Merge pull request #5 from robert-mcdermott/win-encoding-bug

Fixed Issue #4 (charmap) issue on Windows systems

Fix Windows encoding issues in knowledge graph visualization

## Issue
Windows users were experiencing UnicodeEncodeError when generating knowledge graphs with certain input files. The error occurred during HTML generation with the message: 'charmap' codec can't encode characters in position 263607-263621: character maps to <undefined>

## Root Cause
The PyVis library was using the system's default encoding when writing HTML files. On Windows, this is typically 'cp1252' (Windows-1252), which has limited Unicode support compared to UTF-8 that's used on Mac/Linux systems.

## Fix
Modified the [_save_and_modify_html](cci:1://file:///Users/rmcdermo/mycode/ai-knowledge-graph/src/knowledge_graph/visualization.py:324:0-362:44) function to:
1. Use PyVis's internal HTML generation capabilities instead of letting it write to a file
2. Directly access the generated HTML content in memory
3. Write the file ourselves with explicit UTF-8 encoding

This bypasses the encoding issue completely by taking control of the file writing process with proper UTF-8 handling, ensuring consistent behavior across all operating systems.

## Testing
Confirmed working on both Mac OS and Windows environments with the same input data.

Author: robert-mcdermott

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ai-knowledge-graph"
-version = "0.6.0"
+version = "0.6.1"
 description = "Takes a text document and generates an interactive knowledge graph"
 authors = [{ name = "Robert McDermott", email = "[email protected]" }, ]
 readme = "README.md"

src/knowledge_graph/visualization.py

@@ -324,12 +324,15 @@ def _get_visualization_options(edge_smooth=False):
 
 def _save_and_modify_html(net, output_file, community_count, all_nodes, triples):
     """Save the network as HTML and modify with custom template."""
-    # Save the network as HTML
-    net.save_graph(output_file)
+    # Instead of letting PyVis write to a file, we'll access its HTML directly
+    # and write it ourselves with explicit UTF-8 encoding
 
-    # Read the generated HTML
-    with open(output_file, 'r', encoding='utf-8') as f:
-        html = f.read()
+    # Generate the HTML content
+    # This happens internally in PyVis without writing to a file
+    net.generate_html()
+    
+    # Get the HTML from PyVis's internal html attribute
+    html = net.html
 
     # Add our custom controls by replacing the div with our template
     html = html.replace('<div id="mynetwork" class="card-body"></div>', _load_html_template())
@@ -341,7 +344,7 @@ def _save_and_modify_html(net, output_file, community_count, all_nodes, triples)
     # Replace the other h1 with our enhanced title
     html = html.replace('<h1></h1>', f'<h1>Knowledge Graph - {len(all_nodes)} Nodes, {len(triples)} Relationships, {community_count} Communities</h1>')
 
-    # Write back the modified HTML
+    # Write the HTML directly to the output file with explicit UTF-8 encoding
     with open(output_file, 'w', encoding='utf-8') as f:
         f.write(html)