@in-the-loop-labs/pair-review 3.5.1 → 3.6.0

package/src/database.js

@@ -21,7 +21,7 @@ function getDbPath() {
 /**
  * Current schema version - increment this when adding new migrations
  */
-const CURRENT_SCHEMA_VERSION = 46;
+const CURRENT_SCHEMA_VERSION = 48;
 
 /**
  * Database schema SQL statements
@@ -270,6 +270,36 @@ const SCHEMA_SQL = {
     )
   `,
 
+  hunk_summaries: `
+    CREATE TABLE IF NOT EXISTS hunk_summaries (
+      id INTEGER PRIMARY KEY AUTOINCREMENT,
+      review_id INTEGER NOT NULL,
+      file_path TEXT NOT NULL,
+      content_hash TEXT NOT NULL,
+      summary_text TEXT,
+      trivial_reason TEXT,
+      provider TEXT,
+      model TEXT,
+      created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+      FOREIGN KEY (review_id) REFERENCES reviews(id) ON DELETE CASCADE,
+      CHECK (summary_text IS NOT NULL OR trivial_reason IS NOT NULL),
+      UNIQUE (review_id, content_hash)
+    )
+  `,
+
+  tours: `
+    CREATE TABLE IF NOT EXISTS tours (
+      id INTEGER PRIMARY KEY AUTOINCREMENT,
+      review_id INTEGER NOT NULL UNIQUE,
+      stops TEXT NOT NULL,
+      diff_hash TEXT NOT NULL,
+      provider TEXT,
+      model TEXT,
+      created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+      FOREIGN KEY (review_id) REFERENCES reviews(id) ON DELETE CASCADE
+    )
+  `,
+
   github_pr_cache: `
     CREATE TABLE IF NOT EXISTS github_pr_cache (
       id INTEGER PRIMARY KEY AUTOINCREMENT,
@@ -365,6 +395,8 @@ const INDEX_SQL = [
   'CREATE INDEX IF NOT EXISTS idx_chat_messages_session ON chat_messages(session_id)',
   // Context files indexes
   'CREATE INDEX IF NOT EXISTS idx_context_files_review ON context_files(review_id)',
+  // Hunk summaries indexes
+  'CREATE INDEX IF NOT EXISTS idx_hunk_summaries_review ON hunk_summaries(review_id)',
   // GitHub PR cache indexes
   'CREATE UNIQUE INDEX IF NOT EXISTS idx_github_pr_cache_unique ON github_pr_cache(collection, owner, repo, number)',
   // Worktree pool indexes
@@ -2078,6 +2110,55 @@ const MIGRATIONS = {
 
     console.log('  Rebuilt external_comments with ON DELETE SET NULL on parent_id');
     console.log('Migration to schema version 46 complete');
+  },
+
+  47: (db) => {
+    console.log('Running migration to schema version 47: Add hunk_summaries table...');
+    if (!tableExists(db, 'hunk_summaries')) {
+      db.exec(`
+        CREATE TABLE IF NOT EXISTS hunk_summaries (
+          id INTEGER PRIMARY KEY AUTOINCREMENT,
+          review_id INTEGER NOT NULL,
+          file_path TEXT NOT NULL,
+          content_hash TEXT NOT NULL,
+          summary_text TEXT,
+          trivial_reason TEXT,
+          provider TEXT,
+          model TEXT,
+          created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+          FOREIGN KEY (review_id) REFERENCES reviews(id) ON DELETE CASCADE,
+          CHECK (summary_text IS NOT NULL OR trivial_reason IS NOT NULL),
+          UNIQUE (review_id, content_hash)
+        )
+      `);
+      console.log('  Created hunk_summaries table');
+    } else {
+      console.log('  Table hunk_summaries already exists');
+    }
+    db.exec('CREATE INDEX IF NOT EXISTS idx_hunk_summaries_review ON hunk_summaries(review_id)');
+    console.log('Migration to schema version 47 complete');
+  },
+
+  48: (db) => {
+    console.log('Running migration to schema version 48: Add tours table...');
+    if (!tableExists(db, 'tours')) {
+      db.exec(`
+        CREATE TABLE IF NOT EXISTS tours (
+          id INTEGER PRIMARY KEY AUTOINCREMENT,
+          review_id INTEGER NOT NULL UNIQUE,
+          stops TEXT NOT NULL,
+          diff_hash TEXT NOT NULL,
+          provider TEXT,
+          model TEXT,
+          created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+          FOREIGN KEY (review_id) REFERENCES reviews(id) ON DELETE CASCADE
+        )
+      `);
+      console.log('  Created tours table');
+    } else {
+      console.log('  Table tours already exists');
+    }
+    console.log('Migration to schema version 48 complete');
   }
 };
 
@@ -5546,6 +5627,204 @@ class ContextFileRepository {
   }
 }
 
+/**
+ * HunkSummaryRepository class for managing per-hunk natural-language summary
+ * records. Summaries are generated by background AI jobs and keyed by content
+ * hash so that unchanged hunks don't get re-enqueued across reloads.
+ *
+ * Trivial hunks (whitespace, imports, version bumps, etc.) are persisted with
+ * `summary_text = NULL` and a non-null `trivial_reason` so the missing-hash
+ * lookup treats them as "already known, don't re-enqueue". The frontend hides
+ * trivial rows.
+ */
+class HunkSummaryRepository {
+  /**
+   * Create a new HunkSummaryRepository instance
+   * @param {Database} db - Database instance
+   */
+  constructor(db) {
+    this.db = db;
+  }
+
+  /**
+   * Get all hunk summary rows for a review, ordered by file_path then id
+   * @param {number} reviewId - Review ID
+   * @returns {Promise<Array<Object>>} Array of hunk summary rows
+   */
+  async getByReview(reviewId) {
+    return query(
+      this.db,
+      'SELECT * FROM hunk_summaries WHERE review_id = ? ORDER BY file_path, id',
+      [reviewId]
+    );
+  }
+
+  /**
+   * Get all hunk summary rows for a review scoped to a single file path,
+   * ordered by id.
+   * @param {number} reviewId - Review ID
+   * @param {string} filePath - File path to scope to
+   * @returns {Promise<Array<Object>>} Array of hunk summary rows
+   */
+  async getByReviewAndFile(reviewId, filePath) {
+    return query(
+      this.db,
+      'SELECT * FROM hunk_summaries WHERE review_id = ? AND file_path = ? ORDER BY id',
+      [reviewId, filePath]
+    );
+  }
+
+  /**
+   * Get hunk summary rows for a specific set of content hashes within a review.
+   * Used to identify which hashes already have persisted summaries and which
+   * still need to be enqueued.
+   * @param {number} reviewId - Review ID
+   * @param {Array<string>} hashes - Content hashes to look up
+   * @returns {Promise<Array<Object>>} Array of matching hunk summary rows
+   */
+  async getByHashes(reviewId, hashes) {
+    if (!Array.isArray(hashes) || hashes.length === 0) {
+      return [];
+    }
+    const placeholders = hashes.map(() => '?').join(',');
+    return query(
+      this.db,
+      `SELECT * FROM hunk_summaries WHERE review_id = ? AND content_hash IN (${placeholders})`,
+      [reviewId, ...hashes]
+    );
+  }
+
+  /**
+   * Insert or update many hunk summary rows in a single transaction.
+   * On conflict by (review_id, content_hash), the existing row's mutable
+   * fields are overwritten.
+   * @param {Array<Object>} rows - Rows to upsert. Each row should have
+   *   {review_id, file_path, content_hash, summary_text?, trivial_reason?, provider?, model?}
+   * @returns {Promise<number>} Number of rows processed
+   */
+  async upsertMany(rows) {
+    if (!Array.isArray(rows) || rows.length === 0) {
+      return 0;
+    }
+    return withTransaction(this.db, () => {
+      const stmt = this.db.prepare(`
+        INSERT INTO hunk_summaries (
+          review_id, file_path, content_hash, summary_text, trivial_reason, provider, model
+        ) VALUES (?, ?, ?, ?, ?, ?, ?)
+        ON CONFLICT(review_id, content_hash) DO UPDATE SET
+          file_path = excluded.file_path,
+          summary_text = excluded.summary_text,
+          trivial_reason = excluded.trivial_reason,
+          provider = excluded.provider,
+          model = excluded.model
+      `);
+      let count = 0;
+      for (const row of rows) {
+        if (row.summary_text == null && row.trivial_reason == null) {
+          throw new Error(
+            `HunkSummaryRepository.upsertMany: row must set summary_text or trivial_reason ` +
+            `(review_id=${row.review_id}, content_hash=${row.content_hash})`
+          );
+        }
+        stmt.run(
+          row.review_id,
+          row.file_path,
+          row.content_hash,
+          row.summary_text ?? null,
+          row.trivial_reason ?? null,
+          row.provider ?? null,
+          row.model ?? null
+        );
+        count++;
+      }
+      return count;
+    });
+  }
+
+  /**
+   * Delete all hunk summary rows for a review
+   * @param {number} reviewId - Review ID
+   * @returns {Promise<Object>} Run result with `changes` count
+   */
+  async deleteByReview(reviewId) {
+    return run(this.db, 'DELETE FROM hunk_summaries WHERE review_id = ?', [reviewId]);
+  }
+}
+
+/**
+ * TourRepository class for managing per-review guided-tour records. A tour is
+ * a single ordered narrative walkthrough (a JSON array of stops). `diff_hash`
+ * is a 16-char SHA-256 prefix of the diff text; the tour-generator uses it to
+ * detect staleness (when the diff changes, the tour is regenerated).
+ *
+ * One row per review (review_id is UNIQUE). On regeneration, upsert replaces
+ * the existing row.
+ */
+class TourRepository {
+  /**
+   * Create a new TourRepository instance
+   * @param {Database} db - Database instance
+   */
+  constructor(db) {
+    this.db = db;
+  }
+
+  /**
+   * Get the tour row for a review.
+   * `stops` is returned as a raw JSON string; the caller parses.
+   * @param {number} reviewId - Review ID
+   * @returns {Promise<Object|undefined>} The tour row or undefined if none
+   */
+  async get(reviewId) {
+    return queryOne(
+      this.db,
+      'SELECT * FROM tours WHERE review_id = ?',
+      [reviewId]
+    );
+  }
+
+  /**
+   * Insert or replace the tour row for a review. `stops` is stored verbatim;
+   * the caller is responsible for JSON.stringify.
+   * @param {Object} row - { review_id, stops, diff_hash, provider?, model? }
+   * @returns {Promise<Object>} Run result with `changes` count
+   */
+  async upsert(row) {
+    if (row.review_id == null) {
+      throw new Error('TourRepository.upsert: row.review_id is required');
+    }
+    return run(
+      this.db,
+      `
+        INSERT INTO tours (review_id, stops, diff_hash, provider, model)
+        VALUES (?, ?, ?, ?, ?)
+        ON CONFLICT(review_id) DO UPDATE SET
+          stops = excluded.stops,
+          diff_hash = excluded.diff_hash,
+          provider = excluded.provider,
+          model = excluded.model,
+          created_at = CURRENT_TIMESTAMP
+      `,
+      [
+        row.review_id,
+        row.stops,
+        row.diff_hash,
+        row.provider ?? null,
+        row.model ?? null,
+      ]
+    );
+  }
+
+  /**
+   * Delete the tour row for a review.
+   * @param {number} reviewId - Review ID
+   * @returns {Promise<Object>} Run result with `changes` count
+   */
+  async deleteByReview(reviewId) {
+    return run(this.db, 'DELETE FROM tours WHERE review_id = ?', [reviewId]);
+  }
+}
+
 module.exports = {
   initializeDatabase,
   closeDatabase,
@@ -5571,6 +5850,8 @@ module.exports = {
   GitHubReviewRepository,
   CouncilRepository,
   ContextFileRepository,
+  HunkSummaryRepository,
+  TourRepository,
   generateWorktreeId,
   migrateExistingWorktrees,
   // Exported for testing only