Keep a Pinecone index in step with a Drive folder

import json
from google.oauth2.credentials import Credentials
from googleapiclient.discovery import build
from clickruntime import on_trigger, report_fetched, credentials, http, state

Q = "'your-folder-id' in parents and trashed=false and mimeType='application/vnd.google-apps.document'"
HOST, NS = "https://docs-abc123.svc.aped-1234.pinecone.io", "company"  # your index host and namespace
def upsert(records):
    http.post(f"{HOST}/records/namespaces/{NS}/upsert", data="\n".join(json.dumps(r) for r in records[:96]), headers={
        "Api-Key": credentials.secret("PINECONE_API_KEY"), "X-Pinecone-API-Version": "2025-04", "Content-Type": "application/x-ndjson"}).raise_for_status()

@on_trigger
def run(event):
    drive = build("drive", "v3", credentials=Credentials(credentials.connection("google", "work")), cache_discovery=False)
    files = {f["id"]: f for f in drive.files().list(q=Q, fields="files(id,name,modifiedTime)", pageSize=1000).execute()["files"]}
    known = state.get("known") or {}
    changed = [f for i, f in files.items() if known.get(i) != f["modifiedTime"]]
    removed = [i for i in known if i not in files]
    report_fetched(len(changed) + len(removed))
    if not changed and not removed:
        return {"changed": 0}
    pc = {"Api-Key": credentials.secret("PINECONE_API_KEY"), "X-Pinecone-API-Version": "2025-04"}
    for fid in removed + [f["id"] for f in changed if f["id"] in known]:
        http.post(f"{HOST}/vectors/delete", headers=pc, json={"namespace": NS, "filter": {"file_id": {"$eq": fid}}})
    for f in changed:
        text = drive.files().export(fileId=f["id"], mimeType="text/plain").execute().decode("utf-8", "ignore")
        upsert([{"_id": f"{f['id']}#{n}", "chunk_text": text[i:i + 1500], "file_id": f["id"], "file_name": f["name"]}
                for n, i in enumerate(range(0, len(text), 1500))])
    state.set("known", {i: f["modifiedTime"] for i, f in files.items()})
    return {"changed": len(changed), "removed": len(removed)}