エンドポイント別AIプロンプト

これらのプロンプトをコピーして、AIエージェントに個別のPROOF APIエンドポイントを呼び出させてください。

ファイルをアップロード

Call the PROOF API to upload a file for table extraction.

POST https://your-domain.com/api/v1/jobs/upload
Headers: X-API-Key: <API_KEY>
Body (multipart/form-data):
  - files: <the file to upload>
  - ocr_model: premium
  - columns: <comma-separated column names, or omit for defaults>
  - prompt: <extra instructions, or omit>

Return the job_id from the response.

ジョブを確認

Call the PROOF API to confirm a job and start processing.

POST https://your-domain.com/api/v1/jobs/<JOB_ID>/confirm
Headers: X-API-Key: <API_KEY>

Return the status and remaining credits.

ジョブステータスを確認

Call the PROOF API to check the status of a job.

GET https://your-domain.com/api/v1/jobs/<JOB_ID>
Headers: X-API-Key: <API_KEY>

If status is "processing", report progress as done_pages/total_pages.
If status is "done", the job is complete.
If status is "error", report the error_log.

結果をダウンロード

Call the PROOF API to get the download URL for a completed job.

GET https://your-domain.com/api/v1/jobs/<JOB_ID>/download?type=xlsx_basic
Headers: X-API-Key: <API_KEY>

Return the download_url from the response.

Office形式に変換

Call the PROOF API to convert a completed job to DOCX.

POST https://your-domain.com/api/v1/jobs/<JOB_ID>/convert
Headers:
  X-API-Key: <API_KEY>
  Content-Type: application/json
Body: {"format": "docx"}

Return the download_url from the response.

アカウント残高を確認

Call the PROOF API to check the account info and point balance.

GET https://your-domain.com/api/v1/account
Headers: X-API-Key: <API_KEY>

Report the credit balance and today's usage.

Python自動化スニペット

import requests
import time

API_KEY = "chu_live_xxxxxxxx"
BASE = "https://your-domain.com/api/v1"
HEADERS = {"X-API-Key": API_KEY}

def process_file(filepath, columns=None, ocr_model="premium", output_type="xlsx_basic"):
    """フルパイプライン: アップロード → 確認 → ポーリング → ダウンロード."""
    # ステップ1: アップロード
    with open(filepath, "rb") as f:
        data = {"ocr_model": ocr_model}
        if columns:
            data["columns"] = columns
        resp = requests.post(f"{BASE}/jobs/upload", headers=HEADERS,
                             files={"files": f}, data=data)
    resp.raise_for_status()
    job_id = resp.json()["job_id"]

    # ステップ2: 確認
    resp = requests.post(f"{BASE}/jobs/{job_id}/confirm", headers=HEADERS)
    resp.raise_for_status()

    # ステップ3: ポーリング
    while True:
        status = requests.get(f"{BASE}/jobs/{job_id}", headers=HEADERS).json()
        if status["status"] in ("done", "error"):
            break
        time.sleep(3)

    if status["status"] == "error":
        raise RuntimeError(f"ジョブ失敗: {status.get('error_log')}")

    # ステップ4: ダウンロード
    url = requests.get(f"{BASE}/jobs/{job_id}/download",
                       params={"type": output_type}, headers=HEADERS).json()
    return url["download_url"]