Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -7,6 +7,13 @@ import PyPDF2
|
|
| 7 |
from docx import Document
|
| 8 |
import io
|
| 9 |
import tempfile
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
class MultiConverter:
|
| 12 |
def convert_excel_to_formatted_text(self, excel_file):
|
|
@@ -90,38 +97,70 @@ class MultiConverter:
|
|
| 90 |
|
| 91 |
def convert_file(file):
|
| 92 |
"""Process uploaded file and convert it to text"""
|
|
|
|
|
|
|
|
|
|
|
|
|
| 93 |
try:
|
|
|
|
| 94 |
converter = MultiConverter()
|
| 95 |
filename = file.name
|
| 96 |
_, file_ext = os.path.splitext(filename)
|
| 97 |
file_ext = file_ext.lower()
|
| 98 |
|
|
|
|
|
|
|
| 99 |
# Create a temporary file
|
| 100 |
with tempfile.NamedTemporaryFile(delete=False) as temp_file:
|
| 101 |
-
|
|
|
|
|
|
|
| 102 |
temp_file_path = temp_file.name
|
|
|
|
| 103 |
|
| 104 |
try:
|
|
|
|
| 105 |
if file_ext in [".xlsx", ".xls"]:
|
|
|
|
| 106 |
result = converter.convert_excel_to_formatted_text(temp_file_path)
|
| 107 |
elif file_ext in [".pptx", ".ppt"]:
|
|
|
|
| 108 |
result = converter.convert_pptx_to_text(temp_file_path, filename)
|
| 109 |
elif file_ext == ".pdf":
|
|
|
|
| 110 |
result = converter.convert_pdf_to_text(temp_file_path, filename)
|
| 111 |
elif file_ext in [".docx", ".doc"]:
|
|
|
|
| 112 |
result = converter.convert_docx_to_text(temp_file_path, filename)
|
| 113 |
else:
|
| 114 |
-
|
| 115 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
# Create and return file for download
|
| 117 |
output_filename = os.path.splitext(filename)[0] + ".txt"
|
| 118 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 119 |
finally:
|
| 120 |
# Clean up the temporary file
|
| 121 |
if os.path.exists(temp_file_path):
|
|
|
|
| 122 |
os.unlink(temp_file_path)
|
|
|
|
|
|
|
| 123 |
except Exception as e:
|
| 124 |
-
|
|
|
|
|
|
|
| 125 |
|
| 126 |
|
| 127 |
# Create Gradio interface
|
|
@@ -132,20 +171,42 @@ with gr.Blocks(title="Multi-Format to TXT Converter") as app:
|
|
| 132 |
with gr.Row():
|
| 133 |
with gr.Column():
|
| 134 |
file_input = gr.File(label="Upload a file (Excel, PowerPoint, PDF, or Word)")
|
| 135 |
-
convert_button = gr.Button("Convert to TXT")
|
| 136 |
|
| 137 |
with gr.Row():
|
| 138 |
with gr.Column():
|
| 139 |
text_output = gr.Textbox(label="Converted Text", lines=15)
|
| 140 |
file_output = gr.File(label="Download Text File")
|
| 141 |
|
|
|
|
|
|
|
|
|
|
| 142 |
# Setup the conversion process
|
| 143 |
convert_button.click(
|
| 144 |
fn=convert_file,
|
| 145 |
inputs=[file_input],
|
| 146 |
-
outputs=[text_output, file_output,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 147 |
)
|
| 148 |
|
| 149 |
# Launch the app
|
| 150 |
if __name__ == "__main__":
|
| 151 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
from docx import Document
|
| 8 |
import io
|
| 9 |
import tempfile
|
| 10 |
+
import logging
|
| 11 |
+
|
| 12 |
+
# Konfiguracja logowania
|
| 13 |
+
logging.basicConfig(
|
| 14 |
+
level=logging.INFO,
|
| 15 |
+
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
|
| 16 |
+
)
|
| 17 |
|
| 18 |
class MultiConverter:
|
| 19 |
def convert_excel_to_formatted_text(self, excel_file):
|
|
|
|
| 97 |
|
| 98 |
def convert_file(file):
|
| 99 |
"""Process uploaded file and convert it to text"""
|
| 100 |
+
if file is None:
|
| 101 |
+
logging.warning("No file provided")
|
| 102 |
+
return "No file uploaded. Please select a file first.", None, None
|
| 103 |
+
|
| 104 |
try:
|
| 105 |
+
logging.info(f"Converting file: {file.name}")
|
| 106 |
converter = MultiConverter()
|
| 107 |
filename = file.name
|
| 108 |
_, file_ext = os.path.splitext(filename)
|
| 109 |
file_ext = file_ext.lower()
|
| 110 |
|
| 111 |
+
logging.info(f"File extension: {file_ext}")
|
| 112 |
+
|
| 113 |
# Create a temporary file
|
| 114 |
with tempfile.NamedTemporaryFile(delete=False) as temp_file:
|
| 115 |
+
file_content = file.read()
|
| 116 |
+
logging.info(f"Read {len(file_content)} bytes from input file")
|
| 117 |
+
temp_file.write(file_content)
|
| 118 |
temp_file_path = temp_file.name
|
| 119 |
+
logging.info(f"Created temporary file at: {temp_file_path}")
|
| 120 |
|
| 121 |
try:
|
| 122 |
+
result = None
|
| 123 |
if file_ext in [".xlsx", ".xls"]:
|
| 124 |
+
logging.info("Processing Excel file")
|
| 125 |
result = converter.convert_excel_to_formatted_text(temp_file_path)
|
| 126 |
elif file_ext in [".pptx", ".ppt"]:
|
| 127 |
+
logging.info("Processing PowerPoint file")
|
| 128 |
result = converter.convert_pptx_to_text(temp_file_path, filename)
|
| 129 |
elif file_ext == ".pdf":
|
| 130 |
+
logging.info("Processing PDF file")
|
| 131 |
result = converter.convert_pdf_to_text(temp_file_path, filename)
|
| 132 |
elif file_ext in [".docx", ".doc"]:
|
| 133 |
+
logging.info("Processing Word file")
|
| 134 |
result = converter.convert_docx_to_text(temp_file_path, filename)
|
| 135 |
else:
|
| 136 |
+
error_msg = f"Unsupported file format: {file_ext}"
|
| 137 |
+
logging.error(error_msg)
|
| 138 |
+
result = error_msg
|
| 139 |
+
|
| 140 |
+
logging.info(f"Conversion successful, text length: {len(result) if result else 0}")
|
| 141 |
+
|
| 142 |
# Create and return file for download
|
| 143 |
output_filename = os.path.splitext(filename)[0] + ".txt"
|
| 144 |
+
|
| 145 |
+
# Convert result to bytes for file download
|
| 146 |
+
if result:
|
| 147 |
+
file_content = result.encode('utf-8')
|
| 148 |
+
logging.info(f"Created output file: {output_filename}, size: {len(file_content)} bytes")
|
| 149 |
+
return result, (output_filename, file_content), result
|
| 150 |
+
else:
|
| 151 |
+
logging.warning("No result generated")
|
| 152 |
+
return "No result generated", None, None
|
| 153 |
finally:
|
| 154 |
# Clean up the temporary file
|
| 155 |
if os.path.exists(temp_file_path):
|
| 156 |
+
logging.info(f"Removing temporary file: {temp_file_path}")
|
| 157 |
os.unlink(temp_file_path)
|
| 158 |
+
else:
|
| 159 |
+
logging.warning(f"Temporary file not found for deletion: {temp_file_path}")
|
| 160 |
except Exception as e:
|
| 161 |
+
error_msg = f"Error converting file: {str(e)}"
|
| 162 |
+
logging.exception(error_msg)
|
| 163 |
+
return error_msg, None, None
|
| 164 |
|
| 165 |
|
| 166 |
# Create Gradio interface
|
|
|
|
| 171 |
with gr.Row():
|
| 172 |
with gr.Column():
|
| 173 |
file_input = gr.File(label="Upload a file (Excel, PowerPoint, PDF, or Word)")
|
| 174 |
+
convert_button = gr.Button("Convert to TXT", variant="primary")
|
| 175 |
|
| 176 |
with gr.Row():
|
| 177 |
with gr.Column():
|
| 178 |
text_output = gr.Textbox(label="Converted Text", lines=15)
|
| 179 |
file_output = gr.File(label="Download Text File")
|
| 180 |
|
| 181 |
+
# Add status message
|
| 182 |
+
status_output = gr.Textbox(label="Status", value="Ready to convert")
|
| 183 |
+
|
| 184 |
# Setup the conversion process
|
| 185 |
convert_button.click(
|
| 186 |
fn=convert_file,
|
| 187 |
inputs=[file_input],
|
| 188 |
+
outputs=[text_output, file_output, status_output]
|
| 189 |
+
)
|
| 190 |
+
|
| 191 |
+
# Dodaj przykładowe pliki do demonstracji (opcjonalnie)
|
| 192 |
+
gr.Examples(
|
| 193 |
+
examples=[
|
| 194 |
+
["example.xlsx"],
|
| 195 |
+
["example.pdf"],
|
| 196 |
+
["example.docx"],
|
| 197 |
+
["example.pptx"],
|
| 198 |
+
],
|
| 199 |
+
inputs=[file_input],
|
| 200 |
+
outputs=[text_output, file_output, status_output],
|
| 201 |
+
fn=convert_file,
|
| 202 |
+
cache_examples=False,
|
| 203 |
)
|
| 204 |
|
| 205 |
# Launch the app
|
| 206 |
if __name__ == "__main__":
|
| 207 |
+
try:
|
| 208 |
+
logging.info("Starting the application")
|
| 209 |
+
app.launch(debug=True, share=False)
|
| 210 |
+
logging.info("Application stopped")
|
| 211 |
+
except Exception as e:
|
| 212 |
+
logging.exception(f"Error launching application: {str(e)}")
|