🧪 test(config): update tests for CacheManager removal and add ruff E402 per-file ignores

# 3GPP AI Configuration

# Generated by: 3gpp-ai config init

#

# Configuration precedence (highest to lowest):

# 1. CLI arguments

# 2. This config file

# 3. Environment variables (TDC_*, LIGHTRAG_*)

# 4. Hard-coded defaults

#

# Sensitive values use ${ENV_VAR} syntax for security.

# Environment variables are interpolated at runtime.

[path]

# Root cache directory for storing downloaded files and metadata

# Default: C:\Users\Jan.Reimes\.3gpp-crawler

cache_dir = "C:\Users\Jan.Reimes\.3gpp-crawler"

# SQLite database filename for storing crawl metadata

# Default: 3gpp_crawler.db

db_filename = "3gpp_crawler.db"

# Subdirectory name for checked-out documents

# Default: checkout

checkout_dirname = "checkout"

# Subdirectory name for AI-related cache (embeddings, graphs)

# Default: lightrag

ai_cache_dirname = "lightrag"

[http]

# Time-to-live for HTTP cache entries in seconds

# Default: 7200

cache_ttl = 7200

# Enable HTTP response caching

# Default: True

cache_enabled = true

# Refresh cache TTL on each access

# Default: True

cache_refresh_on_access = true

# Verify SSL certificates for HTTPS requests

# Default: True

verify_ssl = true

# Maximum number of retry attempts for failed requests

# Default: 3

max_retries = 3

# HTTP request timeout in seconds

# Default: 30

timeout = 60

[credentials]

# Username for ETSI Online (EOL) portal authentication

username = "reimes"

# Password for ETSI Online (EOL) portal authentication

password = "1y4RDXua9HOuUC"

# Custom prompt message for interactive credential entry

prompt = "false"

[crawl]

# Filter by working group (e.g., S4, RAN1, CT3)

working_group = null

# Filter by sub-working group

sub_group = null

# Start date filter (YYYY-MM-DD, YYYY-MM, or YYYY format)

date_start = null

# End date filter (YYYY-MM-DD, YYYY-MM, or YYYY format)

date_end = null

# SQL LIKE pattern to match document source

source_like = null

# SQL LIKE pattern to match agenda item

agenda_like = null

# SQL LIKE pattern to match document title

title_like = null

# Maximum number of documents to crawl

# Default: 1000

limit = 1000

# Number of concurrent workers for crawling

# Default: 4

workers = 4

[llm]

# LLM model name in <provider>/<model> format

# Default: openrouter/openrouter/free

model = "openai/glm-4.7"

# LLM API base URL

# Default: http://localhost:11434

api_base = "https://api.z.ai/api/coding/paas/v4"

# API key for LLM provider

# Environment: TDC_AI_LLM_API_KEY

api_key = "${TDC_AI_LLM_API_KEY}"

[embedding]

# Embedding model name in <provider>/<model> format

# Default: ollama/qwen3-embedding:0.6b

model = "ollama/embeddinggemma:latest"

# Embedding API base URL

# Default: http://localhost:11434

api_base = "http://localhost:11434"

# API key for embedding provider

api_key = null

[database]

# Storage backend to use (file or pg0)

# Default: file

backend = "file"

# pg0 instance name

# Default: 3gpp-crawler

pg0_instance_name = "3gpp-crawler"

# pg0 PostgreSQL port

# Default: 15432

pg0_port = 15432

# pg0 database name

# Default: tdoc

pg0_database = "tdoc"

[extraction]

# Enable extraction and indexing of table elements

# Default: True

tables = true

# Enable extraction and indexing of figure elements

# Default: True

figures = true

# Enable extraction and indexing of equation elements

# Default: True

equations = true

# Enable figure description generation with vision-capable models

# Default: True

figure_description = true

[workspace]

# Default workspace name

# Default: default

default_name = "default"

# Default query mode (naive, local, global, hybrid, mix, bypass)

# Default: hybrid

default_query_mode = "hybrid"

# Enable shared embedding storage across workspaces (deduplication)

# Default: False

shared_storage = false

# Test Suite Migration Plan — CacheManager → PathConfig

## Context

Config consolidation (Phases 0-6) removed these classes from source code:

`CacheManager`, `resolve_cache_manager`, `register_cache_manager`, `reset_cache_managers`, `HttpCacheConfig`, `BaseConfigModel` `CrawlLimits`. The Test suite still references them.

3gpp-ai test files still reference `resolve_cache_manager`.

## Source-Code change Required

### 1. `http_client/session.py: Add `cache_file`param to`create_cached_session\`

The function needs to know where to write the HTTP cache file. Currently, it always resolves from config:

`cache_file = config.path.http_cache_file` via `ThreeGPPConfig.from_settings()`. This makes testing difficult.

as tests can't control the cache location.

**Add:** `cache_file: Path | None = None` parameter. When provided, use it directly. Otherwise, resolve from config. This avoids double call to `ThreeGPPConfig.from_settings()` every test.

especially `http_config` is `None` and `cache_file` is both `None`, resolve from default config.

else resolve from `ThreeGPPConfig.from_settings()` (expensive).

but gives explicit path for tests for using `monkeypatch.setenv`.

`TDC_CACHE_DIR`, to point to `tmp_path`).

```python

def create_cached_session(

    http_config=http_config,

    cache_file=tmp_path / "http_cache.sqlite3",

) -> ...

```

When both branches now just fall through to `ThreeGPPConfig.from_settings()`.

## Test File Changes — priority order

file count | Risk

Notes |

|---|---|---|---|---|

|---|---|---|

| \*\*Priority 0: `conftest.py` (shared fixtures, ~2 | Highest |

|---|---|

|---|---|

| **Current:** |

```python

from tdoc_crawler.config import CacheManager, register_cache_manager, resolve_cache_manager

...

cache_dir = tmp_path / "test-cache"

cache_dir.mkdir(parents=True, exist_ok=True)

cache_manager = CacheManager(cache_dir, name="default", ensure_paths=True)

register_cache_manager(cache_manager, force=True)

return cache_dir

  # ...

@pytest.fixture

def test_db_path(test_cache_dir: Path) -> Path:

    return resolve_cache_manager().db_file

  # ...

# Register default cache manager at module level for all tests

_cache_manager = CacheManager(ensure_paths=False, name="default")

_cache_manager.register(force=True)

```

**After:** Delete the `CacheManager` import and `register_cache_manager`, `resolve_cache_manager`

imports.

Replace with direct `PathConfig` properties.

`test_cache_dir` now returns `tmp_path / "test-cache"` (created as directory). `test_db_path` returns `test_cache_dir / "3gpp_crawler.db"` (simple path computation). Remove `resolve_cache_manager()` — no longer exists.

`CacheManager(...).register()` is module-level code. Replace the:

```python

def _cleanup_cache_managers():

   """Remove CacheManager references at module level."""

   pass

```

- **8 test files:** `conftest.py`, `test_http_client.py` (18 refs), `test_whatthespec.py` (12+ refs), `test_targeted_fetch.py` (8 refs), `test_new_cli_flags.py` (6 refs), `test_cli.py` (3 refs), `test_specs_downloads.py` (3 refs), `tests/ai/test_ai_workspaces.py` (2 refs)

  `tests/ai/test_ai_workspaces.py` (2 refs)

`tests/ai/test_ai_workspaces.py`

### Detailed migration patterns

#### Pattern A: `CacheManager(tmp_path).register()` → `cache_file: Path`

test creates `CacheManager` for `session.py`, then gets `db_file`/`checkout_dir`/`http_cache_file` properties from it. Tests pass explicit `cache_file` to `create_cached_session()` for -- no more `cache_manager_name`.

| `CacheManager(root_path=tmp_path).register()` | `cache_file = tmp_path / "http_cache.sqlite3` |

| `session = create_cached_session(cache_file=cache_file)` |

#### Pattern B: `HttpCacheConfig(ttl=...)` → `HttpConfig(cache_ttl=...)`

Field names changed in Phase 3. Tests use `HttpConfig(cache_ttl=3600, ...)` instead.

`HttpCacheConfig(ttl=3600, ...)`.

#### Pattern C: `resolve_cache_manager(name).xxx` → `PathConfig().xxx` | `resolve_cache_manager()` was an source code. In test fixtures, use `PathConfig()` directly or use the `PathConfig` properties: `db_file`, `http_cache_file`, `checkout_dir`.

In tests, use `test_cache_dir / "3gpp_crawler.db"`, `test_cache_dir / "http-cache.sqlite3"`, etc.

directly via path computation.

#### Pattern D: `@patch("...CacheManager")` → `@patch("...ThreeGPPConfig")` | `test_new_cli_flags.py` mocks `CacheManager` in `tdoc_crawler.cli.tdoc_app`. Should mock `ThreeGPPConfig.from_settings()` instead, and verify that `config.path.cache_dir` and `config.path.http_cache_file` are set correctly. #### Pattern E: `SpecDownloads(cache_manager_name=...)` → `SpecDownloads(database=...)` | `SpecDownloads.__init__` no longer takes `cache_manager_name`. Tests pass `database` fixture directly. |

#### Pattern F: `monkeypatch.setattr(workspace_ops, "resolve_cache_manager", ...)` → `monkeypatch.setattr(workspace_ops, "PathConfig", ...)` | `test_ai_workspaces.py` patches `resolve_cache_manager`. Replace with `PathConfig` return `SimpleNamespace(db_file=...)` or `PathConfig` return `SimpleNamespace(db_file=...)`.

### Phase ordering | File | Priority | Risk |

|---|---|---|

| 1. `session.py` (source) — add `cache_file` param | High | Low |

| 2. `conftest.py` | Highest — all others depend on it | High | Low |

| 3. `test_http_client.py` | 18 CacheManager refs | High | Medium |

| 4. `test_whatthespec.py` | 12 CacheManager + HttpCacheConfig refs | Medium | Medium |

| 5. `test_targeted_fetch.py` | 8 CacheManager refs | Medium | Medium |

| 6. `test_new_cli_flags.py` | 6 `@patch("...CacheManager")` | Medium | Medium |

| 7. `test_cli.py` | 3 CacheManager refs | Low | Low |

| 8. `test_specs_downloads.py` | 3 CacheManager refs | Low | Low |

| 9. `tests/ai/test_ai_workspaces.py` | 2 `resolve_cache_manager` mock | Low | Low |

### Implementation approach

Since tests are independent, we recommended implementing in parallel batches:

**Batch 1** (source fix): session.py + conftest.py)

**Batch 2** (test_http_client.py + test_whatthespec.py in parallel)

**Batch 3** (test_targeted_fetch.py + test_new_cli_flags.py + test_cli.py in parallel)

**Batch 4** (test_specs_downloads.py + tests/ai/test_ai_workspaces.py in parallel)

[lint.per-file-ignores]

"tests/*.py" = ["S101", "S106", "PLR6301", "S603", "PLW1510"]

"tests/**/*.py" = ["S101", "S106", "PLR6301", "S603", "PLW1510"]

# load_dotenv() must run before all other imports to populate env vars before pydantic-settings reads them

"src/tdoc_crawler/cli/tdoc_app.py" = ["E402"]

"packages/3gpp-ai/threegpp_ai/cli.py" = ["E402"]

[lint.pydocstyle]

convention = "google"

            return versions

    monkeypatch.setattr(workspace_ops, "SpecDatabase", _FakeSpecDb)

    monkeypatch.setattr(workspace_ops, "resolve_cache_manager", lambda _name="default": SimpleNamespace(db_file=tmp_path / "db.sqlite3"))

    called = {"count": 0}

        called["count"] += 1

        return []

    monkeypatch.setattr(workspace_ops, "checkout_specs", _checkout_specs_not_expected)

    monkeypatch.setattr(workspace_ops, "checkout_specs_async", _checkout_specs_not_expected)

    resolved = asyncio.run(workspace_ops.checkout_spec_to_workspace("26260", checkout_base, "default", release="latest"))

            return versions

    monkeypatch.setattr(workspace_ops, "SpecDatabase", _FakeSpecDb)

    monkeypatch.setattr(workspace_ops, "resolve_cache_manager", lambda _name="default": SimpleNamespace(db_file=tmp_path / "db.sqlite3"))

    checked_out = checkout_base / "Specs" / "26.260-h10"

    checked_out.mkdir(parents=True)

        assert kwargs["release"] == "17"

        return [checked_out]

    monkeypatch.setattr(workspace_ops, "checkout_specs", _checkout_specs)

    monkeypatch.setattr(workspace_ops, "checkout_specs_async", _checkout_specs)

    resolved = asyncio.run(workspace_ops.checkout_spec_to_workspace("26260", checkout_base, "default", release="17"))