commit

app.py

@@ -1,7 +1,7 @@
 import os
 import re
 from dataclasses import dataclass
-from typing import Dict, Iterable, List, Optional, Tuple
+from typing import Dict, Iterable, List, Optional
 
 import gradio as gr
 import requests
@@ -20,7 +20,7 @@ class StorageHit:
     snippet: Optional[str]
 
 
-VECTOR_STORE_ID_RE = re.compile(r"(vs_[a-zA-Z0-9]{8,})")
+VECTOR_STORE_ID_RE = re.compile(r"vs_[a-zA-Z0-9]{8,}")
 
 
 class OpenAIStorageClient:
@@ -113,19 +113,19 @@ class OpenAIStorageClient:
                 attempts += 1
                 if attempts > retries:
                     raise
+def get_vector_store_ids_from_env() -> List[str]:
+    """Read VECTOR_STORE_ID env var and return validated vector store IDs."""
+    raw = os.getenv("VECTOR_STORE_ID", "").strip()
+    if not raw:
+        return []
 
-
-def extract_vector_store_context(question: str) -> Tuple[str, List[str]]:
-    """Extract vector store IDs and return the cleaned question text."""
     ids: List[str] = []
-    for match in VECTOR_STORE_ID_RE.finditer(question):
-        candidate = match.group(1)
-        if candidate not in ids:
-            ids.append(candidate)
-
-    cleaned = VECTOR_STORE_ID_RE.sub(" ", question)
-    cleaned = re.sub(r"\s+", " ", cleaned).strip()
-    return cleaned, ids
+    for token in re.split(r"[,\s]+", raw):
+        if not token:
+            continue
+        if VECTOR_STORE_ID_RE.fullmatch(token) and token not in ids:
+            ids.append(token)
+    return ids
 
 
 def generate_search_queries(question: str) -> List[str]:
@@ -184,17 +184,18 @@ def produce_answer(question: str, top_k: int) -> str:
 
     client = OpenAIStorageClient(api_key=api_key)
 
-    cleaned_question, vector_store_ids = extract_vector_store_context(question)
-    queries = generate_search_queries(cleaned_question or question)
+    vector_store_ids = get_vector_store_ids_from_env()
+    if not vector_store_ids:
+        return "VECTOR_STORE_ID が設定されていません。環境変数を確認してください。"
+
+    queries = generate_search_queries(question)
     if not queries:
         return "該当データがありません。"
 
     try:
         hits_by_query = []
         for query in queries:
-            filters = None
-            if vector_store_ids:
-                filters = {"vector_store_ids": vector_store_ids}
+            filters = {"vector_store_ids": vector_store_ids}
             hits = client.storage_search(query=query, top_k=top_k, filters=filters)
             if hits:
                 hits_by_query.append((query, hits))
@@ -250,8 +251,7 @@ with gr.Blocks() as demo:
         """
         ## MCP Storage レスポンダ
         OpenAI Storage (Files / Vector Store) に登録されたデータを検索し、根拠付きで回答します。
-        質問に `vs_xxxxx` の形式で Vector Store ID を含めると、そのストアを優先して検索します。
-        OPENAI_API_KEY を環境変数に設定してからご利用ください。
+        OPENAI_API_KEY と VECTOR_STORE_ID を環境変数に設定してからご利用ください。
         """
     )