chore: repo hygiene — fix connection leaks, clear all lint findings, drop stray root script

CHANGELOG.md

@@ -35,6 +35,13 @@ public API may still change between minor versions.
 
 ### Fixed
 
+- **A failed store open no longer leaks its SQLite connection.** `SQLiteConnection` opened the
+  underlying `sqlite3` connection before applying its setup PRAGMAs, and `SQLiteTraceStore`
+  opened the connection before creating its schema — so pointing either at a file that is not a
+  usable database raised the expected `DatabaseError` but left the already-open connection (and
+  its file descriptor) to the garbage collector. Both constructors now close the connection
+  before re-raising, so error paths (a typo'd `--db`, a corrupted golden baseline) clean up
+  after themselves.
 - **`RegressionRisk` no longer misses materially-changed or skipped critical steps.** The risk
   verdict was derived only from `removed`/`reordered` alignment states, but a critical step that
   was changed or skipped binds to a same-type event (type match alone clears the matcher

conformance/conformance_event.py

@@ -12,25 +12,18 @@
 
 from __future__ import annotations
 
-import os
-import sys
 import uuid
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from typing import Dict, List, Tuple
 
-# Make the package importable whether this module is loaded by the generator (run from
-# anywhere) or by pytest (which adds ``src`` via conftest, but we must not depend on that).
-_SRC = os.path.join(os.path.dirname(__file__), "..", "src")
-
-from dprovenancekit import (  # noqa: E402
-    AlignmentConfiguration,
+from dprovenancekit import (
     AnyEquivalenceEvaluator,
     TraceableEvent,
     TraceEvent,
     TracePriority,
     TraceRun,
 )
-from dprovenancekit.query import (  # noqa: E402
+from dprovenancekit.query import (
     AfterNode,
     AndNode,
     BeforeNode,

demo/demo_gif.py

@@ -87,7 +87,7 @@ def main() -> None:
     candidate = record(store, "research-agent · PR-42", CANDIDATE)
 
     def line(label, run, color):
-        parts, prev = [], None
+        parts = []
         for e in run.events:
             a = e.payload.type_identifier
             if parts and parts[-1][0] == a:
@@ -105,7 +105,7 @@ def line(label, run, color):
     out(f"{DIM}$ dprovenancekit gate --golden main --candidate PR-42{R}", 0.7)
     report = RegressionGate().check(golden, candidate)
     rules = [ToolDropRule("verify"), LoopingRule("search", max_repeats=3)]
-    anomalies = AnomalyDetector(store).detect_anomalies(rules)
+    AnomalyDetector(store).detect_anomalies(rules)
 
     verdict = "PASS" if report.passed else "REGRESSION"
     out(f"  {RED}{B}✗ {verdict}{R}  {GREY}severity {R}{RED}{report.regression_level.value.upper()}{R}"

dprovenancekit/sqlite_store.py

@@ -46,9 +46,15 @@ class SQLiteConnection:
     def __init__(self, path: str):
         self._lock = threading.RLock()
         self._db = sqlite3.connect(path, isolation_level=None, check_same_thread=False)
-        self.execute("PRAGMA journal_mode=WAL;")
-        self.execute("PRAGMA synchronous=NORMAL;")
-        self.execute("PRAGMA temp_store=MEMORY;")
+        try:
+            self.execute("PRAGMA journal_mode=WAL;")
+            self.execute("PRAGMA synchronous=NORMAL;")
+            self.execute("PRAGMA temp_store=MEMORY;")
+        except BaseException:
+            # The first PRAGMA is where a non-database file surfaces as DatabaseError;
+            # the connection is already open by then and would leak without this.
+            self._db.close()
+            raise
 
     def execute(self, sql: str, params=()) -> None:
         with self._lock:
@@ -340,16 +346,22 @@ def __init__(
     ):
         self._event_type = event_type
         self._db = SQLiteConnection(path)
-        self._buffer = TraceWriteBuffer(
-            max_global_buffer=max_global_buffer, max_per_run_buffer=max_per_run_buffer
-        )
-        self._drop_tally = TraceDropTally()
-        self._writer = SQLiteWriter(self._db, self._buffer, self._drop_tally)
+        try:
+            self._buffer = TraceWriteBuffer(
+                max_global_buffer=max_global_buffer, max_per_run_buffer=max_per_run_buffer
+            )
+            self._drop_tally = TraceDropTally()
+            self._writer = SQLiteWriter(self._db, self._buffer, self._drop_tally)
 
-        self._create_schema()
+            self._create_schema()
 
-        if start_writer:
-            self._writer.start()
+            if start_writer:
+                self._writer.start()
+        except BaseException:
+            # A failed __init__ never hands the store to the caller, so nobody else can
+            # close the connection we just opened.
+            self._db.close()
+            raise
 
     def _create_schema(self) -> None:
         db = self._db

examples/basic_agent_trace.py

@@ -1,4 +1,3 @@
-import uuid
 from dprovenancekit import (
     DProvenanceKit,
     SQLiteTraceStore,

examples/pr_gate_demo/agent.py

@@ -1,6 +1,4 @@
 import argparse
-import sys
-from pathlib import Path
 from langchain_core.runnables import RunnableLambda
 from dprovenancekit import SQLiteTraceStore
 from dprovenancekit.integrations.langchain import DProvenanceTracer, LangChainTraceEvent

examples/regression_testing.py

@@ -32,7 +32,6 @@
 import sys
 import uuid
 from dataclasses import dataclass
-from typing import Optional
 
 # Make the package importable when run straight from a checkout (no install needed).
 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))

tests/conftest.py

@@ -6,8 +6,6 @@
 
 from __future__ import annotations
 
-import os
-import sys
 import uuid
 from dataclasses import dataclass
 

tests/integrations/test_fastapi.py

@@ -4,7 +4,6 @@
 from dprovenancekit.kit import DProvenanceKit
 from dprovenancekit.store import InMemoryTraceStore
 from dprovenancekit.event import TraceableEvent
-from dataclasses import dataclass
 
 pytest.importorskip("fastapi")
 pytest.importorskip("httpx")

tests/integrations/test_jupyter.py

@@ -2,7 +2,6 @@
 from dprovenancekit.kit import DProvenanceKit
 from dprovenancekit.store import InMemoryTraceStore
 from dprovenancekit.event import TraceableEvent
-from dataclasses import dataclass
 
 pytest.importorskip("IPython")
 

tests/integrations/test_mcp.py

@@ -2,7 +2,6 @@
 from dprovenancekit.kit import DProvenanceKit
 from dprovenancekit.store import InMemoryTraceStore
 from dprovenancekit.event import TraceableEvent
-from dataclasses import dataclass
 import asyncio
 
 from dprovenancekit.integrations.mcp import traced_mcp_tool

tests/test_cli_export.py

@@ -35,8 +35,8 @@ def trace_db(tmp_path):
 def test_export_emits_valid_ordered_jsonl(trace_db, capsys):
     db, run_id = trace_db
     assert main(["export", "--db", db, "--run", str(run_id)]) == 0
-    lines = [l for l in capsys.readouterr().out.splitlines() if l.strip()]
-    rows = [json.loads(l) for l in lines]  # every line must be valid JSON
+    lines = [ln for ln in capsys.readouterr().out.splitlines() if ln.strip()]
+    rows = [json.loads(ln) for ln in lines]  # every line must be valid JSON
     assert [r["type"] for r in rows] == ["retrieved", "decided"]
     assert [r["sequence"] for r in rows] == [0, 1]
     assert all(r["run_id"] == str(run_id) for r in rows)

tests/test_conformance.py

@@ -90,6 +90,7 @@ def test_run_fingerprint(desc, case):
         rows = store._db.query(
             "SELECT fingerprint FROM runs WHERE run_id = ?", (str(run.run_id),)
         )
+        store.close()
     assert rows[0][0] == case["fingerprint"]
 
 
@@ -131,6 +132,7 @@ def test_query_semantics(backend, desc, case):
 
         dsl = dsl_from_wire_dsl(case["dsl"])
         matched = sorted(r.context_id for r in store.query_runs(dsl))
+        store.close()
     assert matched == case["expected_context_ids"]
 
 

tests/test_facade.py

@@ -78,10 +78,10 @@ def test_facade_jsonl_save_is_one_event_per_line(tmp_path):
     path = tmp_path / "run.jsonl"
     trace.save(path)
 
-    lines = [json.loads(l) for l in path.read_text().splitlines() if l.strip()]
+    lines = [json.loads(ln) for ln in path.read_text().splitlines() if ln.strip()]
     # Three spans, a .start/.end pair each.
     assert len(lines) == 6
-    types = [l["type"] for l in lines]
+    types = [ln["type"] for ln in lines]
     assert "Agent Workflow.start" in types
     assert "Verify Claims.end" in types
     # Every event carries the envelope fields the loader needs.

tests/test_instrument.py

@@ -419,6 +419,7 @@ def b():
         fp1 = _fingerprint_after(store, r1.run_id)
         fp2 = _fingerprint_after(store, r2.run_id)
         fp3 = _fingerprint_after(store, r3.run_id)
+        store.close()
 
     assert fp1 == fp2
     assert fp1 != fp3

tests/test_integration_langchain.py

@@ -389,6 +389,7 @@ def test_identical_paths_share_a_fingerprint_divergent_paths_do_not():
         fp_a = _fingerprint_after(store, fa)
         fp_b = _fingerprint_after(store, fb)
         fp_c = _fingerprint_after(store, fc)
+        store.close()
 
     assert fp_a == fp_b  # same path → same fingerprint
     assert fp_a != fp_c  # reordered path → different fingerprint

tests/test_integration_openai_agents.py

@@ -289,6 +289,7 @@ def test_force_flush_writes_open_trace():
         rows = store._db.query(
             "SELECT event_count FROM runs WHERE run_id = ?", (str(run_id),)
         )
+        store.close()
     assert rows and rows[0][0] == 1
 
 
@@ -356,6 +357,7 @@ def test_same_path_shares_fingerprint_divergent_path_differs():
             _fingerprint_after(store, b),
             _fingerprint_after(store, c),
         )
+        store.close()
 
     assert fp_a == fp_b
     assert fp_a != fp_c

tests/test_otel_ingest.py

@@ -9,7 +9,6 @@
 
 from dprovenancekit.diff import ChangeKind, TraceDiffEngine
 from dprovenancekit.otel_ingest import (
-    IngestedRun,
     OTelSpanEvent,
     ingest_otlp,
     run_id_for_trace,

tests/test_public_api.py

@@ -9,7 +9,6 @@
 
 from __future__ import annotations
 
-import importlib
 
 import dprovenancekit
 

tests/test_raw_store_roundtrip.py

@@ -17,9 +17,11 @@ def test_written_run_survives_raw_trace_store_reopen(temp_db_path):
         kit.record(TestEvent.error_detected())
         kit.record(TestEvent.process_finished())
     store.flush()
+    store.close()
 
     reader = RawTraceStore(temp_db_path)
     runs = reader.fetch_all_runs()
+    reader.close()
 
     assert len(runs) == 1
     run = runs[0]

tests/test_sqlite_get_run.py

@@ -41,11 +41,14 @@ def test_get_run_returns_run_by_id(tmp_path):
         kit.record(_E("b"))
 
     got = store.get_run(run.run_id)  # flushes, then indexed fetch
+    store.close()
     assert got is not None
     assert got.context_id == "c"
     assert [e.payload.type_identifier for e in got.events] == ["a", "b"]
 
 
 def test_get_run_missing_is_none(tmp_path):
     store = SQLiteTraceStore(_E, str(tmp_path / "t.sqlite"), start_writer=False)
-    assert store.get_run(uuid.uuid4()) is None
+    missing = store.get_run(uuid.uuid4())
+    store.close()
+    assert missing is None

tests/test_sqlite_insert_failure_drop.py

@@ -72,6 +72,7 @@ def test_failed_batch_insert_is_tallied_and_breaks_integrity(temp_db_path):
     assert not stats.preserved_integrity
 
     assert _count(db, "SELECT COUNT(*) FROM runs;") == 0
+    db.close()
 
 
 def test_successful_insert_tallies_nothing_and_records_accurate_metadata(temp_db_path):
@@ -94,3 +95,4 @@ def test_successful_insert_tallies_nothing_and_records_accurate_metadata(temp_db
 
     assert _count(db, "SELECT COUNT(*) FROM trace_events;") == 3
     assert _count(db, "SELECT event_count FROM runs WHERE run_id = 'run-1';") == 3
+    db.close()

tests/test_visualizer.py

@@ -1,6 +1,5 @@
 import uuid
-import pytest
-from dprovenancekit import TraceGraph, TraceEdge, TraceEdgeType, InMemoryTraceStore
+from dprovenancekit import TraceGraph, TraceEdge, TraceEdgeType
 from dprovenancekit.visualizer import render_trace_html
 from conftest import TestEvent