sachin commited on
Commit ·
7fe792a
1
Parent(s): da18e21
init
Browse files- app.py +121 -0
- requirements.txt +2 -0
app.py
ADDED
|
@@ -0,0 +1,121 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import requests
|
| 3 |
+
import os
|
| 4 |
+
import dwani
|
| 5 |
+
import logging
|
| 6 |
+
|
| 7 |
+
# Set up logging
|
| 8 |
+
logging.basicConfig(level=logging.INFO)
|
| 9 |
+
logger = logging.getLogger(__name__)
|
| 10 |
+
|
| 11 |
+
# Configure dwani API settings
|
| 12 |
+
dwani.api_key = os.getenv("DWANI_API_KEY")
|
| 13 |
+
dwani.api_base = os.getenv("DWANI_API_BASE_URL")
|
| 14 |
+
|
| 15 |
+
# Log API configuration for debugging
|
| 16 |
+
logger.debug("DWANI_API_KEY: %s", "Set" if dwani.api_key else "Not set")
|
| 17 |
+
logger.debug("DWANI_API_BASE_URL: %s", dwani.api_base)
|
| 18 |
+
|
| 19 |
+
# Language options as simple array
|
| 20 |
+
language_options = ["english", "kannada", "hindi"]
|
| 21 |
+
|
| 22 |
+
def process_pdf(pdf_file, page_number, prompt, src_lang, tgt_lang):
|
| 23 |
+
logger.debug("Received inputs - PDF: %s, Page: %s, Prompt: %s, Source Lang: %s, Target Lang: %s",
|
| 24 |
+
pdf_file, page_number, prompt, src_lang, tgt_lang)
|
| 25 |
+
|
| 26 |
+
# Validate inputs
|
| 27 |
+
if not pdf_file:
|
| 28 |
+
logger.error("No PDF file provided")
|
| 29 |
+
return {"error": "Please upload a PDF file"}
|
| 30 |
+
|
| 31 |
+
if not prompt.strip():
|
| 32 |
+
logger.error("Prompt is empty")
|
| 33 |
+
return {"error": "Please provide a non-empty prompt"}
|
| 34 |
+
|
| 35 |
+
try:
|
| 36 |
+
page_number = int(page_number)
|
| 37 |
+
if page_number < 1:
|
| 38 |
+
raise ValueError("Page number must be at least 1")
|
| 39 |
+
except (ValueError, TypeError):
|
| 40 |
+
logger.error("Invalid page number: %s", page_number)
|
| 41 |
+
return {"error": "Page number must be a positive integer"}
|
| 42 |
+
|
| 43 |
+
# Validate language codes
|
| 44 |
+
if src_lang not in language_options or tgt_lang not in language_options:
|
| 45 |
+
logger.error("Invalid language selection - Source: %s, Target: %s", src_lang, tgt_lang)
|
| 46 |
+
return {"error": "Invalid source or target language selection"}
|
| 47 |
+
|
| 48 |
+
# Get file path from Gradio File object
|
| 49 |
+
file_path = pdf_file.name if hasattr(pdf_file, 'name') else pdf_file
|
| 50 |
+
|
| 51 |
+
logger.debug("Calling API with file: %s, page: %d, prompt: %s, src_lang: %s, tgt_lang: %s",
|
| 52 |
+
file_path, page_number, prompt, src_lang, tgt_lang)
|
| 53 |
+
|
| 54 |
+
# Call the API
|
| 55 |
+
try:
|
| 56 |
+
result = dwani.Documents.run_doc_query(
|
| 57 |
+
file_path=file_path,
|
| 58 |
+
prompt=prompt,
|
| 59 |
+
page_number=page_number,
|
| 60 |
+
src_lang=src_lang,
|
| 61 |
+
tgt_lang=tgt_lang
|
| 62 |
+
)
|
| 63 |
+
logger.debug("API response: %s", result)
|
| 64 |
+
return {
|
| 65 |
+
"Original Text": result.get("original_text", "N/A"),
|
| 66 |
+
"Response": result.get("response", "N/A"),
|
| 67 |
+
"Processed Page": result.get("processed_page", "N/A"),
|
| 68 |
+
"Translated Response": result.get("translated_response", "N/A")
|
| 69 |
+
}
|
| 70 |
+
except dwani.exceptions.DhwaniAPIError as e:
|
| 71 |
+
logger.error("Dhwani API error: %s", str(e))
|
| 72 |
+
return {"error": f"API error: {str(e)}"}
|
| 73 |
+
except Exception as e:
|
| 74 |
+
logger.error("Unexpected error: %s", str(e))
|
| 75 |
+
return {"error": f"Unexpected error: {str(e)}"}
|
| 76 |
+
|
| 77 |
+
# Define Gradio interface
|
| 78 |
+
with gr.Blocks(title="PDF Custom Prompt Processor") as demo:
|
| 79 |
+
gr.Markdown("# PDF Custom Prompt Processor")
|
| 80 |
+
gr.Markdown("Upload a PDF, specify a page number, enter a prompt, and select source and target languages.")
|
| 81 |
+
|
| 82 |
+
with gr.Row():
|
| 83 |
+
with gr.Column():
|
| 84 |
+
pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
|
| 85 |
+
page_number = gr.Number(label="Page Number", value=1, minimum=1, precision=0)
|
| 86 |
+
prompt = gr.Textbox(
|
| 87 |
+
label="Custom Prompt",
|
| 88 |
+
placeholder="e.g., List the key points",
|
| 89 |
+
value="List the key points",
|
| 90 |
+
lines=3
|
| 91 |
+
)
|
| 92 |
+
src_lang_input = gr.Dropdown(
|
| 93 |
+
label="Source Language",
|
| 94 |
+
choices=language_options,
|
| 95 |
+
value="english"
|
| 96 |
+
)
|
| 97 |
+
tgt_lang_input = gr.Dropdown(
|
| 98 |
+
label="Target Language",
|
| 99 |
+
choices=language_options,
|
| 100 |
+
value="kannada"
|
| 101 |
+
)
|
| 102 |
+
submit_btn = gr.Button("Process")
|
| 103 |
+
|
| 104 |
+
with gr.Column():
|
| 105 |
+
output = gr.JSON(label="Response")
|
| 106 |
+
|
| 107 |
+
submit_btn.click(
|
| 108 |
+
fn=process_pdf,
|
| 109 |
+
inputs=[pdf_input, page_number, prompt, src_lang_input, tgt_lang_input],
|
| 110 |
+
outputs=output
|
| 111 |
+
)
|
| 112 |
+
|
| 113 |
+
# Launch the interface
|
| 114 |
+
if __name__ == "__main__":
|
| 115 |
+
# Test API configuration
|
| 116 |
+
if not dwani.api_key or not dwani.api_base:
|
| 117 |
+
logger.error("API key or base URL not set. Please set DWANI_API_KEY and DWANI_API_BASE_URL environment variables.")
|
| 118 |
+
print("Error: Please set DWANI_API_KEY and DWANI_API_BASE_URL environment variables.")
|
| 119 |
+
else:
|
| 120 |
+
logger.debug("Starting Gradio interface...")
|
| 121 |
+
demo.launch()
|
requirements.txt
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
gradio
|
| 2 |
+
dwani
|