@timmeck/brain 1.8.1 → 1.8.2

package/src/learning/rule-generator.ts

@@ -1,74 +1,74 @@
-import type { LearningConfig } from '../types/config.types.js';
-import type { RuleRepository } from '../db/repositories/rule.repository.js';
-import type { ErrorPattern } from './pattern-extractor.js';
-import { getLogger } from '../utils/logger.js';
-
-export interface GeneratedRule {
-  pattern: string;
-  action: string;
-  description: string;
-  confidence: number;
-  sourceErrorIds: number[];
-}
-
-/**
- * Generate prevention rules from extracted patterns.
- */
-export function generateRules(
-  patterns: ErrorPattern[],
-  config: LearningConfig,
-): GeneratedRule[] {
-  return patterns
-    .filter(p =>
-      p.occurrences >= config.minOccurrences &&
-      p.confidence >= config.minConfidence,
-    )
-    .map(pattern => ({
-      pattern: pattern.messageRegex,
-      action: pattern.confidence >= 0.90
-        ? `Auto-fix available for ${pattern.errorType}`
-        : `Suggestion: check ${pattern.errorType} pattern (${pattern.occurrences} occurrences)`,
-      description: `Auto-generated from ${pattern.occurrences} occurrences of ${pattern.errorType}`,
-      confidence: pattern.confidence,
-      sourceErrorIds: pattern.errorIds,
-    }));
-}
-
-/**
- * Persist generated rules to the database.
- */
-export function persistRules(
-  rules: GeneratedRule[],
-  ruleRepo: RuleRepository,
-  projectId?: number,
-): number {
-  const logger = getLogger();
-  let created = 0;
-
-  for (const rule of rules) {
-    // Check if similar rule already exists
-    const existing = ruleRepo.findByPattern(rule.pattern);
-    if (existing.length > 0) {
-      // Update confidence of existing rule
-      const best = existing[0]!;
-      if (rule.confidence > best.confidence) {
-        ruleRepo.update(best.id, { confidence: rule.confidence });
-      }
-      continue;
-    }
-
-    ruleRepo.create({
-      pattern: rule.pattern,
-      action: rule.action,
-      description: rule.description,
-      confidence: rule.confidence,
-      occurrences: 0,
-      active: 1,
-      project_id: projectId ?? null,
-    });
-    created++;
-    logger.info(`New rule generated: ${rule.pattern.substring(0, 50)}...`);
-  }
-
-  return created;
-}
+import type { LearningConfig } from '../types/config.types.js';
+import type { RuleRepository } from '../db/repositories/rule.repository.js';
+import type { ErrorPattern } from './pattern-extractor.js';
+import { getLogger } from '../utils/logger.js';
+
+export interface GeneratedRule {
+  pattern: string;
+  action: string;
+  description: string;
+  confidence: number;
+  sourceErrorIds: number[];
+}
+
+/**
+ * Generate prevention rules from extracted patterns.
+ */
+export function generateRules(
+  patterns: ErrorPattern[],
+  config: LearningConfig,
+): GeneratedRule[] {
+  return patterns
+    .filter(p =>
+      p.occurrences >= config.minOccurrences &&
+      p.confidence >= config.minConfidence,
+    )
+    .map(pattern => ({
+      pattern: pattern.messageRegex,
+      action: pattern.confidence >= 0.90
+        ? `Auto-fix available for ${pattern.errorType}`
+        : `Suggestion: check ${pattern.errorType} pattern (${pattern.occurrences} occurrences)`,
+      description: `Auto-generated from ${pattern.occurrences} occurrences of ${pattern.errorType}`,
+      confidence: pattern.confidence,
+      sourceErrorIds: pattern.errorIds,
+    }));
+}
+
+/**
+ * Persist generated rules to the database.
+ */
+export function persistRules(
+  rules: GeneratedRule[],
+  ruleRepo: RuleRepository,
+  projectId?: number,
+): number {
+  const logger = getLogger();
+  let created = 0;
+
+  for (const rule of rules) {
+    // Check if similar rule already exists
+    const existing = ruleRepo.findByPattern(rule.pattern);
+    if (existing.length > 0) {
+      // Update confidence of existing rule
+      const best = existing[0]!;
+      if (rule.confidence > best.confidence) {
+        ruleRepo.update(best.id, { confidence: rule.confidence });
+      }
+      continue;
+    }
+
+    ruleRepo.create({
+      pattern: rule.pattern,
+      action: rule.action,
+      description: rule.description,
+      confidence: rule.confidence,
+      occurrences: 0,
+      active: 1,
+      project_id: projectId ?? null,
+    });
+    created++;
+    logger.info(`New rule generated: ${rule.pattern.substring(0, 50)}...`);
+  }
+
+  return created;
+}

package/src/matching/error-matcher.ts

@@ -23,12 +23,12 @@ interface MatchSignal {
 
 // Base signals (used when vector search is NOT available)
 const SIGNALS_BASE: MatchSignal[] = [
-  { name: 'fingerprint', weight: 0.30, compute: fingerprintMatch },
-  { name: 'message_similarity', weight: 0.20, compute: messageSimilarity },
+  { name: 'fingerprint', weight: 0.20, compute: fingerprintMatch },
+  { name: 'message_similarity', weight: 0.25, compute: messageSimilarity },
   { name: 'type_match', weight: 0.15, compute: typeMatch },
   { name: 'stack_similarity', weight: 0.15, compute: stackSimilarity },
-  { name: 'file_similarity', weight: 0.10, compute: fileSimilarity },
-  { name: 'context_similarity', weight: 0.10, compute: contextSimilarity },
+  { name: 'file_similarity', weight: 0.12, compute: fileSimilarity },
+  { name: 'context_similarity', weight: 0.13, compute: contextSimilarity },
 ];
 
 // Hybrid signals (used when vector search IS available — vector gets 20% weight)
@@ -42,7 +42,7 @@ const SIGNALS_HYBRID: MatchSignal[] = [
 ];
 
 const VECTOR_WEIGHT = 0.20;
-const MATCH_THRESHOLD = 0.70;
+const MATCH_THRESHOLD = 0.55;
 const STRONG_MATCH_THRESHOLD = 0.90;
 
 /**

package/src/matching/fingerprint.ts

@@ -1,29 +1,34 @@
-import path from 'node:path';
-import { sha256 } from '../utils/hash.js';
-import type { StackFrame } from '../parsing/types.js';
-
-export function templateMessage(msg: string): string {
-  return msg
-    .replace(/[A-Z]:\\[\w\-.\\ ]+\.\w+/g, '<PATH>')
-    .replace(/\/[\w\-./ ]+\.\w+/g, '<PATH>')
-    .replace(/:(\d+):(\d+)/g, ':<LINE>:<COL>')
-    .replace(/line \d+/gi, 'line <LINE>')
-    .replace(/0x[0-9a-fA-F]+/g, '<ADDR>')
-    .replace(/[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}/gi, '<UUID>')
-    .replace(/https?:\/\/[^\s]+/g, '<URL>')
-    .replace(/\d{4}-\d{2}-\d{2}[T ]\d{2}:\d{2}:\d{2}/g, '<TIMESTAMP>');
-}
-
-export function generateFingerprint(
-  errorType: string,
-  message: string,
-  frames: StackFrame[],
-): string {
-  const template = templateMessage(message);
-  const topFrames = frames
-    .slice(0, 3)
-    .map(f => `${f.function_name || '<anon>'}@${path.basename(f.file_path || '<unknown>')}`)
-    .join('|');
-  const input = `${errorType}::${template}::${topFrames}`;
-  return sha256(input);
-}
+import path from 'node:path';
+import { sha256 } from '../utils/hash.js';
+import type { StackFrame } from '../parsing/types.js';
+
+export function templateMessage(msg: string): string {
+  return msg
+    .replace(/[A-Z]:\\[\w\-.\\ ]+\.\w+/g, '<PATH>')
+    .replace(/\/[\w\-./ ]+\.\w+/g, '<PATH>')
+    .replace(/:(\d+):(\d+)/g, ':<LINE>:<COL>')
+    .replace(/line \d+/gi, 'line <LINE>')
+    .replace(/0x[0-9a-fA-F]+/g, '<ADDR>')
+    .replace(/[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}/gi, '<UUID>')
+    .replace(/https?:\/\/[^\s]+/g, '<URL>')
+    .replace(/\d{4}-\d{2}-\d{2}[T ]\d{2}:\d{2}:\d{2}/g, '<TIMESTAMP>')
+    // Normalize JS/TS property access patterns so "reading 'map'" ≈ "reading 'forEach'"
+    .replace(/\(reading ['"][^'"]*['"]\)/g, "(reading '<PROP>')")
+    .replace(/\(writing ['"][^'"]*['"]\)/g, "(writing '<PROP>')")
+    // Normalize quoted identifiers (e.g., 'someVar', "someFunc")
+    .replace(/['"][a-zA-Z_$][\w$]*['"]/g, "'<IDENT>'");
+}
+
+export function generateFingerprint(
+  errorType: string,
+  message: string,
+  frames: StackFrame[],
+): string {
+  const template = templateMessage(message);
+  const topFrames = frames
+    .slice(0, 3)
+    .map(f => `${f.function_name || '<anon>'}@${path.basename(f.file_path || '<unknown>')}`)
+    .join('|');
+  const input = `${errorType}::${template}::${topFrames}`;
+  return sha256(input);
+}

package/src/matching/similarity.ts

@@ -1,61 +1,61 @@
-export function levenshteinDistance(a: string, b: string): number {
-  if (a === b) return 1.0;
-  if (a.length === 0 || b.length === 0) return 0.0;
-
-  const dp: number[][] = Array(b.length + 1)
-    .fill(0)
-    .map(() => Array(a.length + 1).fill(0) as number[]);
-
-  for (let i = 0; i <= a.length; i++) dp[0]![i] = i;
-  for (let j = 0; j <= b.length; j++) dp[j]![0] = j;
-
-  for (let i = 1; i <= b.length; i++) {
-    for (let j = 1; j <= a.length; j++) {
-      const cost = a[j - 1] === b[i - 1] ? 0 : 1;
-      dp[i]![j] = Math.min(
-        dp[i - 1]![j]! + 1,
-        dp[i]![j - 1]! + 1,
-        dp[i - 1]![j - 1]! + cost,
-      );
-    }
-  }
-
-  return 1 - dp[b.length]![a.length]! / Math.max(a.length, b.length);
-}
-
-export function cosineSimilarity(tokensA: string[], tokensB: string[]): number {
-  if (tokensA.length === 0 || tokensB.length === 0) return 0.0;
-
-  const vocab = new Set([...tokensA, ...tokensB]);
-  const vecA = new Map<string, number>();
-  const vecB = new Map<string, number>();
-
-  for (const t of tokensA) vecA.set(t, (vecA.get(t) ?? 0) + 1);
-  for (const t of tokensB) vecB.set(t, (vecB.get(t) ?? 0) + 1);
-
-  let dot = 0;
-  let magA = 0;
-  let magB = 0;
-
-  for (const word of vocab) {
-    const a = vecA.get(word) ?? 0;
-    const b = vecB.get(word) ?? 0;
-    dot += a * b;
-    magA += a * a;
-    magB += b * b;
-  }
-
-  const denom = Math.sqrt(magA) * Math.sqrt(magB);
-  return denom === 0 ? 0 : dot / denom;
-}
-
-export function jaccardSimilarity(tokensA: string[], tokensB: string[]): number {
-  if (tokensA.length === 0 && tokensB.length === 0) return 0.0;
-
-  const setA = new Set(tokensA);
-  const setB = new Set(tokensB);
-  const intersection = new Set([...setA].filter(x => setB.has(x)));
-  const union = new Set([...setA, ...setB]);
-
-  return union.size === 0 ? 0 : intersection.size / union.size;
-}
+export function levenshteinDistance(a: string, b: string): number {
+  if (a === b) return 1.0;
+  if (a.length === 0 || b.length === 0) return 0.0;
+
+  const dp: number[][] = Array(b.length + 1)
+    .fill(0)
+    .map(() => Array(a.length + 1).fill(0) as number[]);
+
+  for (let i = 0; i <= a.length; i++) dp[0]![i] = i;
+  for (let j = 0; j <= b.length; j++) dp[j]![0] = j;
+
+  for (let i = 1; i <= b.length; i++) {
+    for (let j = 1; j <= a.length; j++) {
+      const cost = a[j - 1] === b[i - 1] ? 0 : 1;
+      dp[i]![j] = Math.min(
+        dp[i - 1]![j]! + 1,
+        dp[i]![j - 1]! + 1,
+        dp[i - 1]![j - 1]! + cost,
+      );
+    }
+  }
+
+  return 1 - dp[b.length]![a.length]! / Math.max(a.length, b.length);
+}
+
+export function cosineSimilarity(tokensA: string[], tokensB: string[]): number {
+  if (tokensA.length === 0 || tokensB.length === 0) return 0.0;
+
+  const vocab = new Set([...tokensA, ...tokensB]);
+  const vecA = new Map<string, number>();
+  const vecB = new Map<string, number>();
+
+  for (const t of tokensA) vecA.set(t, (vecA.get(t) ?? 0) + 1);
+  for (const t of tokensB) vecB.set(t, (vecB.get(t) ?? 0) + 1);
+
+  let dot = 0;
+  let magA = 0;
+  let magB = 0;
+
+  for (const word of vocab) {
+    const a = vecA.get(word) ?? 0;
+    const b = vecB.get(word) ?? 0;
+    dot += a * b;
+    magA += a * a;
+    magB += b * b;
+  }
+
+  const denom = Math.sqrt(magA) * Math.sqrt(magB);
+  return denom === 0 ? 0 : dot / denom;
+}
+
+export function jaccardSimilarity(tokensA: string[], tokensB: string[]): number {
+  if (tokensA.length === 0 && tokensB.length === 0) return 0.0;
+
+  const setA = new Set(tokensA);
+  const setB = new Set(tokensB);
+  const intersection = new Set([...setA].filter(x => setB.has(x)));
+  const union = new Set([...setA, ...setB]);
+
+  return union.size === 0 ? 0 : intersection.size / union.size;
+}

package/src/matching/tfidf.ts

@@ -1,74 +1,74 @@
-export class TfIdfIndex {
-  private documents = new Map<number, string[]>();
-  private df = new Map<string, number>();
-  private idf = new Map<string, number>();
-  private documentCount = 0;
-
-  addDocument(id: number, tokens: string[]): void {
-    if (this.documents.has(id)) {
-      this.removeDocument(id);
-    }
-    const unique = new Set(tokens);
-    for (const token of unique) {
-      this.df.set(token, (this.df.get(token) ?? 0) + 1);
-    }
-    this.documents.set(id, tokens);
-    this.documentCount++;
-    this.recomputeIdfForTerms(unique);
-  }
-
-  removeDocument(id: number): void {
-    const tokens = this.documents.get(id);
-    if (!tokens) return;
-
-    const unique = new Set(tokens);
-    for (const token of unique) {
-      const count = this.df.get(token) ?? 0;
-      if (count <= 1) {
-        this.df.delete(token);
-        this.idf.delete(token);
-      } else {
-        this.df.set(token, count - 1);
-      }
-    }
-    this.documents.delete(id);
-    this.documentCount--;
-  }
-
-  query(tokens: string[], topK: number = 10): Array<{ id: number; score: number }> {
-    const scores = new Map<number, number>();
-
-    for (const token of tokens) {
-      const idfVal = this.idf.get(token) ?? 0;
-      if (idfVal === 0) continue;
-
-      for (const [docId, docTokens] of this.documents) {
-        const tf = docTokens.filter(t => t === token).length / docTokens.length;
-        const score = (scores.get(docId) ?? 0) + tf * idfVal;
-        scores.set(docId, score);
-      }
-    }
-
-    return Array.from(scores.entries())
-      .map(([id, score]) => ({ id, score }))
-      .sort((a, b) => b.score - a.score)
-      .slice(0, topK);
-  }
-
-  getDocumentCount(): number {
-    return this.documentCount;
-  }
-
-  getIdf(): ReadonlyMap<string, number> {
-    return this.idf;
-  }
-
-  private recomputeIdfForTerms(terms: Set<string>): void {
-    for (const term of terms) {
-      const dfVal = this.df.get(term) ?? 0;
-      if (dfVal > 0 && this.documentCount > 0) {
-        this.idf.set(term, Math.log(this.documentCount / dfVal));
-      }
-    }
-  }
-}
+export class TfIdfIndex {
+  private documents = new Map<number, string[]>();
+  private df = new Map<string, number>();
+  private idf = new Map<string, number>();
+  private documentCount = 0;
+
+  addDocument(id: number, tokens: string[]): void {
+    if (this.documents.has(id)) {
+      this.removeDocument(id);
+    }
+    const unique = new Set(tokens);
+    for (const token of unique) {
+      this.df.set(token, (this.df.get(token) ?? 0) + 1);
+    }
+    this.documents.set(id, tokens);
+    this.documentCount++;
+    this.recomputeIdfForTerms(unique);
+  }
+
+  removeDocument(id: number): void {
+    const tokens = this.documents.get(id);
+    if (!tokens) return;
+
+    const unique = new Set(tokens);
+    for (const token of unique) {
+      const count = this.df.get(token) ?? 0;
+      if (count <= 1) {
+        this.df.delete(token);
+        this.idf.delete(token);
+      } else {
+        this.df.set(token, count - 1);
+      }
+    }
+    this.documents.delete(id);
+    this.documentCount--;
+  }
+
+  query(tokens: string[], topK: number = 10): Array<{ id: number; score: number }> {
+    const scores = new Map<number, number>();
+
+    for (const token of tokens) {
+      const idfVal = this.idf.get(token) ?? 0;
+      if (idfVal === 0) continue;
+
+      for (const [docId, docTokens] of this.documents) {
+        const tf = docTokens.filter(t => t === token).length / docTokens.length;
+        const score = (scores.get(docId) ?? 0) + tf * idfVal;
+        scores.set(docId, score);
+      }
+    }
+
+    return Array.from(scores.entries())
+      .map(([id, score]) => ({ id, score }))
+      .sort((a, b) => b.score - a.score)
+      .slice(0, topK);
+  }
+
+  getDocumentCount(): number {
+    return this.documentCount;
+  }
+
+  getIdf(): ReadonlyMap<string, number> {
+    return this.idf;
+  }
+
+  private recomputeIdfForTerms(terms: Set<string>): void {
+    for (const term of terms) {
+      const dfVal = this.df.get(term) ?? 0;
+      if (dfVal > 0 && this.documentCount > 0) {
+        this.idf.set(term, Math.log(this.documentCount / dfVal));
+      }
+    }
+  }
+}

package/src/matching/tokenizer.ts

@@ -1,41 +1,41 @@
-const STOPWORDS = new Set([
-  'the', 'is', 'are', 'a', 'an', 'and', 'or', 'not', 'in', 'at', 'by', 'for',
-  'from', 'of', 'on', 'to', 'with', 'as', 'error', 'exception', 'throw', 'catch',
-  'was', 'be', 'been', 'being', 'have', 'has', 'had', 'do', 'does', 'did',
-  'will', 'would', 'could', 'should', 'may', 'might', 'can', 'it', 'its',
-  'this', 'that', 'these', 'those', 'i', 'we', 'you', 'he', 'she', 'they',
-]);
-
-export function splitCamelCase(text: string): string[] {
-  return text
-    .replace(/([A-Z]+)([A-Z][a-z])/g, '$1 $2')
-    .replace(/([a-z\d])([A-Z])/g, '$1 $2')
-    .split(/\s+/)
-    .filter(t => t.length > 0);
-}
-
-export function splitSnakeCase(text: string): string[] {
-  return text.split(/[_\-]+/).filter(t => t.length > 0);
-}
-
-export function removeStopwords(tokens: string[]): string[] {
-  return tokens.filter(t => !STOPWORDS.has(t.toLowerCase()));
-}
-
-export function tokenize(text: string): string[] {
-  const words = text
-    .replace(/[^\w\s]/g, ' ')
-    .split(/\s+/)
-    .filter(t => t.length > 0);
-
-  const tokens: string[] = [];
-  for (const word of words) {
-    tokens.push(...splitCamelCase(word));
-    if (word.includes('_') || word.includes('-')) {
-      tokens.push(...splitSnakeCase(word));
-    }
-  }
-
-  const cleaned = removeStopwords(tokens);
-  return [...new Set(cleaned.map(t => t.toLowerCase()))];
-}
+const STOPWORDS = new Set([
+  'the', 'is', 'are', 'a', 'an', 'and', 'or', 'not', 'in', 'at', 'by', 'for',
+  'from', 'of', 'on', 'to', 'with', 'as', 'error', 'exception', 'throw', 'catch',
+  'was', 'be', 'been', 'being', 'have', 'has', 'had', 'do', 'does', 'did',
+  'will', 'would', 'could', 'should', 'may', 'might', 'can', 'it', 'its',
+  'this', 'that', 'these', 'those', 'i', 'we', 'you', 'he', 'she', 'they',
+]);
+
+export function splitCamelCase(text: string): string[] {
+  return text
+    .replace(/([A-Z]+)([A-Z][a-z])/g, '$1 $2')
+    .replace(/([a-z\d])([A-Z])/g, '$1 $2')
+    .split(/\s+/)
+    .filter(t => t.length > 0);
+}
+
+export function splitSnakeCase(text: string): string[] {
+  return text.split(/[_\-]+/).filter(t => t.length > 0);
+}
+
+export function removeStopwords(tokens: string[]): string[] {
+  return tokens.filter(t => !STOPWORDS.has(t.toLowerCase()));
+}
+
+export function tokenize(text: string): string[] {
+  const words = text
+    .replace(/[^\w\s]/g, ' ')
+    .split(/\s+/)
+    .filter(t => t.length > 0);
+
+  const tokens: string[] = [];
+  for (const word of words) {
+    tokens.push(...splitCamelCase(word));
+    if (word.includes('_') || word.includes('-')) {
+      tokens.push(...splitSnakeCase(word));
+    }
+  }
+
+  const cleaned = removeStopwords(tokens);
+  return [...new Set(cleaned.map(t => t.toLowerCase()))];
+}