]> git.madduck.net Git - etc/vim.git/blobdiff - src/black/trans.py

madduck's git repository

Every one of the projects in this repository is available at the canonical URL git://git.madduck.net/madduck/pub/<projectpath> — see each project's metadata for the exact URL.

All patches and comments are welcome. Please squash your changes to logical commits before using git-format-patch and git-send-email to patches@git.madduck.net. If you'd read over the Git project's submission guidelines and adhered to them, I'd be especially grateful.

SSH access, as well as push access can be individually arranged.

If you use my repositories frequently, consider adding the following snippet to ~/.gitconfig and using the third clone URL listed for each project:

[url "git://git.madduck.net/madduck/"]
  insteadOf = madduck:

Assignment to env var in Jupyter Notebook doesn't round-trip (#2642)
[etc/vim.git] / src / black / trans.py
index 7ecc31d6d31fbb9d0b3db8c027dd1094f936bbf0..d918ef111a21993b517f10b3ba9c30fad05e1a8a 100644 (file)
@@ -8,6 +8,7 @@ import regex as re
 from typing import (
     Any,
     Callable,
 from typing import (
     Any,
     Callable,
+    ClassVar,
     Collection,
     Dict,
     Iterable,
     Collection,
     Dict,
     Iterable,
@@ -15,17 +16,26 @@ from typing import (
     List,
     Optional,
     Sequence,
     List,
     Optional,
     Sequence,
+    Set,
     Tuple,
     TypeVar,
     Union,
 )
     Tuple,
     TypeVar,
     Union,
 )
+import sys
+
+if sys.version_info < (3, 8):
+    from typing_extensions import Final
+else:
+    from typing import Final
+
+from mypy_extensions import trait
 
 from black.rusty import Result, Ok, Err
 
 from black.mode import Feature
 from black.nodes import syms, replace_child, parent_type
 from black.nodes import is_empty_par, is_empty_lpar, is_empty_rpar
 
 from black.rusty import Result, Ok, Err
 
 from black.mode import Feature
 from black.nodes import syms, replace_child, parent_type
 from black.nodes import is_empty_par, is_empty_lpar, is_empty_rpar
-from black.nodes import CLOSING_BRACKETS, STANDALONE_COMMENT
+from black.nodes import OPENING_BRACKETS, CLOSING_BRACKETS, STANDALONE_COMMENT
 from black.lines import Line, append_leaves
 from black.brackets import BracketMatchError
 from black.comments import contains_pragma_comment
 from black.lines import Line, append_leaves
 from black.brackets import BracketMatchError
 from black.comments import contains_pragma_comment
@@ -61,7 +71,6 @@ def TErr(err_msg: str) -> Err[CannotTransform]:
     return Err(cant_transform)
 
 
     return Err(cant_transform)
 
 
-@dataclass  # type: ignore
 class StringTransformer(ABC):
     """
     An implementation of the Transformer protocol that relies on its
 class StringTransformer(ABC):
     """
     An implementation of the Transformer protocol that relies on its
@@ -89,9 +98,13 @@ class StringTransformer(ABC):
         as much as possible.
     """
 
         as much as possible.
     """
 
-    line_length: int
-    normalize_strings: bool
-    __name__ = "StringTransformer"
+    __name__: Final = "StringTransformer"
+
+    # Ideally this would be a dataclass, but unfortunately mypyc breaks when used with
+    # `abc.ABC`.
+    def __init__(self, line_length: int, normalize_strings: bool) -> None:
+        self.line_length = line_length
+        self.normalize_strings = normalize_strings
 
     @abstractmethod
     def do_match(self, line: Line) -> TMatchResult:
 
     @abstractmethod
     def do_match(self, line: Line) -> TMatchResult:
@@ -183,6 +196,7 @@ class CustomSplit:
     break_idx: int
 
 
     break_idx: int
 
 
+@trait
 class CustomSplitMapMixin:
     """
     This mixin class is used to map merged strings to a sequence of
 class CustomSplitMapMixin:
     """
     This mixin class is used to map merged strings to a sequence of
@@ -190,8 +204,10 @@ class CustomSplitMapMixin:
     the resultant substrings go over the configured max line length.
     """
 
     the resultant substrings go over the configured max line length.
     """
 
-    _Key = Tuple[StringID, str]
-    _CUSTOM_SPLIT_MAP: Dict[_Key, Tuple[CustomSplit, ...]] = defaultdict(tuple)
+    _Key: ClassVar = Tuple[StringID, str]
+    _CUSTOM_SPLIT_MAP: ClassVar[Dict[_Key, Tuple[CustomSplit, ...]]] = defaultdict(
+        tuple
+    )
 
     @staticmethod
     def _get_key(string: str) -> "CustomSplitMapMixin._Key":
 
     @staticmethod
     def _get_key(string: str) -> "CustomSplitMapMixin._Key":
@@ -242,7 +258,7 @@ class CustomSplitMapMixin:
         return key in self._CUSTOM_SPLIT_MAP
 
 
         return key in self._CUSTOM_SPLIT_MAP
 
 
-class StringMerger(CustomSplitMapMixin, StringTransformer):
+class StringMerger(StringTransformer, CustomSplitMapMixin):
     """StringTransformer that merges strings together.
 
     Requirements:
     """StringTransformer that merges strings together.
 
     Requirements:
@@ -411,7 +427,7 @@ class StringMerger(CustomSplitMapMixin, StringTransformer):
             and is_valid_index(next_str_idx)
             and LL[next_str_idx].type == token.STRING
         ):
             and is_valid_index(next_str_idx)
             and LL[next_str_idx].type == token.STRING
         ):
-            prefix = get_string_prefix(LL[next_str_idx].value)
+            prefix = get_string_prefix(LL[next_str_idx].value).lower()
             next_str_idx += 1
 
         # The next loop merges the string group. The final string will be
             next_str_idx += 1
 
         # The next loop merges the string group. The final string will be
@@ -431,7 +447,7 @@ class StringMerger(CustomSplitMapMixin, StringTransformer):
             num_of_strings += 1
 
             SS = LL[next_str_idx].value
             num_of_strings += 1
 
             SS = LL[next_str_idx].value
-            next_prefix = get_string_prefix(SS)
+            next_prefix = get_string_prefix(SS).lower()
 
             # If this is an f-string group but this substring is not prefixed
             # with 'f'...
 
             # If this is an f-string group but this substring is not prefixed
             # with 'f'...
@@ -541,7 +557,7 @@ class StringMerger(CustomSplitMapMixin, StringTransformer):
                 return TErr("StringMerger does NOT merge multiline strings.")
 
             num_of_strings += 1
                 return TErr("StringMerger does NOT merge multiline strings.")
 
             num_of_strings += 1
-            prefix = get_string_prefix(leaf.value)
+            prefix = get_string_prefix(leaf.value).lower()
             if "r" in prefix:
                 return TErr("StringMerger does NOT merge raw strings.")
 
             if "r" in prefix:
                 return TErr("StringMerger does NOT merge raw strings.")
 
@@ -738,6 +754,18 @@ class BaseStringSplitter(StringTransformer):
         * The target string is not a multiline (i.e. triple-quote) string.
     """
 
         * The target string is not a multiline (i.e. triple-quote) string.
     """
 
+    STRING_OPERATORS: Final = [
+        token.EQEQUAL,
+        token.GREATER,
+        token.GREATEREQUAL,
+        token.LESS,
+        token.LESSEQUAL,
+        token.NOTEQUAL,
+        token.PERCENT,
+        token.PLUS,
+        token.STAR,
+    ]
+
     @abstractmethod
     def do_splitter_match(self, line: Line) -> TMatchResult:
         """
     @abstractmethod
     def do_splitter_match(self, line: Line) -> TMatchResult:
         """
@@ -847,15 +875,15 @@ class BaseStringSplitter(StringTransformer):
                 p_idx -= 1
 
             P = LL[p_idx]
                 p_idx -= 1
 
             P = LL[p_idx]
-            if P.type == token.PLUS:
-                # WMA4 a space and a '+' character (e.g. `+ STRING`).
-                offset += 2
+            if P.type in self.STRING_OPERATORS:
+                # WMA4 a space and a string operator (e.g. `+ STRING` or `== STRING`).
+                offset += len(str(P)) + 1
 
             if P.type == token.COMMA:
                 # WMA4 a space, a comma, and a closing bracket [e.g. `), STRING`].
                 offset += 3
 
 
             if P.type == token.COMMA:
                 # WMA4 a space, a comma, and a closing bracket [e.g. `), STRING`].
                 offset += 3
 
-            if P.type in [token.COLON, token.EQUAL, token.NAME]:
+            if P.type in [token.COLON, token.EQUAL, token.PLUSEQUAL, token.NAME]:
                 # This conditional branch is meant to handle dictionary keys,
                 # variable assignments, 'return STRING' statement lines, and
                 # 'else STRING' ternary expression lines.
                 # This conditional branch is meant to handle dictionary keys,
                 # variable assignments, 'return STRING' statement lines, and
                 # 'else STRING' ternary expression lines.
@@ -914,15 +942,15 @@ class BaseStringSplitter(StringTransformer):
         return max_string_length
 
 
         return max_string_length
 
 
-class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
+class StringSplitter(BaseStringSplitter, CustomSplitMapMixin):
     """
     StringTransformer that splits "atom" strings (i.e. strings which exist on
     lines by themselves).
 
     Requirements:
     """
     StringTransformer that splits "atom" strings (i.e. strings which exist on
     lines by themselves).
 
     Requirements:
-        * The line consists ONLY of a single string (with the exception of a
-        '+' symbol which MAY exist at the start of the line), MAYBE a string
-        trailer, and MAYBE a trailing comma.
+        * The line consists ONLY of a single string (possibly prefixed by a
+        string operator [e.g. '+' or '==']), MAYBE a string trailer, and MAYBE
+        a trailing comma.
             AND
         * All of the requirements listed in BaseStringSplitter's docstring.
 
             AND
         * All of the requirements listed in BaseStringSplitter's docstring.
 
@@ -952,17 +980,17 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
         CustomSplit objects and add them to the custom split map.
     """
 
         CustomSplit objects and add them to the custom split map.
     """
 
-    MIN_SUBSTR_SIZE = 6
+    MIN_SUBSTR_SIZE: Final = 6
     # Matches an "f-expression" (e.g. {var}) that might be found in an f-string.
     # Matches an "f-expression" (e.g. {var}) that might be found in an f-string.
-    RE_FEXPR = r"""
+    RE_FEXPR: Final = r"""
     (?<!\{) (?:\{\{)* \{ (?!\{)
         (?:
             [^\{\}]
             | \{\{
             | \}\}
             | (?R)
     (?<!\{) (?:\{\{)* \{ (?!\{)
         (?:
             [^\{\}]
             | \{\{
             | \}\}
             | (?R)
-        )+?
-    (?<!\}) \} (?:\}\})* (?!\})
+        )+
+    \}
     """
 
     def do_splitter_match(self, line: Line) -> TMatchResult:
     """
 
     def do_splitter_match(self, line: Line) -> TMatchResult:
@@ -972,8 +1000,20 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
 
         idx = 0
 
 
         idx = 0
 
-        # The first leaf MAY be a '+' symbol...
-        if is_valid_index(idx) and LL[idx].type == token.PLUS:
+        # The first two leaves MAY be the 'not in' keywords...
+        if (
+            is_valid_index(idx)
+            and is_valid_index(idx + 1)
+            and [LL[idx].type, LL[idx + 1].type] == [token.NAME, token.NAME]
+            and str(LL[idx]) + str(LL[idx + 1]) == "not in"
+        ):
+            idx += 2
+        # Else the first leaf MAY be a string operator symbol or the 'in' keyword...
+        elif is_valid_index(idx) and (
+            LL[idx].type in self.STRING_OPERATORS
+            or LL[idx].type == token.NAME
+            and str(LL[idx]) == "in"
+        ):
             idx += 1
 
         # The next/first leaf MAY be an empty LPAR...
             idx += 1
 
         # The next/first leaf MAY be an empty LPAR...
@@ -1012,7 +1052,7 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
         is_valid_index = is_valid_index_factory(LL)
         insert_str_child = insert_str_child_factory(LL[string_idx])
 
         is_valid_index = is_valid_index_factory(LL)
         insert_str_child = insert_str_child_factory(LL[string_idx])
 
-        prefix = get_string_prefix(LL[string_idx].value)
+        prefix = get_string_prefix(LL[string_idx].value).lower()
 
         # We MAY choose to drop the 'f' prefix from substrings that don't
         # contain any f-expressions, but ONLY if the original f-string
 
         # We MAY choose to drop the 'f' prefix from substrings that don't
         # contain any f-expressions, but ONLY if the original f-string
@@ -1023,23 +1063,26 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
         )
 
         first_string_line = True
         )
 
         first_string_line = True
-        starts_with_plus = LL[0].type == token.PLUS
 
 
-        def line_needs_plus() -> bool:
-            return first_string_line and starts_with_plus
+        string_op_leaves = self._get_string_operator_leaves(LL)
+        string_op_leaves_length = (
+            sum([len(str(prefix_leaf)) for prefix_leaf in string_op_leaves]) + 1
+            if string_op_leaves
+            else 0
+        )
 
 
-        def maybe_append_plus(new_line: Line) -> None:
+        def maybe_append_string_operators(new_line: Line) -> None:
             """
             Side Effects:
             """
             Side Effects:
-                If @line starts with a plus and this is the first line we are
-                constructing, this function appends a PLUS leaf to @new_line
-                and replaces the old PLUS leaf in the node structure. Otherwise
-                this function does nothing.
+                If @line starts with a string operator and this is the first
+                line we are constructing, this function appends the string
+                operator to @new_line and replaces the old string operator leaf
+                in the node structure. Otherwise this function does nothing.
             """
             """
-            if line_needs_plus():
-                plus_leaf = Leaf(token.PLUS, "+")
-                replace_child(LL[0], plus_leaf)
-                new_line.append(plus_leaf)
+            maybe_prefix_leaves = string_op_leaves if first_string_line else []
+            for i, prefix_leaf in enumerate(maybe_prefix_leaves):
+                replace_child(LL[i], prefix_leaf)
+                new_line.append(prefix_leaf)
 
         ends_with_comma = (
             is_valid_index(string_idx + 1) and LL[string_idx + 1].type == token.COMMA
 
         ends_with_comma = (
             is_valid_index(string_idx + 1) and LL[string_idx + 1].type == token.COMMA
@@ -1054,7 +1097,7 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
             result = self.line_length
             result -= line.depth * 4
             result -= 1 if ends_with_comma else 0
             result = self.line_length
             result -= line.depth * 4
             result -= 1 if ends_with_comma else 0
-            result -= 2 if line_needs_plus() else 0
+            result -= string_op_leaves_length
             return result
 
         # --- Calculate Max Break Index (for string value)
             return result
 
         # --- Calculate Max Break Index (for string value)
@@ -1103,7 +1146,7 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
                 break_idx = csplit.break_idx
             else:
                 # Algorithmic Split (automatic)
                 break_idx = csplit.break_idx
             else:
                 # Algorithmic Split (automatic)
-                max_bidx = max_break_idx - 2 if line_needs_plus() else max_break_idx
+                max_bidx = max_break_idx - string_op_leaves_length
                 maybe_break_idx = self._get_break_idx(rest_value, max_bidx)
                 if maybe_break_idx is None:
                     # If we are unable to algorithmically determine a good split
                 maybe_break_idx = self._get_break_idx(rest_value, max_bidx)
                 if maybe_break_idx is None:
                     # If we are unable to algorithmically determine a good split
@@ -1124,21 +1167,32 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
 
             # --- Construct `next_value`
             next_value = rest_value[:break_idx] + QUOTE
 
             # --- Construct `next_value`
             next_value = rest_value[:break_idx] + QUOTE
+
+            # HACK: The following 'if' statement is a hack to fix the custom
+            # breakpoint index in the case of either: (a) substrings that were
+            # f-strings but will have the 'f' prefix removed OR (b) substrings
+            # that were not f-strings but will now become f-strings because of
+            # redundant use of the 'f' prefix (i.e. none of the substrings
+            # contain f-expressions but one or more of them had the 'f' prefix
+            # anyway; in which case, we will prepend 'f' to _all_ substrings).
+            #
+            # There is probably a better way to accomplish what is being done
+            # here...
+            #
+            # If this substring is an f-string, we _could_ remove the 'f'
+            # prefix, and the current custom split did NOT originally use a
+            # prefix...
             if (
             if (
-                # Are we allowed to try to drop a pointless 'f' prefix?
-                drop_pointless_f_prefix
-                # If we are, will we be successful?
-                and next_value != self._normalize_f_string(next_value, prefix)
+                next_value != self._normalize_f_string(next_value, prefix)
+                and use_custom_breakpoints
+                and not csplit.has_prefix
             ):
             ):
-                # If the current custom split did NOT originally use a prefix,
-                # then `csplit.break_idx` will be off by one after removing
+                # Then `csplit.break_idx` will be off by one after removing
                 # the 'f' prefix.
                 # the 'f' prefix.
-                break_idx = (
-                    break_idx + 1
-                    if use_custom_breakpoints and not csplit.has_prefix
-                    else break_idx
-                )
+                break_idx += 1
                 next_value = rest_value[:break_idx] + QUOTE
                 next_value = rest_value[:break_idx] + QUOTE
+
+            if drop_pointless_f_prefix:
                 next_value = self._normalize_f_string(next_value, prefix)
 
             # --- Construct `next_leaf`
                 next_value = self._normalize_f_string(next_value, prefix)
 
             # --- Construct `next_leaf`
@@ -1148,7 +1202,7 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
 
             # --- Construct `next_line`
             next_line = line.clone()
 
             # --- Construct `next_line`
             next_line = line.clone()
-            maybe_append_plus(next_line)
+            maybe_append_string_operators(next_line)
             next_line.append(next_leaf)
             string_line_results.append(Ok(next_line))
 
             next_line.append(next_leaf)
             string_line_results.append(Ok(next_line))
 
@@ -1169,7 +1223,7 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
         self._maybe_normalize_string_quotes(rest_leaf)
 
         last_line = line.clone()
         self._maybe_normalize_string_quotes(rest_leaf)
 
         last_line = line.clone()
-        maybe_append_plus(last_line)
+        maybe_append_string_operators(last_line)
 
         # If there are any leaves to the right of the target string...
         if is_valid_index(string_idx + 1):
 
         # If there are any leaves to the right of the target string...
         if is_valid_index(string_idx + 1):
@@ -1205,6 +1259,61 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
             last_line.comments = line.comments.copy()
             yield Ok(last_line)
 
             last_line.comments = line.comments.copy()
             yield Ok(last_line)
 
+    def _iter_nameescape_slices(self, string: str) -> Iterator[Tuple[Index, Index]]:
+        """
+        Yields:
+            All ranges of @string which, if @string were to be split there,
+            would result in the splitting of an \\N{...} expression (which is NOT
+            allowed).
+        """
+        # True - the previous backslash was unescaped
+        # False - the previous backslash was escaped *or* there was no backslash
+        previous_was_unescaped_backslash = False
+        it = iter(enumerate(string))
+        for idx, c in it:
+            if c == "\\":
+                previous_was_unescaped_backslash = not previous_was_unescaped_backslash
+                continue
+            if not previous_was_unescaped_backslash or c != "N":
+                previous_was_unescaped_backslash = False
+                continue
+            previous_was_unescaped_backslash = False
+
+            begin = idx - 1  # the position of backslash before \N{...}
+            for idx, c in it:
+                if c == "}":
+                    end = idx
+                    break
+            else:
+                # malformed nameescape expression?
+                # should have been detected by AST parsing earlier...
+                raise RuntimeError(f"{self.__class__.__name__} LOGIC ERROR!")
+            yield begin, end
+
+    def _iter_fexpr_slices(self, string: str) -> Iterator[Tuple[Index, Index]]:
+        """
+        Yields:
+            All ranges of @string which, if @string were to be split there,
+            would result in the splitting of an f-expression (which is NOT
+            allowed).
+        """
+        if "f" not in get_string_prefix(string).lower():
+            return
+
+        for match in re.finditer(self.RE_FEXPR, string, re.VERBOSE):
+            yield match.span()
+
+    def _get_illegal_split_indices(self, string: str) -> Set[Index]:
+        illegal_indices: Set[Index] = set()
+        iterators = [
+            self._iter_fexpr_slices(string),
+            self._iter_nameescape_slices(string),
+        ]
+        for it in iterators:
+            for begin, end in it:
+                illegal_indices.update(range(begin, end + 1))
+        return illegal_indices
+
     def _get_break_idx(self, string: str, max_break_idx: int) -> Optional[int]:
         """
         This method contains the algorithm that StringSplitter uses to
     def _get_break_idx(self, string: str, max_break_idx: int) -> Optional[int]:
         """
         This method contains the algorithm that StringSplitter uses to
@@ -1234,40 +1343,15 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
         assert is_valid_index(max_break_idx)
         assert_is_leaf_string(string)
 
         assert is_valid_index(max_break_idx)
         assert_is_leaf_string(string)
 
-        _fexpr_slices: Optional[List[Tuple[Index, Index]]] = None
+        _illegal_split_indices = self._get_illegal_split_indices(string)
 
 
-        def fexpr_slices() -> Iterator[Tuple[Index, Index]]:
-            """
-            Yields:
-                All ranges of @string which, if @string were to be split there,
-                would result in the splitting of an f-expression (which is NOT
-                allowed).
-            """
-            nonlocal _fexpr_slices
-
-            if _fexpr_slices is None:
-                _fexpr_slices = []
-                for match in re.finditer(self.RE_FEXPR, string, re.VERBOSE):
-                    _fexpr_slices.append(match.span())
-
-            yield from _fexpr_slices
-
-        is_fstring = "f" in get_string_prefix(string)
-
-        def breaks_fstring_expression(i: Index) -> bool:
+        def breaks_unsplittable_expression(i: Index) -> bool:
             """
             Returns:
                 True iff returning @i would result in the splitting of an
             """
             Returns:
                 True iff returning @i would result in the splitting of an
-                f-expression (which is NOT allowed).
+                unsplittable expression (which is NOT allowed).
             """
             """
-            if not is_fstring:
-                return False
-
-            for (start, end) in fexpr_slices():
-                if start <= i < end:
-                    return True
-
-            return False
+            return i in _illegal_split_indices
 
         def passes_all_checks(i: Index) -> bool:
             """
 
         def passes_all_checks(i: Index) -> bool:
             """
@@ -1291,7 +1375,7 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
                 is_space
                 and is_not_escaped
                 and is_big_enough
                 is_space
                 and is_not_escaped
                 and is_big_enough
-                and not breaks_fstring_expression(i)
+                and not breaks_unsplittable_expression(i)
             )
 
         # First, we check all indices BELOW @max_break_idx.
             )
 
         # First, we check all indices BELOW @max_break_idx.
@@ -1345,8 +1429,19 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter):
         else:
             return string
 
         else:
             return string
 
+    def _get_string_operator_leaves(self, leaves: Iterable[Leaf]) -> List[Leaf]:
+        LL = list(leaves)
+
+        string_op_leaves = []
+        i = 0
+        while LL[i].type in self.STRING_OPERATORS + [token.NAME]:
+            prefix_leaf = Leaf(LL[i].type, str(LL[i]).strip())
+            string_op_leaves.append(prefix_leaf)
+            i += 1
+        return string_op_leaves
 
 
-class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter):
+
+class StringParenWrapper(BaseStringSplitter, CustomSplitMapMixin):
     """
     StringTransformer that splits non-"atom" strings (i.e. strings that do not
     exist on lines by themselves).
     """
     StringTransformer that splits non-"atom" strings (i.e. strings that do not
     exist on lines by themselves).
@@ -1398,6 +1493,11 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter):
     def do_splitter_match(self, line: Line) -> TMatchResult:
         LL = line.leaves
 
     def do_splitter_match(self, line: Line) -> TMatchResult:
         LL = line.leaves
 
+        if line.leaves[-1].type in OPENING_BRACKETS:
+            return TErr(
+                "Cannot wrap parens around a line that ends in an opening bracket."
+            )
+
         string_idx = (
             self._return_match(LL)
             or self._else_match(LL)
         string_idx = (
             self._return_match(LL)
             or self._else_match(LL)
@@ -1665,9 +1765,10 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter):
                 right_leaves.pop()
 
             if old_parens_exist:
                 right_leaves.pop()
 
             if old_parens_exist:
-                assert (
-                    right_leaves and right_leaves[-1].type == token.RPAR
-                ), "Apparently, old parentheses do NOT exist?!"
+                assert right_leaves and right_leaves[-1].type == token.RPAR, (
+                    "Apparently, old parentheses do NOT exist?!"
+                    f" (left_leaves={left_leaves}, right_leaves={right_leaves})"
+                )
                 old_rpar_leaf = right_leaves.pop()
 
             append_leaves(string_line, line, right_leaves)
                 old_rpar_leaf = right_leaves.pop()
 
             append_leaves(string_line, line, right_leaves)
@@ -1725,20 +1826,20 @@ class StringParser:
         ```
     """
 
         ```
     """
 
-    DEFAULT_TOKEN = -1
+    DEFAULT_TOKEN: Final = 20210605
 
     # String Parser States
 
     # String Parser States
-    START = 1
-    DOT = 2
-    NAME = 3
-    PERCENT = 4
-    SINGLE_FMT_ARG = 5
-    LPAR = 6
-    RPAR = 7
-    DONE = 8
+    START: Final = 1
+    DOT: Final = 2
+    NAME: Final = 3
+    PERCENT: Final = 4
+    SINGLE_FMT_ARG: Final = 5
+    LPAR: Final = 6
+    RPAR: Final = 7
+    DONE: Final = 8
 
     # Lookup Table for Next State
 
     # Lookup Table for Next State
-    _goto: Dict[Tuple[ParserState, NodeType], ParserState] = {
+    _goto: Final[Dict[Tuple[ParserState, NodeType], ParserState]] = {
         # A string trailer may start with '.' OR '%'.
         (START, token.DOT): DOT,
         (START, token.PERCENT): PERCENT,
         # A string trailer may start with '.' OR '%'.
         (START, token.DOT): DOT,
         (START, token.PERCENT): PERCENT,