@staticn0va/wigolo 0.4.0 → 0.5.1

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.
Files changed (135) hide show
  1. package/SKILL.md +97 -50
  2. package/dist/agent/executor.d.ts +13 -0
  3. package/dist/agent/executor.d.ts.map +1 -0
  4. package/dist/agent/executor.js +128 -0
  5. package/dist/agent/executor.js.map +1 -0
  6. package/dist/agent/pipeline.d.ts +5 -0
  7. package/dist/agent/pipeline.d.ts.map +1 -0
  8. package/dist/agent/pipeline.js +198 -0
  9. package/dist/agent/pipeline.js.map +1 -0
  10. package/dist/agent/planner.d.ts +9 -0
  11. package/dist/agent/planner.d.ts.map +1 -0
  12. package/dist/agent/planner.js +190 -0
  13. package/dist/agent/planner.js.map +1 -0
  14. package/dist/cache/db.d.ts.map +1 -1
  15. package/dist/cache/db.js +32 -0
  16. package/dist/cache/db.js.map +1 -1
  17. package/dist/cache/store.d.ts +14 -0
  18. package/dist/cache/store.d.ts.map +1 -1
  19. package/dist/cache/store.js +69 -0
  20. package/dist/cache/store.js.map +1 -1
  21. package/dist/cli/warmup.d.ts +4 -0
  22. package/dist/cli/warmup.d.ts.map +1 -1
  23. package/dist/cli/warmup.js +58 -0
  24. package/dist/cli/warmup.js.map +1 -1
  25. package/dist/config.d.ts +8 -0
  26. package/dist/config.d.ts.map +1 -1
  27. package/dist/config.js +8 -0
  28. package/dist/config.js.map +1 -1
  29. package/dist/embedding/embed.d.ts +19 -0
  30. package/dist/embedding/embed.d.ts.map +1 -0
  31. package/dist/embedding/embed.js +131 -0
  32. package/dist/embedding/embed.js.map +1 -0
  33. package/dist/embedding/key-terms.d.ts +12 -0
  34. package/dist/embedding/key-terms.d.ts.map +1 -0
  35. package/dist/embedding/key-terms.js +138 -0
  36. package/dist/embedding/key-terms.js.map +1 -0
  37. package/dist/embedding/subprocess.d.ts +31 -0
  38. package/dist/embedding/subprocess.d.ts.map +1 -0
  39. package/dist/embedding/subprocess.js +213 -0
  40. package/dist/embedding/subprocess.js.map +1 -0
  41. package/dist/embedding/vector-index.d.ts +26 -0
  42. package/dist/embedding/vector-index.d.ts.map +1 -0
  43. package/dist/embedding/vector-index.js +78 -0
  44. package/dist/embedding/vector-index.js.map +1 -0
  45. package/dist/fetch/browser-pool.d.ts.map +1 -1
  46. package/dist/fetch/browser-pool.js +61 -0
  47. package/dist/fetch/browser-pool.js.map +1 -1
  48. package/dist/fetch/browser-types.js +1 -1
  49. package/dist/fetch/browser-types.js.map +1 -1
  50. package/dist/fetch/lightpanda.d.ts +28 -0
  51. package/dist/fetch/lightpanda.d.ts.map +1 -0
  52. package/dist/fetch/lightpanda.js +177 -0
  53. package/dist/fetch/lightpanda.js.map +1 -0
  54. package/dist/instructions.d.ts +9 -6
  55. package/dist/instructions.d.ts.map +1 -1
  56. package/dist/instructions.js +95 -20
  57. package/dist/instructions.js.map +1 -1
  58. package/dist/logger.d.ts +1 -1
  59. package/dist/logger.d.ts.map +1 -1
  60. package/dist/repl/commands/agent.d.ts +5 -0
  61. package/dist/repl/commands/agent.d.ts.map +1 -0
  62. package/dist/repl/commands/agent.js +48 -0
  63. package/dist/repl/commands/agent.js.map +1 -0
  64. package/dist/repl/commands/find-similar.d.ts +5 -0
  65. package/dist/repl/commands/find-similar.d.ts.map +1 -0
  66. package/dist/repl/commands/find-similar.js +61 -0
  67. package/dist/repl/commands/find-similar.js.map +1 -0
  68. package/dist/repl/commands/research.d.ts +5 -0
  69. package/dist/repl/commands/research.d.ts.map +1 -0
  70. package/dist/repl/commands/research.js +50 -0
  71. package/dist/repl/commands/research.js.map +1 -0
  72. package/dist/repl/formatters.d.ts +4 -1
  73. package/dist/repl/formatters.d.ts.map +1 -1
  74. package/dist/repl/formatters.js +73 -0
  75. package/dist/repl/formatters.js.map +1 -1
  76. package/dist/repl/shell.d.ts.map +1 -1
  77. package/dist/repl/shell.js +22 -1
  78. package/dist/repl/shell.js.map +1 -1
  79. package/dist/research/decompose.d.ts +7 -0
  80. package/dist/research/decompose.d.ts.map +1 -0
  81. package/dist/research/decompose.js +195 -0
  82. package/dist/research/decompose.js.map +1 -0
  83. package/dist/research/pipeline.d.ts +5 -0
  84. package/dist/research/pipeline.d.ts.map +1 -0
  85. package/dist/research/pipeline.js +135 -0
  86. package/dist/research/pipeline.js.map +1 -0
  87. package/dist/research/synthesize.d.ts +10 -0
  88. package/dist/research/synthesize.d.ts.map +1 -0
  89. package/dist/research/synthesize.js +119 -0
  90. package/dist/research/synthesize.js.map +1 -0
  91. package/dist/search/answer-synthesis.d.ts +13 -0
  92. package/dist/search/answer-synthesis.d.ts.map +1 -0
  93. package/dist/search/answer-synthesis.js +120 -0
  94. package/dist/search/answer-synthesis.js.map +1 -0
  95. package/dist/search/find-similar.d.ts +5 -0
  96. package/dist/search/find-similar.d.ts.map +1 -0
  97. package/dist/search/find-similar.js +329 -0
  98. package/dist/search/find-similar.js.map +1 -0
  99. package/dist/search/multi-query.d.ts +22 -0
  100. package/dist/search/multi-query.d.ts.map +1 -0
  101. package/dist/search/multi-query.js +157 -0
  102. package/dist/search/multi-query.js.map +1 -0
  103. package/dist/search/rrf.d.ts +17 -0
  104. package/dist/search/rrf.d.ts.map +1 -0
  105. package/dist/search/rrf.js +48 -0
  106. package/dist/search/rrf.js.map +1 -0
  107. package/dist/search/sampling.d.ts +25 -0
  108. package/dist/search/sampling.d.ts.map +1 -0
  109. package/dist/search/sampling.js +52 -0
  110. package/dist/search/sampling.js.map +1 -0
  111. package/dist/server.d.ts.map +1 -1
  112. package/dist/server.js +165 -4
  113. package/dist/server.js.map +1 -1
  114. package/dist/tools/agent.d.ts +5 -0
  115. package/dist/tools/agent.d.ts.map +1 -0
  116. package/dist/tools/agent.js +67 -0
  117. package/dist/tools/agent.js.map +1 -0
  118. package/dist/tools/fetch.d.ts.map +1 -1
  119. package/dist/tools/fetch.js +10 -0
  120. package/dist/tools/fetch.js.map +1 -1
  121. package/dist/tools/find-similar.d.ts +5 -0
  122. package/dist/tools/find-similar.d.ts.map +1 -0
  123. package/dist/tools/find-similar.js +48 -0
  124. package/dist/tools/find-similar.js.map +1 -0
  125. package/dist/tools/research.d.ts +5 -0
  126. package/dist/tools/research.d.ts.map +1 -0
  127. package/dist/tools/research.js +50 -0
  128. package/dist/tools/research.js.map +1 -0
  129. package/dist/tools/search.d.ts +2 -1
  130. package/dist/tools/search.d.ts.map +1 -1
  131. package/dist/tools/search.js +169 -13
  132. package/dist/tools/search.js.map +1 -1
  133. package/dist/types.d.ts +100 -3
  134. package/dist/types.d.ts.map +1 -1
  135. package/package.json +9 -3
@@ -0,0 +1,19 @@
1
+ import { VectorIndex, type SimilarResult } from './vector-index.js';
2
+ export declare class EmbeddingService {
3
+ private subprocess;
4
+ private index;
5
+ private available;
6
+ private initialized;
7
+ constructor();
8
+ init(): Promise<void>;
9
+ isAvailable(): boolean;
10
+ setAvailable(value: boolean): void;
11
+ getIndex(): VectorIndex;
12
+ embedAndStore(url: string, markdown: string): Promise<void>;
13
+ embedAsync(url: string, markdown: string): void;
14
+ findSimilar(queryText: string, topK: number, excludeUrls?: Set<string>): Promise<SimilarResult[]>;
15
+ shutdown(): void;
16
+ }
17
+ export declare function getEmbeddingService(): EmbeddingService;
18
+ export declare function resetEmbeddingService(): void;
19
+ //# sourceMappingURL=embed.d.ts.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"embed.d.ts","sourceRoot":"","sources":["../../src/embedding/embed.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,WAAW,EAAE,KAAK,aAAa,EAAE,MAAM,mBAAmB,CAAC;AAOpE,qBAAa,gBAAgB;IAC3B,OAAO,CAAC,UAAU,CAAsB;IACxC,OAAO,CAAC,KAAK,CAAc;IAC3B,OAAO,CAAC,SAAS,CAAS;IAC1B,OAAO,CAAC,WAAW,CAAS;;IAOtB,IAAI,IAAI,OAAO,CAAC,IAAI,CAAC;IAuB3B,WAAW,IAAI,OAAO;IAItB,YAAY,CAAC,KAAK,EAAE,OAAO,GAAG,IAAI;IAIlC,QAAQ,IAAI,WAAW;IAIjB,aAAa,CAAC,GAAG,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAoCjE,UAAU,CAAC,GAAG,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,IAAI;IAQzC,WAAW,CACf,SAAS,EAAE,MAAM,EACjB,IAAI,EAAE,MAAM,EACZ,WAAW,CAAC,EAAE,GAAG,CAAC,MAAM,CAAC,GACxB,OAAO,CAAC,aAAa,EAAE,CAAC;IAsB3B,QAAQ,IAAI,IAAI;CAWjB;AAID,wBAAgB,mBAAmB,IAAI,gBAAgB,CAKtD;AAED,wBAAgB,qBAAqB,IAAI,IAAI,CAK5C"}
@@ -0,0 +1,131 @@
1
+ import { randomUUID } from 'node:crypto';
2
+ import { EmbeddingSubprocess } from './subprocess.js';
3
+ import { VectorIndex } from './vector-index.js';
4
+ import { updateCacheEmbedding, getAllEmbeddings, normalizeUrl } from '../cache/store.js';
5
+ import { getConfig } from '../config.js';
6
+ import { createLogger } from '../logger.js';
7
+ const log = createLogger('embedding');
8
+ export class EmbeddingService {
9
+ subprocess;
10
+ index;
11
+ available = false;
12
+ initialized = false;
13
+ constructor() {
14
+ this.subprocess = new EmbeddingSubprocess();
15
+ this.index = new VectorIndex();
16
+ }
17
+ async init() {
18
+ try {
19
+ const stored = getAllEmbeddings();
20
+ if (stored.length > 0) {
21
+ const entries = stored
22
+ .filter(e => e.embedding && e.dims > 0)
23
+ .map(e => ({
24
+ url: e.normalizedUrl,
25
+ embedding: e.embedding,
26
+ dims: e.dims,
27
+ }));
28
+ const loaded = this.index.loadFromBuffers(entries);
29
+ log.info('loaded embeddings into index', { count: loaded });
30
+ }
31
+ this.available = true;
32
+ this.initialized = true;
33
+ }
34
+ catch (err) {
35
+ log.error('EmbeddingService init failed', { error: String(err) });
36
+ this.available = false;
37
+ }
38
+ }
39
+ isAvailable() {
40
+ return this.available;
41
+ }
42
+ setAvailable(value) {
43
+ this.available = value;
44
+ }
45
+ getIndex() {
46
+ return this.index;
47
+ }
48
+ async embedAndStore(url, markdown) {
49
+ if (!this.available) {
50
+ log.debug('embedding skipped: service not available', { url });
51
+ return;
52
+ }
53
+ try {
54
+ const requestId = randomUUID();
55
+ const response = await this.subprocess.embed(requestId, markdown);
56
+ if (!response.vector || response.error) {
57
+ log.warn('embedding failed for URL', { url, error: response.error });
58
+ return;
59
+ }
60
+ const vector = new Float32Array(response.vector);
61
+ const buffer = Buffer.from(vector.buffer);
62
+ const model = this.subprocess.getModel() ?? getConfig().embeddingModel;
63
+ const dims = this.subprocess.getDims() ?? response.vector.length;
64
+ let normalizedUrl;
65
+ try {
66
+ normalizedUrl = normalizeUrl(url);
67
+ }
68
+ catch {
69
+ normalizedUrl = url;
70
+ }
71
+ updateCacheEmbedding(normalizedUrl, buffer, model, dims);
72
+ this.index.add(normalizedUrl, vector);
73
+ log.debug('embedded and stored', { url: normalizedUrl, dims });
74
+ }
75
+ catch (err) {
76
+ log.warn('embedAndStore failed', { url, error: String(err) });
77
+ }
78
+ }
79
+ embedAsync(url, markdown) {
80
+ if (!this.available)
81
+ return;
82
+ this.embedAndStore(url, markdown).catch(err => {
83
+ log.warn('async embedding failed', { url, error: String(err) });
84
+ });
85
+ }
86
+ async findSimilar(queryText, topK, excludeUrls) {
87
+ if (!this.available || this.index.size() === 0) {
88
+ return [];
89
+ }
90
+ try {
91
+ const requestId = randomUUID();
92
+ const response = await this.subprocess.embed(requestId, queryText);
93
+ if (!response.vector || response.error) {
94
+ log.warn('query embedding failed', { error: response.error });
95
+ return [];
96
+ }
97
+ const queryVector = new Float32Array(response.vector);
98
+ return this.index.findSimilar(queryVector, topK, excludeUrls);
99
+ }
100
+ catch (err) {
101
+ log.warn('findSimilar failed', { error: String(err) });
102
+ return [];
103
+ }
104
+ }
105
+ shutdown() {
106
+ try {
107
+ this.subprocess.shutdown();
108
+ this.index.clear();
109
+ this.available = false;
110
+ this.initialized = false;
111
+ log.info('EmbeddingService shut down');
112
+ }
113
+ catch (err) {
114
+ log.error('EmbeddingService shutdown error', { error: String(err) });
115
+ }
116
+ }
117
+ }
118
+ let globalInstance = null;
119
+ export function getEmbeddingService() {
120
+ if (!globalInstance) {
121
+ globalInstance = new EmbeddingService();
122
+ }
123
+ return globalInstance;
124
+ }
125
+ export function resetEmbeddingService() {
126
+ if (globalInstance) {
127
+ globalInstance.shutdown();
128
+ globalInstance = null;
129
+ }
130
+ }
131
+ //# sourceMappingURL=embed.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"embed.js","sourceRoot":"","sources":["../../src/embedding/embed.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAC;AACtD,OAAO,EAAE,WAAW,EAAsB,MAAM,mBAAmB,CAAC;AACpE,OAAO,EAAE,oBAAoB,EAAE,gBAAgB,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AACzF,OAAO,EAAE,SAAS,EAAE,MAAM,cAAc,CAAC;AACzC,OAAO,EAAE,YAAY,EAAE,MAAM,cAAc,CAAC;AAE5C,MAAM,GAAG,GAAG,YAAY,CAAC,WAAW,CAAC,CAAC;AAEtC,MAAM,OAAO,gBAAgB;IACnB,UAAU,CAAsB;IAChC,KAAK,CAAc;IACnB,SAAS,GAAG,KAAK,CAAC;IAClB,WAAW,GAAG,KAAK,CAAC;IAE5B;QACE,IAAI,CAAC,UAAU,GAAG,IAAI,mBAAmB,EAAE,CAAC;QAC5C,IAAI,CAAC,KAAK,GAAG,IAAI,WAAW,EAAE,CAAC;IACjC,CAAC;IAED,KAAK,CAAC,IAAI;QACR,IAAI,CAAC;YACH,MAAM,MAAM,GAAG,gBAAgB,EAAE,CAAC;YAClC,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBACtB,MAAM,OAAO,GAAG,MAAM;qBACnB,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,SAAS,IAAI,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC;qBACtC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;oBACT,GAAG,EAAE,CAAC,CAAC,aAAa;oBACpB,SAAS,EAAE,CAAC,CAAC,SAAS;oBACtB,IAAI,EAAE,CAAC,CAAC,IAAI;iBACb,CAAC,CAAC,CAAC;gBACN,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,eAAe,CAAC,OAAO,CAAC,CAAC;gBACnD,GAAG,CAAC,IAAI,CAAC,8BAA8B,EAAE,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC;YAC9D,CAAC;YAED,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC;YACtB,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC;QAC1B,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,GAAG,CAAC,KAAK,CAAC,8BAA8B,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YAClE,IAAI,CAAC,SAAS,GAAG,KAAK,CAAC;QACzB,CAAC;IACH,CAAC;IAED,WAAW;QACT,OAAO,IAAI,CAAC,SAAS,CAAC;IACxB,CAAC;IAED,YAAY,CAAC,KAAc;QACzB,IAAI,CAAC,SAAS,GAAG,KAAK,CAAC;IACzB,CAAC;IAED,QAAQ;QACN,OAAO,IAAI,CAAC,KAAK,CAAC;IACpB,CAAC;IAED,KAAK,CAAC,aAAa,CAAC,GAAW,EAAE,QAAgB;QAC/C,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,CAAC;YACpB,GAAG,CAAC,KAAK,CAAC,0CAA0C,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC;YAC/D,OAAO;QACT,CAAC;QAED,IAAI,CAAC;YACH,MAAM,SAAS,GAAG,UAAU,EAAE,CAAC;YAC/B,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,SAAS,EAAE,QAAQ,CAAC,CAAC;YAElE,IAAI,CAAC,QAAQ,CAAC,MAAM,IAAI,QAAQ,CAAC,KAAK,EAAE,CAAC;gBACvC,GAAG,CAAC,IAAI,CAAC,0BAA0B,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,QAAQ,CAAC,KAAK,EAAE,CAAC,CAAC;gBACrE,OAAO;YACT,CAAC;YAED,MAAM,MAAM,GAAG,IAAI,YAAY,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YACjD,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;YAC1C,MAAM,KAAK,GAAG,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,IAAI,SAAS,EAAE,CAAC,cAAc,CAAC;YACvE,MAAM,IAAI,GAAG,IAAI,CAAC,UAAU,CAAC,OAAO,EAAE,IAAI,QAAQ,CAAC,MAAM,CAAC,MAAM,CAAC;YAEjE,IAAI,aAAqB,CAAC;YAC1B,IAAI,CAAC;gBACH,aAAa,GAAG,YAAY,CAAC,GAAG,CAAC,CAAC;YACpC,CAAC;YAAC,MAAM,CAAC;gBACP,aAAa,GAAG,GAAG,CAAC;YACtB,CAAC;YAED,oBAAoB,CAAC,aAAa,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,CAAC,CAAC;YACzD,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,aAAa,EAAE,MAAM,CAAC,CAAC;YAEtC,GAAG,CAAC,KAAK,CAAC,qBAAqB,EAAE,EAAE,GAAG,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,CAAC;QACjE,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,GAAG,CAAC,IAAI,CAAC,sBAAsB,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAChE,CAAC;IACH,CAAC;IAED,UAAU,CAAC,GAAW,EAAE,QAAgB;QACtC,IAAI,CAAC,IAAI,CAAC,SAAS;YAAE,OAAO;QAE5B,IAAI,CAAC,aAAa,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE;YAC5C,GAAG,CAAC,IAAI,CAAC,wBAAwB,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAClE,CAAC,CAAC,CAAC;IACL,CAAC;IAED,KAAK,CAAC,WAAW,CACf,SAAiB,EACjB,IAAY,EACZ,WAAyB;QAEzB,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,EAAE,CAAC;YAC/C,OAAO,EAAE,CAAC;QACZ,CAAC;QAED,IAAI,CAAC;YACH,MAAM,SAAS,GAAG,UAAU,EAAE,CAAC;YAC/B,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,SAAS,EAAE,SAAS,CAAC,CAAC;YAEnE,IAAI,CAAC,QAAQ,CAAC,MAAM,IAAI,QAAQ,CAAC,KAAK,EAAE,CAAC;gBACvC,GAAG,CAAC,IAAI,CAAC,wBAAwB,EAAE,EAAE,KAAK,EAAE,QAAQ,CAAC,KAAK,EAAE,CAAC,CAAC;gBAC9D,OAAO,EAAE,CAAC;YACZ,CAAC;YAED,MAAM,WAAW,GAAG,IAAI,YAAY,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YACtD,OAAO,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,WAAW,EAAE,IAAI,EAAE,WAAW,CAAC,CAAC;QAChE,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,GAAG,CAAC,IAAI,CAAC,oBAAoB,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YACvD,OAAO,EAAE,CAAC;QACZ,CAAC;IACH,CAAC;IAED,QAAQ;QACN,IAAI,CAAC;YACH,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC;YAC3B,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;YACnB,IAAI,CAAC,SAAS,GAAG,KAAK,CAAC;YACvB,IAAI,CAAC,WAAW,GAAG,KAAK,CAAC;YACzB,GAAG,CAAC,IAAI,CAAC,4BAA4B,CAAC,CAAC;QACzC,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,GAAG,CAAC,KAAK,CAAC,iCAAiC,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QACvE,CAAC;IACH,CAAC;CACF;AAED,IAAI,cAAc,GAA4B,IAAI,CAAC;AAEnD,MAAM,UAAU,mBAAmB;IACjC,IAAI,CAAC,cAAc,EAAE,CAAC;QACpB,cAAc,GAAG,IAAI,gBAAgB,EAAE,CAAC;IAC1C,CAAC;IACD,OAAO,cAAc,CAAC;AACxB,CAAC;AAED,MAAM,UAAU,qBAAqB;IACnC,IAAI,cAAc,EAAE,CAAC;QACnB,cAAc,CAAC,QAAQ,EAAE,CAAC;QAC1B,cAAc,GAAG,IAAI,CAAC;IACxB,CAAC;AACH,CAAC"}
@@ -0,0 +1,12 @@
1
+ /**
2
+ * Extract key terms from page content and title for FTS5 query building.
3
+ * Prioritizes: title words > headings > bold text > first paragraph.
4
+ * Returns up to 20 deduplicated, lowercased, stopword-free terms.
5
+ */
6
+ export declare function extractKeyTerms(content: string, title: string): string[];
7
+ export declare function removeStopwords(words: string[]): string[];
8
+ export declare function extractHeadings(content: string): string[];
9
+ export declare function extractBoldText(content: string): string[];
10
+ export declare function extractFirstParagraph(content: string): string;
11
+ export declare function buildFTS5Query(terms: string[]): string;
12
+ //# sourceMappingURL=key-terms.d.ts.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"key-terms.d.ts","sourceRoot":"","sources":["../../src/embedding/key-terms.ts"],"names":[],"mappings":"AAwBA;;;;GAIG;AACH,wBAAgB,eAAe,CAAC,OAAO,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,GAAG,MAAM,EAAE,CAkDxE;AAiBD,wBAAgB,eAAe,CAAC,KAAK,EAAE,MAAM,EAAE,GAAG,MAAM,EAAE,CAEzD;AAED,wBAAgB,eAAe,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,EAAE,CAazD;AAED,wBAAgB,eAAe,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,EAAE,CAazD;AAED,wBAAgB,qBAAqB,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,CAqB7D;AAED,wBAAgB,cAAc,CAAC,KAAK,EAAE,MAAM,EAAE,GAAG,MAAM,CAKtD"}
@@ -0,0 +1,138 @@
1
+ import { createLogger } from '../logger.js';
2
+ const log = createLogger('embedding');
3
+ const STOPWORDS = new Set([
4
+ 'a', 'an', 'the', 'and', 'or', 'but', 'in', 'on', 'at', 'to', 'for',
5
+ 'of', 'with', 'by', 'from', 'is', 'it', 'as', 'be', 'was', 'were',
6
+ 'been', 'are', 'am', 'has', 'have', 'had', 'do', 'does', 'did', 'will',
7
+ 'would', 'could', 'should', 'may', 'might', 'can', 'shall', 'not', 'no',
8
+ 'nor', 'so', 'yet', 'both', 'each', 'few', 'more', 'most', 'other',
9
+ 'some', 'such', 'than', 'too', 'very', 'just', 'about', 'above', 'after',
10
+ 'again', 'all', 'also', 'any', 'because', 'before', 'below', 'between',
11
+ 'during', 'further', 'here', 'how', 'into', 'its', 'itself', 'me', 'my',
12
+ 'myself', 'once', 'only', 'our', 'ours', 'ourselves', 'out', 'over',
13
+ 'own', 'same', 'she', 'he', 'her', 'him', 'his', 'hers', 'that', 'their',
14
+ 'theirs', 'them', 'themselves', 'then', 'there', 'these', 'they', 'this',
15
+ 'those', 'through', 'under', 'until', 'up', 'we', 'what', 'when', 'where',
16
+ 'which', 'while', 'who', 'whom', 'why', 'you', 'your', 'yours', 'yourself',
17
+ 'i', 'if',
18
+ ]);
19
+ const MAX_TERMS = 20;
20
+ const FIRST_PARAGRAPH_MAX_CHARS = 200;
21
+ /**
22
+ * Extract key terms from page content and title for FTS5 query building.
23
+ * Prioritizes: title words > headings > bold text > first paragraph.
24
+ * Returns up to 20 deduplicated, lowercased, stopword-free terms.
25
+ */
26
+ export function extractKeyTerms(content, title) {
27
+ try {
28
+ if (!content.trim() && !title.trim()) {
29
+ return [];
30
+ }
31
+ const allTerms = [];
32
+ const titleWords = tokenize(title);
33
+ allTerms.push(...titleWords);
34
+ const headings = extractHeadings(content);
35
+ for (const heading of headings) {
36
+ allTerms.push(...tokenize(heading));
37
+ }
38
+ const boldPhrases = extractBoldText(content);
39
+ for (const phrase of boldPhrases) {
40
+ allTerms.push(...tokenize(phrase));
41
+ }
42
+ const firstPara = extractFirstParagraph(content);
43
+ if (firstPara) {
44
+ allTerms.push(...tokenize(firstPara));
45
+ }
46
+ const cleaned = removeStopwords(allTerms);
47
+ const seen = new Set();
48
+ const unique = [];
49
+ for (const term of cleaned) {
50
+ if (!seen.has(term)) {
51
+ seen.add(term);
52
+ unique.push(term);
53
+ }
54
+ if (unique.length >= MAX_TERMS)
55
+ break;
56
+ }
57
+ log.debug('extracted key terms', {
58
+ titleTerms: titleWords.length,
59
+ headingTerms: headings.length,
60
+ boldTerms: boldPhrases.length,
61
+ uniqueTerms: unique.length,
62
+ });
63
+ return unique;
64
+ }
65
+ catch (err) {
66
+ log.error('key term extraction failed', { error: String(err) });
67
+ return [];
68
+ }
69
+ }
70
+ function tokenize(text) {
71
+ if (!text)
72
+ return [];
73
+ let cleaned = text.replace(/https?:\/\/[^\s)]+/g, '');
74
+ cleaned = cleaned.replace(/```[\s\S]*?```/g, '');
75
+ cleaned = cleaned.replace(/`[^`]+`/g, '');
76
+ cleaned = cleaned.replace(/^#{1,6}\s+/gm, '');
77
+ cleaned = cleaned.replace(/\*{1,3}|_{1,3}/g, '');
78
+ return cleaned
79
+ .toLowerCase()
80
+ .split(/[^a-z0-9-]+/)
81
+ .filter(w => w.length > 1 && !/^\d+$/.test(w));
82
+ }
83
+ export function removeStopwords(words) {
84
+ return words.filter(w => w.length > 1 && !STOPWORDS.has(w.toLowerCase()));
85
+ }
86
+ export function extractHeadings(content) {
87
+ if (!content)
88
+ return [];
89
+ const headings = [];
90
+ const regex = /^#{1,3}\s+(.+)$/gm;
91
+ let match;
92
+ while ((match = regex.exec(content)) !== null) {
93
+ const text = match[1].trim();
94
+ if (text)
95
+ headings.push(text);
96
+ }
97
+ return headings;
98
+ }
99
+ export function extractBoldText(content) {
100
+ if (!content)
101
+ return [];
102
+ const bold = [];
103
+ const regex = /\*\*(.+?)\*\*|__(.+?)__/g;
104
+ let match;
105
+ while ((match = regex.exec(content)) !== null) {
106
+ const text = (match[1] || match[2]).trim();
107
+ if (text)
108
+ bold.push(text);
109
+ }
110
+ return bold;
111
+ }
112
+ export function extractFirstParagraph(content) {
113
+ if (!content)
114
+ return '';
115
+ const lines = content.split('\n');
116
+ const paragraphLines = [];
117
+ for (const line of lines) {
118
+ const trimmed = line.trim();
119
+ if (!trimmed || trimmed.startsWith('#'))
120
+ continue;
121
+ if (trimmed.startsWith('```'))
122
+ continue;
123
+ paragraphLines.push(trimmed);
124
+ const joined = paragraphLines.join(' ');
125
+ if (joined.length >= FIRST_PARAGRAPH_MAX_CHARS) {
126
+ return joined.slice(0, FIRST_PARAGRAPH_MAX_CHARS);
127
+ }
128
+ break;
129
+ }
130
+ return paragraphLines.join(' ').slice(0, FIRST_PARAGRAPH_MAX_CHARS);
131
+ }
132
+ export function buildFTS5Query(terms) {
133
+ if (terms.length === 0)
134
+ return '';
135
+ const escaped = terms.map(t => t.replace(/['"()]/g, ''));
136
+ return escaped.filter(Boolean).join(' OR ');
137
+ }
138
+ //# sourceMappingURL=key-terms.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"key-terms.js","sourceRoot":"","sources":["../../src/embedding/key-terms.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,cAAc,CAAC;AAE5C,MAAM,GAAG,GAAG,YAAY,CAAC,WAAW,CAAC,CAAC;AAEtC,MAAM,SAAS,GAAG,IAAI,GAAG,CAAC;IACxB,GAAG,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK;IACnE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM;IACjE,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM;IACtE,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,IAAI;IACvE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO;IAClE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO;IACxE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,SAAS,EAAE,QAAQ,EAAE,OAAO,EAAE,SAAS;IACtE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,IAAI;IACvE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,WAAW,EAAE,KAAK,EAAE,MAAM;IACnE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO;IACxE,QAAQ,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM;IACxE,OAAO,EAAE,SAAS,EAAE,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO;IACzE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,OAAO,EAAE,UAAU;IAC1E,GAAG,EAAE,IAAI;CACV,CAAC,CAAC;AAEH,MAAM,SAAS,GAAG,EAAE,CAAC;AACrB,MAAM,yBAAyB,GAAG,GAAG,CAAC;AAEtC;;;;GAIG;AACH,MAAM,UAAU,eAAe,CAAC,OAAe,EAAE,KAAa;IAC5D,IAAI,CAAC;QACH,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC;YACrC,OAAO,EAAE,CAAC;QACZ,CAAC;QAED,MAAM,QAAQ,GAAa,EAAE,CAAC;QAE9B,MAAM,UAAU,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC;QACnC,QAAQ,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,CAAC;QAE7B,MAAM,QAAQ,GAAG,eAAe,CAAC,OAAO,CAAC,CAAC;QAC1C,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;YAC/B,QAAQ,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAC;QACtC,CAAC;QAED,MAAM,WAAW,GAAG,eAAe,CAAC,OAAO,CAAC,CAAC;QAC7C,KAAK,MAAM,MAAM,IAAI,WAAW,EAAE,CAAC;YACjC,QAAQ,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC;QACrC,CAAC;QAED,MAAM,SAAS,GAAG,qBAAqB,CAAC,OAAO,CAAC,CAAC;QACjD,IAAI,SAAS,EAAE,CAAC;YACd,QAAQ,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAC;QACxC,CAAC;QAED,MAAM,OAAO,GAAG,eAAe,CAAC,QAAQ,CAAC,CAAC;QAC1C,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAC;QAC/B,MAAM,MAAM,GAAa,EAAE,CAAC;QAE5B,KAAK,MAAM,IAAI,IAAI,OAAO,EAAE,CAAC;YAC3B,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;gBACpB,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;gBACf,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACpB,CAAC;YACD,IAAI,MAAM,CAAC,MAAM,IAAI,SAAS;gBAAE,MAAM;QACxC,CAAC;QAED,GAAG,CAAC,KAAK,CAAC,qBAAqB,EAAE;YAC/B,UAAU,EAAE,UAAU,CAAC,MAAM;YAC7B,YAAY,EAAE,QAAQ,CAAC,MAAM;YAC7B,SAAS,EAAE,WAAW,CAAC,MAAM;YAC7B,WAAW,EAAE,MAAM,CAAC,MAAM;SAC3B,CAAC,CAAC;QAEH,OAAO,MAAM,CAAC;IAChB,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,GAAG,CAAC,KAAK,CAAC,4BAA4B,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAChE,OAAO,EAAE,CAAC;IACZ,CAAC;AACH,CAAC;AAED,SAAS,QAAQ,CAAC,IAAY;IAC5B,IAAI,CAAC,IAAI;QAAE,OAAO,EAAE,CAAC;IAErB,IAAI,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,qBAAqB,EAAE,EAAE,CAAC,CAAC;IACtD,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC,iBAAiB,EAAE,EAAE,CAAC,CAAC;IACjD,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;IAC1C,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC,cAAc,EAAE,EAAE,CAAC,CAAC;IAC9C,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC,iBAAiB,EAAE,EAAE,CAAC,CAAC;IAEjD,OAAO,OAAO;SACX,WAAW,EAAE;SACb,KAAK,CAAC,aAAa,CAAC;SACpB,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;AACnD,CAAC;AAED,MAAM,UAAU,eAAe,CAAC,KAAe;IAC7C,OAAO,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC;AAC5E,CAAC;AAED,MAAM,UAAU,eAAe,CAAC,OAAe;IAC7C,IAAI,CAAC,OAAO;QAAE,OAAO,EAAE,CAAC;IAExB,MAAM,QAAQ,GAAa,EAAE,CAAC;IAC9B,MAAM,KAAK,GAAG,mBAAmB,CAAC;IAClC,IAAI,KAA6B,CAAC;IAElC,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;QAC9C,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QAC7B,IAAI,IAAI;YAAE,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAChC,CAAC;IAED,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED,MAAM,UAAU,eAAe,CAAC,OAAe;IAC7C,IAAI,CAAC,OAAO;QAAE,OAAO,EAAE,CAAC;IAExB,MAAM,IAAI,GAAa,EAAE,CAAC;IAC1B,MAAM,KAAK,GAAG,0BAA0B,CAAC;IACzC,IAAI,KAA6B,CAAC;IAElC,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;QAC9C,MAAM,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QAC3C,IAAI,IAAI;YAAE,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC5B,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC;AAED,MAAM,UAAU,qBAAqB,CAAC,OAAe;IACnD,IAAI,CAAC,OAAO;QAAE,OAAO,EAAE,CAAC;IAExB,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAClC,MAAM,cAAc,GAAa,EAAE,CAAC;IAEpC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;QAC5B,IAAI,CAAC,OAAO,IAAI,OAAO,CAAC,UAAU,CAAC,GAAG,CAAC;YAAE,SAAS;QAClD,IAAI,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC;YAAE,SAAS;QAExC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QAC7B,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACxC,IAAI,MAAM,CAAC,MAAM,IAAI,yBAAyB,EAAE,CAAC;YAC/C,OAAO,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,yBAAyB,CAAC,CAAC;QACpD,CAAC;QAED,MAAM;IACR,CAAC;IAED,OAAO,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,yBAAyB,CAAC,CAAC;AACtE,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,KAAe;IAC5C,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IAElC,MAAM,OAAO,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC,CAAC;IACzD,OAAO,OAAO,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AAC9C,CAAC"}
@@ -0,0 +1,31 @@
1
+ export interface EmbeddingResponse {
2
+ id: string;
3
+ vector?: number[];
4
+ error?: string;
5
+ }
6
+ export interface SubprocessOptions {
7
+ readyTimeoutMs?: number;
8
+ requestTimeoutMs?: number;
9
+ }
10
+ export declare class EmbeddingSubprocess {
11
+ private proc;
12
+ private pending;
13
+ private available;
14
+ private dims;
15
+ private modelName;
16
+ private spawnPromise;
17
+ private stdoutBuffer;
18
+ private idleTimer;
19
+ private readyTimeoutMs;
20
+ private requestTimeoutMs;
21
+ constructor(options?: SubprocessOptions);
22
+ isAvailable(): boolean;
23
+ getDims(): number | null;
24
+ getModel(): string | null;
25
+ embed(id: string, text: string): Promise<EmbeddingResponse>;
26
+ shutdown(): void;
27
+ private spawnProcess;
28
+ private handleStdoutData;
29
+ private resetIdleTimer;
30
+ }
31
+ //# sourceMappingURL=subprocess.d.ts.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"subprocess.d.ts","sourceRoot":"","sources":["../../src/embedding/subprocess.ts"],"names":[],"mappings":"AAWA,MAAM,WAAW,iBAAiB;IAChC,EAAE,EAAE,MAAM,CAAC;IACX,MAAM,CAAC,EAAE,MAAM,EAAE,CAAC;IAClB,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAQD,MAAM,WAAW,iBAAiB;IAChC,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,gBAAgB,CAAC,EAAE,MAAM,CAAC;CAC3B;AAED,qBAAa,mBAAmB;IAC9B,OAAO,CAAC,IAAI,CAA6B;IACzC,OAAO,CAAC,OAAO,CAAqC;IACpD,OAAO,CAAC,SAAS,CAAwB;IACzC,OAAO,CAAC,IAAI,CAAuB;IACnC,OAAO,CAAC,SAAS,CAAuB;IACxC,OAAO,CAAC,YAAY,CAA8B;IAClD,OAAO,CAAC,YAAY,CAAM;IAC1B,OAAO,CAAC,SAAS,CAA8C;IAC/D,OAAO,CAAC,cAAc,CAAS;IAC/B,OAAO,CAAC,gBAAgB,CAAS;gBAErB,OAAO,CAAC,EAAE,iBAAiB;IAKvC,WAAW,IAAI,OAAO;IAItB,OAAO,IAAI,MAAM,GAAG,IAAI;IAIxB,QAAQ,IAAI,MAAM,GAAG,IAAI;IAInB,KAAK,CAAC,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,iBAAiB,CAAC;IAmCjE,QAAQ,IAAI,IAAI;YA0BF,YAAY;IAgG1B,OAAO,CAAC,gBAAgB;IAoCxB,OAAO,CAAC,cAAc;CAWvB"}
@@ -0,0 +1,213 @@
1
+ import { spawn } from 'node:child_process';
2
+ import { dirname, join } from 'node:path';
3
+ import { fileURLToPath } from 'node:url';
4
+ import { getConfig } from '../config.js';
5
+ import { createLogger } from '../logger.js';
6
+ const log = createLogger('embedding');
7
+ const __dirname = dirname(fileURLToPath(import.meta.url));
8
+ const SCRIPT_PATH = join(__dirname, '..', 'scripts', 'embedding_server.py');
9
+ export class EmbeddingSubprocess {
10
+ proc = null;
11
+ pending = new Map();
12
+ available = null;
13
+ dims = null;
14
+ modelName = null;
15
+ spawnPromise = null;
16
+ stdoutBuffer = '';
17
+ idleTimer = null;
18
+ readyTimeoutMs;
19
+ requestTimeoutMs;
20
+ constructor(options) {
21
+ this.readyTimeoutMs = options?.readyTimeoutMs ?? 60000;
22
+ this.requestTimeoutMs = options?.requestTimeoutMs ?? 30000;
23
+ }
24
+ isAvailable() {
25
+ return this.available === true;
26
+ }
27
+ getDims() {
28
+ return this.dims;
29
+ }
30
+ getModel() {
31
+ return this.modelName;
32
+ }
33
+ async embed(id, text) {
34
+ try {
35
+ if (!this.proc && !this.spawnPromise) {
36
+ this.spawnPromise = this.spawnProcess();
37
+ }
38
+ if (this.spawnPromise) {
39
+ await this.spawnPromise;
40
+ }
41
+ if (!this.proc || this.available === false) {
42
+ throw new Error('Embedding subprocess not available');
43
+ }
44
+ this.resetIdleTimer();
45
+ const config = getConfig();
46
+ const truncatedText = text.slice(0, config.embeddingMaxTextLength);
47
+ return await new Promise((resolve, reject) => {
48
+ const timeoutHandle = setTimeout(() => {
49
+ this.pending.delete(id);
50
+ reject(new Error(`Embedding request ${id} timed out after ${this.requestTimeoutMs}ms`));
51
+ }, this.requestTimeoutMs);
52
+ this.pending.set(id, { resolve, reject, timeoutHandle });
53
+ const request = JSON.stringify({ id, text: truncatedText }) + '\n';
54
+ this.proc.stdin.write(request);
55
+ });
56
+ }
57
+ catch (err) {
58
+ log.error('embed failed', { id, error: String(err) });
59
+ throw err;
60
+ }
61
+ }
62
+ shutdown() {
63
+ try {
64
+ if (this.idleTimer) {
65
+ clearTimeout(this.idleTimer);
66
+ this.idleTimer = null;
67
+ }
68
+ for (const [id, pending] of this.pending.entries()) {
69
+ clearTimeout(pending.timeoutHandle);
70
+ pending.reject(new Error('Subprocess shutting down'));
71
+ this.pending.delete(id);
72
+ }
73
+ if (this.proc) {
74
+ this.proc.stdin?.end();
75
+ this.proc.kill();
76
+ this.proc = null;
77
+ }
78
+ this.spawnPromise = null;
79
+ log.info('embedding subprocess shut down');
80
+ }
81
+ catch (err) {
82
+ log.error('shutdown error', { error: String(err) });
83
+ }
84
+ }
85
+ async spawnProcess() {
86
+ const config = getConfig();
87
+ try {
88
+ log.info('spawning embedding subprocess', { model: config.embeddingModel });
89
+ const proc = spawn('python3', [
90
+ SCRIPT_PATH,
91
+ config.embeddingModel,
92
+ String(config.embeddingMaxTextLength),
93
+ ], {
94
+ stdio: ['pipe', 'pipe', 'pipe'],
95
+ });
96
+ this.proc = proc;
97
+ await new Promise((resolve, reject) => {
98
+ const readyTimeout = setTimeout(() => {
99
+ reject(new Error(`Embedding subprocess READY timeout after ${this.readyTimeoutMs}ms`));
100
+ proc.kill();
101
+ }, this.readyTimeoutMs);
102
+ let stderrBuf = '';
103
+ proc.stderr.on('data', (data) => {
104
+ stderrBuf += data.toString();
105
+ const lines = stderrBuf.split('\n');
106
+ for (const line of lines) {
107
+ if (line.startsWith('READY')) {
108
+ clearTimeout(readyTimeout);
109
+ const modelMatch = line.match(/model=(\S+)/);
110
+ const dimsMatch = line.match(/dims=(\d+)/);
111
+ if (modelMatch)
112
+ this.modelName = modelMatch[1];
113
+ if (dimsMatch)
114
+ this.dims = parseInt(dimsMatch[1], 10);
115
+ this.available = true;
116
+ log.info('embedding subprocess ready', {
117
+ model: this.modelName,
118
+ dims: this.dims,
119
+ });
120
+ resolve();
121
+ return;
122
+ }
123
+ if (line.startsWith('ERROR')) {
124
+ clearTimeout(readyTimeout);
125
+ this.available = false;
126
+ reject(new Error(`Embedding subprocess: ${line}`));
127
+ return;
128
+ }
129
+ }
130
+ });
131
+ proc.on('error', (err) => {
132
+ clearTimeout(readyTimeout);
133
+ this.available = false;
134
+ this.proc = null;
135
+ this.spawnPromise = null;
136
+ log.error('embedding subprocess error', { error: String(err) });
137
+ reject(err);
138
+ });
139
+ proc.on('close', (code) => {
140
+ clearTimeout(readyTimeout);
141
+ this.available = false;
142
+ this.proc = null;
143
+ this.spawnPromise = null;
144
+ for (const [id, pending] of this.pending.entries()) {
145
+ clearTimeout(pending.timeoutHandle);
146
+ pending.reject(new Error(`Embedding subprocess exited with code ${code}`));
147
+ this.pending.delete(id);
148
+ }
149
+ if (code !== 0 && code !== null) {
150
+ log.warn('embedding subprocess exited', { code });
151
+ reject(new Error(`Embedding subprocess exited with code ${code}`));
152
+ }
153
+ });
154
+ });
155
+ proc.stdout.on('data', (data) => {
156
+ this.handleStdoutData(data.toString());
157
+ });
158
+ this.resetIdleTimer();
159
+ }
160
+ catch (err) {
161
+ this.available = false;
162
+ this.proc = null;
163
+ this.spawnPromise = null;
164
+ log.error('failed to spawn embedding subprocess', { error: String(err) });
165
+ throw err;
166
+ }
167
+ }
168
+ handleStdoutData(data) {
169
+ this.stdoutBuffer += data;
170
+ const lines = this.stdoutBuffer.split('\n');
171
+ this.stdoutBuffer = lines.pop() ?? '';
172
+ for (const line of lines) {
173
+ const trimmed = line.trim();
174
+ if (!trimmed)
175
+ continue;
176
+ try {
177
+ const response = JSON.parse(trimmed);
178
+ const id = response.id;
179
+ if (id && this.pending.has(id)) {
180
+ const pending = this.pending.get(id);
181
+ this.pending.delete(id);
182
+ clearTimeout(pending.timeoutHandle);
183
+ if (response.error) {
184
+ pending.reject(new Error(response.error));
185
+ }
186
+ else {
187
+ pending.resolve(response);
188
+ }
189
+ }
190
+ else {
191
+ log.warn('received response for unknown request', { id });
192
+ }
193
+ }
194
+ catch (err) {
195
+ log.warn('failed to parse subprocess stdout line', {
196
+ line: trimmed.slice(0, 200),
197
+ error: String(err),
198
+ });
199
+ }
200
+ }
201
+ }
202
+ resetIdleTimer() {
203
+ if (this.idleTimer) {
204
+ clearTimeout(this.idleTimer);
205
+ }
206
+ const config = getConfig();
207
+ this.idleTimer = setTimeout(() => {
208
+ log.info('embedding subprocess idle timeout, shutting down');
209
+ this.shutdown();
210
+ }, config.embeddingIdleTimeoutMs);
211
+ }
212
+ }
213
+ //# sourceMappingURL=subprocess.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"subprocess.js","sourceRoot":"","sources":["../../src/embedding/subprocess.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,EAAqB,MAAM,oBAAoB,CAAC;AAC9D,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AAC1C,OAAO,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AACzC,OAAO,EAAE,SAAS,EAAE,MAAM,cAAc,CAAC;AACzC,OAAO,EAAE,YAAY,EAAE,MAAM,cAAc,CAAC;AAE5C,MAAM,GAAG,GAAG,YAAY,CAAC,WAAW,CAAC,CAAC;AAEtC,MAAM,SAAS,GAAG,OAAO,CAAC,aAAa,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;AAC1D,MAAM,WAAW,GAAG,IAAI,CAAC,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE,qBAAqB,CAAC,CAAC;AAmB5E,MAAM,OAAO,mBAAmB;IACtB,IAAI,GAAwB,IAAI,CAAC;IACjC,OAAO,GAAG,IAAI,GAAG,EAA0B,CAAC;IAC5C,SAAS,GAAmB,IAAI,CAAC;IACjC,IAAI,GAAkB,IAAI,CAAC;IAC3B,SAAS,GAAkB,IAAI,CAAC;IAChC,YAAY,GAAyB,IAAI,CAAC;IAC1C,YAAY,GAAG,EAAE,CAAC;IAClB,SAAS,GAAyC,IAAI,CAAC;IACvD,cAAc,CAAS;IACvB,gBAAgB,CAAS;IAEjC,YAAY,OAA2B;QACrC,IAAI,CAAC,cAAc,GAAG,OAAO,EAAE,cAAc,IAAI,KAAK,CAAC;QACvD,IAAI,CAAC,gBAAgB,GAAG,OAAO,EAAE,gBAAgB,IAAI,KAAK,CAAC;IAC7D,CAAC;IAED,WAAW;QACT,OAAO,IAAI,CAAC,SAAS,KAAK,IAAI,CAAC;IACjC,CAAC;IAED,OAAO;QACL,OAAO,IAAI,CAAC,IAAI,CAAC;IACnB,CAAC;IAED,QAAQ;QACN,OAAO,IAAI,CAAC,SAAS,CAAC;IACxB,CAAC;IAED,KAAK,CAAC,KAAK,CAAC,EAAU,EAAE,IAAY;QAClC,IAAI,CAAC;YACH,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,CAAC;gBACrC,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC,YAAY,EAAE,CAAC;YAC1C,CAAC;YACD,IAAI,IAAI,CAAC,YAAY,EAAE,CAAC;gBACtB,MAAM,IAAI,CAAC,YAAY,CAAC;YAC1B,CAAC;YAED,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,SAAS,KAAK,KAAK,EAAE,CAAC;gBAC3C,MAAM,IAAI,KAAK,CAAC,oCAAoC,CAAC,CAAC;YACxD,CAAC;YAED,IAAI,CAAC,cAAc,EAAE,CAAC;YAEtB,MAAM,MAAM,GAAG,SAAS,EAAE,CAAC;YAC3B,MAAM,aAAa,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,MAAM,CAAC,sBAAsB,CAAC,CAAC;YAEnE,OAAO,MAAM,IAAI,OAAO,CAAoB,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;gBAC9D,MAAM,aAAa,GAAG,UAAU,CAAC,GAAG,EAAE;oBACpC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;oBACxB,MAAM,CAAC,IAAI,KAAK,CAAC,qBAAqB,EAAE,oBAAoB,IAAI,CAAC,gBAAgB,IAAI,CAAC,CAAC,CAAC;gBAC1F,CAAC,EAAE,IAAI,CAAC,gBAAgB,CAAC,CAAC;gBAE1B,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,OAAO,EAAE,MAAM,EAAE,aAAa,EAAE,CAAC,CAAC;gBAEzD,MAAM,OAAO,GAAG,IAAI,CAAC,SAAS,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,aAAa,EAAE,CAAC,GAAG,IAAI,CAAC;gBACnE,IAAI,CAAC,IAAK,CAAC,KAAM,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;YACnC,CAAC,CAAC,CAAC;QACL,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,GAAG,CAAC,KAAK,CAAC,cAAc,EAAE,EAAE,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YACtD,MAAM,GAAG,CAAC;QACZ,CAAC;IACH,CAAC;IAED,QAAQ;QACN,IAAI,CAAC;YACH,IAAI,IAAI,CAAC,SAAS,EAAE,CAAC;gBACnB,YAAY,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;gBAC7B,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC;YACxB,CAAC;YAED,KAAK,MAAM,CAAC,EAAE,EAAE,OAAO,CAAC,IAAI,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC;gBACnD,YAAY,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC;gBACpC,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,0BAA0B,CAAC,CAAC,CAAC;gBACtD,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;YAC1B,CAAC;YAED,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC;gBACd,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,GAAG,EAAE,CAAC;gBACvB,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;gBACjB,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC;YACnB,CAAC;YAED,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC;YACzB,GAAG,CAAC,IAAI,CAAC,gCAAgC,CAAC,CAAC;QAC7C,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,GAAG,CAAC,KAAK,CAAC,gBAAgB,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QACtD,CAAC;IACH,CAAC;IAEO,KAAK,CAAC,YAAY;QACxB,MAAM,MAAM,GAAG,SAAS,EAAE,CAAC;QAE3B,IAAI,CAAC;YACH,GAAG,CAAC,IAAI,CAAC,+BAA+B,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,cAAc,EAAE,CAAC,CAAC;YAE5E,MAAM,IAAI,GAAG,KAAK,CAAC,SAAS,EAAE;gBAC5B,WAAW;gBACX,MAAM,CAAC,cAAc;gBACrB,MAAM,CAAC,MAAM,CAAC,sBAAsB,CAAC;aACtC,EAAE;gBACD,KAAK,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;aAChC,CAAC,CAAC;YAEH,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC;YAEjB,MAAM,IAAI,OAAO,CAAO,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;gBAC1C,MAAM,YAAY,GAAG,UAAU,CAAC,GAAG,EAAE;oBACnC,MAAM,CAAC,IAAI,KAAK,CAAC,4CAA4C,IAAI,CAAC,cAAc,IAAI,CAAC,CAAC,CAAC;oBACvF,IAAI,CAAC,IAAI,EAAE,CAAC;gBACd,CAAC,EAAE,IAAI,CAAC,cAAc,CAAC,CAAC;gBAExB,IAAI,SAAS,GAAG,EAAE,CAAC;gBAEnB,IAAI,CAAC,MAAO,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAY,EAAE,EAAE;oBACvC,SAAS,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;oBAC7B,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;oBAEpC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;wBACzB,IAAI,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,EAAE,CAAC;4BAC7B,YAAY,CAAC,YAAY,CAAC,CAAC;4BAC3B,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC;4BAC7C,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC;4BAC3C,IAAI,UAAU;gCAAE,IAAI,CAAC,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC;4BAC/C,IAAI,SAAS;gCAAE,IAAI,CAAC,IAAI,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;4BAEtD,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC;4BACtB,GAAG,CAAC,IAAI,CAAC,4BAA4B,EAAE;gCACrC,KAAK,EAAE,IAAI,CAAC,SAAS;gCACrB,IAAI,EAAE,IAAI,CAAC,IAAI;6BAChB,CAAC,CAAC;4BACH,OAAO,EAAE,CAAC;4BACV,OAAO;wBACT,CAAC;wBACD,IAAI,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,EAAE,CAAC;4BAC7B,YAAY,CAAC,YAAY,CAAC,CAAC;4BAC3B,IAAI,CAAC,SAAS,GAAG,KAAK,CAAC;4BACvB,MAAM,CAAC,IAAI,KAAK,CAAC,yBAAyB,IAAI,EAAE,CAAC,CAAC,CAAC;4BACnD,OAAO;wBACT,CAAC;oBACH,CAAC;gBACH,CAAC,CAAC,CAAC;gBAEH,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE;oBACvB,YAAY,CAAC,YAAY,CAAC,CAAC;oBAC3B,IAAI,CAAC,SAAS,GAAG,KAAK,CAAC;oBACvB,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC;oBACjB,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC;oBACzB,GAAG,CAAC,KAAK,CAAC,4BAA4B,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;oBAChE,MAAM,CAAC,GAAG,CAAC,CAAC;gBACd,CAAC,CAAC,CAAC;gBAEH,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,IAAI,EAAE,EAAE;oBACxB,YAAY,CAAC,YAAY,CAAC,CAAC;oBAC3B,IAAI,CAAC,SAAS,GAAG,KAAK,CAAC;oBACvB,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC;oBACjB,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC;oBAEzB,KAAK,MAAM,CAAC,EAAE,EAAE,OAAO,CAAC,IAAI,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC;wBACnD,YAAY,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC;wBACpC,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,yCAAyC,IAAI,EAAE,CAAC,CAAC,CAAC;wBAC3E,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;oBAC1B,CAAC;oBAED,IAAI,IAAI,KAAK,CAAC,IAAI,IAAI,KAAK,IAAI,EAAE,CAAC;wBAChC,GAAG,CAAC,IAAI,CAAC,6BAA6B,EAAE,EAAE,IAAI,EAAE,CAAC,CAAC;wBAClD,MAAM,CAAC,IAAI,KAAK,CAAC,yCAAyC,IAAI,EAAE,CAAC,CAAC,CAAC;oBACrE,CAAC;gBACH,CAAC,CAAC,CAAC;YACL,CAAC,CAAC,CAAC;YAEH,IAAI,CAAC,MAAO,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAY,EAAE,EAAE;gBACvC,IAAI,CAAC,gBAAgB,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,CAAC;YACzC,CAAC,CAAC,CAAC;YAEH,IAAI,CAAC,cAAc,EAAE,CAAC;QAExB,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,IAAI,CAAC,SAAS,GAAG,KAAK,CAAC;YACvB,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC;YACjB,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC;YACzB,GAAG,CAAC,KAAK,CAAC,sCAAsC,EAAE,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YAC1E,MAAM,GAAG,CAAC;QACZ,CAAC;IACH,CAAC;IAEO,gBAAgB,CAAC,IAAY;QACnC,IAAI,CAAC,YAAY,IAAI,IAAI,CAAC;QAE1B,MAAM,KAAK,GAAG,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAC5C,IAAI,CAAC,YAAY,GAAG,KAAK,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC;QAEtC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YACzB,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;YAC5B,IAAI,CAAC,OAAO;gBAAE,SAAS;YAEvB,IAAI,CAAC;gBACH,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAsB,CAAC;gBAC1D,MAAM,EAAE,GAAG,QAAQ,CAAC,EAAE,CAAC;gBAEvB,IAAI,EAAE,IAAI,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,CAAC;oBAC/B,MAAM,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAE,CAAC;oBACtC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;oBACxB,YAAY,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC;oBAEpC,IAAI,QAAQ,CAAC,KAAK,EAAE,CAAC;wBACnB,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC;oBAC5C,CAAC;yBAAM,CAAC;wBACN,OAAO,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC;oBAC5B,CAAC;gBACH,CAAC;qBAAM,CAAC;oBACN,GAAG,CAAC,IAAI,CAAC,uCAAuC,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC;gBAC5D,CAAC;YACH,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACb,GAAG,CAAC,IAAI,CAAC,wCAAwC,EAAE;oBACjD,IAAI,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC;oBAC3B,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC;iBACnB,CAAC,CAAC;YACL,CAAC;QACH,CAAC;IACH,CAAC;IAEO,cAAc;QACpB,IAAI,IAAI,CAAC,SAAS,EAAE,CAAC;YACnB,YAAY,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;QAC/B,CAAC;QAED,MAAM,MAAM,GAAG,SAAS,EAAE,CAAC;QAC3B,IAAI,CAAC,SAAS,GAAG,UAAU,CAAC,GAAG,EAAE;YAC/B,GAAG,CAAC,IAAI,CAAC,kDAAkD,CAAC,CAAC;YAC7D,IAAI,CAAC,QAAQ,EAAE,CAAC;QAClB,CAAC,EAAE,MAAM,CAAC,sBAAsB,CAAC,CAAC;IACpC,CAAC;CACF"}
@@ -0,0 +1,26 @@
1
+ export interface VectorEntry {
2
+ url: string;
3
+ vector: Float32Array;
4
+ }
5
+ export interface SimilarResult {
6
+ url: string;
7
+ score: number;
8
+ }
9
+ export declare class VectorIndex {
10
+ private entries;
11
+ size(): number;
12
+ add(url: string, vector: Float32Array): void;
13
+ remove(url: string): boolean;
14
+ has(url: string): boolean;
15
+ get(url: string): Float32Array | undefined;
16
+ findSimilar(queryVector: Float32Array, topK: number, excludeUrls?: Set<string>): SimilarResult[];
17
+ clear(): void;
18
+ loadFromBuffers(entries: Array<{
19
+ url: string;
20
+ embedding: Buffer;
21
+ dims: number;
22
+ }>): number;
23
+ getAllUrls(): string[];
24
+ }
25
+ export declare function cosineSimilarity(a: Float32Array, b: Float32Array): number;
26
+ //# sourceMappingURL=vector-index.d.ts.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"vector-index.d.ts","sourceRoot":"","sources":["../../src/embedding/vector-index.ts"],"names":[],"mappings":"AAIA,MAAM,WAAW,WAAW;IAC1B,GAAG,EAAE,MAAM,CAAC;IACZ,MAAM,EAAE,YAAY,CAAC;CACtB;AAED,MAAM,WAAW,aAAa;IAC5B,GAAG,EAAE,MAAM,CAAC;IACZ,KAAK,EAAE,MAAM,CAAC;CACf;AAED,qBAAa,WAAW;IACtB,OAAO,CAAC,OAAO,CAAmC;IAElD,IAAI,IAAI,MAAM;IAId,GAAG,CAAC,GAAG,EAAE,MAAM,EAAE,MAAM,EAAE,YAAY,GAAG,IAAI;IAI5C,MAAM,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO;IAI5B,GAAG,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO;IAIzB,GAAG,CAAC,GAAG,EAAE,MAAM,GAAG,YAAY,GAAG,SAAS;IAI1C,WAAW,CAAC,WAAW,EAAE,YAAY,EAAE,IAAI,EAAE,MAAM,EAAE,WAAW,CAAC,EAAE,GAAG,CAAC,MAAM,CAAC,GAAG,aAAa,EAAE;IAoBhG,KAAK,IAAI,IAAI;IAIb,eAAe,CAAC,OAAO,EAAE,KAAK,CAAC;QAAE,GAAG,EAAE,MAAM,CAAC;QAAC,SAAS,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC,GAAG,MAAM;IAkBzF,UAAU,IAAI,MAAM,EAAE;CAGvB;AAED,wBAAgB,gBAAgB,CAAC,CAAC,EAAE,YAAY,EAAE,CAAC,EAAE,YAAY,GAAG,MAAM,CAkBzE"}