- "in" for India
|
|
37
|
+
`.trim();
|
|
38
|
+
const querySchema = z.string().describe(DEFAULT_QUERY_DESCRIPTION);
|
|
39
|
+
const dateSchema = z
|
|
40
|
+
.nativeEnum(DATE_RANGE)
|
|
41
|
+
.optional()
|
|
42
|
+
.describe('Date range for search results.');
|
|
43
|
+
const countrySchema = z
|
|
44
|
+
.string()
|
|
45
|
+
.optional()
|
|
46
|
+
.describe(DEFAULT_COUNTRY_DESCRIPTION);
|
|
47
|
+
const imagesSchema = z
|
|
48
|
+
.boolean()
|
|
49
|
+
.optional()
|
|
50
|
+
.describe('Whether to also run an image search.');
|
|
51
|
+
const videosSchema = z
|
|
52
|
+
.boolean()
|
|
53
|
+
.optional()
|
|
54
|
+
.describe('Whether to also run a video search.');
|
|
55
|
+
const newsSchema = z
|
|
56
|
+
.boolean()
|
|
57
|
+
.optional()
|
|
58
|
+
.describe('Whether to also run a news search.');
|
|
59
|
+
|
|
60
|
+
export { DATE_RANGE, DEFAULT_COUNTRY_DESCRIPTION, DEFAULT_QUERY_DESCRIPTION, countrySchema, dateSchema, imagesSchema, newsSchema, querySchema, videosSchema };
|
|
61
|
+
//# sourceMappingURL=schema.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"schema.mjs","sources":["../../../../src/tools/search/schema.ts"],"sourcesContent":["import { z } from 'zod';\n\nexport enum DATE_RANGE {\n PAST_HOUR = 'h',\n PAST_24_HOURS = 'd',\n PAST_WEEK = 'w',\n PAST_MONTH = 'm',\n PAST_YEAR = 'y',\n}\n\nexport const DEFAULT_QUERY_DESCRIPTION = `\nGUIDELINES:\n- Start broad, then narrow: Begin with key concepts, then refine with specifics\n- Think like sources: Use terminology experts would use in the field\n- Consider perspective: Frame queries from different viewpoints for better results\n- Quality over quantity: A precise 3-4 word query often beats lengthy sentences\n\nTECHNIQUES (combine for power searches):\n- EXACT PHRASES: Use quotes (\"climate change report\")\n- EXCLUDE TERMS: Use minus to remove unwanted results (-wikipedia)\n- SITE-SPECIFIC: Restrict to websites (site:edu research)\n- FILETYPE: Find specific documents (filetype:pdf study)\n- OR OPERATOR: Find alternatives (electric OR hybrid cars)\n- DATE RANGE: Recent information (data after:2020)\n- WILDCARDS: Use * for unknown terms (how to * bread)\n- SPECIFIC QUESTIONS: Use who/what/when/where/why/how\n- DOMAIN TERMS: Include technical terminology for specialized topics\n- CONCISE TERMS: Prioritize keywords over sentences\n`.trim();\n\nexport const DEFAULT_COUNTRY_DESCRIPTION =\n `Country code to localize search results.\nUse standard 2-letter country codes: \"us\", \"uk\", \"ca\", \"de\", \"fr\", \"jp\", \"br\", etc.\nProvide this when the search should return results specific to a particular country.\nExamples:\n- \"us\" for United States (default)\n- \"de\" for Germany\n- \"in\" for India\n`.trim();\n\nexport const querySchema = z.string().describe(DEFAULT_QUERY_DESCRIPTION);\nexport const dateSchema = z\n .nativeEnum(DATE_RANGE)\n .optional()\n .describe('Date range for search results.');\nexport const countrySchema = z\n .string()\n .optional()\n .describe(DEFAULT_COUNTRY_DESCRIPTION);\nexport const imagesSchema = z\n .boolean()\n .optional()\n .describe('Whether to also run an image search.');\n\nexport const videosSchema = z\n .boolean()\n .optional()\n .describe('Whether to also run a video search.');\n\nexport const newsSchema = z\n .boolean()\n .optional()\n .describe('Whether to also run a news search.');\n"],"names":[],"mappings":";;IAEY;AAAZ,CAAA,UAAY,UAAU,EAAA;AACpB,IAAA,UAAA,CAAA,WAAA,CAAA,GAAA,GAAe;AACf,IAAA,UAAA,CAAA,eAAA,CAAA,GAAA,GAAmB;AACnB,IAAA,UAAA,CAAA,WAAA,CAAA,GAAA,GAAe;AACf,IAAA,UAAA,CAAA,YAAA,CAAA,GAAA,GAAgB;AAChB,IAAA,UAAA,CAAA,WAAA,CAAA,GAAA,GAAe;AACjB,CAAC,EANW,UAAU,KAAV,UAAU,GAMrB,EAAA,CAAA,CAAA;AAEY,MAAA,yBAAyB,GAAG;;;;;;;;;;;;;;;;;;CAkBxC,CAAC,IAAI;AAEO,MAAA,2BAA2B,GACtC,CAAA;;;;;;;CAOD,CAAC,IAAI;AAEC,MAAM,WAAW,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,yBAAyB;AACjE,MAAM,UAAU,GAAG;KACvB,UAAU,CAAC,UAAU;AACrB,KAAA,QAAQ;KACR,QAAQ,CAAC,gCAAgC;AACrC,MAAM,aAAa,GAAG;AAC1B,KAAA,MAAM;AACN,KAAA,QAAQ;KACR,QAAQ,CAAC,2BAA2B;AAChC,MAAM,YAAY,GAAG;AACzB,KAAA,OAAO;AACP,KAAA,QAAQ;KACR,QAAQ,CAAC,sCAAsC;AAE3C,MAAM,YAAY,GAAG;AACzB,KAAA,OAAO;AACP,KAAA,QAAQ;KACR,QAAQ,CAAC,qCAAqC;AAE1C,MAAM,UAAU,GAAG;AACvB,KAAA,OAAO;AACP,KAAA,QAAQ;KACR,QAAQ,CAAC,oCAAoC;;;;"}
|