feat: add Qdrant recording workflow

scripts/build_digest_payload.py

@@ -0,0 +1,22 @@
+import json, uuid, os
+
+summary = open('/tmp/summary.txt').read()
+vector = json.load(open('/tmp/embed_response.json'))['embeddings'][0]
+point_id = str(uuid.uuid4())
+
+payload = {
+    "points": [{
+        "id": point_id,
+        "vector": vector,
+        "payload": {
+            "date": os.environ.get('TODAY', ''),
+            "type": "gitea_daily_digest",
+            "record_count": int(os.environ.get('COUNT', '0')),
+            "timestamp": os.environ.get('TIMESTAMP', ''),
+            "content": summary
+        }
+    }]
+}
+with open('/tmp/digest_payload.json', 'w') as f:
+    json.dump(payload, f)
+print(f"Digest Point ID: {point_id}")

scripts/build_embed_input.py

@@ -0,0 +1,3 @@
+import json
+summary = open('/tmp/summary.txt').read()
+open('/tmp/embed_input.json','w').write(json.dumps({'model':'bge-m3:latest','input':summary}))

scripts/build_gen_payload.py

@@ -0,0 +1,6 @@
+import json, sys, os
+today = os.environ.get('TODAY', '')
+prompt = open('/tmp/texts.txt').read()
+full_prompt = f"以下は{today}のGitea操作ログです。重要な変更点と傾向を日本語300字以内でまとめてください:\n{prompt}"
+payload = {'model': 'qwen2.5:7b', 'prompt': full_prompt, 'stream': False}
+open('/tmp/gen_payload.json','w').write(json.dumps(payload))

scripts/extract_texts.py

@@ -0,0 +1,4 @@
+import json
+d = json.load(open('/tmp/scroll.json'))
+texts = '\n'.join(p.get('payload',{}).get('content','') for p in d.get('result',{}).get('points',[]))
+open('/tmp/texts.txt','w').write(texts)

scripts/record_event.py

@@ -0,0 +1,25 @@
+import json, uuid, sys, os
+
+content = open('/tmp/content.txt').read()
+embed_resp = json.load(open('/tmp/embed_response.json'))
+vector = embed_resp['embeddings'][0]
+point_id = str(uuid.uuid4())
+
+payload = {
+    "points": [{
+        "id": point_id,
+        "vector": vector,
+        "payload": {
+            "run_id": os.environ.get('RUN_ID', ''),
+            "event": os.environ.get('EVENT', ''),
+            "repository": os.environ.get('REPO', ''),
+            "branch": os.environ.get('BRANCH', ''),
+            "actor": os.environ.get('ACTOR', ''),
+            "timestamp": os.environ.get('TIMESTAMP', ''),
+            "content": content
+        }
+    }]
+}
+with open('/tmp/payload.json', 'w') as f:
+    json.dump(payload, f)
+print(f"Point ID: {point_id}")