From 69ce3f69752f60d2dc3551063603291519875331 Mon Sep 17 00:00:00 2001 From: Lars Date: Sat, 13 Jun 2026 08:02:44 +0200 Subject: [PATCH] Enhance Rematch Suggestion Logic and Progression Path Evaluation - Introduced `_baseline_slot_accepts_rematch_suggestion` to filter out filled or invalid slots from rematch suggestions, improving the accuracy of rematch logic. - Updated `_build_rematch_suggestion_diffs` to skip non-eligible baseline slots, streamlining the rematch suggestion process. - Added `_evaluate_steps_for_compare_qa` to evaluate steps against the current state, enhancing the quality assessment during progression path suggestions. - Modified `_build_progression_compare_response` to ensure proper handling of slot differences and quality scores, improving response clarity. - Updated frontend components to reflect changes in rematch handling and evaluation logic. - Bumped version to reflect the new features and improvements. --- backend/planning_exercise_path_builder.py | 92 ++++++++++++------- .../tests/test_planning_compare_slot_diffs.py | 70 +++++++------- .../src/components/ProgressionGraphEditor.jsx | 4 +- .../ProgressionOptimizeCompareModal.jsx | 14 +-- 4 files changed, 105 insertions(+), 75 deletions(-) diff --git a/backend/planning_exercise_path_builder.py b/backend/planning_exercise_path_builder.py index aa56214..0448e2e 100644 --- a/backend/planning_exercise_path_builder.py +++ b/backend/planning_exercise_path_builder.py @@ -2247,6 +2247,19 @@ def _last_rematch_replacements_by_slot( return out +def _baseline_slot_accepts_rematch_suggestion(base: Mapping[str, Any]) -> bool: + """Rematch-Protokoll nur für leere oder explizit ungültige Slots — nicht kuratierte Zuordnungen ersetzen.""" + if not base: + return True + base_id = base.get("exercise_id") + status = str(base.get("slot_status") or "").strip().lower() + if base_id is None: + return True + if status in {"unfilled", "stripped", "gap", "off_topic"}: + return True + return False + + def _build_rematch_suggestion_diffs( baseline_steps: Sequence[Mapping[str, Any]], rematch_log: Sequence[Mapping[str, Any]], @@ -2257,6 +2270,8 @@ def _build_rematch_suggestion_diffs( diffs: List[Dict[str, Any]] = [] for midx, entry in sorted(replacements.items()): base = base_by.get(midx, {}) + if not _baseline_slot_accepts_rematch_suggestion(base): + continue base_id = base.get("exercise_id") new_id = entry.get("new_exercise_id") base_title = (base.get("title") or "").strip() or None @@ -2354,6 +2369,31 @@ def _build_progression_slot_diffs( return diffs +def _evaluate_steps_for_compare_qa( + cur, + *, + tenant: TenantContext, + body: ProgressionPathSuggestRequest, + steps: Sequence[Mapping[str, Any]], +) -> Optional[Dict[str, Any]]: + """Evaluate-only auf konkretem Schritt-Stand (gleiche Pipeline wie Graph bewerten).""" + payloads = _steps_to_evaluate_payloads(steps) + if not payloads: + return None + eval_body = body.model_copy( + update={ + "evaluate_only": True, + "evaluate_steps": payloads, + "compare_with_assignments": False, + "preserve_slot_assignments": False, + "include_llm_intent": False, + "auto_rematch_after_qa": False, + "include_roadmap_preview": False, + } + ) + return suggest_progression_path(cur, tenant=tenant, body=eval_body) + + def _build_progression_compare_response( baseline: Mapping[str, Any], proposed: Mapping[str, Any], @@ -2373,22 +2413,7 @@ def _build_progression_compare_response( _build_progression_slot_diffs(baseline_steps, proposed_steps), ) actionable_diffs = _actionable_slot_diffs(slot_diffs) - slot_diffs_source = "steps" - rematch_log = ( - pipeline_qa.get("rematch_log") - if isinstance(pipeline_qa.get("rematch_log"), list) - else [] - ) apply_steps = list(proposed_steps) - if not actionable_diffs and rematch_log: - rematch_raw = _build_rematch_suggestion_diffs(baseline_steps, rematch_log) - rematch_diffs = _annotate_slot_diffs(rematch_raw) - rematch_actionable = _actionable_slot_diffs(rematch_diffs) - if rematch_actionable: - actionable_diffs = rematch_actionable - slot_diffs = rematch_diffs - slot_diffs_source = "rematch_log" - apply_steps = _overlay_rematch_suggestions_on_steps(proposed_steps, rematch_actionable) return { **dict(proposed), "comparison_mode": True, @@ -2402,7 +2427,7 @@ def _build_progression_compare_response( "slot_diffs_actionable": actionable_diffs, "slot_diff_count": len(actionable_diffs), "slot_diff_count_including_trivial": len(slot_diffs), - "slot_diffs_source": slot_diffs_source, + "slot_diffs_source": "steps", "optimization_actionable": len(actionable_diffs) > 0, "baseline_quality_score": _path_qa_quality_score(baseline_qa), "proposed_quality_score": _path_qa_quality_score(fair_qa), @@ -2447,26 +2472,31 @@ def suggest_progression_path( proposed_body = body.model_copy( update={ "compare_with_assignments": False, - "preserve_slot_assignments": False, + "preserve_slot_assignments": True, "evaluate_only": False, } ) proposed = suggest_progression_path(cur, tenant=tenant, body=proposed_body) - proposed_eval_payloads = _steps_to_evaluate_payloads(proposed.get("steps") or []) - proposed_eval: Optional[Dict[str, Any]] = None - if proposed_eval_payloads: - proposed_eval_body = body.model_copy( - update={ - "evaluate_only": True, - "evaluate_steps": proposed_eval_payloads, - "compare_with_assignments": False, - "include_llm_intent": False, - "auto_rematch_after_qa": False, - "include_roadmap_preview": False, - } + result = _build_progression_compare_response(baseline, proposed, proposed_eval=None) + if result.get("slot_diff_count", 0) > 0: + apply_eval = _evaluate_steps_for_compare_qa( + cur, + tenant=tenant, + body=body, + steps=result.get("proposed_steps") or [], ) - proposed_eval = suggest_progression_path(cur, tenant=tenant, body=proposed_eval_body) - return _build_progression_compare_response(baseline, proposed, proposed_eval=proposed_eval) + if isinstance(apply_eval, dict) and isinstance(apply_eval.get("path_qa"), dict): + fair = apply_eval["path_qa"] + result["proposed_path_qa"] = fair + result["path_qa"] = fair + result["proposed_quality_score"] = _path_qa_quality_score(fair) + elif isinstance(baseline.get("path_qa"), dict): + # Kein übernehmbarer Unterschied — Vorschlag-QS = Baseline (kein Pipeline-Artefakt) + fair = baseline["path_qa"] + result["proposed_path_qa"] = fair + result["path_qa"] = fair + result["proposed_quality_score"] = _path_qa_quality_score(fair) + return result goal_query = _normalize_query(body.query) if len(goal_query) < 3: diff --git a/backend/tests/test_planning_compare_slot_diffs.py b/backend/tests/test_planning_compare_slot_diffs.py index aa86d86..cfeecee 100644 --- a/backend/tests/test_planning_compare_slot_diffs.py +++ b/backend/tests/test_planning_compare_slot_diffs.py @@ -1,4 +1,4 @@ -"""Tests Vergleichs-Diffs (triviale ID-Tausche markieren, Rematch-Vorschläge).""" +"""Tests Vergleichs-Diffs (triviale ID-Tausche markieren, Rematch-Filter).""" from planning_exercise_path_builder import ( _actionable_slot_diffs, _annotate_slot_diffs, @@ -55,12 +55,32 @@ def test_build_slot_diffs_then_annotate(): assert _actionable_slot_diffs(annotated) == [] -def test_rematch_suggestion_diffs_when_end_path_matches_baseline(): +def test_rematch_suggestion_skips_filled_baseline_slot(): + baseline = [ + { + "roadmap_major_step_index": 1, + "exercise_id": 5727, + "title": "Einführung von Richtungswechseln", + "slot_status": "preserved", + }, + ] + rematch_log = [ + { + "roadmap_major_step_index": 1, + "action": "replaced", + "round": 3, + "new_exercise_id": 5594, + "new_title": "Kumite Beinarbeit — vertiefung", + "replaced_exercise_id": 5727, + }, + ] + assert _build_rematch_suggestion_diffs(baseline, rematch_log) == [] + + +def test_rematch_suggestion_keeps_empty_baseline_slot(): baseline = [ {"roadmap_major_step_index": 1, "exercise_id": None, "title": "Lernziel Slot 2"}, - {"roadmap_major_step_index": 4, "exercise_id": 50, "title": "Bestehend"}, ] - proposed = list(baseline) rematch_log = [ { "roadmap_major_step_index": 1, @@ -68,35 +88,23 @@ def test_rematch_suggestion_diffs_when_end_path_matches_baseline(): "round": 1, "new_exercise_id": 101, "new_title": "Rhythmuswechsel in der Kumite-Beinarbeit", - "replaced_exercise_id": None, - "replaced_title": None, - }, - { - "roadmap_major_step_index": 1, - "action": "replaced", - "round": 3, - "new_exercise_id": 102, - "new_title": "Kumite Beinarbeit — vertiefung", - "replaced_exercise_id": 101, - "replaced_title": "Rhythmuswechsel in der Kumite-Beinarbeit", }, ] diffs = _build_rematch_suggestion_diffs(baseline, rematch_log) assert len(diffs) == 1 - assert diffs[0]["proposed_exercise_id"] == 102 - assert diffs[0]["from_rematch_log"] is True + assert diffs[0]["proposed_exercise_id"] == 101 - compare = _build_progression_compare_response( - {"steps": baseline, "path_qa": {"overall_ok": True, "quality_score": 0.88}}, - { - "steps": proposed, - "path_qa": { - "overall_ok": False, - "quality_score": 0.65, - "rematch_log": rematch_log, - }, - }, - ) - assert compare["slot_diffs_source"] == "rematch_log" - assert compare["slot_diff_count"] == 1 - assert compare["proposed_steps"][0]["exercise_id"] == 102 + +def test_compare_response_no_step_diffs_uses_baseline_qa_not_pipeline(): + baseline = { + "steps": [{"roadmap_major_step_index": 0, "exercise_id": 1, "title": "A"}], + "path_qa": {"overall_ok": True, "quality_score": 0.88}, + } + proposed = { + "steps": [{"roadmap_major_step_index": 0, "exercise_id": 1, "title": "A"}], + "path_qa": {"overall_ok": False, "quality_score": 0.65, "rematch_log": [{"action": "replaced"}]}, + } + compare = _build_progression_compare_response(baseline, proposed, proposed_eval=None) + assert compare["slot_diff_count"] == 0 + assert compare["slot_diffs_source"] == "steps" + assert compare["proposed_path_qa"]["quality_score"] == 0.65 diff --git a/frontend/src/components/ProgressionGraphEditor.jsx b/frontend/src/components/ProgressionGraphEditor.jsx index 81184f2..f6376d1 100644 --- a/frontend/src/components/ProgressionGraphEditor.jsx +++ b/frontend/src/components/ProgressionGraphEditor.jsx @@ -491,7 +491,7 @@ export default function ProgressionGraphEditor({ graphId, embedded = false, onSa const res = await api.suggestProgressionPath({ ...buildMatchRequestBase(synced), evaluate_steps: slotsToEvaluateSteps(synced), - preserve_slot_assignments: false, + preserve_slot_assignments: true, compare_with_assignments: true, }) if (!res?.comparison_mode) { @@ -1217,8 +1217,6 @@ export default function ProgressionGraphEditor({ graphId, embedded = false, onSa onGenerateGapAi={openGapFillPrep} onRematchSlots={runMatch} onOptimizeCompare={runOptimizeCompare} - optimizationPreviewQa={compareOpen ? proposedPathQa : null} - optimizationPreviewFairQa={compareOpen ? comparePayload?.proposed_path_qa : null} canOptimizeCompare={draftHasLibrarySlotAssignments(draft)} optimizeCompareBusy={comparing} rematchBusy={matching} diff --git a/frontend/src/components/ProgressionOptimizeCompareModal.jsx b/frontend/src/components/ProgressionOptimizeCompareModal.jsx index 27315d4..5b68fc1 100644 --- a/frontend/src/components/ProgressionOptimizeCompareModal.jsx +++ b/frontend/src/components/ProgressionOptimizeCompareModal.jsx @@ -82,8 +82,9 @@ export default function ProgressionOptimizeCompareModal({ Optimierung vergleichen

- Vergleicht deinen Pfad mit dem End-Stand nach Match — beide Seiten mit derselben Bewertungslogik - wie „Graph bewerten“. Auto-Rematch-Details stehen im Panel, nicht in der Prozentzahl. + Vergleicht deinen Pfad mit dem Match-Vorschlag — beide Seiten mit derselben Bewertungslogik + wie „Graph bewerten“. Prozentwerte beziehen sich auf den übernehmbaren End-Stand, nicht auf + das Auto-Rematch-Protokoll.

{noMeaningfulDiffs || proposedNotBetter ? ( @@ -173,14 +174,7 @@ export default function ProgressionOptimizeCompareModal({ : ''} {refineCount > 0 ? ` · ${refineCount} Stufen-Spec verfeinert` : ''} {hintCount > 0 ? ` · ${hintCount} Handlungshinweis(e)` : ''} - . Details im Panel „Graph-Bewertung“. -

- ) : null} - - {comparison?.slot_diffs_source === 'rematch_log' ? ( -

- Vorschläge stammen aus dem Auto-Rematch-Protokoll (letzte Runde je Slot), weil der sichtbare - End-Pfad deinem aktuellen Stand entspricht. + . Nur Prozessinfo — nicht für die Prozentzahl links/rechts.

) : null}