Remove Stellenanteil matching, use name and Funktion hint only

src/taetigkeitsnachweis/matcher.py

@@ -1,4 +1,4 @@
-"""Record matching logic across allocation and master data."""
+"""Record matching logic across allocation and master data."""
 
 from __future__ import annotations
 
@@ -15,7 +15,6 @@ def calculate_foerderquote_percent(allocation: AllocationRecord, precision: int
         raise ValidationError(
             "contract_hours_weekly must be greater than 0 for foerderquote calculation."
         )
-
     raw_share = (allocation.funded_hours_weekly / allocation.contract_hours_weekly) * 100.0
     return round(raw_share, precision)
 
@@ -31,7 +30,6 @@ def _normalize_name(name: str) -> str:
 def _name_keys(name: str) -> set[str]:
     normalized = _normalize_name(name)
     keys = {normalized}
-
     if "," in normalized:
         last_name, first_name = [part.strip() for part in normalized.split(",", 1)]
         if first_name and last_name:
@@ -44,16 +42,17 @@ def _name_keys(name: str) -> set[str]:
             last_name = tokens[-1]
             keys.add(f"{first_name} {last_name}")
             keys.add(f"{last_name}, {first_name}")
-
     return {re.sub(r"\s+", " ", key).strip() for key in keys if key.strip()}
 
 
-def _with_tolerance(
-    value: float,
-    target: float,
-    tolerance: float,
-) -> bool:
-    return abs(value - target) <= tolerance
+def _disambiguate_by_project_key(
+    candidates: list,
+    project_key: str,
+) -> object:
+    """Return the single candidate whose Funktion contains the project key, or None."""
+    key = project_key.strip().casefold()
+    matched = [c for c in candidates if key in c.funktion.casefold()]
+    return matched[0] if len(matched) == 1 else None
 
 
 def match_assignments(
@@ -61,25 +60,33 @@ def match_assignments(
     masterdata_records: list[MasterDataRecord],
     config: dict[str, Any],
 ) -> tuple[list[MatchedAssignment], list[ValidationIssue]]:
-    matching_cfg = config.get("matching", {})
-    tolerance = float(matching_cfg.get("share_tolerance_percent", 0.5))
-    prefer_function_hint = bool(matching_cfg.get("prefer_function_hint", True))
-
+    # Build lookup by all normalized name variants
     master_by_name: dict[str, list[MasterDataRecord]] = {}
     for record in masterdata_records:
         for key in _name_keys(record.name):
             master_by_name.setdefault(key, []).append(record)
 
+    # Deduplicate so the same record does not appear twice for one name key
+    for key in master_by_name:
+        seen_ids: set[str] = set()
+        deduped: list[MasterDataRecord] = []
+        for rec in master_by_name[key]:
+            if rec.record_id not in seen_ids:
+                seen_ids.add(rec.record_id)
+                deduped.append(rec)
+        master_by_name[key] = deduped
+
     matched: list[MatchedAssignment] = []
     issues: list[ValidationIssue] = []
 
     for allocation in allocation_records:
+        # Collect all master-data candidates for this employee name
         candidate_map: dict[str, MasterDataRecord] = {}
         for key in _name_keys(allocation.employee_name):
             for candidate in master_by_name.get(key, []):
                 candidate_map[candidate.record_id] = candidate
-
         candidates = list(candidate_map.values())
+
         if not candidates:
             issues.append(
                 ValidationIssue(
@@ -94,66 +101,36 @@ def match_assignments(
             )
             continue
 
-        foerderquote = calculate_foerderquote_percent(allocation)
-        share_matches = [
-            candidate
-            for candidate in candidates
-            if _with_tolerance(candidate.stellenanteil_percent, foerderquote, tolerance)
-        ]
-
-        if len(share_matches) == 1:
-            selected = share_matches[0]
-        elif len(share_matches) > 1 and prefer_function_hint:
-            key = allocation.project_key.casefold()
-            hinted = [c for c in share_matches if key in c.funktion.casefold()]
-            if len(hinted) == 1:
-                selected = hinted[0]
-            else:
+        # Single candidate: use directly
+        if len(candidates) == 1:
+            selected = candidates[0]
+        else:
+            # Multiple candidates: disambiguate by project key in Funktion
+            selected = _disambiguate_by_project_key(candidates, allocation.project_key)
+            if selected is None:
                 issues.append(
                     ValidationIssue(
                         code="AMBIGUOUS_MASTERDATA",
                         message=(
-                            f"Ambiguous match for employee '{allocation.employee_name}' "
-                            f"(project '{allocation.project_key}', foerderquote {foerderquote:.2f}%)."
+                            f"Ambiguous master-data for '{allocation.employee_name}' / "
+                            f"'{allocation.project_key}': {len(candidates)} rows, "
+                            f"none or multiple match Funktion. "
+                            f"Funktionen: {[c.funktion for c in candidates]}"
                         ),
                         severity="error",
                         context=allocation.employee_name,
                     )
                 )
                 continue
-        elif len(share_matches) > 1:
-            issues.append(
-                ValidationIssue(
-                    code="AMBIGUOUS_MASTERDATA",
-                    message=(
-                        f"Ambiguous match for employee '{allocation.employee_name}' "
-                        f"(project '{allocation.project_key}', foerderquote {foerderquote:.2f}%)."
-                    ),
-                    severity="error",
-                    context=allocation.employee_name,
-                )
-            )
-            continue
-        else:
-            issues.append(
-                ValidationIssue(
-                    code="NO_SHARE_MATCH",
-                    message=(
-                        f"No Stellenanteil match for employee '{allocation.employee_name}' "
-                        f"with foerderquote {foerderquote:.2f}% (project '{allocation.project_key}')."
-                    ),
-                    severity="error",
-                    context=allocation.employee_name,
-                )
-            )
-            continue
+
+        foerderquote = calculate_foerderquote_percent(allocation)
 
         if allocation.notes:
             issues.append(
                 ValidationIssue(
                     code="ALLOCATION_NOTE",
                     message=(
-                        f"Allocation note for '{allocation.employee_name}' / "
+                        f"Note for '{allocation.employee_name}' / "
                         f"'{allocation.project_key}': {allocation.notes}"
                     ),
                     severity="warning",