feat(tests): add --preprocess-dir CLI option to pytest

- conftest.py: registra --preprocess-dir via pytest_addoption
- test_preprocessors.py: usa pytest_generate_tests per leggere i casi
  a collection time con accesso a config; _content e _fixtures_dir
  accettano path dinamico

Usage: pytest tests/test_preprocessors.py --preprocess-dir /my/folder

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

tests/conftest.py

@@ -4,6 +4,14 @@ Provides an async SQLite in-memory engine that auto-creates all tables,
 a per-test session, and a FastAPI ``TestClient`` wired to use it.
 """
 
+
+def pytest_addoption(parser):
+    parser.addoption(
+        "--preprocess-dir",
+        default=None,
+        help="Override fixture folder for preprocessor tests (must contain cases.yaml + data/)",
+    )
+
 from __future__ import annotations
 
 import json

tests/test_preprocessors.py

@@ -1,9 +1,10 @@
 """Tests for the preprocessor system (Step 1 — Local Agent V2).
 
-Fixtures: tests/fixtures/preprocessors/cases.yaml + data/
-
 Run:
     pytest tests/test_preprocessors.py -v
+    pytest tests/test_preprocessors.py -v --preprocess-dir /path/to/folder
+
+The folder must contain cases.yaml + data/.
 """
 
 from __future__ import annotations
@@ -17,40 +18,56 @@ import yaml
 from app.core.langfuse_client import get_langfuse
 from app.core.preprocessors import detect_content_type, preprocess
 
-_DATA_DIR = Path(__file__).parent / "fixtures" / "preprocessors" / "data"
-_CASES_FILE = Path(__file__).parent / "fixtures" / "preprocessors" / "cases.yaml"
+_DEFAULT_DIR = Path(__file__).parent / "fixtures" / "preprocessors"
 
 _GENERATORS = {
     "binary_noise": "some\x00\x01\x02\x03\x04\x05content" * 20,
 }
 
 
-def _cases():
-    return yaml.safe_load(_CASES_FILE.read_text(encoding="utf-8"))
+def _fixtures_dir(config) -> Path:
+    override = config.getoption("--preprocess-dir")
+    return Path(override) if override else _DEFAULT_DIR
 
 
-def _content(case: dict) -> str:
+def _load_cases(config) -> list[dict]:
+    return yaml.safe_load((_fixtures_dir(config) / "cases.yaml").read_text(encoding="utf-8"))
+
+
+def _content(case: dict, data_dir: Path) -> str:
     if "generate" in case:
         return _GENERATORS[case["generate"]]
-    return (_DATA_DIR / case["file"]).read_text(encoding="utf-8")
+    return (data_dir / case["file"]).read_text(encoding="utf-8")
 
 
-def _lf_score(name: str, value: float, comment: str = "") -> None:
+def _lf_score(name: str, value: float) -> None:
     lf = get_langfuse()
     if lf:
         trace = lf.trace(name=f"eval-{name}")
-        lf.score(trace_id=trace.id, name=name, value=value, data_type="NUMERIC", comment=comment)
+        lf.score(trace_id=trace.id, name=name, value=value, data_type="NUMERIC")
         lf.flush()
 
 
+# ── parametrize at collection time via pytest hook ────────────────────
+
+def pytest_generate_tests(metafunc):
+    if "preprocess_case" not in metafunc.fixturenames:
+        return
+    cases = _load_cases(metafunc.config)
+    test_name = metafunc.function.__name__
+    if test_name == "test_detect":
+        subset = [c for c in cases if "detect" in c]
+    else:
+        subset = [c for c in cases if "process" in c]
+    metafunc.parametrize("preprocess_case", subset, ids=[c["id"] for c in subset])
+
+
 # ── detect ────────────────────────────────────────────────────────────
 
-_detect = [c for c in _cases() if "detect" in c]
-
-
-@pytest.mark.parametrize("case", _detect, ids=[c["id"] for c in _detect])
-def test_detect(case: dict) -> None:
-    raw = _content(case)
+def test_detect(preprocess_case, pytestconfig) -> None:
+    case = preprocess_case
+    data_dir = _fixtures_dir(pytestconfig) / "data"
+    raw = _content(case, data_dir)
     filename = case.get("filename", case.get("file", ""))
     ct = detect_content_type(filename, raw)
     expected = case["detect"]
@@ -60,12 +77,10 @@ def test_detect(case: dict) -> None:
 
 # ── preprocess ────────────────────────────────────────────────────────
 
-_process = [c for c in _cases() if "process" in c]
-
-
-@pytest.mark.parametrize("case", _process, ids=[c["id"] for c in _process])
-def test_preprocess(case: dict) -> None:
-    raw = _content(case)
+def test_preprocess(preprocess_case, pytestconfig) -> None:
+    case = preprocess_case
+    data_dir = _fixtures_dir(pytestconfig) / "data"
+    raw = _content(case, data_dir)
     result = preprocess(case["process"], raw)
 
     if case.get("no_html"):