1

#!/usr/bin/env python

     2

# vim: set fileencoding=utf-8 :

     3

     4

"""\

     5

YML/YSLT 2 processor version 5.8

     6

Copyleft (c), 2009-2011 Volker Birk  http://fdik.org/yml/

     7

     8

"""

     9

    10

import sys, os, codecs, locale

    11

import fileinput, unicodedata

    12

from optparse import OptionParser

    13

    14

try:

    15

    from lxml import etree

    16

except:

    17

    sys.stderr.write("This program needs lxml, see http://codespeak.net/lxml/\n")

    18

    sys.exit(1)

    19

    20

from yml2 import ymlCStyle, comment, oldSyntax

    21

from pyPEG import parse, u

    22

import backend

    23

    24

def printInfo(option, opt_str, value, parser):

    25

    sys.stdout.write(__doc__)

    26

    sys.exit(0)

    27

    28

class YMLAssert(Exception): pass

    29

    30

def w(msg):

    31

    if isinstance(msg, BaseException):

    32

        try:

    33

            msg = str(msg) + "\n"

    34

        except:

    35

            msg = u(msg) + u"\n"

    36

    if type(msg) is unicode:

    37

        msg = codecs.encode(msg, sys.stderr.encoding)

    38

    sys.stderr.write(msg)

    39

    44

optParser = OptionParser()

    45

optParser.add_option("-C", "--old-syntax", action="store_true", dest="old_syntax",

    46

        help="syntax of YML 2 version 1.x (compatibility mode)", default=False)

    47

optParser.add_option("-D", "--emit-linenumbers", action="store_true", dest="emitlinenumbers",

    48

        help="emit line numbers into the resulting XML for debugging purposes", default=False)

    49

optParser.add_option("--debug", action="store_true", dest="trace",

    50

        help="switch on tracing to stderr", default=False)

    51

optParser.add_option("-d", "--paramdict", dest="params", metavar="PARAMS",

    52

        help="call X/YSLT script with dictionary PARAMS as parameters")

    53

optParser.add_option("-e", "--xpath", dest="xpath", metavar="XPATH",

    54

        help="execute XPath expression XPATH and print result")

    55

optParser.add_option("-E", "--encoding", dest="encoding", metavar="ENCODING", default=locale.getdefaultlocale()[1],

    56

        help="encoding of input files (default to locale)")

    57

optParser.add_option("-I", "--include", dest="includePathText", metavar="INCLUDE_PATH",

    58

        help="precede YML_PATH by a colon separated INCLUDE_PATH to search for include files")

    59

optParser.add_option("-m", "--omit-empty-parm-tags", action="store_true", dest="omitemptyparm",

    60

        help="does nothing (only there for compatibility reasons)", default=False)

    61

optParser.add_option("-M", "--empty-input-document", action="store_true", dest="emptyinput",

    62

        help="use an empty input document", default=False)

    63

optParser.add_option("-n", "--normalization", dest="normalization", metavar="NORMALIZATION", default="NFC",

    64

        help="Unicode normalization (none, NFD, NFKD, NFC, NFKC, FCD, default is NFC)")

    65

optParser.add_option("-o", "--output", dest="outputFile", metavar="FILE",

    66

        help="place output in file FILE")

    67

optParser.add_option("-p", "--parse-only", action="store_true", dest="parseonly",

    68

        help="parse only, then output pyAST as text to stdout", default=False)

    69

optParser.add_option("-P", "--pretty", action="store_true", default=False,

    70

        help="pretty print output adding whitespace")

    71

optParser.add_option("-s", "--stringparamdict", dest="stringparams", metavar="STRINGPARAMS",

    72

        help="call X/YSLT script with dictionary STRINGPARAMS as string parameters")

    73

optParser.add_option("-x", "--xml", action="store_true", default=False,

    74

        help="input document is XML already")

    75

optParser.add_option("-X", "--xslt", dest="xslt", metavar="XSLTSCRIPT",

    76

        help="execute XSLT script XSLTSCRIPT")

    77

optParser.add_option("-y", "--yslt", dest="yslt", metavar="YSLTSCRIPT",

    78

        help="execute YSLT script YSLTSCRIPT")

    79

optParser.add_option("-Y", "--xml2yml", action="store_true", default=False,

    80

        help="convert XML to normalized YML code")

    81

optParser.add_option("-V", "--version", action="callback", callback=printInfo, help="show version info and exit")

    82

(options, args) = optParser.parse_args()

    83

    84

if options.old_syntax:

    85

    oldSyntax()

    86

    87

if options.trace:

    88

    backend.enable_tracing = True

    89

    90

if options.emitlinenumbers:

    91

    backend.emitlinenumbers = True

    92

    93

if options.includePathText:

    94

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

    95

    96

backend.encoding = options.encoding

    97

    98

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

    99

backend.includePath.extend(dirs)

   100

   101

if options.xml2yml:

   102

    for directory in backend.includePath:

   103

        try:

   104

            name = directory + "/xml2yml.ysl2"

   105

            f = open(name, "r")

   106

            f.close()

   107

            break

   108

        except:

   109

            pass

   110

   111

    options.yslt = name

   112

    options.xml = True

   113

   114

if  (options.xslt and options.yslt) or (options.xslt and options.xpath) or (options.yslt and options.xpath):

   115

    sys.stderr.write("Cannot combine --xpath, --xslt and --yslt params\n")

   116

    sys.exit(1)

   117

   118

try:

   119

    ymlC = ymlCStyle()

   120

   121

    rtext = u""

   122

   123

    if not options.emptyinput:

   124

        files = fileinput.input(args, mode="rU", openhook=fileinput.hook_encoded(options.encoding))

   125

   126

        if options.xml:

   127

            rtext = ""

   128

            for line in files:

   129

                rtext += line

   130

        else:

   131

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

   132

            if options.parseonly:

   133

                print(result)

   134

                sys.exit(0)

   135

            else:

   136

                rtext = backend.finish(result)

   137

   138

    if not rtext:

   139

        rtext = u"<empty/>"

   140

   141

    def ymldebug(context, text):

   142

        if options.trace:

   143

            sys.stderr.write("Debug: " + codecs.encode(u(text), options.encoding) + "\n")

   144

        return ""

   145

   146

    def ymlassert(context, value, msg):

   147

        if options.trace:

   148

            if not value:

   149

                raise YMLAssert(msg)

   150

        return ""

   151

   152

    ymlns = etree.FunctionNamespace("http://fdik.org/yml")

   153

    ymlns.prefix = "yml"

   154

    ymlns['debug'] = ymldebug

   155

    ymlns['assert'] = ymlassert

   156

   157

    if options.xpath:

   158

        tree = etree.fromstring(rtext)

   159

        ltree = tree.xpath(codecs.decode(options.xpath, options.encoding))

   160

        rtext = u""

   161

        try:

   162

            for rtree in ltree:

   163

                rtext += etree.tostring(rtree, pretty_print=options.pretty, encoding=unicode)

   164

        except:

   165

            rtext = ltree

   166

   167

    elif options.yslt or options.xslt:

   168

        params = {}

   169

   170

        if options.yslt:

   171

            backend.clearAll()

   172

            yscript = fileinput.input(options.yslt, mode="rU", openhook=fileinput.hook_encoded(options.encoding))

   173

            yresult = parse(ymlC, yscript, True, comment)

   174

            ytext = backend.finish(yresult)

   175

        else:

   176

            yscript = fileinput.input(options.xslt, mode="rU")

   177

            ytext = ""

   178

            for line in yscript:

   179

                ytext += line

   180

   181

        doc = etree.fromstring(rtext)

   182

   183

        xsltree = etree.XML(ytext, base_url=os.path.abspath(yscript.filename()))

   184

        transform = etree.XSLT(xsltree)

   185

   186

        if options.params:

   187

            params = eval(options.params)

   188

            for key, value in params.iteritems():

   189

                if type(value) != unicode:

   190

                    params[key] = u(value)

   191

        if options.stringparams:

   192

            for key, value in eval(options.stringparams).iteritems():

   193

                params[key] = u"'" + u(value) + u"'"

   194

   195

        rresult = transform(doc, **params)

   196

        # lxml is somewhat buggy

   197

        try:

   198

            rtext = u(rresult)

   199

        except:

   200

            rtext = etree.tostring(rresult, encoding=unicode)

   201

            if not rtext:

   202

                rtext = codecs.decode(str(rresult), "utf-8")

   203

   204

    if options.normalization != "none":

   205

        rtext = unicodedata.normalize(options.normalization, rtext)

   206

   207

    if options.pretty:

   208

        plaintext = etree.tostring(etree.fromstring(rtext), pretty_print=True, xml_declaration=True, encoding=options.encoding)

   209

    else:

   210

        if isinstance(rtext, unicode):

   211

            plaintext = codecs.encode(rtext, options.encoding)

   212

        else:

   213

            plaintext = str(rtext)

   214

   215

    try:

   216

        if plaintext[-1] == "\n":

   217

            plaintext = plaintext[:-1]

   218

    except: pass

   219

   220

    if options.outputFile and options.outputFile != "-":

   221

        outfile = open(options.outputFile, "w")

   222

        outfile.write(plaintext)

   223

        outfile.close()

   224

    else:

   225

        print(plaintext)

   226

   227

except KeyboardInterrupt:

   228

    w("\n")

   229

    sys.exit(1)

   230

except YMLAssert as msg:

   231

    w(u"YML Assertion failed: " + u(msg) + u"\n")

   232

    sys.exit(2)

   233

except KeyError as msg:

   234

    w(u"not found: " + u(msg) + u"\n")

   235

    sys.exit(4)

   236

except LookupError as msg:

   237

    w(u"not found: " + u(msg) + u"\n")

   238

    sys.exit(4)

   239

except etree.XMLSyntaxError as e:

   240

    log = e.error_log.filter_from_level(etree.ErrorLevels.FATAL)

   241

    for entry in log:

   242

        w(u"XML error: " + u(entry.message) + u"\n")

   243

    sys.exit(5)

   244

except Exception as msg:

   245

    w(msg)

   246

    sys.exit(5)