1

#!/usr/bin/env python3

     2

# vim: set fileencoding=utf-8 :

     3

     4

"""\

     5

YML/YSLT 2 processor version 6.2

     6

Copyleft (c), 2009-2020 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

    21

from yml2.pyPEG import parse, u

    22

import yml2.backend as backend

    23

    24

YML_DEFAULT_PATH = [os.path.dirname(backend.__file__)]

    25

    26

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

    27

    sys.stdout.write(__doc__)

    28

    sys.exit(0)

    29

    30

class YMLAssert(Exception): pass

    31

    32

def w(msg):

    33

    if isinstance(msg, BaseException):

    34

        try:

    35

            msg = str(msg) + "\n"

    36

        except:

    37

            msg = u(msg) + "\n"

    38

    sys.stderr.write(msg)

    39

    40

def main():

    41

    optParser = OptionParser()

    42

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

    43

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

    44

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

    45

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

    46

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

    47

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

    48

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

    49

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

    50

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

    51

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

    52

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

    53

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

    54

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

    55

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

    56

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

    57

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

    58

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

    59

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

    60

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

    61

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

    62

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

    63

            help="place output in file FILE")

    64

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

    65

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

    66

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

    67

            help="pretty print output adding whitespace")

    68

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

    69

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

    70

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

    71

            help="input document is XML already")

    72

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

    73

            help="execute XSLT script XSLTSCRIPT")

    74

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

    75

            help="execute YSLT script YSLTSCRIPT")

    76

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

    77

            help="convert XML to normalized YML code")

    78

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

    79

    (options, args) = optParser.parse_args()

    80

    81

    if options.old_syntax:

    82

        oldSyntax()

    83

    84

    if options.trace:

    85

        backend.enable_tracing = True

    86

    87

    if options.emitlinenumbers:

    88

        backend.emitlinenumbers = True

    89

    90

    if options.includePathText:

    91

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

    92

    93

    backend.encoding = options.encoding

    94

    95

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

    96

    backend.includePath.extend(dirs)

    97

    98

    if options.xml2yml:

    99

        for directory in backend.includePath:

   100

            name = os.path.join(directory, "xml2yml.ysl2")

   101

            if os.path.isfile(name):

   102

                options.yslt = name

   103

                options.xml = True

   104

                break

   105

        else:

   106

            sys.stderr.write("Error: Stylesheet xml2yml.ysl2 required for --xml2yml not found\n")

   107

            sys.stderr.write("Please check your YML_PATH\n")

   108

            sys.exit(1)

   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 = ""

   118

   119

        if not options.emptyinput:

   120

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

   121

   122

            if options.xml:

   123

                rtext = ""

   124

                for line in files:

   125

                    rtext += line

   126

            else:

   127

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

   128

                if options.parseonly:

   129

                    print(result)

   130

                    sys.exit(0)

   131

                else:

   132

                    rtext = backend.finish(result)

   133

   134

        if not rtext:

   135

            rtext = "<empty/>"

   136

   137

        def ymldebug(context, text):

   138

            if options.trace:

   139

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

   140

            return ""

   141

   142

        def ymlassert(context, value, msg):

   143

            if options.trace:

   144

                if not value:

   145

                    raise YMLAssert(msg)

   146

            return ""

   147

   148

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

   149

        ymlns.prefix = "yml"

   150

        ymlns['debug'] = ymldebug

   151

        ymlns['assert'] = ymlassert

   152

   153

        if options.xpath:

   154

            tree = etree.fromstring(rtext)

   155

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

   156

            rtext = ""

   157

            try:

   158

                for rtree in ltree:

   159

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

   160

            except:

   161

                rtext = ltree

   162

   163

        elif options.yslt or options.xslt:

   164

            params = {}

   165

   166

            if options.yslt:

   167

                backend.clearAll()

   168

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

   169

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

   170

                ytext = backend.finish(yresult)

   171

            else:

   172

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

   173

                ytext = ""

   174

                for line in yscript:

   175

                    ytext += line

   176

   177

            doc = etree.fromstring(rtext)

   178

   179

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

   180

            transform = etree.XSLT(xsltree)

   181

   182

            if options.params:

   183

                params = eval(options.params)

   184

                for key, value in params.items():

   185

                    if type(value) is not str:

   186

                        params[key] = u(value)

   187

            if options.stringparams:

   188

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

   189

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

   190

   191

            rresult = transform(doc, **params)

   192

            # lxml is somewhat buggy

   193

            try:

   194

                rtext = u(rresult)

   195

            except:

   196

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

   197

                if not rtext:

   198

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

   199

   200

        if options.normalization != "none":

   201

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

   202

   203

        if options.pretty:

   204

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

   205

        else:

   206

            if isinstance(rtext, str):

   207

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

   208

            else:

   209

                plaintext = rtext

   210

   211

        try:

   212

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

   213

                plaintext = plaintext[:-1]

   214

        except: pass

   215

   216

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

   217

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

   218

            outfile.write(plaintext)

   219

            outfile.close()

   220

        else:

   221

            sys.stdout.buffer.write(plaintext)

   222

            if not options.pretty:

   223

                print()

   224

   225

    except KeyboardInterrupt:

   226

        w("\n")

   227

        sys.exit(1)

   228

    except YMLAssert as msg:

   229

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

   230

        sys.exit(2)

   231

    except KeyError as msg:

   232

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

   233

        sys.exit(4)

   234

    except LookupError as msg:

   235

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

   236

        sys.exit(4)

   237

    except etree.XMLSyntaxError as e:

   238

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

   239

        for entry in log:

   240

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

   241

        sys.exit(5)

   242

    except Exception as msg:

   243

        w(msg)

   244

        sys.exit(5)

   245

   246

   247

if __name__ == "__main__":

   248

    main()

   249