Spaces:
Sleeping
Sleeping
File size: 5,850 Bytes
7dec78f e92fbe1 7dec78f 17d36dc b8d5f22 6544d14 b8d5f22 4f7269c b8d5f22 17d36dc b8d5f22 6544d14 0a3a380 6544d14 b8d5f22 0a3a380 b8d5f22 7dec78f 7c51401 7dec78f 5e94ef1 17d36dc |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 |
import json
import gradio as gr
from pdfminer.high_level import extract_pages
from pdfminer.layout import LTTextBoxHorizontal, LTFigure, LTImage
import os
import io
from PIL import Image
import pandas as pd
import pdfplumber
def parse_pdf(pdf_file, output_format, progress=gr.Progress()):
"""
Parses a PDF file, extracts text, tables, and images, and formats the output.
Args:
pdf_file: Path to the uploaded PDF file.
output_format: Desired output format ("JSON", "Markdown", or "HTML").
progress: Gradio Progress object for displaying progress.
Returns:
tuple: Extracted text and download data in the specified format.
Returns an empty string and None if there is an error.
"""
try:
with open(pdf_file, 'rb') as file:
pages = list(extract_pages(file)) # Convert generator to list
text = ""
tables = []
images = []
# Iterate through pages and extract text and images
for i, page in enumerate(pages):
progress(i / len(pages)) # Update progress bar
for element in page:
if isinstance(element, LTTextBoxHorizontal):
text += element.get_text()
elif isinstance(element, (LTFigure, LTImage)):
try:
if hasattr(element, 'stream'):
image_data = element.stream.read()
image = Image.open(io.BytesIO(image_data))
image_filename = f"extracted_image_{len(images)}.png"
image.save(image_filename)
images.append({"filename": image_filename})
else:
for child in element:
if isinstance(child, LTImage):
image_data = child.stream.read()
image = Image.open(io.BytesIO(image_data))
image_filename = f"extracted_image_{len(images)}.png"
image.save(image_filename)
images.append({"filename": image_filename})
except Exception as e:
print(f"Error extracting image: {e}")
# Enhanced table extraction using pdfplumber
with pdfplumber.open(pdf_file) as pdf:
for page_num, page in enumerate(pdf.pages):
for table in page.extract_tables():
# Handle potential duplicate columns
if len(table) > 0 and len(set(table[0])) != len(table[0]):
# If duplicate columns exist, try to create unique column names
unique_columns = []
for col in table[0]:
if col in unique_columns:
col = f"{col}_{unique_columns.count(col)}" # Append a counter
unique_columns.append(col)
df = pd.DataFrame(table[1:], columns=unique_columns)
else:
df = pd.DataFrame(table[1:], columns=table[0] if table[0] else None)
tables.append(df)
# Format extracted data based on user selection
if output_format == "JSON":
json_data = {
"text": text,
"tables": [table.to_dict(orient='records') for table in tables], # Use 'records' for better handling of duplicate columns
"images": images
}
download_data = json.dumps(json_data, indent=4) # Add indentation for readability
elif output_format == "Markdown":
markdown_text = f"# Extracted Text\n\n{text}\n\n# Tables\n"
for i, table in enumerate(tables):
markdown_text += f"## Table {i+1}\n"
markdown_text += table.to_markdown(index=False) + "\n\n"
# Image embedding in Markdown (using relative paths)
markdown_text += "\n\n# Images\n\n"
for image in images:
image_path = os.path.join(os.getcwd(), image["filename"])
markdown_text += f'\n'
download_data = markdown_text
elif output_format == "HTML":
html_text = f"<p>{text}</p>\n\n<h2>Tables</h2>\n"
for i, table in enumerate(tables):
html_text += f"<h2>Table {i+1}</h2>\n"
html_text += table.to_html() + "<br>"
# Image embedding in HTML (using relative paths)
html_text += "\n\n<h2>Images</h2>\n\n"
for image in images:
image_path = os.path.join(os.getcwd(), image["filename"])
html_text += f'<img src="{image_path}" alt="Image"><br>\n'
download_data = html_text.encode("utf-8") # Encode for HTML download
return text, download_data
except Exception as main_e:
print(f"A main error occurred: {main_e}")
return "", None # Return empty string and None in case of error
iface = gr.Interface(
fn=parse_pdf,
inputs=["file", gr.Dropdown(["JSON", "Markdown", "HTML"])], # Remove gr.Progress() from inputs
outputs=[
gr.Text(label="Output Text"),
gr.File(label="Download Output")
],
title="PDF Parser",
description="Parse a PDF and choose the output format."
)
if __name__ == "__main__":
iface.launch(share=False) |