"""Tests for HTMLParser.py.""" import html.parser import pprint import unittest from test import support class EventCollector(html.parser.HTMLParser): def __init__(self, *args, **kw): self.events = [] self.append = self.events.append html.parser.HTMLParser.__init__(self, *args, **kw) def get_events(self): # Normalize the list of events so that buffer artefacts don't # separate runs of contiguous characters. L = [] prevtype = None for event in self.events: type = event[0] if type == prevtype == "data": L[-1] = ("data", L[-1][1] + event[1]) else: L.append(event) prevtype = type self.events = L return L # structure markup def handle_starttag(self, tag, attrs): self.append(("starttag", tag, attrs)) def handle_startendtag(self, tag, attrs): self.append(("startendtag", tag, attrs)) def handle_endtag(self, tag): self.append(("endtag", tag)) # all other markup def handle_comment(self, data): self.append(("comment", data)) def handle_charref(self, data): self.append(("charref", data)) def handle_data(self, data): self.append(("data", data)) def handle_decl(self, data): self.append(("decl", data)) def handle_entityref(self, data): self.append(("entityref", data)) def handle_pi(self, data): self.append(("pi", data)) def unknown_decl(self, decl): self.append(("unknown decl", decl)) class EventCollectorExtra(EventCollector): def handle_starttag(self, tag, attrs): EventCollector.handle_starttag(self, tag, attrs) self.append(("starttag_text", self.get_starttag_text())) class TestCaseBase(unittest.TestCase): def get_collector(self): raise NotImplementedError def _run_check(self, source, expected_events, collector=None): if collector is None: collector = self.get_collector() parser = collector for s in source: parser.feed(s) parser.close() events = parser.get_events() if events != expected_events: self.fail("received events did not match expected events\n" "Expected:\n" + pprint.pformat(expected_events) + "\nReceived:\n" + pprint.pformat(events)) def _run_check_extra(self, source, events): self._run_check(source, events, EventCollectorExtra()) def _parse_error(self, source): def parse(source=source): parser = self.get_collector() parser.feed(source) parser.close() with self.assertRaises(html.parser.HTMLParseError): with self.assertWarns(DeprecationWarning): parse() class HTMLParserStrictTestCase(TestCaseBase): def get_collector(self): with support.check_warnings(("", DeprecationWarning), quite=False): return EventCollector(strict=True) def test_processing_instruction_only(self): self._run_check("", [ ("pi", "processing instruction"), ]) self._run_check("", [ ("pi", "processing instruction ?"), ]) def test_simple_html(self): self._run_check(""" &entity; sample text “ """, [ ("data", "\n"), ("decl", "DOCTYPE html PUBLIC 'foo'"), ("data", "\n"), ("starttag", "html", []), ("entityref", "entity"), ("charref", "32"), ("data", "\n"), ("comment", "comment1a\n-><&#bad;

", [ ("starttag", "p", []), ("data", "&#bad;"), ("endtag", "p"), ]) def test_unclosed_entityref(self): self._run_check("&entityref foo", [ ("entityref", "entityref"), ("data", " foo"), ]) def test_bad_nesting(self): # Strangely, this *is* supposed to test that overlapping # elements are allowed. HTMLParser is more geared toward # lexing the input that parsing the structure. self._run_check("", [ ("starttag", "a", []), ("starttag", "b", []), ("endtag", "a"), ("endtag", "b"), ]) def test_bare_ampersands(self): self._run_check("this text & contains & ampersands &", [ ("data", "this text & contains & ampersands &"), ]) def test_bare_pointy_brackets(self): self._run_check("this < text > contains < bare>pointy< brackets", [ ("data", "this < text > contains < bare>pointy< brackets"), ]) def test_illegal_declarations(self): self._parse_error('') def test_starttag_end_boundary(self): self._run_check("""""", [("starttag", "a", [("b", "<")])]) self._run_check("""""", [("starttag", "a", [("b", ">")])]) def test_buffer_artefacts(self): output = [("starttag", "a", [("b", "<")])] self._run_check([""], output) self._run_check([""], output) self._run_check([""], output) self._run_check([""], output) self._run_check([""], output) self._run_check([""], output) output = [("starttag", "a", [("b", ">")])] self._run_check([""], output) self._run_check([""], output) self._run_check([""], output) self._run_check(["'>"], output) self._run_check([""], output) self._run_check([""], output) output = [("comment", "abc")] self._run_check(["", ""], output) self._run_check(["<", "!--abc-->"], output) self._run_check([""], output) self._run_check([""], output) self._run_check([""], output) self._run_check([""], output) self._run_check([""], output) self._run_check([""], output) self._run_check(["", ""], output) def test_starttag_junk_chars(self): self._parse_error("") self._parse_error("") self._parse_error("") self._parse_error("") self._parse_error("'") self._parse_error("" % dtd, [('decl', 'DOCTYPE ' + dtd)]) def test_declaration_junk_chars(self): self._parse_error("") def test_startendtag(self): self._run_check("

", [ ("startendtag", "p", []), ]) self._run_check("

", [ ("starttag", "p", []), ("endtag", "p"), ]) self._run_check("

", [ ("starttag", "p", []), ("startendtag", "img", [("src", "foo")]), ("endtag", "p"), ]) def test_get_starttag_text(self): s = """""" self._run_check_extra(s, [ ("starttag", "foo:bar", [("one", "1"), ("two", "2")]), ("starttag_text", s)]) def test_cdata_content(self): contents = [ ' ¬-an-entity-ref;', "", '

', 'foo = "";', 'foo = "";', 'foo = <\n/script> ', '', ('\n//<\\/s\'+\'cript>\');\n//]]>'), '\n\n', 'foo = "";', '', # these two should be invalid according to the HTML 5 spec, # section 8.1.2.2 #'foo = ', #'foo = ', ] elements = ['script', 'style', 'SCRIPT', 'STYLE', 'Script', 'Style'] for content in contents: for element in elements: element_lower = element.lower() s = '<{element}>{content}'.format(element=element, content=content) self._run_check(s, [("starttag", element_lower, []), ("data", content), ("endtag", element_lower)]) def test_cdata_with_closing_tags(self): # see issue #13358 # make sure that HTMLParser calls handle_data only once for each CDATA. # The normal event collector normalizes the events in get_events, # so we override it to return the original list of events. class Collector(EventCollector): def get_events(self): return self.events content = """ ¬-an-entity-ref;

''""" for element in [' script', 'script ', ' script ', '\nscript', 'script\n', '\nscript\n']: element_lower = element.lower().strip() s = '