scripts/process_remaining.py

import requests, os, time

backend = "http://localhost:8000"
uploads = os.path.join(os.path.dirname(__file__), "..", "uploads")
reports = os.path.join(os.path.dirname(__file__), "..", "reports")
os.makedirs(reports, exist_ok=True)

for i in range(14, 21):
    paper = f"paper_{i:02d}.pdf"
    path = os.path.join(uploads, paper)
    report = os.path.join(reports, f"{paper[:-4]}_report.pdf")

    print(f"{paper}: uploading...")
    with open(path, "rb") as f:
        r = requests.post(
            f"{backend}/analyze", files={"file": (paper, f, "application/pdf")}
        )
    if r.ok:
        job_id = r.json()["job_id"]
        print(f"  job: {job_id}")
        for _ in range(90):
            time.sleep(5)
            j = requests.get(f"{backend}/jobs/{job_id}").json()
            status = j.get("status", "")
            print(f"  status: {status}")
            if status == "completed":
                r2 = requests.get(f"{backend}/reports/{job_id}/pdf", stream=True)
                with open(report, "wb") as out:
                    for chunk in r2.iter_content():
                        out.write(chunk)
                print(f"  saved: {report}")
                break
            elif status in ["failed"]:
                break
    print()