Spaces:
Sleeping
Sleeping
Upload app.py with huggingface_hub
Browse files
app.py
ADDED
|
@@ -0,0 +1,122 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import pipeline
|
| 3 |
+
import re
|
| 4 |
+
|
| 5 |
+
# Load your fine-tuned model
|
| 6 |
+
try:
|
| 7 |
+
ner_pipeline = pipeline("ner",
|
| 8 |
+
model="MuneebAbro/ecommerce-ner-model",
|
| 9 |
+
aggregation_strategy="simple")
|
| 10 |
+
model_loaded = True
|
| 11 |
+
except:
|
| 12 |
+
# Fallback to a general NER model if yours isn't ready
|
| 13 |
+
ner_pipeline = pipeline("ner",
|
| 14 |
+
model="dbmdz/bert-large-cased-finetuned-conll03-english",
|
| 15 |
+
aggregation_strategy="simple")
|
| 16 |
+
model_loaded = False
|
| 17 |
+
|
| 18 |
+
def extract_product_info(text):
|
| 19 |
+
"""Extract product information and format results"""
|
| 20 |
+
if not text.strip():
|
| 21 |
+
return "Please enter some text!"
|
| 22 |
+
|
| 23 |
+
try:
|
| 24 |
+
# Enhanced regex patterns for better extraction
|
| 25 |
+
result = {
|
| 26 |
+
"product_name": "",
|
| 27 |
+
"brand": "",
|
| 28 |
+
"price": "",
|
| 29 |
+
"quantities": []
|
| 30 |
+
}
|
| 31 |
+
|
| 32 |
+
# Extract brand
|
| 33 |
+
brand_patterns = [
|
| 34 |
+
r'\b(Samsung|Apple|iPhone|Google|OnePlus|Xiaomi|Huawei|Sony|LG|Dell|HP|Lenovo|Microsoft|Nintendo)\b'
|
| 35 |
+
]
|
| 36 |
+
for pattern in brand_patterns:
|
| 37 |
+
match = re.search(pattern, text, re.IGNORECASE)
|
| 38 |
+
if match:
|
| 39 |
+
result["brand"] = match.group(1)
|
| 40 |
+
break
|
| 41 |
+
|
| 42 |
+
# Extract product name
|
| 43 |
+
product_patterns = [
|
| 44 |
+
r'(Galaxy\s+\w+(?:\s+\w+)?)',
|
| 45 |
+
r'(iPhone\s+\d+(?:\s+\w+)?)',
|
| 46 |
+
r'(Pixel\s+\d+(?:\s+\w+)?)',
|
| 47 |
+
r'(\w+\s+\d+(?:\s+\w+)?)'
|
| 48 |
+
]
|
| 49 |
+
for pattern in product_patterns:
|
| 50 |
+
match = re.search(pattern, text)
|
| 51 |
+
if match:
|
| 52 |
+
result["product_name"] = match.group(1).strip()
|
| 53 |
+
break
|
| 54 |
+
|
| 55 |
+
# Extract price
|
| 56 |
+
price_patterns = [
|
| 57 |
+
r'Price\s*:?\s*\$(\d+(?:,\d{3})*(?:\.\d{2})?)',
|
| 58 |
+
r'\$(\d+(?:,\d{3})*(?:\.\d{2})?)'
|
| 59 |
+
]
|
| 60 |
+
for pattern in price_patterns:
|
| 61 |
+
match = re.search(pattern, text, re.IGNORECASE)
|
| 62 |
+
if match:
|
| 63 |
+
result["price"] = f"${match.group(1)}"
|
| 64 |
+
break
|
| 65 |
+
|
| 66 |
+
# Extract quantities
|
| 67 |
+
quantity_matches = re.findall(r'(\d+(?:GB|TB|MB|RAM|Storage))', text, re.IGNORECASE)
|
| 68 |
+
result["quantities"] = [q for q in quantity_matches if q]
|
| 69 |
+
|
| 70 |
+
# Try NER model too
|
| 71 |
+
if model_loaded:
|
| 72 |
+
ner_results = ner_pipeline(text)
|
| 73 |
+
ner_info = "\nπ€ **NER Model Results:**\n"
|
| 74 |
+
for entity in ner_results:
|
| 75 |
+
if entity.get('score', 0) > 0.3:
|
| 76 |
+
ner_info += f"- {entity['word']} β {entity['entity_group']} ({entity['score']:.2f})\n"
|
| 77 |
+
else:
|
| 78 |
+
ner_info = "\nπ€ **NER Model:** Using fallback model\n"
|
| 79 |
+
|
| 80 |
+
# Format output
|
| 81 |
+
output = f"""
|
| 82 |
+
π·οΈ **Brand:** {result['brand'] or 'Not found'}
|
| 83 |
+
π± **Product:** {result['product_name'] or 'Not found'}
|
| 84 |
+
π° **Price:** {result['price'] or 'Not found'}
|
| 85 |
+
π **Quantities:** {', '.join(result['quantities']) or 'Not found'}
|
| 86 |
+
|
| 87 |
+
{ner_info}
|
| 88 |
+
"""
|
| 89 |
+
|
| 90 |
+
return output
|
| 91 |
+
|
| 92 |
+
except Exception as e:
|
| 93 |
+
return f"β Error: {str(e)}"
|
| 94 |
+
|
| 95 |
+
# Create interface
|
| 96 |
+
demo = gr.Interface(
|
| 97 |
+
fn=extract_product_info,
|
| 98 |
+
inputs=gr.Textbox(
|
| 99 |
+
label="Product Description",
|
| 100 |
+
placeholder="Enter product description here...",
|
| 101 |
+
lines=3,
|
| 102 |
+
value="Samsung Galaxy S23 Ultra, 12GB RAM, 256GB Storage, Price: $1199."
|
| 103 |
+
),
|
| 104 |
+
outputs=gr.Textbox(
|
| 105 |
+
label="Extracted Information",
|
| 106 |
+
lines=12
|
| 107 |
+
),
|
| 108 |
+
title="π E-commerce Product Information Extractor",
|
| 109 |
+
description="Extract product names, brands, prices, and quantities from product descriptions using fine-tuned NER model.",
|
| 110 |
+
examples=[
|
| 111 |
+
["Samsung Galaxy S23 Ultra, 12GB RAM, 256GB Storage, Price: $1199."],
|
| 112 |
+
["iPhone 14 Pro Max 256GB $1299 Apple"],
|
| 113 |
+
["Google Pixel 7 Pro 12GB RAM $899"],
|
| 114 |
+
["Sony WH-1000XM4 wireless headphones $349"],
|
| 115 |
+
["Dell XPS 13 laptop 16GB DDR4 $1299"]
|
| 116 |
+
],
|
| 117 |
+
theme="default",
|
| 118 |
+
allow_flagging="never"
|
| 119 |
+
)
|
| 120 |
+
|
| 121 |
+
if __name__ == "__main__":
|
| 122 |
+
demo.launch()
|