@librechat/agents 2.4.320 → 2.4.322
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/tools/search/firecrawl.cjs +6 -4
- package/dist/cjs/tools/search/firecrawl.cjs.map +1 -1
- package/dist/cjs/tools/search/format.cjs +117 -80
- package/dist/cjs/tools/search/format.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 +125 -52
- package/dist/cjs/tools/search/search.cjs.map +1 -1
- package/dist/cjs/tools/search/tool.cjs +162 -47
- package/dist/cjs/tools/search/tool.cjs.map +1 -1
- package/dist/cjs/tools/search/utils.cjs +34 -5
- package/dist/cjs/tools/search/utils.cjs.map +1 -1
- package/dist/esm/tools/search/firecrawl.mjs +6 -4
- package/dist/esm/tools/search/firecrawl.mjs.map +1 -1
- package/dist/esm/tools/search/format.mjs +118 -81
- package/dist/esm/tools/search/format.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 +126 -53
- package/dist/esm/tools/search/search.mjs.map +1 -1
- package/dist/esm/tools/search/tool.mjs +161 -46
- package/dist/esm/tools/search/tool.mjs.map +1 -1
- package/dist/esm/tools/search/utils.mjs +33 -6
- package/dist/esm/tools/search/utils.mjs.map +1 -1
- package/dist/types/tools/search/firecrawl.d.ts +1 -0
- 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/tool.d.ts +13 -0
- package/dist/types/tools/search/types.d.ts +36 -0
- package/dist/types/tools/search/utils.d.ts +9 -2
- package/package.json +3 -2
- package/src/scripts/search.ts +3 -0
- package/src/tools/search/firecrawl.ts +9 -4
- package/src/tools/search/format.ts +157 -87
- package/src/tools/search/rerankers.ts +57 -36
- package/src/tools/search/schema.ts +63 -0
- package/src/tools/search/search.ts +165 -52
- package/src/tools/search/tool.ts +217 -44
- package/src/tools/search/types.ts +37 -0
- package/src/tools/search/utils.ts +37 -5
- package/src/utils/llmConfig.ts +1 -1
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"firecrawl.mjs","sources":["../../../../src/tools/search/firecrawl.ts"],"sourcesContent":["
|
|
1
|
+
{"version":3,"file":"firecrawl.mjs","sources":["../../../../src/tools/search/firecrawl.ts"],"sourcesContent":["import axios from 'axios';\nimport { processContent } from './content';\nimport type * as t from './types';\nimport { createDefaultLogger } from './utils';\n\n/**\n * Firecrawl scraper implementation\n * Uses the Firecrawl API to scrape web pages\n */\nexport class FirecrawlScraper {\n private apiKey: string;\n private apiUrl: string;\n private defaultFormats: string[];\n private timeout: number;\n private logger: t.Logger;\n\n constructor(config: t.FirecrawlScraperConfig = {}) {\n this.apiKey = config.apiKey ?? process.env.FIRECRAWL_API_KEY ?? '';\n\n const baseUrl =\n config.apiUrl ??\n process.env.FIRECRAWL_BASE_URL ??\n 'https://api.firecrawl.dev';\n this.apiUrl = `${baseUrl.replace(/\\/+$/, '')}/v1/scrape`;\n\n this.defaultFormats = config.formats ?? ['markdown', 'html'];\n this.timeout = config.timeout ?? 15000;\n\n this.logger = config.logger || createDefaultLogger();\n\n if (!this.apiKey) {\n this.logger.warn('FIRECRAWL_API_KEY is not set. Scraping will not work.');\n }\n\n this.logger.debug(\n `Firecrawl scraper initialized with API URL: ${this.apiUrl}`\n );\n }\n\n /**\n * Scrape a single URL\n * @param url URL to scrape\n * @param options Scrape options\n * @returns Scrape response\n */\n async scrapeUrl(\n url: string,\n options: t.FirecrawlScrapeOptions = {}\n ): Promise<[string, t.FirecrawlScrapeResponse]> {\n if (!this.apiKey) {\n return [\n url,\n {\n success: false,\n error: 'FIRECRAWL_API_KEY is not set',\n },\n ];\n }\n\n try {\n const response = await axios.post(\n this.apiUrl,\n {\n url,\n formats: options.formats || this.defaultFormats,\n includeTags: options.includeTags,\n excludeTags: options.excludeTags,\n headers: options.headers,\n waitFor: options.waitFor,\n timeout: options.timeout ?? this.timeout,\n },\n {\n headers: {\n 'Content-Type': 'application/json',\n Authorization: `Bearer ${this.apiKey}`,\n },\n timeout: this.timeout,\n }\n );\n\n return [url, response.data];\n } catch (error) {\n const errorMessage =\n error instanceof Error ? error.message : String(error);\n return [\n url,\n {\n success: false,\n error: `Firecrawl API request failed: ${errorMessage}`,\n },\n ];\n }\n }\n\n /**\n * Extract content from scrape response\n * @param response Scrape response\n * @returns Extracted content or empty string if not available\n */\n extractContent(\n response: t.FirecrawlScrapeResponse\n ): [string, undefined | t.References] {\n if (!response.success || !response.data) {\n return ['', undefined];\n }\n\n if (response.data.markdown != null && response.data.html != null) {\n try {\n const { markdown, ...rest } = processContent(\n response.data.html,\n response.data.markdown\n );\n return [markdown, rest];\n } catch (error) {\n this.logger.error('Error processing content:', error);\n return [response.data.markdown, undefined];\n }\n } else if (response.data.markdown != null) {\n return [response.data.markdown, undefined];\n }\n\n // Fall back to HTML content\n if (response.data.html != null) {\n return [response.data.html, undefined];\n }\n\n // Fall back to raw HTML content\n if (response.data.rawHtml != null) {\n return [response.data.rawHtml, undefined];\n }\n\n return ['', undefined];\n }\n\n /**\n * Extract metadata from scrape response\n * @param response Scrape response\n * @returns Metadata object\n */\n extractMetadata(response: t.FirecrawlScrapeResponse): t.ScrapeMetadata {\n if (!response.success || !response.data || !response.data.metadata) {\n return {};\n }\n\n return response.data.metadata;\n }\n}\n\n/**\n * Create a Firecrawl scraper instance\n * @param config Scraper configuration\n * @returns Firecrawl scraper instance\n */\nexport const createFirecrawlScraper = (\n config: t.FirecrawlScraperConfig = {}\n): FirecrawlScraper => {\n return new FirecrawlScraper(config);\n};\n"],"names":[],"mappings":";;;;AAKA;;;AAGG;MACU,gBAAgB,CAAA;AACnB,IAAA,MAAM;AACN,IAAA,MAAM;AACN,IAAA,cAAc;AACd,IAAA,OAAO;AACP,IAAA,MAAM;AAEd,IAAA,WAAA,CAAY,SAAmC,EAAE,EAAA;AAC/C,QAAA,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,IAAI,OAAO,CAAC,GAAG,CAAC,iBAAiB,IAAI,EAAE;AAElE,QAAA,MAAM,OAAO,GACX,MAAM,CAAC,MAAM;YACb,OAAO,CAAC,GAAG,CAAC,kBAAkB;AAC9B,YAAA,2BAA2B;AAC7B,QAAA,IAAI,CAAC,MAAM,GAAG,CAAA,EAAG,OAAO,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,YAAY;AAExD,QAAA,IAAI,CAAC,cAAc,GAAG,MAAM,CAAC,OAAO,IAAI,CAAC,UAAU,EAAE,MAAM,CAAC;QAC5D,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO,IAAI,KAAK;QAEtC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,IAAI,mBAAmB,EAAE;AAEpD,QAAA,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;AAChB,YAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,uDAAuD,CAAC;;QAG3E,IAAI,CAAC,MAAM,CAAC,KAAK,CACf,CAA+C,4CAAA,EAAA,IAAI,CAAC,MAAM,CAAE,CAAA,CAC7D;;AAGH;;;;;AAKG;AACH,IAAA,MAAM,SAAS,CACb,GAAW,EACX,UAAoC,EAAE,EAAA;AAEtC,QAAA,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;YAChB,OAAO;gBACL,GAAG;AACH,gBAAA;AACE,oBAAA,OAAO,EAAE,KAAK;AACd,oBAAA,KAAK,EAAE,8BAA8B;AACtC,iBAAA;aACF;;AAGH,QAAA,IAAI;YACF,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,IAAI,CAAC,MAAM,EACX;gBACE,GAAG;AACH,gBAAA,OAAO,EAAE,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,cAAc;gBAC/C,WAAW,EAAE,OAAO,CAAC,WAAW;gBAChC,WAAW,EAAE,OAAO,CAAC,WAAW;gBAChC,OAAO,EAAE,OAAO,CAAC,OAAO;gBACxB,OAAO,EAAE,OAAO,CAAC,OAAO;AACxB,gBAAA,OAAO,EAAE,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO;aACzC,EACD;AACE,gBAAA,OAAO,EAAE;AACP,oBAAA,cAAc,EAAE,kBAAkB;AAClC,oBAAA,aAAa,EAAE,CAAA,OAAA,EAAU,IAAI,CAAC,MAAM,CAAE,CAAA;AACvC,iBAAA;gBACD,OAAO,EAAE,IAAI,CAAC,OAAO;AACtB,aAAA,CACF;AAED,YAAA,OAAO,CAAC,GAAG,EAAE,QAAQ,CAAC,IAAI,CAAC;;QAC3B,OAAO,KAAK,EAAE;AACd,YAAA,MAAM,YAAY,GAChB,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;YACxD,OAAO;gBACL,GAAG;AACH,gBAAA;AACE,oBAAA,OAAO,EAAE,KAAK;oBACd,KAAK,EAAE,CAAiC,8BAAA,EAAA,YAAY,CAAE,CAAA;AACvD,iBAAA;aACF;;;AAIL;;;;AAIG;AACH,IAAA,cAAc,CACZ,QAAmC,EAAA;QAEnC,IAAI,CAAC,QAAQ,CAAC,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE;AACvC,YAAA,OAAO,CAAC,EAAE,EAAE,SAAS,CAAC;;AAGxB,QAAA,IAAI,QAAQ,CAAC,IAAI,CAAC,QAAQ,IAAI,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,EAAE;AAChE,YAAA,IAAI;gBACF,MAAM,EAAE,QAAQ,EAAE,GAAG,IAAI,EAAE,GAAG,cAAc,CAC1C,QAAQ,CAAC,IAAI,CAAC,IAAI,EAClB,QAAQ,CAAC,IAAI,CAAC,QAAQ,CACvB;AACD,gBAAA,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC;;YACvB,OAAO,KAAK,EAAE;gBACd,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,2BAA2B,EAAE,KAAK,CAAC;gBACrD,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,SAAS,CAAC;;;aAEvC,IAAI,QAAQ,CAAC,IAAI,CAAC,QAAQ,IAAI,IAAI,EAAE;YACzC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,SAAS,CAAC;;;QAI5C,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,EAAE;YAC9B,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE,SAAS,CAAC;;;QAIxC,IAAI,QAAQ,CAAC,IAAI,CAAC,OAAO,IAAI,IAAI,EAAE;YACjC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,EAAE,SAAS,CAAC;;AAG3C,QAAA,OAAO,CAAC,EAAE,EAAE,SAAS,CAAC;;AAGxB;;;;AAIG;AACH,IAAA,eAAe,CAAC,QAAmC,EAAA;AACjD,QAAA,IAAI,CAAC,QAAQ,CAAC,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE;AAClE,YAAA,OAAO,EAAE;;AAGX,QAAA,OAAO,QAAQ,CAAC,IAAI,CAAC,QAAQ;;AAEhC;AAED;;;;AAIG;MACU,sBAAsB,GAAG,CACpC,MAAmC,GAAA,EAAE,KACjB;AACpB,IAAA,OAAO,IAAI,gBAAgB,CAAC,MAAM,CAAC;AACrC;;;;"}
|
|
@@ -1,9 +1,91 @@
|
|
|
1
|
-
import { getDomainName } from './utils.mjs';
|
|
1
|
+
import { getDomainName, fileExtRegex } from './utils.mjs';
|
|
2
2
|
|
|
3
|
+
function addHighlightSection() {
|
|
4
|
+
return ['\n## Highlights', ''];
|
|
5
|
+
}
|
|
6
|
+
// Helper function to format a source (organic or top story)
|
|
7
|
+
function formatSource(source, index, turn, sourceType, references) {
|
|
8
|
+
/** Array of all lines to include in the output */
|
|
9
|
+
const outputLines = [];
|
|
10
|
+
// Add the title
|
|
11
|
+
outputLines.push(`# ${sourceType.charAt(0).toUpperCase() + sourceType.slice(1)} ${index}: ${source.title != null && source.title ? `"${source.title}"` : '(no title)'}`);
|
|
12
|
+
outputLines.push(`\nAnchor: \\ue202turn${turn}${sourceType}${index}`);
|
|
13
|
+
outputLines.push(`URL: ${source.link}`);
|
|
14
|
+
// Add optional fields
|
|
15
|
+
if ('snippet' in source && source.snippet != null) {
|
|
16
|
+
outputLines.push(`Summary: ${source.snippet}`);
|
|
17
|
+
}
|
|
18
|
+
if (source.date != null) {
|
|
19
|
+
outputLines.push(`Date: ${source.date}`);
|
|
20
|
+
}
|
|
21
|
+
if (source.attribution != null) {
|
|
22
|
+
outputLines.push(`Source: ${source.attribution}`);
|
|
23
|
+
}
|
|
24
|
+
// Add highlight section or empty line
|
|
25
|
+
if ((source.highlights?.length ?? 0) > 0) {
|
|
26
|
+
outputLines.push(...addHighlightSection());
|
|
27
|
+
}
|
|
28
|
+
else {
|
|
29
|
+
outputLines.push('');
|
|
30
|
+
}
|
|
31
|
+
// Process highlights if they exist
|
|
32
|
+
(source.highlights ?? [])
|
|
33
|
+
.filter((h) => h.text.trim().length > 0)
|
|
34
|
+
.forEach((h, hIndex) => {
|
|
35
|
+
outputLines.push(`### Highlight ${hIndex + 1} [Relevance: ${h.score.toFixed(2)}]`);
|
|
36
|
+
outputLines.push('');
|
|
37
|
+
outputLines.push('```text');
|
|
38
|
+
outputLines.push(h.text.trim());
|
|
39
|
+
outputLines.push('```');
|
|
40
|
+
outputLines.push('');
|
|
41
|
+
if (h.references != null && h.references.length) {
|
|
42
|
+
let hasHeader = false;
|
|
43
|
+
const refLines = [];
|
|
44
|
+
for (let j = 0; j < h.references.length; j++) {
|
|
45
|
+
const ref = h.references[j];
|
|
46
|
+
if (ref.reference.originalUrl.includes('mailto:')) {
|
|
47
|
+
continue;
|
|
48
|
+
}
|
|
49
|
+
references.push({
|
|
50
|
+
type: ref.type,
|
|
51
|
+
link: ref.reference.originalUrl,
|
|
52
|
+
attribution: getDomainName(ref.reference.originalUrl),
|
|
53
|
+
title: (((ref.reference.title ?? '') || ref.reference.text) ??
|
|
54
|
+
'').split('\n')[0],
|
|
55
|
+
});
|
|
56
|
+
if (ref.type !== 'link') {
|
|
57
|
+
continue;
|
|
58
|
+
}
|
|
59
|
+
if (fileExtRegex.test(ref.reference.originalUrl)) {
|
|
60
|
+
continue;
|
|
61
|
+
}
|
|
62
|
+
if (!hasHeader) {
|
|
63
|
+
refLines.push('Core References:');
|
|
64
|
+
hasHeader = true;
|
|
65
|
+
}
|
|
66
|
+
refLines.push(`- ${ref.type}#${ref.originalIndex + 1}: ${ref.reference.originalUrl}`);
|
|
67
|
+
refLines.push(`\t- Anchor: \\ue202turn${turn}ref${references.length - 1}`);
|
|
68
|
+
}
|
|
69
|
+
if (hasHeader) {
|
|
70
|
+
outputLines.push(...refLines);
|
|
71
|
+
outputLines.push('');
|
|
72
|
+
}
|
|
73
|
+
}
|
|
74
|
+
if (hIndex < (source.highlights?.length ?? 0) - 1) {
|
|
75
|
+
outputLines.push('---');
|
|
76
|
+
outputLines.push('');
|
|
77
|
+
}
|
|
78
|
+
});
|
|
79
|
+
outputLines.push('');
|
|
80
|
+
return outputLines.join('\n');
|
|
81
|
+
}
|
|
3
82
|
function formatResultsForLLM(turn, results) {
|
|
4
|
-
|
|
83
|
+
/** Array to collect all output lines */
|
|
84
|
+
const outputLines = [];
|
|
5
85
|
const addSection = (title) => {
|
|
6
|
-
|
|
86
|
+
outputLines.push('');
|
|
87
|
+
outputLines.push(`=== ${title} ===`);
|
|
88
|
+
outputLines.push('');
|
|
7
89
|
};
|
|
8
90
|
const references = [];
|
|
9
91
|
// Organic (web) results
|
|
@@ -11,82 +93,38 @@ function formatResultsForLLM(turn, results) {
|
|
|
11
93
|
addSection(`Web Results, Turn ${turn}`);
|
|
12
94
|
for (let i = 0; i < results.organic.length; i++) {
|
|
13
95
|
const r = results.organic[i];
|
|
14
|
-
|
|
15
|
-
`# Source ${i}: "${r.title ?? '(no title)'}"`,
|
|
16
|
-
`Anchor: \\ue202turn${turn}search${i}`,
|
|
17
|
-
`URL: ${r.link}`,
|
|
18
|
-
r.snippet != null ? `Summary: ${r.snippet}` : '',
|
|
19
|
-
r.date != null ? `Date: ${r.date}` : '',
|
|
20
|
-
r.attribution != null ? `Source: ${r.attribution}` : '',
|
|
21
|
-
'',
|
|
22
|
-
'\n## Highlights\n\n',
|
|
23
|
-
'',
|
|
24
|
-
'',
|
|
25
|
-
]
|
|
26
|
-
.filter(Boolean)
|
|
27
|
-
.join('\n');
|
|
28
|
-
(r.highlights ?? [])
|
|
29
|
-
.filter((h) => h.text.trim().length > 0)
|
|
30
|
-
.forEach((h, hIndex) => {
|
|
31
|
-
output += `### Highlight ${hIndex + 1} [Relevance: ${h.score.toFixed(2)}]\n\n`;
|
|
32
|
-
output += '```text\n' + h.text.trim() + '\n```\n\n';
|
|
33
|
-
if (h.references != null && h.references.length) {
|
|
34
|
-
output += 'Core References:\n';
|
|
35
|
-
output += h.references
|
|
36
|
-
.map((ref) => {
|
|
37
|
-
references.push({
|
|
38
|
-
link: ref.reference.originalUrl,
|
|
39
|
-
attribution: getDomainName(ref.reference.originalUrl),
|
|
40
|
-
title: (((ref.reference.title ?? '') || ref.reference.text) ??
|
|
41
|
-
'').split('\n')[0],
|
|
42
|
-
});
|
|
43
|
-
return `- ${ref.type}#${ref.originalIndex + 1}: ${ref.reference.originalUrl}\n\t- Anchor: \\ue202turn${turn}ref${references.length - 1}`;
|
|
44
|
-
})
|
|
45
|
-
.join('\n');
|
|
46
|
-
output += '\n\n';
|
|
47
|
-
}
|
|
48
|
-
if (hIndex < (r.highlights?.length ?? 0) - 1) {
|
|
49
|
-
output += '---\n\n';
|
|
50
|
-
}
|
|
51
|
-
});
|
|
96
|
+
outputLines.push(formatSource(r, i, turn, 'search', references));
|
|
52
97
|
delete results.organic[i].highlights;
|
|
53
|
-
output += '\n';
|
|
54
98
|
}
|
|
55
99
|
}
|
|
56
|
-
//
|
|
57
|
-
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
|
|
61
|
-
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
// r.attribution != null ? `Source: ${r.attribution}` : '',
|
|
69
|
-
// ''
|
|
70
|
-
// ].filter(Boolean).join('\n');
|
|
71
|
-
// });
|
|
72
|
-
// }
|
|
100
|
+
// Top stories (news)
|
|
101
|
+
const topStories = results.topStories ?? [];
|
|
102
|
+
if (topStories.length) {
|
|
103
|
+
addSection('News Results');
|
|
104
|
+
for (let i = 0; i < topStories.length; i++) {
|
|
105
|
+
const r = topStories[i];
|
|
106
|
+
outputLines.push(formatSource(r, i, turn, 'news', references));
|
|
107
|
+
if (results.topStories?.[i]?.highlights) {
|
|
108
|
+
delete results.topStories[i].highlights;
|
|
109
|
+
}
|
|
110
|
+
}
|
|
111
|
+
}
|
|
73
112
|
// // Images
|
|
74
113
|
// const images = results.images ?? [];
|
|
75
114
|
// if (images.length) {
|
|
76
115
|
// addSection('Image Results');
|
|
77
|
-
// images.
|
|
78
|
-
//
|
|
79
|
-
//
|
|
80
|
-
//
|
|
81
|
-
//
|
|
82
|
-
//
|
|
83
|
-
//
|
|
84
|
-
// });
|
|
116
|
+
// const imageLines = images.map((img, i) => [
|
|
117
|
+
// `Anchor: \ue202turn0image${i}`,
|
|
118
|
+
// `Title: ${img.title ?? '(no title)'}`,
|
|
119
|
+
// `Image URL: ${img.imageUrl}`,
|
|
120
|
+
// ''
|
|
121
|
+
// ].join('\n'));
|
|
122
|
+
// outputLines.push(imageLines.join('\n'));
|
|
85
123
|
// }
|
|
86
124
|
// Knowledge Graph
|
|
87
125
|
if (results.knowledgeGraph != null) {
|
|
88
126
|
addSection('Knowledge Graph');
|
|
89
|
-
|
|
127
|
+
const kgLines = [
|
|
90
128
|
`**Title:** ${results.knowledgeGraph.title ?? '(no title)'}`,
|
|
91
129
|
results.knowledgeGraph.type != null
|
|
92
130
|
? `**Type:** ${results.knowledgeGraph.type}`
|
|
@@ -110,14 +148,13 @@ function formatResultsForLLM(turn, results) {
|
|
|
110
148
|
? `**Attributes:**\n\`\`\`json\n${JSON.stringify(results.knowledgeGraph.attributes, null, 2)}\n\`\`\``
|
|
111
149
|
: '',
|
|
112
150
|
'',
|
|
113
|
-
]
|
|
114
|
-
|
|
115
|
-
.join('\n\n');
|
|
151
|
+
].filter(Boolean);
|
|
152
|
+
outputLines.push(kgLines.join('\n\n'));
|
|
116
153
|
}
|
|
117
154
|
// Answer Box
|
|
118
155
|
if (results.answerBox != null) {
|
|
119
156
|
addSection('Answer Box');
|
|
120
|
-
|
|
157
|
+
const abLines = [
|
|
121
158
|
results.answerBox.title != null
|
|
122
159
|
? `**Title:** ${results.answerBox.title}`
|
|
123
160
|
: '',
|
|
@@ -133,29 +170,29 @@ function formatResultsForLLM(turn, results) {
|
|
|
133
170
|
? `**Link:** ${results.answerBox.link}`
|
|
134
171
|
: '',
|
|
135
172
|
'',
|
|
136
|
-
]
|
|
137
|
-
|
|
138
|
-
.join('\n\n');
|
|
173
|
+
].filter(Boolean);
|
|
174
|
+
outputLines.push(abLines.join('\n\n'));
|
|
139
175
|
}
|
|
140
176
|
// People also ask
|
|
141
177
|
const peopleAlsoAsk = results.peopleAlsoAsk ?? [];
|
|
142
178
|
if (peopleAlsoAsk.length) {
|
|
143
179
|
addSection('People Also Ask');
|
|
180
|
+
const paaLines = [];
|
|
144
181
|
peopleAlsoAsk.forEach((p, i) => {
|
|
145
|
-
|
|
182
|
+
const questionLines = [
|
|
146
183
|
`### Question ${i + 1}:`,
|
|
147
184
|
`"${p.question}"`,
|
|
148
|
-
`${p.snippet != null && p.snippet ? `Snippet: ${p.snippet}
|
|
185
|
+
`${p.snippet != null && p.snippet ? `Snippet: ${p.snippet}` : ''}`,
|
|
149
186
|
`${p.title != null && p.title ? `Title: ${p.title}` : ''}`,
|
|
150
187
|
`${p.link != null && p.link ? `Link: ${p.link}` : ''}`,
|
|
151
188
|
'',
|
|
152
|
-
]
|
|
153
|
-
|
|
154
|
-
.join('\n\n');
|
|
189
|
+
].filter(Boolean);
|
|
190
|
+
paaLines.push(questionLines.join('\n\n'));
|
|
155
191
|
});
|
|
192
|
+
outputLines.push(paaLines.join(''));
|
|
156
193
|
}
|
|
157
194
|
return {
|
|
158
|
-
output:
|
|
195
|
+
output: outputLines.join('\n').trim(),
|
|
159
196
|
references,
|
|
160
197
|
};
|
|
161
198
|
}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"format.mjs","sources":["../../../../src/tools/search/format.ts"],"sourcesContent":["import type * as t from './types';\nimport { getDomainName } from './utils';\n\nexport function formatResultsForLLM(\n turn: number,\n results: t.SearchResultData\n): { output: string; references: t.ResultReference[] } {\n let output = '';\n\n const addSection = (title: string): void => {\n output += `\\n=== ${title} ===\\n`;\n };\n\n const references: t.ResultReference[] = [];\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 output += [\n `# Source ${i}: \"${r.title ?? '(no title)'}\"`,\n `Anchor: \\\\ue202turn${turn}search${i}`,\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 '\\n## Highlights\\n\\n',\n '',\n '',\n ]\n .filter(Boolean)\n .join('\\n');\n\n (r.highlights ?? [])\n .filter((h) => h.text.trim().length > 0)\n .forEach((h, hIndex) => {\n output += `### Highlight ${hIndex + 1} [Relevance: ${h.score.toFixed(2)}]\\n\\n`;\n output += '```text\\n' + h.text.trim() + '\\n```\\n\\n';\n\n if (h.references != null && h.references.length) {\n output += 'Core References:\\n';\n output += h.references\n .map((ref) => {\n references.push({\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 return `- ${ref.type}#${ref.originalIndex + 1}: ${ref.reference.originalUrl}\\n\\t- Anchor: \\\\ue202turn${turn}ref${references.length - 1}`;\n })\n .join('\\n');\n output += '\\n\\n';\n }\n\n if (hIndex < (r.highlights?.length ?? 0) - 1) {\n output += '---\\n\\n';\n }\n });\n\n delete results.organic[i].highlights;\n output += '\\n';\n }\n }\n\n // Ignoring these sections for now\n // // Top stories (news)\n // const topStores = results.topStories ?? [];\n // if (topStores.length) {\n // addSection('News Results');\n // topStores.forEach((r, i) => {\n // output += [\n // `Anchor: \\ue202turn0news${i}`,\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}`,\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.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 ]\n .filter(Boolean)\n .join('\\n\\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.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 ]\n .filter(Boolean)\n .join('\\n\\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 += [\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 ]\n .filter(Boolean)\n .join('\\n\\n');\n });\n }\n return {\n output: output.trim(),\n references,\n };\n}\n"],"names":[],"mappings":";;AAGgB,SAAA,mBAAmB,CACjC,IAAY,EACZ,OAA2B,EAAA;IAE3B,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;IAED,MAAM,UAAU,GAAwB,EAAE;;AAE1C,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,MAAM,IAAI;AACR,gBAAA,CAAA,SAAA,EAAY,CAAC,CAAM,GAAA,EAAA,CAAC,CAAC,KAAK,IAAI,YAAY,CAAG,CAAA,CAAA;gBAC7C,CAAsB,mBAAA,EAAA,IAAI,CAAS,MAAA,EAAA,CAAC,CAAE,CAAA;gBACtC,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,qBAAqB;gBACrB,EAAE;gBACF,EAAE;AACH;iBACE,MAAM,CAAC,OAAO;iBACd,IAAI,CAAC,IAAI,CAAC;AAEb,YAAA,CAAC,CAAC,CAAC,UAAU,IAAI,EAAE;AAChB,iBAAA,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;AACtC,iBAAA,OAAO,CAAC,CAAC,CAAC,EAAE,MAAM,KAAI;AACrB,gBAAA,MAAM,IAAI,CAAA,cAAA,EAAiB,MAAM,GAAG,CAAC,CAAgB,aAAA,EAAA,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO;gBAC9E,MAAM,IAAI,WAAW,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,GAAG,WAAW;AAEnD,gBAAA,IAAI,CAAC,CAAC,UAAU,IAAI,IAAI,IAAI,CAAC,CAAC,UAAU,CAAC,MAAM,EAAE;oBAC/C,MAAM,IAAI,oBAAoB;oBAC9B,MAAM,IAAI,CAAC,CAAC;AACT,yBAAA,GAAG,CAAC,CAAC,GAAG,KAAI;wBACX,UAAU,CAAC,IAAI,CAAC;AACd,4BAAA,IAAI,EAAE,GAAG,CAAC,SAAS,CAAC,WAAW;4BAC/B,WAAW,EAAE,aAAa,CAAC,GAAG,CAAC,SAAS,CAAC,WAAW,CAAC;AACrD,4BAAA,KAAK,EAAE,CACL,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,IAAI,EAAE,KAAK,GAAG,CAAC,SAAS,CAAC,IAAI;gCAClD,EAAE,EACF,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;AACjB,yBAAA,CAAC;wBACF,OAAO,CAAA,EAAA,EAAK,GAAG,CAAC,IAAI,CAAA,CAAA,EAAI,GAAG,CAAC,aAAa,GAAG,CAAC,CAAK,EAAA,EAAA,GAAG,CAAC,SAAS,CAAC,WAAW,CAAA,yBAAA,EAA4B,IAAI,CAAA,GAAA,EAAM,UAAU,CAAC,MAAM,GAAG,CAAC,CAAA,CAAE;AAC1I,qBAAC;yBACA,IAAI,CAAC,IAAI,CAAC;oBACb,MAAM,IAAI,MAAM;;AAGlB,gBAAA,IAAI,MAAM,GAAG,CAAC,CAAC,CAAC,UAAU,EAAE,MAAM,IAAI,CAAC,IAAI,CAAC,EAAE;oBAC5C,MAAM,IAAI,SAAS;;AAEvB,aAAC,CAAC;YAEJ,OAAO,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,UAAU;YACpC,MAAM,IAAI,IAAI;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAqClB,IAAA,IAAI,OAAO,CAAC,cAAc,IAAI,IAAI,EAAE;QAClC,UAAU,CAAC,iBAAiB,CAAC;AAC7B,QAAA,MAAM,IAAI;AACR,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;aACE,MAAM,CAAC,OAAO;aACd,IAAI,CAAC,MAAM,CAAC;;;AAIjB,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,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;aACE,MAAM,CAAC,OAAO;aACd,IAAI,CAAC,MAAM,CAAC;;;AAIjB,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,CAAC,KAAI;AAC7B,YAAA,MAAM,IAAI;gBACR,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,GAAG,GAAG,EAAE,CAAE,CAAA;gBACnE,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;iBACE,MAAM,CAAC,OAAO;iBACd,IAAI,CAAC,MAAM,CAAC;AACjB,SAAC,CAAC;;IAEJ,OAAO;AACL,QAAA,MAAM,EAAE,MAAM,CAAC,IAAI,EAAE;QACrB,UAAU;KACX;AACH;;;;"}
|
|
1
|
+
{"version":3,"file":"format.mjs","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":[],"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,EAAE,aAAa,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,IAAI,YAAY,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;;;;"}
|
|
@@ -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;;;;"}
|