fix(anonymizer): cover CHCB real-world staff layouts

tests/unit/test_real_world_identifier_layouts.py

@@ -107,3 +107,132 @@ def test_ogc_pdf_redaction_does_not_mask_numeric_substrings(tmp_path):
     redacted.close()
     assert "07C141" in text
     assert "142 : La facturation" in text
+
+
+def test_crop_epi_header_name_is_masked():
+    cfg = load_dictionaries(None)
+    text = (
+        "CROp Epi -  NOMTEST, Jean-Michel\n"
+        "Compte rendu opératoire\n"
+    )
+
+    anon = anonymise_document_regex([text], [[]], cfg)
+
+    assert "NOMTEST" not in anon.text_out
+    assert "Jean-Michel" not in anon.text_out
+    assert "CROp Epi -  [NOM]" in anon.text_out
+    assert any(
+        h.kind == "NOM_FORCE" and "NOMTEST" in h.original
+        for h in anon.audit
+    )
+
+
+def test_crop_epi_header_name_with_spaced_suffix_is_masked():
+    cfg = load_dictionaries(None)
+    text = (
+        "CROp Epi -  NOMTEST, Marie     NOMSUFFIX\n"
+        "Compte rendu opératoire\n"
+    )
+
+    anon = anonymise_document_regex([text], [[]], cfg)
+
+    assert "NOMTEST" not in anon.text_out
+    assert "Marie" not in anon.text_out
+    assert "NOMSUFFIX" not in anon.text_out
+    assert "CROp Epi -  [NOM]" in anon.text_out
+    assert any(
+        h.kind == "NOM_FORCE" and "NOMSUFFIX" in h.original
+        for h in anon.audit
+    )
+
+
+def test_standalone_compound_signature_name_is_masked():
+    cfg = load_dictionaries(None)
+    text = (
+        "Observation clinique stable.\n"
+        "Alix-Pierre Nomtest\n"
+        "Suite de la prise en charge.\n"
+    )
+
+    anon = anonymise_document_regex([text], [[]], cfg)
+
+    assert "Alix-Pierre" not in anon.text_out
+    assert "Nomtest" not in anon.text_out
+    assert "[NOM]" in anon.text_out
+    assert any(
+        h.kind == "NOM_FORCE" and "Alix-Pierre Nomtest" in h.original
+        for h in anon.audit
+    )
+
+
+def test_modified_by_application_banner_name_is_masked():
+    cfg = load_dictionaries(None)
+    text = (
+        ">>>CRO type 10/04/23 14 : 19   "
+        "(mod. le 13/04/23 15:58 par NOMTEST Fanny, statut : complet)\n"
+    )
+
+    anon = anonymise_document_regex([text], [[]], cfg)
+
+    assert "NOMTEST" not in anon.text_out
+    assert "Fanny" not in anon.text_out
+    assert "par [NOM], statut" in anon.text_out
+    assert any(
+        h.kind == "NOM_FORCE" and "NOMTEST Fanny" in h.original
+        for h in anon.audit
+    )
+
+
+def test_ref_initials_are_page_local_for_pdf_redaction():
+    cfg = load_dictionaries(None)
+    text = "Réf : RBG/FL\n"
+
+    anon = anonymise_document_regex([text], [[]], cfg)
+
+    assert "Réf : [NOM]/[NOM]" in anon.text_out
+    assert any(h.kind == "NOM_INITIAL" and h.original == "RBG" and h.page == 0 for h in anon.audit)
+    assert any(h.kind == "NOM_INITIAL" and h.original == "FL" and h.page == 0 for h in anon.audit)
+
+
+def test_ref_initials_with_suffix_label_are_masked():
+    cfg = load_dictionaries(None)
+    text = "Réf_CRO : EG/PB\nCOMPTE-RENDU OPÉRATOIRE\n"
+
+    anon = anonymise_document_regex([text], [[]], cfg)
+
+    assert "EG/PB" not in anon.text_out
+    assert "Réf_CRO : [NOM]/[NOM]" in anon.text_out
+    assert any(h.kind == "NOM_INITIAL" and h.original == "EG" and h.page == 0 for h in anon.audit)
+    assert any(h.kind == "NOM_INITIAL" and h.original == "PB" and h.page == 0 for h in anon.audit)
+
+
+def test_staff_role_with_interne_name_is_masked():
+    cfg = load_dictionaries(None)
+    text = "Aide : l'interne Charles NOMTEST\nAnesthésiste : Docteur [NOM]\n"
+
+    anon = anonymise_document_regex([text], [[]], cfg)
+
+    assert "Charles" not in anon.text_out
+    assert "NOMTEST" not in anon.text_out
+    assert "Aide : l'interne [NOM]" in anon.text_out
+    assert any(
+        h.kind == "NOM_FORCE" and "Charles NOMTEST" in h.original
+        for h in anon.audit
+    )
+
+
+def test_trackare_iao_multiline_staff_name_is_masked():
+    cfg = load_dictionaries(None)
+    text = (
+        "Heure d'orientation\n"
+        "18 : 48\n"
+        "IAO\n"
+        "NOMTEST Marlène\n"
+        "Priorité\n"
+    )
+
+    anon = anonymise_document_regex([text], [[]], cfg)
+
+    assert "NOMTEST" not in anon.text_out
+    assert "Marlène" not in anon.text_out
+    assert any(h.kind in {"NOM", "NOM_FORCE", "NOM_GLOBAL"} for h in anon.audit)