@timmeck/brain 1.8.1 → 1.8.2

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 (162) hide show
  1. package/BRAIN_PLAN.md +3324 -3324
  2. package/LICENSE +21 -21
  3. package/dist/cli/commands/dashboard.js +595 -595
  4. package/dist/dashboard/server.js +25 -25
  5. package/dist/db/migrations/001_core_schema.js +115 -115
  6. package/dist/db/migrations/002_learning_schema.js +33 -33
  7. package/dist/db/migrations/003_code_schema.js +48 -48
  8. package/dist/db/migrations/004_synapses_schema.js +52 -52
  9. package/dist/db/migrations/005_fts_indexes.js +73 -73
  10. package/dist/db/migrations/007_feedback.js +8 -8
  11. package/dist/db/migrations/008_git_integration.js +33 -33
  12. package/dist/db/migrations/009_embeddings.js +3 -3
  13. package/dist/db/repositories/antipattern.repository.js +3 -3
  14. package/dist/db/repositories/code-module.repository.js +32 -32
  15. package/dist/db/repositories/notification.repository.js +3 -3
  16. package/dist/db/repositories/project.repository.js +21 -21
  17. package/dist/db/repositories/rule.repository.js +24 -24
  18. package/dist/db/repositories/solution.repository.js +50 -50
  19. package/dist/db/repositories/synapse.repository.js +18 -18
  20. package/dist/db/repositories/terminal.repository.js +24 -24
  21. package/dist/ipc/server.d.ts +8 -0
  22. package/dist/ipc/server.js +67 -1
  23. package/dist/ipc/server.js.map +1 -1
  24. package/dist/matching/error-matcher.js +5 -5
  25. package/dist/matching/fingerprint.js +6 -1
  26. package/dist/matching/fingerprint.js.map +1 -1
  27. package/dist/services/error.service.js +4 -3
  28. package/dist/services/error.service.js.map +1 -1
  29. package/dist/services/git.service.js +14 -14
  30. package/package.json +49 -49
  31. package/src/api/server.ts +395 -395
  32. package/src/brain.ts +266 -266
  33. package/src/cli/colors.ts +116 -116
  34. package/src/cli/commands/config.ts +169 -169
  35. package/src/cli/commands/dashboard.ts +755 -755
  36. package/src/cli/commands/doctor.ts +118 -118
  37. package/src/cli/commands/explain.ts +83 -83
  38. package/src/cli/commands/export.ts +31 -31
  39. package/src/cli/commands/import.ts +199 -199
  40. package/src/cli/commands/insights.ts +65 -65
  41. package/src/cli/commands/learn.ts +24 -24
  42. package/src/cli/commands/modules.ts +53 -53
  43. package/src/cli/commands/network.ts +67 -67
  44. package/src/cli/commands/projects.ts +42 -42
  45. package/src/cli/commands/query.ts +120 -120
  46. package/src/cli/commands/start.ts +62 -62
  47. package/src/cli/commands/status.ts +75 -75
  48. package/src/cli/commands/stop.ts +34 -34
  49. package/src/cli/ipc-helper.ts +22 -22
  50. package/src/cli/update-check.ts +63 -63
  51. package/src/code/fingerprint.ts +87 -87
  52. package/src/code/parsers/generic.ts +29 -29
  53. package/src/code/parsers/python.ts +54 -54
  54. package/src/code/parsers/typescript.ts +65 -65
  55. package/src/code/registry.ts +60 -60
  56. package/src/dashboard/server.ts +142 -142
  57. package/src/db/connection.ts +22 -22
  58. package/src/db/migrations/001_core_schema.ts +120 -120
  59. package/src/db/migrations/002_learning_schema.ts +38 -38
  60. package/src/db/migrations/003_code_schema.ts +53 -53
  61. package/src/db/migrations/004_synapses_schema.ts +57 -57
  62. package/src/db/migrations/005_fts_indexes.ts +78 -78
  63. package/src/db/migrations/006_synapses_phase3.ts +17 -17
  64. package/src/db/migrations/007_feedback.ts +13 -13
  65. package/src/db/migrations/008_git_integration.ts +38 -38
  66. package/src/db/migrations/009_embeddings.ts +8 -8
  67. package/src/db/repositories/antipattern.repository.ts +66 -66
  68. package/src/db/repositories/code-module.repository.ts +142 -142
  69. package/src/db/repositories/notification.repository.ts +66 -66
  70. package/src/db/repositories/project.repository.ts +93 -93
  71. package/src/db/repositories/rule.repository.ts +108 -108
  72. package/src/db/repositories/solution.repository.ts +154 -154
  73. package/src/db/repositories/synapse.repository.ts +153 -153
  74. package/src/db/repositories/terminal.repository.ts +101 -101
  75. package/src/embeddings/engine.ts +238 -238
  76. package/src/index.ts +63 -63
  77. package/src/ipc/client.ts +118 -118
  78. package/src/ipc/protocol.ts +35 -35
  79. package/src/ipc/router.ts +133 -133
  80. package/src/ipc/server.ts +176 -110
  81. package/src/learning/decay.ts +46 -46
  82. package/src/learning/pattern-extractor.ts +90 -90
  83. package/src/learning/rule-generator.ts +74 -74
  84. package/src/matching/error-matcher.ts +5 -5
  85. package/src/matching/fingerprint.ts +34 -29
  86. package/src/matching/similarity.ts +61 -61
  87. package/src/matching/tfidf.ts +74 -74
  88. package/src/matching/tokenizer.ts +41 -41
  89. package/src/mcp/auto-detect.ts +93 -93
  90. package/src/mcp/http-server.ts +140 -140
  91. package/src/mcp/server.ts +73 -73
  92. package/src/parsing/error-parser.ts +28 -28
  93. package/src/parsing/parsers/compiler.ts +93 -93
  94. package/src/parsing/parsers/generic.ts +28 -28
  95. package/src/parsing/parsers/go.ts +97 -97
  96. package/src/parsing/parsers/node.ts +69 -69
  97. package/src/parsing/parsers/python.ts +62 -62
  98. package/src/parsing/parsers/rust.ts +50 -50
  99. package/src/parsing/parsers/shell.ts +42 -42
  100. package/src/parsing/types.ts +47 -47
  101. package/src/research/gap-analyzer.ts +135 -135
  102. package/src/research/insight-generator.ts +123 -123
  103. package/src/research/research-engine.ts +116 -116
  104. package/src/research/synergy-detector.ts +126 -126
  105. package/src/research/template-extractor.ts +130 -130
  106. package/src/research/trend-analyzer.ts +127 -127
  107. package/src/services/code.service.ts +271 -271
  108. package/src/services/error.service.ts +4 -3
  109. package/src/services/git.service.ts +132 -132
  110. package/src/services/notification.service.ts +41 -41
  111. package/src/services/synapse.service.ts +59 -59
  112. package/src/services/terminal.service.ts +81 -81
  113. package/src/synapses/activation.ts +80 -80
  114. package/src/synapses/decay.ts +38 -38
  115. package/src/synapses/hebbian.ts +69 -69
  116. package/src/synapses/pathfinder.ts +81 -81
  117. package/src/synapses/synapse-manager.ts +109 -109
  118. package/src/types/code.types.ts +52 -52
  119. package/src/types/error.types.ts +67 -67
  120. package/src/types/ipc.types.ts +8 -8
  121. package/src/types/mcp.types.ts +53 -53
  122. package/src/types/research.types.ts +28 -28
  123. package/src/types/solution.types.ts +30 -30
  124. package/src/utils/events.ts +45 -45
  125. package/src/utils/hash.ts +5 -5
  126. package/src/utils/logger.ts +48 -48
  127. package/src/utils/paths.ts +19 -19
  128. package/tests/e2e/test_code_intelligence.py +1015 -0
  129. package/tests/e2e/test_error_memory.py +451 -0
  130. package/tests/e2e/test_full_integration.py +534 -0
  131. package/tests/fixtures/code-modules/modules.ts +83 -83
  132. package/tests/fixtures/errors/go.ts +9 -9
  133. package/tests/fixtures/errors/node.ts +24 -24
  134. package/tests/fixtures/errors/python.ts +21 -21
  135. package/tests/fixtures/errors/rust.ts +25 -25
  136. package/tests/fixtures/errors/shell.ts +15 -15
  137. package/tests/fixtures/solutions/solutions.ts +27 -27
  138. package/tests/helpers/setup-db.ts +52 -52
  139. package/tests/integration/code-flow.test.ts +86 -86
  140. package/tests/integration/error-flow.test.ts +83 -83
  141. package/tests/integration/ipc-flow.test.ts +166 -166
  142. package/tests/integration/learning-cycle.test.ts +82 -82
  143. package/tests/integration/synapse-flow.test.ts +117 -117
  144. package/tests/unit/code/analyzer.test.ts +58 -58
  145. package/tests/unit/code/fingerprint.test.ts +51 -51
  146. package/tests/unit/code/scorer.test.ts +55 -55
  147. package/tests/unit/learning/confidence-scorer.test.ts +60 -60
  148. package/tests/unit/learning/decay.test.ts +45 -45
  149. package/tests/unit/learning/pattern-extractor.test.ts +50 -50
  150. package/tests/unit/matching/error-matcher.test.ts +69 -69
  151. package/tests/unit/matching/fingerprint.test.ts +47 -47
  152. package/tests/unit/matching/similarity.test.ts +65 -65
  153. package/tests/unit/matching/tfidf.test.ts +71 -71
  154. package/tests/unit/matching/tokenizer.test.ts +83 -83
  155. package/tests/unit/parsing/parsers.test.ts +113 -113
  156. package/tests/unit/research/gap-analyzer.test.ts +45 -45
  157. package/tests/unit/research/trend-analyzer.test.ts +45 -45
  158. package/tests/unit/synapses/activation.test.ts +80 -80
  159. package/tests/unit/synapses/decay.test.ts +27 -27
  160. package/tests/unit/synapses/hebbian.test.ts +96 -96
  161. package/tests/unit/synapses/pathfinder.test.ts +72 -72
  162. package/tsconfig.json +18 -18
package/src/ipc/server.ts CHANGED
@@ -1,110 +1,176 @@
1
- import net from 'node:net';
2
- import { randomUUID } from 'node:crypto';
3
- import { getLogger } from '../utils/logger.js';
4
-
5
- const logger = getLogger();
6
- import type { IpcMessage } from '../types/ipc.types.js';
7
- import { encodeMessage, MessageDecoder } from './protocol.js';
8
- import type { IpcRouter } from './router.js';
9
-
10
- export class IpcServer {
11
- private server: net.Server | null = null;
12
- private clients = new Map<string, net.Socket>();
13
-
14
- constructor(
15
- private router: IpcRouter,
16
- private pipeName: string,
17
- ) {}
18
-
19
- start(): void {
20
- this.server = net.createServer((socket) => {
21
- const clientId = randomUUID();
22
- this.clients.set(clientId, socket);
23
- const decoder = new MessageDecoder();
24
-
25
- logger.info(`IPC client connected: ${clientId}`);
26
-
27
- socket.on('data', (chunk) => {
28
- const messages = decoder.feed(chunk);
29
- for (const msg of messages) {
30
- this.handleMessage(clientId, msg, socket);
31
- }
32
- });
33
-
34
- socket.on('close', () => {
35
- logger.info(`IPC client disconnected: ${clientId}`);
36
- this.clients.delete(clientId);
37
- });
38
-
39
- socket.on('error', (err) => {
40
- logger.error(`IPC client ${clientId} error:`, err);
41
- this.clients.delete(clientId);
42
- });
43
- });
44
-
45
- this.server.on('error', (err) => {
46
- logger.error('IPC server error:', err);
47
- });
48
-
49
- this.server.listen(this.pipeName, () => {
50
- logger.info(`IPC server listening on ${this.pipeName}`);
51
- });
52
- }
53
-
54
- private handleMessage(clientId: string, msg: IpcMessage, socket: net.Socket): void {
55
- if (msg.type !== 'request' || !msg.method) return;
56
-
57
- try {
58
- const result = this.router.handle(msg.method, msg.params);
59
- const response: IpcMessage = {
60
- id: msg.id,
61
- type: 'response',
62
- result,
63
- };
64
- socket.write(encodeMessage(response));
65
- } catch (err) {
66
- const response: IpcMessage = {
67
- id: msg.id,
68
- type: 'response',
69
- error: { code: -1, message: err instanceof Error ? err.message : String(err) },
70
- };
71
- socket.write(encodeMessage(response));
72
- }
73
- }
74
-
75
- notify(clientId: string | null, notification: Omit<IpcMessage, 'id' | 'type'>): void {
76
- const msg: IpcMessage = {
77
- id: randomUUID(),
78
- type: 'notification',
79
- ...notification,
80
- };
81
- const encoded = encodeMessage(msg);
82
-
83
- if (clientId) {
84
- const socket = this.clients.get(clientId);
85
- if (socket && !socket.destroyed) {
86
- socket.write(encoded);
87
- }
88
- } else {
89
- for (const socket of this.clients.values()) {
90
- if (!socket.destroyed) {
91
- socket.write(encoded);
92
- }
93
- }
94
- }
95
- }
96
-
97
- getClientCount(): number {
98
- return this.clients.size;
99
- }
100
-
101
- stop(): void {
102
- for (const socket of this.clients.values()) {
103
- socket.destroy();
104
- }
105
- this.clients.clear();
106
- this.server?.close();
107
- this.server = null;
108
- logger.info('IPC server stopped');
109
- }
110
- }
1
+ import net from 'node:net';
2
+ import fs from 'node:fs';
3
+ import { randomUUID } from 'node:crypto';
4
+ import { getLogger } from '../utils/logger.js';
5
+
6
+ const logger = getLogger();
7
+ import type { IpcMessage } from '../types/ipc.types.js';
8
+ import { encodeMessage, MessageDecoder } from './protocol.js';
9
+ import type { IpcRouter } from './router.js';
10
+
11
+ export class IpcServer {
12
+ private server: net.Server | null = null;
13
+ private clients = new Map<string, net.Socket>();
14
+
15
+ constructor(
16
+ private router: IpcRouter,
17
+ private pipeName: string,
18
+ ) {}
19
+
20
+ start(): void {
21
+ this.createServer();
22
+ this.listen();
23
+ }
24
+
25
+ private createServer(): void {
26
+ this.server = net.createServer((socket) => {
27
+ const clientId = randomUUID();
28
+ this.clients.set(clientId, socket);
29
+ const decoder = new MessageDecoder();
30
+
31
+ logger.info(`IPC client connected: ${clientId}`);
32
+
33
+ socket.on('data', (chunk) => {
34
+ const messages = decoder.feed(chunk);
35
+ for (const msg of messages) {
36
+ this.handleMessage(clientId, msg, socket);
37
+ }
38
+ });
39
+
40
+ socket.on('close', () => {
41
+ logger.info(`IPC client disconnected: ${clientId}`);
42
+ this.clients.delete(clientId);
43
+ });
44
+
45
+ socket.on('error', (err) => {
46
+ logger.error(`IPC client ${clientId} error:`, err);
47
+ this.clients.delete(clientId);
48
+ });
49
+ });
50
+ }
51
+
52
+ private listen(retried = false): void {
53
+ if (!this.server) return;
54
+
55
+ this.server.on('error', (err: NodeJS.ErrnoException) => {
56
+ if (err.code === 'EADDRINUSE' && !retried) {
57
+ logger.warn(`IPC pipe in use, attempting to recover stale pipe: ${this.pipeName}`);
58
+ this.recoverStalePipe();
59
+ } else {
60
+ logger.error('IPC server error:', err);
61
+ }
62
+ });
63
+
64
+ this.server.listen(this.pipeName, () => {
65
+ logger.info(`IPC server listening on ${this.pipeName}`);
66
+ });
67
+ }
68
+
69
+ /**
70
+ * On Windows, named pipes can remain after a crashed daemon.
71
+ * Try to connect as a client — if it fails, the pipe is stale and we can reclaim it.
72
+ * On Unix, simply unlink the socket file and retry.
73
+ */
74
+ private recoverStalePipe(): void {
75
+ const probe = net.createConnection(this.pipeName);
76
+
77
+ probe.on('connect', () => {
78
+ // Pipe is alive — another daemon is actually running
79
+ probe.destroy();
80
+ logger.error('IPC pipe is held by another running daemon. Stop it first with: brain stop');
81
+ });
82
+
83
+ probe.on('error', () => {
84
+ // Pipe is stale — no one is listening. Clean up and retry.
85
+ probe.destroy();
86
+ logger.info('Stale IPC pipe detected, reclaiming...');
87
+
88
+ // On Unix, unlink the socket file
89
+ if (process.platform !== 'win32') {
90
+ try { fs.unlinkSync(this.pipeName); } catch { /* ignore */ }
91
+ }
92
+
93
+ // Recreate server and retry (Windows auto-reclaims dead named pipes on re-listen)
94
+ this.createServer();
95
+ this.server!.on('error', (err) => {
96
+ logger.error('IPC server error after recovery:', err);
97
+ });
98
+ this.server!.listen(this.pipeName, () => {
99
+ logger.info(`IPC server recovered and listening on ${this.pipeName}`);
100
+ });
101
+ });
102
+
103
+ // Timeout: if probe hangs, treat pipe as stale
104
+ probe.setTimeout(2000, () => {
105
+ probe.destroy();
106
+ logger.warn('IPC pipe probe timed out, treating as stale');
107
+ if (process.platform !== 'win32') {
108
+ try { fs.unlinkSync(this.pipeName); } catch { /* ignore */ }
109
+ }
110
+ this.createServer();
111
+ this.server!.on('error', (err) => {
112
+ logger.error('IPC server error after timeout recovery:', err);
113
+ });
114
+ this.server!.listen(this.pipeName, () => {
115
+ logger.info(`IPC server recovered (timeout) and listening on ${this.pipeName}`);
116
+ });
117
+ });
118
+ }
119
+
120
+ private handleMessage(clientId: string, msg: IpcMessage, socket: net.Socket): void {
121
+ if (msg.type !== 'request' || !msg.method) return;
122
+
123
+ try {
124
+ const result = this.router.handle(msg.method, msg.params);
125
+ const response: IpcMessage = {
126
+ id: msg.id,
127
+ type: 'response',
128
+ result,
129
+ };
130
+ socket.write(encodeMessage(response));
131
+ } catch (err) {
132
+ const response: IpcMessage = {
133
+ id: msg.id,
134
+ type: 'response',
135
+ error: { code: -1, message: err instanceof Error ? err.message : String(err) },
136
+ };
137
+ socket.write(encodeMessage(response));
138
+ }
139
+ }
140
+
141
+ notify(clientId: string | null, notification: Omit<IpcMessage, 'id' | 'type'>): void {
142
+ const msg: IpcMessage = {
143
+ id: randomUUID(),
144
+ type: 'notification',
145
+ ...notification,
146
+ };
147
+ const encoded = encodeMessage(msg);
148
+
149
+ if (clientId) {
150
+ const socket = this.clients.get(clientId);
151
+ if (socket && !socket.destroyed) {
152
+ socket.write(encoded);
153
+ }
154
+ } else {
155
+ for (const socket of this.clients.values()) {
156
+ if (!socket.destroyed) {
157
+ socket.write(encoded);
158
+ }
159
+ }
160
+ }
161
+ }
162
+
163
+ getClientCount(): number {
164
+ return this.clients.size;
165
+ }
166
+
167
+ stop(): void {
168
+ for (const socket of this.clients.values()) {
169
+ socket.destroy();
170
+ }
171
+ this.clients.clear();
172
+ this.server?.close();
173
+ this.server = null;
174
+ logger.info('IPC server stopped');
175
+ }
176
+ }
@@ -1,46 +1,46 @@
1
- import type { SolutionRepository } from '../db/repositories/solution.repository.js';
2
- import { wilsonScore } from './confidence-scorer.js';
3
-
4
- /**
5
- * Update confidence scores for all solutions based on their attempt history.
6
- */
7
- export function updateSolutionConfidences(solutionRepo: SolutionRepository): number {
8
- const solutions = solutionRepo.getAll();
9
- let updated = 0;
10
-
11
- for (const sol of solutions) {
12
- const total = sol.success_count + sol.fail_count;
13
- if (total === 0) continue;
14
-
15
- const newConfidence = wilsonScore(sol.success_count, total);
16
- if (Math.abs(newConfidence - sol.confidence) > 0.001) {
17
- solutionRepo.update(sol.id, { confidence: newConfidence });
18
- updated++;
19
- }
20
- }
21
-
22
- return updated;
23
- }
24
-
25
- /**
26
- * Compute relevance decay factor for a timestamp.
27
- */
28
- export function relevanceDecay(timestamp: string, halfLifeDays: number): number {
29
- const ageDays = (Date.now() - new Date(timestamp).getTime()) / (1000 * 60 * 60 * 24);
30
- return Math.pow(0.5, ageDays / halfLifeDays);
31
- }
32
-
33
- /**
34
- * Determine if a rule should be pruned based on its performance.
35
- */
36
- export function shouldPruneRule(
37
- confidence: number,
38
- rejectionCount: number,
39
- totalUsage: number,
40
- pruneThreshold: number,
41
- maxRejectionRate: number,
42
- ): boolean {
43
- if (confidence < pruneThreshold) return true;
44
- if (totalUsage > 0 && rejectionCount / totalUsage > maxRejectionRate) return true;
45
- return false;
46
- }
1
+ import type { SolutionRepository } from '../db/repositories/solution.repository.js';
2
+ import { wilsonScore } from './confidence-scorer.js';
3
+
4
+ /**
5
+ * Update confidence scores for all solutions based on their attempt history.
6
+ */
7
+ export function updateSolutionConfidences(solutionRepo: SolutionRepository): number {
8
+ const solutions = solutionRepo.getAll();
9
+ let updated = 0;
10
+
11
+ for (const sol of solutions) {
12
+ const total = sol.success_count + sol.fail_count;
13
+ if (total === 0) continue;
14
+
15
+ const newConfidence = wilsonScore(sol.success_count, total);
16
+ if (Math.abs(newConfidence - sol.confidence) > 0.001) {
17
+ solutionRepo.update(sol.id, { confidence: newConfidence });
18
+ updated++;
19
+ }
20
+ }
21
+
22
+ return updated;
23
+ }
24
+
25
+ /**
26
+ * Compute relevance decay factor for a timestamp.
27
+ */
28
+ export function relevanceDecay(timestamp: string, halfLifeDays: number): number {
29
+ const ageDays = (Date.now() - new Date(timestamp).getTime()) / (1000 * 60 * 60 * 24);
30
+ return Math.pow(0.5, ageDays / halfLifeDays);
31
+ }
32
+
33
+ /**
34
+ * Determine if a rule should be pruned based on its performance.
35
+ */
36
+ export function shouldPruneRule(
37
+ confidence: number,
38
+ rejectionCount: number,
39
+ totalUsage: number,
40
+ pruneThreshold: number,
41
+ maxRejectionRate: number,
42
+ ): boolean {
43
+ if (confidence < pruneThreshold) return true;
44
+ if (totalUsage > 0 && rejectionCount / totalUsage > maxRejectionRate) return true;
45
+ return false;
46
+ }
@@ -1,90 +1,90 @@
1
- import type { ErrorRecord } from '../types/error.types.js';
2
- import { tokenize } from '../matching/tokenizer.js';
3
- import { cosineSimilarity } from '../matching/similarity.js';
4
-
5
- export interface ErrorPattern {
6
- errorType: string;
7
- messageTemplate: string;
8
- messageRegex: string;
9
- filePattern: string | null;
10
- occurrences: number;
11
- errorIds: number[];
12
- solutionIds: number[];
13
- confidence: number;
14
- successRate: number;
15
- }
16
-
17
- interface Centroid {
18
- errorType: string;
19
- tokens: string[];
20
- errorIds: number[];
21
- filePattern: string | null;
22
- }
23
-
24
- /**
25
- * Extract patterns from error records using centroid-based clustering.
26
- */
27
- export function extractPatterns(
28
- errors: ErrorRecord[],
29
- similarityThreshold: number = 0.7,
30
- ): ErrorPattern[] {
31
- const centroids: Centroid[] = [];
32
-
33
- for (const error of errors) {
34
- const tokens = tokenize(`${error.type} ${error.message}`);
35
- let merged = false;
36
-
37
- for (const centroid of centroids) {
38
- if (centroid.errorType !== error.type) continue;
39
-
40
- const sim = cosineSimilarity(centroid.tokens, tokens);
41
- if (sim >= similarityThreshold) {
42
- // Merge into existing centroid (running average)
43
- const allTokens = [...centroid.tokens, ...tokens];
44
- centroid.tokens = [...new Set(allTokens)];
45
- centroid.errorIds.push(error.id);
46
- if (!centroid.filePattern && error.file_path) {
47
- centroid.filePattern = extractFilePattern(error.file_path);
48
- }
49
- merged = true;
50
- break;
51
- }
52
- }
53
-
54
- if (!merged) {
55
- centroids.push({
56
- errorType: error.type,
57
- tokens,
58
- errorIds: [error.id],
59
- filePattern: error.file_path ? extractFilePattern(error.file_path) : null,
60
- });
61
- }
62
- }
63
-
64
- return centroids
65
- .filter(c => c.errorIds.length >= 2)
66
- .map(c => ({
67
- errorType: c.errorType,
68
- messageTemplate: c.tokens.join(' '),
69
- messageRegex: buildRegex(c.tokens),
70
- filePattern: c.filePattern,
71
- occurrences: c.errorIds.length,
72
- errorIds: c.errorIds,
73
- solutionIds: [],
74
- confidence: 0,
75
- successRate: 0,
76
- }));
77
- }
78
-
79
- function extractFilePattern(filePath: string): string {
80
- // Extract the meaningful part: last directory + extension
81
- const parts = filePath.replace(/\\/g, '/').split('/');
82
- const fileName = parts[parts.length - 1] ?? '';
83
- const ext = fileName.split('.').pop() ?? '';
84
- return ext ? `*.${ext}` : '*';
85
- }
86
-
87
- function buildRegex(tokens: string[]): string {
88
- const escaped = tokens.map(t => t.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'));
89
- return escaped.join('.*');
90
- }
1
+ import type { ErrorRecord } from '../types/error.types.js';
2
+ import { tokenize } from '../matching/tokenizer.js';
3
+ import { cosineSimilarity } from '../matching/similarity.js';
4
+
5
+ export interface ErrorPattern {
6
+ errorType: string;
7
+ messageTemplate: string;
8
+ messageRegex: string;
9
+ filePattern: string | null;
10
+ occurrences: number;
11
+ errorIds: number[];
12
+ solutionIds: number[];
13
+ confidence: number;
14
+ successRate: number;
15
+ }
16
+
17
+ interface Centroid {
18
+ errorType: string;
19
+ tokens: string[];
20
+ errorIds: number[];
21
+ filePattern: string | null;
22
+ }
23
+
24
+ /**
25
+ * Extract patterns from error records using centroid-based clustering.
26
+ */
27
+ export function extractPatterns(
28
+ errors: ErrorRecord[],
29
+ similarityThreshold: number = 0.7,
30
+ ): ErrorPattern[] {
31
+ const centroids: Centroid[] = [];
32
+
33
+ for (const error of errors) {
34
+ const tokens = tokenize(`${error.type} ${error.message}`);
35
+ let merged = false;
36
+
37
+ for (const centroid of centroids) {
38
+ if (centroid.errorType !== error.type) continue;
39
+
40
+ const sim = cosineSimilarity(centroid.tokens, tokens);
41
+ if (sim >= similarityThreshold) {
42
+ // Merge into existing centroid (running average)
43
+ const allTokens = [...centroid.tokens, ...tokens];
44
+ centroid.tokens = [...new Set(allTokens)];
45
+ centroid.errorIds.push(error.id);
46
+ if (!centroid.filePattern && error.file_path) {
47
+ centroid.filePattern = extractFilePattern(error.file_path);
48
+ }
49
+ merged = true;
50
+ break;
51
+ }
52
+ }
53
+
54
+ if (!merged) {
55
+ centroids.push({
56
+ errorType: error.type,
57
+ tokens,
58
+ errorIds: [error.id],
59
+ filePattern: error.file_path ? extractFilePattern(error.file_path) : null,
60
+ });
61
+ }
62
+ }
63
+
64
+ return centroids
65
+ .filter(c => c.errorIds.length >= 2)
66
+ .map(c => ({
67
+ errorType: c.errorType,
68
+ messageTemplate: c.tokens.join(' '),
69
+ messageRegex: buildRegex(c.tokens),
70
+ filePattern: c.filePattern,
71
+ occurrences: c.errorIds.length,
72
+ errorIds: c.errorIds,
73
+ solutionIds: [],
74
+ confidence: 0,
75
+ successRate: 0,
76
+ }));
77
+ }
78
+
79
+ function extractFilePattern(filePath: string): string {
80
+ // Extract the meaningful part: last directory + extension
81
+ const parts = filePath.replace(/\\/g, '/').split('/');
82
+ const fileName = parts[parts.length - 1] ?? '';
83
+ const ext = fileName.split('.').pop() ?? '';
84
+ return ext ? `*.${ext}` : '*';
85
+ }
86
+
87
+ function buildRegex(tokens: string[]): string {
88
+ const escaped = tokens.map(t => t.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'));
89
+ return escaped.join('.*');
90
+ }