yml2/yml2c

#!/usr/bin/env python
# vim: set fileencoding=utf-8 :

"""\
YML 2 compiler version 5.5
Copyleft (c), 2009-2011, Volker Birk  http://fdik.org/yml/

"""

import sys, os, codecs, locale
import fileinput, unicodedata
from optparse import OptionParser

from pyPEG import parse, u
from yml2 import ymlCStyle, comment, oldSyntax
import backend

def printInfo(option, opt_str, value, parser):
    sys.stdout.write(__doc__)

def w(msg):
    if isinstance(msg, BaseException):
        try:
            msg = str(msg) + "\n"
        except:
            msg = u(msg) + u"\n"
    if type(msg) is unicode:
        msg = codecs.encode(msg, sys.stderr.encoding)
    sys.stderr.write(msg)

optParser = OptionParser()
optParser.add_option("-C", "--old-syntax", action="store_true", dest="old_syntax",
        help="syntax of YML 2 version 1.x (compatibility mode)", default=False)
optParser.add_option("-D", "--emit-linenumbers", action="store_true", dest="emitlinenumbers",
        help="emit line numbers into the resulting XML for debugging purposes", default=False)
optParser.add_option("-E", "--encoding", dest="encoding", metavar="ENCODING", default=locale.getdefaultlocale()[1],
        help="encoding of input files (default to locale)")
optParser.add_option("-I", "--include", dest="includePathText", metavar="INCLUDE_PATH",
        help="precede YML_PATH by a colon separated INCLUDE_PATH to search for include files")
optParser.add_option("-m", "--omit-empty-parm-tags", action="store_true", dest="omitemptyparm",
        help="does nothing (only there for compatibility reasons)", default=False)
optParser.add_option("-n", "--normalization", dest="normalization", metavar="NORMALIZATION", default="NFC",
        help="Unicode normalization (none, NFD, NFKD, NFC, NFKC, FCD, default is NFC)")
optParser.add_option("-o", "--output", dest="outputFile", metavar="FILE",
        help="place output in file FILE")
optParser.add_option("-p", "--parse-only", action="store_true", dest="parseonly",
        help="parse only, then output pyAST as text to stdout", default=False)
optParser.add_option("-V", "--version", action="callback", callback=printInfo, help="show version info")
(options, args) = optParser.parse_args()

if options.old_syntax:
    oldSyntax()

if options.emitlinenumbers:
    backend.emitlinenumbers = True

backend.encoding = options.encoding

try:
    if options.includePathText:
        backend.includePath = options.includePathText.split(':')

    dirs = os.environ.get('YML_PATH', '.').split(':')
    backend.includePath.extend(dirs)

    files = fileinput.input(args, mode="rU", openhook=fileinput.hook_encoded(options.encoding))
    # fileinput suffer from two nasty bugs :
    # - ignoring open hook with stdin
    # - iterator requires ctrl-D to be pressed twice on some platform
    if args in [[],['-']] :
        files._files=[]
        files._buffer=[unicode(line, options.encoding) for line in sys.stdin.readlines()]

    ymlC = ymlCStyle()
    result = parse(ymlC, files, True, comment, packrat=True)

    if options.parseonly:
        print(result)
    else:
        result = backend.finish(result)
        if options.normalization != "none":
            result = unicodedata.normalize(options.normalization, result)

        if options.outputFile and options.outputFile != "-":
            outfile = open(options.outputFile, "w")
            outfile.write(codecs.encode(result, options.encoding))
            outfile.close()
        else:
            print(codecs.encode(result, options.encoding))

except KeyboardInterrupt:
    w("\n")
    sys.exit(1)
except KeyError as msg:
    w(u"not found: " + u(msg) + u"\n")
    sys.exit(4)
except LookupError as msg:
    w(u"not found: " + u(msg) + u"\n")
    sys.exit(4)
except Exception as msg:
    w(msg)
    sys.exit(5)