feat(parser): add midas lexer to test script

the lexer and token structures where adapted from another project (see docstring on the Lexer class)

.gitignore

@@ -1 +1,6 @@
-.vscode
+.vscode
+__pycache__
+.env
+venv
+.venv
+*.pyc

examples/00_syntax_prototype/02_custom_types.midas

@@ -21,4 +21,4 @@ type Age<int + (0 <= _ < 150)>
 // Predefined custom constraints that can be referenced in other definitions
 constraint Positive = _ >= 0
 constraint StrictlyPositive = _ > 0
-constraint Even = _ % 2 == 0
+//constraint Even = _ % 2 == 0

lexer/annotations.py

@@ -0,0 +1,81 @@
+from lexer.base import Lexer
+from lexer.token import TokenType
+
+
+class AnnotationLexer(Lexer):
+    def scan_token(self) -> None:
+        char: str = self.advance()
+        match char:
+            case "(":
+                self.add_token(TokenType.LEFT_PAREN)
+            case ")":
+                self.add_token(TokenType.RIGHT_PAREN)
+            case "[":
+                self.add_token(TokenType.LEFT_BRACKET)
+            case "]":
+                self.add_token(TokenType.RIGHT_BRACKET)
+            case ":":
+                self.add_token(TokenType.COLON)
+            case ",":
+                self.add_token(TokenType.COMMA)
+            case "_":
+                self.add_token(TokenType.UNDERSCORE)
+            case "+":
+                self.add_token(TokenType.PLUS)
+            case "#":
+                self.scan_comment()
+            case "\n":
+                self.add_token(TokenType.NEWLINE)
+            case " " | "\r" | "\t":
+                # Consume all whitespace characters until EOL or EOF
+                while (
+                    self.peek().isspace()
+                    and self.peek() != "\n"
+                    and not self.is_at_end()
+                ):
+                    self.advance()
+                self.add_token(TokenType.WHITESPACE)
+            case _:
+                if char.isdigit():
+                    self.scan_number()
+                elif char.isalpha():
+                    self.scan_identifier()
+                else:
+                    self.error("Unexpected character")
+        return None
+
+    def scan_number(self):
+        """Scan the rest of number and add it as a token
+
+        This method handles both simple integers and floats. Scientific notation
+        and base prefixes (0x, 0b, 0o) are not supported
+        """
+        while self.peek().isdigit():
+            self.advance()
+
+        if self.peek() == "." and self.peek_next().isdigit():
+            self.advance()
+            while self.peek().isdigit():
+                self.advance()
+
+        value: float = float(self.source[self.start : self.idx])
+        self.add_token(TokenType.NUMBER, value)
+
+    def scan_identifier(self):
+        """Scan the rest of an identifier and add it as a token
+
+        An identifier starts with a letter, followed by any number of
+        alphanumerical characters or underscores
+        """
+        while self.peek().isalnum() or self.peek() == "_":
+            self.advance()
+        self.add_token(TokenType.IDENTIFIER)
+
+    def scan_comment(self):
+        """Scan the rest of a comment and add it as a token
+
+        A comment starts with a `#` character and ends at the EOL/EOF
+        """
+        while self.peek() != "\n" and not self.is_at_end():
+            self.advance()
+        self.add_token(TokenType.COMMENT)

lexer/base.py

@@ -0,0 +1,166 @@
+from abc import ABC, abstractmethod
+from typing import Any, Callable, Optional
+
+from lexer.position import Position
+from lexer.token import Token, TokenType
+
+
+class Lexer(ABC):
+    """An abstract lexer which provides methods to easily extend it into a concrete one
+
+    This implementation is based on the [_Crafting Interpreters_][1] book by Robert Nystrom,
+    more specifically on my [previous Python implementation](https://git.kb28.ch/HEL/pebble)
+
+    [1]: https://craftinginterpreters.com/
+    """
+
+    def __init__(self, source: str, file: Optional[str] = None) -> None:
+        """Create a new lexer to scan for tokens in the given source
+
+        Args:
+            source (str): the source to scan
+            file (Optional[str], optional): the path of the given source. Can be a file path or any string identifier. Defaults to None.
+        """
+        self.source: str = source
+        self.file: Optional[str] = file
+        self.tokens: list[Token] = []
+        self.start: int = 0
+        self.idx: int = 0
+        self.length: int = len(self.source)
+        self.line: int = 1
+        self.column: int = 1
+        self.start_pos: Position = self.get_position()
+
+    def error(self, msg: str):
+        """Raise a syntax error
+
+        Args:
+            msg (str): the error message
+
+        Raises:
+            SyntaxError
+        """
+        raise SyntaxError(f"[ERROR] Error at {self.start_pos}: {msg}")
+
+    def process(self) -> list[Token]:
+        """Scan tokens out of the source text
+
+        Returns:
+            list[Token]: all the tokens that could be scanned
+
+        Raises:
+            SyntaxError: if a syntax error is found
+        """
+        self.scan_tokens()
+        self.tokens.append(Token(TokenType.EOF, "", None, self.get_position()))
+        return self.tokens
+
+    def is_at_end(self) -> bool:
+        """Whether the lexer is at the end of the source
+
+        Returns:
+            bool: True if the current index is at the end of the source
+        """
+        return self.idx >= self.length
+
+    def get_position(self) -> Position:
+        """Get the current position
+
+        Returns:
+            Position: the current position
+        """
+        return Position(file=self.file, line=self.line, column=self.column)
+
+    def peek(self) -> str:
+        """Get the current character without advancing, if any
+
+        Returns:
+            str: the current character, or an empty string if at EOF
+        """
+        if self.idx < self.length:
+            return self.source[self.idx]
+        return ""
+
+    def peek_next(self) -> str:
+        """Get the next character without advancing, if any
+
+        Returns:
+            str: the next character, or an empty string if at EOF
+        """
+        if self.idx + 1 < self.length:
+            return self.source[self.idx + 1]
+        return ""
+
+    def advance(self) -> str:
+        """Get the new character and advance
+
+        Returns:
+            str: the current character, before advancing
+        """
+        char: str = self.peek()
+        self.idx += 1
+        self.column += 1
+        if char == "\n":
+            self.newline()
+        return char
+
+    def newline(self):
+        """Update the current position after encountering a newline character"""
+        self.line += 1
+        self.column = 1
+
+    def match(self, expected: str) -> bool:
+        """Consume the next character if it matches the given value
+
+        Args:
+            expected (str): the expected character
+
+        Returns:
+            bool: whether a character was matched and consumed
+        """
+        if self.peek() == expected:
+            self.advance()
+            return True
+        return False
+
+    def update_start(self):
+        """Update the starting position of the current lexeme
+
+        The cursor marking the start of the lexeme currently being scanned is
+        moved to the current position
+        """
+        self.start_pos = self.get_position()
+        self.start = self.idx
+
+    def add_token(self, token_type: TokenType, value: Optional[Any] = None):
+        """Add the current lexeme to the list of scanned tokens
+
+        Args:
+            token_type (TokenType): the type of token to add
+            value (Optional[Any], optional): the value of the token (useful for numbers or constants). Defaults to None.
+        """
+        lexeme: str = self.source[self.start : self.idx]
+        self.tokens.append(
+            Token(position=self.start_pos, type=token_type, lexeme=lexeme, value=value)
+        )
+
+    def scan_tokens(self, condition: Optional[Callable[[], bool]] = None):
+        """Scan tokens until EOF is reached or the given condition becomes False
+
+        Args:
+            condition (Optional[Callable[[], bool]], optional): the condition to continue scanning tokens.
+                If None, defaults to always being True, effectively scanning tokens until EOF is reached. Defaults to None.
+        """
+        if condition is None:
+            condition = lambda: True  # noqa: E731
+        while condition() and not self.is_at_end():
+            self.update_start()
+            self.scan_token()
+
+    @abstractmethod
+    def scan_token(self) -> None:
+        """Scan a token
+
+        This function should (at least) consume the current character and produce the appropriate token(s), using `add_token`
+        """
+        pass

lexer/keyword.py

@@ -0,0 +1,9 @@
+from lexer.token import TokenType
+
+KEYWORDS: dict[str, TokenType] = {
+    "type": TokenType.TYPE,
+    "op": TokenType.OP,
+    "constraint": TokenType.CONSTRAINT,
+    "true": TokenType.TRUE,
+    "false": TokenType.FALSE,
+}

lexer/midas.py

@@ -0,0 +1,126 @@
+from lexer.base import Lexer
+from lexer.keyword import KEYWORDS
+from lexer.token import TokenType
+
+
+class MidasLexer(Lexer):
+    def scan_token(self) -> None:
+        char: str = self.advance()
+        match char:
+            case "(":
+                self.add_token(TokenType.LEFT_PAREN)
+            case ")":
+                self.add_token(TokenType.RIGHT_PAREN)
+            case "[":
+                self.add_token(TokenType.LEFT_BRACKET)
+            case "]":
+                self.add_token(TokenType.RIGHT_BRACKET)
+            case "{":
+                self.add_token(TokenType.LEFT_BRACE)
+            case "}":
+                self.add_token(TokenType.RIGHT_BRACE)
+            case "<":
+                self.add_token(
+                    TokenType.LESS_EQUAL if self.match("=") else TokenType.LESS
+                )
+            case ">":
+                self.add_token(
+                    TokenType.GREATER_EQUAL if self.match("=") else TokenType.GREATER
+                )
+            case "=":
+                self.add_token(
+                    TokenType.EQUAL_EQUAL if self.match("=") else TokenType.EQUAL
+                )
+            case ":":
+                self.add_token(TokenType.COLON)
+            case ",":
+                self.add_token(TokenType.COMMA)
+            case "_":
+                self.add_token(TokenType.UNDERSCORE)
+            case "+":
+                self.add_token(TokenType.PLUS)
+            case "-":
+                self.add_token(TokenType.MINUS)
+            case "*":
+                self.add_token(TokenType.STAR)
+            case "/":
+                if self.match("/"):
+                    self.scan_comment()
+                elif self.match("*"):
+                    self.scan_comment_multiline()
+                else:
+                    self.add_token(TokenType.SLASH)
+            case "\n":
+                self.add_token(TokenType.NEWLINE)
+            case " " | "\r" | "\t":
+                # Consume all whitespace characters until EOL or EOF
+                while (
+                    self.peek().isspace()
+                    and self.peek() != "\n"
+                    and not self.is_at_end()
+                ):
+                    self.advance()
+                self.add_token(TokenType.WHITESPACE)
+            case _:
+                if char.isdigit():
+                    self.scan_number()
+                elif char.isalpha():
+                    self.scan_identifier()
+                else:
+                    self.error("Unexpected character")
+        return None
+
+    def scan_number(self):
+        """Scan the rest of number and add it as a token
+
+        This method handles both simple integers and floats. Scientific notation
+        and base prefixes (0x, 0b, 0o) are not supported
+        """
+        while self.peek().isdigit():
+            self.advance()
+
+        if self.peek() == "." and self.peek_next().isdigit():
+            self.advance()
+            while self.peek().isdigit():
+                self.advance()
+
+        value: float = float(self.source[self.start : self.idx])
+        self.add_token(TokenType.NUMBER, value)
+
+    def scan_identifier(self):
+        """Scan the rest of an identifier and add it as a token
+
+        An identifier starts with a letter, followed by any number of
+        alphanumerical characters or underscores
+        """
+        while self.peek().isalnum() or self.peek() == "_":
+            self.advance()
+
+        lexeme: str = self.source[self.start : self.idx]
+        token_type: TokenType = KEYWORDS.get(lexeme, TokenType.IDENTIFIER)
+        self.add_token(token_type)
+
+    def scan_comment(self):
+        """Scan the rest of a comment and add it as a token
+
+        A comment starts with `//` and ends at the EOL/EOF
+        """
+        while self.peek() != "\n" and not self.is_at_end():
+            self.advance()
+        self.add_token(TokenType.COMMENT)
+
+    def scan_comment_multiline(self):
+        """Scan the rest of a multiline comment and add it as a token
+
+        A multiline comment starts with `/*` and ends with `*/` or at the EOF
+        """
+        while (
+            not (self.peek() == "*" and self.peek_next() == "/")
+            and not self.is_at_end()
+        ):
+            self.advance()
+        if not self.is_at_end():
+            self.advance()
+        if not self.is_at_end():
+            self.advance()
+        self.add_token(TokenType.COMMENT)

lexer/position.py

@@ -0,0 +1,13 @@
+from dataclasses import dataclass
+from typing import Optional
+
+
+@dataclass(frozen=True)
+class Position:
+    """A simple structure to store the position of a token"""
+    file: Optional[str]
+    line: int
+    column: int
+
+    def __repr__(self):
+        return f"{self.file or ''}L{self.line}:{self.column}"

lexer/token.py

@@ -0,0 +1,58 @@
+from dataclasses import dataclass
+from enum import Enum, auto
+from typing import Any
+
+from lexer.position import Position
+
+
+class TokenType(Enum):
+    # Punctuation
+    LEFT_PAREN = auto()
+    RIGHT_PAREN = auto()
+    LEFT_BRACKET = auto()
+    RIGHT_BRACKET = auto()
+    LEFT_BRACE = auto()
+    RIGHT_BRACE = auto()
+    COLON = auto()
+    COMMA = auto()
+    UNDERSCORE = auto()
+
+    # Operators
+    PLUS = auto()
+    MINUS = auto()
+    STAR = auto()
+    SLASH = auto()
+    GREATER = auto()
+    GREATER_EQUAL = auto()
+    LESS = auto()
+    LESS_EQUAL = auto()
+    EQUAL = auto()
+    EQUAL_EQUAL = auto()
+
+    # Literals
+    IDENTIFIER = auto()
+    NUMBER = auto()
+    TRUE = auto()
+    FALSE = auto()
+    NONE = auto()
+
+    # Keywords
+    TYPE = auto()
+    OP = auto()
+    CONSTRAINT = auto()
+
+    # Misc
+    COMMENT = auto()
+    WHITESPACE = auto()
+    EOF = auto()
+    NEWLINE = auto()
+
+
+@dataclass(frozen=True)
+class Token:
+    """A scanned token"""
+
+    type: TokenType
+    lexeme: str
+    value: Any
+    position: Position

test.py

@@ -0,0 +1,28 @@
+import importlib
+from pathlib import Path
+
+from lexer.annotations import AnnotationLexer
+from lexer.midas import MidasLexer
+from lexer.token import Token
+
+
+# Frame annotation
+mod = importlib.import_module("examples.00_syntax_prototype.01_simple_types")
+
+annotation: str = mod.__annotations__["df"]
+lexer: AnnotationLexer = AnnotationLexer(annotation, "01_simple_types.py")
+tokens: list[Token] = lexer.process()
+print([
+    f"{t.type.name}('{t.lexeme}')"
+    for t in tokens
+])
+
+# Midas type definitions
+path: Path = Path("examples") / "00_syntax_prototype" / "02_custom_types.midas"
+definitions: str = path.read_text()
+midas_lexer: MidasLexer = MidasLexer(definitions, path.name)
+tokens = midas_lexer.process()
+print([
+    f"{t.type.name}('{t.lexeme}')"
+    for t in tokens
+])