test(classifier): ✅ add tests that assert that fixtures must exist with their associated existing pattern

a log line cannot match multiple patterns. Again a fixture can only exist in codebase if its associated pattern also exists.

tests/test_classifier.py

@@ -11,19 +11,28 @@ ORIGINAL_LOGS_DIR = FIXTURE_PATH / "original_logs"
 
 
 def _pattern_name_to_filename(name: str) -> str:
-    """STORE_CODE_NOT_FOUND → CR_store_code_not_found.txt"""
+    """
+        Returns the fixture corresponding to the error code name
+    """
     return f"CR_{name.lower()}.txt"
 
 def _read_fixture(path: Path) -> str:
-    """Lit un fichier de fixture avec fallback Windows-1252."""
+    """
+        Read a file with fallback Windows-1252.
+    """
     try:
-        return path.read_text(encoding="utf-8")
+        contenu = path.read_text(encoding="utf-8")
     except UnicodeDecodeError:
         return path.read_text(encoding="windows-1252")
+    if "\ufffd" not in contenu:
+        return contenu
+    return path.read_text(encoding="windows-1252")
 
 @pytest.mark.parametrize("name, pattern", N2_PATTERNS.items())
 def test_pattern_matches_fixture(name, pattern):
-    """Chaque pattern N2 matche au moins une ligne de son fichier de fixture."""
+    """
+        Assert each N2 pattern must match at least one line in its fixture.
+    """
 
     fixture = ORIGINAL_LOGS_DIR / _pattern_name_to_filename(name)
     if not fixture.exists():
@@ -49,4 +58,31 @@ def test_pattern_matches_fixture(name, pattern):
         f"{name} : no match in file {fixture.name}\n"
         f"pattern: {pattern}"
     )
-  
+  
+
+def test_patterns_do_not_overlap():
+    """
+        Assert that a log line can match only one N2 pattern.
+    """
+    fixtures_dir = FIXTURE_PATH
+    for fixture in fixtures_dir.glob("CR_*.txt"):
+        line_list = [line for line in _read_fixture(fixture).splitlines() if line.strip()]
+        for line in line_list:
+            matching = [
+                name for name, raw in N2_PATTERNS.items()
+                if re.compile(raw).search(line)
+            ]
+            assert len(matching) <= 1, (
+                f"Chevauchement détecté sur {fixture.name} : "
+                f"{matching} match line : {line[:80]}"
+            )
+
+def test_no_orphan_fixtures():
+    """
+        Assert that a fixture exist only if its associated pattern exists.
+        Aucun fichier de fixture ne doit exister sans pattern associé.
+    """
+    expected = {f"CR_{name.lower()}.txt" for name in N2_PATTERNS}
+    actual = {f.name for f in FIXTURE_PATH.glob("CR_*.txt")}
+    orphan = actual - expected
+    assert not orphan, f"Fixtures orphelines (sans pattern) : {sorted(orphan)}"