def __tidy(self, content):
tidy = Tidy()
tidy.setIndentAttributes(False)
tidy.setIndentContent(False)
tidy.setPrintBodyOnly(True)
tidy.setSmartIndent(False)
tidy.setWraplen(0)
tidy.setXHTML(False)
tidy.setNumEntities(True)
out = ByteArrayOutputStream()
doc = tidy.parseDOM(ByteArrayInputStream(String(content).getBytes()), out)
content = out.toString("UTF-8")
return content, doc
def __tidy(self, content):
tidy = Tidy()
tidy.setIndentAttributes(False)
tidy.setIndentContent(False)
tidy.setPrintBodyOnly(True)
tidy.setSmartIndent(False)
tidy.setWraplen(0)
tidy.setXHTML(True)
tidy.setNumEntities(True)
tidy.setShowWarnings(False)
tidy.setQuiet(True)
out = ByteArrayOutputStream()
tidy.parse(IOUtils.toInputStream(content, "UTF-8"), out)
return out.toString("UTF-8")
def __getContent(self, oid):
slash = oid.rfind("/")
pid = os.path.splitext(oid[slash+1:])[0] + ".htm"
payload = Services.storage.getObject(oid).getPayload(pid)
tidy = Tidy()
tidy.setIndentAttributes(False)
tidy.setIndentContent(False)
tidy.setPrintBodyOnly(True)
tidy.setSmartIndent(False)
tidy.setWraplen(0)
tidy.setXHTML(False)
tidy.setNumEntities(True)
out = ByteArrayOutputStream()
try:
doc = tidy.parseDOM(payload.getInputStream(), out)
content = out.toString("UTF-8")
content = self.__processMedia(oid, doc, content)
#print "[\n%s\n]" % content
except Exception, e:
print " * blog.py: Failed to get content: %s" % e.getMessage()