cmlenz@336: # -*- coding: utf-8 -*-
cmlenz@336: #
cmlenz@897: # Copyright (C) 2006-2010 Edgewall Software
cmlenz@336: # All rights reserved.
cmlenz@336: #
cmlenz@336: # This software is licensed as described in the file COPYING, which
cmlenz@336: # you should have received as part of this distribution. The terms
cmlenz@336: # are also available at http://genshi.edgewall.org/wiki/License.
cmlenz@336: #
cmlenz@336: # This software consists of voluntary contributions made by many
cmlenz@336: # individuals. For the exact contribution history, see the revision
cmlenz@336: # history and logs, available at http://genshi.edgewall.org/log/.
cmlenz@336: 
cmlenz@336: """Markup templating engine."""
cmlenz@336: 
cmlenz@336: from itertools import chain
cmlenz@336: 
cmlenz@636: from genshi.core import Attrs, Markup, Namespace, Stream, StreamEventKind
cmlenz@405: from genshi.core import START, END, START_NS, END_NS, TEXT, PI, COMMENT
cmlenz@336: from genshi.input import XMLParser
cmlenz@400: from genshi.template.base import BadDirectiveError, Template, \
cmlenz@475:                                  TemplateSyntaxError, _apply_directives, \
cmlenz@609:                                  EXEC, INCLUDE, SUB
cmlenz@405: from genshi.template.eval import Suite
cmlenz@407: from genshi.template.interpolation import interpolate
cmlenz@336: from genshi.template.directives import *
cmlenz@610: from genshi.template.text import NewTextTemplate
cmlenz@336: 
cmlenz@425: __all__ = ['MarkupTemplate']
cmlenz@425: __docformat__ = 'restructuredtext en'
cmlenz@425: 
cmlenz@336: 
cmlenz@336: class MarkupTemplate(Template):
cmlenz@336:     """Implementation of the template language for XML-based templates.
cmlenz@336:     
cmlenz@336:     >>> tmpl = MarkupTemplate('''<ul xmlns:py="http://genshi.edgewall.org/">
cmlenz@336:     ...   <li py:for="item in items">${item}</li>
cmlenz@336:     ... </ul>''')
cmlenz@853:     >>> print(tmpl.generate(items=[1, 2, 3]))
cmlenz@336:     <ul>
cmlenz@336:       <li>1</li><li>2</li><li>3</li>
cmlenz@336:     </ul>
cmlenz@336:     """
cmlenz@427: 
cmlenz@790:     DIRECTIVE_NAMESPACE = 'http://genshi.edgewall.org/'
cmlenz@790:     XINCLUDE_NAMESPACE = 'http://www.w3.org/2001/XInclude'
cmlenz@336: 
cmlenz@336:     directives = [('def', DefDirective),
cmlenz@336:                   ('match', MatchDirective),
cmlenz@336:                   ('when', WhenDirective),
cmlenz@336:                   ('otherwise', OtherwiseDirective),
cmlenz@336:                   ('for', ForDirective),
cmlenz@336:                   ('if', IfDirective),
cmlenz@336:                   ('choose', ChooseDirective),
cmlenz@336:                   ('with', WithDirective),
cmlenz@336:                   ('replace', ReplaceDirective),
cmlenz@336:                   ('content', ContentDirective),
cmlenz@336:                   ('attrs', AttrsDirective),
cmlenz@336:                   ('strip', StripDirective)]
cmlenz@605:     serializer = 'xml'
cmlenz@636:     _number_conv = Markup
cmlenz@336: 
cmlenz@790:     def __init__(self, source, filepath=None, filename=None, loader=None,
cmlenz@790:                  encoding=None, lookup='strict', allow_exec=True):
cmlenz@790:         Template.__init__(self, source, filepath=filepath, filename=filename,
cmlenz@790:                           loader=loader, encoding=encoding, lookup=lookup,
cmlenz@790:                           allow_exec=allow_exec)
cmlenz@790:         self.add_directives(self.DIRECTIVE_NAMESPACE, self)
cmlenz@790: 
cmlenz@715:     def _init_filters(self):
cmlenz@715:         Template._init_filters(self)
cmlenz@496:         # Make sure the include filter comes after the match filter
cmlenz@876:         self.filters.remove(self._include)
cmlenz@876:         self.filters += [self._match, self._include]
cmlenz@336: 
cmlenz@374:     def _parse(self, source, encoding):
cmlenz@374:         if not isinstance(source, Stream):
cmlenz@434:             source = XMLParser(source, filename=self.filename,
cmlenz@374:                                encoding=encoding)
cmlenz@790:         stream = []
cmlenz@374: 
cmlenz@374:         for kind, data, pos in source:
cmlenz@336: 
cmlenz@790:             if kind is TEXT:
cmlenz@790:                 for kind, data, pos in interpolate(data, self.filepath, pos[1],
cmlenz@790:                                                    pos[2], lookup=self.lookup):
cmlenz@336:                     stream.append((kind, data, pos))
cmlenz@336: 
cmlenz@790:             elif kind is PI and data[0] == 'python':
cmlenz@790:                 if not self.allow_exec:
cmlenz@790:                     raise TemplateSyntaxError('Python code blocks not allowed',
cmlenz@790:                                               self.filepath, *pos[1:])
cmlenz@790:                 try:
cmlenz@790:                     suite = Suite(data[1], self.filepath, pos[1],
cmlenz@790:                                   lookup=self.lookup)
cmlenz@790:                 except SyntaxError, err:
cmlenz@790:                     raise TemplateSyntaxError(err, self.filepath,
cmlenz@790:                                               pos[1] + (err.lineno or 1) - 1,
cmlenz@790:                                               pos[2] + (err.offset or 0))
cmlenz@790:                 stream.append((EXEC, suite, pos))
cmlenz@790: 
cmlenz@790:             elif kind is COMMENT:
cmlenz@790:                 if not data.lstrip().startswith('!'):
cmlenz@336:                     stream.append((kind, data, pos))
cmlenz@336: 
cmlenz@790:             else:
cmlenz@790:                 stream.append((kind, data, pos))
cmlenz@790: 
cmlenz@790:         return stream
cmlenz@790: 
cmlenz@790:     def _extract_directives(self, stream, namespace, factory):
cmlenz@790:         depth = 0
cmlenz@790:         dirmap = {} # temporary mapping of directives to elements
cmlenz@790:         new_stream = []
cmlenz@790:         ns_prefix = {} # namespace prefixes in use
cmlenz@790: 
cmlenz@790:         for kind, data, pos in stream:
cmlenz@790: 
cmlenz@790:             if kind is START:
cmlenz@363:                 tag, attrs = data
cmlenz@336:                 directives = []
cmlenz@336:                 strip = False
cmlenz@336: 
cmlenz@790:                 if tag.namespace == namespace:
cmlenz@790:                     cls = factory.get_directive(tag.localname)
cmlenz@336:                     if cls is None:
cmlenz@790:                         raise BadDirectiveError(tag.localname,
cmlenz@790:                                                 self.filepath, pos[1])
cmlenz@552:                     args = dict([(name.localname, value) for name, value
cmlenz@552:                                  in attrs if not name.namespace])
cmlenz@847:                     directives.append((factory.get_directive_index(cls), cls,
cmlenz@847:                                        args, ns_prefix.copy(), pos))
cmlenz@336:                     strip = True
cmlenz@336: 
cmlenz@363:                 new_attrs = []
cmlenz@363:                 for name, value in attrs:
cmlenz@790:                     if name.namespace == namespace:
cmlenz@790:                         cls = factory.get_directive(name.localname)
cmlenz@336:                         if cls is None:
cmlenz@336:                             raise BadDirectiveError(name.localname,
cmlenz@336:                                                     self.filepath, pos[1])
cmlenz@790:                         if type(value) is list and len(value) == 1:
cmlenz@790:                             value = value[0][1]
cmlenz@847:                         directives.append((factory.get_directive_index(cls),
cmlenz@847:                                            cls, value, ns_prefix.copy(), pos))
cmlenz@336:                     else:
cmlenz@363:                         new_attrs.append((name, value))
cmlenz@363:                 new_attrs = Attrs(new_attrs)
cmlenz@336: 
cmlenz@336:                 if directives:
cmlenz@847:                     directives.sort()
cmlenz@790:                     dirmap[(depth, tag)] = (directives, len(new_stream),
cmlenz@790:                                             strip)
cmlenz@336: 
cmlenz@790:                 new_stream.append((kind, (tag, new_attrs), pos))
cmlenz@790:                 depth += 1
cmlenz@790: 
cmlenz@790:             elif kind is END:
cmlenz@790:                 depth -= 1
cmlenz@790:                 new_stream.append((kind, data, pos))
cmlenz@790: 
cmlenz@790:                 # If there have have directive attributes with the
cmlenz@790:                 # corresponding start tag, move the events inbetween into
cmlenz@790:                 # a "subprogram"
cmlenz@790:                 if (depth, data) in dirmap:
cmlenz@790:                     directives, offset, strip = dirmap.pop((depth, data))
cmlenz@790:                     substream = new_stream[offset:]
cmlenz@790:                     if strip:
cmlenz@790:                         substream = substream[1:-1]
cmlenz@790:                     new_stream[offset:] = [
cmlenz@790:                         (SUB, (directives, substream), pos)
cmlenz@790:                     ]
cmlenz@790: 
cmlenz@790:             elif kind is SUB:
cmlenz@790:                 directives, substream = data
cmlenz@790:                 substream = self._extract_directives(substream, namespace,
cmlenz@790:                                                      factory)
cmlenz@790: 
cmlenz@790:                 if len(substream) == 1 and substream[0][0] is SUB:
cmlenz@790:                     added_directives, substream = substream[0][1]
cmlenz@790:                     directives += added_directives
cmlenz@790: 
cmlenz@790:                 new_stream.append((kind, (directives, substream), pos))
cmlenz@790: 
cmlenz@790:             elif kind is START_NS:
cmlenz@790:                 # Strip out the namespace declaration for template
cmlenz@790:                 # directives
cmlenz@790:                 prefix, uri = data
cmlenz@790:                 ns_prefix[prefix] = uri
cmlenz@790:                 if uri != namespace:
cmlenz@790:                     new_stream.append((kind, data, pos))
cmlenz@790: 
cmlenz@790:             elif kind is END_NS:
cmlenz@790:                 uri = ns_prefix.pop(data, None)
cmlenz@790:                 if uri and uri != namespace:
cmlenz@790:                     new_stream.append((kind, data, pos))
cmlenz@790: 
cmlenz@790:             else:
cmlenz@790:                 new_stream.append((kind, data, pos))
cmlenz@790: 
cmlenz@790:         return new_stream
cmlenz@790: 
cmlenz@790:     def _extract_includes(self, stream):
cmlenz@790:         streams = [[]] # stacked lists of events of the "compiled" template
cmlenz@790:         prefixes = {}
cmlenz@790:         fallbacks = []
cmlenz@790:         includes = []
cmlenz@790:         xinclude_ns = Namespace(self.XINCLUDE_NAMESPACE)
cmlenz@790: 
cmlenz@790:         for kind, data, pos in stream:
cmlenz@790:             stream = streams[-1]
cmlenz@790: 
cmlenz@790:             if kind is START:
cmlenz@790:                 # Record any directive attributes in start tags
cmlenz@790:                 tag, attrs = data
cmlenz@790:                 if tag in xinclude_ns:
cmlenz@363:                     if tag.localname == 'include':
cmlenz@790:                         include_href = attrs.get('href')
cmlenz@363:                         if not include_href:
cmlenz@363:                             raise TemplateSyntaxError('Include misses required '
cmlenz@422:                                                       'attribute "href"',
cmlenz@422:                                                       self.filepath, *pos[1:])
cmlenz@790:                         includes.append((include_href, attrs.get('parse')))
cmlenz@381:                         streams.append([])
cmlenz@363:                     elif tag.localname == 'fallback':
cmlenz@590:                         streams.append([])
cmlenz@590:                         fallbacks.append(streams[-1])
cmlenz@363:                 else:
cmlenz@790:                     stream.append((kind, (tag, attrs), pos))
cmlenz@336: 
cmlenz@336:             elif kind is END:
cmlenz@790:                 if fallbacks and data == xinclude_ns['fallback']:
cmlenz@590:                     assert streams.pop() is fallbacks[-1]
cmlenz@790:                 elif data == xinclude_ns['include']:
cmlenz@590:                     fallback = None
cmlenz@590:                     if len(fallbacks) == len(includes):
cmlenz@590:                         fallback = fallbacks.pop()
cmlenz@590:                     streams.pop() # discard anything between the include tags
cmlenz@590:                                   # and the fallback element
cmlenz@381:                     stream = streams[-1]
cmlenz@610:                     href, parse = includes.pop()
cmlenz@610:                     try:
cmlenz@610:                         cls = {
cmlenz@610:                             'xml': MarkupTemplate,
cmlenz@610:                             'text': NewTextTemplate
cmlenz@880:                         }.get(parse) or self.__class__
cmlenz@610:                     except KeyError:
cmlenz@610:                         raise TemplateSyntaxError('Invalid value for "parse" '
cmlenz@610:                                                   'attribute of include',
cmlenz@610:                                                   self.filepath, *pos[1:])
cmlenz@610:                     stream.append((INCLUDE, (href, cls, fallback), pos))
cmlenz@363:                 else:
cmlenz@363:                     stream.append((kind, data, pos))
cmlenz@336: 
cmlenz@790:             elif kind is START_NS and data[1] == xinclude_ns:
cmlenz@790:                 # Strip out the XInclude namespace
cmlenz@790:                 prefixes[data[0]] = data[1]
cmlenz@336: 
cmlenz@790:             elif kind is END_NS and data in prefixes:
cmlenz@790:                 prefixes.pop(data)
cmlenz@336: 
cmlenz@336:             else:
cmlenz@336:                 stream.append((kind, data, pos))
cmlenz@336: 
cmlenz@381:         assert len(streams) == 1
cmlenz@381:         return streams[0]
cmlenz@381: 
cmlenz@790:     def _interpolate_attrs(self, stream):
cmlenz@790:         for kind, data, pos in stream:
cmlenz@790: 
cmlenz@790:             if kind is START:
cmlenz@790:                 # Record any directive attributes in start tags
cmlenz@790:                 tag, attrs = data
cmlenz@790:                 new_attrs = []
cmlenz@790:                 for name, value in attrs:
cmlenz@790:                     if value:
cmlenz@790:                         value = list(interpolate(value, self.filepath, pos[1],
cmlenz@790:                                                  pos[2], lookup=self.lookup))
cmlenz@790:                         if len(value) == 1 and value[0][0] is TEXT:
cmlenz@790:                             value = value[0][1]
cmlenz@790:                     new_attrs.append((name, value))
cmlenz@790:                 data = tag, Attrs(new_attrs)
cmlenz@790: 
cmlenz@790:             yield kind, data, pos
cmlenz@790: 
cmlenz@790:     def _prepare(self, stream):
cmlenz@790:         return Template._prepare(self,
cmlenz@790:             self._extract_includes(self._interpolate_attrs(stream))
cmlenz@790:         )
cmlenz@790: 
cmlenz@790:     def add_directives(self, namespace, factory):
cmlenz@790:         """Register a custom `DirectiveFactory` for a given namespace.
cmlenz@790:         
cmlenz@790:         :param namespace: the namespace URI
cmlenz@790:         :type namespace: `basestring`
cmlenz@790:         :param factory: the directive factory to register
cmlenz@790:         :type factory: `DirectiveFactory`
cmlenz@790:         :since: version 0.6
cmlenz@790:         """
cmlenz@790:         assert not self._prepared, 'Too late for adding directives, ' \
cmlenz@790:                                    'template already prepared'
cmlenz@790:         self._stream = self._extract_directives(self._stream, namespace,
cmlenz@790:                                                 factory)
cmlenz@790: 
cmlenz@766:     def _match(self, stream, ctxt, start=0, end=None, **vars):
cmlenz@336:         """Internal stream filter that applies any defined match templates
cmlenz@336:         to the stream.
cmlenz@336:         """
cmlenz@766:         match_templates = ctxt._match_templates
cmlenz@336: 
hodgestar@921:         def _strip(stream, append):
cmlenz@336:             depth = 1
cmlenz@843:             next = stream.next
cmlenz@336:             while 1:
cmlenz@843:                 event = next()
cmlenz@336:                 if event[0] is START:
cmlenz@336:                     depth += 1
cmlenz@336:                 elif event[0] is END:
cmlenz@336:                     depth -= 1
cmlenz@336:                 if depth > 0:
cmlenz@336:                     yield event
cmlenz@336:                 else:
cmlenz@843:                     append(event)
cmlenz@336:                     break
cmlenz@336: 
cmlenz@336:         for event in stream:
cmlenz@336: 
cmlenz@809:             # We (currently) only care about start and end events for matching
cmlenz@336:             # We might care about namespace events in the future, though
cmlenz@809:             if not match_templates or (event[0] is not START and
cmlenz@809:                                        event[0] is not END):
cmlenz@336:                 yield event
cmlenz@336:                 continue
cmlenz@336: 
cmlenz@602:             for idx, (test, path, template, hints, namespaces, directives) \
cmlenz@602:                     in enumerate(match_templates):
cmlenz@766:                 if idx < start or end is not None and idx >= end:
cmlenz@758:                     continue
cmlenz@336: 
cmlenz@336:                 if test(event, namespaces, ctxt) is True:
cmlenz@602:                     if 'match_once' in hints:
cmlenz@602:                         del match_templates[idx]
cmlenz@602:                         idx -= 1
cmlenz@336: 
cmlenz@336:                     # Let the remaining match templates know about the event so
cmlenz@336:                     # they get a chance to update their internal state
cmlenz@336:                     for test in [mt[0] for mt in match_templates[idx + 1:]]:
cmlenz@336:                         test(event, namespaces, ctxt, updateonly=True)
cmlenz@336: 
cmlenz@336:                     # Consume and store all events until an end event
cmlenz@336:                     # corresponding to this start event is encountered
cmlenz@766:                     pre_end = idx + 1
cmlenz@694:                     if 'match_once' not in hints and 'not_recursive' in hints:
cmlenz@766:                         pre_end -= 1
hodgestar@921:                     tail = []
hodgestar@921:                     inner = _strip(stream, tail.append)
cmlenz@766:                     if pre_end > 0:
cmlenz@870:                         inner = self._match(inner, ctxt, start=start,
cmlenz@870:                                             end=pre_end, **vars)
cmlenz@700:                     content = self._include(chain([event], inner, tail), ctxt)
cmlenz@700:                     if 'not_buffered' not in hints:
cmlenz@700:                         content = list(content)
cmlenz@843:                     content = Stream(content)
cmlenz@336: 
cmlenz@336:                     # Make the select() function available in the body of the
cmlenz@336:                     # match template
cmlenz@771:                     selected = [False]
cmlenz@336:                     def select(path):
cmlenz@771:                         selected[0] = True
cmlenz@843:                         return content.select(path, namespaces, ctxt)
cmlenz@700:                     vars = dict(select=select)
cmlenz@336: 
cmlenz@336:                     # Recursively process the output
cmlenz@700:                     template = _apply_directives(template, directives, ctxt,
cmlenz@827:                                                  vars)
cmlenz@813:                     for event in self._match(self._flatten(template, ctxt,
cmlenz@813:                                                            **vars),
cmlenz@813:                                              ctxt, start=idx + 1, **vars):
cmlenz@336:                         yield event
cmlenz@336: 
cmlenz@771:                     # If the match template did not actually call select to
cmlenz@771:                     # consume the matched stream, the original events need to
cmlenz@771:                     # be consumed here or they'll get appended to the output
cmlenz@771:                     if not selected[0]:
cmlenz@771:                         for event in content:
cmlenz@771:                             pass
cmlenz@771: 
cmlenz@771:                     # Let the remaining match templates know about the last
cmlenz@771:                     # event in the matched content, so they can update their
cmlenz@771:                     # internal state accordingly
cmlenz@874:                     for test in [mt[0] for mt in match_templates[idx + 1:]]:
cmlenz@771:                         test(tail[0], namespaces, ctxt, updateonly=True)
cmlenz@771: 
cmlenz@336:                     break
cmlenz@336: 
cmlenz@336:             else: # no matches
cmlenz@336:                 yield event