etisserant@0: #
etisserant@0: # minixsv, Release 0.3
etisserant@0: # file: pyxsval.py
etisserant@0: #
etisserant@0: # API for XML schema validator
etisserant@0: #
etisserant@0: # history:
etisserant@0: # 2004-09-09 rl   created
etisserant@0: # 2004-09-29 rl   adapted to re-designed XML interface classes,
etisserant@0: #                 ErrorHandler separated, URL processing added, some bugs fixed
etisserant@0: # 2004-10-07 rl   Validator classes extracted into separate files
etisserant@0: # 2004-10-12 rl   API re-worked, XML text processing added
etisserant@0: #
etisserant@0: # Copyright (c) 2004 by Roland Leuthe.  All rights reserved.
etisserant@0: #
etisserant@0: # --------------------------------------------------------------------
etisserant@0: # The minixsv XML schema validator is
etisserant@0: #
etisserant@0: # Copyright (c) 2004 by Roland Leuthe
etisserant@0: #
etisserant@0: # By obtaining, using, and/or copying this software and/or its
etisserant@0: # associated documentation, you agree that you have read, understood,
etisserant@0: # and will comply with the following terms and conditions:
etisserant@0: #
etisserant@0: # Permission to use, copy, modify, and distribute this software and
etisserant@0: # its associated documentation for any purpose and without fee is
etisserant@0: # hereby granted, provided that the above copyright notice appears in
etisserant@0: # all copies, and that both that copyright notice and this permission
etisserant@0: # notice appear in supporting documentation, and that the name of
etisserant@0: # the author not be used in advertising or publicity
etisserant@0: # pertaining to distribution of the software without specific, written
etisserant@0: # prior permission.
etisserant@0: #
etisserant@0: # THE AUTHOR DISCLAIMS ALL WARRANTIES WITH REGARD
etisserant@0: # TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANT-
etisserant@0: # ABILITY AND FITNESS.  IN NO EVENT SHALL THE AUTHOR
etisserant@0: # BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY
etisserant@0: # DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS,
etisserant@0: # WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
etisserant@0: # ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE
etisserant@0: # OF THIS SOFTWARE.
etisserant@0: # --------------------------------------------------------------------
etisserant@0: 
etisserant@0: __all__ = [
etisserant@0:     # public symbols
etisserant@0:     "parseAndValidate",
etisserant@0:     "parseAndValidateString",
etisserant@0:     "parseAndValidateXmlInput",
etisserant@0:     "parseAndValidateXmlInputString",
etisserant@0:     "parseAndValidateXmlSchema",
etisserant@0:     "parseAndValidateXmlSchemaString",
etisserant@0:     "XsValidator",
etisserant@0:     "XMLIF_MINIDOM",
etisserant@0:     "XMLIF_ELEMENTTREE"
etisserant@0:     ]
etisserant@0: 
etisserant@0: 
etisserant@0: from xsvalErrorHandler import *
etisserant@0: from xsvalBase import XsValBase
etisserant@0: from xsvalSchema import XsValSchema
etisserant@0: 
etisserant@0: 
etisserant@0: XMLIF_MINIDOM     = "XMLIF_MINIDOM"
etisserant@0: XMLIF_ELEMENTTREE = "XMLIF_ELEMENTTREE"
etisserant@0: 
etisserant@0: 
etisserant@0: _XS_VAL_DEFAULT_ERROR_LIMIT = 20
etisserant@0: 
etisserant@0: _OWNDIR = os.path.dirname(__file__)
etisserant@0: 
etisserant@0: 
etisserant@0: ########################################
etisserant@0: # convenience function for validating 
etisserant@0: # 1. XML schema file
etisserant@0: # 2. XML input file
etisserant@0: # If xsdFile is specified, it will be used for validation
etisserant@0: # If xsdFile=None, the schemaLocation attribute is expected in the root tag of the XML input file
etisserant@0: #
etisserant@0: def parseAndValidate (inputFile, xsdFile=None, **kw):
etisserant@0:     return parseAndValidateXmlInput (inputFile, xsdFile, 1, **kw)
etisserant@0: 
etisserant@0: 
etisserant@0: ########################################
etisserant@0: # convenience function for validating 
etisserant@0: # 1. text string containing the XML schema
etisserant@0: # 2. text string containing the XML input
etisserant@0: # If xsdText is given, it will be used for validation
etisserant@0: # If xsdText=None, the schemaLocation attribute is expected in the root tag of the XML input
etisserant@0: #
etisserant@0: def parseAndValidateString (inputText, xsdText=None, **kw):
etisserant@0:     return parseAndValidateXmlInputString (inputText, xsdText, 1, **kw)
etisserant@0: 
etisserant@0: 
etisserant@0: ########################################
etisserant@0: # factory for validating 
etisserant@0: # 1. XML schema file (only if validateSchema=1)
etisserant@0: # 2. XML input file
etisserant@0: # If xsdFile is specified, it will be used for validation
etisserant@0: # If xsdFile=None, the schemaLocation attribute is expected in the root tag of the XML input file
etisserant@0: #
etisserant@0: def parseAndValidateXmlInput (inputFile, xsdFile=None, validateSchema=0, **kw):
etisserant@0:     xsValidator = XsValidator (**kw)
etisserant@0:     # parse XML input file
etisserant@0:     inputTreeWrapper = xsValidator.parse (inputFile)
etisserant@0:     # validate XML input file
etisserant@0:     return xsValidator.validateXmlInput (inputFile, inputTreeWrapper, xsdFile, validateSchema)
etisserant@0: 
etisserant@0: 
etisserant@0: ########################################
etisserant@0: # factory for validating 
etisserant@0: # 1. text string containing the XML schema (only if validateSchema=1)
etisserant@0: # 2. text string containing the XML input
etisserant@0: # If xsdText is given, it will be used for validation
etisserant@0: # If xsdText=None, the schemaLocation attribute is expected in the root tag of the XML input
etisserant@0: #
etisserant@0: def parseAndValidateXmlInputString (inputText, xsdText=None, validateSchema=0, **kw):
etisserant@0:     xsValidator = XsValidator (**kw)
etisserant@0:     # parse XML input text string
etisserant@0:     inputTreeWrapper = xsValidator.parseString (inputText)
etisserant@0:     # validate XML input text string
etisserant@0:     return xsValidator.validateXmlInputString (inputTreeWrapper, xsdText, validateSchema)
etisserant@0: 
etisserant@0: 
etisserant@0: ########################################
etisserant@0: # factory for validating only given XML schema file
etisserant@0: #
etisserant@0: def parseAndValidateXmlSchema (xsdFile, **kw):
etisserant@0:     xsValidator = XsValidator (**kw)
etisserant@0:     # parse XML schema file
etisserant@0:     xsdTreeWrapper = xsValidator.parse (xsdFile)
etisserant@0:     # validate XML schema file
etisserant@0:     return xsValidator.validateXmlSchema (xsdFile, xsdTreeWrapper)
etisserant@0: 
etisserant@0: 
etisserant@0: ########################################
etisserant@0: # factory for validating only given XML schema file
etisserant@0: #
etisserant@0: def parseAndValidateXmlSchemaString (xsdText, **kw):
etisserant@0:     xsValidator = XsValidator (**kw)
etisserant@0:     # parse XML schema
etisserant@0:     xsdTreeWrapper = xsValidator.parseString (xsdText)
etisserant@0:     # validate XML schema
etisserant@0:     return xsValidator.validateXmlSchema ("", xsdTreeWrapper)
etisserant@0: 
etisserant@0: 
etisserant@0: ########################################
etisserant@0: # XML schema validator class
etisserant@0: #
etisserant@0: class XsValidator:
etisserant@0:     def __init__(self, xmlIfClass=XMLIF_MINIDOM,
etisserant@0:                  warningProc=IGNORE_WARNINGS, errorLimit=_XS_VAL_DEFAULT_ERROR_LIMIT, verbose=0):
etisserant@0: 
etisserant@0:         self.warningProc   = warningProc
etisserant@0:         self.errorLimit    = errorLimit
etisserant@0:         self.verbose       = verbose
etisserant@0: 
etisserant@0:         # enable interface to minidom
etisserant@0:         if xmlIfClass == XMLIF_MINIDOM:
etisserant@0:             import minidomif
etisserant@0:             self.xmlIf = minidomif.MiniDomInterface(verbose)
etisserant@0: 
etisserant@0:         # enable interface to elementtree
etisserant@0:         elif xmlIfClass == XMLIF_ELEMENTTREE:
etisserant@0:             import elemtreeif
etisserant@0:             self.xmlIf = elemtreeif.ElementTreeInterface(verbose)
etisserant@0: 
etisserant@0:         # create error handler
etisserant@0:         self.errorHandler  = ErrorHandler (errorLimit, warningProc, verbose)
etisserant@0: 
etisserant@0: 
etisserant@0:     ########################################
etisserant@0:     # parse XML file
etisserant@0:     # 'file' may be a filepath or an URI
etisserant@0:     #
etisserant@0:     def parse (self, file, baseUrl=""):
etisserant@0:         try:
etisserant@0:             self._verbosePrint ("Parsing %s..." %(file))
etisserant@0:             treeWrapper = self.xmlIf.extParse(file, baseUrl)
etisserant@0:         except AssertionError, errstr:
etisserant@0:             self.errorHandler.raiseError (errstr)
etisserant@0:         return treeWrapper
etisserant@0: 
etisserant@0: 
etisserant@0:     ########################################
etisserant@0:     # parse text string containing XML 
etisserant@0:     #
etisserant@0:     def parseString (self, text):
etisserant@0:         try:
etisserant@0:             self._verbosePrint ("Parsing XML text string...")
etisserant@0:             treeWrapper = self.xmlIf.extParseString(text)
etisserant@0:         except AssertionError, errstr:
etisserant@0:             self.errorHandler.raiseError (errstr)
etisserant@0:         return treeWrapper
etisserant@0: 
etisserant@0: 
etisserant@0:     ########################################
etisserant@0:     # validate XML input
etisserant@0:     #
etisserant@0:     def validateXmlInput (self, xmlInputFile, inputTreeWrapper, xsdFile=None, validateSchema=0):
etisserant@0:         # parse XML schema file
etisserant@0:         if xsdFile != None:
etisserant@0:             xsdTreeWrapper = self.parse (xsdFile)
etisserant@0:         else:
etisserant@0:             # a schemaLocation attribute is expected in the root tag of the XML input file
etisserant@0:             xsdFile = self._retrieveReferencedXsdFile (inputTreeWrapper)
etisserant@0:             xsdTreeWrapper = self.parse (xsdFile, inputTreeWrapper.getRootNode().getAbsUrl())
etisserant@0: 
etisserant@0:         return self._validateXmlInput (xmlInputFile, xsdFile, inputTreeWrapper, xsdTreeWrapper, validateSchema)
etisserant@0: 
etisserant@0: 
etisserant@0:     ########################################
etisserant@0:     # validate XML input
etisserant@0:     #
etisserant@0:     def validateXmlInputString (self, inputTreeWrapper, xsdText=None, validateSchema=0):
etisserant@0:         # parse XML schema file
etisserant@0:         if xsdText != None:
etisserant@0:             xsdFile = "schema text"
etisserant@0:             xsdTreeWrapper = self.parseString (xsdText)
etisserant@0:         else:
etisserant@0:             # a schemaLocation attribute is expected in the root tag of the XML input file
etisserant@0:             xsdFile = self._retrieveReferencedXsdFile (inputTreeWrapper)
etisserant@0:             xsdTreeWrapper = self.parse (xsdFile, inputTreeWrapper.getRootNode().getAbsUrl())
etisserant@0: 
etisserant@0:         return self._validateXmlInput ("input text", xsdFile, inputTreeWrapper, xsdTreeWrapper, validateSchema)
etisserant@0: 
etisserant@0: 
etisserant@0:     ########################################
etisserant@0:     # validate XML schema separately
etisserant@0:     #
etisserant@0:     def validateXmlSchema (self, xsdFile, xsdTreeWrapper):
etisserant@0:         # parse minixsv internal schema
etisserant@0:         rulesTreeWrapper = self.parse(os.path.join (_OWNDIR, "xsStructs.xsd"))
etisserant@0: 
etisserant@0:         self._verbosePrint ("Validating %s..." %(xsdFile))
etisserant@0:         xsvGivenXsdFile = XsValSchema (self.xmlIf, self.errorHandler)
etisserant@0:         xsvGivenXsdFile.validate(xsdTreeWrapper, rulesTreeWrapper)
etisserant@0:         self.errorHandler.flushOutput()
etisserant@0:         return xsdTreeWrapper
etisserant@0: 
etisserant@0: 
etisserant@0:     ########################################
etisserant@0:     # validate XML input tree and xsd tree if requested
etisserant@0:     #
etisserant@0:     def _validateXmlInput (self, xmlInputFile, xsdFile, inputTreeWrapper, xsdTreeWrapper, validateSchema=0):
etisserant@0:         # validate XML schema file if requested
etisserant@0:         if validateSchema:
etisserant@0:             self.validateXmlSchema (xsdFile, xsdTreeWrapper)
etisserant@0: 
etisserant@0:         try:
etisserant@0:             xsdTreeWrapper.insertSubtree (xsdTreeWrapper.getRootNode().getFirstChild(), os.path.join (_OWNDIR, "datatypes2.xsd"), None)
etisserant@0:         except AssertionError, errstr:
etisserant@0:             self.errorHandler.raiseError (errstr)
etisserant@0: 
etisserant@0:         self._verbosePrint ("Validating %s..." %(xmlInputFile))
etisserant@0:         xsvInputFile = XsValBase (self.xmlIf, self.errorHandler)
etisserant@0:         xsvInputFile.validate(inputTreeWrapper, xsdTreeWrapper)
etisserant@0:         self.errorHandler.flushOutput()
etisserant@0:         return inputTreeWrapper
etisserant@0: 
etisserant@0: 
etisserant@0:     ########################################
etisserant@0:     # retrieve XML schema location from XML input tree
etisserant@0:     #
etisserant@0:     def _retrieveReferencedXsdFile (self, inputTreeWrapper):
etisserant@0:         # a schemaLocation attribute is expected in the root tag of the XML input file
etisserant@0:         for attributeName, attributeValue in inputTreeWrapper.getRootNode().getAttributeDict().items():
etisserant@0:             if self.xmlIf.extractLocalName(attributeName) in ("noNamespaceSchemaLocation", "schemaLocation"):
etisserant@0:                 xsdFile = attributeValue
etisserant@0:                 return xsdFile
etisserant@0:         else:
etisserant@0:             self.errorHandler.raiseError ("No schema file specified!")
etisserant@0: 
etisserant@0: 
etisserant@0:     ########################################
etisserant@0:     # print if verbose flag is set
etisserant@0:     #
etisserant@0:     def _verbosePrint (self, text):
etisserant@0:         if self.verbose:
etisserant@0:             print text
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: