Externalize dictionaries and add anonymization review corpus

tests/unit/test_config_externalization.py

@@ -0,0 +1,92 @@
+#!/usr/bin/env python3
+"""
+Tests de non-régression pour la config externalisée.
+"""
+from pathlib import Path
+
+import anonymizer_core_refactored_onnx as core
+from config_defaults import (
+    deep_merge_dict,
+    ensure_runtime_dictionaries_config,
+    load_effective_dictionaries_dict,
+    read_default_dictionaries_text,
+    read_runtime_dictionaries_overlay_text,
+)
+
+
+def test_default_config_template_is_externalized():
+    text = read_default_dictionaries_text()
+
+    assert "blacklist:" in text
+    assert "whitelist_phrases:" in text
+
+    cfg = core.load_dictionaries(None)
+    assert "CHCB" in cfg["blacklist"]["force_mask_terms"]
+
+
+def test_runtime_overlay_template_is_minimal():
+    text = read_runtime_dictionaries_overlay_text()
+
+    assert "dictionnaires.default.yml" in text
+    assert "{}" in text
+
+
+def test_deep_merge_dict_preserves_nested_defaults():
+    base = {
+        "whitelist": {
+            "sections_titres": ["DIM"],
+            "org_gpe_keep": False,
+        },
+        "flags": {
+            "case_insensitive": True,
+            "regex_engine": "python",
+        },
+    }
+    override = {
+        "whitelist": {
+            "sections_titres": ["GHM"],
+            "org_gpe_keep": True,
+        },
+        "flags": {
+            "regex_engine": "re2",
+        },
+    }
+
+    merged = deep_merge_dict(base, override)
+
+    assert merged["whitelist"]["sections_titres"] == ["DIM", "GHM"]
+    assert merged["whitelist"]["org_gpe_keep"] is True
+    assert merged["flags"]["case_insensitive"] is True
+    assert merged["flags"]["regex_engine"] == "re2"
+
+
+def test_additional_stopwords_refresh_and_reset(tmp_path: Path):
+    cfg_path = tmp_path / "cfg.yml"
+    cfg_path.write_text("additional_stopwords:\n  - xyzzymed\n", encoding="utf-8")
+
+    core.load_dictionaries(cfg_path)
+    assert "xyzzymed" in core._MEDICAL_STOP_WORDS_SET
+    assert "xyzzymed" in core._MEDICAL_STOP_WORDS
+
+    core.load_dictionaries(None)
+    assert "xyzzymed" not in core._MEDICAL_STOP_WORDS_SET
+    assert "xyzzymed" not in core._MEDICAL_STOP_WORDS
+
+
+def test_runtime_overlay_is_created_and_effective_merge_works(tmp_path: Path):
+    cfg_path = tmp_path / "dictionnaires.yml"
+
+    created = ensure_runtime_dictionaries_config(cfg_path)
+    assert created == cfg_path
+    assert cfg_path.exists()
+
+    effective = load_effective_dictionaries_dict(cfg_path)
+    assert "CHCB" in effective["blacklist"]["force_mask_terms"]
+
+    cfg_path.write_text(
+        "blacklist:\n  force_mask_terms:\n    - LOCAL_SIGLE\n",
+        encoding="utf-8",
+    )
+    effective = load_effective_dictionaries_dict(cfg_path)
+    assert "CHCB" in effective["blacklist"]["force_mask_terms"]
+    assert "LOCAL_SIGLE" in effective["blacklist"]["force_mask_terms"]