Spaces:
Sleeping
Sleeping
Upload 2 files
Browse files- app.py +59 -0
- requirements.txt +5 -0
app.py
ADDED
|
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import pipeline
|
| 3 |
+
import PyPDF2
|
| 4 |
+
from docx import Document
|
| 5 |
+
|
| 6 |
+
# Load pipelines
|
| 7 |
+
classifier = pipeline("zero-shot-classification", model="facebook/bart-large-mnli")
|
| 8 |
+
ner = pipeline("ner", model="Jean-Baptiste/roberta-large-ner-english", grouped_entities=True)
|
| 9 |
+
|
| 10 |
+
# File reading
|
| 11 |
+
def read_file(file_obj):
|
| 12 |
+
name = file_obj.name
|
| 13 |
+
if name.endswith(".txt"):
|
| 14 |
+
return file_obj.read().decode("utf-8")
|
| 15 |
+
elif name.endswith(".pdf"):
|
| 16 |
+
reader = PyPDF2.PdfReader(file_obj)
|
| 17 |
+
return " ".join([page.extract_text() for page in reader.pages if page.extract_text()])
|
| 18 |
+
elif name.endswith(".docx"):
|
| 19 |
+
doc = Document(file_obj)
|
| 20 |
+
return "\n".join([para.text for para in doc.paragraphs])
|
| 21 |
+
else:
|
| 22 |
+
return "Unsupported file format"
|
| 23 |
+
|
| 24 |
+
# Contract classification
|
| 25 |
+
def is_contract(text):
|
| 26 |
+
result = classifier(text[:1000], ["contract", "not a contract"])
|
| 27 |
+
return result['labels'][0] == 'contract', result
|
| 28 |
+
|
| 29 |
+
# Party extraction
|
| 30 |
+
def extract_parties(text):
|
| 31 |
+
entities = ner(text[:1000])
|
| 32 |
+
return list(set(ent['word'] for ent in entities if ent['entity_group'] in ['ORG', 'PER']))
|
| 33 |
+
|
| 34 |
+
# Main logic
|
| 35 |
+
def process_file(file):
|
| 36 |
+
text = read_file(file)
|
| 37 |
+
if not text.strip():
|
| 38 |
+
return "Empty or unreadable file.", None
|
| 39 |
+
|
| 40 |
+
is_contract_flag, classification = is_contract(text)
|
| 41 |
+
if is_contract_flag:
|
| 42 |
+
parties = extract_parties(text)
|
| 43 |
+
return "✅ This is a contract.", parties
|
| 44 |
+
else:
|
| 45 |
+
return "❌ This is NOT a contract.", []
|
| 46 |
+
|
| 47 |
+
# Gradio interface
|
| 48 |
+
iface = gr.Interface(
|
| 49 |
+
fn=process_file,
|
| 50 |
+
inputs=gr.File(file_types=[".txt", ".pdf", ".docx"], label="Upload a document"),
|
| 51 |
+
outputs=[
|
| 52 |
+
gr.Textbox(label="Classification Result"),
|
| 53 |
+
gr.Label(label="Detected Parties")
|
| 54 |
+
],
|
| 55 |
+
title="Contract Classifier with RoBERTa",
|
| 56 |
+
description="Upload a document (.pdf, .txt, .docx) to detect if it's a contract and extract involved parties using RoBERTa."
|
| 57 |
+
)
|
| 58 |
+
|
| 59 |
+
iface.launch()
|
requirements.txt
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
gradio
|
| 2 |
+
transformers
|
| 3 |
+
torch
|
| 4 |
+
python-docx
|
| 5 |
+
PyPDF2
|