diff options
Diffstat (limited to 'src/prompt_toolkit/contrib/regular_languages')
6 files changed, 1177 insertions, 0 deletions
diff --git a/src/prompt_toolkit/contrib/regular_languages/__init__.py b/src/prompt_toolkit/contrib/regular_languages/__init__.py new file mode 100644 index 0000000..3f30614 --- /dev/null +++ b/src/prompt_toolkit/contrib/regular_languages/__init__.py @@ -0,0 +1,77 @@ +r""" +Tool for expressing the grammar of an input as a regular language. +================================================================== + +The grammar for the input of many simple command line interfaces can be +expressed by a regular language. Examples are PDB (the Python debugger); a +simple (bash-like) shell with "pwd", "cd", "cat" and "ls" commands; arguments +that you can pass to an executable; etc. It is possible to use regular +expressions for validation and parsing of such a grammar. (More about regular +languages: http://en.wikipedia.org/wiki/Regular_language) + +Example +------- + +Let's take the pwd/cd/cat/ls example. We want to have a shell that accepts +these three commands. "cd" is followed by a quoted directory name and "cat" is +followed by a quoted file name. (We allow quotes inside the filename when +they're escaped with a backslash.) We could define the grammar using the +following regular expression:: + + grammar = \s* ( + pwd | + ls | + (cd \s+ " ([^"]|\.)+ ") | + (cat \s+ " ([^"]|\.)+ ") + ) \s* + + +What can we do with this grammar? +--------------------------------- + +- Syntax highlighting: We could use this for instance to give file names + different colour. +- Parse the result: .. We can extract the file names and commands by using a + regular expression with named groups. +- Input validation: .. Don't accept anything that does not match this grammar. + When combined with a parser, we can also recursively do + filename validation (and accept only existing files.) +- Autocompletion: .... Each part of the grammar can have its own autocompleter. + "cat" has to be completed using file names, while "cd" + has to be completed using directory names. + +How does it work? +----------------- + +As a user of this library, you have to define the grammar of the input as a +regular expression. The parts of this grammar where autocompletion, validation +or any other processing is required need to be marked using a regex named +group. Like ``(?P<varname>...)`` for instance. + +When the input is processed for validation (for instance), the regex will +execute, the named group is captured, and the validator associated with this +named group will test the captured string. + +There is one tricky bit: + + Often we operate on incomplete input (this is by definition the case for + autocompletion) and we have to decide for the cursor position in which + possible state the grammar it could be and in which way variables could be + matched up to that point. + +To solve this problem, the compiler takes the original regular expression and +translates it into a set of other regular expressions which each match certain +prefixes of the original regular expression. We generate one prefix regular +expression for every named variable (with this variable being the end of that +expression). + + +TODO: some examples of: + - How to create a highlighter from this grammar. + - How to create a validator from this grammar. + - How to create an autocompleter from this grammar. + - How to create a parser from this grammar. +""" +from .compiler import compile + +__all__ = ["compile"] diff --git a/src/prompt_toolkit/contrib/regular_languages/compiler.py b/src/prompt_toolkit/contrib/regular_languages/compiler.py new file mode 100644 index 0000000..fed7849 --- /dev/null +++ b/src/prompt_toolkit/contrib/regular_languages/compiler.py @@ -0,0 +1,574 @@ +r""" +Compiler for a regular grammar. + +Example usage:: + + # Create and compile grammar. + p = compile('add \s+ (?P<var1>[^\s]+) \s+ (?P<var2>[^\s]+)') + + # Match input string. + m = p.match('add 23 432') + + # Get variables. + m.variables().get('var1') # Returns "23" + m.variables().get('var2') # Returns "432" + + +Partial matches are possible:: + + # Create and compile grammar. + p = compile(''' + # Operators with two arguments. + ((?P<operator1>[^\s]+) \s+ (?P<var1>[^\s]+) \s+ (?P<var2>[^\s]+)) | + + # Operators with only one arguments. + ((?P<operator2>[^\s]+) \s+ (?P<var1>[^\s]+)) + ''') + + # Match partial input string. + m = p.match_prefix('add 23') + + # Get variables. (Notice that both operator1 and operator2 contain the + # value "add".) This is because our input is incomplete, and we don't know + # yet in which rule of the regex we we'll end up. It could also be that + # `operator1` and `operator2` have a different autocompleter and we want to + # call all possible autocompleters that would result in valid input.) + m.variables().get('var1') # Returns "23" + m.variables().get('operator1') # Returns "add" + m.variables().get('operator2') # Returns "add" + +""" +import re +from typing import Callable, Dict, Iterable, Iterator, List +from typing import Match as RegexMatch +from typing import Optional, Pattern, Tuple + +from .regex_parser import ( + AnyNode, + Lookahead, + Node, + NodeSequence, + Regex, + Repeat, + Variable, + parse_regex, + tokenize_regex, +) + +__all__ = [ + "compile", +] + + +# Name of the named group in the regex, matching trailing input. +# (Trailing input is when the input contains characters after the end of the +# expression has been matched.) +_INVALID_TRAILING_INPUT = "invalid_trailing" + +EscapeFuncDict = Dict[str, Callable[[str], str]] + + +class _CompiledGrammar: + """ + Compiles a grammar. This will take the parse tree of a regular expression + and compile the grammar. + + :param root_node: :class~`.regex_parser.Node` instance. + :param escape_funcs: `dict` mapping variable names to escape callables. + :param unescape_funcs: `dict` mapping variable names to unescape callables. + """ + + def __init__( + self, + root_node: Node, + escape_funcs: Optional[EscapeFuncDict] = None, + unescape_funcs: Optional[EscapeFuncDict] = None, + ) -> None: + + self.root_node = root_node + self.escape_funcs = escape_funcs or {} + self.unescape_funcs = unescape_funcs or {} + + #: Dictionary that will map the regex names to Node instances. + self._group_names_to_nodes: Dict[ + str, str + ] = {} # Maps regex group names to varnames. + counter = [0] + + def create_group_func(node: Variable) -> str: + name = "n%s" % counter[0] + self._group_names_to_nodes[name] = node.varname + counter[0] += 1 + return name + + # Compile regex strings. + self._re_pattern = "^%s$" % self._transform(root_node, create_group_func) + self._re_prefix_patterns = list( + self._transform_prefix(root_node, create_group_func) + ) + + # Compile the regex itself. + flags = re.DOTALL # Note that we don't need re.MULTILINE! (^ and $ + # still represent the start and end of input text.) + self._re = re.compile(self._re_pattern, flags) + self._re_prefix = [re.compile(t, flags) for t in self._re_prefix_patterns] + + # We compile one more set of regexes, similar to `_re_prefix`, but accept any trailing + # input. This will ensure that we can still highlight the input correctly, even when the + # input contains some additional characters at the end that don't match the grammar.) + self._re_prefix_with_trailing_input = [ + re.compile( + r"(?:{})(?P<{}>.*?)$".format(t.rstrip("$"), _INVALID_TRAILING_INPUT), + flags, + ) + for t in self._re_prefix_patterns + ] + + def escape(self, varname: str, value: str) -> str: + """ + Escape `value` to fit in the place of this variable into the grammar. + """ + f = self.escape_funcs.get(varname) + return f(value) if f else value + + def unescape(self, varname: str, value: str) -> str: + """ + Unescape `value`. + """ + f = self.unescape_funcs.get(varname) + return f(value) if f else value + + @classmethod + def _transform( + cls, root_node: Node, create_group_func: Callable[[Variable], str] + ) -> str: + """ + Turn a :class:`Node` object into a regular expression. + + :param root_node: The :class:`Node` instance for which we generate the grammar. + :param create_group_func: A callable which takes a `Node` and returns the next + free name for this node. + """ + + def transform(node: Node) -> str: + # Turn `AnyNode` into an OR. + if isinstance(node, AnyNode): + return "(?:%s)" % "|".join(transform(c) for c in node.children) + + # Concatenate a `NodeSequence` + elif isinstance(node, NodeSequence): + return "".join(transform(c) for c in node.children) + + # For Regex and Lookahead nodes, just insert them literally. + elif isinstance(node, Regex): + return node.regex + + elif isinstance(node, Lookahead): + before = "(?!" if node.negative else "(=" + return before + transform(node.childnode) + ")" + + # A `Variable` wraps the children into a named group. + elif isinstance(node, Variable): + return "(?P<{}>{})".format( + create_group_func(node), + transform(node.childnode), + ) + + # `Repeat`. + elif isinstance(node, Repeat): + if node.max_repeat is None: + if node.min_repeat == 0: + repeat_sign = "*" + elif node.min_repeat == 1: + repeat_sign = "+" + else: + repeat_sign = "{%i,%s}" % ( + node.min_repeat, + ("" if node.max_repeat is None else str(node.max_repeat)), + ) + + return "(?:{}){}{}".format( + transform(node.childnode), + repeat_sign, + ("" if node.greedy else "?"), + ) + else: + raise TypeError(f"Got {node!r}") + + return transform(root_node) + + @classmethod + def _transform_prefix( + cls, root_node: Node, create_group_func: Callable[[Variable], str] + ) -> Iterable[str]: + """ + Yield all the regular expressions matching a prefix of the grammar + defined by the `Node` instance. + + For each `Variable`, one regex pattern will be generated, with this + named group at the end. This is required because a regex engine will + terminate once a match is found. For autocompletion however, we need + the matches for all possible paths, so that we can provide completions + for each `Variable`. + + - So, in the case of an `Any` (`A|B|C)', we generate a pattern for each + clause. This is one for `A`, one for `B` and one for `C`. Unless some + groups don't contain a `Variable`, then these can be merged together. + - In the case of a `NodeSequence` (`ABC`), we generate a pattern for + each prefix that ends with a variable, and one pattern for the whole + sequence. So, that's one for `A`, one for `AB` and one for `ABC`. + + :param root_node: The :class:`Node` instance for which we generate the grammar. + :param create_group_func: A callable which takes a `Node` and returns the next + free name for this node. + """ + + def contains_variable(node: Node) -> bool: + if isinstance(node, Regex): + return False + elif isinstance(node, Variable): + return True + elif isinstance(node, (Lookahead, Repeat)): + return contains_variable(node.childnode) + elif isinstance(node, (NodeSequence, AnyNode)): + return any(contains_variable(child) for child in node.children) + + return False + + def transform(node: Node) -> Iterable[str]: + # Generate separate pattern for all terms that contain variables + # within this OR. Terms that don't contain a variable can be merged + # together in one pattern. + if isinstance(node, AnyNode): + # If we have a definition like: + # (?P<name> .*) | (?P<city> .*) + # Then we want to be able to generate completions for both the + # name as well as the city. We do this by yielding two + # different regular expressions, because the engine won't + # follow multiple paths, if multiple are possible. + children_with_variable = [] + children_without_variable = [] + for c in node.children: + if contains_variable(c): + children_with_variable.append(c) + else: + children_without_variable.append(c) + + for c in children_with_variable: + yield from transform(c) + + # Merge options without variable together. + if children_without_variable: + yield "|".join( + r for c in children_without_variable for r in transform(c) + ) + + # For a sequence, generate a pattern for each prefix that ends with + # a variable + one pattern of the complete sequence. + # (This is because, for autocompletion, we match the text before + # the cursor, and completions are given for the variable that we + # match right before the cursor.) + elif isinstance(node, NodeSequence): + # For all components in the sequence, compute prefix patterns, + # as well as full patterns. + complete = [cls._transform(c, create_group_func) for c in node.children] + prefixes = [list(transform(c)) for c in node.children] + variable_nodes = [contains_variable(c) for c in node.children] + + # If any child is contains a variable, we should yield a + # pattern up to that point, so that we are sure this will be + # matched. + for i in range(len(node.children)): + if variable_nodes[i]: + for c_str in prefixes[i]: + yield "".join(complete[:i]) + c_str + + # If there are non-variable nodes, merge all the prefixes into + # one pattern. If the input is: "[part1] [part2] [part3]", then + # this gets compiled into: + # (complete1 + (complete2 + (complete3 | partial3) | partial2) | partial1 ) + # For nodes that contain a variable, we skip the "|partial" + # part here, because thees are matched with the previous + # patterns. + if not all(variable_nodes): + result = [] + + # Start with complete patterns. + for i in range(len(node.children)): + result.append("(?:") + result.append(complete[i]) + + # Add prefix patterns. + for i in range(len(node.children) - 1, -1, -1): + if variable_nodes[i]: + # No need to yield a prefix for this one, we did + # the variable prefixes earlier. + result.append(")") + else: + result.append("|(?:") + # If this yields multiple, we should yield all combinations. + assert len(prefixes[i]) == 1 + result.append(prefixes[i][0]) + result.append("))") + + yield "".join(result) + + elif isinstance(node, Regex): + yield "(?:%s)?" % node.regex + + elif isinstance(node, Lookahead): + if node.negative: + yield "(?!%s)" % cls._transform(node.childnode, create_group_func) + else: + # Not sure what the correct semantics are in this case. + # (Probably it's not worth implementing this.) + raise Exception("Positive lookahead not yet supported.") + + elif isinstance(node, Variable): + # (Note that we should not append a '?' here. the 'transform' + # method will already recursively do that.) + for c_str in transform(node.childnode): + yield f"(?P<{create_group_func(node)}>{c_str})" + + elif isinstance(node, Repeat): + # If we have a repetition of 8 times. That would mean that the + # current input could have for instance 7 times a complete + # match, followed by a partial match. + prefix = cls._transform(node.childnode, create_group_func) + + if node.max_repeat == 1: + yield from transform(node.childnode) + else: + for c_str in transform(node.childnode): + if node.max_repeat: + repeat_sign = "{,%i}" % (node.max_repeat - 1) + else: + repeat_sign = "*" + yield "(?:{}){}{}{}".format( + prefix, + repeat_sign, + ("" if node.greedy else "?"), + c_str, + ) + + else: + raise TypeError("Got %r" % node) + + for r in transform(root_node): + yield "^(?:%s)$" % r + + def match(self, string: str) -> Optional["Match"]: + """ + Match the string with the grammar. + Returns a :class:`Match` instance or `None` when the input doesn't match the grammar. + + :param string: The input string. + """ + m = self._re.match(string) + + if m: + return Match( + string, [(self._re, m)], self._group_names_to_nodes, self.unescape_funcs + ) + return None + + def match_prefix(self, string: str) -> Optional["Match"]: + """ + Do a partial match of the string with the grammar. The returned + :class:`Match` instance can contain multiple representations of the + match. This will never return `None`. If it doesn't match at all, the "trailing input" + part will capture all of the input. + + :param string: The input string. + """ + # First try to match using `_re_prefix`. If nothing is found, use the patterns that + # also accept trailing characters. + for patterns in [self._re_prefix, self._re_prefix_with_trailing_input]: + matches = [(r, r.match(string)) for r in patterns] + matches2 = [(r, m) for r, m in matches if m] + + if matches2 != []: + return Match( + string, matches2, self._group_names_to_nodes, self.unescape_funcs + ) + + return None + + +class Match: + """ + :param string: The input string. + :param re_matches: List of (compiled_re_pattern, re_match) tuples. + :param group_names_to_nodes: Dictionary mapping all the re group names to the matching Node instances. + """ + + def __init__( + self, + string: str, + re_matches: List[Tuple[Pattern[str], RegexMatch[str]]], + group_names_to_nodes: Dict[str, str], + unescape_funcs: Dict[str, Callable[[str], str]], + ): + self.string = string + self._re_matches = re_matches + self._group_names_to_nodes = group_names_to_nodes + self._unescape_funcs = unescape_funcs + + def _nodes_to_regs(self) -> List[Tuple[str, Tuple[int, int]]]: + """ + Return a list of (varname, reg) tuples. + """ + + def get_tuples() -> Iterable[Tuple[str, Tuple[int, int]]]: + for r, re_match in self._re_matches: + for group_name, group_index in r.groupindex.items(): + if group_name != _INVALID_TRAILING_INPUT: + regs = re_match.regs + reg = regs[group_index] + node = self._group_names_to_nodes[group_name] + yield (node, reg) + + return list(get_tuples()) + + def _nodes_to_values(self) -> List[Tuple[str, str, Tuple[int, int]]]: + """ + Returns list of (Node, string_value) tuples. + """ + + def is_none(sl: Tuple[int, int]) -> bool: + return sl[0] == -1 and sl[1] == -1 + + def get(sl: Tuple[int, int]) -> str: + return self.string[sl[0] : sl[1]] + + return [ + (varname, get(slice), slice) + for varname, slice in self._nodes_to_regs() + if not is_none(slice) + ] + + def _unescape(self, varname: str, value: str) -> str: + unwrapper = self._unescape_funcs.get(varname) + return unwrapper(value) if unwrapper else value + + def variables(self) -> "Variables": + """ + Returns :class:`Variables` instance. + """ + return Variables( + [(k, self._unescape(k, v), sl) for k, v, sl in self._nodes_to_values()] + ) + + def trailing_input(self) -> Optional["MatchVariable"]: + """ + Get the `MatchVariable` instance, representing trailing input, if there is any. + "Trailing input" is input at the end that does not match the grammar anymore, but + when this is removed from the end of the input, the input would be a valid string. + """ + slices: List[Tuple[int, int]] = [] + + # Find all regex group for the name _INVALID_TRAILING_INPUT. + for r, re_match in self._re_matches: + for group_name, group_index in r.groupindex.items(): + if group_name == _INVALID_TRAILING_INPUT: + slices.append(re_match.regs[group_index]) + + # Take the smallest part. (Smaller trailing text means that a larger input has + # been matched, so that is better.) + if slices: + slice = (max(i[0] for i in slices), max(i[1] for i in slices)) + value = self.string[slice[0] : slice[1]] + return MatchVariable("<trailing_input>", value, slice) + return None + + def end_nodes(self) -> Iterable["MatchVariable"]: + """ + Yields `MatchVariable` instances for all the nodes having their end + position at the end of the input string. + """ + for varname, reg in self._nodes_to_regs(): + # If this part goes until the end of the input string. + if reg[1] == len(self.string): + value = self._unescape(varname, self.string[reg[0] : reg[1]]) + yield MatchVariable(varname, value, (reg[0], reg[1])) + + +class Variables: + def __init__(self, tuples: List[Tuple[str, str, Tuple[int, int]]]) -> None: + #: List of (varname, value, slice) tuples. + self._tuples = tuples + + def __repr__(self) -> str: + return "{}({})".format( + self.__class__.__name__, + ", ".join(f"{k}={v!r}" for k, v, _ in self._tuples), + ) + + def get(self, key: str, default: Optional[str] = None) -> Optional[str]: + items = self.getall(key) + return items[0] if items else default + + def getall(self, key: str) -> List[str]: + return [v for k, v, _ in self._tuples if k == key] + + def __getitem__(self, key: str) -> Optional[str]: + return self.get(key) + + def __iter__(self) -> Iterator["MatchVariable"]: + """ + Yield `MatchVariable` instances. + """ + for varname, value, slice in self._tuples: + yield MatchVariable(varname, value, slice) + + +class MatchVariable: + """ + Represents a match of a variable in the grammar. + + :param varname: (string) Name of the variable. + :param value: (string) Value of this variable. + :param slice: (start, stop) tuple, indicating the position of this variable + in the input string. + """ + + def __init__(self, varname: str, value: str, slice: Tuple[int, int]) -> None: + self.varname = varname + self.value = value + self.slice = slice + + self.start = self.slice[0] + self.stop = self.slice[1] + + def __repr__(self) -> str: + return f"{self.__class__.__name__}({self.varname!r}, {self.value!r})" + + +def compile( + expression: str, + escape_funcs: Optional[EscapeFuncDict] = None, + unescape_funcs: Optional[EscapeFuncDict] = None, +) -> _CompiledGrammar: + """ + Compile grammar (given as regex string), returning a `CompiledGrammar` + instance. + """ + return _compile_from_parse_tree( + parse_regex(tokenize_regex(expression)), + escape_funcs=escape_funcs, + unescape_funcs=unescape_funcs, + ) + + +def _compile_from_parse_tree( + root_node: Node, + escape_funcs: Optional[EscapeFuncDict] = None, + unescape_funcs: Optional[EscapeFuncDict] = None, +) -> _CompiledGrammar: + """ + Compile grammar (given as parse tree), returning a `CompiledGrammar` + instance. + """ + return _CompiledGrammar( + root_node, escape_funcs=escape_funcs, unescape_funcs=unescape_funcs + ) diff --git a/src/prompt_toolkit/contrib/regular_languages/completion.py b/src/prompt_toolkit/contrib/regular_languages/completion.py new file mode 100644 index 0000000..3cebcc0 --- /dev/null +++ b/src/prompt_toolkit/contrib/regular_languages/completion.py @@ -0,0 +1,93 @@ +""" +Completer for a regular grammar. +""" +from typing import Dict, Iterable, List + +from prompt_toolkit.completion import CompleteEvent, Completer, Completion +from prompt_toolkit.document import Document + +from .compiler import Match, _CompiledGrammar + +__all__ = [ + "GrammarCompleter", +] + + +class GrammarCompleter(Completer): + """ + Completer which can be used for autocompletion according to variables in + the grammar. Each variable can have a different autocompleter. + + :param compiled_grammar: `GrammarCompleter` instance. + :param completers: `dict` mapping variable names of the grammar to the + `Completer` instances to be used for each variable. + """ + + def __init__( + self, compiled_grammar: _CompiledGrammar, completers: Dict[str, Completer] + ) -> None: + + self.compiled_grammar = compiled_grammar + self.completers = completers + + def get_completions( + self, document: Document, complete_event: CompleteEvent + ) -> Iterable[Completion]: + m = self.compiled_grammar.match_prefix(document.text_before_cursor) + + if m: + completions = self._remove_duplicates( + self._get_completions_for_match(m, complete_event) + ) + + yield from completions + + def _get_completions_for_match( + self, match: Match, complete_event: CompleteEvent + ) -> Iterable[Completion]: + """ + Yield all the possible completions for this input string. + (The completer assumes that the cursor position was at the end of the + input string.) + """ + for match_variable in match.end_nodes(): + varname = match_variable.varname + start = match_variable.start + + completer = self.completers.get(varname) + + if completer: + text = match_variable.value + + # Unwrap text. + unwrapped_text = self.compiled_grammar.unescape(varname, text) + + # Create a document, for the completions API (text/cursor_position) + document = Document(unwrapped_text, len(unwrapped_text)) + + # Call completer + for completion in completer.get_completions(document, complete_event): + new_text = ( + unwrapped_text[: len(text) + completion.start_position] + + completion.text + ) + + # Wrap again. + yield Completion( + text=self.compiled_grammar.escape(varname, new_text), + start_position=start - len(match.string), + display=completion.display, + display_meta=completion.display_meta, + ) + + def _remove_duplicates(self, items: Iterable[Completion]) -> List[Completion]: + """ + Remove duplicates, while keeping the order. + (Sometimes we have duplicates, because the there several matches of the + same grammar, each yielding similar completions.) + """ + result: List[Completion] = [] + for i in items: + if i not in result: + result.append(i) + return result diff --git a/src/prompt_toolkit/contrib/regular_languages/lexer.py b/src/prompt_toolkit/contrib/regular_languages/lexer.py new file mode 100644 index 0000000..1ddeede --- /dev/null +++ b/src/prompt_toolkit/contrib/regular_languages/lexer.py @@ -0,0 +1,92 @@ +""" +`GrammarLexer` is compatible with other lexers and can be used to highlight +the input using a regular grammar with annotations. +""" +from typing import Callable, Dict, Optional + +from prompt_toolkit.document import Document +from prompt_toolkit.formatted_text.base import StyleAndTextTuples +from prompt_toolkit.formatted_text.utils import split_lines +from prompt_toolkit.lexers import Lexer + +from .compiler import _CompiledGrammar + +__all__ = [ + "GrammarLexer", +] + + +class GrammarLexer(Lexer): + """ + Lexer which can be used for highlighting of fragments according to variables in the grammar. + + (It does not actual lexing of the string, but it exposes an API, compatible + with the Pygments lexer class.) + + :param compiled_grammar: Grammar as returned by the `compile()` function. + :param lexers: Dictionary mapping variable names of the regular grammar to + the lexers that should be used for this part. (This can + call other lexers recursively.) If you wish a part of the + grammar to just get one fragment, use a + `prompt_toolkit.lexers.SimpleLexer`. + """ + + def __init__( + self, + compiled_grammar: _CompiledGrammar, + default_style: str = "", + lexers: Optional[Dict[str, Lexer]] = None, + ) -> None: + + self.compiled_grammar = compiled_grammar + self.default_style = default_style + self.lexers = lexers or {} + + def _get_text_fragments(self, text: str) -> StyleAndTextTuples: + m = self.compiled_grammar.match_prefix(text) + + if m: + characters: StyleAndTextTuples = [(self.default_style, c) for c in text] + + for v in m.variables(): + # If we have a `Lexer` instance for this part of the input. + # Tokenize recursively and apply tokens. + lexer = self.lexers.get(v.varname) + + if lexer: + document = Document(text[v.start : v.stop]) + lexer_tokens_for_line = lexer.lex_document(document) + text_fragments: StyleAndTextTuples = [] + for i in range(len(document.lines)): + text_fragments.extend(lexer_tokens_for_line(i)) + text_fragments.append(("", "\n")) + if text_fragments: + text_fragments.pop() + + i = v.start + for t, s, *_ in text_fragments: + for c in s: + if characters[i][0] == self.default_style: + characters[i] = (t, characters[i][1]) + i += 1 + + # Highlight trailing input. + trailing_input = m.trailing_input() + if trailing_input: + for i in range(trailing_input.start, trailing_input.stop): + characters[i] = ("class:trailing-input", characters[i][1]) + + return characters + else: + return [("", text)] + + def lex_document(self, document: Document) -> Callable[[int], StyleAndTextTuples]: + lines = list(split_lines(self._get_text_fragments(document.text))) + + def get_line(lineno: int) -> StyleAndTextTuples: + try: + return lines[lineno] + except IndexError: + return [] + + return get_line diff --git a/src/prompt_toolkit/contrib/regular_languages/regex_parser.py b/src/prompt_toolkit/contrib/regular_languages/regex_parser.py new file mode 100644 index 0000000..87e39d4 --- /dev/null +++ b/src/prompt_toolkit/contrib/regular_languages/regex_parser.py @@ -0,0 +1,281 @@ +""" +Parser for parsing a regular expression. +Take a string representing a regular expression and return the root node of its +parse tree. + +usage:: + + root_node = parse_regex('(hello|world)') + +Remarks: +- The regex parser processes multiline, it ignores all whitespace and supports + multiple named groups with the same name and #-style comments. + +Limitations: +- Lookahead is not supported. +""" +import re +from typing import List, Optional + +__all__ = [ + "Repeat", + "Variable", + "Regex", + "Lookahead", + "tokenize_regex", + "parse_regex", +] + + +class Node: + """ + Base class for all the grammar nodes. + (You don't initialize this one.) + """ + + def __add__(self, other_node: "Node") -> "NodeSequence": + return NodeSequence([self, other_node]) + + def __or__(self, other_node: "Node") -> "AnyNode": + return AnyNode([self, other_node]) + + +class AnyNode(Node): + """ + Union operation (OR operation) between several grammars. You don't + initialize this yourself, but it's a result of a "Grammar1 | Grammar2" + operation. + """ + + def __init__(self, children: List[Node]) -> None: + self.children = children + + def __or__(self, other_node: Node) -> "AnyNode": + return AnyNode(self.children + [other_node]) + + def __repr__(self) -> str: + return f"{self.__class__.__name__}({self.children!r})" + + +class NodeSequence(Node): + """ + Concatenation operation of several grammars. You don't initialize this + yourself, but it's a result of a "Grammar1 + Grammar2" operation. + """ + + def __init__(self, children: List[Node]) -> None: + self.children = children + + def __add__(self, other_node: Node) -> "NodeSequence": + return NodeSequence(self.children + [other_node]) + + def __repr__(self) -> str: + return f"{self.__class__.__name__}({self.children!r})" + + +class Regex(Node): + """ + Regular expression. + """ + + def __init__(self, regex: str) -> None: + re.compile(regex) # Validate + + self.regex = regex + + def __repr__(self) -> str: + return f"{self.__class__.__name__}(/{self.regex}/)" + + +class Lookahead(Node): + """ + Lookahead expression. + """ + + def __init__(self, childnode: Node, negative: bool = False) -> None: + self.childnode = childnode + self.negative = negative + + def __repr__(self) -> str: + return f"{self.__class__.__name__}({self.childnode!r})" + + +class Variable(Node): + """ + Mark a variable in the regular grammar. This will be translated into a + named group. Each variable can have his own completer, validator, etc.. + + :param childnode: The grammar which is wrapped inside this variable. + :param varname: String. + """ + + def __init__(self, childnode: Node, varname: str = "") -> None: + self.childnode = childnode + self.varname = varname + + def __repr__(self) -> str: + return "{}(childnode={!r}, varname={!r})".format( + self.__class__.__name__, + self.childnode, + self.varname, + ) + + +class Repeat(Node): + def __init__( + self, + childnode: Node, + min_repeat: int = 0, + max_repeat: Optional[int] = None, + greedy: bool = True, + ) -> None: + self.childnode = childnode + self.min_repeat = min_repeat + self.max_repeat = max_repeat + self.greedy = greedy + + def __repr__(self) -> str: + return f"{self.__class__.__name__}(childnode={self.childnode!r})" + + +def tokenize_regex(input: str) -> List[str]: + """ + Takes a string, representing a regular expression as input, and tokenizes + it. + + :param input: string, representing a regular expression. + :returns: List of tokens. + """ + # Regular expression for tokenizing other regular expressions. + p = re.compile( + r"""^( + \(\?P\<[a-zA-Z0-9_-]+\> | # Start of named group. + \(\?#[^)]*\) | # Comment + \(\?= | # Start of lookahead assertion + \(\?! | # Start of negative lookahead assertion + \(\?<= | # If preceded by. + \(\?< | # If not preceded by. + \(?: | # Start of group. (non capturing.) + \( | # Start of group. + \(?[iLmsux] | # Flags. + \(?P=[a-zA-Z]+\) | # Back reference to named group + \) | # End of group. + \{[^{}]*\} | # Repetition + \*\? | \+\? | \?\?\ | # Non greedy repetition. + \* | \+ | \? | # Repetition + \#.*\n | # Comment + \\. | + + # Character group. + \[ + ( [^\]\\] | \\.)* + \] | + + [^(){}] | + . + )""", + re.VERBOSE, + ) + + tokens = [] + + while input: + m = p.match(input) + if m: + token, input = input[: m.end()], input[m.end() :] + if not token.isspace(): + tokens.append(token) + else: + raise Exception("Could not tokenize input regex.") + + return tokens + + +def parse_regex(regex_tokens: List[str]) -> Node: + """ + Takes a list of tokens from the tokenizer, and returns a parse tree. + """ + # We add a closing brace because that represents the final pop of the stack. + tokens: List[str] = [")"] + regex_tokens[::-1] + + def wrap(lst: List[Node]) -> Node: + """Turn list into sequence when it contains several items.""" + if len(lst) == 1: + return lst[0] + else: + return NodeSequence(lst) + + def _parse() -> Node: + or_list: List[List[Node]] = [] + result: List[Node] = [] + + def wrapped_result() -> Node: + if or_list == []: + return wrap(result) + else: + or_list.append(result) + return AnyNode([wrap(i) for i in or_list]) + + while tokens: + t = tokens.pop() + + if t.startswith("(?P<"): + variable = Variable(_parse(), varname=t[4:-1]) + result.append(variable) + + elif t in ("*", "*?"): + greedy = t == "*" + result[-1] = Repeat(result[-1], greedy=greedy) + + elif t in ("+", "+?"): + greedy = t == "+" + result[-1] = Repeat(result[-1], min_repeat=1, greedy=greedy) + + elif t in ("?", "??"): + if result == []: + raise Exception("Nothing to repeat." + repr(tokens)) + else: + greedy = t == "?" + result[-1] = Repeat( + result[-1], min_repeat=0, max_repeat=1, greedy=greedy + ) + + elif t == "|": + or_list.append(result) + result = [] + + elif t in ("(", "(?:"): + result.append(_parse()) + + elif t == "(?!": + result.append(Lookahead(_parse(), negative=True)) + + elif t == "(?=": + result.append(Lookahead(_parse(), negative=False)) + + elif t == ")": + return wrapped_result() + + elif t.startswith("#"): + pass + + elif t.startswith("{"): + # TODO: implement! + raise Exception(f"{t}-style repetition not yet supported") + + elif t.startswith("(?"): + raise Exception("%r not supported" % t) + + elif t.isspace(): + pass + else: + result.append(Regex(t)) + + raise Exception("Expecting ')' token") + + result = _parse() + + if len(tokens) != 0: + raise Exception("Unmatched parentheses.") + else: + return result diff --git a/src/prompt_toolkit/contrib/regular_languages/validation.py b/src/prompt_toolkit/contrib/regular_languages/validation.py new file mode 100644 index 0000000..71d3434 --- /dev/null +++ b/src/prompt_toolkit/contrib/regular_languages/validation.py @@ -0,0 +1,60 @@ +""" +Validator for a regular language. +""" +from typing import Dict + +from prompt_toolkit.document import Document +from prompt_toolkit.validation import ValidationError, Validator + +from .compiler import _CompiledGrammar + +__all__ = [ + "GrammarValidator", +] + + +class GrammarValidator(Validator): + """ + Validator which can be used for validation according to variables in + the grammar. Each variable can have its own validator. + + :param compiled_grammar: `GrammarCompleter` instance. + :param validators: `dict` mapping variable names of the grammar to the + `Validator` instances to be used for each variable. + """ + + def __init__( + self, compiled_grammar: _CompiledGrammar, validators: Dict[str, Validator] + ) -> None: + + self.compiled_grammar = compiled_grammar + self.validators = validators + + def validate(self, document: Document) -> None: + # Parse input document. + # We use `match`, not `match_prefix`, because for validation, we want + # the actual, unambiguous interpretation of the input. + m = self.compiled_grammar.match(document.text) + + if m: + for v in m.variables(): + validator = self.validators.get(v.varname) + + if validator: + # Unescape text. + unwrapped_text = self.compiled_grammar.unescape(v.varname, v.value) + + # Create a document, for the completions API (text/cursor_position) + inner_document = Document(unwrapped_text, len(unwrapped_text)) + + try: + validator.validate(inner_document) + except ValidationError as e: + raise ValidationError( + cursor_position=v.start + e.cursor_position, + message=e.message, + ) from e + else: + raise ValidationError( + cursor_position=len(document.text), message="Invalid command" + ) |