rgbds anonymous label

#!/usr/bin/env python3
"""
Indentation translator for RGBASM
Copyright 2018 Damian Yerrick
(insert zlib License here)

bgrdedent.py is a preprocessor for LR35902 assembly language source
code files intended to be assembled using the RGBDS assembler
([RGBASM]).  It translates the "followed by colon" convention for
denoting labels that [ca65] and other modern assemblers use to the
"begins in first column" convention that punch-card-era assemblers
used and RGBASM continues to use.

First all leading and trailing whitespace is removed.  Then decide
whether or not to re-add a leading space based on the first of the
following rules that applies for each line, where a "word" is a run
of non-whitespace.

* If the previous line ended with a backslash, it is a line
  continuation. Start this line at column 2.
* If the line is empty, is blank, output a blank line.
* If the first word is `SECTION`, `EXPORT`, `GLOBAL`, `UNION`,
  `NEXTU`, or `ENDU` (case insensitive), start this line at column 1.
* If the first word does not contain a quotation mark or semicolon,
  and the second word is `EQU`, `SET`, `RB`, `RW`, `RL`, or `EQUS`
  (case insensitive), start this line at column 1.
* If the first word contains a colon or equal sign,
  start this line at column 1.
* If the second word begins with a colon or equal sign,
  start this line at column 1.
* Start everything else at column 2.


[RGBASM]: https://rednex.github.io/rgbds/rgbasm.5.html
[ca65]: https://cc65.github.io/doc/ca65.html

"""
import sys
import argparse

anonymous_counter = 0

def anonymous_label(counter):
    return ".anon%d" % counter

def fixcolumns(lines):
    global anonymous_counter
    last_was_continue = False
    # first words that signal no indenting
    word0s = {'section', 'export', 'global', 'union', 'nextu', 'endu'}
    # second words that signal no indenting
    word1s = {'equ', 'set', 'rb', 'rw', 'rl', 'equs'}
    # instruction parts
    jumps = {'jr', 'jp', 'call'}
    conditions = {'z', 'nz', 'c', 'nc'}

    for line in lines:
        line = line.strip()            # remove indenting
        owords = line.split()          # keep the original words around
        lwords = line.lower().split()  # separate it into words and lowercase it
        start = ' '                    # start off with an automatic indent
        # process indenting
        if last_was_continue:
            pass
        elif len(lwords) == 0:         # empty line
            start = ''
        elif lwords[0] in word0s:
            start = ''
        elif (len(lwords) > 1 and lwords[1] in word1s
              and ";" not in lwords[0] and ":" not in lwords[0]):
            start = ''
        elif lwords[0] == ':':         # anonymous label
            anonymous_counter += 1
            start = anonymous_label(anonymous_counter)
        elif ':' in lwords[0] or '=' in lwords[0]:
            start = ''
        elif len(lwords) > 1 and lwords[1].startswith((':', '=')) and lwords[0] not in jumps:
            start = ''

        # process the actual instructions
        instruction_i = 0
        if len(lwords) >= 1:
            if ':' in lwords[0]:
                instruction_i = 1
            if len(lwords)-instruction_i >= 2 and lwords[instruction_i] in jumps:
                # find the destination
                destination_i = instruction_i+1
                if len(lwords)-instruction_i >= 3 \
                and lwords[instruction_i+1].rstrip(',') in conditions:
                    destination_i += 1

                # parse destination
                destination = lwords[destination_i]
                if destination[0] == ':' and len(destination) > 1:
                    anon_destination = anonymous_counter
                    if destination[1] == '+':
                        anon_destination += len(destination)-1
                    elif destination[1] == '-':
                        anon_destination -= len(destination)-2
                    else:
                        print("bad anonymous label reference")
                    # patch the destination and recreate the line
                    owords[destination_i] = anonymous_label(anon_destination)
                    line = ' '.join(owords)

        yield "".join((start, line, "\n"))
        last_was_continue = line.endswith("\\")

def parse_argv(argv):
    parser = argparse.ArgumentParser()
    parser.add_argument("input", default="-", nargs='?',
                        help="file to dedent (standard input if omitted)")
    parser.add_argument("-o", "--output", default="-",
                        help="file to write (standard output if omitted)")
    return parser.parse_args(argv[1:])

def main(argv=None):
    args = parse_argv(argv or sys.argv)

    infp, outfp = sys.stdin, sys.stdout
    try:
        if args.input != '-':
            infp = open(args.input, "r")
        if args.output != '-':
            outfp = open(args.output, "w")
        outfp.writelines(fixcolumns(infp))
    finally:
        if outfp is not sys.stdout:
            outfp.close()
        if infp is not sys.stdin:
            infp.close()

if __name__=='__main__':
    main()