fix: parse cv bullets and education entries

Normalize leading bullet markers before parsing and extract simple education entries from heading-led or keyword-plus-institution lines.

src/job_research/profile/cv_extractor.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import re
 from pathlib import Path
 
 from pypdf import PdfReader
@@ -50,6 +51,40 @@ EXPERIENCE_PROSE_MARKERS = {
     "working",
 }
 
+EDUCATION_TITLE_KEYWORDS = {
+    "bachelor",
+    "degree",
+    "diploma",
+    "engineering school",
+    "licence",
+    "master",
+}
+
+EDUCATION_INSTITUTION_MARKERS = {
+    "academy",
+    "college",
+    "centrale",
+    "ecole",
+    "école",
+    "ens",
+    "epita",
+    "epitech",
+    "hec",
+    "imt",
+    "institute",
+    "institut",
+    "insa",
+    "polytech",
+    "school",
+    "university",
+    "universite",
+    "université",
+}
+
+EDUCATION_ENTRY_SEPARATORS = (" at ", ", ", " - ", " – ", " — ")
+
+LEADING_BULLET_MARKERS = {"-", "*", "+", "•", "‣", "∙", "●", "◦"}
+
 
 def extract_pdf_text(path: Path) -> str:
     reader = PdfReader(str(path))
@@ -62,7 +97,7 @@ def extract_pdf_text(path: Path) -> str:
 
 
 def extract_cv_signals(text: str) -> dict[str, object]:
-    lines = [line.strip() for line in text.splitlines()]
+    lines = [_normalize_line(line) for line in text.splitlines()]
     non_empty_lines = [line for line in lines if line]
 
     name = non_empty_lines[0] if non_empty_lines else None
@@ -70,9 +105,28 @@ def extract_cv_signals(text: str) -> dict[str, object]:
     languages: list[str] = []
     skills: list[str] = []
     experience_entries: list[dict[str, str]] = []
+    education_entries: list[dict[str, str]] = []
+    in_education_section = False
 
     for line in non_empty_lines[1:]:
         lowered = line.lower()
+
+        if lowered.startswith("education:"):
+            in_education_section = True
+            remainder = line.split(":", 1)[1].strip()
+            if remainder:
+                education_entry = _parse_education_entry(remainder)
+                if education_entry:
+                    education_entries.append(education_entry)
+            continue
+
+        if in_education_section:
+            education_entry = _parse_education_entry(line)
+            if education_entry:
+                education_entries.append(education_entry)
+                continue
+            in_education_section = False
+
         if lowered.startswith("location:"):
             location = line.split(":", 1)[1].strip() or None
             continue
@@ -82,6 +136,10 @@ def extract_cv_signals(text: str) -> dict[str, object]:
         if lowered.startswith("skills:"):
             skills = _parse_csv_field(line)
             continue
+        education_entry = _parse_education_entry(line)
+        if education_entry:
+            education_entries.append(education_entry)
+            continue
         if _looks_like_experience_line(line):
             title, company = line.split(" at ", 1)
             experience_entries.append(
@@ -97,7 +155,7 @@ def extract_cv_signals(text: str) -> dict[str, object]:
         "languages": languages,
         "skills": skills,
         "experience_entries": experience_entries,
-        "education_entries": [],
+        "education_entries": education_entries,
     }
 
 
@@ -106,6 +164,48 @@ def _parse_csv_field(line: str) -> list[str]:
     return [item.strip() for item in value.split(",") if item.strip()]
 
 
+def _normalize_line(line: str) -> str:
+    stripped = line.strip()
+    if not stripped:
+        return ""
+
+    if stripped[0] in LEADING_BULLET_MARKERS:
+        return stripped[1:].lstrip()
+
+    return stripped
+
+
+def _parse_education_entry(line: str) -> dict[str, str] | None:
+    for separator in EDUCATION_ENTRY_SEPARATORS:
+        if separator not in line:
+            continue
+
+        left, right = (part.strip() for part in line.split(separator, 1))
+        if not left or not right:
+            continue
+
+        if _looks_like_education_credential(left) and _looks_like_institution(right):
+            return {"credential": left, "institution": right}
+
+        if _looks_like_education_credential(right) and _looks_like_institution(left):
+            return {"credential": right, "institution": left}
+
+    return None
+
+
+def _looks_like_education_credential(text: str) -> bool:
+    lowered = text.lower()
+    return any(
+        re.search(rf"\b{re.escape(keyword)}\b", lowered)
+        for keyword in EDUCATION_TITLE_KEYWORDS
+    )
+
+
+def _looks_like_institution(text: str) -> bool:
+    lowered = text.lower()
+    return any(marker in lowered for marker in EDUCATION_INSTITUTION_MARKERS)
+
+
 def _looks_like_experience_line(line: str) -> bool:
     if line.count(" at ") != 1:
         return False

tests/profile/test_cv_extractor.py

@@ -73,6 +73,23 @@ def test_extract_cv_signals_ignores_prose_after_company_name() -> None:
     assert extracted["experience_entries"] == []
 
 
+def test_extract_cv_signals_normalizes_bullet_prefixed_fields_and_experience() -> None:
+    text = dedent(
+        """
+        Tonio
+        - Location: France
+        - Data Engineer at Company A
+        """
+    ).strip()
+
+    extracted = extract_cv_signals(text)
+
+    assert extracted["location"] == "France"
+    assert extracted["experience_entries"] == [
+        {"title": "Data Engineer", "company": "Company A"}
+    ]
+
+
 def test_extract_cv_signals_ignores_in_paris_prose_tail() -> None:
     text = dedent(
         """
@@ -87,6 +104,26 @@ def test_extract_cv_signals_ignores_in_paris_prose_tail() -> None:
     assert extracted["experience_entries"] == []
 
 
+def test_extract_cv_signals_extracts_education_entries_after_heading() -> None:
+    text = dedent(
+        """
+        Tonio
+        Location: France
+        Education:
+        - Master of Science in Data Science at University of Paris
+        """
+    ).strip()
+
+    extracted = extract_cv_signals(text)
+
+    assert extracted["education_entries"] == [
+        {
+            "credential": "Master of Science in Data Science",
+            "institution": "University of Paris",
+        }
+    ]
+
+
 def test_extract_cv_signals_ignores_before_moving_prose_tail() -> None:
     text = dedent(
         """