vb@31: #!/usr/bin/env python3
vb@0: # vim: set fileencoding=utf-8 :
vb@0: 
vb@0: """\
vb@40: YML/YSLT 2 processor version 6.2
h@50: Copyleft (c), 2009-2020 Volker Birk  http://fdik.org/yml/
vb@0: 
vb@0: """
vb@0: 
vb@0: import sys, os, codecs, locale
vb@0: import fileinput, unicodedata
vb@0: from optparse import OptionParser
vb@0: 
vb@0: try:
vb@0:     from lxml import etree
vb@0: except:
vb@0:     sys.stderr.write("This program needs lxml, see http://codespeak.net/lxml/\n")
vb@0:     sys.exit(1)
vb@0: 
vb@0: from yml2 import ymlCStyle, comment, oldSyntax
h@41: from yml2.pyPEG import parse, u
h@41: import yml2.backend as backend
vb@0: 
h@43: YML_DEFAULT_PATH = [os.path.dirname(backend.__file__)]
h@43: 
vb@0: def printInfo(option, opt_str, value, parser):
vb@0:     sys.stdout.write(__doc__)
vb@0:     sys.exit(0)
vb@0: 
vb@0: class YMLAssert(Exception): pass
vb@0: 
vb@0: def w(msg):
vb@0:     if isinstance(msg, BaseException):
vb@0:         try:
vb@0:             msg = str(msg) + "\n"
vb@0:         except:
vb@32:             msg = u(msg) + "\n"
vb@0:     sys.stderr.write(msg)
vb@0: 
claudio@64: def main():
claudio@64:     optParser = OptionParser()
claudio@64:     optParser.add_option("-C", "--old-syntax", action="store_true", dest="old_syntax",
claudio@64:             help="syntax of YML 2 version 1.x (compatibility mode)", default=False)
claudio@64:     optParser.add_option("-D", "--emit-linenumbers", action="store_true", dest="emitlinenumbers",
claudio@64:             help="emit line numbers into the resulting XML for debugging purposes", default=False)
claudio@64:     optParser.add_option("--debug", action="store_true", dest="trace",
claudio@64:             help="switch on tracing to stderr", default=False)
claudio@64:     optParser.add_option("-d", "--paramdict", dest="params", metavar="PARAMS",
claudio@64:             help="call X/YSLT script with dictionary PARAMS as parameters")
claudio@64:     optParser.add_option("-e", "--xpath", dest="xpath", metavar="XPATH",
claudio@64:             help="execute XPath expression XPATH and print result")
claudio@64:     optParser.add_option("-E", "--encoding", dest="encoding", metavar="ENCODING", default=locale.getdefaultlocale()[1],
claudio@64:             help="encoding of input files (default to locale)")
claudio@64:     optParser.add_option("-I", "--include", dest="includePathText", metavar="INCLUDE_PATH",
claudio@64:             help="precede YML_PATH by a colon separated INCLUDE_PATH to search for include files")
claudio@64:     optParser.add_option("-m", "--omit-empty-parm-tags", action="store_true", dest="omitemptyparm",
claudio@64:             help="does nothing (only there for compatibility reasons)", default=False)
claudio@64:     optParser.add_option("-M", "--empty-input-document", action="store_true", dest="emptyinput",
claudio@64:             help="use an empty input document", default=False)
claudio@64:     optParser.add_option("-n", "--normalization", dest="normalization", metavar="NORMALIZATION", default="NFC",
claudio@64:             help="Unicode normalization (none, NFD, NFKD, NFC, NFKC, FCD, default is NFC)")
claudio@64:     optParser.add_option("-o", "--output", dest="outputFile", metavar="FILE",
claudio@64:             help="place output in file FILE")
claudio@64:     optParser.add_option("-p", "--parse-only", action="store_true", dest="parseonly",
claudio@64:             help="parse only, then output pyAST as text to stdout", default=False)
claudio@64:     optParser.add_option("-P", "--pretty", action="store_true", default=False,
claudio@64:             help="pretty print output adding whitespace")
claudio@64:     optParser.add_option("-s", "--stringparamdict", dest="stringparams", metavar="STRINGPARAMS",
claudio@64:             help="call X/YSLT script with dictionary STRINGPARAMS as string parameters")
claudio@64:     optParser.add_option("-x", "--xml", action="store_true", default=False,
claudio@64:             help="input document is XML already")
claudio@64:     optParser.add_option("-X", "--xslt", dest="xslt", metavar="XSLTSCRIPT",
claudio@64:             help="execute XSLT script XSLTSCRIPT")
claudio@64:     optParser.add_option("-y", "--yslt", dest="yslt", metavar="YSLTSCRIPT",
claudio@64:             help="execute YSLT script YSLTSCRIPT")
claudio@64:     optParser.add_option("-Y", "--xml2yml", action="store_true", default=False,
claudio@64:             help="convert XML to normalized YML code")
claudio@64:     optParser.add_option("-V", "--version", action="callback", callback=printInfo, help="show version info and exit")
claudio@64:     (options, args) = optParser.parse_args()
claudio@64: 
claudio@64:     if options.old_syntax:
claudio@64:         oldSyntax()
claudio@64: 
claudio@64:     if options.trace:
claudio@64:         backend.enable_tracing = True
claudio@64: 
claudio@64:     if options.emitlinenumbers:
claudio@64:         backend.emitlinenumbers = True
claudio@64: 
claudio@64:     if options.includePathText:
claudio@64:         backend.includePath = options.includePathText.split(':')
claudio@64: 
claudio@64:     backend.encoding = options.encoding
claudio@64: 
claudio@64:     dirs = os.environ.get('YML_PATH', '.').split(':') + YML_DEFAULT_PATH
claudio@64:     backend.includePath.extend(dirs)
claudio@64: 
claudio@64:     if options.xml2yml:
claudio@64:         for directory in backend.includePath:
claudio@64:             name = os.path.join(directory, "xml2yml.ysl2")
claudio@64:             if os.path.isfile(name):
claudio@64:                 options.yslt = name
claudio@64:                 options.xml = True
claudio@64:                 break
claudio@64:         else:
claudio@64:             sys.stderr.write("Error: Stylesheet xml2yml.ysl2 required for --xml2yml not found\n")
claudio@64:             sys.stderr.write("Please check your YML_PATH\n")
claudio@64:             sys.exit(1)
claudio@64: 
claudio@64:     if  (options.xslt and options.yslt) or (options.xslt and options.xpath) or (options.yslt and options.xpath):
claudio@64:         sys.stderr.write("Cannot combine --xpath, --xslt and --yslt params\n")
h@42:         sys.exit(1)
vb@0: 
claudio@64:     try:
claudio@64:         ymlC = ymlCStyle()
claudio@64: 
claudio@64:         rtext = ""
claudio@64: 
claudio@64:         if not options.emptyinput:
claudio@64:             files = fileinput.input(args, mode="rU", openhook=fileinput.hook_encoded(options.encoding))
claudio@64: 
claudio@64:             if options.xml:
claudio@64:                 rtext = ""
claudio@64:                 for line in files:
claudio@64:                     rtext += line
claudio@64:             else:
claudio@64:                 result = parse(ymlC, files, True, comment)
claudio@64:                 if options.parseonly:
claudio@64:                     print(result)
claudio@64:                     sys.exit(0)
claudio@64:                 else:
claudio@64:                     rtext = backend.finish(result)
claudio@64: 
claudio@64:         if not rtext:
claudio@64:             rtext = "<empty/>"
claudio@64: 
claudio@64:         def ymldebug(context, text):
claudio@64:             if options.trace:
claudio@64:                 sys.stderr.write("Debug: " + codecs.encode(u(text), options.encoding) + "\n")
claudio@64:             return ""
claudio@64: 
claudio@64:         def ymlassert(context, value, msg):
claudio@64:             if options.trace:
claudio@64:                 if not value:
claudio@64:                     raise YMLAssert(msg)
claudio@64:             return ""
claudio@64: 
claudio@64:         ymlns = etree.FunctionNamespace("http://fdik.org/yml")
claudio@64:         ymlns.prefix = "yml"
claudio@64:         ymlns['debug'] = ymldebug
claudio@64:         ymlns['assert'] = ymlassert
claudio@64: 
claudio@64:         if options.xpath:
claudio@64:             tree = etree.fromstring(rtext)
claudio@64:             ltree = tree.xpath(codecs.decode(options.xpath, options.encoding))
vb@0:             rtext = ""
claudio@64:             try:
claudio@64:                 for rtree in ltree:
claudio@64:                     rtext += etree.tostring(rtree, pretty_print=options.pretty, encoding=unicode)
claudio@64:             except:
claudio@64:                 rtext = ltree
claudio@64: 
claudio@64:         elif options.yslt or options.xslt:
claudio@64:             params = {}
claudio@64: 
claudio@64:             if options.yslt:
claudio@64:                 backend.clearAll()
claudio@64:                 yscript = fileinput.input(options.yslt, mode="rU", openhook=fileinput.hook_encoded(options.encoding))
claudio@64:                 yresult = parse(ymlC, yscript, True, comment)
claudio@64:                 ytext = backend.finish(yresult)
claudio@64:             else:
claudio@64:                 yscript = fileinput.input(options.xslt, mode="rU")
claudio@64:                 ytext = ""
claudio@64:                 for line in yscript:
claudio@64:                     ytext += line
claudio@64: 
claudio@64:             doc = etree.fromstring(rtext)
claudio@64: 
claudio@64:             xsltree = etree.XML(ytext, base_url=os.path.abspath(yscript.filename()))
claudio@64:             transform = etree.XSLT(xsltree)
claudio@64:             
claudio@64:             if options.params:
claudio@64:                 params = eval(options.params)
claudio@64:                 for key, value in params.iteritems():
claudio@64:                     if type(value) is not str:
claudio@64:                         params[key] = u(value)
claudio@64:             if options.stringparams:
claudio@64:                 for key, value in eval(options.stringparams).iteritems():
claudio@64:                     params[key] = "'" + u(value) + "'"
claudio@64: 
claudio@64:             rresult = transform(doc, **params)
claudio@64:             # lxml is somewhat buggy
claudio@64:             try:
claudio@64:                 rtext = u(rresult)
claudio@64:             except:
claudio@64:                 rtext = etree.tostring(rresult, encoding=unicode)
claudio@64:                 if not rtext:
claudio@64:                     rtext = codecs.decode(str(rresult), "utf-8")
claudio@64: 
claudio@64:         if options.normalization != "none":
claudio@64:             rtext = unicodedata.normalize(options.normalization, rtext)
claudio@64: 
claudio@64:         if options.pretty:
claudio@64:             plaintext = etree.tostring(etree.fromstring(rtext), pretty_print=True, xml_declaration=True, encoding=options.encoding)
vb@0:         else:
claudio@64:             if isinstance(rtext, str):
claudio@64:                 plaintext = codecs.encode(rtext, options.encoding)
vb@0:             else:
claudio@64:                 plaintext = rtext
claudio@64: 
vb@0:         try:
claudio@64:             if plaintext[-1] == "\n":
claudio@64:                 plaintext = plaintext[:-1]
claudio@64:         except: pass
claudio@64: 
claudio@64:         if options.outputFile and options.outputFile != "-":
claudio@64:             outfile = open(options.outputFile, "wb")
claudio@64:             outfile.write(plaintext)
claudio@64:             outfile.close()
vb@0:         else:
claudio@64:             sys.stdout.buffer.write(plaintext)
claudio@64:             if not options.pretty:
claudio@64:                 print()
claudio@64: 
claudio@64:     except KeyboardInterrupt:
claudio@64:         w("\n")
claudio@64:         sys.exit(1)
claudio@64:     except YMLAssert as msg:
claudio@64:         w("YML Assertion failed: " + u(msg) + "\n")
claudio@64:         sys.exit(2)
claudio@64:     except KeyError as msg:
claudio@64:         w("not found: " + u(msg) + "\n")
claudio@64:         sys.exit(4)
claudio@64:     except LookupError as msg:
claudio@64:         w("not found: " + u(msg) + "\n")
claudio@64:         sys.exit(4)
claudio@64:     except etree.XMLSyntaxError as e:
claudio@64:         log = e.error_log.filter_from_level(etree.ErrorLevels.FATAL)
claudio@64:         for entry in log:
claudio@64:             w("XML error: " + u(entry.message) + "\n")
claudio@64:         sys.exit(5)
claudio@64:     except Exception as msg:
claudio@64:         w(msg)
claudio@64:         sys.exit(5)
claudio@64: 
claudio@64: 
claudio@64: if __name__ == "__main__":
claudio@64:     sys.exit(main())
claudio@64: