directly to avoid highlight.js auto-detection escaped_body = html.escape(body) return f'''

{tool_svg_small} {html.escape(header)}

{escaped_body}

''' def build_thinking_block(thinking_content, message_id, has_remaining_content, thinking_index=0): """Build HTML for a thinking block.""" if thinking_content is None: return None # Process the thinking content through markdown thinking_html = process_markdown_content(thinking_content) # Generate unique ID for the thinking block block_id = f"thinking-{message_id}-{thinking_index}" # Check if thinking is complete or still in progress is_streaming = not has_remaining_content title_text = "Thinking..." if is_streaming else "Thought" return f'''

{info_svg_small} {title_text}

{thinking_html}

''' def build_main_content_block(content): """Build HTML for the main content block.""" if not content: return "" return process_markdown_content(content) def process_markdown_content(string): """ Process a string through the markdown conversion pipeline. Uses robust manual parsing to ensure correct LaTeX and Code Block rendering. """ if not string: return "" # Define unique placeholders for LaTeX asterisks and underscores LATEX_ASTERISK_PLACEHOLDER = "LATEXASTERISKPLACEHOLDER" LATEX_UNDERSCORE_PLACEHOLDER = "LATEXUNDERSCOREPLACEHOLDER" def protect_asterisks_underscores_in_latex(match): """A replacer function for re.sub to protect asterisks and underscores in multiple LaTeX formats.""" # Check which delimiter group was captured if match.group(1) is not None: # Content from $$...$$ content = match.group(1) modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER) modified_content = modified_content.replace('_', LATEX_UNDERSCORE_PLACEHOLDER) return f'{modified_content}' elif match.group(2) is not None: # Content from \[...\] content = match.group(2) modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER) modified_content = modified_content.replace('_', LATEX_UNDERSCORE_PLACEHOLDER) return f'\\[{modified_content}\\]' elif match.group(3) is not None: # Content from $...$ content = match.group(3) modified_content = content.replace('*', LATEX_ASTERISK_PLACEHOLDER) modified_content = modified_content.replace('_', LATEX_UNDERSCORE_PLACEHOLDER) return f'\${modified_content}\$' return match.group(0) # Fallback # Make \[ \] LaTeX equations inline pattern = r'^\s*\\\[\s*\n([\s\S]*?)\n\s*\\\]\s*$' replacement = r'\\[ \1 \\]' string = re.sub(pattern, replacement, string, flags=re.MULTILINE) # Escape backslashes string = string.replace('\\', '\\\\') # Quote to string = replace_quotes(string) # Blockquote string = re.sub(r'(^|[\n])>', r'\1>', string) pattern = re.compile(r'\\begin{blockquote}(.*?)\\end{blockquote}', re.DOTALL) string = pattern.sub(replace_blockquote, string) # Code block standardization string = string.replace('\\begin{code}', '```') string = string.replace('\\end{code}', '```') string = string.replace('\\begin{align*}', '$$') string = string.replace('\\end{align*}', '$$') string = string.replace('\\begin{align}', '$$') string = string.replace('\\end{align}', '$$') string = string.replace('\\begin{equation}', '$$') string = string.replace('\\end{equation}', '$$') string = string.replace('\\begin{equation*}', '$$') string = string.replace('\\end{equation*}', '$$') string = re.sub(r"(.)```", r"\1\n```", string) # Protect asterisks and underscores within all LaTeX blocks before markdown conversion latex_pattern = re.compile(r'((?:^|[\r\n\s])\$\$[^`]*?\$\$)|\\\[(.*?)\\\]|\\$(.*?)\\$', re.DOTALL) string = latex_pattern.sub(protect_asterisks_underscores_in_latex, string) result = '' is_code = False is_latex = False # Manual line iteration for robust structure parsing for line in string.split('\n'): stripped_line = line.strip() if stripped_line.startswith('```'): is_code = not is_code elif stripped_line.startswith('$$') and (stripped_line == "$$" or not stripped_line.endswith('$$')): is_latex = not is_latex elif stripped_line.endswith('$$'): is_latex = False elif stripped_line.startswith('\\\\[') and not stripped_line.endswith('\\\\]'): is_latex = True elif stripped_line.startswith('\\\\]'): is_latex = False elif stripped_line.endswith('\\\\]'): is_latex = False result += line # Don't add an extra \n for code, LaTeX, or tables if is_code or is_latex or line.startswith('|'): result += '\n' # Also don't add an extra \n for lists elif stripped_line.startswith('-') or stripped_line.startswith('*') or stripped_line.startswith('+') or stripped_line.startswith('>') or re.match(r'\d+\.', stripped_line): result += ' \n' else: result += ' \n' result = result.strip() if is_code: result += '\n```' # Unfinished code block # Unfinished list, like "\n1.". A |delete| string is added and then # removed to force a

. list_item_pattern = r'(\n\d+\.?|\n\s*[-*+]\s*([*_~]{1,3})?)$' if re.search(list_item_pattern, result): delete_str = '|delete|' if re.search(r'(\d+\.?)$', result) and not result.endswith('.'): result += '.' # Add the delete string after the list item result = re.sub(list_item_pattern, r'\g<1> ' + delete_str, result) # Convert to HTML using markdown html_output = markdown.markdown(result, extensions=['fenced_code', 'tables', SaneListExtension()]) # Remove the delete string from the HTML output pos = html_output.rfind(delete_str) if pos > -1: html_output = html_output[:pos] + html_output[pos + len(delete_str):] else: # Convert to HTML using markdown html_output = markdown.markdown(result, extensions=['fenced_code', 'tables', SaneListExtension()]) # Restore the LaTeX asterisks and underscores after markdown conversion html_output = html_output.replace(LATEX_ASTERISK_PLACEHOLDER, '*') html_output = html_output.replace(LATEX_UNDERSCORE_PLACEHOLDER, '_') # Remove extra newlines before html_output = re.sub(r'\s*', '', html_output) # Unescape code blocks pattern = re.compile(r']*>(.*?)', re.DOTALL) html_output = pattern.sub(lambda x: html.unescape(x.group()), html_output) # Unescape backslashes html_output = html_output.replace('\\\\', '\\') # Wrap tables in a scrollable div html_output = html_output.replace('', '
').replace('
', '') return html_output @functools.lru_cache(maxsize=None) def convert_to_markdown(string, message_id=None): """ Convert a string to markdown HTML with support for multiple block types. Blocks are assembled in order: thinking, main content, etc. """ if not string: return "" # Use a default message ID if none provided if message_id is None: message_id = "unknown" # Find tool call blocks by position, then process the text segments # between them using extract_thinking_block (which supports all # THINKING_FORMATS, including end-only variants like Qwen's). tool_call_pattern = re.compile(r'(.*?)\n(.*?)\n', re.DOTALL) tool_calls = list(tool_call_pattern.finditer(string)) if not tool_calls: # No tool calls — use original single-pass extraction thinking_content, remaining_content = extract_thinking_block(string) blocks = [] thinking_html = build_thinking_block(thinking_content, message_id, bool(remaining_content)) if thinking_html: blocks.append(thinking_html) main_html = build_main_content_block(remaining_content) if main_html: blocks.append(main_html) return ''.join(blocks) # Split string into text segments around tool_call blocks and # run extract_thinking_block on each segment for full format support. html_parts = [] last_end = 0 tool_idx = 0 think_idx = 0 def process_text_segment(text, is_last_segment): """Process a text segment between tool_call blocks for thinking content.""" nonlocal think_idx if not text.strip(): return while text.strip(): thinking_content, remaining = extract_thinking_block(text) if thinking_content is None: break has_remaining = bool(remaining.strip()) or not is_last_segment html_parts.append(build_thinking_block(thinking_content, message_id, has_remaining, think_idx)) think_idx += 1 text = remaining if text.strip(): html_parts.append(process_markdown_content(text)) for tc in tool_calls: # Process text before this tool_call process_text_segment(string[last_end:tc.start()], is_last_segment=False) # Add tool call accordion header = tc.group(1).strip() body = tc.group(2).strip() html_parts.append(build_tool_call_block(header, body, message_id, tool_idx)) tool_idx += 1 last_end = tc.end() # Process text after the last tool_call process_text_segment(string[last_end:], is_last_segment=True) return ''.join(html_parts) def convert_to_markdown_wrapped(string, message_id=None, use_cache=True): ''' Used to avoid caching convert_to_markdown calls during streaming. ''' if use_cache: return convert_to_markdown(string, message_id=message_id) return convert_to_markdown.__wrapped__(string, message_id=message_id) def generate_basic_html(string): convert_to_markdown.cache_clear() string = convert_to_markdown(string) string = f'

{string}

{timestamp}

{current_idx + 1}/{len(versions)}

' f'{action_buttons}' f'{info_message}' f'

' else: output = "" def create_message(role, content, raw_content): """Inner function that captures variables from outer scope.""" class_name = "user-message" if role == "user" else "assistant-message" # Get role-specific data timestamp = format_message_timestamp(history, role, i) attachments = format_message_attachments(history, role, i) # Create info button if timestamp exists info_message = "" if timestamp: tooltip_text = get_message_tooltip(history, role, i) info_message = info_button.replace('title="message"', f'title="{html.escape(tooltip_text)}"') return ( f'

' f'

{content}

' f'{attachments}' f'{actions_html(history, i, role, info_message)}' f'

' f'

' ) # Determine range start_idx = len(history['visible']) - 1 if last_message_only else 0 end_idx = len(history['visible']) for i in range(start_idx, end_idx): row_visible = history['visible'][i] row_internal = history['internal'][i] # Convert content if last_message_only: converted_visible = [None, convert_to_markdown_wrapped(row_visible[1], message_id=i, use_cache=i != len(history['visible']) - 1)] else: converted_visible = [convert_to_markdown_wrapped(entry, message_id=i, use_cache=i != len(history['visible']) - 1) for entry in row_visible] # Generate messages if not last_message_only and converted_visible[0]: output += create_message("user", converted_visible[0], row_internal[0]) output += create_message("assistant", converted_visible[1], row_internal[1]) if not last_message_only: output += "

' else: output = "" img_bot = get_character_image_with_cache_buster() def create_message(role, content, raw_content): """Inner function for CAI-style messages.""" circle_class = "circle-you" if role == "user" else "circle-bot" name = name1 if role == "user" else name2 # Get role-specific data timestamp = format_message_timestamp(history, role, i, tooltip_include_timestamp=False) attachments = format_message_attachments(history, role, i) # Get appropriate image if role == "user": img = (f'

' if (shared.user_data_dir / "cache" / "pfp_me.png").exists() else '') else: img = img_bot return ( f'

' f'

{img}

' f'

{name}{timestamp}

' f'

{content}

' f'{attachments}' f'{actions_html(history, i, role)}' f'

' f'

{time_greeting()} How can I help you today?

{greeting}