mkaramb commited on
Commit
bc9779f
·
verified ·
1 Parent(s): 07e2341

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +27 -14
app.py CHANGED
@@ -64,14 +64,11 @@ def unzip_and_find_jpgs(file_path):
64
  def process_images(uploaded_file):
65
  # Reinitialize the DataFrame each time this function is called
66
  results_df = pd.DataFrame(columns=["Filename", "Extracted Text", "Translated Text"])
67
- print("DataFrame reinitialized:", results_df) # Debugging statement
68
 
69
  file_path = uploaded_file.name # Gradio provides the file path through the .name attribute
70
- print("Processing file:", file_path) # Debugging statement
71
 
72
  try:
73
  image_files = unzip_and_find_jpgs(file_path)
74
- print("Found image files:", image_files) # Debugging statement
75
 
76
  if not image_files:
77
  return "No JPG files found in the zip."
@@ -84,20 +81,36 @@ def process_images(uploaded_file):
84
  "Translated Text": translated_text
85
  }])
86
  results_df = pd.concat([results_df, new_row], ignore_index=True)
87
- print("Current DataFrame state:", results_df) # Debugging statement
88
  except Exception as e:
89
  return f"An error occurred: {str(e)}"
90
 
91
- return results_df.to_html()
92
-
93
-
94
- interface = gr.Interface(
95
- fn=process_images,
96
- inputs="file",
97
- outputs="html",
98
- title="Document AI Translation",
99
- description="Upload a ZIP file containing JPEG/JPG images, and the system will extract and translate text from each image."
100
- )
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
101
 
102
  if __name__ == "__main__":
103
  interface.launch(debug=True)
 
64
  def process_images(uploaded_file):
65
  # Reinitialize the DataFrame each time this function is called
66
  results_df = pd.DataFrame(columns=["Filename", "Extracted Text", "Translated Text"])
 
67
 
68
  file_path = uploaded_file.name # Gradio provides the file path through the .name attribute
 
69
 
70
  try:
71
  image_files = unzip_and_find_jpgs(file_path)
 
72
 
73
  if not image_files:
74
  return "No JPG files found in the zip."
 
81
  "Translated Text": translated_text
82
  }])
83
  results_df = pd.concat([results_df, new_row], ignore_index=True)
 
84
  except Exception as e:
85
  return f"An error occurred: {str(e)}"
86
 
87
+ html_output = results_df.to_html()
88
+
89
+ # Save DataFrame to a temporary CSV file for download
90
+ temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".csv") # Create a temp file
91
+ results_df.to_csv(temp_file.name, index=False) # Save DataFrame to CSV
92
+
93
+ # Read the contents back to verify
94
+ temp_file.seek(0) # Move file pointer to the beginning of the file
95
+ print(pd.read_csv(temp_file.name)) # Read and print the CSV file to verify its contents
96
+
97
+ temp_file.close() # Close the file
98
+
99
+ # Return HTML and the path to the CSV file
100
+ return html_output, temp_file.name
101
+
102
+ with gr.Blocks() as interface:
103
+ with gr.Row():
104
+ gr.Markdown("# Document AI Translation")
105
+ gr.Markdown("Upload a ZIP file containing JPEG/JPG images, and the system will extract and translate text from each image.")
106
+ with gr.Row():
107
+ file_input = gr.File(label="Upload ZIP File")
108
+ with gr.Row():
109
+ html_output = gr.HTML()
110
+ with gr.Row():
111
+ file_output = gr.File()
112
+
113
+ file_input.change(process_images, inputs=file_input, outputs=[html_output, file_output])
114
 
115
  if __name__ == "__main__":
116
  interface.launch(debug=True)