claudio@55: #!/usr/bin/env python claudio@55: # vim: set fileencoding=utf-8 : claudio@55: claudio@55: """\ claudio@55: YML/YSLT 2 processor version 5.8 claudio@55: Copyleft (c), 2009-2011 Volker Birk http://fdik.org/yml/ claudio@55: claudio@55: """ claudio@55: claudio@55: import sys, os, codecs, locale claudio@55: import fileinput, unicodedata claudio@55: from optparse import OptionParser claudio@55: claudio@55: try: claudio@55: from lxml import etree claudio@55: except: claudio@55: sys.stderr.write("This program needs lxml, see http://codespeak.net/lxml/\n") claudio@55: sys.exit(1) claudio@55: claudio@55: from yml2 import ymlCStyle, comment, oldSyntax claudio@55: from pyPEG import parse, u claudio@55: import backend claudio@55: claudio@55: def printInfo(option, opt_str, value, parser): claudio@55: sys.stdout.write(__doc__) claudio@55: sys.exit(0) claudio@55: claudio@55: class YMLAssert(Exception): pass claudio@55: claudio@55: def w(msg): claudio@55: if isinstance(msg, BaseException): claudio@55: try: claudio@55: msg = str(msg) + "\n" claudio@55: except: claudio@55: msg = u(msg) + u"\n" claudio@55: if type(msg) is unicode: claudio@55: msg = codecs.encode(msg, sys.stderr.encoding) claudio@55: sys.stderr.write(msg) claudio@55: claudio@55: claudio@55: def main(): claudio@55: optParser = OptionParser() claudio@55: optParser.add_option("-C", "--old-syntax", action="store_true", dest="old_syntax", claudio@55: help="syntax of YML 2 version 1.x (compatibility mode)", default=False) claudio@55: optParser.add_option("-D", "--emit-linenumbers", action="store_true", dest="emitlinenumbers", claudio@55: help="emit line numbers into the resulting XML for debugging purposes", default=False) claudio@55: optParser.add_option("--debug", action="store_true", dest="trace", claudio@55: help="switch on tracing to stderr", default=False) claudio@55: optParser.add_option("-d", "--paramdict", dest="params", metavar="PARAMS", claudio@55: help="call X/YSLT script with dictionary PARAMS as parameters") claudio@55: optParser.add_option("-e", "--xpath", dest="xpath", metavar="XPATH", claudio@55: help="execute XPath expression XPATH and print result") claudio@55: optParser.add_option("-E", "--encoding", dest="encoding", metavar="ENCODING", default=locale.getdefaultlocale()[1], claudio@55: help="encoding of input files (default to locale)") claudio@55: optParser.add_option("-I", "--include", dest="includePathText", metavar="INCLUDE_PATH", claudio@55: help="precede YML_PATH by a colon separated INCLUDE_PATH to search for include files") claudio@55: optParser.add_option("-m", "--omit-empty-parm-tags", action="store_true", dest="omitemptyparm", claudio@55: help="does nothing (only there for compatibility reasons)", default=False) claudio@55: optParser.add_option("-M", "--empty-input-document", action="store_true", dest="emptyinput", claudio@55: help="use an empty input document", default=False) claudio@55: optParser.add_option("-n", "--normalization", dest="normalization", metavar="NORMALIZATION", default="NFC", claudio@55: help="Unicode normalization (none, NFD, NFKD, NFC, NFKC, FCD, default is NFC)") claudio@55: optParser.add_option("-o", "--output", dest="outputFile", metavar="FILE", claudio@55: help="place output in file FILE") claudio@55: optParser.add_option("-p", "--parse-only", action="store_true", dest="parseonly", claudio@55: help="parse only, then output pyAST as text to stdout", default=False) claudio@55: optParser.add_option("-P", "--pretty", action="store_true", default=False, claudio@55: help="pretty print output adding whitespace") claudio@55: optParser.add_option("-s", "--stringparamdict", dest="stringparams", metavar="STRINGPARAMS", claudio@55: help="call X/YSLT script with dictionary STRINGPARAMS as string parameters") claudio@55: optParser.add_option("-x", "--xml", action="store_true", default=False, claudio@55: help="input document is XML already") claudio@55: optParser.add_option("-X", "--xslt", dest="xslt", metavar="XSLTSCRIPT", claudio@55: help="execute XSLT script XSLTSCRIPT") claudio@55: optParser.add_option("-y", "--yslt", dest="yslt", metavar="YSLTSCRIPT", claudio@55: help="execute YSLT script YSLTSCRIPT") claudio@55: optParser.add_option("-Y", "--xml2yml", action="store_true", default=False, claudio@55: help="convert XML to normalized YML code") claudio@55: optParser.add_option("-V", "--version", action="callback", callback=printInfo, help="show version info and exit") claudio@55: (options, args) = optParser.parse_args() claudio@55: claudio@55: if options.old_syntax: claudio@55: oldSyntax() claudio@55: claudio@55: if options.trace: claudio@55: backend.enable_tracing = True claudio@55: claudio@55: if options.emitlinenumbers: claudio@55: backend.emitlinenumbers = True claudio@55: claudio@55: if options.includePathText: claudio@55: backend.includePath = options.includePathText.split(':') claudio@55: claudio@55: backend.encoding = options.encoding claudio@55: claudio@55: dirs = os.environ.get('YML_PATH', '.').split(':') claudio@55: backend.includePath.extend(dirs) claudio@55: claudio@55: if options.xml2yml: claudio@55: for directory in backend.includePath: claudio@55: try: claudio@55: name = directory + "/xml2yml.ysl2" claudio@55: f = open(name, "r") claudio@55: f.close() claudio@55: break claudio@55: except: claudio@55: pass claudio@55: claudio@55: options.yslt = name claudio@55: options.xml = True claudio@55: claudio@55: if (options.xslt and options.yslt) or (options.xslt and options.xpath) or (options.yslt and options.xpath): claudio@55: sys.stderr.write("Cannot combine --xpath, --xslt and --yslt params\n") claudio@55: sys.exit(1) claudio@55: claudio@55: try: claudio@55: ymlC = ymlCStyle() claudio@55: claudio@55: rtext = u"" claudio@55: claudio@55: if not options.emptyinput: claudio@55: files = fileinput.input(args, mode="rU", openhook=fileinput.hook_encoded(options.encoding)) claudio@55: claudio@55: if options.xml: claudio@55: rtext = "" claudio@55: for line in files: claudio@55: rtext += line claudio@55: else: claudio@55: result = parse(ymlC, files, True, comment) claudio@55: if options.parseonly: claudio@55: print(result) claudio@55: sys.exit(0) claudio@55: else: claudio@55: rtext = backend.finish(result) claudio@55: claudio@55: if not rtext: claudio@55: rtext = u"" claudio@55: claudio@55: def ymldebug(context, text): claudio@55: if options.trace: claudio@55: sys.stderr.write("Debug: " + codecs.encode(u(text), options.encoding) + "\n") claudio@55: return "" claudio@55: claudio@55: def ymlassert(context, value, msg): claudio@55: if options.trace: claudio@55: if not value: claudio@55: raise YMLAssert(msg) claudio@55: return "" claudio@55: claudio@55: ymlns = etree.FunctionNamespace("http://fdik.org/yml") claudio@55: ymlns.prefix = "yml" claudio@55: ymlns['debug'] = ymldebug claudio@55: ymlns['assert'] = ymlassert claudio@55: claudio@55: if options.xpath: claudio@55: tree = etree.fromstring(rtext) claudio@55: ltree = tree.xpath(codecs.decode(options.xpath, options.encoding)) claudio@55: rtext = u"" claudio@55: try: claudio@55: for rtree in ltree: claudio@55: rtext += etree.tostring(rtree, pretty_print=options.pretty, encoding=unicode) claudio@55: except: claudio@55: rtext = ltree claudio@55: claudio@55: elif options.yslt or options.xslt: claudio@55: params = {} claudio@55: claudio@55: if options.yslt: claudio@55: backend.clearAll() claudio@55: yscript = fileinput.input(options.yslt, mode="rU", openhook=fileinput.hook_encoded(options.encoding)) claudio@55: yresult = parse(ymlC, yscript, True, comment) claudio@55: ytext = backend.finish(yresult) claudio@55: else: claudio@55: yscript = fileinput.input(options.xslt, mode="rU") claudio@55: ytext = "" claudio@55: for line in yscript: claudio@55: ytext += line claudio@55: claudio@55: doc = etree.fromstring(rtext) claudio@55: claudio@55: xsltree = etree.XML(ytext, base_url=os.path.abspath(yscript.filename())) claudio@55: transform = etree.XSLT(xsltree) claudio@55: claudio@55: if options.params: claudio@55: params = eval(options.params) claudio@55: for key, value in params.iteritems(): claudio@55: if type(value) != unicode: claudio@55: params[key] = u(value) claudio@55: if options.stringparams: claudio@55: for key, value in eval(options.stringparams).iteritems(): claudio@55: params[key] = u"'" + u(value) + u"'" claudio@55: claudio@55: rresult = transform(doc, **params) claudio@55: # lxml is somewhat buggy claudio@55: try: claudio@55: rtext = u(rresult) claudio@55: except: claudio@55: rtext = etree.tostring(rresult, encoding=unicode) claudio@55: if not rtext: claudio@55: rtext = codecs.decode(str(rresult), "utf-8") claudio@55: claudio@55: if options.normalization != "none": claudio@55: rtext = unicodedata.normalize(options.normalization, rtext) claudio@55: claudio@55: if options.pretty: claudio@55: plaintext = etree.tostring(etree.fromstring(rtext), pretty_print=True, xml_declaration=True, encoding=options.encoding) claudio@55: else: claudio@55: if isinstance(rtext, unicode): claudio@55: plaintext = codecs.encode(rtext, options.encoding) claudio@55: else: claudio@55: plaintext = str(rtext) claudio@55: claudio@55: try: claudio@55: if plaintext[-1] == "\n": claudio@55: plaintext = plaintext[:-1] claudio@55: except: pass claudio@55: claudio@55: if options.outputFile and options.outputFile != "-": claudio@55: outfile = open(options.outputFile, "w") claudio@55: outfile.write(plaintext) claudio@55: outfile.close() claudio@55: else: claudio@55: print(plaintext) claudio@55: claudio@55: except KeyboardInterrupt: claudio@55: w("\n") claudio@55: sys.exit(1) claudio@55: except YMLAssert as msg: claudio@55: w(u"YML Assertion failed: " + u(msg) + u"\n") claudio@55: sys.exit(2) claudio@55: except KeyError as msg: claudio@55: w(u"not found: " + u(msg) + u"\n") claudio@55: sys.exit(4) claudio@55: except LookupError as msg: claudio@55: w(u"not found: " + u(msg) + u"\n") claudio@55: sys.exit(4) claudio@55: except etree.XMLSyntaxError as e: claudio@55: log = e.error_log.filter_from_level(etree.ErrorLevels.FATAL) claudio@55: for entry in log: claudio@55: w(u"XML error: " + u(entry.message) + u"\n") claudio@55: sys.exit(5) claudio@55: except Exception as msg: claudio@55: w(msg) claudio@55: sys.exit(5) claudio@55: claudio@55: claudio@55: if __name__ == "__main__": claudio@55: sys.exit(main()) claudio@55: