Spaces:
Sleeping
Sleeping
refactor: Simplify main application entry point and remove unused interface functions
Browse files
app.py
CHANGED
|
@@ -1,172 +1,35 @@
|
|
| 1 |
"""
|
| 2 |
-
|
| 3 |
-
|
| 4 |
"""
|
| 5 |
|
| 6 |
-
import
|
| 7 |
-
from
|
| 8 |
-
from ui
|
| 9 |
-
from
|
| 10 |
-
create_header, create_upload_section, create_action_button,
|
| 11 |
-
create_text_display, create_action_buttons, create_image_gallery, apply_custom_css
|
| 12 |
-
)
|
| 13 |
-
from gradio_pdf import PDF # <-- Added import
|
| 14 |
|
| 15 |
-
def
|
| 16 |
-
"""
|
| 17 |
-
Create a simple interface for when the API key is not configured.
|
| 18 |
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
"""
|
| 22 |
-
with gr.Blocks(title="PDF Text Extractor") as interface:
|
| 23 |
-
gr.Markdown("""
|
| 24 |
-
# 🔍 PDF Text Extractor
|
| 25 |
-
|
| 26 |
-
⚠️ **API key not configured.** Please set MISTRAL_API_KEY environment variable and restart the application.
|
| 27 |
-
""")
|
| 28 |
-
|
| 29 |
-
with gr.Row():
|
| 30 |
-
gr.File(label="Upload PDF", file_types=[".pdf"])
|
| 31 |
-
|
| 32 |
-
with gr.Row():
|
| 33 |
-
gr.Button("Extract Text", variant="primary", interactive=False)
|
| 34 |
-
|
| 35 |
-
with gr.Row():
|
| 36 |
-
gr.Textbox(
|
| 37 |
-
label="Extracted Text",
|
| 38 |
-
lines=10,
|
| 39 |
-
value="API key not configured. Text extraction is unavailable.",
|
| 40 |
-
interactive=False
|
| 41 |
-
)
|
| 42 |
-
|
| 43 |
-
with gr.Row():
|
| 44 |
-
gr.Textbox(
|
| 45 |
-
label="Status",
|
| 46 |
-
lines=2,
|
| 47 |
-
value="❌ MISTRAL_API_KEY environment variable is not set. Please set it and restart the application."
|
| 48 |
-
)
|
| 49 |
-
|
| 50 |
-
with gr.Row():
|
| 51 |
-
gr.Button("📋 Copy to Clipboard", interactive=False)
|
| 52 |
-
gr.Button("📥 Download as Text File", interactive=False)
|
| 53 |
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
def create_main_interface(extractor: PDFTextExtractor) -> gr.Blocks:
|
| 59 |
-
"""
|
| 60 |
-
Create the main application interface.
|
| 61 |
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
Returns:
|
| 66 |
-
gr.Blocks: Gradio interface with full functionality """
|
| 67 |
-
# Make the extractor a local function attribute
|
| 68 |
-
def process_pdf_wrapper(pdf_file):
|
| 69 |
-
"""Process PDF with the extractor from closure"""
|
| 70 |
-
extracted_text, status, images_data = extractor.extract_text_from_pdf(pdf_file)
|
| 71 |
-
# Get PDF file path for image extraction
|
| 72 |
-
pdf_path = pdf_file.name if hasattr(pdf_file, 'name') else pdf_file if pdf_file else None
|
| 73 |
-
gallery_images = process_images_for_display(images_data, pdf_path)
|
| 74 |
-
return extracted_text, status, gallery_images
|
| 75 |
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
#
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
submit_btn = create_action_button()
|
| 87 |
-
# Add status display
|
| 88 |
-
with gr.Row():
|
| 89 |
-
status_output = gr.Textbox(
|
| 90 |
-
label="Status",
|
| 91 |
-
lines=2,
|
| 92 |
-
placeholder="Upload a PDF to see status..."
|
| 93 |
-
)
|
| 94 |
-
# Create tabs for text and images
|
| 95 |
-
with gr.Tabs():
|
| 96 |
-
with gr.TabItem("Extracted Text"):
|
| 97 |
-
text_output = gr.Textbox(
|
| 98 |
-
label="Extracted Text",
|
| 99 |
-
lines=15,
|
| 100 |
-
max_lines=30,
|
| 101 |
-
placeholder="Extracted text will appear here...",
|
| 102 |
-
show_copy_button=True
|
| 103 |
-
)
|
| 104 |
-
with gr.Row():
|
| 105 |
-
copy_btn, download_btn = create_action_buttons()
|
| 106 |
-
with gr.TabItem("Extracted Images"):
|
| 107 |
-
image_gallery = create_image_gallery()
|
| 108 |
-
image_info = gr.Markdown("Images extracted from the PDF will appear here.")
|
| 109 |
-
# Set up function calls
|
| 110 |
-
# Update PDF viewer when a file is uploaded
|
| 111 |
-
pdf_input.change(
|
| 112 |
-
fn=lambda f: f,
|
| 113 |
-
inputs=pdf_input,
|
| 114 |
-
outputs=pdf_viewer
|
| 115 |
-
)
|
| 116 |
-
submit_btn.click(
|
| 117 |
-
fn=process_pdf_wrapper,
|
| 118 |
-
inputs=[pdf_input],
|
| 119 |
-
outputs=[text_output, status_output, image_gallery]
|
| 120 |
-
)
|
| 121 |
-
copy_btn.click(
|
| 122 |
-
fn=copy_text,
|
| 123 |
-
inputs=text_output,
|
| 124 |
-
outputs=None,
|
| 125 |
-
js="""
|
| 126 |
-
function(text) {
|
| 127 |
-
if (text) {
|
| 128 |
-
navigator.clipboard.writeText(text);
|
| 129 |
-
// Show a temporary notification
|
| 130 |
-
var notification = document.createElement('div');
|
| 131 |
-
notification.textContent = 'Text copied to clipboard!';
|
| 132 |
-
notification.style.position = 'fixed';
|
| 133 |
-
notification.style.bottom = '20px';
|
| 134 |
-
notification.style.left = '50%';
|
| 135 |
-
notification.style.transform = 'translateX(-50%)';
|
| 136 |
-
notification.style.padding = '10px 20px';
|
| 137 |
-
notification.style.background = '#4CAF50';
|
| 138 |
-
notification.style.color = 'white';
|
| 139 |
-
notification.style.borderRadius = '4px';
|
| 140 |
-
notification.style.zIndex = '1000';
|
| 141 |
-
document.body.appendChild(notification);
|
| 142 |
-
setTimeout(function() {
|
| 143 |
-
document.body.removeChild(notification);
|
| 144 |
-
}, 2000);
|
| 145 |
-
}
|
| 146 |
-
return text;
|
| 147 |
-
}
|
| 148 |
-
"""
|
| 149 |
-
)
|
| 150 |
-
download_btn.click(
|
| 151 |
-
fn=download_text,
|
| 152 |
-
inputs=text_output,
|
| 153 |
-
outputs=gr.File(label="Download", elem_id="download_output"),
|
| 154 |
-
show_progress=False
|
| 155 |
-
)
|
| 156 |
-
apply_custom_css()
|
| 157 |
-
return interface
|
| 158 |
|
| 159 |
-
|
| 160 |
-
|
| 161 |
-
Create and configure the Gradio interface.
|
| 162 |
-
|
| 163 |
-
Returns:
|
| 164 |
-
gr.Blocks: Configured Gradio interface
|
| 165 |
-
"""
|
| 166 |
-
# Initialize the PDF extractor
|
| 167 |
-
try:
|
| 168 |
-
extractor = PDFTextExtractor()
|
| 169 |
-
return create_main_interface(extractor)
|
| 170 |
-
except ValueError as e:
|
| 171 |
-
# Create a dummy interface if API key is missing
|
| 172 |
-
return create_dummy_interface()
|
|
|
|
| 1 |
"""
|
| 2 |
+
PDF Text Extractor Application
|
| 3 |
+
Main entry point for the PDF Text Extractor application.
|
| 4 |
"""
|
| 5 |
|
| 6 |
+
import os
|
| 7 |
+
from dotenv import load_dotenv
|
| 8 |
+
from ui import create_interface
|
| 9 |
+
from utils.config import check_api_key, get_app_config
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
+
def main():
|
| 12 |
+
"""Main function to launch the application."""
|
|
|
|
| 13 |
|
| 14 |
+
# Load environment variables from .env file
|
| 15 |
+
load_dotenv()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
+
# Check for API key
|
| 18 |
+
check_api_key()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
|
| 20 |
+
# Create and launch the interface
|
| 21 |
+
interface = create_interface()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
+
# Get application configuration
|
| 24 |
+
app_config = get_app_config()
|
| 25 |
+
|
| 26 |
+
# Launch with appropriate settings
|
| 27 |
+
interface.launch(
|
| 28 |
+
# server_port=app_config["server_port"],
|
| 29 |
+
debug=app_config["debug"],
|
| 30 |
+
quiet=app_config["quiet"],
|
| 31 |
+
max_file_size=app_config["max_file_size"]
|
| 32 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
|
| 34 |
+
if __name__ == "__main__":
|
| 35 |
+
main()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|