X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/a4e35b314977baae2e930abd24fa1013c7235e39..37542e64855ce21bd580f973ae5ce1ed86812a7a:/src/black/trans.py diff --git a/src/black/trans.py b/src/black/trans.py index bc6e93b..ec07f5e 100644 --- a/src/black/trans.py +++ b/src/black/trans.py @@ -1,13 +1,15 @@ """ String transformers that can split and merge strings. """ +import re +import sys from abc import ABC, abstractmethod from collections import defaultdict from dataclasses import dataclass -import regex as re from typing import ( Any, Callable, + ClassVar, Collection, Dict, Iterable, @@ -15,25 +17,43 @@ from typing import ( List, Optional, Sequence, + Set, Tuple, TypeVar, Union, ) -from black.rusty import Result, Ok, Err +if sys.version_info < (3, 8): + from typing_extensions import Final, Literal +else: + from typing import Literal, Final -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 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.strings import has_triple_quotes, get_string_prefix, assert_is_leaf_string -from black.strings import normalize_string_quotes +from mypy_extensions import trait -from blib2to3.pytree import Leaf, Node +from black.comments import contains_pragma_comment +from black.lines import Line, append_leaves +from black.mode import Feature +from black.nodes import ( + CLOSING_BRACKETS, + OPENING_BRACKETS, + STANDALONE_COMMENT, + is_empty_lpar, + is_empty_par, + is_empty_rpar, + is_part_of_annotation, + parent_type, + replace_child, + syms, +) +from black.rusty import Err, Ok, Result +from black.strings import ( + assert_is_leaf_string, + get_string_prefix, + has_triple_quotes, + normalize_string_quotes, +) from blib2to3.pgen2 import token +from blib2to3.pytree import Leaf, Node class CannotTransform(Exception): @@ -49,7 +69,7 @@ NodeType = int ParserState = int StringID = int TResult = Result[T, CannotTransform] # (T)ransform Result -TMatchResult = TResult[Index] +TMatchResult = TResult[List[Index]] def TErr(err_msg: str) -> Err[CannotTransform]: @@ -61,7 +81,84 @@ def TErr(err_msg: str) -> Err[CannotTransform]: return Err(cant_transform) -@dataclass # type: ignore +def hug_power_op(line: Line, features: Collection[Feature]) -> Iterator[Line]: + """A transformer which normalizes spacing around power operators.""" + + # Performance optimization to avoid unnecessary Leaf clones and other ops. + for leaf in line.leaves: + if leaf.type == token.DOUBLESTAR: + break + else: + raise CannotTransform("No doublestar token was found in the line.") + + def is_simple_lookup(index: int, step: Literal[1, -1]) -> bool: + # Brackets and parentheses indicate calls, subscripts, etc. ... + # basically stuff that doesn't count as "simple". Only a NAME lookup + # or dotted lookup (eg. NAME.NAME) is OK. + if step == -1: + disallowed = {token.RPAR, token.RSQB} + else: + disallowed = {token.LPAR, token.LSQB} + + while 0 <= index < len(line.leaves): + current = line.leaves[index] + if current.type in disallowed: + return False + if current.type not in {token.NAME, token.DOT} or current.value == "for": + # If the current token isn't disallowed, we'll assume this is simple as + # only the disallowed tokens are semantically attached to this lookup + # expression we're checking. Also, stop early if we hit the 'for' bit + # of a comprehension. + return True + + index += step + + return True + + def is_simple_operand(index: int, kind: Literal["base", "exponent"]) -> bool: + # An operand is considered "simple" if's a NAME, a numeric CONSTANT, a simple + # lookup (see above), with or without a preceding unary operator. + start = line.leaves[index] + if start.type in {token.NAME, token.NUMBER}: + return is_simple_lookup(index, step=(1 if kind == "exponent" else -1)) + + if start.type in {token.PLUS, token.MINUS, token.TILDE}: + if line.leaves[index + 1].type in {token.NAME, token.NUMBER}: + # step is always one as bases with a preceding unary op will be checked + # for simplicity starting from the next token (so it'll hit the check + # above). + return is_simple_lookup(index + 1, step=1) + + return False + + new_line = line.clone() + should_hug = False + for idx, leaf in enumerate(line.leaves): + new_leaf = leaf.clone() + if should_hug: + new_leaf.prefix = "" + should_hug = False + + should_hug = ( + (0 < idx < len(line.leaves) - 1) + and leaf.type == token.DOUBLESTAR + and is_simple_operand(idx - 1, kind="base") + and line.leaves[idx - 1].value != "lambda" + and is_simple_operand(idx + 1, kind="exponent") + ) + if should_hug: + new_leaf.prefix = "" + + # We have to be careful to make a new line properly: + # - bracket related metadata must be maintained (handled by Line.append) + # - comments need to copied over, updating the leaf IDs they're attached to + new_line.append(new_leaf, preformatted=True) + for comment_leaf in line.comments_after(leaf): + new_line.append(comment_leaf, preformatted=True) + + yield new_line + + class StringTransformer(ABC): """ An implementation of the Transformer protocol that relies on its @@ -89,22 +186,31 @@ class StringTransformer(ABC): 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: """ Returns: - * Ok(string_idx) such that `line.leaves[string_idx]` is our target - string, if a match was able to be made. + * Ok(string_indices) such that for each index, `line.leaves[index]` + is our target string if a match was able to be made. For + transformers that don't result in more lines (e.g. StringMerger, + StringParenStripper), multiple matches and transforms are done at + once to reduce the complexity. OR - * Err(CannotTransform), if a match was not able to be made. + * Err(CannotTransform), if no match could be made. """ @abstractmethod - def do_transform(self, line: Line, string_idx: int) -> Iterator[TResult[Line]]: + def do_transform( + self, line: Line, string_indices: List[int] + ) -> Iterator[TResult[Line]]: """ Yields: * Ok(new_line) where new_line is the new transformed line. @@ -145,9 +251,9 @@ class StringTransformer(ABC): " this line as one that it can transform." ) from cant_transform - string_idx = match_result.ok() + string_indices = match_result.ok() - for line_result in self.do_transform(line, string_idx): + for line_result in self.do_transform(line, string_indices): if isinstance(line_result, Err): cant_transform = line_result.err() raise CannotTransform( @@ -183,6 +289,7 @@ class CustomSplit: break_idx: int +@trait class CustomSplitMapMixin: """ This mixin class is used to map merged strings to a sequence of @@ -190,8 +297,10 @@ class CustomSplitMapMixin: 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": @@ -242,12 +351,12 @@ class CustomSplitMapMixin: return key in self._CUSTOM_SPLIT_MAP -class StringMerger(CustomSplitMapMixin, StringTransformer): +class StringMerger(StringTransformer, CustomSplitMapMixin): """StringTransformer that merges strings together. Requirements: (A) The line contains adjacent strings such that ALL of the validation checks - listed in StringMerger.__validate_msg(...)'s docstring pass. + listed in StringMerger._validate_msg(...)'s docstring pass. OR (B) The line contains a string which uses line continuation backslashes. @@ -267,28 +376,50 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): is_valid_index = is_valid_index_factory(LL) - for (i, leaf) in enumerate(LL): + string_indices = [] + idx = 0 + while is_valid_index(idx): + leaf = LL[idx] if ( leaf.type == token.STRING - and is_valid_index(i + 1) - and LL[i + 1].type == token.STRING + and is_valid_index(idx + 1) + and LL[idx + 1].type == token.STRING ): - return Ok(i) + if not is_part_of_annotation(leaf): + string_indices.append(idx) + + # Advance to the next non-STRING leaf. + idx += 2 + while is_valid_index(idx) and LL[idx].type == token.STRING: + idx += 1 - if leaf.type == token.STRING and "\\\n" in leaf.value: - return Ok(i) + elif leaf.type == token.STRING and "\\\n" in leaf.value: + string_indices.append(idx) + # Advance to the next non-STRING leaf. + idx += 1 + while is_valid_index(idx) and LL[idx].type == token.STRING: + idx += 1 - return TErr("This line has no strings that need merging.") + else: + idx += 1 + + if string_indices: + return Ok(string_indices) + else: + return TErr("This line has no strings that need merging.") - def do_transform(self, line: Line, string_idx: int) -> Iterator[TResult[Line]]: + def do_transform( + self, line: Line, string_indices: List[int] + ) -> Iterator[TResult[Line]]: new_line = line + rblc_result = self._remove_backslash_line_continuation_chars( - new_line, string_idx + new_line, string_indices ) if isinstance(rblc_result, Ok): new_line = rblc_result.ok() - msg_result = self._merge_string_group(new_line, string_idx) + msg_result = self._merge_string_group(new_line, string_indices) if isinstance(msg_result, Ok): new_line = msg_result.ok() @@ -309,7 +440,7 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): @staticmethod def _remove_backslash_line_continuation_chars( - line: Line, string_idx: int + line: Line, string_indices: List[int] ) -> TResult[Line]: """ Merge strings that were split across multiple lines using @@ -323,34 +454,44 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): """ LL = line.leaves - string_leaf = LL[string_idx] - if not ( - string_leaf.type == token.STRING - and "\\\n" in string_leaf.value - and not has_triple_quotes(string_leaf.value) - ): + indices_to_transform = [] + for string_idx in string_indices: + string_leaf = LL[string_idx] + if ( + string_leaf.type == token.STRING + and "\\\n" in string_leaf.value + and not has_triple_quotes(string_leaf.value) + ): + indices_to_transform.append(string_idx) + + if not indices_to_transform: return TErr( - f"String leaf {string_leaf} does not contain any backslash line" - " continuation characters." + "Found no string leaves that contain backslash line continuation" + " characters." ) new_line = line.clone() new_line.comments = line.comments.copy() append_leaves(new_line, line, LL) - new_string_leaf = new_line.leaves[string_idx] - new_string_leaf.value = new_string_leaf.value.replace("\\\n", "") + for string_idx in indices_to_transform: + new_string_leaf = new_line.leaves[string_idx] + new_string_leaf.value = new_string_leaf.value.replace("\\\n", "") return Ok(new_line) - def _merge_string_group(self, line: Line, string_idx: int) -> TResult[Line]: + def _merge_string_group( + self, line: Line, string_indices: List[int] + ) -> TResult[Line]: """ - Merges string group (i.e. set of adjacent strings) where the first - string in the group is `line.leaves[string_idx]`. + Merges string groups (i.e. set of adjacent strings). + + Each index from `string_indices` designates one string group's first + leaf in `line.leaves`. Returns: Ok(new_line), if ALL of the validation checks found in - __validate_msg(...) pass. + _validate_msg(...) pass. OR Err(CannotTransform), otherwise. """ @@ -358,10 +499,54 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): is_valid_index = is_valid_index_factory(LL) - vresult = self._validate_msg(line, string_idx) - if isinstance(vresult, Err): - return vresult + # A dict of {string_idx: tuple[num_of_strings, string_leaf]}. + merged_string_idx_dict: Dict[int, Tuple[int, Leaf]] = {} + for string_idx in string_indices: + vresult = self._validate_msg(line, string_idx) + if isinstance(vresult, Err): + continue + merged_string_idx_dict[string_idx] = self._merge_one_string_group( + LL, string_idx, is_valid_index + ) + + if not merged_string_idx_dict: + return TErr("No string group is merged") + + # Build the final line ('new_line') that this method will later return. + new_line = line.clone() + previous_merged_string_idx = -1 + previous_merged_num_of_strings = -1 + for i, leaf in enumerate(LL): + if i in merged_string_idx_dict: + previous_merged_string_idx = i + previous_merged_num_of_strings, string_leaf = merged_string_idx_dict[i] + new_line.append(string_leaf) + + if ( + previous_merged_string_idx + <= i + < previous_merged_string_idx + previous_merged_num_of_strings + ): + for comment_leaf in line.comments_after(LL[i]): + new_line.append(comment_leaf, preformatted=True) + continue + + append_leaves(new_line, line, [leaf]) + + return Ok(new_line) + + def _merge_one_string_group( + self, LL: List[Leaf], string_idx: int, is_valid_index: Callable[[int], bool] + ) -> Tuple[int, Leaf]: + """ + Merges one string group where the first string in the group is + `LL[string_idx]`. + Returns: + A tuple of `(num_of_strings, leaf)` where `num_of_strings` is the + number of strings merged and `leaf` is the newly merged string + to be replaced in the new line. + """ # If the string group is wrapped inside an Atom node, we must make sure # to later replace that Atom with our new (merged) string leaf. atom_node = LL[string_idx].parent @@ -411,7 +596,7 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): 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 @@ -431,13 +616,13 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): 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 "f" in prefix and "f" not in next_prefix: # Then we must escape any braces contained in this substring. - SS = re.subf(r"(\{|\})", "{1}{1}", SS) + SS = re.sub(r"(\{|\})", r"\1\1", SS) NSS = make_naked(SS, next_prefix) @@ -449,6 +634,9 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): next_str_idx += 1 + # Take a note on the index of the non-STRING leaf. + non_string_idx = next_str_idx + S_leaf = Leaf(token.STRING, S) if self.normalize_strings: S_leaf.value = normalize_string_quotes(S_leaf.value) @@ -468,29 +656,27 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): string_leaf = Leaf(token.STRING, S_leaf.value.replace(BREAK_MARK, "")) if atom_node is not None: - replace_child(atom_node, string_leaf) - - # Build the final line ('new_line') that this method will later return. - new_line = line.clone() - for (i, leaf) in enumerate(LL): - if i == string_idx: - new_line.append(string_leaf) - - if string_idx <= i < string_idx + num_of_strings: - for comment_leaf in line.comments_after(LL[i]): - new_line.append(comment_leaf, preformatted=True) - continue - - append_leaves(new_line, line, [leaf]) + # If not all children of the atom node are merged (this can happen + # when there is a standalone comment in the middle) ... + if non_string_idx - string_idx < len(atom_node.children): + # We need to replace the old STRING leaves with the new string leaf. + first_child_idx = LL[string_idx].remove() + for idx in range(string_idx + 1, non_string_idx): + LL[idx].remove() + if first_child_idx is not None: + atom_node.insert_child(first_child_idx, string_leaf) + else: + # Else replace the atom node with the new string leaf. + replace_child(atom_node, string_leaf) self.add_custom_splits(string_leaf.value, custom_splits) - return Ok(new_line) + return num_of_strings, string_leaf @staticmethod def _validate_msg(line: Line, string_idx: int) -> TResult[None]: """Validate (M)erge (S)tring (G)roup - Transform-time string validation logic for __merge_string_group(...). + Transform-time string validation logic for _merge_string_group(...). Returns: * Ok(None), if ALL validation checks (listed below) pass. @@ -504,6 +690,11 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): - The set of all string prefixes in the string group is of length greater than one and is not equal to {"", "f"}. - The string group consists of raw strings. + - The string group is stringified type annotations. We don't want to + process stringified type annotations since pyright doesn't support + them spanning multiple string values. (NOTE: mypy, pytype, pyre do + support them, so we can change if pyright also gains support in the + future. See https://github.com/microsoft/pyright/issues/4359.) """ # We first check for "inner" stand-alone comments (i.e. stand-alone # comments that have a string leaf before them AND after them). @@ -541,7 +732,7 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): 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.") @@ -593,7 +784,15 @@ class StringParenStripper(StringTransformer): is_valid_index = is_valid_index_factory(LL) - for (idx, leaf) in enumerate(LL): + string_indices = [] + + idx = -1 + while True: + idx += 1 + if idx >= len(LL): + break + leaf = LL[idx] + # Should be a string... if leaf.type != token.STRING: continue @@ -675,45 +874,73 @@ class StringParenStripper(StringTransformer): }: continue - return Ok(string_idx) + string_indices.append(string_idx) + idx = string_idx + while idx < len(LL) - 1 and LL[idx + 1].type == token.STRING: + idx += 1 + if string_indices: + return Ok(string_indices) return TErr("This line has no strings wrapped in parens.") - def do_transform(self, line: Line, string_idx: int) -> Iterator[TResult[Line]]: + def do_transform( + self, line: Line, string_indices: List[int] + ) -> Iterator[TResult[Line]]: LL = line.leaves - string_parser = StringParser() - rpar_idx = string_parser.parse(LL, string_idx) + string_and_rpar_indices: List[int] = [] + for string_idx in string_indices: + string_parser = StringParser() + rpar_idx = string_parser.parse(LL, string_idx) + + should_transform = True + for leaf in (LL[string_idx - 1], LL[rpar_idx]): + if line.comments_after(leaf): + # Should not strip parentheses which have comments attached + # to them. + should_transform = False + break + if should_transform: + string_and_rpar_indices.extend((string_idx, rpar_idx)) - for leaf in (LL[string_idx - 1], LL[rpar_idx]): - if line.comments_after(leaf): - yield TErr( - "Will not strip parentheses which have comments attached to them." - ) - return + if string_and_rpar_indices: + yield Ok(self._transform_to_new_line(line, string_and_rpar_indices)) + else: + yield Err( + CannotTransform("All string groups have comments attached to them.") + ) + + def _transform_to_new_line( + self, line: Line, string_and_rpar_indices: List[int] + ) -> Line: + LL = line.leaves new_line = line.clone() new_line.comments = line.comments.copy() - try: - append_leaves(new_line, line, LL[: string_idx - 1]) - except BracketMatchError: - # HACK: I believe there is currently a bug somewhere in - # right_hand_split() that is causing brackets to not be tracked - # properly by a shared BracketTracker. - append_leaves(new_line, line, LL[: string_idx - 1], preformatted=True) - - string_leaf = Leaf(token.STRING, LL[string_idx].value) - LL[string_idx - 1].remove() - replace_child(LL[string_idx], string_leaf) - new_line.append(string_leaf) - - append_leaves( - new_line, line, LL[string_idx + 1 : rpar_idx] + LL[rpar_idx + 1 :] - ) - LL[rpar_idx].remove() + previous_idx = -1 + # We need to sort the indices, since string_idx and its matching + # rpar_idx may not come in order, e.g. in + # `("outer" % ("inner".join(items)))`, the "inner" string's + # string_idx is smaller than "outer" string's rpar_idx. + for idx in sorted(string_and_rpar_indices): + leaf = LL[idx] + lpar_or_rpar_idx = idx - 1 if leaf.type == token.STRING else idx + append_leaves(new_line, line, LL[previous_idx + 1 : lpar_or_rpar_idx]) + if leaf.type == token.STRING: + string_leaf = Leaf(token.STRING, LL[idx].value) + LL[lpar_or_rpar_idx].remove() # Remove lpar. + replace_child(LL[idx], string_leaf) + new_line.append(string_leaf) + else: + LL[lpar_or_rpar_idx].remove() # This is a rpar. + + previous_idx = idx - yield Ok(new_line) + # Append the leaves after the last idx: + append_leaves(new_line, line, LL[idx + 1 :]) + + return new_line class BaseStringSplitter(StringTransformer): @@ -738,7 +965,7 @@ class BaseStringSplitter(StringTransformer): * The target string is not a multiline (i.e. triple-quote) string. """ - STRING_OPERATORS = [ + STRING_OPERATORS: Final = [ token.EQEQUAL, token.GREATER, token.GREATEREQUAL, @@ -766,7 +993,12 @@ class BaseStringSplitter(StringTransformer): if isinstance(match_result, Err): return match_result - string_idx = match_result.ok() + string_indices = match_result.ok() + assert len(string_indices) == 1, ( + f"{self.__class__.__name__} should only find one match at a time, found" + f" {len(string_indices)}" + ) + string_idx = string_indices[0] vresult = self._validate(line, string_idx) if isinstance(vresult, Err): return vresult @@ -867,7 +1099,7 @@ class BaseStringSplitter(StringTransformer): # 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. @@ -925,8 +1157,90 @@ class BaseStringSplitter(StringTransformer): max_string_length = self.line_length - offset return max_string_length + @staticmethod + def _prefer_paren_wrap_match(LL: List[Leaf]) -> Optional[int]: + """ + Returns: + string_idx such that @LL[string_idx] is equal to our target (i.e. + matched) string, if this line matches the "prefer paren wrap" statement + requirements listed in the 'Requirements' section of the StringParenWrapper + class's docstring. + OR + None, otherwise. + """ + # The line must start with a string. + if LL[0].type != token.STRING: + return None + + # If the string is surrounded by commas (or is the first/last child)... + prev_sibling = LL[0].prev_sibling + next_sibling = LL[0].next_sibling + if not prev_sibling and not next_sibling and parent_type(LL[0]) == syms.atom: + # If it's an atom string, we need to check the parent atom's siblings. + parent = LL[0].parent + assert parent is not None # For type checkers. + prev_sibling = parent.prev_sibling + next_sibling = parent.next_sibling + if (not prev_sibling or prev_sibling.type == token.COMMA) and ( + not next_sibling or next_sibling.type == token.COMMA + ): + return 0 + + return None -class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): + +def iter_fexpr_spans(s: str) -> Iterator[Tuple[int, int]]: + """ + Yields spans corresponding to expressions in a given f-string. + Spans are half-open ranges (left inclusive, right exclusive). + Assumes the input string is a valid f-string, but will not crash if the input + string is invalid. + """ + stack: List[int] = [] # our curly paren stack + i = 0 + while i < len(s): + if s[i] == "{": + # if we're in a string part of the f-string, ignore escaped curly braces + if not stack and i + 1 < len(s) and s[i + 1] == "{": + i += 2 + continue + stack.append(i) + i += 1 + continue + + if s[i] == "}": + if not stack: + i += 1 + continue + j = stack.pop() + # we've made it back out of the expression! yield the span + if not stack: + yield (j, i + 1) + i += 1 + continue + + # if we're in an expression part of the f-string, fast forward through strings + # note that backslashes are not legal in the expression portion of f-strings + if stack: + delim = None + if s[i : i + 3] in ("'''", '"""'): + delim = s[i : i + 3] + elif s[i] in ("'", '"'): + delim = s[i] + if delim: + i += len(delim) + while i < len(s) and s[i : i + len(delim)] != delim: + i += 1 + i += len(delim) + continue + i += 1 + + +def fstring_contains_expr(s: str) -> bool: + return any(iter_fexpr_spans(s)) + + +class StringSplitter(BaseStringSplitter, CustomSplitMapMixin): """ StringTransformer that splits "atom" strings (i.e. strings which exist on lines by themselves). @@ -964,22 +1278,14 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): CustomSplit objects and add them to the custom split map. """ - MIN_SUBSTR_SIZE = 6 - # Matches an "f-expression" (e.g. {var}) that might be found in an f-string. - RE_FEXPR = r""" - (? TMatchResult: LL = line.leaves + if self._prefer_paren_wrap_match(LL) is not None: + return TErr("Line needs to be wrapped in parens first.") + is_valid_index = is_valid_index_factory(LL) idx = 0 @@ -1026,31 +1332,38 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): if is_valid_index(idx): return TErr("This line does not end with a string.") - return Ok(string_idx) + return Ok([string_idx]) - def do_transform(self, line: Line, string_idx: int) -> Iterator[TResult[Line]]: + def do_transform( + self, line: Line, string_indices: List[int] + ) -> Iterator[TResult[Line]]: LL = line.leaves + assert len(string_indices) == 1, ( + f"{self.__class__.__name__} should only find one match at a time, found" + f" {len(string_indices)}" + ) + string_idx = string_indices[0] QUOTE = LL[string_idx].value[-1] 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 # contains at least one f-expression. Otherwise, we will alter the AST # of the program. - drop_pointless_f_prefix = ("f" in prefix) and re.search( - self.RE_FEXPR, LL[string_idx].value, re.VERBOSE + drop_pointless_f_prefix = ("f" in prefix) and fstring_contains_expr( + LL[string_idx].value ) first_string_line = True 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 + sum(len(str(prefix_leaf)) for prefix_leaf in string_op_leaves) + 1 if string_op_leaves else 0 ) @@ -1151,21 +1464,37 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): # --- 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 ( - # 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) + use_custom_breakpoints + and not csplit.has_prefix + and ( + # `next_value == prefix + QUOTE` happens when the custom + # split is an empty string. + next_value == prefix + QUOTE + or next_value != self._normalize_f_string(next_value, 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. - 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 + + if drop_pointless_f_prefix: next_value = self._normalize_f_string(next_value, prefix) # --- Construct `next_leaf` @@ -1232,6 +1561,59 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): 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 + yield from iter_fexpr_spans(string) + + 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 @@ -1261,40 +1643,15 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): assert is_valid_index(max_break_idx) assert_is_leaf_string(string) - _fexpr_slices: Optional[List[Tuple[Index, Index]]] = None - - 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) + _illegal_split_indices = self._get_illegal_split_indices(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 - 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: """ @@ -1318,7 +1675,7 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): 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. @@ -1360,7 +1717,7 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): """ assert_is_leaf_string(string) - if "f" in prefix and not re.search(self.RE_FEXPR, string, re.VERBOSE): + if "f" in prefix and not fstring_contains_expr(string): new_prefix = prefix.replace("f", "") temp = string[len(prefix) :] @@ -1384,10 +1741,9 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): 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 wraps strings in parens and then splits at the LPAR. Requirements: All of the requirements listed in BaseStringSplitter's docstring in @@ -1407,6 +1763,12 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): OR * The line is a dictionary key assignment where some valid key is being assigned the value of some string. + OR + * The line is an lambda expression and the value is a string. + OR + * The line starts with an "atom" string that prefers to be wrapped in + parens. It's preferred to be wrapped when the string is surrounded by + commas (or is the first/last child). Transformations: The chosen string is wrapped in parentheses and then split at the LPAR. @@ -1431,6 +1793,9 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): changed such that it no longer needs to be given its own line, StringParenWrapper relies on StringParenStripper to clean up the parentheses it created. + + For "atom" strings that prefers to be wrapped in parens, it requires + StringSplitter to hold the split until the string is wrapped in parens. """ def do_splitter_match(self, line: Line) -> TMatchResult: @@ -1446,7 +1811,8 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): or self._else_match(LL) or self._assert_match(LL) or self._assign_match(LL) - or self._dict_match(LL) + or self._dict_or_lambda_match(LL) + or self._prefer_paren_wrap_match(LL) ) if string_idx is not None: @@ -1464,7 +1830,7 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): " resultant line would still be over the specified line" " length and can't be split further by StringSplitter." ) - return Ok(string_idx) + return Ok([string_idx]) return TErr("This line does not contain any non-atomic strings.") @@ -1536,7 +1902,7 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): if parent_type(LL[0]) == syms.assert_stmt and LL[0].value == "assert": is_valid_index = is_valid_index_factory(LL) - for (i, leaf) in enumerate(LL): + for i, leaf in enumerate(LL): # We MUST find a comma... if leaf.type == token.COMMA: idx = i + 2 if is_empty_par(LL[i + 1]) else i + 1 @@ -1574,7 +1940,7 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): ): is_valid_index = is_valid_index_factory(LL) - for (i, leaf) in enumerate(LL): + for i, leaf in enumerate(LL): # We MUST find either an '=' or '+=' symbol... if leaf.type in [token.EQUAL, token.PLUSEQUAL]: idx = i + 2 if is_empty_par(LL[i + 1]) else i + 1 @@ -1603,23 +1969,24 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): return None @staticmethod - def _dict_match(LL: List[Leaf]) -> Optional[int]: + def _dict_or_lambda_match(LL: List[Leaf]) -> Optional[int]: """ Returns: string_idx such that @LL[string_idx] is equal to our target (i.e. matched) string, if this line matches the dictionary key assignment - statement requirements listed in the 'Requirements' section of this - classes' docstring. + statement or lambda expression requirements listed in the + 'Requirements' section of this classes' docstring. OR None, otherwise. """ - # If this line is apart of a dictionary key assignment... - if syms.dictsetmaker in [parent_type(LL[0]), parent_type(LL[0].parent)]: + # If this line is a part of a dictionary key assignment or lambda expression... + parent_types = [parent_type(LL[0]), parent_type(LL[0].parent)] + if syms.dictsetmaker in parent_types or syms.lambdef in parent_types: is_valid_index = is_valid_index_factory(LL) - for (i, leaf) in enumerate(LL): - # We MUST find a colon... - if leaf.type == token.COLON: + for i, leaf in enumerate(LL): + # We MUST find a colon, it can either be dict's or lambda's colon... + if leaf.type == token.COLON and i < len(LL) - 1: idx = i + 2 if is_empty_par(LL[i + 1]) else i + 1 # That colon MUST be followed by a string... @@ -1640,8 +2007,15 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): return None - def do_transform(self, line: Line, string_idx: int) -> Iterator[TResult[Line]]: + def do_transform( + self, line: Line, string_indices: List[int] + ) -> Iterator[TResult[Line]]: LL = line.leaves + assert len(string_indices) == 1, ( + f"{self.__class__.__name__} should only find one match at a time, found" + f" {len(string_indices)}" + ) + string_idx = string_indices[0] is_valid_index = is_valid_index_factory(LL) insert_str_child = insert_str_child_factory(LL[string_idx]) @@ -1713,6 +2087,25 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): f" (left_leaves={left_leaves}, right_leaves={right_leaves})" ) old_rpar_leaf = right_leaves.pop() + elif right_leaves and right_leaves[-1].type == token.RPAR: + # Special case for lambda expressions as dict's value, e.g.: + # my_dict = { + # "key": lambda x: f"formatted: {x}, + # } + # After wrapping the dict's value with parentheses, the string is + # followed by a RPAR but its opening bracket is lambda's, not + # the string's: + # "key": (lambda x: f"formatted: {x}), + opening_bracket = right_leaves[-1].opening_bracket + if opening_bracket is not None and opening_bracket in left_leaves: + index = left_leaves.index(opening_bracket) + if ( + index > 0 + and index < len(left_leaves) - 1 + and left_leaves[index - 1].type == token.COLON + and left_leaves[index + 1].value == "lambda" + ): + right_leaves.pop() append_leaves(string_line, line, right_leaves) @@ -1769,20 +2162,20 @@ class StringParser: ``` """ - DEFAULT_TOKEN = -1 + DEFAULT_TOKEN: Final = 20210605 # 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 - _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,