SyntaxHighlighter.py

# SyntaxHighlighter.py
# compile once at import time
import re
import tkinter as tk

_pattern_comment = re.compile(
    r"/\*[\s\S]*?\*/"         # /* … */
  r"|//.*?$"                  # // …
  r"|#.*?$"                   # # …
  r"|'''[\s\S]*?'''"          # ''' … '''
  r"|<!--[\s\S]*?-->",     # HTML comment
  re.MULTILINE
)
_pattern_string = re.compile(
    r'"(?:\\.|[^"\\])*"'         # "…"
    r"|'(?:\\.|[^'\\])*'"        # '…'
    r"|`(?:\\.|[^`\\])*`",       # `...` backtick
  re.MULTILINE
)
_pattern_number = re.compile(
    r"\b\d+(\.\d+)?\b" # numbers, including floats
)
# _pattern_operator = re.compile(
#     r"(\+\+|--|\+|-|\*|\/|%|->|%|==|!=|>=|<=|>|<|!==|<>|&&|\|\||@|=|\+=|-=|\*=|===|%=)"
# )

class SyntaxHighlighter:
    """Attach to a Tk.Text widget to color comments and strings."""
    def __init__(self, text_widget: tk.Text):
        self.text = text_widget
        # set up tags - these colors meant for DARK background
        self.text.tag_configure('comment', foreground='gray')
        self.text.tag_configure('string',  foreground='goldenrod3')
        self.text.tag_configure('number',  foreground='LightGoldenrod3')
        # self.text.tag_configure('operator',foreground='cyan')

        # bind change event
        self.text.bind('<<Modified>>', self._on_modified)

    def _on_modified(self, event=None):
        if self.text.edit_modified():
            self._highlight_all()
            # reset modified flag
            self.text.edit_modified(False)

    def _highlight_all(self):
        """Remove old tags and re‐apply to entire content."""
        content = self.text.get('1.0', 'end-1c')
        # clear
        self.text.tag_remove('comment', '1.0', 'end')
        self.text.tag_remove('string',  '1.0', 'end')
        self.text.tag_remove('number',  '1.0', 'end')
        # self.text.tag_remove('operator','1.0', 'end')
        # apply strings first, comments after so that comments override strings
        for m in _pattern_string.finditer(content):
            self._apply_tag('string', m.start(), m.end())
        for m in _pattern_comment.finditer(content):
            self._apply_tag('comment', m.start(), m.end())
        for m in _pattern_number.finditer(content):
            self._apply_tag('number', m.start(), m.end())
        # for m in _pattern_operator.finditer(content):
        #     self._apply_tag('operator', m.start(), m.end())

    def _apply_tag(self, tag: str, start_idx: int, end_idx: int):
        """Convert character‐offset indices to Tk indices and add tag."""
        start = self.text.index(f'1.0+{start_idx}c')
        end   = self.text.index(f'1.0+{end_idx}c')
        self.text.tag_add(tag, start, end)