Spaces:
Sleeping
Sleeping
File size: 4,633 Bytes
145c622 12ff613 145c622 cbd7d88 145c622 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 |
# Running on local URL: http://127.0.0.1:7860
import gradio as gr # Gradio: for creating web-based user interfaces
import PyPDF2 # PyPDF2: for reading PDF files
import tempfile # tempfile: to safely handle temporary files
from langchain.prompts import PromptTemplate # LangChain: for managing prompt templates
from langchain_huggingface.llms import HuggingFacePipeline # LangChain integration with HuggingFace models
# Define a summarization class
class TextSummarizer:
def __init__(self):
# Define the model to use for summarization
self.model_id = "facebook/bart-large-cnn"
def summarize_text(self, article_text, max_length=150, min_length=30):
# Load a summarization pipeline with custom length settings
llm = HuggingFacePipeline.from_model_id(
model_id=self.model_id,
task="summarization",
pipeline_kwargs={
#"max_length": 250,
#"do_sample": True,
#"temperature": 0.7, # More creative
#"top_k": 50, # Limit to top 50 tokens
#"top_p": 0.95 # Use nucleus sampling
"max_length": max_length,
"min_length": min_length,
"do_sample": False # Deterministic output
}
)
# Create a basic prompt template that just passes the text
prompt = PromptTemplate(input_variables=["document"], template="""{document}""")
# Format the article text into the prompt
prompt_input = prompt.format(document=article_text)
# Generate the summary using the model
summary = llm.__call__(prompt_input)
# If the model returns a list of summaries, extract the actual summary text
if isinstance(summary, list):
return summary[0]['summary_text'] if 'summary_text' in summary[0] else str(summary[0])
return str(summary) # Fallback for other formats
# Function to extract text from an uploaded PDF
def pdf_to_text(pdf_file):
try:
# Create a temporary file to write the uploaded PDF bytes
with tempfile.NamedTemporaryFile(delete=False) as tmp:
tmp.write(pdf_file) # Write raw bytes directly
tmp.flush() # Make sure data is written to disk
# Use PyPDF2 to read and extract text
reader = PyPDF2.PdfReader(tmp.name)
text = "\n".join(page.extract_text() or "" for page in reader.pages)
# Return cleaned-up text or a message if extraction fails
return text.strip() if text.strip() else "No extractable text found in the PDF."
except Exception as e:
return f"Error reading PDF: {str(e)}" # Return readable error message
# Instantiate the summarizer class
summarizer = TextSummarizer()
# Summarize input with user-defined maximum length
def summarize_input(text, max_words):
if not text.strip():
return "Please enter or extract some text first."
try:
# Convert max_words input to integer
max_length = int(max_words)
# Set a safe minimum length for quality summaries
min_length = max(30, max_length // 4)
# Generate the summary
return summarizer.summarize_text(text, max_length=max_length, min_length=min_length)
except Exception as e:
return f"Error during summarization: {str(e)}"
# Build the Gradio UI
with gr.Blocks() as demo:
gr.Markdown("## 📝 Text & PDF Summarizer with Length Control")
with gr.Row():
# Text input for manually entering article
text_input = gr.Textbox(label="Enter article text", lines=15, placeholder="Paste your article here...")
# Upload input for PDF files
pdf_file = gr.File(label="Or upload PDF", file_types=[".pdf"], type="binary")
# User input for controlling max summary length
max_words = gr.Number(label="Max summary word count", value=150, precision=0)
with gr.Row():
# Button to convert PDF to text
convert_btn = gr.Button("Convert PDF to Text")
# Button to generate the summary
summary_btn = gr.Button("Summarize Text")
# Textbox to display the summary output
output_text = gr.Textbox(label="Summary", lines=10)
# Link buttons to their respective functions
convert_btn.click(fn=pdf_to_text, inputs=pdf_file, outputs=text_input)
summary_btn.click(fn=summarize_input, inputs=[text_input, max_words], outputs=output_text)
# Launch the app if run directly
if __name__ == "__main__":
demo.launch()
|