| 1 | #!/usr/bin/python -u
|
|---|
| 2 | #
|
|---|
| 3 | # The exercise of rewriting xsltproc on top of the python
|
|---|
| 4 | # bindings, not complete yet and shows up the things missing
|
|---|
| 5 | # from the existing python interfaces
|
|---|
| 6 | #
|
|---|
| 7 | import sys
|
|---|
| 8 | import time
|
|---|
| 9 | import posix
|
|---|
| 10 | import string
|
|---|
| 11 | import libxml2
|
|---|
| 12 | # Memory debug specific
|
|---|
| 13 | libxml2.debugMemory(1)
|
|---|
| 14 | import libxslt
|
|---|
| 15 |
|
|---|
| 16 | debug = 0
|
|---|
| 17 | repeat = 0
|
|---|
| 18 | timing = 0
|
|---|
| 19 | novalid = 0
|
|---|
| 20 | noout = 0
|
|---|
| 21 | docbook = 0
|
|---|
| 22 | html = 0
|
|---|
| 23 | xinclude = 0
|
|---|
| 24 | profile = 0
|
|---|
| 25 | params = {}
|
|---|
| 26 | output = None
|
|---|
| 27 | errorno = 0
|
|---|
| 28 |
|
|---|
| 29 | #
|
|---|
| 30 | # timing
|
|---|
| 31 | #
|
|---|
| 32 | begin = 0
|
|---|
| 33 | endtime = 0
|
|---|
| 34 | def startTimer():
|
|---|
| 35 | global begin
|
|---|
| 36 |
|
|---|
| 37 | begin = time.time()
|
|---|
| 38 |
|
|---|
| 39 | def endTimer(msg):
|
|---|
| 40 | global begin
|
|---|
| 41 | global endtime
|
|---|
| 42 |
|
|---|
| 43 | endtime = time.time()
|
|---|
| 44 | print "%s took %d ms" % (msg, (endtime - begin) * 1000)
|
|---|
| 45 |
|
|---|
| 46 | def xsltProcess(doc, cur, filename):
|
|---|
| 47 | global timing
|
|---|
| 48 | global xinclude
|
|---|
| 49 | global params
|
|---|
| 50 | global html
|
|---|
| 51 |
|
|---|
| 52 | if xinclude:
|
|---|
| 53 | if timing:
|
|---|
| 54 | startTimer()
|
|---|
| 55 | doc.XIncludeProcess()
|
|---|
| 56 | if timing:
|
|---|
| 57 | endTimer("XInclude processing %s" % (filename))
|
|---|
| 58 |
|
|---|
| 59 | if timing:
|
|---|
| 60 | startTimer()
|
|---|
| 61 | if output == None:
|
|---|
| 62 | if repeat != 0:
|
|---|
| 63 | for j in range(1, repeat):
|
|---|
| 64 | res = cur.applyStylesheet(doc, params)
|
|---|
| 65 | res.freeDoc()
|
|---|
| 66 | doc.freeDoc()
|
|---|
| 67 | if html == 1:
|
|---|
| 68 | doc = libxml2.htmlParseFile(filename, None)
|
|---|
| 69 | else:
|
|---|
| 70 | doc = libxml2.parseFile(filename, None)
|
|---|
| 71 | # ctxt = libxslt.newTransformContext(doc)
|
|---|
| 72 | # if ctxt == None:
|
|---|
| 73 | # return
|
|---|
| 74 | if profile:
|
|---|
| 75 | print "TODO: Profiling not yet supported"
|
|---|
| 76 | else:
|
|---|
| 77 | res = cur.applyStylesheet(doc, params)
|
|---|
| 78 | if timing:
|
|---|
| 79 | if repeat != 0:
|
|---|
| 80 | endTimer("Applying stylesheet %d times" % (repeat))
|
|---|
| 81 | else:
|
|---|
| 82 | endTimer("Applying stylesheet")
|
|---|
| 83 | doc.freeDoc()
|
|---|
| 84 | if res == None:
|
|---|
| 85 | print "no result for %s" % (filename)
|
|---|
| 86 | return
|
|---|
| 87 | if noout != 0:
|
|---|
| 88 | res.freeDoc()
|
|---|
| 89 | return
|
|---|
| 90 | if debug == 1:
|
|---|
| 91 | res.debugDumpDocument(None)
|
|---|
| 92 | else:
|
|---|
| 93 | if timing:
|
|---|
| 94 | startTimer()
|
|---|
| 95 | cur.saveResultToFilename("-", res, 0)
|
|---|
| 96 | if timing:
|
|---|
| 97 | endTimer("Saving result")
|
|---|
| 98 | res.freeDoc()
|
|---|
| 99 | else:
|
|---|
| 100 | print "TODO: xsltRunStylesheet not yet mapped"
|
|---|
| 101 |
|
|---|
| 102 | def usage(name = 'pyxsltproc'):
|
|---|
| 103 | print "Usage: %s [options] stylesheet file [file ...]" % (name)
|
|---|
| 104 | print "a reimplementation of xsltproc(1) on top of libxslt-python"
|
|---|
| 105 | print " Options:"
|
|---|
| 106 | print "\t--version or -V: show the version of libxml and libxslt used"
|
|---|
| 107 | print "\t--verbose or -v: show logs of what's happening"
|
|---|
| 108 | print "\t--output file or -o file: save to a given file"
|
|---|
| 109 | print "\t--timing: display the time used"
|
|---|
| 110 | print "\t--repeat: run the transformation 20 times"
|
|---|
| 111 | print "\t--debug: dump the tree of the result instead"
|
|---|
| 112 | print "\t--novalid skip the Dtd loading phase"
|
|---|
| 113 | print "\t--noout: do not dump the result"
|
|---|
| 114 | print "\t--maxdepth val : increase the maximum depth"
|
|---|
| 115 | print "\t--html: the input document is(are) an HTML file(s)"
|
|---|
| 116 | print "\t--param name value : pass a (parameter,value) pair"
|
|---|
| 117 | print "\t value is an XPath expression."
|
|---|
| 118 | print "\t string values must be quoted like \"'string'\""
|
|---|
| 119 | print "\t or use stringparam to avoid it"
|
|---|
| 120 | print "\t--stringparam name value : pass a (parameter,string value) pair"
|
|---|
| 121 | print "\t--nonet refuse to fetch DTDs or entities over network"
|
|---|
| 122 | print "\t--catalogs : use SGML catalogs from $SGML_CATALOG_FILES"
|
|---|
| 123 | print "\t otherwise XML Catalogs starting from "
|
|---|
| 124 | print "\t file:///etc/xml/catalog are activated by default"
|
|---|
| 125 | print "\t--xinclude : do XInclude processing on document input"
|
|---|
| 126 | print "\t--profile or --norman : dump profiling informations "
|
|---|
| 127 | print "\nProject libxslt home page: http://xmlsoft.org/XSLT/"
|
|---|
| 128 | print "To report bugs and get help: http://xmlsoft.org/XSLT/bugs.html"
|
|---|
| 129 |
|
|---|
| 130 | def main(args = None):
|
|---|
| 131 | global debug
|
|---|
| 132 | global repeat
|
|---|
| 133 | global timing
|
|---|
| 134 | global novalid
|
|---|
| 135 | global noout
|
|---|
| 136 | global docbook
|
|---|
| 137 | global html
|
|---|
| 138 | global xinclude
|
|---|
| 139 | global profile
|
|---|
| 140 | global params
|
|---|
| 141 | global output
|
|---|
| 142 | global errorno
|
|---|
| 143 |
|
|---|
| 144 | done = 0
|
|---|
| 145 | cur = None
|
|---|
| 146 |
|
|---|
| 147 | if not args:
|
|---|
| 148 | args = sys.argv[1:]
|
|---|
| 149 | if len(args) <= 0:
|
|---|
| 150 | usage(sys.argv[0])
|
|---|
| 151 |
|
|---|
| 152 |
|
|---|
| 153 | i = 0
|
|---|
| 154 | while i < len(args):
|
|---|
| 155 | if args[i] == "-":
|
|---|
| 156 | break
|
|---|
| 157 | if args[i][0] != '-':
|
|---|
| 158 | i = i + 1
|
|---|
| 159 | continue
|
|---|
| 160 | if args[i] == "-timing" or args[i] == "--timing":
|
|---|
| 161 | timing = 1
|
|---|
| 162 | elif args[i] == "-debug" or args[i] == "--debug":
|
|---|
| 163 | debug = 1
|
|---|
| 164 | elif args[i] == "-verbose" or args[i] == "--verbose" or \
|
|---|
| 165 | args[i] == "-v":
|
|---|
| 166 | print "TODO: xsltSetGenericDebugFunc() mapping missing"
|
|---|
| 167 | elif args[i] == "-version" or args[i] == "--version" or \
|
|---|
| 168 | args[i] == "-V":
|
|---|
| 169 | print "TODO: version informations mapping missing"
|
|---|
| 170 | elif args[i] == "-verbose" or args[i] == "--verbose" or \
|
|---|
| 171 | args[i] == "-v":
|
|---|
| 172 | if repeat == 0:
|
|---|
| 173 | repeat = 20
|
|---|
| 174 | else:
|
|---|
| 175 | repeat = 100
|
|---|
| 176 | elif args[i] == "-novalid" or args[i] == "--novalid":
|
|---|
| 177 | print "TODO: xmlLoadExtDtdDefaultValue mapping missing"
|
|---|
| 178 | novalid = 1
|
|---|
| 179 | elif args[i] == "-noout" or args[i] == "--noout":
|
|---|
| 180 | noout = 1
|
|---|
| 181 | elif args[i] == "-html" or args[i] == "--html":
|
|---|
| 182 | html = 1
|
|---|
| 183 | elif args[i] == "-nonet" or args[i] == "--nonet":
|
|---|
| 184 | print "TODO: xmlSetExternalEntityLoader mapping missing"
|
|---|
| 185 | nonet = 1
|
|---|
| 186 | elif args[i] == "-catalogs" or args[i] == "--catalogs":
|
|---|
| 187 | try:
|
|---|
| 188 | catalogs = posix.environ['SGML_CATALOG_FILES']
|
|---|
| 189 | except:
|
|---|
| 190 | catalogs = None
|
|---|
| 191 | if catalogs != none:
|
|---|
| 192 | libxml2.xmlLoadCatalogs(catalogs)
|
|---|
| 193 | else:
|
|---|
| 194 | print "Variable $SGML_CATALOG_FILES not set"
|
|---|
| 195 | elif args[i] == "-xinclude" or args[i] == "--xinclude":
|
|---|
| 196 | xinclude = 1
|
|---|
| 197 | libxslt.setXIncludeDefault(1)
|
|---|
| 198 | elif args[i] == "-param" or args[i] == "--param":
|
|---|
| 199 | i = i + 1
|
|---|
| 200 | params[args[i]] = args[i + 1]
|
|---|
| 201 | i = i + 1
|
|---|
| 202 | elif args[i] == "-stringparam" or args[i] == "--stringparam":
|
|---|
| 203 | i = i + 1
|
|---|
| 204 | params[args[i]] = "'%s'" % (args[i + 1])
|
|---|
| 205 | i = i + 1
|
|---|
| 206 | elif args[i] == "-maxdepth" or args[i] == "--maxdepth":
|
|---|
| 207 | print "TODO: xsltMaxDepth mapping missing"
|
|---|
| 208 | else:
|
|---|
| 209 | print "Unknown option %s" % (args[i])
|
|---|
| 210 | usage()
|
|---|
| 211 | return(3)
|
|---|
| 212 |
|
|---|
| 213 |
|
|---|
| 214 |
|
|---|
| 215 |
|
|---|
| 216 | i = i + 1
|
|---|
| 217 |
|
|---|
| 218 | libxml2.lineNumbersDefault(1)
|
|---|
| 219 | libxml2.substituteEntitiesDefault(1)
|
|---|
| 220 | # TODO: xmlLoadExtDtdDefaultValue = XML_DETECT_IDS | XML_COMPLETE_ATTRS
|
|---|
| 221 | # if novalid:
|
|---|
| 222 | # TODO: xmlLoadExtDtdDefaultValue = 0
|
|---|
| 223 |
|
|---|
| 224 | # TODO libxslt.exsltRegisterAll();
|
|---|
| 225 | libxslt.registerTestModule()
|
|---|
| 226 |
|
|---|
| 227 | i = 0
|
|---|
| 228 | while i < len(args) and done == 0:
|
|---|
| 229 | if args[i] == "-maxdepth" or args[i] == "--maxdepth":
|
|---|
| 230 | i = i + 2
|
|---|
| 231 | continue
|
|---|
| 232 | if args[i] == "-o" or args[i] == "-output" or args[i] == "--output":
|
|---|
| 233 | i = i + 2
|
|---|
| 234 | continue
|
|---|
| 235 | if args[i] == "-param" or args[i] == "--param":
|
|---|
| 236 | i = i + 3
|
|---|
| 237 | continue
|
|---|
| 238 | if args[i] == "-stringparam" or args[i] == "--stringparam":
|
|---|
| 239 | i = i + 3
|
|---|
| 240 | continue
|
|---|
| 241 | if args[i] != "-" and args[i][0] == '-':
|
|---|
| 242 | i = i + 1
|
|---|
| 243 | continue
|
|---|
| 244 | if timing:
|
|---|
| 245 | startTimer()
|
|---|
| 246 | style = libxml2.parseFile(args[i])
|
|---|
| 247 | if timing:
|
|---|
| 248 | endTimer("Parsing stylesheet %s" % (args[i]))
|
|---|
| 249 | if style == None:
|
|---|
| 250 | print "cannot parse %s" % (args[i])
|
|---|
| 251 | cur = None
|
|---|
| 252 | errorno = 4
|
|---|
| 253 | done = 1
|
|---|
| 254 | else:
|
|---|
| 255 | cur = libxslt.loadStylesheetPI(style)
|
|---|
| 256 | if cur != None:
|
|---|
| 257 | xsltProcess(style, cur, args[i])
|
|---|
| 258 | cur = None
|
|---|
| 259 | else:
|
|---|
| 260 | cur = libxslt.parseStylesheetDoc(style)
|
|---|
| 261 | if cur == None:
|
|---|
| 262 | style.freeDoc()
|
|---|
| 263 | errorno = 5
|
|---|
| 264 | done = 1
|
|---|
| 265 | i = i + 1
|
|---|
| 266 | break
|
|---|
| 267 |
|
|---|
| 268 | while i < len(args) and done == 0 and cur != None:
|
|---|
| 269 | if timing:
|
|---|
| 270 | startTimer()
|
|---|
| 271 | if html:
|
|---|
| 272 | doc = libxml2.htmlParseFile(args[i], None)
|
|---|
| 273 | else:
|
|---|
| 274 | doc = libxml2.parseFile(args[i])
|
|---|
| 275 | if doc == None:
|
|---|
| 276 | print "unable to parse %s" % (args[i])
|
|---|
| 277 | errorno = 6
|
|---|
| 278 | i = i + 1
|
|---|
| 279 | continue
|
|---|
| 280 | if timing:
|
|---|
| 281 | endTimer("Parsing document %s" % (args[i]))
|
|---|
| 282 | xsltProcess(doc, cur, args[i])
|
|---|
| 283 | i = i + 1
|
|---|
| 284 |
|
|---|
| 285 | if cur != None:
|
|---|
| 286 | cur.freeStylesheet()
|
|---|
| 287 | params = None
|
|---|
| 288 |
|
|---|
| 289 | if __name__ == "__main__":
|
|---|
| 290 | main()
|
|---|
| 291 |
|
|---|
| 292 | # Memory debug specific
|
|---|
| 293 | libxslt.cleanup()
|
|---|
| 294 | if libxml2.debugMemory(1) != 0:
|
|---|
| 295 | print "Memory leak %d bytes" % (libxml2.debugMemory(1))
|
|---|
| 296 | libxml2.dumpMemory()
|
|---|
| 297 |
|
|---|
| 298 | sys.exit(errorno)
|
|---|
