--- MPIWGWeb/MPIWGProjects.py 2006/08/03 09:43:40 1.47.2.56 +++ MPIWGWeb/MPIWGProjects.py 2006/08/15 13:27:43 1.47.2.57 @@ -47,6 +47,30 @@ def getTextFromNode(nodename): rc = rc + node.data return rc + +def sortStopWordsF(self,xo,yo): + if not hasattr(self,'_v_stopWords'): + self._v_stopWords=self.stopwords_en.data.split("\n") + + x=str(xo[1]) + y=str(yo[1]) + + strx=x.split(" ") + stry=y.split(" ") + + for tmp in strx: + if tmp.lower() in self._v_stopWords: + del strx[strx.index(tmp)] + + for tmp in stry: + if tmp.lower() in self._v_stopWords: + del stry[stry.index(tmp)] + + return cmp(" ".join(strx)," ".join(stry)) + +def sortStopWords(self): + return lambda x,y : sortStopWordsF(self,x,y) + def sortF(x,y): try: return cmp(x[1],y[1]) @@ -270,6 +294,23 @@ class MPIWGRoot(ZSQLExtendFolder): folders=['MPIWGProject','Folder','ECHO_Navigation'] meta_type='MPIWGRoot' + def harvestProjects(self): + """harvest""" + folder="/tmp" + try: + os.mkdir("/tmp/harvest_MPIWG") + except: + pass + founds=self.ZopeFind(self.aq_parent.projects,obj_metatypes=['MPIWGProject'],search_sub=1) + for found in founds: + txt=found[1].harvest_page() + + if txt and (txt != ""): + name=found[0].replace("/","_") + fh=file("/tmp/harvest_MPIWG/"+name,"w") + fh.write(txt) + fh.close() + def decode(self,str): """decoder""" if not str: @@ -520,6 +561,22 @@ class MPIWGRoot(ZSQLExtendFolder): self.id=id self.title=title + def removeStopWords(self,xo): + """remove stop words from xo""" + if not hasattr(self,'_v_stopWords'): + self._v_stopWords=self.stopwords_en.data.split("\n") + + x=str(xo) + + strx=x.split(" ") + + for tmp in strx: + + if tmp.lower() in self._v_stopWords: + del strx[strx.index(tmp)] + + return " ".join(strx) + def urlQuote(self,str): """quote""" return urllib.quote(str) @@ -718,6 +775,10 @@ class MPIWGRoot(ZSQLExtendFolder): if sort=="int": ret.sort(sortI) + elif sort=="stopWords": + print "sortStop" + ret.sort(sortStopWords(self)) + else: ret.sort(sortF) @@ -1451,6 +1512,7 @@ class MPIWGProject(CatalogAware,Folder): meta_type='MPIWGProject' default_catalog='ProjectCatalog' + def sortedByPlace(self,metatype): """find metatype and sort by place""" def sort(x,y):