Merge pull request #15 from wagesj45/codex/fix-operational-error-in-report-generation

Fix view generation for domain reports

scripts/generate_reports.py

@@ -36,6 +36,7 @@ def _get_domains() -> List[str]:
     conn.close()
     return domains
 
+
 def _load_config() -> List[Dict]:
     if not REPORT_CONFIG.exists():
         typer.echo(f"Config file not found: {REPORT_CONFIG}")
@@ -47,10 +48,12 @@ def _load_config() -> List[Dict]:
         raise typer.Exit(1)
     return data
 
+
 def _save_json(path: Path, data: List[Dict]) -> None:
     path.parent.mkdir(parents=True, exist_ok=True)
     path.write_text(json.dumps(data, indent=2))
 
+
 def _render_html(interval: str, reports: List[Dict], out_path: Path) -> None:
     env = Environment(loader=FileSystemLoader(TEMPLATE_DIR))
     template = env.get_template("report.html")
@@ -65,6 +68,7 @@ def _bucket_expr(interval: str) -> str:
         raise typer.Exit(1)
     return f"strftime('{fmt}', datetime(time))"
 
+
 def _generate_interval(interval: str, domain: Optional[str] = None) -> None:
     cfg = _load_config()
     if not cfg:
@@ -79,9 +83,12 @@ def _generate_interval(interval: str, domain: Optional[str] = None) -> None:
     # Create a temporary view so queries can easily be filtered by domain
     cur.execute("DROP VIEW IF EXISTS logs_view")
     if domain:
+        # Parameters are not allowed in CREATE VIEW statements, so we must
+        # safely interpolate the domain value ourselves. Escape any single
+        # quotes to prevent malformed queries.
+        safe_domain = domain.replace("'", "''")
         cur.execute(
-            "CREATE TEMP VIEW logs_view AS SELECT * FROM logs WHERE host = ?",
+            f"CREATE TEMP VIEW logs_view AS SELECT * FROM logs WHERE host = '{safe_domain}'"
-            (domain,),
         )
         out_dir = OUTPUT_DIR / domain / interval
     else:
@@ -101,12 +108,14 @@ def _generate_interval(interval: str, domain: Optional[str] = None) -> None:
         data = [dict(zip(headers, row)) for row in rows]
         json_path = out_dir / f"{name}.json"
         _save_json(json_path, data)
-        report_list.append({
+        report_list.append(
-            "name": name,
+            {
-            "label": definition.get("label", name.title()),
+                "name": name,
-            "chart": definition.get("chart", "line"),
+                "label": definition.get("label", name.title()),
-            "json": f"{name}.json",
+                "chart": definition.get("chart", "line"),
-        })
+                "json": f"{name}.json",
+            }
+        )
 
     _save_json(out_dir / "reports.json", report_list)
     _render_html(interval, report_list, out_dir / "index.html")
@@ -118,6 +127,7 @@ def _generate_all_domains(interval: str) -> None:
     for domain in _get_domains():
         _generate_interval(interval, domain)
 
+
 @app.command()
 def hourly(
     domain: Optional[str] = typer.Option(
@@ -133,6 +143,7 @@ def hourly(
     else:
         _generate_interval("hourly", domain)
 
+
 @app.command()
 def daily(
     domain: Optional[str] = typer.Option(
@@ -148,6 +159,7 @@ def daily(
     else:
         _generate_interval("daily", domain)
 
+
 @app.command()
 def weekly(
     domain: Optional[str] = typer.Option(
@@ -163,6 +175,7 @@ def weekly(
     else:
         _generate_interval("weekly", domain)
 
+
 @app.command()
 def monthly(
     domain: Optional[str] = typer.Option(
@@ -178,5 +191,6 @@ def monthly(
     else:
         _generate_interval("monthly", domain)
 
+
 if __name__ == "__main__":
     app()

tests/test_reports.py

@@ -32,11 +32,31 @@ def setup_db(path: Path):
     )
     cur.execute(
         "INSERT INTO logs (ip, host, time, request, status, bytes_sent, referer, user_agent, cache_status) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)",
-        ("127.0.0.1", "example.com", "2024-01-01 10:00:00", "GET / HTTP/1.1", 200, 100, "-", "curl", "MISS"),
+        (
+            "127.0.0.1",
+            "example.com",
+            "2024-01-01 10:00:00",
+            "GET / HTTP/1.1",
+            200,
+            100,
+            "-",
+            "curl",
+            "MISS",
+        ),
     )
     cur.execute(
         "INSERT INTO logs (ip, host, time, request, status, bytes_sent, referer, user_agent, cache_status) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)",
-        ("127.0.0.1", "example.com", "2024-01-01 10:05:00", "GET /err HTTP/1.1", 500, 100, "-", "curl", "MISS"),
+        (
+            "127.0.0.1",
+            "example.com",
+            "2024-01-01 10:05:00",
+            "GET /err HTTP/1.1",
+            500,
+            100,
+            "-",
+            "curl",
+            "MISS",
+        ),
     )
     conn.commit()
     conn.close()
@@ -72,14 +92,36 @@ def test_generate_interval(tmp_path, sample_reports, monkeypatch):
     monkeypatch.setattr(gr, "DB_PATH", db_path)
     monkeypatch.setattr(gr, "OUTPUT_DIR", tmp_path / "output")
     monkeypatch.setattr(gr, "REPORT_CONFIG", sample_reports)
-    monkeypatch.setattr(gr, "TEMPLATE_DIR", Path(__file__).resolve().parents[1] / "templates")
+    monkeypatch.setattr(
+        gr, "TEMPLATE_DIR", Path(__file__).resolve().parents[1] / "templates"
+    )
 
     gr._generate_interval("hourly")
 
     hits = json.loads((tmp_path / "output" / "hourly" / "hits.json").read_text())
     assert hits[0]["value"] == 2
-    error_rate = json.loads((tmp_path / "output" / "hourly" / "error_rate.json").read_text())
+    error_rate = json.loads(
+        (tmp_path / "output" / "hourly" / "error_rate.json").read_text()
+    )
     assert error_rate[0]["value"] == pytest.approx(50.0)
     reports = json.loads((tmp_path / "output" / "hourly" / "reports.json").read_text())
     assert {r["name"] for r in reports} == {"hits", "error_rate"}
 
+
+def test_generate_interval_domain_filter(tmp_path, sample_reports, monkeypatch):
+    db_path = tmp_path / "database" / "ngxstat.db"
+    setup_db(db_path)
+
+    monkeypatch.setattr(gr, "DB_PATH", db_path)
+    monkeypatch.setattr(gr, "OUTPUT_DIR", tmp_path / "output")
+    monkeypatch.setattr(gr, "REPORT_CONFIG", sample_reports)
+    monkeypatch.setattr(
+        gr, "TEMPLATE_DIR", Path(__file__).resolve().parents[1] / "templates"
+    )
+
+    gr._generate_interval("hourly", "example.com")
+
+    hits = json.loads(
+        (tmp_path / "output" / "example.com" / "hourly" / "hits.json").read_text()
+    )
+    assert hits[0]["value"] == 2