def extract_inline_source(self):
source = self.get_latest_source()
doc = pyQ(source)
sheets = doc.find('style')
source = ''
for sheet in sheets:
source += pyQ(sheet).html()
def sync_stylesheets(self):
source = self.get_latest_source()
doc = pyQ(source)
links = doc.find('link[rel="stylesheet"]')
inlineSheets = doc.find('style')
for sheet in links:
sheet = pyQ(sheet)
href = sheet.attr('href')
url = resolve_url(href, self.project.url);
sheet = Stylesheet.objects.filter(project=self.project, url=url)
if len(sheet) != 0:
sheet = sheet[0]
else:
sheet = Stylesheet(url=url)
sheet.user = self.user
sheet.project = self.project
sheet.save()
sheet.paths.add(self)
sheet.save()
def extract_paths(self):
project = self.project
user = self.user
base = ApplicationPath.objects.filter(project=project)
pyQ = self.py_query()
links = pyQ('a')
projectUrl = self.project.url
netloc = urlparse(projectUrl).netloc
for link in links:
if 'href' in link.attrib:
href = link.attrib['href']
href = resolve_url(href, projectUrl)
path = base.filter(url=href)
if len(path) == 0:
parsed = urlparse(href)
if parsed.netloc == netloc:
path = ApplicationPath(url=href)
path.project=project
path.user=user
path.save()
def py_query(self): if not hasattr(self, 'pyPath'): source = self.get_latest_source() self.pyPath = pyQ(source) return self.pyPath