plexlab.py

"""

input :
123 8 0x7ff 45.
23.67 .89000
0x5A 6.2

output:
(venv) ionianironist@debian:~/PycharmProjects/untitled$ python lab.py
INT_TOKEN 123
INT_TOKEN 8
HEX_TOKEN 0x7ff
FLOAT_TOKEN 45.
FLOAT_TOKEN 23.67
FLOAT_TOKEN .89000
HEX_TOKEN 0x5A
FLOAT_TOKEN 6.2


"""


import plex

# define patterns
# letter = plex.Range("AZaz")
digit = plex.Range("09")
dot = plex.Str(".")
integer = plex.Rep1(digit)
floating = (integer + dot + plex.Opt(integer)) | (dot + integer)
hexPrefix = plex.Str("0x")
hexDigit = plex.Range("09AFaf")
hexadecimal = hexPrefix + plex.Rep1(hexDigit)
# the scanner lexicon - constructor argument is a list of (pattern,action ) tuples
lexicon = plex.Lexicon([
    (integer, "INT_TOKEN"),
    (floating, "FLOAT_TOKEN"),
    (hexadecimal, "HEX_TOKEN"),
    (plex.Rep1(plex.Any(" \t\n")), plex.IGNORE)
    ])

with open("identifiers.txr", "r") as fp:
    scanner = plex.Scanner(lexicon, fp)

    while True:
        try:
            token, matched = scanner.read()

        except plex.errors.PlexError:

            _, lineno, charno = scanner.position()  # lineno is 1-based, charno is 0-based
            print("Scanner Error at line {} char {}".format(lineno, charno + 1))
            break  # lexical analysis ends after error

        if not token:
            break  # reached end-of-text (EOT)

        print(token, matched)