Gemini API 文件搜索实现多模态 RAG：图文检索一体 + 精确页码溯源（开发者指南）

pip install -U google-genai

from google import genai
from google.genai import types
import time

client = genai.Client()

store = client.file_search_stores.create(
    config={
        "display_name": "product-catalog",
        "embedding_model": "models/gemini-embedding-2"
    }
)
print(f"Created store: {store.name}")

# upload_to_file_search_store 一步完成上传 + 索引
for file_name in ["catalog.pdf", "sneaker_red.png", "sneaker_blue.jpeg"]:
    op = client.file_search_stores.upload_to_file_search_store(
        file_search_store_name=store.name,
        file=file_name,
    )
    while not op.done:
        time.sleep(5)
        op = client.operations.get(op)

response = client.models.generate_content(
    model="gemini-3-flash-preview",
    contents="Which sneakers come in red?",
    config={
        "tools": [{
            "file_search": {
                "file_search_store_names": [store.name]
            }
        }]
    }
)

# 解析 grounding metadata
for chunk in response.grounding_metadata.grounding_chunks:
    ctx = chunk.retrieved_context
    if ctx.media_id:
        blob = client.file_search_stores.download_media(media_id=ctx.media_id)
        with open(f"cited_{ctx.title}.png", "wb") as f:
            f.write(blob)
        print(f"Downloaded cited image: {ctx.title}")
    elif ctx.page_number:
        print(f"Page {ctx.page_number}: {ctx.text[:200]}...")

response = client.models.generate_content(
    model="gemini-3-flash-preview",
    contents="Do you have blue spring shoes?",
    config={
        "tools": [{
            "file_search": {
                "file_search_store_names": [store.name],
                "metadata_filter": 'category="footwear" AND season="spring-2026"',
            }
        }]
    }
)

# 列出所有存储
for s in client.file_search_stores.list():
    print(f"{s.name} — {s.display_name}")

# 列出存储内文档
for doc in client.file_search_stores.documents.list(parent=store.name):
    print(f"  {doc.name}")

# 删除存储（force=True 同时删除所有文档）
client.file_search_stores.delete(name=store.name, config={"force": True})