Legal_AI_Agent / components /document_viewer.py
cryogenic22's picture
Update components/document_viewer.py
68a9ffc verified
import streamlit as st
from typing import Dict, List, Optional
import pandas as pd
from datetime import datetime
import json
import base64
from pathlib import Path
import pypdf
from PIL import Image
import io
class DocumentViewer:
def __init__(self, case_manager, document_processor):
"""Initialize DocumentViewer with required components."""
self.case_manager = case_manager
self.document_processor = document_processor
# Initialize session state for viewer
if 'current_document' not in st.session_state:
st.session_state.current_document = None
if 'current_page' not in st.session_state:
st.session_state.current_page = 1
if 'chunk_view' not in st.session_state:
st.session_state.chunk_view = False
def render(self, case_id: str):
"""Render enhanced document viewer interface."""
case = self.case_manager.get_case(case_id)
if not case:
st.error("Case not found")
return
# Add custom CSS for document viewer
self._add_viewer_styles()
# Document selection and viewing area
col1, col2 = st.columns([1, 3])
with col1:
self._render_document_list(case)
with col2:
if st.session_state.current_document:
self._render_document_viewer(st.session_state.current_document)
def _add_viewer_styles(self):
"""Add custom CSS styles for document viewer."""
st.markdown("""
<style>
.document-list {
border-right: 1px solid #ddd;
padding-right: 1rem;
height: 100%;
}
.document-viewer {
padding: 1rem;
background: white;
border-radius: 5px;
box-shadow: 0 2px 4px rgba(0,0,0,0.1);
}
.chunk-card {
background: #f8f9fa;
border-left: 3px solid #2196F3;
padding: 1rem;
margin: 0.5rem 0;
border-radius: 0 5px 5px 0;
}
.document-controls {
display: flex;
justify-content: space-between;
align-items: center;
padding: 1rem;
background: #f8f9fa;
border-radius: 5px;
margin-bottom: 1rem;
}
.page-viewer {
text-align: center;
padding: 1rem;
background: #fff;
border: 1px solid #ddd;
border-radius: 5px;
}
.metadata-section {
background: #f8f9fa;
padding: 1rem;
border-radius: 5px;
margin-top: 1rem;
}
.highlight {
background-color: #fff3cd;
padding: 0.2rem;
border-radius: 3px;
}
</style>
""", unsafe_allow_html=True)
def _render_document_list(self, case: Dict):
"""Render enhanced document list with filtering and sorting."""
st.markdown("### ๐Ÿ“‘ Documents")
# Search and filter options
search = st.text_input("๐Ÿ” Search documents", key="doc_search")
sort_by = st.selectbox(
"Sort by",
["Recent first", "Name A-Z", "Name Z-A", "Size", "Type"],
key="doc_sort"
)
# Get and sort documents
docs = case['documents']
if search:
docs = [d for d in docs if search.lower() in d['title'].lower()]
if sort_by == "Recent first":
docs.sort(key=lambda x: x['added_at'], reverse=True)
elif sort_by == "Name A-Z":
docs.sort(key=lambda x: x['title'])
elif sort_by == "Name Z-A":
docs.sort(key=lambda x: x['title'], reverse=True)
# Display document list
for doc in docs:
doc_container = st.container()
with doc_container:
if st.button(
f"๐Ÿ“„ {doc['title'][:30]}...",
key=f"select_{doc['id']}",
help=f"Click to view {doc['title']}"
):
st.session_state.current_document = doc
st.session_state.current_page = 1
st.session_state.chunk_view = False
def _render_document_viewer(self, document: Dict):
"""Render enhanced document viewer with multiple view modes."""
st.markdown(f"### ๐Ÿ“„ {document['title']}")
# Document controls
col1, col2, col3 = st.columns([2, 1, 1])
with col1:
view_mode = st.radio(
"View Mode",
["Document", "Chunks", "Analysis"],
horizontal=True,
key="view_mode"
)
with col2:
if st.button("Download", key="download_doc"):
self._handle_download(document)
with col3:
if st.button("Process Again", key="reprocess_doc"):
self._reprocess_document(document)
# Render selected view mode
if view_mode == "Document":
self._render_document_content(document)
elif view_mode == "Chunks":
self._render_chunk_view(document)
else:
self._render_analysis_view(document)
# Document metadata
with st.expander("๐Ÿ“‹ Document Metadata", expanded=False):
self._render_metadata(document)
def _render_document_content(self, document: Dict):
"""Render document content with page navigation for PDFs."""
content = document.get('content', '')
file_type = document.get('metadata', {}).get('file_type', '').lower()
if file_type == 'pdf':
self._render_pdf_viewer(document)
elif file_type in ['png', 'jpg', 'jpeg']:
self._render_image_viewer(document)
else:
st.text_area(
"Content",
value=content[:10000] + ("..." if len(content) > 10000 else ""),
height=400,
disabled=True
)
def _render_pdf_viewer(self, document: Dict):
"""Render PDF viewer with navigation controls."""
pdf_path = document.get('file_path')
if not pdf_path:
st.error("PDF file path not found")
return
try:
reader = pypdf.PdfReader(pdf_path)
num_pages = len(reader.pages)
# Navigation controls
col1, col2, col3 = st.columns([1, 2, 1])
with col1:
if st.button("โ—€๏ธ Previous") and st.session_state.current_page > 1:
st.session_state.current_page -= 1
with col2:
st.session_state.current_page = st.select_slider(
"Page",
range(1, num_pages + 1),
value=st.session_state.current_page
)
with col3:
if st.button("Next โ–ถ๏ธ") and st.session_state.current_page < num_pages:
st.session_state.current_page += 1
# Display current page
page = reader.pages[st.session_state.current_page - 1]
text = page.extract_text()
st.text_area("Page Content", value=text, height=400, disabled=True)
except Exception as e:
st.error(f"Error displaying PDF: {str(e)}")
def _render_chunk_view(self, document: Dict):
"""Render document chunks with enhanced visualization."""
chunks = document.get('chunks', [])
if not chunks:
st.warning("No chunks available. Try reprocessing the document.")
return
# Chunk navigation and filtering
search = st.text_input("๐Ÿ” Search in chunks", key="chunk_search")
# Display chunks
for idx, chunk in enumerate(chunks):
if not search or search.lower() in chunk['text'].lower():
with st.container():
st.markdown(f"""
<div class="chunk-card">
<strong>Chunk {idx + 1}</strong>
<p>{self._highlight_text(chunk['text'], search) if search else chunk['text']}</p>
<div style="font-size: 0.8em; color: #666;">
Characters: {len(chunk['text'])} |
Location: {chunk.get('start_idx', 'N/A')} - {chunk.get('end_idx', 'N/A')}
</div>
</div>
""", unsafe_allow_html=True)
def _render_analysis_view(self, document: Dict):
"""Render document analysis with insights."""
metadata = document.get('metadata', {})
# Document statistics
st.markdown("### ๐Ÿ“Š Document Analysis")
col1, col2 = st.columns(2)
with col1:
st.metric("Total Pages", metadata.get('page_count', 'N/A'))
st.metric("Word Count", metadata.get('word_count', 'N/A'))
with col2:
st.metric("Chunks", len(document.get('chunks', [])))
st.metric("File Size", self._format_file_size(metadata.get('file_size', 0)))
# Entity extraction
if 'entities' in metadata:
st.markdown("### ๐Ÿท๏ธ Named Entities")
entities_df = pd.DataFrame(metadata['entities'])
st.dataframe(entities_df)
# Key phrases
if 'key_phrases' in metadata:
st.markdown("### ๐Ÿ”‘ Key Phrases")
st.write(", ".join(metadata['key_phrases']))
def _render_metadata(self, document: Dict):
"""Render detailed document metadata."""
metadata = document.get('metadata', {})
col1, col2 = st.columns(2)
with col1:
st.markdown("**Basic Information**")
st.write(f"Type: {metadata.get('file_type', 'Unknown')}")
st.write(f"Added: {document.get('added_at', 'Unknown')}")
st.write(f"Size: {self._format_file_size(metadata.get('file_size', 0))}")
with col2:
st.markdown("**Processing Information**")
st.write(f"Chunks: {len(document.get('chunks', []))}")
st.write(f"Processing Status: {metadata.get('processing_status', 'Unknown')}")
st.write(f"Last Updated: {metadata.get('updated_at', 'Unknown')}")
def _highlight_text(self, text: str, search: str) -> str:
"""Highlight search terms in text."""
if not search:
return text
return text.replace(
search,
f'<span class="highlight">{search}</span>'
)
def _format_file_size(self, size_bytes: int) -> str:
"""Format file size in human-readable format."""
for unit in ['B', 'KB', 'MB', 'GB']:
if size_bytes < 1024:
return f"{size_bytes:.1f} {unit}"
size_bytes /= 1024
return f"{size_bytes:.1f} TB"
def _handle_download(self, document: Dict):
"""Handle document download with progress bar."""
try:
file_path = document.get('file_path')
if file_path and Path(file_path).exists():
with open(file_path, 'rb') as f:
data = f.read()
st.download_button(
"Download File",
data,
file_name=document['title'],
mime=document.get('metadata', {}).get('mime_type', 'application/octet-stream')
)
else:
st.error("File not found")
except Exception as e:
st.error(f"Error downloading file: {str(e)}")
def _reprocess_document(self, document: Dict):
"""Reprocess document to update chunks and metadata."""
try:
with st.spinner("Reprocessing document..."):
# Get file path
file_path = document.get('file_path')
if not file_path:
st.error("File not found")
return
# Reprocess document
with open(file_path, 'rb') as f:
text, chunks, metadata = self.document_processor.process_and_tag_document(f)
# Update document in case manager
document.update({
'content': text,
'chunks': chunks,
'metadata': {**document.get('metadata', {}), **metadata},
'updated_at': datetime.now().isoformat()
})
self.case_manager.update_document(document['id'], document)
st.success("Document reprocessed successfully")
st.experimental_rerun()
except Exception as e:
st.error(f"Error reprocessing document: {str(e)}")
def render_chunk_references(self, references: List[Dict]):
"""Render referenced chunks with enhanced visualization."""
st.markdown("### ๐Ÿ“Ž Referenced Chunks")
for ref in references:
with st.container():
st.markdown(f"""
<div class="chunk-card">
<strong>{ref['title']}</strong> (Chunk {ref['chunk_id']})
<p>{ref['snippet']}</p>
<div style="font-size: 0.8em; color: #666;">
Page: {ref.get('page', 'N/A')} |
Confidence: {ref.get('confidence', 'N/A')}
</div>
</div>
""", unsafe_allow_html=True)