scheme_tokens.py

"""The scheme_tokens module provides functions tokenize_line and tokenize_lines
for converting (iterators producing) strings into (iterators producing) lists
of tokens.  A token may be:

  * A number (represented as an int or float)
  * A boolean (represented as a bool)
  * A symbol (represented as a string)
  * A delimiter, including parentheses, dots, and single quotes

This file also includes some features of Scheme that have not been addressed
in the course, such as Scheme strings.
"""

from ucb import main
import sys

_NUMERAL_STARTS = set("0123456789") | set('+-.')
_SYMBOL_CHARS = (set('!$%&*/:<=>?@^_~') | set("abcdefghijklmnopqrstuvwxyz") |
                 set("ABCDEFGHIJKLMNOPQRSTUVWXYZ") | _NUMERAL_STARTS)
_STRING_DELIMS = set('"')
_WHITESPACE = set(' \t\n\r')
_SINGLE_CHAR_TOKENS = set("()[]'`")
_TOKEN_END = _WHITESPACE | _SINGLE_CHAR_TOKENS | _STRING_DELIMS | {',', ',@'}
DELIMITERS = _SINGLE_CHAR_TOKENS | {'.', ',', ',@'}
_MAX_TOKEN_LENGTH = 50


def chain(*iters):
    for iter in iters:
        yield from iter


def valid_symbol(s):
    """Returns whether s is a well-formed symbol."""
    if len(s) == 0:
        return False
    for c in s:
        if c not in _SYMBOL_CHARS:
            return False
    return True


def next_candidate_token(line, k):
    """A tuple (tok, k'), where tok is the next substring of line at or
    after position k that could be a token (assuming it passes a validity
    check), and k' is the position in line following that token.  Returns
    (None, len(line)) when there are no more tokens."""
    while k < len(line):
        c = line[k]
        if c == ';':
            return None, len(line)
        elif c in _WHITESPACE:
            k += 1
        elif c in _SINGLE_CHAR_TOKENS:
            if c == ']': c = ')'
            if c == '[': c = '('
            return c, k + 1
        elif c == '#':  # Boolean values #t and #f
            return line[k:k + 2], min(k + 2, len(line))
        elif c == ',':  # Unquote; check for @
            if k + 1 < len(line) and line[k + 1] == '@':
                return ',@', k + 2
            return c, k + 1
        elif c in _STRING_DELIMS:
            if k + 1 < len(line) and line[k + 1] == c:  # No triple quotes in Scheme
                return c + c, k + 2
            s = ""
            k += 1
            while k < len(line):
                c = line[k]
                if c == "\"":
                    check_token_length_warning(s, len(s) + 2)
                    return "\"" + s + "\"", k + 1
                elif c == "\\":
                    if k + 1 == len(line):
                        raise SyntaxError("String ended abruptly")
                    next = line[k + 1]
                    if next == "n":
                        s += "\n"
                    else:
                        s += next
                    k += 2
                else:
                    s += c
                    k += 1
            raise SyntaxError("String ended abruptly")
        else:
            j = k
            while j < len(line) and line[j] not in _TOKEN_END:
                j += 1
            check_token_length_warning(line[k:j], min(j, len(line)) - k)
            return line[k:j], min(j, len(line))
    return None, len(line)


def tokenize_line(line):
    """The list of Scheme tokens on line.  Excludes comments and whitespace."""
    result = []
    text, i = next_candidate_token(line, 0)
    while text is not None:
        if text in DELIMITERS:
            result.append(text)
        elif text == '#t' or text.lower() == 'true':
            result.append(True)
        elif text == '#f' or text.lower() == 'false':
            result.append(False)
        elif text == 'nil':
            result.append(text)
        elif text[0] in _SYMBOL_CHARS:
            number = False
            if text[0] in _NUMERAL_STARTS:
                try:
                    result.append(int(text))
                    number = True
                except ValueError:
                    try:
                        result.append(float(text))
                        number = True
                    except ValueError:
                        pass
            if not number:
                if valid_symbol(text):
                    result.append(text.lower())
                else:
                    raise ValueError("invalid numeral or symbol: {0}".format(text))
        elif text[0] in _STRING_DELIMS:
            result.append(text)
        else:
            error_message = [
                "warning: invalid token: {0}".format(text),
                " " * 4 + line,
                " " * (i + 4) + "^"
            ]
            raise ValueError("\n".join(error_message))
        text, i = next_candidate_token(line, i)
    return result


def check_token_length_warning(token, length):
    if length > _MAX_TOKEN_LENGTH:
        import warnings
        warnings.warn("Token {} has exceeded the maximum token length {}".format(token, _MAX_TOKEN_LENGTH, length))


def tokenize_lines(inp):
    """An iterator over lists of tokens, one for each line of the iterable
    input sequence inp."""
    return (tokenize_line(line) for line in inp)


def count_tokens(inp):
    """Count the number of non-delimiter tokens in inp."""
    return len(list(chain(*tokenize_lines(inp))))


@main
def run(*args):
    import argparse
    parser = argparse.ArgumentParser(description='Count Scheme tokens.')
    parser.add_argument('file', nargs='?',
                        type=argparse.FileType('r'), default=sys.stdin,
                        help='input file to be counted')
    args = parser.parse_args()
    print('counted', count_tokens(args.file), 'tokens')