Django haystack prepare() method using proposed backend.extract() for rich-content handling

content_extraction_index.py

class MyIndex(SearchIndex):
    text = fields.CharField(document=True, use_template=False)

    def prepare(self, obj):
        data = super(MyIndex, self).prepare(obj)

        extracted_data = self.backend.extract(open(obj.file_field.path, "rb"))

        if extracted_data is not None:
            for k, v in extracted_data['metadata'].items():
                data["attr_%s" % k] = k
        else:
            self.log.warning("Metadata extraction failed for %s", obj)

        # Now we'll allow human-entered data to overwrite any conflicting
        # values:
        for k, v in obj.attributes.items():
            data["attr_%s" % k] = v

        # Now we'll finally perform the template processing to render the
        # text field with *all* of our metadata visible for templating:
        t = loader.select_template(('search/indexes/myapp/mymodel_text.txt',))
        data['text'] = t.render(Context({'object': obj,
                                            'extracted': extracted_data}))

        return data