Read invoices saved to Dropbox and post the details

import io, json
from pypdf import PdfReader
from clickruntime import on_trigger, report_fetched, state, notify, http, credentials, ai

SCHEMA = {"type": "object", "properties": {k: {"type": "string"} for k in ("vendor", "invoice_date", "total", "due_date")},
          "required": ["vendor", "invoice_date", "total", "due_date"]}

@on_trigger
def run(event):
    auth = {"Authorization": "Bearer " + credentials.connection("dropbox", "main")}
    cursor = state.get("cursor") or http.post("https://api.dropboxapi.com/2/files/list_folder/get_latest_cursor", headers=auth, json={"path": "/Invoices"}).json()["cursor"]
    page = http.post("https://api.dropboxapi.com/2/files/list_folder/continue", headers=auth, json={"cursor": cursor}).json()
    pdfs = [e for e in page["entries"] if e[".tag"] == "file" and e["name"].lower().endswith(".pdf")]
    report_fetched(len(pdfs))
    if not pdfs:
        return {"summary": "No new invoices"}
    for f in pdfs:
        raw = http.post("https://content.dropboxapi.com/2/files/download", headers={**auth, "Dropbox-API-Arg": json.dumps({"path": f["path_lower"]})}).content
        text = "\n".join(p.extract_text() or "" for p in PdfReader(io.BytesIO(raw)).pages)[:8000]
        x = ai.generate(f"Extract vendor, invoice date, total (with currency) and due date from this invoice.\n\n{text}", json_schema=SCHEMA, max_output_tokens=120).json
        notify.slack(f"Invoice {f['name']}\nVendor: {x['vendor']} · Date: {x['invoice_date']} · Total: {x['total']} · Due: {x['due_date']}", channel="#finance")
    state.set("cursor", page["cursor"])
    return {"invoices": len(pdfs)}