pdf_data / src /app.py
Ai-Bharti's picture
Upload app.py
9a3a58b verified
Raw History Blame Contribute Delete
8.3 kB
import gradio as gr
import pandas as pd
from datetime import datetime
import time
import os
from scraper import get_pdf_urls_from_screener
from pdf_reader import extract_text_from_bse_pdf, extract_key_info
# ============================================================
# HuggingFace Inference API use karenge (FREE!)
# ============================================================
import requests as req
HF_API_URL = "https://api-inference.huggingface.co/models/facebook/bart-large-cnn"
def summarize_via_hf_api(text, hf_token):
"""HuggingFace ke FREE API se summary banata hai."""
if not hf_token or hf_token.strip() == "":
return "⚠️ HuggingFace token dalo pehle!"
short_text = " ".join(text.split()[:700])
if len(short_text) < 50:
return "Text bahut chhota hai - summary nahi ban sakti"
headers = {"Authorization": f"Bearer {hf_token}"}
payload = {
"inputs": short_text,
"parameters": {"max_length": 130, "min_length": 40, "do_sample": False}
}
try:
response = req.post(HF_API_URL, headers=headers, json=payload, timeout=30)
if response.status_code == 503:
time.sleep(10)
response = req.post(HF_API_URL, headers=headers, json=payload, timeout=30)
result = response.json()
if isinstance(result, list) and result:
return result[0].get("summary_text", "Summary nahi aayi")
elif isinstance(result, dict) and "error" in result:
return f"API Error: {result['error']}"
return str(result)
except Exception as e:
return f"❌ API call failed: {str(e)}"
# ============================================================
# MAIN FUNCTION
# ============================================================
def run_scraper(page_number, hf_token, max_pdfs, session_id, csrf_token):
"""
Poora pipeline: Screener β†’ PDF URLs β†’ Text β†’ AI Summary β†’ Table
session_id / csrf_token:
- Agar UI mein dala hai β†’ woh use hoga
- Agar khali chhoda β†’ HuggingFace Secrets se automatically aayega
"""
logs = []
def log(msg):
logs.append(f"[{datetime.now().strftime('%H:%M:%S')}] {msg}")
return "\n".join(logs)
# --- Cookies decide karo ---
# Priority: UI input > HuggingFace Secrets > kuch nahi (guest mode)
cookies = {}
if session_id and session_id.strip():
# User ne UI mein dala
cookies = {
"sessionid": session_id.strip(),
"csrftoken": csrf_token.strip() if csrf_token else ""
}
current_log = log("πŸ”‘ UI se cookies use ho rahi hain")
elif os.environ.get("SCREENER_SESSION_ID"):
# HuggingFace Secrets se auto-load
cookies = {
"sessionid": os.environ.get("SCREENER_SESSION_ID", ""),
"csrftoken": os.environ.get("SCREENER_CSRF_TOKEN", "")
}
current_log = log("πŸ”’ HuggingFace Secrets se cookies load hui!")
else:
current_log = log("⚠️ Cookies nahi mili - guest mode mein chalega")
yield current_log, None, None
# Step 1 - Screener se PDF links nikalo
current_log = log(f"πŸ”„ Screener.in page {int(page_number)} se links nikal raha hoon...")
yield current_log, None, None
pdf_data = get_pdf_urls_from_screener(page_num=int(page_number), cookies=cookies if cookies else None)
if not pdf_data:
current_log = log("❌ Koi PDF links nahi mile!")
current_log = log("πŸ’‘ Tip: Cookies check karo ya Screener pe login karo")
yield current_log, None, None
return
total = min(len(pdf_data), int(max_pdfs))
current_log = log(f"βœ… {len(pdf_data)} PDF links mile! Pehle {total} process karunga.")
yield current_log, None, None
# Step 2 - Har PDF process karo
results = []
for i, (company, pdf_url) in enumerate(pdf_data[:total]):
current_log = log(f"πŸ“„ [{i+1}/{total}] {pdf_url[:65]}...")
yield current_log, None, None
text = extract_text_from_bse_pdf(pdf_url)
if "❌" in text:
current_log = log(" ⚠️ Skip - PDF read nahi hua")
yield current_log, None, None
continue
key_info = extract_key_info(text)
current_log = log(" πŸ€– AI Summary ban raha hai...")
yield current_log, None, None
summary = summarize_via_hf_api(text, hf_token)
results.append({
"πŸ“… Date": datetime.now().strftime("%Y-%m-%d"),
"🏒 Company": company[:40],
"πŸ“‹ Order Type": key_info.get("order_type", "Order"),
"πŸ’° Amounts": ", ".join(key_info.get("amounts_found", []))[:60],
"πŸ€– AI Summary": summary,
"πŸ”— PDF URL": pdf_url
})
current_log = log(" βœ… Done!")
yield current_log, None, None
time.sleep(1)
# Step 3 - Results ready
if not results:
current_log = log("πŸ˜” Koi results nahi aayi!")
yield current_log, None, None
return
df = pd.DataFrame(results)
csv_path = f"/tmp/orders_{datetime.now().strftime('%Y%m%d_%H%M%S')}.csv"
df.to_csv(csv_path, index=False)
current_log = log(f"πŸŽ‰ COMPLETE! {len(results)} orders process hue!")
current_log = log("πŸ’Ύ CSV ready for download!")
yield current_log, df, csv_path
# ============================================================
# GRADIO UI
# ============================================================
with gr.Blocks(title="πŸ“ˆ BSE Order Tracker", theme=gr.themes.Soft()) as app:
gr.Markdown("""
# πŸ“ˆ BSE Order Tracker
### Screener.in se orders dhundo β†’ PDF padhho β†’ AI Summary pao!
""")
# --- Row 1: HF Token + Page + Count ---
with gr.Row():
hf_token_input = gr.Textbox(
label="πŸ”‘ HuggingFace API Token",
placeholder="hf_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
type="password",
scale=3,
info="huggingface.co β†’ Settings β†’ Access Tokens β†’ New Token (Read)"
)
page_input = gr.Slider(
label="πŸ“„ Screener Page", minimum=1, maximum=50, value=1, step=1, scale=1
)
max_pdf_input = gr.Slider(
label="πŸ“Š Kitne PDFs?", minimum=1, maximum=20, value=5, step=1, scale=1
)
# --- Row 2: Screener Cookies (Optional - agar Secrets set hain toh khali chhodo) ---
with gr.Accordion("πŸͺ Screener Cookies (Optional - HuggingFace Secrets use karo)", open=False):
gr.Markdown("""
> **Yeh tab use karo jab:** Aap locally test kar rahe ho ya Secrets set nahi kiye hain.
>
> **Recommended:** HuggingFace Space β†’ **Settings β†’ Secrets** mein daalo:
> - `SCREENER_SESSION_ID`
> - `SCREENER_CSRF_TOKEN`
>
> Tab yeh boxes khali chhod sakte ho - automatically load ho jaayega! πŸ”’
""")
with gr.Row():
session_id_input = gr.Textbox(
label="πŸͺ sessionid",
placeholder="Screener F12 β†’ Application β†’ Cookies β†’ sessionid",
type="password"
)
csrf_token_input = gr.Textbox(
label="πŸͺ csrftoken",
placeholder="Screener F12 β†’ Application β†’ Cookies β†’ csrftoken",
type="password"
)
run_btn = gr.Button("πŸš€ Scraping Shuru Karo!", variant="primary", size="lg")
gr.Markdown("---")
status_box = gr.Textbox(
label="πŸ“Š Live Status & Logs", lines=12,
placeholder="Button dabao - yahan sab dikhega step by step...",
interactive=False
)
results_table = gr.Dataframe(label="πŸ“‹ Orders Results Table", wrap=True, interactive=False)
download_btn = gr.File(label="⬇️ CSV Download Karo")
run_btn.click(
fn=run_scraper,
inputs=[page_input, hf_token_input, max_pdf_input,
session_id_input, csrf_token_input],
outputs=[status_box, results_table, download_btn]
)
gr.Markdown("""
---
> πŸ› οΈ **Built with:** Python + Gradio + HuggingFace BART |
> πŸ“Š **Data:** Screener.in + BSE India |
> πŸ”’ **Cookies:** HuggingFace Secrets se secure tarike se load hoti hain
""")
app.launch()