libxml.py

libxml,在UNIX/LINUX下非常重要的一个库,为XML相关应用提供方便.目前上载的是最新版本,若要取得最新版本,请参考里面的readme.

import libxml2mod
import types

# The root of all libxml2 errors.
class libxmlError(Exception): pass

#
# Errors raised by the wrappers when some tree handling failed.
#
class treeError(libxmlError):
    def __init__(self, msg):
        self.msg = msg
    def __str__(self):
        return self.msg

class parserError(libxmlError):
    def __init__(self, msg):
        self.msg = msg
    def __str__(self):
        return self.msg

class uriError(libxmlError):
    def __init__(self, msg):
        self.msg = msg
    def __str__(self):
        return self.msg

class xpathError(libxmlError):
    def __init__(self, msg):
        self.msg = msg
    def __str__(self):
        return self.msg

class ioWrapper:
    def __init__(self, _obj):
        self.__io = _obj
        self._o = None

    def io_close(self):
        if self.__io == None:
            return(-1)
        self.__io.close()
        self.__io = None
        return(0)

    def io_flush(self):
        if self.__io == None:
            return(-1)
        self.__io.flush()
        return(0)

    def io_read(self, len = -1):
        if self.__io == None:
            return(-1)
        if len < 0:
            return(self.__io.read())
        return(self.__io.read(len))

    def io_write(self, str, len = -1):
        if self.__io == None:
            return(-1)
        if len < 0:
            return(self.__io.write(str))
        return(self.__io.write(str, len))

class ioReadWrapper(ioWrapper):
    def __init__(self, _obj, enc = ""):
        ioWrapper.__init__(self, _obj)
        self._o = libxml2mod.xmlCreateInputBuffer(self, enc)

    def __del__(self):
        print "__del__"
        self.io_close()
        if self._o != None:
            libxml2mod.xmlFreeParserInputBuffer(self._o)
        self._o = None

    def close(self):
        self.io_close()
        if self._o != None:
            libxml2mod.xmlFreeParserInputBuffer(self._o)
        self._o = None

class ioWriteWrapper(ioWrapper):
    def __init__(self, _obj, enc = ""):
#        print "ioWriteWrapper.__init__", _obj
        if type(_obj) == type(''):
            print "write io from a string"
            self.o = None
        elif type(_obj) == types.InstanceType:
            print "write io from instance of %s" % (_obj.__class__)
            ioWrapper.__init__(self, _obj)
            self._o = libxml2mod.xmlCreateOutputBuffer(self, enc)
        else:
            file = libxml2mod.outputBufferGetPythonFile(_obj)
            if file != None:
                ioWrapper.__init__(self, file)
            else:
                ioWrapper.__init__(self, _obj)
            self._o = _obj

    def __del__(self):
#        print "__del__"
        self.io_close()
        if self._o != None:
            libxml2mod.xmlOutputBufferClose(self._o)
        self._o = None

    def flush(self):
        self.io_flush()
        if self._o != None:
            libxml2mod.xmlOutputBufferClose(self._o)
        self._o = None

    def close(self):
        self.io_flush()
        if self._o != None:
            libxml2mod.xmlOutputBufferClose(self._o)
        self._o = None

#
# Example of a class to handle SAX events
#
class SAXCallback:
    """Base class for SAX handlers"""
    def startDocument(self):
        """called at the start of the document"""
        pass

    def endDocument(self):
        """called at the end of the document"""
        pass

    def startElement(self, tag, attrs):
        """called at the start of every element, tag is the name of
           the element, attrs is a dictionary of the element's attributes"""
        pass

    def endElement(self, tag):
        """called at the start of every element, tag is the name of
           the element"""
        pass

    def characters(self, data):
        """called when character data have been read, data is the string
           containing the data, multiple consecutive characters() callback
           are possible."""
        pass

    def cdataBlock(self, data):
        """called when CDATA section have been read, data is the string
           containing the data, multiple consecutive cdataBlock() callback
           are possible."""
        pass

    def reference(self, name):
        """called when an entity reference has been found"""
        pass

    def ignorableWhitespace(self, data):
        """called when potentially ignorable white spaces have been found"""
        pass

    def processingInstruction(self, target, data):
        """called when a PI has been found, target contains the PI name and
           data is the associated data in the PI"""
        pass

    def comment(self, content):
        """called when a comment has been found, content contains the comment"""
        pass

    def externalSubset(self, name, externalID, systemID):
        """called when a DOCTYPE declaration has been found, name is the
           DTD name and externalID, systemID are the DTD public and system
           identifier for that DTd if available"""
        pass

    def internalSubset(self, name, externalID, systemID):
        """called when a DOCTYPE declaration has been found, name is the
           DTD name and externalID, systemID are the DTD public and system
           identifier for that DTD if available"""
        pass

    def entityDecl(self, name, type, externalID, systemID, content):
        """called when an ENTITY declaration has been found, name is the
           entity name and externalID, systemID are the entity public and
           system identifier for that entity if available, type indicates
           the entity type, and content reports it's string content"""
        pass

    def notationDecl(self, name, externalID, systemID):
        """called when an NOTATION declaration has been found, name is the
           notation name and externalID, systemID are the notation public and
           system identifier for that notation if available"""
        pass

    def attributeDecl(self, elem, name, type, defi, defaultValue, nameList):
        """called when an ATTRIBUTE definition has been found"""
        pass

    def elementDecl(self, name, type, content):
        """called when an ELEMENT definition has been found"""
        pass

    def entityDecl(self, name, publicId, systemID, notationName):
        """called when an unparsed ENTITY declaration has been found,
           name is the entity name and publicId,, systemID are the entity
           public and system identifier for that entity if available,
           and notationName indicate the associated NOTATION"""
        pass

    def warning(self, msg):
        print msg

    def error(self, msg):
        raise parserError(msg)

    def fatalError(self, msg):
        raise parserError(msg)

#
# This class is the ancestor of all the Node classes. It provides
# the basic functionalities shared by all nodes (and handle
# gracefylly the exception), like name, navigation in the tree,
# doc reference, content access and serializing to a string or URI
#
class xmlCore:
    def __init__(self, _obj=None):
        if _obj != None: 
            self._o = _obj;
            return
        self._o = None
    def __str__(self):
        return self.serialize()
    def get_parent(self):
        ret = libxml2mod.parent(self._o)
        if ret == None:
            return None
        return xmlNode(_obj=ret)
    def get_children(self):
        ret = libxml2mod.children(self._o)
        if ret == None:
            return None
        return xmlNode(_obj=ret)
    def get_last(self):
        ret = libxml2mod.last(self._o)
        if ret == None:
            return None
        return xmlNode(_obj=ret)
    def get_next(self):
        ret = libxml2mod.next(self._o)
        if ret == None:
            return None
        return xmlNode(_obj=ret)
    def get_properties(self):
        ret = libxml2mod.properties(self._o)
        if ret == None:
            return None
        return xmlAttr(_obj=ret)
    def get_prev(self):
        ret = libxml2mod.prev(self._o)
        if ret == None:
            return None
        return xmlNode(_obj=ret)
    def get_content(self):
        return libxml2mod.xmlNodeGetContent(self._o)
    getContent = get_content  # why is this duplicate naming needed ?
    def get_name(self):
        return libxml2mod.name(self._o)
    def get_type(self):
        return libxml2mod.type(self._o)
    def get_doc(self):
        ret = libxml2mod.doc(self._o)
        if ret == None:
            if self.type in ["document_xml", "document_html"]:
                return xmlDoc(_obj=self._o)
            else:
                return None
        return xmlDoc(_obj=ret)
    #
    # Those are common attributes to nearly all type of nodes
    # defined as python2 properties
    # 
    import sys
    if float(sys.version[0:3]) < 2.2:
        def __getattr__(self, attr):
            if attr == "parent":
                ret = libxml2mod.parent(self._o)
                if ret == None:
                    return None
                return xmlNode(_obj=ret)
            elif attr == "properties":
                ret = libxml2mod.properties(self._o)
                if ret == None:
                    return None
                return xmlAttr(_obj=ret)
            elif attr == "children":
                ret = libxml2mod.children(self._o)
                if ret == None:
                    return None
                return xmlNode(_obj=ret)
            elif attr == "last":
                ret = libxml2mod.last(self._o)
                if ret == None:
                    return None
                return xmlNode(_obj=ret)
            elif attr == "next":
                ret = libxml2mod.next(self._o)
                if ret == None:
                    return None
                return xmlNode(_obj=ret)
            elif attr == "prev":
                ret = libxml2mod.prev(self._o)
                if ret == None:
                    return None
                return xmlNode(_obj=ret)
            elif attr == "content":
                return libxml2mod.xmlNodeGetContent(self._o)
            elif attr == "name":
                return libxml2mod.name(self._o)
            elif attr == "type":
                return libxml2mod.type(self._o)
            elif attr == "doc":
                ret = libxml2mod.doc(self._o)
                if ret == None:
                    if self.type == "document_xml" or self.type == "document_html":
                        return xmlDoc(_obj=self._o)
                    else:
                        return None
                return xmlDoc(_obj=ret)
            raise AttributeError,attr
    else:
        parent = property(get_parent, None, None, "Parent node")
        children = property(get_children, None, None, "First child node")
        last = property(get_last, None, None, "Last sibling node")
        next = property(get_next, None, None, "Next sibling node")
        prev = property(get_prev, None, None, "Previous sibling node")
        properties = property(get_properties, None, None, "List of properies")