gistfile1.py

class MyTemplateMaker(TemplateMaker):
  def annotate_fragment_attribute(self, index, field, attribute='content'):
		for f in self.htmlpage.parsed_body[index::-1]:
			if isinstance(f, HtmlTag) and f.tag_type in [HtmlTagType.OPEN_TAG, HtmlTagType.UNPAIRED_TAG]:
				if 'data-scrapy-annotate' in f.attributes:
					fstr = self.htmlpage.fragment_data(f)
					raise FragmentAlreadyAnnotated("Fragment already annotated: %s" % fstr)
				d = {'annotations': {attribute or "content": field}}
				a = ' data-scrapy-annotate="%s"' % json.dumps(d).replace('"', '"')
				p = self.htmlpage
				p.body = p.body[:f.end-1] + a + p.body[f.end-1:]
				break

	def annotate_attribute(self, field, score_func, best_match=True, attribute=None):
		"""Annotate a field.

		``score_func`` is a callable that receives two arguments: (fragment,
		htmlpage) and returns a relevancy score (float) indicating how relevant
		is the fragment. 0 means the fragment is irrelevant. Higher scores
		means the fragment is more relevant. Otherwise, the closest opening tag
		(to the left) is annotated with the given attribute.

		If ``best_match`` is ``True``, only the best fragment is annotated.
		Otherwise, all fragments (with a positive relevancy) are annotated.

		"""
		indexes = self.select(score_func)
		if not indexes:
			raise FragmentNotFound("Fragment not found annotating %r using: %s" %
			(field, score_func))
		if best_match:
			del indexes[1:]
		for i in indexes:
			self.annotate_fragment_attribute(i, field, attribute)