refactor(db): use shared pooled DatabaseClient singleton instead of per-call instances

- Replace get_db_client() creating new DatabaseClient on every call with a
  module-level singleton initialized once at startup via init_db_client()
- Add init_db_client() and close_db_client() lifecycle functions called
  from FastAPI lifespan handler
- Migrate all DatabaseClient methods from legacy self.connect()/self.conn
  to pooled self.get_conn() context manager for thread-safe connection reuse
- Pool is properly torn down on application shutdown

Closes leeworks-agents/SPARC#7

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

README.md

@@ -54,6 +54,21 @@ docker-compose up -d
 # - API Docs: http://localhost:8000/docs
 ```
 
+#### Patent PDF Storage
+
+The API stores downloaded patent PDFs in a `patents/` directory. In Docker,
+this is mounted as a bind mount (`./patents:/app/patents`) so that PDFs persist
+across container restarts.
+
+If you deploy to a different environment, ensure the `patents/` directory is a
+persistent volume. Without it, PDFs will be re-downloaded on every analysis.
+
+```yaml
+# docker-compose.yml excerpt
+volumes:
+  - ./patents:/app/patents
+```
+
 ### NixOS
 
 ```bash

SPARC/analyzer.py

@@ -104,21 +104,33 @@ class CompanyAnalyzer:
     def analyze_single_patent(self, patent_id: str, company_name: str) -> str:
         """Analyze a single patent by ID.
 
-        Useful for focused analysis of specific innovations.
+        Prerequisite:
+            The patent PDF must already exist at ``patents/{patent_id}.pdf``
+            before calling this method. PDFs are downloaded automatically when
+            using the batch analysis pipeline (``analyze_company`` or the
+            ``/analyze/batch`` API endpoint). For standalone usage, download
+            the PDF manually or call ``SERP.save_patents()`` first.
 
         Args:
-          patent_id: Publication ID of the patent
+          patent_id: Publication ID of the patent (e.g. "US-11234567-B2")
           company_name: Name of the company (for context)
 
         Returns:
           Analysis of the specific patent's innovation quality
+
+        Raises:
+          FileNotFoundError: If the patent PDF is not found at the expected path.
         """
-        # Note: This simplified version assumes the patent PDF is already downloaded
-        # A more complete implementation would support direct patent ID lookup
-        print(f"Analyzing patent {patent_id} for {company_name}...")
+        import os
 
         patent_path = f"patents/{patent_id}.pdf"
 
+        if not os.path.exists(patent_path):
+            raise FileNotFoundError(
+                f"Patent PDF not found at '{patent_path}'. "
+                f"Download the PDF first using SERP.save_patents() or the batch analysis pipeline."
+            )
+
         try:
             sections = SERP.parse_patent_pdf(patent_path)
             minimized_content = SERP.minimize_patent_for_llm(sections)
@@ -129,6 +141,8 @@ class CompanyAnalyzer:
 
             return analysis
 
+        except FileNotFoundError:
+            raise
         except Exception as e:
             return f"Failed to analyze patent {patent_id}: {e}"
 

SPARC/api.py

@@ -7,9 +7,13 @@ from contextlib import asynccontextmanager
 from datetime import datetime
 from typing import Annotated, List
 
-from fastapi import BackgroundTasks, Depends, FastAPI, HTTPException, Query
+from fastapi import BackgroundTasks, Depends, FastAPI, HTTPException, Query, Request
 from fastapi.middleware.cors import CORSMiddleware
+from fastapi.responses import JSONResponse
 from pydantic import BaseModel, EmailStr, Field
+from slowapi import Limiter
+from slowapi.errors import RateLimitExceeded
+from slowapi.util import get_remote_address
 
 from SPARC import config
 from SPARC.analyzer import CompanyAnalyzer
@@ -17,11 +21,13 @@ from SPARC.auth import (
     TokenResponse,
     UserResponse,
     check_jwt_secret,
+    close_db_client,
     create_tokens,
     decode_token,
     get_current_admin,
     get_current_user,
     get_db_client,
+    init_db_client,
 )
 from SPARC.types import BatchAnalysisResult, CompanyAnalysisResult
 
@@ -115,8 +121,7 @@ class AnalyticsResponse(BaseModel):
     period_days: int
 
 
-# In-memory job storage (for demo; production would use Redis/DB)
-_jobs: dict[str, JobStatus] = {}
+# Job counter for generating unique IDs (the actual state is in PostgreSQL)
 _job_counter = 0
 
 
@@ -149,13 +154,25 @@ _analyzer: CompanyAnalyzer | None = None
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    """Initialize resources on startup."""
+    """Initialize resources on startup, clean up on shutdown."""
     global _analyzer
     check_jwt_secret()
+    init_db_client()
     _analyzer = CompanyAnalyzer()
+    # Mark any jobs that were running/pending before the restart as failed
+    from SPARC.database import DatabaseClient
+    _db = DatabaseClient(config.database_url)
+    _db.connect()
+    _db.initialize_schema()
+    stale = _db.mark_stale_jobs_failed()
+    if stale:
+        import logging
+        logging.getLogger(__name__).warning("Marked %d stale jobs as failed on startup", stale)
+    _db.close()
     yield
-    # Cleanup if needed
+    # Cleanup
     _analyzer = None
+    close_db_client()
 
 
 app = FastAPI(
@@ -166,6 +183,22 @@ app = FastAPI(
     root_path=config.root_path,
 )
 
+# Rate limiter (in-memory storage, suitable for single-instance deployments)
+limiter = Limiter(key_func=get_remote_address)
+app.state.limiter = limiter
+
+
+@app.exception_handler(RateLimitExceeded)
+async def rate_limit_handler(request: Request, exc: RateLimitExceeded):
+    """Return 429 with Retry-After header when rate limit is exceeded."""
+    retry_after = getattr(exc, "retry_after", 60)
+    return JSONResponse(
+        status_code=429,
+        content={"detail": "Rate limit exceeded. Please try again later."},
+        headers={"Retry-After": str(retry_after)},
+    )
+
+
 # Add CORS middleware for React frontend
 app.add_middleware(
     CORSMiddleware,
@@ -180,7 +213,8 @@ app.add_middleware(
 
 
 @app.post("/auth/register", response_model=UserResponse, tags=["Auth"])
-async def register(request: RegisterRequest):
+@limiter.limit("5/minute")
+async def register(request: Request, body: RegisterRequest):
     """Register a new user.
 
     The first registered user automatically becomes an admin.
@@ -192,8 +226,8 @@ async def register(request: RegisterRequest):
     role = "admin" if user_count == 0 else "user"
 
     user = db.create_user(
-        email=request.email,
-        password=request.password,
+        email=body.email,
+        password=body.password,
         role=role,
     )
 
@@ -212,11 +246,12 @@ async def register(request: RegisterRequest):
 
 
 @app.post("/auth/login", response_model=TokenResponse, tags=["Auth"])
-async def login(request: LoginRequest):
+@limiter.limit("10/minute")
+async def login(request: Request, body: LoginRequest):
     """Authenticate user and return JWT tokens."""
     db = get_db_client()
 
-    user = db.authenticate_user(request.email, request.password)
+    user = db.authenticate_user(body.email, body.password)
 
     if not user:
         raise HTTPException(
@@ -424,20 +459,52 @@ async def analyze_companies_batch(
     return _convert_batch_result(result)
 
 
+def _get_job_db() -> "DatabaseClient":
+    """Get a DatabaseClient for job persistence."""
+    from SPARC.database import DatabaseClient
+    db = DatabaseClient(config.database_url)
+    return db
+
+
+def _job_row_to_status(row: dict) -> JobStatus:
+    """Convert a database job row to a JobStatus model."""
+    import json as _json
+    result = None
+    if row.get("result_json"):
+        result_data = row["result_json"]
+        if isinstance(result_data, str):
+            result_data = _json.loads(result_data)
+        result = BatchAnalysisResponse(**result_data)
+    return JobStatus(
+        job_id=row["job_id"],
+        status=row["status"],
+        progress=row["progress"],
+        total_companies=row["total_companies"],
+        completed_companies=row["completed_companies"],
+        result=result,
+        error=row.get("error"),
+    )
+
+
 def _run_batch_job(job_id: str, companies: list[str], max_workers: int):
     """Background task for batch analysis."""
-    global _jobs, _analyzer
+    import json as _json
+    global _analyzer
+
+    db = _get_job_db()
 
     if not _analyzer:
-        _jobs[job_id].status = "failed"
-        _jobs[job_id].error = "Analyzer not initialized"
+        db.update_job(job_id, status="failed", error="Analyzer not initialized")
         return
 
-    _jobs[job_id].status = "running"
+    db.update_job(job_id, status="running")
 
     def progress_callback(company: str, completed: int, total: int):
-        _jobs[job_id].completed_companies = completed
-        _jobs[job_id].progress = int((completed / total) * 100)
+        db.update_job(
+            job_id,
+            completed_companies=completed,
+            progress=int((completed / total) * 100),
+        )
 
     try:
         result = _analyzer.analyze_companies(
@@ -445,12 +512,15 @@ def _run_batch_job(job_id: str, companies: list[str], max_workers: int):
             max_workers=max_workers,
             progress_callback=progress_callback,
         )
-        _jobs[job_id].status = "completed"
-        _jobs[job_id].progress = 100
-        _jobs[job_id].result = _convert_batch_result(result)
+        batch_response = _convert_batch_result(result)
+        db.update_job(
+            job_id,
+            status="completed",
+            progress=100,
+            result_json=_json.dumps(batch_response.model_dump(), default=str),
+        )
     except Exception as e:
-        _jobs[job_id].status = "failed"
-        _jobs[job_id].error = str(e)
+        db.update_job(job_id, status="failed", error=str(e))
 
 
 @app.post("/analyze/batch/async", response_model=JobStatus, tags=["Analysis"])
@@ -475,19 +545,14 @@ async def analyze_companies_async(
     _job_counter += 1
     job_id = f"job_{_job_counter}_{datetime.now().strftime('%Y%m%d%H%M%S')}"
 
-    _jobs[job_id] = JobStatus(
-        job_id=job_id,
-        status="pending",
-        progress=0,
-        total_companies=len(request.companies),
-        completed_companies=0,
-    )
+    db = _get_job_db()
+    job_row = db.create_job(job_id=job_id, total_companies=len(request.companies))
 
     background_tasks.add_task(
         _run_batch_job, job_id, request.companies, request.max_workers
     )
 
-    return _jobs[job_id]
+    return _job_row_to_status(job_row)
 
 
 @app.get("/jobs/{job_id}", response_model=JobStatus, tags=["Jobs"])
@@ -503,10 +568,13 @@ async def get_job_status(
     Returns:
         Current job status including progress and results when complete
     """
-    if job_id not in _jobs:
+    db = _get_job_db()
+    job_row = db.get_job(job_id)
+
+    if not job_row:
         raise HTTPException(status_code=404, detail=f"Job {job_id} not found")
 
-    return _jobs[job_id]
+    return _job_row_to_status(job_row)
 
 
 @app.get("/jobs", response_model=list[JobStatus], tags=["Jobs"])
@@ -527,12 +595,6 @@ async def list_jobs(
     Returns:
         List of job statuses
     """
-    jobs = list(_jobs.values())
-
-    if status:
-        jobs = [j for j in jobs if j.status == status]
-
-    # Return most recent first
-    jobs.sort(key=lambda j: j.job_id, reverse=True)
-
-    return jobs[:limit]
+    db = _get_job_db()
+    job_rows = db.list_jobs(status=status, limit=limit)
+    return [_job_row_to_status(row) for row in job_rows]

SPARC/auth.py

@@ -146,11 +146,36 @@ def decode_token(token: str) -> Optional[TokenPayload]:
         return None
 
 
+# Shared database client singleton, initialized at startup via init_db_client()
+_db_client: DatabaseClient | None = None
+
+
+def init_db_client() -> None:
+    """Initialize the shared database client. Call once at app startup."""
+    global _db_client
+    _db_client = DatabaseClient(config.database_url)
+    _db_client.connect()
+
+
+def close_db_client() -> None:
+    """Close the shared database client. Call at app shutdown."""
+    global _db_client
+    if _db_client:
+        _db_client.close()
+        _db_client = None
+
+
 def get_db_client() -> DatabaseClient:
-    """Get database client for auth operations."""
-    client = DatabaseClient(config.database_url)
-    client.connect()
-    return client
+    """Get the shared pooled database client for auth operations.
+
+    Returns the module-level singleton DatabaseClient. If not yet initialized
+    (e.g., during tests), creates a new instance as a fallback.
+    """
+    global _db_client
+    if _db_client is None:
+        _db_client = DatabaseClient(config.database_url)
+        _db_client.connect()
+    return _db_client
 
 
 async def get_current_user(

SPARC/database.py

@@ -171,6 +171,26 @@ class DatabaseClient:
                 ON serp_queries(query_hash)
             """)
 
+            # Create jobs table for persisting async batch job state
+            cursor.execute("""
+                CREATE TABLE IF NOT EXISTS jobs (
+                    job_id VARCHAR(128) PRIMARY KEY,
+                    status VARCHAR(20) NOT NULL DEFAULT 'pending',
+                    progress INTEGER NOT NULL DEFAULT 0,
+                    total_companies INTEGER NOT NULL DEFAULT 0,
+                    completed_companies INTEGER NOT NULL DEFAULT 0,
+                    result_json JSONB,
+                    error TEXT,
+                    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
+                    updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
+                )
+            """)
+
+            cursor.execute("""
+                CREATE INDEX IF NOT EXISTS idx_jobs_status
+                ON jobs(status)
+            """)
+
             self.conn.commit()
 
     @staticmethod
@@ -201,8 +221,6 @@ class DatabaseClient:
         Returns:
             Cached message dict if found, None otherwise
         """
-        self.connect()
-
         prompt_hash = self.hash_prompt(prompt)
 
         query = """
@@ -225,10 +243,11 @@ class DatabaseClient:
 
         query += " ORDER BY timestamp DESC LIMIT 1"
 
-        with self.conn.cursor(cursor_factory=RealDictCursor) as cursor:
-            cursor.execute(query, params)
-            result = cursor.fetchone()
-            return dict(result) if result else None
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute(query, params)
+                result = cursor.fetchone()
+                return dict(result) if result else None
 
     def store_message(
         self,
@@ -256,33 +275,32 @@ class DatabaseClient:
         Returns:
             The ID of the inserted record
         """
-        self.connect()
-
         prompt_hash = self.hash_prompt(prompt)
 
-        with self.conn.cursor() as cursor:
-            cursor.execute(
-                """
-                INSERT INTO llm_messages
-                (prompt, prompt_hash, response, company_name, analysis_type, model, metadata, token_usage, is_cached)
-                VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s)
-                RETURNING id
-                """,
-                (
-                    prompt,
-                    prompt_hash,
-                    response,
-                    company_name,
-                    analysis_type,
-                    model,
-                    json.dumps(metadata) if metadata else None,
-                    json.dumps(token_usage) if token_usage else None,
-                    is_cached,
-                ),
-            )
+        with self.get_conn() as conn:
+            with conn.cursor() as cursor:
+                cursor.execute(
+                    """
+                    INSERT INTO llm_messages
+                    (prompt, prompt_hash, response, company_name, analysis_type, model, metadata, token_usage, is_cached)
+                    VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s)
+                    RETURNING id
+                    """,
+                    (
+                        prompt,
+                        prompt_hash,
+                        response,
+                        company_name,
+                        analysis_type,
+                        model,
+                        json.dumps(metadata) if metadata else None,
+                        json.dumps(token_usage) if token_usage else None,
+                        is_cached,
+                    ),
+                )
 
-            message_id = cursor.fetchone()[0]
-            self.conn.commit()
+                message_id = cursor.fetchone()[0]
+            conn.commit()
 
             return message_id
 
@@ -304,8 +322,6 @@ class DatabaseClient:
         Returns:
             List of message dictionaries
         """
-        self.connect()
-
         query = "SELECT * FROM llm_messages WHERE 1=1"
         params = []
 
@@ -320,9 +336,10 @@ class DatabaseClient:
         query += " ORDER BY timestamp DESC LIMIT %s OFFSET %s"
         params.extend([limit, offset])
 
-        with self.conn.cursor(cursor_factory=RealDictCursor) as cursor:
-            cursor.execute(query, params)
-            return [dict(row) for row in cursor.fetchall()]
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute(query, params)
+                return [dict(row) for row in cursor.fetchall()]
 
     def get_analytics(self, days: int = 30) -> Dict:
         """Get analytics on message usage.
@@ -333,53 +350,52 @@ class DatabaseClient:
         Returns:
             Dictionary with analytics data
         """
-        self.connect()
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                # Total messages
+                cursor.execute(
+                    """
+                    SELECT COUNT(*) as total_messages
+                    FROM llm_messages
+                    WHERE timestamp >= NOW() - INTERVAL '%s days'
+                    """,
+                    (days,),
+                )
+                total = cursor.fetchone()["total_messages"]
 
-        with self.conn.cursor(cursor_factory=RealDictCursor) as cursor:
-            # Total messages
-            cursor.execute(
-                """
-                SELECT COUNT(*) as total_messages
-                FROM llm_messages
-                WHERE timestamp >= NOW() - INTERVAL '%s days'
-                """,
-                (days,),
-            )
-            total = cursor.fetchone()["total_messages"]
+                # Messages by company
+                cursor.execute(
+                    """
+                    SELECT company_name, COUNT(*) as count
+                    FROM llm_messages
+                    WHERE timestamp >= NOW() - INTERVAL '%s days'
+                    GROUP BY company_name
+                    ORDER BY count DESC
+                    LIMIT 10
+                    """,
+                    (days,),
+                )
+                by_company = cursor.fetchall()
 
-            # Messages by company
-            cursor.execute(
-                """
-                SELECT company_name, COUNT(*) as count
-                FROM llm_messages
-                WHERE timestamp >= NOW() - INTERVAL '%s days'
-                GROUP BY company_name
-                ORDER BY count DESC
-                LIMIT 10
-                """,
-                (days,),
-            )
-            by_company = cursor.fetchall()
+                # Messages by type
+                cursor.execute(
+                    """
+                    SELECT analysis_type, COUNT(*) as count
+                    FROM llm_messages
+                    WHERE timestamp >= NOW() - INTERVAL '%s days'
+                    GROUP BY analysis_type
+                    ORDER BY count DESC
+                    """,
+                    (days,),
+                )
+                by_type = cursor.fetchall()
 
-            # Messages by type
-            cursor.execute(
-                """
-                SELECT analysis_type, COUNT(*) as count
-                FROM llm_messages
-                WHERE timestamp >= NOW() - INTERVAL '%s days'
-                GROUP BY analysis_type
-                ORDER BY count DESC
-                """,
-                (days,),
-            )
-            by_type = cursor.fetchall()
-
-            return {
-                "total_messages": total,
-                "by_company": [dict(row) for row in by_company],
-                "by_type": [dict(row) for row in by_type],
-                "period_days": days,
-            }
+                return {
+                    "total_messages": total,
+                    "by_company": [dict(row) for row in by_company],
+                    "by_type": [dict(row) for row in by_type],
+                    "period_days": days,
+                }
 
     # Patent Cache Methods
 
@@ -462,6 +478,131 @@ class DatabaseClient:
                 )
             conn.commit()
 
+    # Job Persistence Methods
+
+    def create_job(
+        self,
+        job_id: str,
+        total_companies: int,
+    ) -> Dict:
+        """Create a new job record.
+
+        Args:
+            job_id: Unique job identifier
+            total_companies: Number of companies in the batch
+
+        Returns:
+            Job dict
+        """
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute(
+                    """
+                    INSERT INTO jobs (job_id, status, progress, total_companies, completed_companies)
+                    VALUES (%s, 'pending', 0, %s, 0)
+                    RETURNING *
+                    """,
+                    (job_id, total_companies),
+                )
+                job = cursor.fetchone()
+            conn.commit()
+            return dict(job)
+
+    def update_job(
+        self,
+        job_id: str,
+        status: Optional[str] = None,
+        progress: Optional[int] = None,
+        completed_companies: Optional[int] = None,
+        result_json: Optional[str] = None,
+        error: Optional[str] = None,
+    ) -> Optional[Dict]:
+        """Update a job's state.
+
+        Only non-None fields are updated.
+        """
+        updates = []
+        params = []
+        if status is not None:
+            updates.append("status = %s")
+            params.append(status)
+        if progress is not None:
+            updates.append("progress = %s")
+            params.append(progress)
+        if completed_companies is not None:
+            updates.append("completed_companies = %s")
+            params.append(completed_companies)
+        if result_json is not None:
+            updates.append("result_json = %s")
+            params.append(result_json)
+        if error is not None:
+            updates.append("error = %s")
+            params.append(error)
+
+        if not updates:
+            return self.get_job(job_id)
+
+        updates.append("updated_at = CURRENT_TIMESTAMP")
+        params.append(job_id)
+
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute(
+                    f"UPDATE jobs SET {', '.join(updates)} WHERE job_id = %s RETURNING *",
+                    params,
+                )
+                job = cursor.fetchone()
+            conn.commit()
+            return dict(job) if job else None
+
+    def get_job(self, job_id: str) -> Optional[Dict]:
+        """Get a job by ID."""
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute("SELECT * FROM jobs WHERE job_id = %s", (job_id,))
+                job = cursor.fetchone()
+                return dict(job) if job else None
+
+    def list_jobs(
+        self,
+        status: Optional[str] = None,
+        limit: int = 10,
+    ) -> List[Dict]:
+        """List jobs, optionally filtered by status."""
+        query = "SELECT * FROM jobs"
+        params: list = []
+        if status:
+            query += " WHERE status = %s"
+            params.append(status)
+        query += " ORDER BY created_at DESC LIMIT %s"
+        params.append(limit)
+
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute(query, params)
+                return [dict(row) for row in cursor.fetchall()]
+
+    def mark_stale_jobs_failed(self) -> int:
+        """Mark any jobs in 'running' or 'pending' state as 'failed'.
+
+        Called at startup to clean up jobs that were interrupted by a restart.
+
+        Returns:
+            Number of jobs marked as failed.
+        """
+        with self.get_conn() as conn:
+            with conn.cursor() as cursor:
+                cursor.execute(
+                    """
+                    UPDATE jobs SET status = 'failed', error = 'Interrupted by server restart',
+                    updated_at = CURRENT_TIMESTAMP
+                    WHERE status IN ('running', 'pending')
+                    """
+                )
+                count = cursor.rowcount
+            conn.commit()
+            return count
+
     # User Authentication Methods
 
     @staticmethod
@@ -505,25 +646,23 @@ class DatabaseClient:
         Returns:
             Created user dict or None if email exists
         """
-        self.connect()
-
         password_hash = self.hash_password(password)
 
         try:
-            with self.conn.cursor(cursor_factory=RealDictCursor) as cursor:
-                cursor.execute(
-                    """
-                    INSERT INTO users (email, password_hash, role)
-                    VALUES (%s, %s, %s)
-                    RETURNING id, email, role, created_at
-                    """,
-                    (email, password_hash, role),
-                )
-                user = cursor.fetchone()
-                self.conn.commit()
+            with self.get_conn() as conn:
+                with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                    cursor.execute(
+                        """
+                        INSERT INTO users (email, password_hash, role)
+                        VALUES (%s, %s, %s)
+                        RETURNING id, email, role, created_at
+                        """,
+                        (email, password_hash, role),
+                    )
+                    user = cursor.fetchone()
+                conn.commit()
                 return dict(user) if user else None
         except psycopg2.errors.UniqueViolation:
-            self.conn.rollback()
             return None
 
     def authenticate_user(self, email: str, password: str) -> Optional[Dict]:
@@ -536,23 +675,22 @@ class DatabaseClient:
         Returns:
             User dict if authenticated, None otherwise
         """
-        self.connect()
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute(
+                    "SELECT * FROM users WHERE email = %s",
+                    (email,),
+                )
+                user = cursor.fetchone()
 
-        with self.conn.cursor(cursor_factory=RealDictCursor) as cursor:
-            cursor.execute(
-                "SELECT * FROM users WHERE email = %s",
-                (email,),
-            )
-            user = cursor.fetchone()
-
-            if user and self.verify_password(password, user["password_hash"]):
-                return {
-                    "id": user["id"],
-                    "email": user["email"],
-                    "role": user["role"],
-                    "created_at": user["created_at"],
-                }
-            return None
+                if user and self.verify_password(password, user["password_hash"]):
+                    return {
+                        "id": user["id"],
+                        "email": user["email"],
+                        "role": user["role"],
+                        "created_at": user["created_at"],
+                    }
+                return None
 
     def get_user_by_id(self, user_id: int) -> Optional[Dict]:
         """Get a user by ID.
@@ -563,15 +701,14 @@ class DatabaseClient:
         Returns:
             User dict or None
         """
-        self.connect()
-
-        with self.conn.cursor(cursor_factory=RealDictCursor) as cursor:
-            cursor.execute(
-                "SELECT id, email, role, created_at FROM users WHERE id = %s",
-                (user_id,),
-            )
-            user = cursor.fetchone()
-            return dict(user) if user else None
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute(
+                    "SELECT id, email, role, created_at FROM users WHERE id = %s",
+                    (user_id,),
+                )
+                user = cursor.fetchone()
+                return dict(user) if user else None
 
     def get_user_by_email(self, email: str) -> Optional[Dict]:
         """Get a user by email.
@@ -582,15 +719,14 @@ class DatabaseClient:
         Returns:
             User dict or None
         """
-        self.connect()
-
-        with self.conn.cursor(cursor_factory=RealDictCursor) as cursor:
-            cursor.execute(
-                "SELECT id, email, role, created_at FROM users WHERE email = %s",
-                (email,),
-            )
-            user = cursor.fetchone()
-            return dict(user) if user else None
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute(
+                    "SELECT id, email, role, created_at FROM users WHERE email = %s",
+                    (email,),
+                )
+                user = cursor.fetchone()
+                return dict(user) if user else None
 
     def get_all_users(self, limit: int = 100, offset: int = 0) -> List[Dict]:
         """Get all users (admin only).
@@ -602,19 +738,18 @@ class DatabaseClient:
         Returns:
             List of user dicts
         """
-        self.connect()
-
-        with self.conn.cursor(cursor_factory=RealDictCursor) as cursor:
-            cursor.execute(
-                """
-                SELECT id, email, role, created_at
-                FROM users
-                ORDER BY created_at DESC
-                LIMIT %s OFFSET %s
-                """,
-                (limit, offset),
-            )
-            return [dict(row) for row in cursor.fetchall()]
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute(
+                    """
+                    SELECT id, email, role, created_at
+                    FROM users
+                    ORDER BY created_at DESC
+                    LIMIT %s OFFSET %s
+                    """,
+                    (limit, offset),
+                )
+                return [dict(row) for row in cursor.fetchall()]
 
     def update_user_role(self, user_id: int, role: str) -> Optional[Dict]:
         """Update a user's role (admin only).
@@ -626,20 +761,19 @@ class DatabaseClient:
         Returns:
             Updated user dict or None
         """
-        self.connect()
-
-        with self.conn.cursor(cursor_factory=RealDictCursor) as cursor:
-            cursor.execute(
-                """
-                UPDATE users
-                SET role = %s, updated_at = CURRENT_TIMESTAMP
-                WHERE id = %s
-                RETURNING id, email, role, created_at
-                """,
-                (role, user_id),
-            )
-            user = cursor.fetchone()
-            self.conn.commit()
+        with self.get_conn() as conn:
+            with conn.cursor(cursor_factory=RealDictCursor) as cursor:
+                cursor.execute(
+                    """
+                    UPDATE users
+                    SET role = %s, updated_at = CURRENT_TIMESTAMP
+                    WHERE id = %s
+                    RETURNING id, email, role, created_at
+                    """,
+                    (role, user_id),
+                )
+                user = cursor.fetchone()
+            conn.commit()
             return dict(user) if user else None
 
     def delete_user(self, user_id: int) -> bool:
@@ -651,12 +785,11 @@ class DatabaseClient:
         Returns:
             True if deleted
         """
-        self.connect()
-
-        with self.conn.cursor() as cursor:
-            cursor.execute("DELETE FROM users WHERE id = %s", (user_id,))
-            deleted = cursor.rowcount > 0
-            self.conn.commit()
+        with self.get_conn() as conn:
+            with conn.cursor() as cursor:
+                cursor.execute("DELETE FROM users WHERE id = %s", (user_id,))
+                deleted = cursor.rowcount > 0
+            conn.commit()
             return deleted
 
     def get_user_count(self) -> int:
@@ -665,8 +798,7 @@ class DatabaseClient:
         Returns:
             Number of users
         """
-        self.connect()
-
-        with self.conn.cursor() as cursor:
-            cursor.execute("SELECT COUNT(*) FROM users")
-            return cursor.fetchone()[0]
+        with self.get_conn() as conn:
+            with conn.cursor() as cursor:
+                cursor.execute("SELECT COUNT(*) FROM users")
+                return cursor.fetchone()[0]

requirements.txt

@@ -14,3 +14,4 @@ numpy
 pandas
 bcrypt
 PyJWT
+slowapi

scripts/init_database.py

@@ -40,6 +40,9 @@ def main():
         print("\nTables created:")
         print("  - llm_messages: Stores all LLM prompts and responses")
         print("  - users: Stores user accounts")
+        print("  - jobs: Stores async batch job state")
+        print("  - patents: Patent PDF cache")
+        print("  - serp_queries: SERP query result cache")
         print("\nIndexes created:")
         print("  - idx_messages_timestamp: For time-based queries")
         print("  - idx_messages_company: For company-specific queries")

tests/test_api.py

@@ -5,7 +5,7 @@ from datetime import datetime
 from unittest.mock import Mock, patch
 from fastapi.testclient import TestClient
 
-from SPARC.api import app, _analyzer, _jobs
+from SPARC.api import app
 from SPARC.types import CompanyAnalysisResult, BatchAnalysisResult
 
 

tests/test_auth.py

@@ -0,0 +1,302 @@
+"""Tests for JWT authentication flow: register, login, protected routes, refresh, admin access."""
+
+from datetime import datetime, timezone
+from unittest.mock import MagicMock, patch
+
+import pytest
+from fastapi.testclient import TestClient
+
+from SPARC.api import app
+from SPARC.auth import create_access_token, create_refresh_token
+
+
+@pytest.fixture
+def client():
+    """Create test client."""
+    return TestClient(app)
+
+
+@pytest.fixture(autouse=True)
+def mock_db(monkeypatch):
+    """Mock the database client used by auth endpoints.
+
+    Returns a MagicMock with all DB methods pre-configured.
+    """
+    db = MagicMock()
+
+    # Default: no users exist
+    db.get_user_count.return_value = 0
+    db.get_user_by_id.return_value = None
+    db.get_user_by_email.return_value = None
+    db.authenticate_user.return_value = None
+    db.create_user.return_value = None
+    db.get_all_users.return_value = []
+    db.update_user_role.return_value = None
+    db.delete_user.return_value = False
+
+    with patch("SPARC.api.get_db_client", return_value=db), \
+         patch("SPARC.auth.get_db_client", return_value=db):
+        yield db
+
+
+def _make_admin_user():
+    return {
+        "id": 1,
+        "email": "admin@test.com",
+        "role": "admin",
+        "created_at": datetime(2025, 1, 1, tzinfo=timezone.utc),
+    }
+
+
+def _make_regular_user():
+    return {
+        "id": 2,
+        "email": "user@test.com",
+        "role": "user",
+        "created_at": datetime(2025, 1, 1, tzinfo=timezone.utc),
+    }
+
+
+def _auth_header(user_dict):
+    """Create an Authorization header with a valid access token for the given user."""
+    token = create_access_token(user_dict["id"], user_dict["email"], user_dict["role"])
+    return {"Authorization": f"Bearer {token}"}
+
+
+class TestRegister:
+    """POST /auth/register"""
+
+    def test_register_first_user_becomes_admin(self, client, mock_db):
+        """First registered user should get admin role."""
+        mock_db.get_user_count.return_value = 0
+        mock_db.create_user.return_value = {
+            "id": 1,
+            "email": "admin@test.com",
+            "role": "admin",
+            "created_at": datetime(2025, 1, 1, tzinfo=timezone.utc),
+        }
+
+        response = client.post(
+            "/auth/register",
+            json={"email": "admin@test.com", "password": "securepass123"},
+        )
+
+        assert response.status_code == 200
+        data = response.json()
+        assert data["email"] == "admin@test.com"
+        assert data["role"] == "admin"
+        mock_db.create_user.assert_called_once_with(
+            email="admin@test.com", password="securepass123", role="admin"
+        )
+
+    def test_register_subsequent_user_gets_user_role(self, client, mock_db):
+        """Non-first user should get regular user role."""
+        mock_db.get_user_count.return_value = 1
+        mock_db.create_user.return_value = _make_regular_user()
+
+        response = client.post(
+            "/auth/register",
+            json={"email": "user@test.com", "password": "securepass123"},
+        )
+
+        assert response.status_code == 200
+        data = response.json()
+        assert data["role"] == "user"
+
+    def test_register_duplicate_email_returns_400(self, client, mock_db):
+        """Registering with an existing email should return 400."""
+        mock_db.get_user_count.return_value = 1
+        mock_db.create_user.return_value = None  # indicates duplicate
+
+        response = client.post(
+            "/auth/register",
+            json={"email": "existing@test.com", "password": "securepass123"},
+        )
+
+        assert response.status_code == 400
+        assert "already registered" in response.json()["detail"].lower()
+
+
+class TestLogin:
+    """POST /auth/login"""
+
+    def test_login_valid_credentials_returns_tokens(self, client, mock_db):
+        """Valid credentials should return access and refresh tokens."""
+        user = _make_regular_user()
+        mock_db.authenticate_user.return_value = user
+
+        response = client.post(
+            "/auth/login",
+            json={"email": "user@test.com", "password": "correctpassword"},
+        )
+
+        assert response.status_code == 200
+        data = response.json()
+        assert "access_token" in data
+        assert "refresh_token" in data
+        assert data["token_type"] == "bearer"
+
+    def test_login_invalid_credentials_returns_401(self, client, mock_db):
+        """Invalid credentials should return 401."""
+        mock_db.authenticate_user.return_value = None
+
+        response = client.post(
+            "/auth/login",
+            json={"email": "user@test.com", "password": "wrongpassword"},
+        )
+
+        assert response.status_code == 401
+        assert "invalid" in response.json()["detail"].lower()
+
+
+class TestGetMe:
+    """GET /auth/me"""
+
+    def test_valid_access_token_returns_user(self, client, mock_db):
+        """A valid access token should return the user's data."""
+        user = _make_regular_user()
+        mock_db.get_user_by_id.return_value = user
+
+        response = client.get("/auth/me", headers=_auth_header(user))
+
+        assert response.status_code == 200
+        data = response.json()
+        assert data["email"] == "user@test.com"
+        assert data["id"] == 2
+
+    def test_missing_token_returns_401(self, client):
+        """No token should return 401 (403 from HTTPBearer)."""
+        response = client.get("/auth/me")
+        assert response.status_code in (401, 403)
+
+    def test_expired_token_returns_401(self, client, mock_db):
+        """An expired token should return 401."""
+        # Create a token that has already expired
+        from datetime import timedelta
+
+        import jwt as pyjwt
+        from SPARC.auth import JWT_ALGORITHM, JWT_SECRET
+
+        payload = {
+            "sub": "1",
+            "email": "user@test.com",
+            "role": "user",
+            "exp": datetime.now(timezone.utc) - timedelta(hours=1),
+            "type": "access",
+        }
+        expired_token = pyjwt.encode(payload, JWT_SECRET, algorithm=JWT_ALGORITHM)
+
+        response = client.get(
+            "/auth/me", headers={"Authorization": f"Bearer {expired_token}"}
+        )
+        assert response.status_code == 401
+
+    def test_refresh_token_as_access_returns_401(self, client, mock_db):
+        """Using a refresh token as an access token should return 401."""
+        user = _make_regular_user()
+        refresh_token = create_refresh_token(user["id"], user["email"], user["role"])
+
+        response = client.get(
+            "/auth/me", headers={"Authorization": f"Bearer {refresh_token}"}
+        )
+        assert response.status_code == 401
+
+
+class TestRefreshToken:
+    """POST /auth/refresh"""
+
+    def test_valid_refresh_token_returns_new_tokens(self, client, mock_db):
+        """A valid refresh token should issue new access and refresh tokens."""
+        user = _make_regular_user()
+        mock_db.get_user_by_id.return_value = user
+        refresh = create_refresh_token(user["id"], user["email"], user["role"])
+
+        response = client.post(
+            "/auth/refresh", json={"refresh_token": refresh}
+        )
+
+        assert response.status_code == 200
+        data = response.json()
+        assert "access_token" in data
+        assert "refresh_token" in data
+
+    def test_invalid_refresh_token_returns_401(self, client, mock_db):
+        """An invalid refresh token should return 401."""
+        response = client.post(
+            "/auth/refresh", json={"refresh_token": "invalid-token-string"}
+        )
+        assert response.status_code == 401
+
+    def test_access_token_as_refresh_returns_401(self, client, mock_db):
+        """Using an access token as a refresh token should return 401."""
+        user = _make_regular_user()
+        access = create_access_token(user["id"], user["email"], user["role"])
+
+        response = client.post(
+            "/auth/refresh", json={"refresh_token": access}
+        )
+        assert response.status_code == 401
+
+
+class TestAdminUsers:
+    """GET /admin/users and PATCH /admin/users/{id}/role"""
+
+    def test_admin_can_list_users(self, client, mock_db):
+        """Admin token should allow listing users."""
+        admin = _make_admin_user()
+        mock_db.get_user_by_id.return_value = admin
+        mock_db.get_all_users.return_value = [admin, _make_regular_user()]
+
+        response = client.get("/admin/users", headers=_auth_header(admin))
+
+        assert response.status_code == 200
+        data = response.json()
+        assert len(data) == 2
+
+    def test_regular_user_cannot_list_users(self, client, mock_db):
+        """Regular user token should be rejected with 403."""
+        user = _make_regular_user()
+        mock_db.get_user_by_id.return_value = user
+
+        response = client.get("/admin/users", headers=_auth_header(user))
+
+        assert response.status_code == 403
+
+    def test_no_token_cannot_list_users(self, client):
+        """No token should be rejected."""
+        response = client.get("/admin/users")
+        assert response.status_code in (401, 403)
+
+    def test_admin_can_change_user_role(self, client, mock_db):
+        """Admin should be able to change another user's role."""
+        admin = _make_admin_user()
+        mock_db.get_user_by_id.return_value = admin
+        mock_db.update_user_role.return_value = {
+            "id": 2,
+            "email": "user@test.com",
+            "role": "admin",
+            "created_at": datetime(2025, 1, 1, tzinfo=timezone.utc),
+        }
+
+        response = client.patch(
+            "/admin/users/2/role",
+            json={"role": "admin"},
+            headers=_auth_header(admin),
+        )
+
+        assert response.status_code == 200
+        assert response.json()["role"] == "admin"
+
+    def test_admin_cannot_change_own_role(self, client, mock_db):
+        """Admin should not be able to change their own role."""
+        admin = _make_admin_user()
+        mock_db.get_user_by_id.return_value = admin
+
+        response = client.patch(
+            "/admin/users/1/role",
+            json={"role": "user"},
+            headers=_auth_header(admin),
+        )
+
+        assert response.status_code == 400
+        assert "own role" in response.json()["detail"].lower()

tests/test_rate_limit.py

@@ -0,0 +1,97 @@
+"""Tests for rate limiting on auth endpoints."""
+
+import pytest
+from unittest.mock import Mock, patch, MagicMock
+from fastapi.testclient import TestClient
+
+from SPARC.api import app
+
+
+@pytest.fixture
+def client():
+    """Create test client with rate limiter enabled."""
+    return TestClient(app)
+
+
+@pytest.fixture(autouse=True)
+def reset_limiter():
+    """Reset rate limiter storage between tests."""
+    from SPARC.api import limiter
+    limiter.reset()
+    yield
+
+
+class TestRateLimiting:
+    """Test rate limiting on login and register endpoints."""
+
+    @patch("SPARC.api.get_db_client")
+    def test_login_allows_requests_under_limit(self, mock_db_client, client):
+        """Login endpoint allows requests under the rate limit."""
+        mock_db = MagicMock()
+        mock_db.authenticate_user.return_value = None
+        mock_db_client.return_value = mock_db
+
+        # Should allow at least a few requests
+        for _ in range(5):
+            response = client.post(
+                "/auth/login",
+                json={"email": "test@example.com", "password": "password123"},
+            )
+            # 401 is expected (invalid credentials), not 429
+            assert response.status_code == 401
+
+    @patch("SPARC.api.get_db_client")
+    def test_login_rate_limited_after_threshold(self, mock_db_client, client):
+        """Login endpoint returns 429 after exceeding rate limit."""
+        mock_db = MagicMock()
+        mock_db.authenticate_user.return_value = None
+        mock_db_client.return_value = mock_db
+
+        # Send more than the limit (10/minute)
+        statuses = []
+        for _ in range(15):
+            response = client.post(
+                "/auth/login",
+                json={"email": "test@example.com", "password": "password123"},
+            )
+            statuses.append(response.status_code)
+
+        # At least one should be 429
+        assert 429 in statuses, f"Expected 429 in statuses but got: {set(statuses)}"
+
+    @patch("SPARC.api.get_db_client")
+    def test_register_rate_limited_after_threshold(self, mock_db_client, client):
+        """Register endpoint returns 429 after exceeding rate limit."""
+        mock_db = MagicMock()
+        mock_db.get_user_count.return_value = 1
+        mock_db.create_user.return_value = None  # triggers 400 (email exists)
+        mock_db_client.return_value = mock_db
+
+        # Send more than the limit (5/minute)
+        statuses = []
+        for _ in range(10):
+            response = client.post(
+                "/auth/register",
+                json={"email": "test@example.com", "password": "password123"},
+            )
+            statuses.append(response.status_code)
+
+        # At least one should be 429
+        assert 429 in statuses, f"Expected 429 in statuses but got: {set(statuses)}"
+
+    @patch("SPARC.api.get_db_client")
+    def test_rate_limit_returns_retry_after_header(self, mock_db_client, client):
+        """Rate limited responses include a Retry-After header."""
+        mock_db = MagicMock()
+        mock_db.authenticate_user.return_value = None
+        mock_db_client.return_value = mock_db
+
+        # Exhaust the limit
+        for _ in range(15):
+            response = client.post(
+                "/auth/login",
+                json={"email": "test@example.com", "password": "password123"},
+            )
+            if response.status_code == 429:
+                assert "Retry-After" in response.headers
+                break