Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,17 +1,14 @@
|
|
| 1 |
import base64
|
| 2 |
import streamlit as st
|
| 3 |
-
|
| 4 |
-
import
|
| 5 |
from PIL import Image
|
| 6 |
import io
|
| 7 |
import tempfile
|
|
|
|
| 8 |
|
| 9 |
-
#
|
| 10 |
-
|
| 11 |
-
# Initialize OpenAI client
|
| 12 |
-
client = OpenAI()
|
| 13 |
-
|
| 14 |
-
|
| 15 |
|
| 16 |
def convert_pdf_to_images(pdf_file):
|
| 17 |
"""Convert PDF to list of images"""
|
|
@@ -19,9 +16,13 @@ def convert_pdf_to_images(pdf_file):
|
|
| 19 |
tmp_file.write(pdf_file.getvalue())
|
| 20 |
pdf_path = tmp_file.name
|
| 21 |
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
def format_response(text):
|
| 27 |
"""Format the analysis response with clean styling"""
|
|
@@ -38,7 +39,7 @@ def format_response(text):
|
|
| 38 |
# Split into pages
|
| 39 |
pages = text.split("Page")
|
| 40 |
|
| 41 |
-
for page_num, page_content in enumerate(pages[1:], 1):
|
| 42 |
# Add page header
|
| 43 |
formatted_text += f'<div style="margin-bottom: 30px;">'
|
| 44 |
formatted_text += f'<h3 style="color: #2c3e50; margin-bottom: 15px;">Page {page_num}</h3>'
|
|
@@ -46,9 +47,7 @@ def format_response(text):
|
|
| 46 |
# Process each line
|
| 47 |
lines = page_content.split('\n')
|
| 48 |
for line in lines:
|
| 49 |
-
# Skip empty lines and lines with asterisks
|
| 50 |
if line.strip() and not line.strip().startswith('*') and not line.strip().startswith('Here'):
|
| 51 |
-
# Remove asterisks and dashes
|
| 52 |
line = line.replace('**', '').replace('- ', '')
|
| 53 |
|
| 54 |
if ':' in line:
|
|
@@ -60,7 +59,6 @@ def format_response(text):
|
|
| 60 |
|
| 61 |
formatted_text += '</div>'
|
| 62 |
|
| 63 |
-
# Add separator between pages except for the last page
|
| 64 |
if page_num < len(pages) - 1:
|
| 65 |
formatted_text += '<hr style="border: 1px solid #eee; margin: 20px 0;">'
|
| 66 |
|
|
@@ -76,8 +74,8 @@ def analyze_image(image):
|
|
| 76 |
|
| 77 |
base64_image = base64.b64encode(img_byte_arr).decode("utf-8")
|
| 78 |
|
| 79 |
-
response =
|
| 80 |
-
model="gpt-
|
| 81 |
messages=[
|
| 82 |
{
|
| 83 |
"role": "user",
|
|
@@ -106,14 +104,14 @@ def analyze_image(image):
|
|
| 106 |
max_tokens=1000
|
| 107 |
)
|
| 108 |
|
| 109 |
-
return response.choices[0].message
|
| 110 |
except Exception as e:
|
|
|
|
| 111 |
return f"An error occurred: {str(e)}"
|
| 112 |
|
| 113 |
def main():
|
| 114 |
st.set_page_config(page_title="Document Analysis App", layout="wide")
|
| 115 |
|
| 116 |
-
# Custom CSS to set light background and improve button styling
|
| 117 |
st.markdown("""
|
| 118 |
<style>
|
| 119 |
.stApp {
|
|
@@ -146,7 +144,6 @@ def main():
|
|
| 146 |
|
| 147 |
if uploaded_file is not None:
|
| 148 |
if uploaded_file.type == "application/pdf":
|
| 149 |
-
# Handle PDF
|
| 150 |
with st.spinner("Processing PDF..."):
|
| 151 |
images = convert_pdf_to_images(uploaded_file)
|
| 152 |
|
|
@@ -161,7 +158,6 @@ def main():
|
|
| 161 |
st.markdown(format_response(combined_results), unsafe_allow_html=True)
|
| 162 |
|
| 163 |
else:
|
| 164 |
-
# Handle single image
|
| 165 |
image = Image.open(uploaded_file)
|
| 166 |
|
| 167 |
if st.button("Extract Information"):
|
|
|
|
| 1 |
import base64
|
| 2 |
import streamlit as st
|
| 3 |
+
import openai
|
| 4 |
+
import os
|
| 5 |
from PIL import Image
|
| 6 |
import io
|
| 7 |
import tempfile
|
| 8 |
+
import pdf2image
|
| 9 |
|
| 10 |
+
# Set OpenAI API key from Streamlit secrets
|
| 11 |
+
openai.api_key = st.secrets["OPENAI_API_KEY"]
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
def convert_pdf_to_images(pdf_file):
|
| 14 |
"""Convert PDF to list of images"""
|
|
|
|
| 16 |
tmp_file.write(pdf_file.getvalue())
|
| 17 |
pdf_path = tmp_file.name
|
| 18 |
|
| 19 |
+
try:
|
| 20 |
+
images = pdf2image.convert_from_path(pdf_path)
|
| 21 |
+
os.unlink(pdf_path)
|
| 22 |
+
return images
|
| 23 |
+
except Exception as e:
|
| 24 |
+
st.error(f"Error converting PDF: {str(e)}")
|
| 25 |
+
return []
|
| 26 |
|
| 27 |
def format_response(text):
|
| 28 |
"""Format the analysis response with clean styling"""
|
|
|
|
| 39 |
# Split into pages
|
| 40 |
pages = text.split("Page")
|
| 41 |
|
| 42 |
+
for page_num, page_content in enumerate(pages[1:], 1):
|
| 43 |
# Add page header
|
| 44 |
formatted_text += f'<div style="margin-bottom: 30px;">'
|
| 45 |
formatted_text += f'<h3 style="color: #2c3e50; margin-bottom: 15px;">Page {page_num}</h3>'
|
|
|
|
| 47 |
# Process each line
|
| 48 |
lines = page_content.split('\n')
|
| 49 |
for line in lines:
|
|
|
|
| 50 |
if line.strip() and not line.strip().startswith('*') and not line.strip().startswith('Here'):
|
|
|
|
| 51 |
line = line.replace('**', '').replace('- ', '')
|
| 52 |
|
| 53 |
if ':' in line:
|
|
|
|
| 59 |
|
| 60 |
formatted_text += '</div>'
|
| 61 |
|
|
|
|
| 62 |
if page_num < len(pages) - 1:
|
| 63 |
formatted_text += '<hr style="border: 1px solid #eee; margin: 20px 0;">'
|
| 64 |
|
|
|
|
| 74 |
|
| 75 |
base64_image = base64.b64encode(img_byte_arr).decode("utf-8")
|
| 76 |
|
| 77 |
+
response = openai.ChatCompletion.create(
|
| 78 |
+
model="gpt-4-vision-preview",
|
| 79 |
messages=[
|
| 80 |
{
|
| 81 |
"role": "user",
|
|
|
|
| 104 |
max_tokens=1000
|
| 105 |
)
|
| 106 |
|
| 107 |
+
return response.choices[0].message['content']
|
| 108 |
except Exception as e:
|
| 109 |
+
st.error(f"API Error: {str(e)}")
|
| 110 |
return f"An error occurred: {str(e)}"
|
| 111 |
|
| 112 |
def main():
|
| 113 |
st.set_page_config(page_title="Document Analysis App", layout="wide")
|
| 114 |
|
|
|
|
| 115 |
st.markdown("""
|
| 116 |
<style>
|
| 117 |
.stApp {
|
|
|
|
| 144 |
|
| 145 |
if uploaded_file is not None:
|
| 146 |
if uploaded_file.type == "application/pdf":
|
|
|
|
| 147 |
with st.spinner("Processing PDF..."):
|
| 148 |
images = convert_pdf_to_images(uploaded_file)
|
| 149 |
|
|
|
|
| 158 |
st.markdown(format_response(combined_results), unsafe_allow_html=True)
|
| 159 |
|
| 160 |
else:
|
|
|
|
| 161 |
image = Image.open(uploaded_file)
|
| 162 |
|
| 163 |
if st.button("Extract Information"):
|