terom@52: """
terom@61:     Generating XHTML output
terom@61: 
terom@61:     XXX: use a 'real' XML builder?
terom@52: """
terom@52: 
terom@52: from cgi import escape
terom@80: import itertools as _itertools, types as _types
terom@52: 
terom@53: class IRenderable (object) :
terom@53:     """
terom@53:         Something that's renderable as the contents of a HTML tag
terom@53:     """
terom@52: 
terom@53:     def render_raw_lines (self, indent=u'\t') :
terom@53:         """
terom@53:             Render the indented lines for tag and contents, without newlines
terom@53:         """
terom@53: 
terom@53:         abstract
terom@53: 
terom@61:     def render_lines (self, indent=u'\t', newline=u'\n') :
terom@61:         """
terom@61:             Render full output lines with given newlines
terom@61: 
terom@61:             >>> list(Tag('xx', 'yy').render_lines())
terom@61:             [u'<xx>\\n', u'\\tyy\\n', u'</xx>\\n']
terom@61:         """
terom@61: 
terom@77:         for line in self.render_raw_lines(indent=indent) :
terom@61:             yield line + newline
terom@61: 
terom@61:     def render_unicode (self, **render_opts) :
terom@61:         """
terom@61:             Render full tag as a single unicode string
terom@61: 
terom@61:             >>> Tag('xx', 'yy').render_unicode()
terom@61:             u'<xx>\\n\\tyy\\n</xx>\\n'
terom@61:         """
terom@61: 
terom@61:         return "".join(self.render_lines(**render_opts))
terom@61: 
terom@61:     def render_str (self, encoding='ascii', **render_opts) :
terom@61:         """
terom@61:             Render full tag as an encoded string
terom@61: 
terom@61:             >>> Tag('xx', 'yy').render_str()
terom@61:             '<xx>\\n\\tyy\\n</xx>\\n'
terom@61:         """
terom@61: 
terom@61:         return self.render_unicode(**render_opts).encode(encoding)
terom@61: 
terom@69:     def render_out (self, stream, encoding=None, **render_opts) :
terom@61:         """
terom@69:             Render output into the given stream, encoding using the given encoding if given.
terom@61: 
terom@61:             >>> from StringIO import StringIO; buf = StringIO(); Tag('xx', 'yy').render_out(buf, 'ascii'); buf.getvalue()
terom@61:             '<xx>\\n\\tyy\\n</xx>\\n'
terom@61:         """
terom@61: 
terom@61:         for line in self.render_lines(**render_opts) :
terom@69:             if encoding :
terom@69:                 line = line.encode(encoding)
terom@69: 
terom@69:             stream.write(line)
terom@61:     
terom@69:     def render_file (self, file, encoding=None, **render_opts) :
terom@69:         """
terom@144:             Render output to given File, overwriteing anything already there
terom@69:         """
terom@69: 
terom@77:         self.render_out(file.open_write(encoding), **render_opts)
terom@69: 
terom@61:     # default output
terom@61:     __str__ = render_str
terom@61:     __unicode__ = render_unicode
terom@61: 
terom@61:     # default .render method
terom@61:     render = render_unicode
terom@61: 
terom@78: class Container (IRenderable) :
terom@78:     """
terom@78:         A container holds a sequence of other renderable items.
terom@78:     """
terom@78: 
terom@78:     @classmethod
terom@78:     def process_contents (cls, contents) :
terom@78:         """
terom@78:             Postprocess contents iterable to return new list.
terom@78: 
terom@78:             Items that are None will be omitted from the return value.
terom@78: 
terom@80:             Certain core sequence types will be recognized and flattened for output: tuples, lists, and generators.
terom@80: 
terom@78:             >>> list(Container.process_contents([]))
terom@78:             []
terom@78:             >>> list(Container.process_contents([None]))
terom@78:             []
terom@78:             >>> list(Container.process_contents([u'foo']))
terom@78:             [u'foo']
terom@78:         """
terom@78:         
terom@78:         for content in contents :
terom@78:             if content is None :
terom@78:                 continue
terom@80:             
terom@80:             # Hardcoded list of special-case nested contents
terom@80:             elif isinstance(content, (_types.TupleType, _types.ListType, _types.GeneratorType)) :
terom@80:                 for subcontent in cls.process_contents(content) :
terom@80:                     yield subcontent
terom@80: 
terom@78:             else :
terom@80:                 # normal, handle as IRenderable/unicode data
terom@78:                 yield content
terom@78: 
terom@78:     def __init__ (self, *contents) :
terom@78:         """
terom@78:             Construct this container with the given sub-items
terom@78:         """
terom@78:         
terom@78:         # store postprocessed
terom@78:         self.contents = list(self.process_contents(contents))
terom@78:     
terom@78:     def render_raw_lines (self, **render_opts) :
terom@78:         """
terom@78:             Render our contents as a series of non-indented lines, with the contents handling indentation themselves.
terom@78: 
terom@78:             >>> list(Container(5).render_raw_lines())
terom@78:             [u'5']
terom@78:             >>> list(Container('line1', 'line2').render_raw_lines())
terom@78:             [u'line1', u'line2']
terom@78:             >>> list(Container('a', Tag('b', 'bb'), 'c').render_raw_lines())
terom@78:             [u'a', u'<b>', u'\\tbb', u'</b>', u'c']
terom@78:             >>> list(Container(Tag('hr'), Tag('foo')('bar')).render_raw_lines())
terom@78:             [u'<hr />', u'<foo>', u'\\tbar', u'</foo>']
terom@78:         """
terom@78: 
terom@78:         for content in self.contents :
terom@78:             if isinstance(content, IRenderable) :
terom@78:                 # sub-items
terom@78:                 for line in content.render_raw_lines(**render_opts) :
terom@78:                     yield line
terom@78:             
terom@78:             else :
terom@78:                 # escape raw values
terom@78:                 yield escape(unicode(content))
terom@78: 
terom@78:     def __repr__ (self) :
terom@78:         return 'Container(%s)' % ', '.join(repr(c) for c in self.contents)
terom@78: 
terom@78: class Tag (Container) :
terom@53:     """
terom@53:         A HTML tag, with attributes and contents, which can a mixture of data and other renderables(tags).
terom@53: 
terom@53:         Provides various kinds of rendering output
terom@53:     """
terom@53: 
terom@53:     @staticmethod
terom@53:     def process_attrs (attrs) :
terom@53:         """
terom@53:             Postprocess attributes.
terom@53: 
terom@53:             Key-value pairs where the value is None will be ommitted, and any trailing underscores in the key omitted.
terom@53: 
terom@53:             TODO: only remove one underscore
terom@53: 
terom@78:             >>> dict(Tag.process_attrs(dict()))
terom@53:             {}
terom@78:             >>> dict(Tag.process_attrs(dict(foo='bar')))
terom@53:             {'foo': 'bar'}
terom@78:             >>> dict(Tag.process_attrs(dict(class_='bar', frob=None)))
terom@53:             {'class': 'bar'}
terom@53:         """
terom@53: 
terom@78:         return ((k.rstrip('_'), v) for k, v in attrs.iteritems() if v is not None)
terom@53: 
terom@53:     def __init__ (self, name, *contents, **attrs) :
terom@53:         """
terom@53:             Construct tag with given name/attributes or contents.
terom@53: 
terom@53:             The filtering rules desribed in process_contents/process_attrs apply.
terom@53: 
terom@53:             >>> Tag('foo')
terom@53:             Tag('foo')
terom@53:             >>> Tag('foo', 'quux')
terom@53:             Tag('foo', 'quux')
terom@53:             >>> Tag('foo', 'quux', bar=5)
terom@53:             Tag('foo', 'quux', bar=5)
terom@53:             >>> Tag('foo', class_='ten')
terom@53:             Tag('foo', class='ten')
terom@53:         """
terom@78:         
terom@78:         # store contents as container
terom@78:         super(Tag, self).__init__(*contents)
terom@78:         
terom@78:         # store postprocessed stuff
terom@53:         self.name = name
terom@78:         self.attrs = dict(self.process_attrs(attrs))
terom@53: 
terom@53:     def __call__ (self, *contents, **attrs) :
terom@53:         """
terom@53:             Return a new Tag with this tag's attributes and contents, as well as the given attributes/contents.
terom@53: 
terom@53:             The filtering rules desribed in process_contents/process_attrs apply.
terom@53: 
terom@53:             >>> Tag('foo')('bar')
terom@53:             Tag('foo', 'bar')
terom@53:             >>> Tag('a', href='index.html')("Home")
terom@53:             Tag('a', 'Home', href='index.html')
terom@53:             >>> Tag('bar', None)(5, foo=None, class_='bar')
terom@53:             Tag('bar', 5, class='bar')
terom@53:             >>> Tag('a')('b')('c')(asdf=5)
terom@53:             Tag('a', 'b', 'c', asdf=5)
terom@53:             >>> t1 = Tag('a'); t2 = t1('b'); t1
terom@53:             Tag('a')
terom@53:         """
terom@53: 
terom@53:         # merge attrs/contents
terom@53:         # XXX: new_attrs is not an iterator...
terom@53:         new_attrs = dict(_itertools.chain(self.attrs.iteritems(), attrs.iteritems()))
terom@53:         new_contents = _itertools.chain(self.contents, contents)
terom@53:         
terom@53:         # build new tag
terom@53:         return Tag(self.name, *new_contents, **new_attrs)
terom@53: 
terom@53:     @staticmethod
terom@53:     def format_attr (name, value) :
terom@53:         """
terom@53:             Format a single HTML tag attribute
terom@53: 
terom@53:             >>> Tag.format_attr('name', 'value')
terom@53:             u'name="value"'
terom@53:             >>> Tag.format_attr('this', '<a"b>')
terom@53:             u'this="&lt;a&quot;b&gt;"'
terom@53:             >>> Tag.format_attr('xx', 1337)
terom@53:             u'xx="1337"'
terom@53:         """
terom@53: 
terom@53:         return u'%s="%s"' % (name, escape(unicode(value), True))
terom@53:  
terom@53:     def render_attrs (self) :
terom@53:         """
terom@53:             Return the HTML attributes string
terom@53: 
terom@53:             >>> Tag('x', foo=5, bar='<').render_attrs()
terom@53:             u'foo="5" bar="&lt;"'
terom@53:         """
terom@53: 
terom@53:         return " ".join(self.format_attr(n, v) for n, v in self.attrs.iteritems())
terom@53: 
terom@53:     def render_raw_lines (self, indent=u'\t') :
terom@53:         """
terom@53:             Render the tag and indented content
terom@53: 
terom@53:             >>> list(Tag('xx', 'yy', zz='foo').render_raw_lines(indent=' '))
terom@53:             [u'<xx zz="foo">', u' yy', u'</xx>']
terom@53:         """
terom@53: 
terom@53:         # render attr string, including preceding space
terom@53:         attrs_stuff = (" " + self.render_attrs()) if self.attrs else ""
terom@53: 
terom@53:         if self.contents :
terom@78:             # wrapping tags
terom@53:             yield u"<%s%s>" % (self.name, attrs_stuff)
terom@78:             
terom@78:             # subcontents
terom@78:             for line in super(Tag, self).render_raw_lines(indent=indent) :
terom@53:                 yield indent + line
terom@53: 
terom@53:             yield u"</%s>" % (self.name, )
terom@53: 
terom@53:         else :
terom@53:             # singleton tag
terom@53:             yield u"<%s%s />" % (self.name, attrs_stuff)
terom@52:     
terom@52: 
terom@53:     def __repr__ (self) :
terom@53:         return 'Tag(%s)' % ', '.join(
terom@53:             [
terom@53:                 repr(self.name)
terom@53:             ] + [
terom@53:                 repr(c) for c in self.contents
terom@53:             ] + [
terom@53:                 '%s=%r' % (name, value) for name, value in self.attrs.iteritems()
terom@53:             ]
terom@53:         )
terom@52: 
terom@53: class Text (IRenderable) :
terom@53:     """
terom@53:         Raw HTML text
terom@53:     """
terom@53: 
terom@53:     def __init__ (self, line) :
terom@53:         """
terom@53:             Initialize to render as the given lines
terom@53:         """
terom@53: 
terom@53:         self.lines = [line]
terom@53:     
terom@53:     def render_raw_lines (self, indent=u'\t') :
terom@53:         return self.lines
terom@52: 
terom@61: class XHTMLDocument (IRenderable) :
terom@61:     """
terom@78:         A full XHTML document with XML header, doctype, head and body.
terom@78: 
terom@78:         XXX: current rendering is a bit of a kludge
terom@78: 
terom@61:         <?xml version="..." encoding="..." ?>
terom@61:         <!DOCTYPE ...>
terom@61:         
terom@61:         <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
terom@61:             <head>
terom@61:                 ...
terom@61:             </head>
terom@61:             <body>
terom@61:                 ...
terom@61:             </body>
terom@61:         </html>
terom@61:     """
terom@61: 
terom@61:     def __init__ (self, 
terom@61:         head, body,
terom@61:         xml_version='1.0', xml_encoding='utf-8', 
terom@61:         doctype='html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd"',
terom@61:         html_xmlns='http://www.w3.org/1999/xhtml', html_lang='en'
terom@61:     ) :
terom@61:         # store
terom@61:         self.xml_version = xml_version
terom@61:         self.xml_encoding = xml_encoding
terom@61:         self.doctype = doctype
terom@61:         
terom@61:         # build the document
terom@61:         self.document = Tag('html', **{'xmlns': html_xmlns, 'xml:lang': html_lang})(
terom@61:             Tag('head', head),
terom@61:             Tag('body', body),
terom@61:         )
terom@61: 
terom@61:     def render_raw_lines (self, **render_opts) :
terom@61:         """
terom@61:             Render the two header lines, and then the document
terom@61:         """
terom@61: 
terom@61:         yield '<?xml version="%s" encoding="%s" ?>' % (self.xml_version, self.xml_encoding)
terom@61:         yield '<!DOCTYPE %s>' % (self.doctype)
terom@61: 
terom@61:         for line in self.document.render_raw_lines(**render_opts) :
terom@61:             yield line
terom@61:     
terom@61:     def _check_encoding (self, encoding) :
terom@61:         if encoding and encoding != self.xml_encoding :
terom@76:             raise ValueError("encoding mismatch: %r should be %r" % (encoding, self.xml_encoding))
terom@61: 
terom@61:     def render_str (self, encoding=None, **render_opts) :
terom@61:         """
terom@61:             Wrap render_str to verify that the right encoding is used
terom@61:         """
terom@61: 
terom@61:         self._check_encoding(encoding)
terom@61:         
terom@61:         return super(XHTMLDocument, self).render_str(self.xml_encoding, **render_opts)
terom@61: 
terom@61:     def render_out (self, stream, encoding=None, **render_opts) :
terom@61:         """
terom@61:             Wrap render_out to verify that the right encoding is used
terom@61:         """
terom@61: 
terom@61:         self._check_encoding(encoding)
terom@61:         
terom@61:         return super(XHTMLDocument, self).render_out(stream, self.xml_encoding, **render_opts)
terom@61: 
terom@52: class TagFactory (object) :
terom@52:     """
terom@53:         Build Tags with names give as attribute names
terom@52:     """
terom@52: 
terom@53:     def __getattr__ (self, name) :
terom@53:         """
terom@53:             Get a Tag object with the given name, but no contents
terom@52: 
terom@53:             >>> TagFactory().a(href='bar')('quux')
terom@53:             Tag('a', 'quux', href='bar')
terom@53:         """
terom@53: 
terom@53:         return Tag(name)
terom@53: 
terom@53: # pretty names
terom@78: container = Container
terom@53: tag = Tag
terom@53: tags = TagFactory()
terom@53: raw = Text
terom@53: 
terom@53: # testing
terom@53: if __name__ == '__main__' :
terom@53:     import doctest
terom@53: 
terom@53:     doctest.testmod()
terom@53: