From 77d2f0f5c68a928b227c2218aa93bd72e1b0e3c9 Mon Sep 17 00:00:00 2001
From: Antoine <antoinevdp@live.fr>
Date: Thu, 28 May 2026 17:05:17 +0200
Subject: [PATCH] fix: tighten french cv heuristics

Reject au sein prose continuations and recognize dotted MSc/BSc degree variants in education parsing.
---
 src/job_research/profile/cv_extractor.py |  5 +++-
 tests/profile/test_cv_extractor.py       | 36 ++++++++++++++++++++++++
 2 files changed, 40 insertions(+), 1 deletion(-)

diff --git a/src/job_research/profile/cv_extractor.py b/src/job_research/profile/cv_extractor.py
index 089ce65..9a0ec88 100644
--- a/src/job_research/profile/cv_extractor.py
+++ b/src/job_research/profile/cv_extractor.py
@@ -223,7 +223,7 @@ def _parse_education_entry(line: str) -> dict[str, str] | None:
 
 
 def _looks_like_education_credential(text: str) -> bool:
-    lowered = text.lower()
+    lowered = text.lower().replace(".", "")
     return any(
         re.search(rf"\b{re.escape(keyword)}\b", lowered)
         for keyword in EDUCATION_TITLE_KEYWORDS
@@ -236,6 +236,9 @@ def _looks_like_institution(text: str) -> bool:
 
 
 def _parse_experience_entry(line: str) -> dict[str, str] | None:
+    if " au sein " in line.lower():
+        return None
+
     for connector in EXPERIENCE_LINE_CONNECTORS:
         if line.count(connector) != 1:
             continue
diff --git a/tests/profile/test_cv_extractor.py b/tests/profile/test_cv_extractor.py
index 279e4d1..c88cb6d 100644
--- a/tests/profile/test_cv_extractor.py
+++ b/tests/profile/test_cv_extractor.py
@@ -125,6 +125,20 @@ def test_extract_cv_signals_parses_french_experience_connectors() -> None:
     ]
 
 
+def test_extract_cv_signals_ignores_french_prose_continuations() -> None:
+    text = dedent(
+        """
+        Tonio
+        Location: France
+        Ingénieur au sein de BNP Paribas
+        """
+    ).strip()
+
+    extracted = extract_cv_signals(text)
+
+    assert extracted["experience_entries"] == []
+
+
 def test_extract_cv_signals_ignores_in_paris_prose_tail() -> None:
     text = dedent(
         """
@@ -189,6 +203,28 @@ def test_extract_cv_signals_extracts_common_french_education_entries() -> None:
     ]
 
 
+def test_extract_cv_signals_extracts_dotted_degree_variants() -> None:
+    text = dedent(
+        """
+        Tonio
+        Location: France
+        Education:
+        M.Sc. in Data Engineering at EPITA
+        B.Sc. in Computer Science at University X
+        """
+    ).strip()
+
+    extracted = extract_cv_signals(text)
+
+    assert extracted["education_entries"] == [
+        {"credential": "M.Sc. in Data Engineering", "institution": "EPITA"},
+        {
+            "credential": "B.Sc. in Computer Science",
+            "institution": "University X",
+        },
+    ]
+
+
 def test_extract_cv_signals_ignores_before_moving_prose_tail() -> None:
     text = dedent(
         """