Sanitized regex markers (#37)

dahlia/constants.py

@@ -131,3 +131,5 @@
     8: "\033[48;5;{}m",
     24: "\033[48;2;{}m",
 }
+
+REGEX_BREAKING_MARKERS = set("^$.|()[]\\*+?")

dahlia/utils.py

@@ -3,7 +3,12 @@
 import re
 from typing import TYPE_CHECKING
 
-from dahlia.constants import ANSI_REGEX, CODE_REGEXES, NO_GROUP_CODES
+from dahlia.constants import (
+    ANSI_REGEX,
+    CODE_REGEXES,
+    NO_GROUP_CODES,
+    REGEX_BREAKING_MARKERS,
+)
 
 if TYPE_CHECKING:
     from collections.abc import Iterator
@@ -52,4 +57,6 @@ def _with_marker(marker: str) -> list[re.Pattern[str]]:
     if len(marker) != 1:
         msg = "The marker has to be a single character"
         raise ValueError(msg)
+    if marker in REGEX_BREAKING_MARKERS:
+        marker = "\\" + marker
     return [re.compile(marker + i) for i in CODE_REGEXES]

tests/test_api.py

@@ -2,6 +2,7 @@
 
 import pytest
 
+from dahlia.constants import REGEX_BREAKING_MARKERS
 from dahlia.lib import Dahlia, Depth
 
 
@@ -46,13 +47,21 @@ def test_conversion(depth: Depth, string: str, expected: str) -> None:
         ("x", "&ee§ee§§_4x"),
     ],
 )
-def test_markers(marker: str, expected: str) -> None:
+def test_marker_clashing(marker: str, expected: str) -> None:
     assert (
         Dahlia(marker=marker, auto_reset=False, depth=Depth.LOW).convert("&ee§ee§§_4x")
         == expected
     )
 
 
+@pytest.mark.parametrize("marker", REGEX_BREAKING_MARKERS)
+def test_regex_markers(marker: str) -> None:
+    assert (
+        Dahlia(depth=Depth.LOW, marker=marker).convert(f"{marker}4xe5")
+        == "\x1b[31mxe5\x1b[0m"
+    )
+
+
 @pytest.mark.parametrize("marker", ["", "&&"])
 def test_invalid_marker(marker: str) -> None:
     with pytest.raises(ValueError, match="The marker has to be a single character"):