1

#!/usr/bin/env python

     2

# vim: set fileencoding=utf-8 :

     3

     4

"""\

     5

YML/YSLT 2 processor version 5.5

     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

    40

optParser = OptionParser()

    41

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

    42

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

    43

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

    44

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

    45

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

    46

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

    47

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

    48

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

    49

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

    50

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

    51

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

    52

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

    53

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

    54

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

    55

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

    56

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

    57

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

    58

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

    59

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

    60

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

    61

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

    62

        help="place output in file FILE")

    63

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

    64

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

    65

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

    66

        help="pretty print output adding whitespace")

    67

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

    68

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

    69

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

    70

        help="input document is XML already")

    71

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

    72

        help="execute XSLT script XSLTSCRIPT")

    73

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

    74

        help="execute YSLT script YSLTSCRIPT")

    75

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

    76

        help="convert XML to normalized YML code")

    77

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

    78

(options, args) = optParser.parse_args()

    79

    80

if options.old_syntax:

    81

    oldSyntax()

    82

    83

if options.trace:

    84

    backend.enable_tracing = True

    85

    86

if options.emitlinenumbers:

    87

    backend.emitlinenumbers = True

    88

    89

if options.includePathText:

    90

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

    91

    92

backend.encoding = options.encoding

    93

    94

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

    95

backend.includePath.extend(dirs)

    96

    97

if options.xml2yml:

    98

    for directory in backend.includePath:

    99

        try:

   100

            name = directory + "/xml2yml.ysl2"

   101

            f = open(name, "r")

   102

            f.close()

   103

            break

   104

        except:

   105

            pass

   106

   107

    options.yslt = name

   108

    options.xml = True

   109

   110

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

   111

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

   112

    sys.exit(1)

   113

   114

try:

   115

    ymlC = ymlCStyle()

   116

   117

    rtext = u""

   118

   119

    if not options.emptyinput:

   120

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

   121

        # fileinput suffer from two nasty bugs :

   122

        # - ignoring open hook with stdin

   123

        # - iterator requires ctrl-D to be pressed twice on siome platform

   124

        if args in [[],['-']] :

   125

            files._files=[]

   126

            files._buffer=[unicode(line, options.encoding) for line in sys.stdin.readlines()]

   127

   128

        if options.xml:

   129

            rtext = ""

   130

            for line in files:

   131

                rtext += line

   132

        else:

   133

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

   134

            if options.parseonly:

   135

                print(result)

   136

                sys.exit(0)

   137

            else:

   138

                rtext = backend.finish(result)

   139

   140

    if not rtext:

   141

        rtext = u"<empty/>"

   142

   143

    def ymldebug(context, text):

   144

        if options.trace:

   145

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

   146

        return ""

   147

   148

    def ymlassert(context, value, msg):

   149

        if options.trace:

   150

            if not value:

   151

                raise YMLAssert(msg)

   152

        return ""

   153

   154

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

   155

    ymlns.prefix = "yml"

   156

    ymlns['debug'] = ymldebug

   157

    ymlns['assert'] = ymlassert

   158

   159

    if options.xpath:

   160

        tree = etree.fromstring(rtext)

   161

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

   162

        rtext = u""

   163

        try:

   164

            for rtree in ltree:

   165

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

   166

        except:

   167

            rtext = ltree

   168

   169

    elif options.yslt or options.xslt:

   170

        params = {}

   171

   172

        if options.yslt:

   173

            backend.clearAll()

   174

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

   175

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

   176

            ytext = backend.finish(yresult)

   177

        else:

   178

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

   179

            ytext = ""

   180

            for line in yscript:

   181

                ytext += line

   182

   183

        doc = etree.fromstring(rtext)

   184

   185

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

   186

        transform = etree.XSLT(xsltree)

   187

   188

        if options.params:

   189

            params = eval(options.params)

   190

            for key, value in params.iteritems():

   191

                if type(value) != unicode:

   192

                    params[key] = u(value)

   193

        if options.stringparams:

   194

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

   195

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

   196

   197

        rresult = transform(doc, **params)

   198

        # lxml is somewhat buggy

   199

        try:

   200

            rtext = u(rresult)

   201

        except:

   202

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

   203

            if not rtext:

   204

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

   205

   206

    if options.normalization != "none":

   207

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

   208

   209

    if options.pretty:

   210

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

   211

    else:

   212

        if isinstance(rtext, unicode):

   213

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

   214

        else:

   215

            plaintext = str(rtext)

   216

   217

    try:

   218

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

   219

            plaintext = plaintext[:-1]

   220

    except: pass

   221

   222

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

   223

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

   224

        outfile.write(plaintext)

   225

        outfile.close()

   226

    else:

   227

        print(plaintext)

   228

   229

except KeyboardInterrupt:

   230

    w("\n")

   231

    sys.exit(1)

   232

except YMLAssert as msg:

   233

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

   234

    sys.exit(2)

   235

except KeyError as msg:

   236

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

   237

    sys.exit(4)

   238

except LookupError as msg:

   239

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

   240

    sys.exit(4)

   241

except etree.XMLSyntaxError as e:

   242

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

   243

    for entry in log:

   244

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

   245

    sys.exit(5)

   246

except Exception as msg:

   247

    w(msg)

   248

    sys.exit(5)