def test_declaration_incomplete(self):
# verify that the lenient declaration can handle incompete tags
doc = " <html>A<!-- bad comment -->B</html>"
# Note unrelated problem: without the initial space above, there
# is problem in parsing the incomplete <html>. Investigate?!
for i in range(1, len(doc)-1):
chunks = [doc[:i], doc[i:]]
#print chunks
fp = ChunkedStringIO(chunks)
tokens = hpp.generate_tokens(fp)
self._test_generator1(
tokens,
[
(TAG, u'html', []),
(DATA, u'A' ),
(DATA, u'B' ),
(ENDTAG, u'html'),
])
def process(fp, out, meta):
""" Return has_html, has_frameset """
has_html = False
has_frameset = False
has_common_tag = False
first_td = False # state for iterating td inside tr
iterator = html_pull_parser.generate_tokens(fp)
# General HTML format
# <html>
# <head>
# <body>
#
# However all elements are optional.
# It is better to use a flat, stateless loop to process elements
for token in iterator:
if token[0] == DATA:
out.out(token[1])
elif token[0] == TAG:
tag = token[1]
id = starttag_dict.get(tag,-1)
if id > 0:
has_common_tag = True
if id == sOUTP:
out.outTag('p')
elif id == sOUTTAG:
out.outTag(tag)
elif id == sTR:
first_td = True
elif id == sTDTH:
if first_td:
first_td = False
else:
out.out(' ')
elif id == sINPUT:
attrs = token[2]
itype = _getvalue(attrs, 'type')
if itype == 'checkbox':
if _hasattr(attrs,'checked'):
out.out('[*] ')
else:
out.out('[ ] ')
elif itype == 'radio':
if _hasattr(attrs,'checked'):
out.out('(*) ')
else:
out.out('( ) ')
elif itype == 'image':
alt = _getvalue(attrs, 'alt') or _getvalue(attrs, 'value')
out.outAlt(saxutils.unescape(alt))
elif itype == 'password':
out.outAlt('***')
elif itype == 'hidden':
pass
else:
value = _getvalue(attrs, 'value')
out.outAlt(saxutils.unescape(value))
elif id == sIMG:
attrs = token[2]
alt = _getvalue(attrs, 'alt')
if alt:
out.outAlt(saxutils.unescape(alt))
elif id == sHTML:
has_html = True
out.notifyHtml()
elif id == sBODY:
out.outHeader(meta)
elif id == sFRAMESET:
has_frameset = True
elif id == sTITLE:
title = ''
for token in iterator:
if token[0] == DATA:
title += token[1]
elif token in [
(ENDTAG, 'title'), # only </title> is valid
(ENDTAG, 'head'), # in case no </title>
(TAG, 'body'), # in case no </title>
]:
break
meta['title'] = _collapse(title)
elif id == sMETA:
attrs = token[2]
name = _getvalue(attrs,'name').lower()
content = _getvalue(attrs,'content')
if name == 'description':
meta['description'] = saxutils.unescape(_collapse(content))
elif name == 'keywords':
meta['keywords'] = saxutils.unescape(_collapse(content))
elif id == sSCRIPT:
for token in iterator:
if token == (ENDTAG, 'script'):
break
elif id == sSTYLE:
for token in iterator:
if token == (ENDTAG, 'style'):
break
elif id == sSELECT:
for token in iterator:
if token == (ENDTAG, 'select'):
break
elif token[0] == ENDTAG:
tag = token[1]
id = endtag_dict.get(tag,-1)
if id == eCLOSE_TAG:
out.outTag('/'+tag)
elif id == eBREAK_LINE:
out.outTag('br')
out.close(meta)
return has_html, has_frameset, has_common_tag
def scan_html(fp, baseuri, append):
token_stream = hpp.generate_tokens(fp, comment=True)
for token in token_stream:
if token[0] != hpp.TAG:
continue
tag = token[1]
# ----------------------------------------------------------------------
# handle <style> block
if tag == 'style':
# HACK: <style> should only be valid inside <head>
styles = []
for token in token_stream:
if token[0] == hpp.DATA:
styles.append(token[1])
elif token[0] == hpp.COMMENT:
styles.append(token[1]) # CSS enclosed by HTML comment!
elif token[0] == hpp.TAG:
if token[1] == 'style':
# <style> follows by <style>??? OK, treat it as <style>.
continue
else:
# No </style>??? Any other open tags would close <style>
break
elif token[0] == hpp.ENDTAG:
break
else:
# the stream is exhausted? Make sure next step knows
# there is no unprocessed token.
token = None
# process the style content
_scan_html_style(tag, ''.join(styles), baseuri, append)
# look at last unprecessed token
if not token:
# TODO: test
break
elif token[0] == hpp.ENDTAG:
# hopefully this ends with a valid </style>
# TODO: test
continue
else:
# we got a (non-style) begin TAG??? OK, we'll process this tag.
tag = token[1]
# TODO: test
# ----------------------------------------------------------------------
# read TAG and its attributes
isLinkTag = (tag == 'link')
uri_attr = LINKABLE_TAGS.get(tag,'')
# run through attribute list to find relevant info
uri = style = rel = ctype = ''
for n, v in token[2]: # TODO: need to XML decode?
if n == uri_attr:
uri = v
elif n == 'style':
style = v
elif isLinkTag:
if n == 'rel':
rel = v
elif n == 'type':
ctype = v
# ----------------------------------------------------------------------
# handle style attribute
if style:
_scan_html_style(tag, style, baseuri, append)
# ----------------------------------------------------------------------
# handle uri attributes (href, src, etc)
if not uri:
continue
#print >>sys.stderr, tag, uri, ctype
# note: the ctype is advisory, content-type from http may not be consistent
# everything not HTML or CSS is APPLICATION
if tag in ['frame','iframe']:
ctype = TEXT_HTML
elif isLinkTag:
if rel.lower() == 'stylesheet': # TODO: rel="alternate stylesheet"?
ctype = TEXT_CSS
elif ctype != TEXT_CSS:
# only want CSS from <link>
continue
else:
ctype = APPLICATION
append(baseuri, uri, ctype, tag)
def _test_generator(self, doc, expect, **args):
fp = StringIO.StringIO(doc)
tokens = hpp.generate_tokens(fp, **args)
self._test_generator1(tokens, expect)
def test_0(self):
tlist = list(hpp.generate_tokens(StringIO.StringIO("å")))
self.assertEqual(tlist, [(DATA, u'\u00e5')])
