parses a patch file and adds highlighting for specific changes made within a block of text. May pass large strings into SequenceMatcher, which scales very poorly.

highlighted-diff.py

def _format_diff_strings(before: str, after: str) -> str:
    before_lines = before.splitlines()
    after_lines = after.splitlines()
    diff = difflib.unified_diff(before_lines, after_lines, lineterm="")
    output = ""
    removed_lines = []
    added_lines = []

    def _format_added_removed():
        # given the current state of the `added` and `removed` variables, add text to the `output`
        # variable. if both added and removed have text, then there's a replacement, which gets
        # special formatting
        nonlocal added_lines, removed_lines, output
        added = "\n".join(added_lines)
        removed = "\n".join(removed_lines)

        def _format_diff_replacement(before: str, after: str) -> str:
            # print before and after lines highlight specific changes on those lines based
            # on the longest common substrings

            before = re.sub(r"(^|\n)-", r"\1", before)
            after = re.sub(r"(^|\n)\+", r"\1", after)

            def _colorize_str(x: str, color: str) -> str:
                return ANSI_RESET + color + x + ANSI_RESET

            matcher = difflib.SequenceMatcher(None, before, after)
            before_output = ""
            after_output = ""
            for tag, i1, i2, j1, j2 in matcher.get_opcodes():
                before_substr = before[i1:i2]
                after_substr = after[j1:j2]
                if tag == "equal":
                    before_output += _colorize_str(before_substr, ANSI_RED)
                    after_output += _colorize_str(after_substr, ANSI_GREEN)
                elif tag == "delete":
                    before_output += _colorize_str(before_substr, ANSI_BG_RED)
                elif tag == "insert":
                    after_output += _colorize_str(after_substr, ANSI_BG_GREEN)
                elif tag == "replace":
                    before_output += _colorize_str(before_substr, ANSI_BG_RED)
                    after_output += _colorize_str(after_substr, ANSI_BG_GREEN)

            before_output = re.sub(r"(^|\n)", rf"\1{ANSI_RED}-{ANSI_RESET}", before_output)
            after_output = re.sub(r"(^|\n)", rf"\1{ANSI_GREEN}+{ANSI_RESET}", after_output)
            return before_output + "\n" + after_output + "\n"

        if added or removed:
            if added and not removed:
                output += f"{ANSI_RESET}{ANSI_GREEN}{added}{ANSI_RESET}\n"
            elif removed and not added:
                output += f"{ANSI_RESET}{ANSI_RED}{removed}{ANSI_RESET}\n"
            elif added and removed:
                output += _format_diff_replacement(removed, added)
            del added_lines, removed_lines
            added_lines = []
            removed_lines = []

    # for line in diff:
    delme = [x for x in diff]
    for line in delme:
        if line.strip() in ["---", "+++"]:
            continue
        if line.startswith("+"):
            added_lines.append(line)
        elif line.startswith("-"):
            removed_lines.append(line)
        else:
            # print any cached added and removed lines, then reset cached lines, then print line
            _format_added_removed()
            if line.startswith("@@") and line.endswith("@@"):
                output += stringc(line, C.COLOR_DIFF_LINES)
            else:
                output += line
            output += "\n"
    _format_added_removed()
    return output