--- ECHO_content/ECHO_language.py 2005/10/26 08:35:53 1.3 +++ ECHO_content/ECHO_language.py 2006/09/10 11:03:07 1.4 @@ -1,5 +1,10 @@ """Methoden fuer Language Technologies""" + +from Products.ECHO_content.analyseAndTag.analyseAndTag import DonatusFile + +import xml.parsers + def donatus(txt2): import xmlrpclib @@ -39,7 +44,17 @@ class ECHO_language: """analyze by donatus""" return donatusVariant2Lemma(donatus(self.lemmatize(nr))) + def tagLex(self,nr="1"): + """gerateLinks""" + txt=self.getPage(_pn=nr) + + df=DonatusFile(txt=self.getPage(_pn=nr)) + + return df.wordsToLinks() + #return DonatusFile(txt=self.getPage(_pn=nr)).convertedXML() + + def tagLex_old(self,nr="1"): """generate Links""" global retLex global toggle @@ -124,7 +139,7 @@ class ECHO_language: def insertW(str): splitted=str.split() wordlist=["%s"%split for split in splitted] - return string.join(wordlist,'\n') + return "\n".join(wordlist) # 3 handler functions def start_element(name, attrs):