in movie_search/final_ver/backend/scene_search.py [0:0]
def search_scene(query: str, top_n: int = 1, model: GenerativeModel = model_flash) -> List[dict]:
"""シーン検索を実行する
Args:
query: 検索クエリ
top_n: 検索対象とする動画の数
model: 利用する Gemini モデル
Returns:
検索結果のリスト
"""
response = search_documents_by_query(query, show_summary=False)
storage_client = storage.Client()
results = []
for doc_id in range(min(top_n, len(response.results))):
# Discovery Engine の検索結果から、動画メタデータの URI とタイトルを取得
meta_uri = response.results[doc_id].document.derived_struct_data['link']
title = response.results[doc_id].document.derived_struct_data['title']
print(f'meta_uri: {meta_uri}')
# URI からバケット名と blob 名を取得
bucket_name = meta_uri.split("//")[1].split("/", 1)[0]
metadata_blob_name = meta_uri.split("//")[1].split("/", 1)[1]
movie_blob_name = metadata_url_to_movie_blob_name(meta_uri)
# Cloud Storage からメタデータを取得
bucket = storage_client.bucket(bucket_name)
blob = bucket.blob(metadata_blob_name)
# download_to_filename を使わずに、blob から直接テキストデータを読み込む
metatext = blob.download_as_text()
prompt = PROMPT_CONTENT_SEARCH.format(query=query, metatext=metatext)
temperature = 0.4
result = None
while temperature < 1.0:
try:
print(f'movie_blob_name: {movie_blob_name}')
signed_url = generate_download_signed_url_v4(bucket_name, movie_blob_name)
# generate_text から直接結果リストを取得
result = generate_text(prompt, model=model, temperature=temperature)
# 結果に signed_url と title を追加
for r in result:
r['signed_url'] = signed_url
r['title'] = title
results.extend(result)
break
except Exception as e:
print(e)
temperature += 0.05
if temperature < 1.0:
print('\n=====')
return results