scripts/bootstrap_qdrant_plans.py aktualisiert
All checks were successful
Deploy Trainer_LLM to llm-node / deploy (push) Successful in 2s
All checks were successful
Deploy Trainer_LLM to llm-node / deploy (push) Successful in 2s
This commit is contained in:
parent
9c955db191
commit
4d67cd9d66
|
|
@ -1,10 +1,12 @@
|
||||||
#!/usr/bin/env python3
|
#!/usr/bin/env python3
|
||||||
# -*- coding: utf-8 -*-
|
# -*- coding: utf-8 -*-
|
||||||
"""
|
"""
|
||||||
Idempotenter Bootstrap für Qdrant-Collections rund um Pläne.
|
Bootstrap (idempotent) für Qdrant-Collections rund um Pläne – v1.1.0
|
||||||
- Legt fehlende Collections NICHT neu an (wir erwarten, dass der erste Upsert sie erzeugt)
|
|
||||||
- Legt fehlende Payload-Indizes als KEYWORD an (wo sinnvoll)
|
- Legt fehlende Payload-Indizes als KEYWORD an (wo sinnvoll)
|
||||||
- Zeitfelder: bleiben textuell (ISO8601); optional kann created_at_ts später ergänzt werden
|
- Nutzt materialisierte Facettenfelder auf Templates (section_*),
|
||||||
|
um verschachtelte Pfade stabil zu indizieren
|
||||||
|
- Collections selbst werden NICHT erstellt; das passiert beim ersten Upsert
|
||||||
|
|
||||||
Start:
|
Start:
|
||||||
pip install qdrant-client --upgrade
|
pip install qdrant-client --upgrade
|
||||||
|
|
@ -22,13 +24,20 @@ TEMPLATES = os.getenv("PLAN_TEMPLATE_COLLECTION", "plan_templates")
|
||||||
SESSIONS = os.getenv("PLAN_SESSION_COLLECTION", "plan_sessions")
|
SESSIONS = os.getenv("PLAN_SESSION_COLLECTION", "plan_sessions")
|
||||||
|
|
||||||
INDEX_SPECS = {
|
INDEX_SPECS = {
|
||||||
|
# Templates: nutzen materialisierte Felder für stabile Indizes
|
||||||
TEMPLATES: [
|
TEMPLATES: [
|
||||||
("discipline", PayloadSchemaType.KEYWORD),
|
("discipline", PayloadSchemaType.KEYWORD),
|
||||||
("age_group", PayloadSchemaType.KEYWORD),
|
("age_group", PayloadSchemaType.KEYWORD),
|
||||||
("target_group", PayloadSchemaType.KEYWORD),
|
("target_group", PayloadSchemaType.KEYWORD),
|
||||||
("sections.name", PayloadSchemaType.KEYWORD),
|
("section_names", PayloadSchemaType.KEYWORD),
|
||||||
|
("section_must_keywords", PayloadSchemaType.KEYWORD),
|
||||||
|
("section_ideal_keywords", PayloadSchemaType.KEYWORD),
|
||||||
|
("section_supplement_keywords", PayloadSchemaType.KEYWORD),
|
||||||
|
("section_forbid_keywords", PayloadSchemaType.KEYWORD),
|
||||||
("goals", PayloadSchemaType.KEYWORD),
|
("goals", PayloadSchemaType.KEYWORD),
|
||||||
],
|
],
|
||||||
|
# Plans: belassen verschachtelte Felder wie bisher; falls deine Qdrant-Version das
|
||||||
|
# nicht unterstützt, können wir analog materialisieren (section_names) und hier ergänzen.
|
||||||
PLANS: [
|
PLANS: [
|
||||||
("discipline", PayloadSchemaType.KEYWORD),
|
("discipline", PayloadSchemaType.KEYWORD),
|
||||||
("age_group", PayloadSchemaType.KEYWORD),
|
("age_group", PayloadSchemaType.KEYWORD),
|
||||||
|
|
@ -38,13 +47,13 @@ INDEX_SPECS = {
|
||||||
("created_by", PayloadSchemaType.KEYWORD),
|
("created_by", PayloadSchemaType.KEYWORD),
|
||||||
("created_at", PayloadSchemaType.KEYWORD),
|
("created_at", PayloadSchemaType.KEYWORD),
|
||||||
("fingerprint", PayloadSchemaType.KEYWORD),
|
("fingerprint", PayloadSchemaType.KEYWORD),
|
||||||
("title", PayloadSchemaType.KEYWORD)
|
("title", PayloadSchemaType.KEYWORD),
|
||||||
],
|
],
|
||||||
SESSIONS: [
|
SESSIONS: [
|
||||||
("plan_id", PayloadSchemaType.KEYWORD),
|
("plan_id", PayloadSchemaType.KEYWORD),
|
||||||
("executed_at", PayloadSchemaType.KEYWORD),
|
("executed_at", PayloadSchemaType.KEYWORD),
|
||||||
("coach", PayloadSchemaType.KEYWORD),
|
("coach", PayloadSchemaType.KEYWORD),
|
||||||
("group_label", PayloadSchemaType.KEYWORD)
|
("group_label", PayloadSchemaType.KEYWORD),
|
||||||
],
|
],
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
@ -53,24 +62,26 @@ def _create_indexes(client: QdrantClient, collection: str, specs):
|
||||||
# Prüfe, ob Collection existiert (wir erwarten sie). Falls nicht, Hinweis, aber nicht fatal.
|
# Prüfe, ob Collection existiert (wir erwarten sie). Falls nicht, Hinweis, aber nicht fatal.
|
||||||
try:
|
try:
|
||||||
client.get_collection(collection)
|
client.get_collection(collection)
|
||||||
print(f"[Bootstrap] Collection '{collection}' ok.")
|
print(f"[Bootstrap v1.1.0] Collection '{collection}' ok.")
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
print(f"[Bootstrap] WARN: Collection '{collection}' nicht gefunden (wird beim ersten Upsert erstellt). Details: {e}")
|
print(f"[Bootstrap v1.1.0] WARN: Collection '{collection}' nicht gefunden (wird beim ersten Upsert erstellt). Details: {e}")
|
||||||
return
|
return
|
||||||
|
|
||||||
for field, schema in specs:
|
for field, schema in specs:
|
||||||
try:
|
try:
|
||||||
client.create_payload_index(collection_name=collection, field_name=field, field_schema=schema)
|
client.create_payload_index(collection_name=collection, field_name=field, field_schema=schema)
|
||||||
print(f"[Bootstrap] Index created: {collection}.{field} ({schema})")
|
print(f"[Bootstrap v1.1.0] Index created: {collection}.{field} ({schema})")
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
print(f"[Bootstrap] Index exists or skipped: {collection}.{field} -> {e}")
|
print(f"[Bootstrap v1.1.0] Index exists or skipped: {collection}.{field} -> {e}")
|
||||||
|
|
||||||
|
|
||||||
def main():
|
def main():
|
||||||
|
print(f"[Bootstrap v1.1.0] Qdrant @ {QDRANT_HOST}:{QDRANT_PORT}")
|
||||||
|
print(f"[Bootstrap v1.1.0] Collections: TEMPLATES={TEMPLATES} PLANS={PLANS} SESSIONS={SESSIONS}")
|
||||||
client = QdrantClient(host=QDRANT_HOST, port=QDRANT_PORT)
|
client = QdrantClient(host=QDRANT_HOST, port=QDRANT_PORT)
|
||||||
for coll, specs in INDEX_SPECS.items():
|
for coll, specs in INDEX_SPECS.items():
|
||||||
_create_indexes(client, coll, specs)
|
_create_indexes(client, coll, specs)
|
||||||
print("[Bootstrap] done.")
|
print("[Bootstrap v1.1.0] done.")
|
||||||
|
|
||||||
|
|
||||||
if __name__ == "__main__":
|
if __name__ == "__main__":
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue
Block a user