Viewing file:  _tokenizer.py (5.15 KB) -rw-r--r--
Select action/file-type:
 (+) |  (+) |  (+) | Code (+) | Session (+) |  (+) | SDB (+) |  (+) |  (+) |  (+) |  (+) |  (+) |
from __future__ import annotations
import contextlib
import re
from dataclasses import dataclass
from typing import Iterator, NoReturn
from .specifiers import Specifier
@dataclass
class Token:
name: str
text: str
position: int
class ParserSyntaxError(Exception):
"""The provided source text could not be parsed correctly."""
def __init__(
self,
message: str,
*,
source: str,
span: tuple[int, int],
) -> None:
self.span = span
self.message = message
self.source = source
super().__init__()
def __str__(self) -> str:
marker = " " * self.span[0] + "~" * (self.span[1] - self.span[0]) + "^"
return "\n ".join([self.message, self.source, marker])
DEFAULT_RULES: dict[str, str | re.Pattern[str]] = {
"LEFT_PARENTHESIS": r"\(",
"RIGHT_PARENTHESIS": r"\)",
"LEFT_BRACKET": r"\[",
"RIGHT_BRACKET": r"\]",
"SEMICOLON": r";",
"COMMA": r",",
"QUOTED_STRING": re.compile(
r"""
(
('[^']*')
|
("[^"]*")
)
""",
re.VERBOSE,
),
"OP": r"(===|==|~=|!=|<=|>=|<|>)",
"BOOLOP": r"\b(or|and)\b",
"IN": r"\bin\b",
"NOT": r"\bnot\b",
"VARIABLE": re.compile(
r"""
\b(
python_version
|python_full_version
|os[._]name
|sys[._]platform
|platform_(release|system)
|platform[._](version|machine|python_implementation)
|python_implementation
|implementation_(name|version)
|extra
)\b
""",
re.VERBOSE,
),
"SPECIFIER": re.compile(
Specifier._operator_regex_str + Specifier._version_regex_str,
re.VERBOSE | re.IGNORECASE,
),
"AT": r"\@",
"URL": r"[^ \t]+",
"IDENTIFIER": r"\b[a-zA-Z0-9][a-zA-Z0-9._-]*\b",
"VERSION_PREFIX_TRAIL": r"\.\*",
"VERSION_LOCAL_LABEL_TRAIL": r"\+[a-z0-9]+(?:[-_\.][a-z0-9]+)*",
"WS": r"[ \t]+",
"END": r"$",
}
class Tokenizer:
"""Context-sensitive token parsing.
Provides methods to examine the input stream to check whether the next token
matches.
"""
def __init__(
self,
source: str,
*,
rules: dict[str, str | re.Pattern[str]],
) -> None:
self.source = source
self.rules: dict[str, re.Pattern[str]] = {
name: re.compile(pattern) for name, pattern in rules.items()
}
self.next_token: Token | None = None
self.position = 0
def consume(self, name: str) -> None:
"""Move beyond provided token name, if at current position."""
if self.check(name):
self.read()
def check(self, name: str, *, peek: bool = False) -> bool:
"""Check whether the next token has the provided name.
By default, if the check succeeds, the token *must* be read before
another check. If `peek` is set to `True`, the token is not loaded and
would need to be checked again.
"""
assert (
self.next_token is None
), f"Cannot check for {name!r}, already have {self.next_token!r}"
assert name in self.rules, f"Unknown token name: {name!r}"
expression = self.rules[name]
match = expression.match(self.source, self.position)
if match is None:
return False
if not peek:
self.next_token = Token(name, match[0], self.position)
return True
def expect(self, name: str, *, expected: str) -> Token:
"""Expect a certain token name next, failing with a syntax error otherwise.
The token is *not* read.
"""
if not self.check(name):
raise self.raise_syntax_error(f"Expected {expected}")
return self.read()
def read(self) -> Token:
"""Consume the next token and return it."""
token = self.next_token
assert token is not None
self.position += len(token.text)
self.next_token = None
return token
def raise_syntax_error(
self,
message: str,
*,
span_start: int | None = None,
span_end: int | None = None,
) -> NoReturn:
"""Raise ParserSyntaxError at the given position."""
span = (
self.position if span_start is None else span_start,
self.position if span_end is None else span_end,
)
raise ParserSyntaxError(
message,
source=self.source,
span=span,
)
@contextlib.contextmanager
def enclosing_tokens(
self, open_token: str, close_token: str, *, around: str
) -> Iterator[None]:
if self.check(open_token):
open_position = self.position
self.read()
else:
open_position = None
yield
if open_position is None:
return
if not self.check(close_token):
self.raise_syntax_error(
f"Expected matching {close_token} for {open_token}, after {around}",
span_start=open_position,
)
self.read()
|