Upload 3 files
Browse files
app.py
CHANGED
|
@@ -1,5 +1,4 @@
|
|
| 1 |
import os
|
| 2 |
-
import re
|
| 3 |
import streamlit as st
|
| 4 |
from io import BytesIO
|
| 5 |
from tempfile import NamedTemporaryFile
|
|
@@ -11,8 +10,17 @@ from langchain_community.document_loaders import PyPDFLoader
|
|
| 11 |
from langchain_community.vectorstores import FAISS
|
| 12 |
from langchain_openai import OpenAIEmbeddings
|
| 13 |
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
| 14 |
-
|
| 15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
# Function to process PDF, run Q&A, and return results
|
| 18 |
def process_pdf(api_key, uploaded_file, questions_path, prompt_path, display_placeholder):
|
|
@@ -71,10 +79,10 @@ def process_pdf(api_key, uploaded_file, questions_path, prompt_path, display_pla
|
|
| 71 |
result = rag_chain.invoke({"input": question})
|
| 72 |
answer = result["answer"]
|
| 73 |
|
| 74 |
-
#
|
| 75 |
-
answer =
|
| 76 |
|
| 77 |
-
qa_text = f"### Question: {question}\n**Answer:**\n
|
| 78 |
qa_results.append(qa_text)
|
| 79 |
# Update the placeholder with each new Q&A pair
|
| 80 |
display_placeholder.markdown("\n".join(qa_results), unsafe_allow_html=True)
|
|
@@ -84,63 +92,6 @@ def process_pdf(api_key, uploaded_file, questions_path, prompt_path, display_pla
|
|
| 84 |
|
| 85 |
return qa_results
|
| 86 |
|
| 87 |
-
# Function to clean up the AI's answer
|
| 88 |
-
def clean_answer(answer):
|
| 89 |
-
# Remove unwanted prefixes like 'markdown'
|
| 90 |
-
answer = answer.strip()
|
| 91 |
-
if answer.lower().startswith('markdown'):
|
| 92 |
-
answer = answer[len('markdown'):].strip()
|
| 93 |
-
|
| 94 |
-
# Additional cleaning if necessary
|
| 95 |
-
# For example, ensure that markdown table syntax is correct
|
| 96 |
-
return answer
|
| 97 |
-
|
| 98 |
-
# Function to convert markdown text to PDF with table support
|
| 99 |
-
def md_to_pdf(md_text):
|
| 100 |
-
# Convert markdown to HTML with table support
|
| 101 |
-
html_content = markdown(md_text, output_format='html5', extensions=['tables'])
|
| 102 |
-
|
| 103 |
-
# Define CSS styles for better table rendering
|
| 104 |
-
css_styles = '''
|
| 105 |
-
<style>
|
| 106 |
-
body {
|
| 107 |
-
font-family: Arial, sans-serif;
|
| 108 |
-
font-size: 12pt;
|
| 109 |
-
}
|
| 110 |
-
table {
|
| 111 |
-
border-collapse: collapse;
|
| 112 |
-
width: 100%;
|
| 113 |
-
}
|
| 114 |
-
th, td {
|
| 115 |
-
border: 1px solid black;
|
| 116 |
-
padding: 8px;
|
| 117 |
-
text-align: left;
|
| 118 |
-
}
|
| 119 |
-
th {
|
| 120 |
-
background-color: #f2f2f2;
|
| 121 |
-
}
|
| 122 |
-
</style>
|
| 123 |
-
'''
|
| 124 |
-
|
| 125 |
-
# Construct the full HTML with CSS and content
|
| 126 |
-
html = f'''
|
| 127 |
-
<html>
|
| 128 |
-
<head>
|
| 129 |
-
{css_styles}
|
| 130 |
-
</head>
|
| 131 |
-
<body>
|
| 132 |
-
{html_content}
|
| 133 |
-
</body>
|
| 134 |
-
</html>
|
| 135 |
-
'''
|
| 136 |
-
|
| 137 |
-
# Generate the PDF
|
| 138 |
-
pdf = BytesIO()
|
| 139 |
-
pisa_status = pisa.CreatePDF(html, dest=pdf)
|
| 140 |
-
if pisa_status.err:
|
| 141 |
-
return None
|
| 142 |
-
return pdf.getvalue()
|
| 143 |
-
|
| 144 |
# Streamlit app layout
|
| 145 |
st.title("Climate Policy Summary Tool")
|
| 146 |
|
|
@@ -171,18 +122,5 @@ if st.button("Generate") and api_key and uploaded_file:
|
|
| 171 |
file_name="qa_results.md",
|
| 172 |
mime="text/markdown"
|
| 173 |
)
|
| 174 |
-
|
| 175 |
-
# Convert markdown to PDF
|
| 176 |
-
pdf_bytes = md_to_pdf(markdown_text)
|
| 177 |
-
if pdf_bytes:
|
| 178 |
-
st.download_button(
|
| 179 |
-
label="Download Results as PDF",
|
| 180 |
-
data=pdf_bytes,
|
| 181 |
-
file_name="qa_results.pdf",
|
| 182 |
-
mime="application/pdf"
|
| 183 |
-
)
|
| 184 |
-
else:
|
| 185 |
-
st.error("Error generating PDF")
|
| 186 |
-
|
| 187 |
except Exception as e:
|
| 188 |
st.error(f"An error occurred: {e}")
|
|
|
|
| 1 |
import os
|
|
|
|
| 2 |
import streamlit as st
|
| 3 |
from io import BytesIO
|
| 4 |
from tempfile import NamedTemporaryFile
|
|
|
|
| 10 |
from langchain_community.vectorstores import FAISS
|
| 11 |
from langchain_openai import OpenAIEmbeddings
|
| 12 |
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
| 13 |
+
import re
|
| 14 |
+
|
| 15 |
+
# Function to remove code block markers from the answer
|
| 16 |
+
def remove_code_blocks(text):
|
| 17 |
+
# Remove starting and ending code block markers
|
| 18 |
+
code_block_pattern = r"^```(?:\w+)?\n(.*?)\n```$"
|
| 19 |
+
match = re.match(code_block_pattern, text, re.DOTALL)
|
| 20 |
+
if match:
|
| 21 |
+
return match.group(1).strip()
|
| 22 |
+
else:
|
| 23 |
+
return text
|
| 24 |
|
| 25 |
# Function to process PDF, run Q&A, and return results
|
| 26 |
def process_pdf(api_key, uploaded_file, questions_path, prompt_path, display_placeholder):
|
|
|
|
| 79 |
result = rag_chain.invoke({"input": question})
|
| 80 |
answer = result["answer"]
|
| 81 |
|
| 82 |
+
# Remove code block markers from the answer
|
| 83 |
+
answer = remove_code_blocks(answer)
|
| 84 |
|
| 85 |
+
qa_text = f"### Question: {question}\n**Answer:**\n{answer}\n"
|
| 86 |
qa_results.append(qa_text)
|
| 87 |
# Update the placeholder with each new Q&A pair
|
| 88 |
display_placeholder.markdown("\n".join(qa_results), unsafe_allow_html=True)
|
|
|
|
| 92 |
|
| 93 |
return qa_results
|
| 94 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 95 |
# Streamlit app layout
|
| 96 |
st.title("Climate Policy Summary Tool")
|
| 97 |
|
|
|
|
| 122 |
file_name="qa_results.md",
|
| 123 |
mime="text/markdown"
|
| 124 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 125 |
except Exception as e:
|
| 126 |
st.error(f"An error occurred: {e}")
|