Download README.md from Menixis/auto-form: direct link, hf CLI and curl.
- Browser
- Download file 4.98 kB
-
https://huggingface.co/spaces/Menixis/auto-form/resolve/main/README.md
- Command line
-
hf download hf://spaces/Menixis/auto-form/README.md
-
curl -L -o README.md https://huggingface.co/spaces/Menixis/auto-form/resolve/main/README.md
title: Auto Form Generator
emoji: 📄
colorFrom: blue
colorTo: indigo
sdk: docker
app_port: 7860
pinned: false
short_description: Auto-generate procurement letters from Excel data
Auto Form Generator
ระบบสร้างหนังสือราชการอัตโนมัติจากแบบฟอร์มมาตรฐาน 5 ประเภท พร้อมดึงข้อมูล PO อัตโนมัติจากไฟล์ Excel
✨ Features
- 📄 รองรับ 5 แบบฟอร์มมาตรฐาน
- 🔍 Auto-fill จาก Excel — ใส่ PO แล้วระบบเติม Vendor, ชื่องาน ฯลฯ ให้อัตโนมัติ
- ✏️ รักษารูปแบบเอกสารต้นฉบับ (Font, สี, layout)
- 🌐 รันบนเว็บได้เลย ไม่ต้องติดตั้งอะไร
- 🚀 ทำงานบน Hugging Face Spaces ฟรี
📦 Tech Stack
- Python 3.11 + Streamlit (web UI)
- python-docx (สร้างเอกสาร Word)
- openpyxl (อ่านไฟล์ Excel)
- Docker (deployment on Hugging Face Spaces)
🚀 Run Locally
pip install -r requirements.txt
streamlit run streamlit_app.py
เปิดเบราว์เซอร์ที่ http://localhost:8501
🔐 Authentication (Microsoft Entra ID)
The app is gated behind Microsoft sign-in using Streamlit's native OIDC auth
(st.login). It reuses an existing Azure App Registration.
Local: copy .streamlit/secrets.toml.template to .streamlit/secrets.toml
and fill in the values (this file is gitignored). Register
http://localhost:8501/oauth2callback as a Web redirect URI in Azure.
Hugging Face Spaces: secrets.toml is built at startup by entrypoint.sh
from these Space secrets — set them in Settings → Variables and secrets:
| Secret | Value |
|---|---|
AUTH_CLIENT_ID |
Azure app (client) id |
AUTH_CLIENT_SECRET |
Azure client secret |
AUTH_COOKIE_SECRET |
any long random hex string |
AUTH_REDIRECT_URI |
https://<space-host>/oauth2callback |
AUTH_SERVER_METADATA_URL |
https://login.microsoftonline.com/<tenant>/v2.0/.well-known/openid-configuration |
DATA_URL |
OneDrive "anyone with the link" URL of the PO-data Excel (Power BI export) |
SMP_TOKEN |
no-reply service-account bearer for the signer lookup (see below) |
🔏 Signer auto-fill (ผู้มีอำนาจอนุมัติ / ประธานกรรมการตรวจรับ)
When the user picks a ผู้ลงนาม role, the app resolves the real name live from the PTT SMP
system — in the user's browser, not on the server. The HF server can't reach the
PTT-internal gateway, but the user's browser (on PTT wifi) can, and the gateway sends
Access-Control-Allow-Origin: *, so the cross-origin fetch is allowed. signer_lookup.py
builds the client-side fetch (PO → SAP PR → approver + receiving-committee chairman);
streamlit-javascript runs it and returns the names.
- Set the
SMP_TOKENSpace secret to the no-reply bearer (mint/refresh it withsmp_api\harvest_token.py; the value is thebearerfield ofte_session_tsopmm.json, with or without theBearerprefix). It is read-only and only works on the PTT network. ⚠️ This token is injected into client-side JS, so it is visible to signed-in users — it's a service account (not a personal login), gated behind Entra sign-in. Rotate it periodically (it expires ~166 days after minting). - Requires the user to be on PTT wifi at generation time. Off-network (or if the token expired), the app falls back to a manual ชื่อผู้ลงนาม text box automatically.
- Local testing: set
SMP_TOKENas an env var andpip install streamlit-javascript.
Register the same https://<space-host>/oauth2callback as a Web redirect
URI in Azure. Sign-in stays broken until both the Space secrets and the Azure
redirect URI are in place.
📋 Project Structure
.
├── Dockerfile # HF Spaces container config
├── README.md # This file (with HF YAML frontmatter)
├── requirements.txt # Python dependencies
├── streamlit_app.py # Web UI (main entry point)
├── form_logic.py # Document generation logic
├── excel_lookup.py # Excel data lookup helper (legacy, unused by web app)
├── sample_po_data.xlsx # Sample data for demo
├── .streamlit/
│ └── config.toml # Theme & toolbar config
└── templates/
└── form1-5.docx # Word templates
📄 License
MIT