diff --git a/AUTHORS b/AUTHORS index 245eafd..f72b6a6 100644 --- a/AUTHORS +++ b/AUTHORS @@ -57,7 +57,8 @@ Julien Phalip Justin Lilly Luis Nell Lukas Lehner -Lukasz Balcerzak +Łukasz Balcerzak +Łukasz Langa Maciek Szczesniak Maor Ben-Dayan Mark Lavin @@ -89,4 +90,4 @@ Ulrich Petri Ulysses V Vladislav Poluhin wesleyb -Wilson Júnior \ No newline at end of file +Wilson Júnior diff --git a/Makefile b/Makefile index 598b837..ae5eec1 100644 --- a/Makefile +++ b/Makefile @@ -2,3 +2,5 @@ test: flake8 compressor --ignore=E501,E128 coverage run --branch --source=compressor `which django-admin.py` test --settings=compressor.test_settings compressor coverage report --omit=compressor/test*,compressor/filters/jsmin/rjsmin*,compressor/filters/cssmin/cssmin*,compressor/utils/stringformat* + +.PHONY: test diff --git a/compressor/parser/html5lib.py b/compressor/parser/html5lib.py index 7fee590..c0c83ef 100644 --- a/compressor/parser/html5lib.py +++ b/compressor/parser/html5lib.py @@ -15,39 +15,45 @@ class Html5LibParser(ParserBase): self.html5lib = html5lib def _serialize(self, elem): - fragment = self.html5lib.treebuilders.simpletree.DocumentFragment() - fragment.appendChild(elem) - return self.html5lib.serialize(fragment, - quote_attr_values=True, omit_optional_tags=False) + return self.html5lib.serialize( + elem, tree="etree", quote_attr_values=True, + omit_optional_tags=False, use_trailing_solidus=True, + ) def _find(self, *names): - for node in self.html.childNodes: - if node.type == 5 and node.name in names: - yield node + for elem in self.html: + if elem.tag in names: + yield elem @cached_property def html(self): try: - return self.html5lib.parseFragment(self.content) - except ImportError, err: + return self.html5lib.parseFragment( + self.content, + treebuilder="etree", + ) + except ImportError as err: raise ImproperlyConfigured("Error while importing html5lib: %s" % err) - except Exception, err: + except Exception as err: raise ParserError("Error while initializing Parser: %s" % err) def css_elems(self): - return self._find('style', 'link') + return self._find('{http://www.w3.org/1999/xhtml}link', + '{http://www.w3.org/1999/xhtml}style') def js_elems(self): - return self._find('script') + return self._find('{http://www.w3.org/1999/xhtml}script') def elem_attribs(self, elem): - return elem.attributes + return elem.attrib def elem_content(self, elem): - return elem.childNodes[0].value + return smart_unicode(elem.text) def elem_name(self, elem): - return elem.name + if '}' in elem.tag: + return elem.tag.split('}')[1] + return elem.tag def elem_str(self, elem): # This method serializes HTML in a way that does not pass all tests. diff --git a/compressor/tests/test_parsers.py b/compressor/tests/test_parsers.py index 04ec924..89950a0 100644 --- a/compressor/tests/test_parsers.py +++ b/compressor/tests/test_parsers.py @@ -20,12 +20,10 @@ except ImportError: from compressor.base import SOURCE_HUNK, SOURCE_FILE from compressor.conf import settings -from compressor.css import CssCompressor from compressor.tests.test_base import CompressorTestCase class ParserTestCase(object): - def setUp(self): self.old_parser = settings.COMPRESS_PARSER settings.COMPRESS_PARSER = self.parser_cls @@ -42,34 +40,86 @@ LxmlParserTests = skipIf(lxml is None, 'lxml not found')(LxmlParserTests) class Html5LibParserTests(ParserTestCase, CompressorTestCase): parser_cls = 'compressor.parser.Html5LibParser' - - def setUp(self): - super(Html5LibParserTests, self).setUp() - # special version of the css since the parser sucks - self.css = """\ - - -""" - self.css_node = CssCompressor(self.css) + # Special test variants required since xml.etree holds attributes + # as a plain dictionary, e.g. key order is unpredictable. def test_css_split(self): - out = [ - (SOURCE_FILE, os.path.join(settings.COMPRESS_ROOT, u'css', u'one.css'), u'css/one.css', u''), - (SOURCE_HUNK, u'p { border:5px solid green;}', None, u''), - (SOURCE_FILE, os.path.join(settings.COMPRESS_ROOT, u'css', u'two.css'), u'css/two.css', u''), - ] split = self.css_node.split_contents() - split = [(x[0], x[1], x[2], self.css_node.parser.elem_str(x[3])) for x in split] - self.assertEqual(out, split) + out0 = ( + SOURCE_FILE, + os.path.join(settings.COMPRESS_ROOT, u'css', u'one.css'), + u'css/one.css', + u'{http://www.w3.org/1999/xhtml}link', + {u'rel': u'stylesheet', u'href': u'/static/css/one.css', + u'type': u'text/css'}, + ) + self.assertEqual(out0, split[0][:3] + (split[0][3].tag, + split[0][3].attrib)) + out1 = ( + SOURCE_HUNK, + u'p { border:5px solid green;}', + None, + u'', + ) + self.assertEqual(out1, split[1][:3] + + (self.css_node.parser.elem_str(split[1][3]),)) + out2 = ( + SOURCE_FILE, + os.path.join(settings.COMPRESS_ROOT, u'css', u'two.css'), + u'css/two.css', + u'{http://www.w3.org/1999/xhtml}link', + {u'rel': u'stylesheet', u'href': u'/static/css/two.css', + u'type': u'text/css'}, + ) + self.assertEqual(out2, split[2][:3] + (split[2][3].tag, + split[2][3].attrib)) def test_js_split(self): - out = [ - (SOURCE_FILE, os.path.join(settings.COMPRESS_ROOT, u'js', u'one.js'), u'js/one.js', u''), - (SOURCE_HUNK, u'obj.value = "value";', None, u''), - ] split = self.js_node.split_contents() - split = [(x[0], x[1], x[2], self.js_node.parser.elem_str(x[3])) for x in split] - self.assertEqual(out, split) + out0 = ( + SOURCE_FILE, + os.path.join(settings.COMPRESS_ROOT, u'js', u'one.js'), + u'js/one.js', + u'{http://www.w3.org/1999/xhtml}script', + {u'src': u'/static/js/one.js', u'type': u'text/javascript'}, + None, + ) + self.assertEqual(out0, split[0][:3] + (split[0][3].tag, + split[0][3].attrib, + split[0][3].text)) + out1 = ( + SOURCE_HUNK, + u'obj.value = "value";', + None, + u'{http://www.w3.org/1999/xhtml}script', + {u'type': u'text/javascript'}, + u'obj.value = "value";', + ) + self.assertEqual(out1, split[1][:3] + (split[1][3].tag, + split[1][3].attrib, + split[1][3].text)) + + def test_css_return_if_off(self): + settings.COMPRESS_ENABLED = False + # Yes, they are semantically equal but attributes might be + # scrambled in unpredictable order. A more elaborate check + # would require parsing both arguments with a different parser + # and then evaluating the result, which no longer is + # a meaningful unit test. + self.assertEqual(len(self.css), len(self.css_node.output())) + + def test_js_return_if_off(self): + try: + enabled = settings.COMPRESS_ENABLED + precompilers = settings.COMPRESS_PRECOMPILERS + settings.COMPRESS_ENABLED = False + settings.COMPRESS_PRECOMPILERS = {} + # As above. + self.assertEqual(len(self.js), len(self.js_node.output())) + finally: + settings.COMPRESS_ENABLED = enabled + settings.COMPRESS_PRECOMPILERS = precompilers + Html5LibParserTests = skipIf( html5lib is None, 'html5lib not found')(Html5LibParserTests)