simplexml.py

##   simplexml.py based on Mattew Allum's xmlstream.py
##
##   Copyright (C) 2003-2005 Alexey "Snake" Nezhdanov
##
##   This program is free software; you can redistribute it and/or modify
##   it under the terms of the GNU General Public License as published by
##   the Free Software Foundation; either version 2, or (at your option)
##   any later version.
##
##   This program is distributed in the hope that it will be useful,
##   but WITHOUT ANY WARRANTY; without even the implied warranty of
##   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
##   GNU General Public License for more details.

# $Id$

"""Simplexml module provides xmpppy library with all needed tools to handle XML nodes and XML streams.
I'm personally using it in many other separate projects. It is designed to be as standalone as possible."""

import xml.parsers.expat
import weakref

def XMLescape(txt):
    """Returns provided string with symbols & < > " replaced by their respective XML entities."""
    # replace also FORM FEED and ESC, because they are not valid XML chars
    return txt.replace("&", "&amp;").replace("<", "&lt;").replace(">", "&gt;").replace('"', "&quot;").replace(u'\x0C', "").replace(u'\x1B', "")

ENCODING='utf-8'
def ustr(what):
    """Converts object "what" to unicode string using it's own __str__ method if accessible or unicode method otherwise."""
    if isinstance(what, unicode): return what
    try: r=what.__str__()
    except AttributeError: r=str(what)
    if not isinstance(r, unicode): return unicode(r,ENCODING)
    return r

class Node(object):
    """ Node class describes syntax of separate XML Node. It have a constructor that permits node creation
        from set of "namespace name", attributes and payload of text strings and other nodes.
        It does not natively support building node from text string and uses NodeBuilder class for that purpose.
        After creation node can be mangled in many ways so it can be completely changed.
        Also node can be serialised into string in one of two modes: default (where the textual representation
        of node describes it exactly) and "fancy" - with whitespace added to make indentation and thus make
        result more readable by human.

        Node class have attribute FORCE_NODE_RECREATION that is defaults to False thus enabling fast node
        replication from the some other node. The drawback of the fast way is that new node shares some
        info with the "original" node that is changing the one node may influence the other. Though it is
        rarely needed (in xmpppy it is never needed at all since I'm usually never using original node after
        replication (and using replication only to move upwards on the classes tree).
    """
    FORCE_NODE_RECREATION=0
    def __init__(self, tag=None, attrs={}, payload=[], parent=None, nsp=None, node_built=False, node=None):
        """ Takes "tag" argument as the name of node (prepended by namespace, if needed and separated from it
            by a space), attrs dictionary as the set of arguments, payload list as the set of textual strings
            and child nodes that this node carries within itself and "parent" argument that is another node
            that this one will be the child of. Also the __init__ can be provided with "node" argument that is
            either a text string containing exactly one node or another Node instance to begin with. If both
            "node" and other arguments is provided then the node initially created as replica of "node"
            provided and then modified to be compliant with other arguments."""
        if node:
            if self.FORCE_NODE_RECREATION and isinstance(node, Node):
                node=str(node)
            if not isinstance(node, Node):
                node=NodeBuilder(node,self)
                node_built = True
            else:
                self.name,self.namespace,self.attrs,self.data,self.kids,self.parent,self.nsd = node.name,node.namespace,{},[],[],node.parent,{}
                for key  in node.attrs.keys(): self.attrs[key]=node.attrs[key]
                for data in node.data: self.data.append(data)
                for kid  in node.kids: self.kids.append(kid)
                for k,v in node.nsd.items(): self.nsd[k] = v
        else: self.name,self.namespace,self.attrs,self.data,self.kids,self.parent,self.nsd = 'tag','',{},[],[],None,{}
        if parent:
            self.parent = parent
        self.nsp_cache = {}
        if nsp:
            for k,v in nsp.items(): self.nsp_cache[k] = v
        for attr,val in attrs.items():
            if attr == 'xmlns':
                self.nsd[u''] = val
            elif attr.startswith('xmlns:'):
                self.nsd[attr[6:]] = val
            self.attrs[attr]=attrs[attr]
        if tag:
            if node_built:
                pfx,self.name = (['']+tag.split(':'))[-2:]
                self.namespace = self.lookup_nsp(pfx)
            else:
                if ' ' in tag:
                    self.namespace,self.name = tag.split()
                else:
                    self.name = tag
        if isinstance(payload, basestring): payload=[payload]
        for i in payload:
            if isinstance(i, Node): self.addChild(node=i)
            else: self.data.append(ustr(i))

    def lookup_nsp(self,pfx=''):
        ns = self.nsd.get(pfx,None)
        if ns is None:
            ns = self.nsp_cache.get(pfx,None)
        if ns is None:
            if self.parent:
                ns = self.parent.lookup_nsp(pfx)
                self.nsp_cache[pfx] = ns
            else:
                return 'http://www.gajim.org/xmlns/undeclared'
        return ns

    def __str__(self,fancy=0):
        """ Method used to dump node into textual representation.
            if "fancy" argument is set to True produces indented output for readability."""
        s = (fancy-1) * 2 * ' ' + "<" + self.name
        if self.namespace:
            if not self.parent or self.parent.namespace!=self.namespace:
                if 'xmlns' not in self.attrs:
                    s = s + ' xmlns="%s"'%self.namespace
        for key in self.attrs.keys():
            val = ustr(self.attrs[key])
            s = s + ' %s="%s"' % ( key, XMLescape(val) )
        s = s + ">"
        cnt = 0
        if self.kids:
            if fancy: s = s + "\n"
            for a in self.kids:
                if not fancy and (len(self.data)-1)>=cnt: s=s+XMLescape(self.data[cnt])
                elif (len(self.data)-1)>=cnt: s=s+XMLescape(self.data[cnt].strip())
                if isinstance(a, Node):
                    s = s + a.__str__(fancy and fancy+1)
                elif a:
                    s = s + a.__str__()
                cnt=cnt+1
        if not fancy and (len(self.data)-1) >= cnt: s = s + XMLescape(self.data[cnt])
        elif (len(self.data)-1) >= cnt: s = s + XMLescape(self.data[cnt].strip())
        if not self.kids and s.endswith('>'):
            s=s[:-1]+' />'
            if fancy: s = s + "\n"
        else:
            if fancy and not self.data: s = s + (fancy-1) * 2 * ' '
            s = s + "</" + self.name + ">"
            if fancy: s = s + "\n"
        return s
    def getCDATA(self):
        """ Serialise node, dropping all tags and leaving CDATA intact.
            That is effectively kills all formatiing, leaving only text were contained in XML.
        """
        s = ""
        cnt = 0
        if self.kids:
            for a in self.kids:
                s=s+self.data[cnt]
                if a: s = s + a.getCDATA()
                cnt=cnt+1
        if (len(self.data)-1) >= cnt: s = s + self.data[cnt]
        return s
    def addChild(self, name=None, attrs={}, payload=[], namespace=None, node=None):
        """ If "node" argument is provided, adds it as child node. Else creates new node from
            the other arguments' values and adds it as well."""
        if 'xmlns' in attrs:
            raise AttributeError("Use namespace=x instead of attrs={'xmlns':x}")
        if node:
            newnode=node
            node.parent = weakref.proxy(self) 
        else: newnode=Node(tag=name, parent=weakref.proxy(self), attrs=attrs, payload=payload)
        if namespace:
            newnode.setNamespace(namespace)
        self.kids.append(newnode)
        self.data.append(u'')
        return newnode
    def addData(self, data):
        """ Adds some CDATA to node. """
        self.data.append(ustr(data))
        self.kids.append(None)
    def clearData(self):
        """ Removes all CDATA from the node. """
        self.data=[]
    def delAttr(self, key):
        """ Deletes an attribute "key" """
        del self.attrs[key]
    def delChild(self, node, attrs={}):
        """ Deletes the "node" from the node's childs list, if "node" is an instance.
            Else deletes the first node that have specified name and (optionally) attributes. """
        if not isinstance(node, Node): node=self.getTag(node,attrs)
        self.kids[self.kids.index(node)]=None
        return node
    def getAttrs(self):
        """ Returns all node's attributes as dictionary. """
        return self.attrs
    def getAttr(self, key):
        """ Returns value of specified attribute. """
        try: return self.attrs[key]
        except: return None
    def getChildren(self):
        """ Returns all node's child nodes as list. """
        return self.kids
    def getData(self):
        """ Returns all node CDATA as string (concatenated). """
        return ''.join(self.data)
    def getName(self):
        """ Returns the name of node """
        return self.name
    def getNamespace(self):
        """ Returns the namespace of node """
        return self.namespace
    def getParent(self):
        """ Returns the parent of node (if present). """
        return self.parent
    def getPayload(self):
        """ Return the payload of node i.e. list of child nodes and CDATA entries.
            F.e. for "<node>text1<nodea/><nodeb/> text2</node>" will be returned list:
            ['text1', <nodea instance>, <nodeb instance>, ' text2']. """
        ret=[]
        for i in range(max(len(self.data),len(self.kids))):
            if i < len(self.data) and self.data[i]: ret.append(self.data[i])
            if i < len(self.kids) and self.kids[i]: ret.append(self.kids[i])
        return ret
    def getTag(self, name, attrs={}, namespace=None):
        """ Filters all child nodes using specified arguments as filter.
            Returns the first found or None if not found. """
        return self.getTags(name, attrs, namespace, one=1)
    def getTagAttr(self,tag,attr):
        """ Returns attribute value of the child with specified name (or None if no such attribute)."""
        try: return self.getTag(tag).attrs[attr]
        except: return None
    def getTagData(self,tag):
        """ Returns cocatenated CDATA of the child with specified name."""
        try: return self.getTag(tag).getData()
        except: return None
    def getTags(self, name, attrs={}, namespace=None, one=0):
        """ Filters all child nodes using specified arguments as filter.
            Returns the list of nodes found. """
        nodes=[]
        for node in self.kids:
            if not node: continue
            if namespace and namespace!=node.getNamespace(): continue
            if node.getName() == name:
                for key in attrs.keys():
                   if key not in node.attrs or node.attrs[key]!=attrs[key]: break
                else: nodes.append(node)
            if one and nodes: return nodes[0]
        if not one: return nodes

    def iterTags(self, name, attrs={}, namespace=None):
        """ Iterate over all children using specified arguments as filter. """
        for node in self.kids:
            if not node: continue
            if namespace is not None and namespace!=node.getNamespace(): continue
            if node.getName() == name:
                for key in attrs.keys():
                    if key not in node.attrs or \
                        node.attrs[key]!=attrs[key]: break
                else:
                    yield node

    def setAttr(self, key, val):
        """ Sets attribute "key" with the value "val". """
        self.attrs[key]=val
    def setData(self, data):
        """ Sets node's CDATA to provided string. Resets all previous CDATA!"""
        self.data=[ustr(data)]
    def setName(self,val):
        """ Changes the node name. """
        self.name = val
    def setNamespace(self, namespace):
        """ Changes the node namespace. """
        self.namespace=namespace
    def setParent(self, node):
        """ Sets node's parent to "node". WARNING: do not checks if the parent already present
            and not removes the node from the list of childs of previous parent. """
        self.parent = weakref.proxy(node) if node else None
    def setPayload(self,payload,add=0):
        """ Sets node payload according to the list specified. WARNING: completely replaces all node's
            previous content. If you wish just to add child or CDATA - use addData or addChild methods. """
        if isinstance(payload, basestring): payload=[payload]
        if add: self.kids+=payload
        else: self.kids=payload
    def setTag(self, name, attrs={}, namespace=None):
        """ Same as getTag but if the node with specified namespace/attributes not found, creates such
            node and returns it. """
        node=self.getTags(name, attrs, namespace=namespace, one=1)
        if node: return node
        else: return self.addChild(name, attrs, namespace=namespace)
    def setTagAttr(self,tag,attr,val):
        """ Creates new node (if not already present) with name "tag"
            and sets it's attribute "attr" to value "val". """
        try: self.getTag(tag).attrs[attr]=val
        except: self.addChild(tag,attrs={attr:val})
    def setTagData(self,tag,val,attrs={}):
        """ Creates new node (if not already present) with name "tag" and (optionally) attributes "attrs"
            and sets it's CDATA to string "val". """
        try: self.getTag(tag,attrs).setData(ustr(val))
        except: self.addChild(tag,attrs,payload=[ustr(val)])
    def has_attr(self,key):
        """ Checks if node have attribute "key"."""
        return key in self.attrs
    def __getitem__(self,item):
        """ Returns node's attribute "item" value. """
        return self.getAttr(item)
    def __setitem__(self,item,val):
        """ Sets node's attribute "item" value. """
        return self.setAttr(item,val)
    def __delitem__(self,item):
        """ Deletes node's attribute "item". """
        return self.delAttr(item)
    def __getattr__(self,attr):
        """ Reduce memory usage caused by T/NT classes - use memory only when needed. """
        if attr=='T':
            self.T=T(self)
            return self.T
        if attr=='NT':
            self.NT=NT(self)
            return self.NT
        raise AttributeError

class T:
    """ Auxiliary class used to quick access to node's child nodes. """
    def __init__(self,node): self.__dict__['node']=node
    def __getattr__(self,attr): return self.node.getTag(attr)
    def __setattr__(self,attr,val):
        if isinstance(val,Node): Node.__init__(self.node.setTag(attr),node=val)
        else: return self.node.setTagData(attr,val)
    def __delattr__(self,attr): return self.node.delChild(attr)

class NT(T):
    """ Auxiliary class used to quick create node's child nodes. """
    def __getattr__(self,attr): return self.node.addChild(attr)
    def __setattr__(self,attr,val):
        if isinstance(val,Node): self.node.addChild(attr,node=val)
        else: return self.node.addChild(attr,payload=[val])

DBG_NODEBUILDER = 'nodebuilder'
class NodeBuilder:
    """ Builds a Node class minidom from data parsed to it. This class used for two purposes:
        1. Creation an XML Node from a textual representation. F.e. reading a config file. See an XML2Node method.
        2. Handling an incoming XML stream. This is done by mangling
           the __dispatch_depth parameter and redefining the dispatch method.
        You do not need to use this class directly if you do not designing your own XML handler."""
    def __init__(self,data=None,initial_node=None):
        """ Takes two optional parameters: "data" and "initial_node".
            By default class initialised with empty Node class instance.
            Though, if "initial_node" is provided it used as "starting point".
            You can think about it as of "node upgrade".
            "data" (if provided) feeded to parser immidiatedly after instance init.
            """
        self.DEBUG(DBG_NODEBUILDER, "Preparing to handle incoming XML stream.", 'start')
        self._parser = xml.parsers.expat.ParserCreate()
        self._parser.StartElementHandler       = self.starttag
        self._parser.EndElementHandler         = self.endtag
        self._parser.CharacterDataHandler      = self.handle_cdata
        self._parser.StartNamespaceDeclHandler = self.handle_namespace_start
        self._parser.buffer_text = True
        self.Parse = self._parser.Parse

        self.__depth = 0
        self.__last_depth = 0
        self.__max_depth = 0
        self._dispatch_depth = 1
        self._document_attrs = None
        self._document_nsp = None
        self._mini_dom=initial_node
        self.last_is_data = 1
        self._ptr=None
        self.data_buffer = None
        self.streamError = ''
        if data:
            self._parser.Parse(data,1)

    def check_data_buffer(self):
        if self.data_buffer:
            self._ptr.data.append(''.join(self.data_buffer))
            del self.data_buffer[:]
            self.data_buffer = None

    def destroy(self):
        """ Method used to allow class instance to be garbage-collected. """
        self.check_data_buffer()
        self._parser.StartElementHandler       = None
        self._parser.EndElementHandler         = None
        self._parser.CharacterDataHandler      = None
        self._parser.StartNamespaceDeclHandler = None

    def starttag(self, tag, attrs):
        """XML Parser callback. Used internally"""
        self.check_data_buffer()
        self._inc_depth()
        self.DEBUG(DBG_NODEBUILDER, "DEPTH -> %i , tag -> %s, attrs -> %s" % (self.__depth, tag, `attrs`), 'down')
        if self.__depth == self._dispatch_depth:
            if not self._mini_dom :
                self._mini_dom = Node(tag=tag, attrs=attrs, nsp = self._document_nsp, node_built=True)
            else:
                Node.__init__(self._mini_dom,tag=tag, attrs=attrs, nsp = self._document_nsp, node_built=True)
            self._ptr = self._mini_dom
        elif self.__depth > self._dispatch_depth:
            self._ptr.kids.append(Node(tag=tag,parent=self._ptr,attrs=attrs, node_built=True))
            self._ptr = self._ptr.kids[-1]
        if self.__depth == 1:
            self._document_attrs = {}
            self._document_nsp = {}
            nsp, name = (['']+tag.split(':'))[-2:]
            for attr,val in attrs.items():
                if attr == 'xmlns':
                    self._document_nsp[u''] = val
                elif attr.startswith('xmlns:'):
                    self._document_nsp[attr[6:]] = val
                else:
                    self._document_attrs[attr] = val
            ns = self._document_nsp.get(nsp, 'http://www.gajim.org/xmlns/undeclared-root')
            try:
                self.stream_header_received(ns, name, attrs)
            except ValueError, e:
                self._document_attrs = None
                raise ValueError(str(e))
        if not self.last_is_data and self._ptr.parent:
            self._ptr.parent.data.append('')
        self.last_is_data = 0

    def endtag(self, tag ):
        """XML Parser callback. Used internally"""
        self.DEBUG(DBG_NODEBUILDER, "DEPTH -> %i , tag -> %s" % (self.__depth, tag), 'up')
        self.check_data_buffer()
        if self.__depth == self._dispatch_depth:
            if self._mini_dom.getName() == 'error':
                self.streamError = self._mini_dom.getChildren()[0].getName()
            self.dispatch(self._mini_dom)
        elif self.__depth > self._dispatch_depth:
            self._ptr = self._ptr.parent
        else:
            self.DEBUG(DBG_NODEBUILDER, "Got higher than dispatch level. Stream terminated?", 'stop')
        self._dec_depth()
        self.last_is_data = 0
        if self.__depth == 0: self.stream_footer_received()

    def handle_cdata(self, data):
        """XML Parser callback. Used internally"""
        self.DEBUG(DBG_NODEBUILDER, data, 'data')
        if self.last_is_data:
            if self.data_buffer:
                self.data_buffer.append(data)
        elif self._ptr:
            self.data_buffer = [data]
            self.last_is_data = 1

    def handle_namespace_start(self, prefix, uri):
        """XML Parser callback. Used internally"""
        self.check_data_buffer()

    def DEBUG(self, level, text, comment=None):
        """ Gets all NodeBuilder walking events. Can be used for debugging if redefined."""
    def getDom(self):
        """ Returns just built Node. """
        self.check_data_buffer()
        return self._mini_dom
    def dispatch(self,stanza):
        """ Gets called when the NodeBuilder reaches some level of depth on it's way up with the built
            node as argument. Can be redefined to convert incoming XML stanzas to program events. """
    def stream_header_received(self,ns,tag,attrs):
        """ Method called when stream just opened. """
        self.check_data_buffer()
    def stream_footer_received(self):
        """ Method called when stream just closed. """
        self.check_data_buffer()

    def has_received_endtag(self, level=0):
        """ Return True if at least one end tag was seen (at level) """
        return self.__depth <= level and self.__max_depth > level

    def _inc_depth(self):
        self.__last_depth = self.__depth
        self.__depth += 1
        self.__max_depth = max(self.__depth, self.__max_depth)

    def _dec_depth(self):
        self.__last_depth = self.__depth
        self.__depth -= 1

def XML2Node(xml):
    """ Converts supplied textual string into XML node. Handy f.e. for reading configuration file.
        Raises xml.parser.expat.parsererror if provided string is not well-formed XML. """
    return NodeBuilder(xml).getDom()

def BadXML2Node(xml):
    """ Converts supplied textual string into XML node. Survives if xml data is cutted half way round.
        I.e. "<html>some text <br>some more text". Will raise xml.parser.expat.parsererror on misplaced
        tags though. F.e. "<b>some text <br>some more text</b>" will not work."""
    return NodeBuilder(xml).getDom()