Viewing file:  BuiltInExtElements.py (23.53 KB) -rw-r--r--
Select action/file-type:
 (+) |  (+) |  (+) | Code (+) | Session (+) |  (+) | SDB (+) |  (+) |  (+) |  (+) |  (+) |  (+) |
########################################################################
# $Header: /var/local/cvsroot/4Suite/Ft/Xml/Xslt/BuiltInExtElements.py,v 1.52 2005/04/06 23:05:47 jkloth Exp $
"""
Fourthought proprietary XSLT extension elements
Copyright 2005 Fourthought, Inc. (USA).
Detailed license and copyright information: http://4suite.org/COPYRIGHT
Project home, documentation, distributions: http://4suite.org/
"""
from xml.dom import Node
from Ft.Xml import EMPTY_NAMESPACE
from Ft.Xml.XPath import Conversions
from Ft.Xml.XPath import FT_EXT_NAMESPACE
from Ft.Xml.Xslt import XSL_NAMESPACE, XsltElement, XsltException, Error
from Ft.Xml.Xslt import CategoryTypes, ContentInfo, AttributeInfo
from Ft.Xml.Xslt import ApplyTemplatesElement, ApplyImportsElement
from Ft.Xml.Xslt import OutputParameters
RESERVED_NAMESPACE = u'http://xmlns.4suite.org/reserved'
__all__ = ['RESERVED_NAMESPACE',
'ExtNamespaces', 'ExtElements',
'AssignElement',
'ChainToElement',
'CreateIndexElement',
'DumpKeysElement',
'DumpVarsElement',
'FtApplyImports',
'FtApplyTemplates',
'FtOutputElement',
'MsgControlElement',
'RawTextOutputElement',
'ReplaceElement',
'UriToElementElement',]
class FtApplyImports(XsltElement):
"""
The f:apply-imports element is an extension of the xsl:apply-imports
element. It differs from xsl:apply-imports in the following way:
The element accepts xsl:with-param children that designate
parameters that will be passed to the applied templates.
"""
category = CategoryTypes.INSTRUCTION
legalAttrs = {}
content = ContentInfo.Rep(
ContentInfo.QName(XSL_NAMESPACE, 'xsl:with-param')
)
doesSetup = 1
def setup(self):
self._params = []
for child in self.children:
if child.expandedName == (XSL_NAMESPACE, 'with-param'):
self._params.append((child, child._name, child._select))
return
def instantiate(self, context, processor):
if not context.stylesheet:
raise XsltRuntimeException(
Error.APPLYIMPORTS_WITH_NULL_CURRENT_TEMPLATE, self)
context.processorNss = self.namespaces
context.currentInstruction = self
with_params = {}
for (param, name, expr) in self._params:
context.processorNss = param.namespaces
context.currentInstruction = param
with_params[name] = expr.evaluate(context)
context.stylesheet.applyTemplates(context, processor,
params=with_params,
maxImport=self.importIndex)
return
class FtApplyTemplates(ApplyTemplatesElement.ApplyTemplatesElement):
"""
The f:apply-templates element is an extension of the xsl:apply-templates
element. It differs from xsl:apply-templates in the following way:
The value of the mode attribute is an attribute value template
rather than a static string. Thus, the mode can be computed at
run time.
"""
legalAttrs = ApplyTemplatesElement.ApplyTemplatesElement.legalAttrs.copy()
legalAttrs['mode'] = AttributeInfo.QNameAvt(description="The mode to be used for template application. In this variation the mode is an AVT and thus can be computed at run time.")
def _instantiate_mode(self, context):
return self._mode.evaluate(context)
class FtOutputElement(XsltElement):
"""
f:output is similar to xsl:output, but it allows you to compute the
output parameters dynamically (as attribute value templates). Unlike
xsl:output, this element is not expected to be empty; the output
parameters apply only to the serialization of the element's content.
"""
content = ContentInfo.Template
legalAttrs = {
'method' : AttributeInfo.QNameAvt(),
'version' : AttributeInfo.NMTokenAvt(),
'encoding' : AttributeInfo.StringAvt(),
'omit-xml-declaration' : AttributeInfo.YesNoAvt(),
'standalone' : AttributeInfo.YesNoAvt(),
'doctype-public' : AttributeInfo.StringAvt(),
'doctype-system' : AttributeInfo.StringAvt(),
'cdata-section-elements' : AttributeInfo.QNamesAvt(),
'indent' : AttributeInfo.YesNoAvt(),
'media-type' : AttributeInfo.StringAvt(),
}
def __init__(self, *args, **kwds):
XsltElement.__init__(self, *args, **kwds)
self._output_parameters = OutputParameters.OutputParameters()
return
def instantiate(self, context, processor):
context.processorNss = self.namespaces
context.currentInstruction = self
# this uses attributes directly from self
self._output_parameters.avtParse(self, context)
processor.addHandler(self._output_parameters,
processor.writers[-1].getStream())
try:
for child in self.children:
child.instantiate(context, processor)
finally:
processor.removeHandler()
return
class AssignElement(XsltElement):
"""The f:assign element works like xsl:variable, but forces both a local
and a global variable binding, replacing any other in-scope bindings
having the same expanded-name. Thus, it can be used to circumvent XSLT's
restriction on variables not being reassignable. However, its use is not
recommended, for reasons explained below.
As with xsl:variable, the name of the variable is given in the mandatory
name attribute, and the new value may be given either by an expression in
the select attribute, or by instantiating the content of the element.
If no select attribute is given, then a body-as-ns attribute may be used
to indicate whether to assign the variable to the contents as a node-set
(value 'yes') or as a result tree fragment (default, or value 'no').
In either case, be aware that the node-set or result tree fragment will
have a root node.
Note that reassignment of variables is generally never actually needed.
Before using f:assign, read the XSL FAQ or ask on xsl-list if there is a
better, more portable way to solve your problem.
XSLT is designed as a language that is free of side-effects, which is
why assignment is not allowed and variables have very specific scope.
When variable assignment is allowed, certain optimizations in the XSLT
processor become impossible. Also, there are some circumstances in which
the order of execution may not be quite what you expect, in which case
f:assign may show anomalous behavior. It does not work predictably when
called from within a tail-recursive template, for example.
That said, f:assign can be a convenient way to create a node-set from
a result tree fragment in XSLT 1.0. The proper way to do this is with
EXSLT: <xsl:variable name="rtf"><foo/></xsl:variable>
<xsl:variable name="ns" select="exsl:node-set($rtf)" xmlns:exsl="http://exslt.org/common"/>
but f:assign can do it in one step:
<f:assign name="ns" body-as-ns="yes"><foo/></f:assign>"""
legalAttrs = {
'name' : AttributeInfo.QName(required=1),
'select' : AttributeInfo.Expression(),
'body-as-ns' : AttributeInfo.YesNoAvt(default='no'),
}
def instantiate(self, context, processor):
context.processorNss = self.namespaces
context.currentInstruction = self
if self._select:
result = self._select.evaluate(context)
else:
processor.pushResultTree(self.baseUri)
try:
for child in self.children:
child.instantiate(context, processor)
finally:
result = processor.popResult()
if self._body_as_ns.evaluate(context):
result = [result]
context.varBindings[self._name] = result
context.processor.stylesheet.globalVars[self._name] = result
return
class DumpKeysElement(XsltElement):
"""
f:dump-keys reports the XSLT keys that have been defined, and the
nodes they identify, for the document that owns the context node.
Keys will only be reported if key() has been evaluated prior to
the instantiation of this element. The key() evaluation must have
been performed with a context node that is from the same document
as the context node for this element.
This extension element is useful for debugging.
By default, the key data is exposed as nodes with this structure:
<zz:KeyDump xmlns:zz="%s">
<zz:Key name="keyName">
<zz:MatchSet value="keyUseValue">
(representation of nodes matched by the key)
</zz:MatchSet>
...
</zz:Key>
...
</zz:KeyDump>
The node representation will be a copy of each of the nodes,
except for attributes. Attribute nodes matched by the key will
manifest as comment nodes with the content "Attribute: name=value".
If raw="yes", the keys will be emitted as a stylesheet message
(as if via xsl:message) and the format will be their Python repr()
representation.
If force-update="yes" all keys will be computed on all documents
that have been loaded into the processor.
4Suite evaluates keys lazily, which means that you could have
situations where f:dump-keys returns unexpected empty results
because the key of interest has not yet been invoked.
""" % RESERVED_NAMESPACE
legalAttrs = {
'raw' : AttributeInfo.YesNoAvt(default='no', description="Present keys in a compact non-XML format"),
'force-update' : AttributeInfo.YesNoAvt(default='no', description="Force evaluation of all keys on all loaded documents"),
}
def instantiate(self, context, processor):
from Ft.Xml.Xslt.CopyOfElement import CopyNode
if self._force_update.evaluate(context):
sheet = context.processor.stylesheet
sheet.updateAllKeys(context, processor)
doc = context.node.rootNode
try:
xkeys = processor.keys[doc]
except KeyError:
xkeys = {}
if (FT_EXT_NAMESPACE, 'indices') in processor.extensionParams:
for k, v in processor.extensionParams[(FT_EXT_NAMESPACE, 'indices')].items():
#Dummy to xsl:key storage format
xkeys[(None, k)] = v
if self._raw.evaluate(context):
processor.xslMessage(repr(xkeys))
else:
writer = processor.writers[-1]
writer.startElement(u'zz:KeyDump', RESERVED_NAMESPACE)
for k, v in xkeys.items():
writer.startElement(u'zz:Key', RESERVED_NAMESPACE)
writer.attribute(u'name', k[1], EMPTY_NAMESPACE)
for kk, vv in v.items():
writer.startElement(u'zz:MatchSet', RESERVED_NAMESPACE)
writer.attribute(u'value', kk, EMPTY_NAMESPACE)
for node in vv:
if node.nodeType == Node.ATTRIBUTE_NODE:
processor.writers[-1].comment(
u"Attribute: %s=%s"%(node.nodeName, node.value))
else:
CopyNode(processor, node)
writer.endElement(u'zz:MatchSet', RESERVED_NAMESPACE)
writer.endElement(u'zz:Key', RESERVED_NAMESPACE)
writer.endElement(u'zz:KeyDump', RESERVED_NAMESPACE)
return
class DumpVarsElement(XsltElement):
"""
f:dump-vars reports the XPath/XSLT variables and parameters in scope.
This extension element is useful for debugging.
By default, the variables are exposed as nodes with this structure:
<zz:varDump xmlns:zz="%s">
<zz:var name="variableName">(representation of object)</zz:var>
...
</zz:varDump>
The representation of the object bound to the variable depends on
the object type: If the object is not a node-set or result tree
fragment, the representation will be the object's string-value.
If the object is a node-set or result tree fragment, the
representation will be a copy of each of the nodes. Attribute nodes
in the node-set or result tree fragment will manifest as comment
nodes with the content "Attribute: name=value".
If raw="yes", the variables will be emitted as a stylesheet message
(as if via xsl:message) and the format will be their Python repr()
representation.
""" % RESERVED_NAMESPACE
legalAttrs = {
'raw' : AttributeInfo.YesNoAvt(default='no', description="Present variables and values in a compact non-XML format"),
}
def instantiate(self, context, processor):
doc = context.node.rootNode
if self._raw.evaluate(context):
processor.xslMessage(repr(context.varBindings))
else:
from Ft.Xml.XPath.Types import g_xpathPrimitiveTypes
from Ft.Xml.Xslt.CopyOfElement import CopyNode
writer = processor.writers[-1]
writer.startElement(u'zz:VarDump', RESERVED_NAMESPACE)
for k, v in context.varBindings.items():
writer.startElement(u'zz:Var', RESERVED_NAMESPACE)
#FIXME: should try to join back prefix to var name
writer.attribute(u'name', k[1], EMPTY_NAMESPACE)
if isinstance(v, list):
# NOTE - this must be before the primitive check due to
# the fact that a node-set is a primitive type
for node in v:
if node.nodeType == Node.ATTRIBUTE_NODE:
processor.writers[-1].comment(
u"Attribute: %s=%s"%(node.nodeName, node.value))
else:
CopyNode(processor, node)
elif type(v) in g_xpathPrimitiveTypes:
writer.text(Conversions.StringValue(v))
elif hasattr(v, 'nodeType'):
CopyNode(processor, v)
writer.endElement(u'zz:Var', RESERVED_NAMESPACE)
writer.endElement(u'zz:VarDump', RESERVED_NAMESPACE)
return
class ReplaceElement(XsltElement):
"""
f:replace performs a search and replace on a string, placing the results
in the output. The content is treated as a template. The string value
of the output from this template is the replacement string.
All instances of the string given by the 'substring' attribute
are replaced with the replacement string.
"""
legalAttrs = {
'string' : AttributeInfo.StringExpression(description="The string to be processed. If not given, the string value of the context node is used."),
'substring' : AttributeInfo.StringExpression(required=1, description="The sub-string to be replaced."),
}
def instantiate(self, context, processor):
context.processorNss = self.namespaces
context.currentInstruction = self
if self._string:
value = self._string.evaluate(context)
else:
value = context.node
string_ = Conversions.StringValue(value)
substring = Conversions.StringValue(self._substring.evaluate(context))
writer = processor.writers[-1]
for chunk in string_.split(substring):
writer.text(chunk)
for child in self.children:
child.instantiate(context, processor)
return
class MsgControlElement(XsltElement):
"""
f:msg-control provides, as a side effect, context-sensitive control
over whether messages (i.e., those produced by xsl:message) and
warnings are output by the processor.
"""
legalAttrs = {
'suppress' : AttributeInfo.YesNoAvt(default='no', description="Disable display of all XSLT messages."),
}
def instantiate(self, context, processor):
processor.messageControl(self._suppress.evaluate(context))
return
class CreateIndexElement(XsltElement):
"""
f:create-index allows one to create an arbitrary key at run time using
any node data. It is similar to xsl:key, except that it is computed
on demand at run-time, and uses an XPath selection rather than an XSLT
match, which gives more flexibility over what is indexed.
These keys can be accessed using the extension function f:lookup().
Avoid making a dynamic index have the same name as a proper xsl:key.
In particular this will confuse tools such as the <f:dump-keys/>
diagnostic extension.
"""
legalAttrs = {
'name' : AttributeInfo.StringAvt(required=1, description='The name of the key to create'),
'select' : AttributeInfo.Expression(required=1, description='Selects which nodes are to be indexed'),
'use' : AttributeInfo.Expression(required=1, description='The expression that computes the index key value for each node'),
}
def instantiate(self, context, processor):
selected = self._select.evaluate(context)
name = self._name.evaluate(context)
index = {}
if not isinstance(selected, list):
return
state = context.copy()
size = len(selected)
pos = 1
for node in selected:
context.node, context.position, context.size = node, pos, size
context.currentNode = node
key = Conversions.StringValue(self._use.evaluate(context))
if key not in index:
index[key] = []
index[key].append(node)
pos += 1
if (FT_EXT_NAMESPACE, 'indices') not in processor.extensionParams:
processor.extensionParams[(FT_EXT_NAMESPACE, 'indices')] = {}
processor.extensionParams[(FT_EXT_NAMESPACE, 'indices')][name] = index
context.set(state)
return
class RawTextOutputElement(XsltElement):
"""
Given a foreign XPath object, f:raw-text-output creates a text node
based on the object, just like xsl:value-of with
disable-output-escaping="yes". Unlike xsl:value-of, however, this
element does not use the string-value of the object; it instead
feeds the object directly to the current output writer. Therefore,
unless a custom output writer is used, the object must be a Python
Unicode string.
The intent is to provide a way to serialize a Unicode string that may
contain characters that are not permitted in an XPath string object.
For example, another extension can convert raw binary data to a
Unicode string, and then this extension can reserialize that string
through the XSLT output stream, without risk of losing any data due
to XPath's restrictions on string content.
"""
legalAttrs = {
'select' : AttributeInfo.Expression(required=1,
description="An XPath expression that returns a Python Unicode object."),
}
def instantiate(self, context, processor):
processor.writers[-1].text(self._select.evaluate(context), escapeOutput=False)
return
class ChainToElement(XsltElement):
"""
f:chain-to tells the processor to apply the output of the current
stylsheet as the input of another stylesheet, establishing a chain of
transforms. The next stylesheet in the chain is specified using an
AVT, which allows for dynamically constructed chains.
Children can be xsl:with-param elements, in which case the specified
values are passed on to the next stylesheet as top-level parameters
Warning: if the href attribute is blank, it will chain back to this
same stylesheet and could lead to an infinite loop.
FIXME: Trap this condition
"""
legalAttrs = {
'href' : AttributeInfo.UriReferenceAvt(
required=1,
description="The URI of the next stylesheet in the chain"),
}
content = ContentInfo.Rep(ContentInfo.QName(XSL_NAMESPACE, 'xsl:with-param'))
validator = ContentInfo.Validator(content)
def instantiate(self, context, processor):
href = self._href.evaluate(context)
params = {}
for child in self.children:
context.processorNss = child.namespaces
context.currentInstruction = child
params[child._name] = child._select.evaluate(context)
base = self.baseUri
processor.chainTo = processor.inputSourceFactory.resolver.normalize(href, base)
processor.chainParams = params
#print "chain to: %s, from base %s and href %s"%(processor.chainTo, base, href)
return
import ElementElement
class UriToElementElement(ElementElement.ElementElement):
"""
Extends xsl:element by deriving the constructed element's QName and
namespace from the supplied URI reference. The URI reference is
first resolved to absolute form. Then, if the resulting URI begins
with an in-scope namespace, that namespace will be used as if it had
been supplied as the 'namespace' attribute to xsl:element, and the
remainder of the URI will be combined with a prefix from the
in-scope namespace bindings and used as if supplied as the 'name'
attribute to xsl:element.
Otherwise, the supplied default-name and default-namespace will be
used, effecting the same result as calling xsl:element with these
values.
The intent is to allow an RDF resource, as identified by a URI with
a fragment component, to be easily converted into an element.
"""
content = ContentInfo.Template
legalAttrs = {
'uri' : AttributeInfo.UriReferenceAvt(description='A URI to be used to create the element. An attempt will be made to split the URI into a head and a tail such that the head matches an in-scope namespace URI. If matched a qname will be constructed from the prefix of that namespace declaration and the tail of the URI and the namespace from the declaration will be used to complete an output element.', required=1),
'default-name' : AttributeInfo.RawQNameAvt(description='Used if the given URI cannot be broken down using in-scope namespaces', required=1),
'default-namespace' : AttributeInfo.UriReferenceAvt(description='Used if the given URI cannot be broken down using in-scope namespaces', isNsName=1),
'use-attribute-sets' : AttributeInfo.QNames(),
}
def instantiate(self, context, processor):
context.currentInstruction = self
matched = 0
uri = self._uri.evaluate(context)
for (prefix, namespace) in self.namespaces.items():
if namespace and uri.startswith(namespace):
local = uri.split(namespace)[1]
if prefix:
qname = prefix + u':' + local
else:
qname = local
matched = 1
break
if matched:
ElementElement.ElementElement.execute(self, context, processor, qname, namespace)
return
else:
self._name = self._default_name
self._namespace = self._default_namespace
return ElementElement.ElementElement.instantiate(self, context, processor)
ExtNamespaces = {
FT_EXT_NAMESPACE : 'f',
}
ExtElements = {
(FT_EXT_NAMESPACE, 'apply-imports'): FtApplyImports,
(FT_EXT_NAMESPACE, 'apply-templates'): FtApplyTemplates,
(FT_EXT_NAMESPACE, 'output'): FtOutputElement,
(FT_EXT_NAMESPACE, 'dump-keys'): DumpKeysElement,
(FT_EXT_NAMESPACE, 'dump-vars'): DumpVarsElement,
(FT_EXT_NAMESPACE, 'assign'): AssignElement,
(FT_EXT_NAMESPACE, 'replace'): ReplaceElement,
(FT_EXT_NAMESPACE, 'message-control'): MsgControlElement,
(FT_EXT_NAMESPACE, 'create-index'): CreateIndexElement,
(FT_EXT_NAMESPACE, 'raw-text-output'): RawTextOutputElement,
(FT_EXT_NAMESPACE, 'chain-to'): ChainToElement,
(FT_EXT_NAMESPACE, 'uri-to-element'): UriToElementElement,
}
|