fix: recompute Jaccard from total file counts during incremental co-change analysis

The incremental upsert was overwriting Jaccard with a value computed
only from new commits, ignoring historical data. Now stores per-file
commit counts in a new file_commit_counts table (migration v6) and
recomputes Jaccard from the accumulated totals after each merge.

Author: github-actions[bot]

src/cochange.js

@@ -124,7 +124,7 @@ export function computeCoChanges(commits, opts = {}) {
     });
   }
 
-  return results;
+  return { pairs: results, fileCommitCounts };
 }
 
 /**
@@ -170,6 +170,7 @@ export function analyzeCoChanges(customDbPath, opts = {}) {
   if (opts.full) {
     db.exec('DELETE FROM co_changes');
     db.exec('DELETE FROM co_change_meta');
+    db.exec('DELETE FROM file_commit_counts');
   }
 
   // Collect known files from the graph for filtering
@@ -182,25 +183,53 @@ export function analyzeCoChanges(customDbPath, opts = {}) {
   }
 
   const { commits } = scanGitHistory(repoRoot, { since, afterSha });
-  const coChanges = computeCoChanges(commits, { minSupport, maxFilesPerCommit, knownFiles });
+  const { pairs: coChanges, fileCommitCounts } = computeCoChanges(commits, {
+    minSupport,
+    maxFilesPerCommit,
+    knownFiles,
+  });
 
-  // Write results
-  const upsert = db.prepare(`
+  // Upsert per-file commit counts so Jaccard can be recomputed from totals
+  const fileCountUpsert = db.prepare(`
+    INSERT INTO file_commit_counts (file, commit_count) VALUES (?, ?)
+    ON CONFLICT(file) DO UPDATE SET commit_count = commit_count + excluded.commit_count
+  `);
+
+  // Upsert pair counts (accumulate commit_count, jaccard placeholder — recomputed below)
+  const pairUpsert = db.prepare(`
     INSERT INTO co_changes (file_a, file_b, commit_count, jaccard, last_commit_epoch)
-    VALUES (?, ?, ?, ?, ?)
+    VALUES (?, ?, ?, 0, ?)
     ON CONFLICT(file_a, file_b) DO UPDATE SET
       commit_count = commit_count + excluded.commit_count,
-      jaccard = excluded.jaccard,
       last_commit_epoch = MAX(co_changes.last_commit_epoch, excluded.last_commit_epoch)
   `);
 
-  const insertMany = db.transaction((pairs) => {
-    for (const [key, data] of pairs) {
+  const insertMany = db.transaction(() => {
+    for (const [file, count] of fileCommitCounts) {
+      fileCountUpsert.run(file, count);
+    }
+    for (const [key, data] of coChanges) {
       const [fileA, fileB] = key.split('\0');
-      upsert.run(fileA, fileB, data.commitCount, data.jaccard, data.lastEpoch);
+      pairUpsert.run(fileA, fileB, data.commitCount, data.lastEpoch);
     }
   });
-  insertMany(coChanges);
+  insertMany();
+
+  // Recompute Jaccard for all affected pairs from total file commit counts
+  const affectedFiles = [...fileCommitCounts.keys()];
+  if (affectedFiles.length > 0) {
+    const ph = affectedFiles.map(() => '?').join(',');
+    db.prepare(`
+      UPDATE co_changes SET jaccard = (
+        SELECT CAST(co_changes.commit_count AS REAL) / (
+          COALESCE(fa.commit_count, 0) + COALESCE(fb.commit_count, 0) - co_changes.commit_count
+        )
+        FROM file_commit_counts fa, file_commit_counts fb
+        WHERE fa.file = co_changes.file_a AND fb.file = co_changes.file_b
+      )
+      WHERE file_a IN (${ph}) OR file_b IN (${ph})
+    `).run(...affectedFiles, ...affectedFiles);
+  }
 
   // Update metadata
   const metaUpsert = db.prepare(`

src/db.js

@@ -92,6 +92,15 @@ export const MIGRATIONS = [
       );
     `,
   },
+  {
+    version: 6,
+    up: `
+      CREATE TABLE IF NOT EXISTS file_commit_counts (
+        file TEXT PRIMARY KEY,
+        commit_count INTEGER NOT NULL DEFAULT 0
+      );
+    `,
+  },
 ];
 
 export function openDb(dbPath) {

tests/integration/cochange.test.js

@@ -32,7 +32,7 @@ describe('computeCoChanges', () => {
     ];
     // a.js appears in 4 commits, b.js in 3, pair(a,b) = 3
     // jaccard(a,b) = 3 / (4 + 3 - 3) = 3/4 = 0.75
-    const result = computeCoChanges(commits, { minSupport: 1 });
+    const { pairs: result } = computeCoChanges(commits, { minSupport: 1 });
     const abKey = 'a.js\0b.js';
     expect(result.has(abKey)).toBe(true);
     expect(result.get(abKey).jaccard).toBeCloseTo(0.75);
@@ -45,7 +45,7 @@ describe('computeCoChanges', () => {
       { sha: 'a2', epoch: 2000, files: ['a.js', 'b.js'] },
       { sha: 'a3', epoch: 3000, files: ['a.js', 'c.js'] },
     ];
-    const result = computeCoChanges(commits, { minSupport: 3 });
+    const { pairs: result } = computeCoChanges(commits, { minSupport: 3 });
     // pair(a,b) only has 2 co-occurrences, pair(a,c) only 1
     expect(result.size).toBe(0);
   });
@@ -57,7 +57,7 @@ describe('computeCoChanges', () => {
       { sha: 'a3', epoch: 3000, files: ['a.js', 'b.js'] },
       { sha: 'a4', epoch: 4000, files: ['a.js', 'b.js'] },
     ];
-    const result = computeCoChanges(commits, { minSupport: 3, maxFilesPerCommit: 3 });
+    const { pairs: result } = computeCoChanges(commits, { minSupport: 3, maxFilesPerCommit: 3 });
     // First commit skipped (4 files > max 3)
     // pair(a,b) = 3 from commits a2,a3,a4; a appears in 3 commits, b in 3
     // jaccard = 3/(3+3-3) = 1.0
@@ -72,14 +72,14 @@ describe('computeCoChanges', () => {
       { sha: 'a2', epoch: 2000, files: ['z.js', 'a.js'] },
       { sha: 'a3', epoch: 3000, files: ['z.js', 'a.js'] },
     ];
-    const result = computeCoChanges(commits, { minSupport: 1 });
+    const { pairs: result } = computeCoChanges(commits, { minSupport: 1 });
     // Should be stored as a.js < z.js
     expect(result.has('a.js\0z.js')).toBe(true);
     expect(result.has('z.js\0a.js')).toBe(false);
   });
 
   test('empty input returns empty map', () => {
-    const result = computeCoChanges([], { minSupport: 1 });
+    const { pairs: result } = computeCoChanges([], { minSupport: 1 });
     expect(result.size).toBe(0);
   });
 
@@ -89,7 +89,7 @@ describe('computeCoChanges', () => {
       { sha: 'a2', epoch: 5000, files: ['a.js', 'b.js'] },
       { sha: 'a3', epoch: 3000, files: ['a.js', 'b.js'] },
     ];
-    const result = computeCoChanges(commits, { minSupport: 1 });
+    const { pairs: result } = computeCoChanges(commits, { minSupport: 1 });
     expect(result.get('a.js\0b.js').lastEpoch).toBe(5000);
   });
 
@@ -100,7 +100,7 @@ describe('computeCoChanges', () => {
       { sha: 'a3', epoch: 3000, files: ['a.js', 'b.js', 'c.js'] },
     ];
     const knownFiles = new Set(['a.js', 'b.js']);
-    const result = computeCoChanges(commits, { minSupport: 1, knownFiles });
+    const { pairs: result } = computeCoChanges(commits, { minSupport: 1, knownFiles });
     expect(result.has('a.js\0b.js')).toBe(true);
     // c.js pairs should not exist
     expect(result.has('a.js\0c.js')).toBe(false);