claude-dev-env 2.15.1 → 2.17.0

package/hooks/blocking/test_plain_language_blocker_allowlist.py

@@ -1,228 +0,0 @@
-"""Tests for the per-project domain-vocabulary allowlist in plain_language_blocker."""
-
-import importlib.util
-import json
-import os
-from pathlib import Path
-
-import pytest
-from types import ModuleType
-from unittest.mock import patch
-
-BLOCKER_PATH = Path(__file__).parent / "plain_language_blocker.py"
-ALLOWLIST_RELATIVE_PATH = Path(".claude") / "plain-language-allow.json"
-ALWAYS_HEAVY_WORD = "utilize"
-_PROSE_STYLE_ENFORCEMENT_ENVIRONMENT = {
-    "CLAUDE_PROSE_STYLE_ENFORCEMENT": "1",
-}
-
-
-def _load_blocker() -> ModuleType:
-    module_spec = importlib.util.spec_from_file_location(
-        "plain_language_blocker_allowlist_under_test", BLOCKER_PATH
-    )
-    assert module_spec is not None and module_spec.loader is not None
-    loaded_module = importlib.util.module_from_spec(module_spec)
-    module_spec.loader.exec_module(loaded_module)
-    return loaded_module
-
-
-_BLOCKER = _load_blocker()
-
-@pytest.fixture(autouse=True)
-def enable_prose_style_enforcement(monkeypatch: pytest.MonkeyPatch) -> None:
-    """Opinionated heavy-word scan is default-off; these tests arm it."""
-    monkeypatch.setenv("CLAUDE_PROSE_STYLE_ENFORCEMENT", "1")
-
-
-
-def _evaluate_advisory(
-    payload_by_key: dict[str, object],
-) -> tuple[str | None, str | None]:
-    """Return (deny_reason, advisory_message) with prose enforcement on.
-
-    Heavy words never deny (OP-07D); allowlist coverage asserts on the advisory
-    message and on ``find_banned_terms`` so the scanner still sees domain terms.
-    """
-    with patch.dict(os.environ, _PROSE_STYLE_ENFORCEMENT_ENVIRONMENT):
-        return _BLOCKER.evaluate_with_advisory(payload_by_key)
-
-
-def _init_repo(root: Path) -> Path:
-    (root / ".git").mkdir(parents=True, exist_ok=True)
-    return root
-
-
-def _write_allowlist(project_root: Path, all_words: list[str]) -> None:
-    allowlist_path = project_root / ALLOWLIST_RELATIVE_PATH
-    allowlist_path.parent.mkdir(parents=True, exist_ok=True)
-    allowlist_path.write_text(json.dumps(all_words), encoding="utf-8")
-
-
-def _write_raw_allowlist(project_root: Path, raw_text: str) -> None:
-    allowlist_path = project_root / ALLOWLIST_RELATIVE_PATH
-    allowlist_path.parent.mkdir(parents=True, exist_ok=True)
-    allowlist_path.write_text(raw_text, encoding="utf-8")
-
-
-def _markdown_write_payload(project_root: Path, prose: str) -> dict[str, object]:
-    document_path = project_root / "docs" / "notes.md"
-    return {
-        "tool_name": "Write",
-        "cwd": str(project_root),
-        "tool_input": {"file_path": str(document_path), "content": prose},
-    }
-
-
-def _ask_user_question_payload(project_root: Path, prose: str) -> dict[str, object]:
-    return {
-        "tool_name": "AskUserQuestion",
-        "cwd": str(project_root),
-        "tool_input": {"questions": [{"question": prose, "options": []}]},
-    }
-
-
-def test_allowlisted_word_passes_in_markdown_write(tmp_path: Path) -> None:
-    prose = "Please submit the release notes."
-    project_without_allowlist = _init_repo(tmp_path / "control")
-    project_with_allowlist = _init_repo(tmp_path / "domain")
-    _write_allowlist(project_with_allowlist, ["submit"])
-
-    control_deny, control_advisory = _evaluate_advisory(
-        _markdown_write_payload(project_without_allowlist, prose)
-    )
-    allowlisted_deny, allowlisted_advisory = _evaluate_advisory(
-        _markdown_write_payload(project_with_allowlist, prose)
-    )
-
-    assert control_deny is None
-    assert control_advisory is not None and "submit" in control_advisory
-    assert allowlisted_deny is None
-    assert allowlisted_advisory is None
-
-
-def test_allowlisted_word_passes_in_ask_user_question(tmp_path: Path) -> None:
-    prose = "Which theme should we identify first?"
-    project_without_allowlist = _init_repo(tmp_path / "control")
-    project_with_allowlist = _init_repo(tmp_path / "domain")
-    _write_allowlist(project_with_allowlist, ["identify"])
-
-    control_deny, control_advisory = _evaluate_advisory(
-        _ask_user_question_payload(project_without_allowlist, prose)
-    )
-    allowlisted_deny, allowlisted_advisory = _evaluate_advisory(
-        _ask_user_question_payload(project_with_allowlist, prose)
-    )
-
-    assert control_deny is None
-    assert control_advisory is not None and "identify" in control_advisory
-    assert allowlisted_deny is None
-    assert allowlisted_advisory is None
-
-
-def test_non_allowlisted_heavy_word_still_flagged(tmp_path: Path) -> None:
-    _init_repo(tmp_path)
-    _write_allowlist(tmp_path, ["submit"])
-
-    deny_reason, advisory_message = _evaluate_advisory(
-        _markdown_write_payload(tmp_path, f"Please {ALWAYS_HEAVY_WORD} the cache now.")
-    )
-
-    assert deny_reason is None
-    assert advisory_message is not None and ALWAYS_HEAVY_WORD in advisory_message
-
-
-def test_allowlist_match_is_case_insensitive(tmp_path: Path) -> None:
-    prose = "Please SUBMIT the release notes."
-    project_without_allowlist = _init_repo(tmp_path / "control")
-    project_with_allowlist = _init_repo(tmp_path / "domain")
-    _write_allowlist(project_with_allowlist, ["Submit"])
-
-    control_deny, control_advisory = _evaluate_advisory(
-        _markdown_write_payload(project_without_allowlist, prose)
-    )
-    allowlisted_deny, allowlisted_advisory = _evaluate_advisory(
-        _markdown_write_payload(project_with_allowlist, prose)
-    )
-
-    assert control_deny is None
-    assert control_advisory is not None and "submit" in control_advisory
-    assert allowlisted_deny is None
-    assert allowlisted_advisory is None
-
-
-def test_malformed_allowlist_json_is_ignored(tmp_path: Path) -> None:
-    _init_repo(tmp_path)
-    _write_raw_allowlist(tmp_path, "{ this is not valid json ")
-
-    deny_reason, advisory_message = _evaluate_advisory(
-        _markdown_write_payload(tmp_path, "Please submit the release notes.")
-    )
-
-    assert deny_reason is None
-    assert advisory_message is not None and "submit" in advisory_message
-
-
-def test_allowlist_in_a_different_project_root_is_not_applied(tmp_path: Path) -> None:
-    project_with_allowlist = _init_repo(tmp_path / "project_a")
-    project_without_allowlist = _init_repo(tmp_path / "project_b")
-    _write_allowlist(project_with_allowlist, ["submit"])
-
-    deny_reason, advisory_message = _evaluate_advisory(
-        _markdown_write_payload(project_without_allowlist, "Please submit the notes.")
-    )
-
-    assert deny_reason is None
-    assert advisory_message is not None and "submit" in advisory_message
-
-
-def test_allowlist_above_the_repo_root_is_not_applied(tmp_path: Path) -> None:
-    _write_allowlist(tmp_path, ["submit"])
-    repository_root = _init_repo(tmp_path / "repo")
-
-    deny_reason, advisory_message = _evaluate_advisory(
-        _markdown_write_payload(repository_root, "Please submit the notes.")
-    )
-
-    assert deny_reason is None
-    assert advisory_message is not None and "submit" in advisory_message
-
-
-def test_allowlist_at_the_repo_root_is_applied(tmp_path: Path) -> None:
-    repository_root = _init_repo(tmp_path / "repo")
-    _write_allowlist(repository_root, ["submit"])
-
-    deny_reason, advisory_message = _evaluate_advisory(
-        _markdown_write_payload(repository_root, "Please submit the notes.")
-    )
-
-    assert deny_reason is None
-    assert advisory_message is None
-
-
-def test_allowlist_without_a_repo_root_is_not_applied(tmp_path: Path) -> None:
-    """An allowlist only applies when the walk finds a ``.git`` repository root.
-
-    The upward walk can hit a personal home ``.git`` on some developer machines,
-    so this case stubs the finder to the no-repository outcome rather than
-    relying on the ambient filesystem.
-    """
-    _write_allowlist(tmp_path, ["submit"])
-
-    with patch.object(_BLOCKER, "_find_project_allowlist_file", return_value=None):
-        deny_reason, advisory_message = _evaluate_advisory(
-            _markdown_write_payload(tmp_path, "Please submit the notes.")
-        )
-
-    assert deny_reason is None
-    assert advisory_message is not None and "submit" in advisory_message
-
-
-def test_find_banned_terms_skips_allowlisted_terms() -> None:
-    matched_without_allowlist = _BLOCKER.find_banned_terms("Please submit the notes.")
-    matched_with_allowlist = _BLOCKER.find_banned_terms(
-        "Please submit the notes.", frozenset({"submit"})
-    )
-
-    assert any(each_term == "submit" for each_term, _ in matched_without_allowlist)
-    assert all(each_term != "submit" for each_term, _ in matched_with_allowlist)

package/hooks/diagnostic/AGENTS.md

@@ -1,75 +0,0 @@
-# hooks/diagnostic
-
-Hooks and scripts that collect, store, and query hook-firing records. The pipeline reads session JSONL transcripts, extracts hook attachment records, and writes them as rows into a Neon (Postgres) `hook_events` table.
-
-## Subdirectories
-
-| Directory | Role |
-|---|---|
-| `migrations/` | SQL migration files for the `hook_events` schema |
-| `queries/` | Parameterized SQL queries for inspecting blocked commands |
-
-## Key files
-
-| File | What it does |
-|---|---|
-| `hook_log_init.py` | One-time setup: creates the Neon schema (runs `schema.sql`), then verifies read-write parity with a sentinel round-trip |
-| `hook_log_extractor.py` | Disabled CLI: exits 0 without reading transcripts or rewriting offset state; body remains for a re-enable path that ingests `hook_*` attachments into Neon |
-| `hook_log_stop_wrapper.py` | Disabled Stop wrapper stub: exits 0; not on the Stop dispatcher roster |
-| `schema.sql` | DDL for the `hook_events` table, `blocked_commands` view, and supporting indexes |
-| `requirements-hook-logs.txt` | Runtime dependencies (`psycopg`) for the extractor |
-| `requirements-hook-logs-dev.txt` | Dev/test dependencies |
-| `test_hook_log_extractor.py` | Tests for the extractor |
-| `test_hook_log_init.py` | Tests for the schema-init script |
-| `test_hook_log_stop_wrapper.py` | Tests for the Stop wrapper |
-
-## Schema overview (`schema.sql`)
-
-The `hook_events` table captures one row per hook firing:
-
-- `hook_event`, `hook_name`, `hook_category` — what fired
-- `outcome` — `allowed`, `blocked`, or `ask`
-- `tool_name`, `command_excerpt` — what tool was called
-- `session_id`, `git_branch`, `cwd` — context
-- `duration_ms`, `exit_code` — timing and result
-- `source_jsonl_path`, `source_line_number` — idempotency key
-
-The `blocked_commands` view filters to `outcome = 'blocked'`.
-
-## Conventions
-
-- Extractor and Stop wrapper mains are disabled and exit 0 with no work.
-- Constants for the extractor (table name, offset state file, timeout) live in `hooks_constants/hook_log_extractor_constants.py`.
-- Tests run with `python -m pytest diagnostic/test_hook_log_*.py`.
-
-## Schema init
-
-Run `hook_log_init.py` once per machine, or after rotating the Neon project.
-
-Prerequisites: Bitwarden Secrets Manager CLI (`bws`) on PATH; a machine-account
-token in `BWS_ACCESS_TOKEN` for the user environment (`setx` on Windows, shell
-profile on macOS/Linux); Neon connection string stored as
-`NEON_HOOK_LOGS_DATABASE_URL`; Python deps from `requirements-hook-logs.txt`.
-
-```
-bws run -- python packages/claude-dev-env/hooks/diagnostic/hook_log_init.py
-```
-
-`bws run` strips `BWS_ACCESS_TOKEN` from the child environment so the Python
-process never sees it. The script verifies `NEON_HOOK_LOGS_DATABASE_URL`,
-connects with a 5-second timeout, applies `schema.sql` with idempotent DDL,
-runs a sentinel insert/select/delete round-trip, and prints the Neon host,
-table name, and row count.
-
-## Operator CLI flags
-
-`hook_log_extractor.py` and `hook_log_stop_wrapper.py` mains exit 0 with no work.
-The extractor body still documents these flags for a re-enable path:
-
-- default / `--incremental`: resume from `~/.claude/logs/hooks/.state/offsets.json`
-- `--full-rebuild`: clear offsets, truncate `hook_events`, re-read every JSONL
-- `--summary`: print the top-10 blockers of the last 24 hours
-- `--query <name>`: run `queries/<name>.sql` (`top_blockers_overall`,
-  `top_blockers_last_24_hours`, `blocks_last_7_days`, `blocks_by_category`,
-  `blocks_by_tool`, `block_details_for_hook`)
-

package/hooks/diagnostic/CLAUDE.md

@@ -1 +0,0 @@
-@AGENTS.md