llm-api/exercise_router.py aktualisiert

llm-api/exercise_router.py

@@ -1,15 +1,17 @@
 # -*- coding: utf-8 -*-
 """
-exercise_router.py – v1.7.0
+exercise_router.py – v1.7.1 (Swagger angereichert)
 
-Neu:
+Ergänzt:
-- Endpoint **POST /exercise/search**: kombinierbare Filter (discipline, duration, equipment any/all, keywords any/all,
+- Aussagekräftige summary/description/response_description je Endpoint
-  capability_geN / capability_eqN + names) + optionaler Vektor-Query (query-Text). Ausgabe inkl. Score.
+- Beispiele (x-codeSamples) für curl-Aufrufe
-- Facetten erweitert: neben capability_ge1..ge5 jetzt auch capability_eq1..eq5.
+- Pydantic-Felder mit description + json_schema_extra (Beispiele)
-- Idempotenz-Fix & Payload-Scroll (aus v1.6.2) beibehalten.
+- Keine API-Signatur-/Pfadänderungen, keine Prefix-Änderungen
-- API-Signaturen bestehender Routen unverändert.
 
-Hinweis: Die „eq/ge“-Felder werden beim Upsert gesetzt; für Alt-Punkte einmal das Backfill laufen lassen.
+Hinweis:
+- Endpunkte bleiben weiterhin unter /exercise/* (weil die Routenstrings bereits /exercise/... enthalten).
+- Falls du später einen APIRouter-Prefix setzen willst, dann bitte die Pfade unten von '/exercise/...' auf relative Pfade ändern,
+  sonst entstehen Doppelpfade.
 """
 
 from fastapi import APIRouter, HTTPException, Query
@@ -27,77 +29,137 @@ from qdrant_client.models import (
     FieldCondition,
     MatchValue,
 )
+import logging
 import os
 
-router = APIRouter()
+logger = logging.getLogger("exercise_router")
+logger.setLevel(logging.INFO)
+
+# Router ohne prefix (Pfadstrings enthalten bereits '/exercise/...')
+router = APIRouter(tags=["exercise"])
 
 # =========================
 # Models
 # =========================
 class Exercise(BaseModel):
-    id: str = Field(default_factory=lambda: str(uuid4()))
+    id: str = Field(default_factory=lambda: str(uuid4()), description="Interne UUID (Qdrant-Punkt-ID)")
     # Upsert-Metadaten
-    external_id: Optional[str] = None
+    external_id: Optional[str] = Field(default=None, description="Upsert-Schlüssel (z. B. 'mw:{pageid}')")
-    fingerprint: Optional[str] = None
+    fingerprint: Optional[str] = Field(default=None, description="sha256 der Kernfelder für Idempotenz/Diff")
-    source: Optional[str] = None
+    source: Optional[str] = Field(default=None, description="Quelle (z. B. 'mediawiki', 'pdf-import', …)")
-    imported_at: Optional[datetime] = None
+    imported_at: Optional[datetime] = Field(default=None, description="Zeitpunkt des Imports (ISO-8601)")
 
     # Domain-Felder
-    title: str
+    title: str = Field(..., description="Übungstitel")
-    summary: str
+    summary: str = Field(..., description="Kurzbeschreibung/Ziel der Übung")
-    short_description: str
+    short_description: str = Field(..., description="Alternative Kurzform / Teaser")
-    keywords: List[str] = []
+    keywords: List[str] = Field(default_factory=list, description="Freie Schlagworte (normalisiert)")
-    link: Optional[str] = None
+    link: Optional[str] = Field(default=None, description="Kanonsiche URL/Permalink zur Quelle")
-    discipline: str
+    discipline: str = Field(..., description="Disziplin (z. B. Karate)")
-    group: Optional[str] = None
+    group: Optional[str] = Field(default=None, description="Optionale Gruppierung/Kategorie")
-    age_group: str
+    age_group: str = Field(..., description="Altersgruppe (z. B. Kinder/Schüler/Teenager/Erwachsene)")
-    target_group: str
+    target_group: str = Field(..., description="Zielgruppe (z. B. Breitensportler)")
-    min_participants: int
+    min_participants: int = Field(..., ge=0, description="Minimale Gruppenstärke")
-    duration_minutes: int
+    duration_minutes: int = Field(..., ge=0, description="Dauer in Minuten")
-    capabilities: Dict[str, int] = {}
+    capabilities: Dict[str, int] = Field(default_factory=dict, description="Fähigkeiten-Map: {Name: Level 1..5}")
-    category: str
+    category: str = Field(..., description="Abschnitt / Kategorie (z. B. Aufwärmen, Grundschule, …)")
-    purpose: str
+    purpose: str = Field(..., description="Zweck/Zielabsicht")
-    execution: str
+    execution: str = Field(..., description="Durchführungsschritte (Markdown/Wiki-ähnlich)")
-    notes: str
+    notes: str = Field(..., description="Hinweise/Coaching-Cues")
-    preparation: str
+    preparation: str = Field(..., description="Vorbereitung/Material")
-    method: str
+    method: str = Field(..., description="Methodik/Didaktik")
-    equipment: List[str] = []
+    equipment: List[str] = Field(default_factory=list, description="Benötigte Hilfsmittel")
+
+    model_config = {
+        "json_schema_extra": {
+            "example": {
+                "external_id": "mw:218",
+                "title": "Affenklatschen",
+                "summary": "Koordination & Aufmerksamkeit mit Ballwechseln",
+                "short_description": "Ballgewöhnung im Stand/Gehen/Laufen",
+                "keywords": ["Hand-Auge-Koordination", "Reaktion"],
+                "link": "https://www.karatetrainer.de/index.php?title=Affenklatschen",
+                "discipline": "Karate",
+                "age_group": "Teenager",
+                "target_group": "Breitensportler",
+                "min_participants": 4,
+                "duration_minutes": 8,
+                "capabilities": {"Reaktionsfähigkeit": 2, "Kopplungsfähigkeit": 2},
+                "category": "Aufwärmen",
+                "purpose": "Aufmerksamkeit & Reaktionskette aktivieren",
+                "execution": "* Paarweise aufstellen …",
+                "notes": "* nicht zu lange werden lassen",
+                "preparation": "* Bälle bereit halten",
+                "method": "* klare Regeln/Strafrunde",
+                "equipment": ["Bälle"]
+            }
+        }
+    }
 
 class DeleteResponse(BaseModel):
-    status: str
+    status: str = Field(..., description="Statusmeldung")
-    count: int
+    count: int = Field(..., ge=0, description="Anzahl betroffener Punkte")
-    collection: str
+    collection: str = Field(..., description="Qdrant-Collection-Name")
 
 class ExerciseSearchRequest(BaseModel):
     # Optionaler Semantik-Query (Vektor)
-    query: Optional[str] = None
+    query: Optional[str] = Field(default=None, description="Freitext für Vektor-Suche (optional)")
-    limit: int = Field(default=20, ge=1, le=200)
+    limit: int = Field(default=20, ge=1, le=200, description="Max. Treffer")
-    offset: int = Field(default=0, ge=0)
+    offset: int = Field(default=0, ge=0, description="Offset/Pagination")
 
     # Einfache Filter
-    discipline: Optional[str] = None
+    discipline: Optional[str] = Field(default=None, description="z. B. Karate")
-    target_group: Optional[str] = None
+    target_group: Optional[str] = Field(default=None, description="z. B. Breitensportler")
-    age_group: Optional[str] = None
+    age_group: Optional[str] = Field(default=None, description="z. B. Teenager")
-    max_duration: Optional[int] = Field(default=None, ge=0)
+    max_duration: Optional[int] = Field(default=None, ge=0, description="Obergrenze Minuten")
 
     # Listen-Filter
-    equipment_any: Optional[List[str]] = None    # mindestens eins muss passen
+    equipment_any: Optional[List[str]] = Field(default=None, description="Mind. eines muss passen")
-    equipment_all: Optional[List[str]] = None    # alle müssen passen
+    equipment_all: Optional[List[str]] = Field(default=None, description="Alle müssen passen")
-    keywords_any: Optional[List[str]] = None
+    keywords_any: Optional[List[str]] = Field(default=None, description="Mind. eines muss passen")
-    keywords_all: Optional[List[str]] = None
+    keywords_all: Optional[List[str]] = Field(default=None, description="Alle müssen passen")
 
     # Capabilities (Namen + Level-Operator)
-    capability_names: Optional[List[str]] = None
+    capability_names: Optional[List[str]] = Field(default=None, description="Capability-Bezeichnungen")
-    capability_ge_level: Optional[int] = Field(default=None, ge=1, le=5)
+    capability_ge_level: Optional[int] = Field(default=None, ge=1, le=5, description="Level ≥ N")
-    capability_eq_level: Optional[int] = Field(default=None, ge=1, le=5)
+    capability_eq_level: Optional[int] = Field(default=None, ge=1, le=5, description="Level == N")
+
+    model_config = {
+        "json_schema_extra": {
+            "examples": [{
+                "discipline": "Karate",
+                "max_duration": 12,
+                "equipment_any": ["Bälle"],
+                "capability_names": ["Reaktionsfähigkeit"],
+                "capability_ge_level": 2,
+                "limit": 5
+            }, {
+                "query": "Aufwärmen Reaktionsfähigkeit 10min Teenager Bälle",
+                "discipline": "Karate",
+                "limit": 3
+            }]
+        }
+    }
 
 class ExerciseSearchHit(BaseModel):
-    id: str
+    id: str = Field(..., description="Qdrant-Punkt-ID")
-    score: Optional[float] = None
+    score: Optional[float] = Field(default=None, description="Ähnlichkeitsscore (nur bei Vektor-Suche)")
-    payload: Exercise
+    payload: Exercise = Field(..., description="Übungsdaten (Payload)")
 
 class ExerciseSearchResponse(BaseModel):
-    hits: List[ExerciseSearchHit]
+    hits: List[ExerciseSearchHit] = Field(..., description="Trefferliste")
+
+    model_config = {
+        "json_schema_extra": {
+            "example": {
+                "hits": [{
+                    "id": "c1f1-…",
+                    "score": 0.78,
+                    "payload": Exercise.model_config["json_schema_extra"]["example"]
+                }]
+            }
+        }
+    }
 
 # =========================
 # Helpers
@@ -160,6 +222,12 @@ def _norm_list(xs: List[Any]) -> List[str]:
 
 
 def _facet_capabilities(caps: Dict[str, Any]) -> Dict[str, List[str]]:
+    """
+    Leitet Facettenfelder aus der capabilities-Map ab:
+    - capability_keys: alle Namen
+    - capability_geN: Namen mit Level >= N (1..5)
+    - capability_eqN: Namen mit Level == N (1..5)
+    """
     caps = caps or {}
 
     def names_where(pred) -> List[str]:
@@ -194,6 +262,7 @@ def _facet_capabilities(caps: Dict[str, Any]) -> Dict[str, List[str]]:
 
 
 def _response_strip_extras(payload: Dict[str, Any]) -> Dict[str, Any]:
+    # Nur definierte Exercise-Felder zurückgeben (saubere API)
     allowed = set(Exercise.model_fields.keys())
     return {k: v for k, v in payload.items() if k in allowed}
 
@@ -209,8 +278,7 @@ def _build_filter(req: ExerciseSearchRequest) -> Filter:
     if req.age_group:
         must.append(FieldCondition(key="age_group", match=MatchValue(value=req.age_group)))
     if req.max_duration is not None:
-        # Range ohne Import zusätzlicher Modelle: Qdrant akzeptiert auch {'range': {'lte': n}} per JSON;
+        # Range in Qdrant: über rohen JSON-Range-Ausdruck (Client-Modell hat keinen Komfort-Wrapper)
-        # über Client-Modell tun wir es hier nicht, da wir Filter primär für Keyword-Felder nutzen.
         must.append({"key": "duration_minutes", "range": {"lte": int(req.max_duration)}})
 
     # equipment
@@ -218,7 +286,6 @@ def _build_filter(req: ExerciseSearchRequest) -> Filter:
         for it in req.equipment_all:
             must.append(FieldCondition(key="equipment", match=MatchValue(value=it)))
     if req.equipment_any:
-        # OR: über 'should' Liste
         for it in req.equipment_any:
             should.append(FieldCondition(key="equipment", match=MatchValue(value=it)))
 
@@ -248,22 +315,55 @@ def _build_filter(req: ExerciseSearchRequest) -> Filter:
 
     flt = Filter(must=must)
     if should:
-        # qdrant: 'should' mit implizitem minimum_should_match=1
+        # Qdrant: 'should' entspricht OR mit minimum_should_match=1
         flt.should = should
     return flt
 
 # =========================
 # Endpoints
 # =========================
-@router.get("/exercise/by-external-id")
+@router.get(
-def get_exercise_by_external_id(external_id: str = Query(..., min_length=3)):
+    "/exercise/by-external-id",
+    summary="Übung per external_id abrufen",
+    description=(
+        "Liefert die Übung mit der gegebenen `external_id` (z. B. `mw:{pageid}`). "
+        "Verwendet einen Qdrant-Filter auf dem Payload-Feld `external_id`."
+    ),
+    response_description="Vollständiger Exercise-Payload oder 404 bei Nichtfund.",
+    openapi_extra={
+        "x-codeSamples": [{
+            "lang": "bash",
+            "label": "curl",
+            "source": "curl -s 'http://localhost:8000/exercise/by-external-id?external_id=mw:218' | jq ."
+        }]
+    }
+)
+def get_exercise_by_external_id(external_id: str = Query(..., min_length=3, description="Upsert-Schlüssel, z. B. 'mw:218'")):
     found = _lookup_by_external_id(external_id)
     if not found:
         raise HTTPException(status_code=404, detail="not found")
     return found
 
 
-@router.post("/exercise", response_model=Exercise)
+@router.post(
+    "/exercise",
+    response_model=Exercise,
+    summary="Create/Update (idempotent per external_id)",
+    description=(
+        "Legt eine Übung an oder aktualisiert sie. Wenn `external_id` vorhanden und bereits in der Collection existiert, "
+        "wird **Update** auf dem bestehenden Punkt ausgeführt (Upsert). `keywords`/`equipment` werden normalisiert, "
+        "Capability-Facetten (`capability_ge1..5`, `capability_eq1..5`, `capability_keys`) automatisch abgeleitet. "
+        "Der Vektor wird aus Kernfeldern (title/summary/short_description/purpose/execution/notes) berechnet."
+    ),
+    response_description="Gespeicherter Exercise-Datensatz (Payload-View).",
+    openapi_extra={
+        "x-codeSamples": [{
+            "lang": "bash",
+            "label": "curl",
+            "source": "curl -s -X POST http://localhost:8000/exercise -H 'Content-Type: application/json' -d @exercise.json | jq ."
+        }]
+    }
+)
 def create_or_update_exercise(ex: Exercise):
     _ensure_collection()
 
@@ -290,7 +390,20 @@ def create_or_update_exercise(ex: Exercise):
     return Exercise(**_response_strip_extras(payload))
 
 
-@router.get("/exercise/{exercise_id}", response_model=Exercise)
+@router.get(
+    "/exercise/{exercise_id}",
+    response_model=Exercise,
+    summary="Übung per interner ID (Qdrant-Punkt-ID) lesen",
+    description="Scrollt nach `id` und gibt den Payload als Exercise zurück.",
+    response_description="Exercise-Payload oder 404 bei Nichtfund.",
+    openapi_extra={
+        "x-codeSamples": [{
+            "lang": "bash",
+            "label": "curl",
+            "source": "curl -s 'http://localhost:8000/exercise/1234-uuid' | jq ."
+        }]
+    }
+)
 def get_exercise(exercise_id: str):
     _ensure_collection()
     pts, _ = qdrant.scroll(
@@ -306,7 +419,32 @@ def get_exercise(exercise_id: str):
     return Exercise(**_response_strip_extras(payload))
 
 
-@router.post("/exercise/search", response_model=ExerciseSearchResponse)
+@router.post(
+    "/exercise/search",
+    response_model=ExerciseSearchResponse,
+    summary="Suche Übungen (Filter + optional Vektor)",
+    description=(
+        "Kombinierbare Filter auf Payload-Feldern (`discipline`, `age_group`, `target_group`, `equipment`, `keywords`, "
+        "`capability_geN/eqN`) und **optional** Vektor-Suche via `query`. "
+        "`should`-Filter (equipment_any/keywords_any) wirken als OR (minimum_should_match=1). "
+        "`max_duration` wird als Range (lte) angewandt. Ergebnis enthält bei Vektor-Suche `score`, sonst `null`."
+    ),
+    response_description="Trefferliste (payload + Score bei Vektor-Suche).",
+    openapi_extra={
+        "x-codeSamples": [
+            {
+                "lang": "bash",
+                "label": "Filter",
+                "source": "curl -s -X POST http://localhost:8000/exercise/search -H 'Content-Type: application/json' -d '{\"discipline\":\"Karate\",\"max_duration\":12,\"equipment_any\":[\"Bälle\"],\"capability_names\":[\"Reaktionsfähigkeit\"],\"capability_ge_level\":2,\"limit\":5}' | jq ."
+            },
+            {
+                "lang": "bash",
+                "label": "Vektor + Filter",
+                "source": "curl -s -X POST http://localhost:8000/exercise/search -H 'Content-Type: application/json' -d '{\"query\":\"Aufwärmen 10min Teenager Bälle\",\"discipline\":\"Karate\",\"limit\":3}' | jq ."
+            }
+        ]
+    }
+)
 def search_exercises(req: ExerciseSearchRequest) -> ExerciseSearchResponse:
     _ensure_collection()
     flt = _build_filter(req)
@@ -314,7 +452,6 @@ def search_exercises(req: ExerciseSearchRequest) -> ExerciseSearchResponse:
     hits: List[ExerciseSearchHit] = []
     if req.query:
         vec = _make_vector_from_query(req.query)
-        # qdrant_client.search unterstützt offset/limit
         res = qdrant.search(
             collection_name=COLLECTION,
             query_vector=vec,
@@ -327,8 +464,7 @@ def search_exercises(req: ExerciseSearchRequest) -> ExerciseSearchResponse:
             payload.setdefault("id", str(h.id))
             hits.append(ExerciseSearchHit(id=str(h.id), score=float(h.score or 0.0), payload=Exercise(**_response_strip_extras(payload))))
     else:
-        # Filter-only: per Scroll (ohne Score); einfache Paginierung via offset/limit
+        # Filter-only: Scroll-Paginierung, Score=None
-        # Hole offset+limit Punkte und simuliere Score=None
         collected = 0
         skipped = 0
         next_offset = None
@@ -357,8 +493,24 @@ def search_exercises(req: ExerciseSearchRequest) -> ExerciseSearchResponse:
     return ExerciseSearchResponse(hits=hits)
 
 
-@router.delete("/exercise/delete-by-external-id", response_model=DeleteResponse)
+@router.delete(
-def delete_by_external_id(external_id: str = Query(...)):
+    "/exercise/delete-by-external-id",
+    response_model=DeleteResponse,
+    summary="Löscht Punkte mit gegebener external_id",
+    description=(
+        "Scrollt nach `external_id` und löscht alle passenden Punkte. "
+        "Idempotent: wenn nichts gefunden → count=0. Vorsicht: **löscht dauerhaft**."
+    ),
+    response_description="Status + Anzahl gelöschter Punkte.",
+    openapi_extra={
+        "x-codeSamples": [{
+            "lang": "bash",
+            "label": "curl",
+            "source": "curl -s 'http://localhost:8000/exercise/delete-by-external-id?external_id=mw:9999' | jq ."
+        }]
+    }
+)
+def delete_by_external_id(external_id: str = Query(..., description="Upsert-Schlüssel, z. B. 'mw:218'")):
     _ensure_collection()
     flt = Filter(must=[FieldCondition(key="external_id", match=MatchValue(value=external_id))])
     pts, _ = qdrant.scroll(collection_name=COLLECTION, scroll_filter=flt, limit=10000, with_payload=False)
@@ -369,8 +521,24 @@ def delete_by_external_id(external_id: str = Query(...)):
     return DeleteResponse(status="🗑️ gelöscht", count=len(ids), collection=COLLECTION)
 
 
-@router.delete("/exercise/delete-collection", response_model=DeleteResponse)
+@router.delete(
-def delete_collection(collection: str = Query(default=COLLECTION)):
+    "/exercise/delete-collection",
+    response_model=DeleteResponse,
+    summary="Collection komplett löschen",
+    description=(
+        "Entfernt die gesamte Collection aus Qdrant. **Gefährlich** – alle Übungen sind danach weg. "
+        "Nutze nur in Testumgebungen oder für einen kompletten Neuaufbau."
+    ),
+    response_description="Status. count=0 (nicht relevant beim Drop).",
+    openapi_extra={
+        "x-codeSamples": [{
+            "lang": "bash",
+            "label": "curl",
+            "source": "curl -s 'http://localhost:8000/exercise/delete-collection?collection=exercises' | jq ."
+        }]
+    }
+)
+def delete_collection(collection: str = Query(default=COLLECTION, description="Collection-Name (Default: 'exercises')")):
     if not qdrant.collection_exists(collection):
         raise HTTPException(status_code=404, detail=f"Collection '{collection}' nicht gefunden.")
     qdrant.delete_collection(collection_name=collection)
@@ -384,7 +552,6 @@ TEST_DOC = """
 Speicher als tests/test_exercise_search.py und mit pytest laufen lassen.
 
 import os, requests
-
 BASE = os.getenv("API_BASE", "http://localhost:8000")
 
 # 1) Filter-only