--- MPIWGWeb/MPIWGProjects.py 2005/10/10 08:42:49 1.47.2.25 +++ MPIWGWeb/MPIWGProjects.py 2006/08/15 13:27:43 1.47.2.57 @@ -2,6 +2,10 @@ for organizing and maintaining the different project pages """ +#TODO: mechanismus fur links to personen ueberarbeiten, da jetzt alle e_mails als members auftauchen unabhaengig vom status publish_the_data +#TODO: was passiert wenn aenderungen von jochen im filemaker nicht mit den aenderungen im sql uebereinstimmen +#TODO: in einzelnen projecte steht als pfad auf die bilder noch wwwneu statt www + from Products.PageTemplates.PageTemplateFile import PageTemplateFile from Products.PageTemplates.PageTemplate import PageTemplate from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate @@ -23,13 +27,50 @@ from Products.ZSQLMethods.SQL import SQL from AccessControl import ClassSecurityInfo from bibliography import * import time +import xml.dom.minidom +import sys +from Ft.Xml.XPath import Evaluate +from Ft.Xml.XPath.Context import Context +from Ft.Xml.Domlette import NonvalidatingReader,PrettyPrint, Print +from Ft.Xml import EMPTY_NAMESPACE definedFields=['WEB_title','xdata_01','xdata_02','xdata_03','xdata_04','xdata_05','xdata_06','xdata_07','xdata_08','xdata_09','xdata_10','xdata_11','xdata_12','xdata_13','WEB_project_header','WEB_project_description','WEB_related_pub'] checkFields = ['xdata_01'] +def getTextFromNode(nodename): + nodelist=nodename.childNodes + rc = "" + for node in nodelist: + if node.nodeType == node.TEXT_NODE: + 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]) @@ -253,6 +294,36 @@ 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: + return "" + if type(str) is StringType: + try: + return str.decode('utf-8') + except: + return str.decode('latin-1') + else: + + return str + def versionHeaderEN(self): """version header text""" @@ -269,6 +340,25 @@ class MPIWGRoot(ZSQLExtendFolder): txt="""
new:%s
\n"%username.encode('utf-8')) + obj=getattr(memberFolder,username) + ret=obj.createNewDBEntry(publish_the_data,id,name, + vorname,username,title,position,e_mail, + e_mail_p,date_from,date_to, + abteilung,heimat_inst,funded_by, + e_mail2,txt,txt_p,stay_at_mpiwg) + RESPONSE.write("""%s
"""%ret[1].encode('utf-8')) + except: + RESPONSE.write("ERROR:%s %s %s
\n"%(username.encode('utf-8'),name.encode('utf-8'),vorname.encode('utf-8'))) + RESPONSE.write(": %s %s"%sys.exc_info()[:2]) + else: + RESPONSE.write("
update:%s
\n"%username.encode('utf-8')) + + xquery=".//dn:ROW[dn:username='%s']"%username + results=dom.xpath(xquery,explicitNss={'dn':'http://www.filemaker.com/fmpdsoresult'}) + if len(results)>1: + RESPONSE.write("WARNING (update): username %s not unique
\n"%username.encode('utf-8')) + xquery=".//dn:ROW[dn:publish_the_data='yes' and dn:username='%s']"%username + rd=dom.xpath(xquery,explicitNss={'dn':'http://www.filemaker.com/fmpdsoresult'}) + if len(rd)>1:#mehrere published + RESPONSE.write("ERROR (update): username %s not unique and more then one date set to be published
\n"%username.encode('utf-8')) +# elif len(rd)==1:#nur einer published, kein published dann mache nichts +# publish_the_data=getTextFromNode(rd[0].xpath('./dn:publish_the_data',explicitNss={'dn':'http://www.filemaker.com/fmpdsoresult'})[0]) +# date_from=getTextFromNode(rd[0].xpath('./dn:Date_from',explicitNss={'dn':'http://www.filemaker.com/fmpdsoresult'})[0]) +# date_to=getTextFromNode(rd[0].xpath('./dn:Date_to',explicitNss={'dn':'http://www.filemaker.com/fmpdsoresult'})[0]) +# + + obj=getattr(memberFolder,username) + + done= obj.updateDBEntry(DBid=id,publish_the_data=publish_the_data, + date_from=date_from, + date_to=date_to,stay_at_mpiwg=stay_at_mpiwg,position=position,abteilung=abteilung) + if not done and (publish_the_data=='yes'): + + ret=obj.createNewDBEntry(publish_the_data,id,name, + vorname,username,title,position,e_mail, + e_mail_p,date_from,date_to, + abteilung,heimat_inst,funded_by, + e_mail2,txt,txt_p,stay_at_mpiwg) + if not ret[0]: + + RESPONSE.write("Error: %s
\n"%repr(ret[1])) + else: + RESPONSE.write("New: %s
\n"%repr(ret[1])) + + #TODO: reindexCatlogs and updatePublications wieder einbaue + #self.reindexCatalogs(RESPONSE) + + #self.updatePublicationDB() + + #self.ZSQLResetConnection() + RESPONSE.write("old:%s (%s,%s)
\n"%(found.username,found.name,found.vorname)) + #delete non existing @@ -733,29 +1026,41 @@ class MPIWGRoot(ZSQLExtendFolder): pass self.ProjectCatalog.manage_catalogReindex(self.REQUEST,RESPONSE,self.REQUEST['URL1']) - + self.updatePublicationDB() + if RESPONSE is not None: RESPONSE.redirect('manage_main') - def getAllMembers(self): - """give list of all members""" - ret=[] - - #for x in self.members.objectValues('MPIWGStaff'): - #print x.title - # ret.append(x.title.decode('utf-8')) +# def getAllMembers(self,reset=None): +# """give list of all members""" +# ret=[] +# +# +# if reset or (getattr(self,'_v_membersList',None) is None): +# for member in self.members._objects: +# if member['meta_type']=='MPIWGStaff': +# memberObj=getattr(self.members,member['id']) +# if memberObj.isPublished(): +# ret.append(memberObj.title.decode('utf-8')) +# +# ret.sort() +# self._v_membersList=ret[0:] +# print ret +# +# return self._v_membersList - for x in self.ZopeFind(self.members,obj_metatypes=['MPIWGStaff']): - ret.append(x[1].title.decode('utf-8')) - - ret.sort() - #print ret - + def getAllMembers(self): + #ret=[] + + results=self.MembersCatalog({'isPublished':True}) + + ret=[", ".join([proj.lastName, proj.firstName]).decode('utf-8') for proj in results] + + ret.sort() return ret - def printAllMembers(self): - """prin""" + """print""" members=self.getAllMembers() ret="" for x in members: @@ -771,7 +1076,7 @@ class MPIWGRoot(ZSQLExtendFolder): return entry - def getTree(self,date=None): + def getTree(self,date=None,onlyActive=None): """generate Tree from project list""" returnList=[] @@ -785,9 +1090,14 @@ class MPIWGRoot(ZSQLExtendFolder): #title=project[0].WEB_title title=[project[0].getContent('WEB_title')] #print title - if idNr[0]!="x": + + if idNr[0]=="x": # kompatibilitt mit alter Konvention, x vor der Nummer macht project inactive + project[0].setActiveFlag(False) + + if onlyActive and project[0].isActiveProject(): #nur active projekte returnList.append((depth,nr,title,project[0])) - + elif not onlyActive: + returnList.append((depth,nr,title,project[0])) return returnList def formatElementsAsList(self,elements): @@ -834,7 +1144,7 @@ class MPIWGRoot(ZSQLExtendFolder): ret+="""""") + + imageURLs=[] + imageCaptions=[] + for split in splitted[1:]: + tmp=split.split("
") + #return repr(splitted[1]) + try: + imageURLs.append(tmp[0].split("\"")[1].encode('utf-8')) + except: + imageURLs.append(tmp[0].split("src=")[1].split(" ")[0].encode('utf-8')) + + split2="".join(tmp[1:]) - filename=self.imageURL.split("/")[-1] - #lege neues images object an, mit leerem bild - if self.ZopeFind(self,obj_ids=[filename]): - #existiert das bild schon, dann neueun filenamen - filename="project_image_"+filename - - self.addImage(None,self.imagecap,filename=filename) - #hole die bilddaten aus der url - data=urllib.urlopen(self.absolute_url()+"/"+self.imageURL).read() + splitted=split2.split("""""") + if len(splitted)>1: + tmp=splitted[1].split("
") + imageCaptions.append(tmp[0].encode('utf-8')) - obj=getattr(self,filename) - obj.update_data(data) + + else: + #keine caption + + imageCaptions.append("") + + + + + + + + #eintragen: + for imageURL in imageURLs: + filename=imageURL.split("/")[-1] + #lege neues images object an, mit leerem bild + + if self.ZopeFind(self,obj_ids=[filename]): + #existiert das bild schon, dann neuen filenamen + filename="project_image_"+filename + + self.addImage(None,imageCaptions[imageURLs.index(imageURL)],filename=filename) + #hole die bilddaten aus der url + url=self.absolute_url()+"/"+imageURL + #url=self.absolute_url()+"/"+filename + + try:#relative url + data=urllib.urlopen(url).read() + except: + try:#absolute + data=urllib.urlopen(self.imageURL).read() + except: + zLOG.LOG("MPIWG Project",zLOG.ERROR,"can't open: %s"%url) + + obj=getattr(self,filename) + obj.update_data(data) if RESPONSE: RESPONSE.redirect('manageImages') @@ -1360,6 +1730,17 @@ class MPIWGProject(CatalogAware,Folder): if RESPONSE: RESPONSE.redirect('manageImages') + + def hasChildren(self,date=None,onlyActive=True): + """check if project has children""" + ct=self.getContexts(childs=self.getContent('xdata_05'), + depth=1,date=date,onlyActive=onlyActive) + + if ct and len(ct)>0: + return True + else: + return False + def addImage(self,fileHd,caption,RESPONSE=None,filename=None): """add an MPIWG_Project_image""" @@ -1517,6 +1898,8 @@ class MPIWGProject(CatalogAware,Folder): self.creationTime=time.strftime("%Y%m%d%H%M%S",time.localtime())[0:] self.id=id self.title=id + self.isActiveFlag=True #Flag is true is the project is still active, False if accomplished + if argv: for arg in definedFields: try: @@ -1537,6 +1920,14 @@ class MPIWGProject(CatalogAware,Folder): ) + def isActiveProject(self): + """check if the project is still active, default is true, set to false is the project is accomplished""" + return getattr(self,'isActiveFlag',True) + + def setActiveFlag(self,status=True): + """set the active flag""" + self.isActiveFlag=status + def checkDate(self,date): """teste ob zum Zeitpunkt date eine andere version existierte""" @@ -1577,7 +1968,106 @@ class MPIWGProject(CatalogAware,Folder): """warnung: project noch nicht existent""" pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','no_project')).__of__(self) return pt() - + + def getGetNeighbourhood(self,wordStr, length=100): + """finde umgebung um die worte in wordStr, zurckgegeben wird eine Array mit den Umgebungen von Fundstellen der Worte + alle Tags werden entfernt, die Fundstellen werden mit XX getaggt, die Umgebungen werden + case insensitive gesucht + @param wordStr: string mit Worten getrennt durch Leerzeichen, Phrasen sind mit " gekennzeichnet + "eine phrase", "*" bezeichnet wildcards und wird ignoriert" + @param length: optional, default wert 100, 2*length ist die gr§e der Umgebung + """ + + ret=[] # nimmt das Array auf, dass spter zurckgegeben wird + ranges=[] #Array mit tupeln x,y wobei x die Position des Anfang und y des Endes der i-ten Umgebung angiebt + + def isInRanges(nr,length): + """test ob eine gegeben Position nr schon irgendwo in einer Umgebung ist, gibt den Index des ersten Wertes aus ranges zurck, + -1, wenn kein Treffer + + @param nr: Position die geprft werden soll + @param length: Lnge des Wortes das geprft werden soll + """ + for x in ranges: + if (x[0]<=nr) and (nr < (x[1]-length)): + return ranges.index(x) + return -1 + + # deal with phrases, in Phrasen werden die Leerzeichen durch "_" ersetzt. + def rep_empty(str): + x= re.sub(" ","_",str.group(0)) + return re.sub("\"","",x) + + wordStr=re.sub("\".*?\"", rep_empty,wordStr)#ersetze leerzeichen in " " durch "_" und loesche " + + #deal with wildcards, for our purposes it is enough to delete the wildcard + wordStr=wordStr.replace("*","") + + words=wordStr.split(" ") + #if not words is ListType: + # words=[words] + + txt=self.harvest_page() + + txt=re.sub("<.*?>", "", txt) # loesche alle Tags + for word in words: + word=re.sub("_"," ",word) # ersetze zurueck "_" durch " " + pos=0 + + n=txt.lower().count(word.lower()) # wie oft tritt das Wort auf + + for i in range(n): + pos=txt.lower().find(word.lower(),pos) + + if pos > 0: + x=max(0,pos-length) + y=min(len(txt),pos+length) + + + #is word already in one of the results + nr=isInRanges(pos,len(word)) + if nr >=0:# word ist in einer schon gefunden Umgebung, dann vergr§ere diese + x=min(ranges[nr][0],x) + y=max(ranges[nr][1],y) + + str=txt[x:y] + + if nr >=0: # word ist in einer schon gefunden Umgebung + ranges[nr]=(x,y) # neue Position der Umgebung + + ret[nr]=str # neue Umgebung + else: # andernfalls neue Umgebung hinzufgen + ranges.append((x,y)) + + ret.append(str) + + pos=pos+len(word) + else: + break; + + # now highlight everything + for x in range(len(ret)): + for word in words: + repl=re.compile(word,re.IGNORECASE) + ret[x]=repl.sub(""" %s"""%word.upper(),ret[x]) + + return ret + + def harvest_page(self,context=None): + """seite fuer harvesting fuer die Projektsuche""" + if not context: + context=self + + if self.isActiveProject() and self.isActual(): + ext=getattr(self,"harvest_main",None) + if ext: + return getattr(self,ext.getId())() + + pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','harvest_main')).__of__(context) + + + return pt() + def index_html(self,request=True,context=None): """show homepage""" if not context: @@ -1716,7 +2206,7 @@ class MPIWGProject(CatalogAware,Folder): start=kupu.find("") end=kupu.find("") newcontent= kupu[start+6:end] - + if preview: return self.preview(newcontent) @@ -1728,7 +2218,7 @@ class MPIWGProject(CatalogAware,Folder): return True - security.declareProtected('View managment screens','edit') + security.declareProtected('View management screens','edit') def edit(self,western=None): """Edit pages""" if western: @@ -1737,6 +2227,16 @@ class MPIWGProject(CatalogAware,Folder): pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','edit_MPIWGProjectNeu.zpt')).__of__(self) return pt() + + edit_MPIWGProject_main = PageTemplateFile('zpt/edit_MPIWGProject_main', globals()) + + def getPathStyle(self, path, selected, style=""): + """returns a string with the given style + 'sel' if path == selected.""" + + if path == selected: + return style + 'sel' + else: + return style def preview(self,description): """preview""" @@ -1747,7 +2247,7 @@ class MPIWGProject(CatalogAware,Folder): for field in definedFields: setattr(tmpPro,field,getattr(self,field)) tmpPro.WEB_project_description=description[0:] - + tmpPro.invisible=True pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','previewFrame.zpt')).__of__(self) return pt() @@ -1796,6 +2296,10 @@ class MPIWGProject(CatalogAware,Folder): setattr(self,x,[self.REQUEST[x].decode('utf-8')]) + if self.REQUEST.has_key('active'): + self.setActiveFlag(True) + else: + self.setActiveFlag(False) if fromEdit and (RESPONSE is not None): RESPONSE.redirect('./editMPIWGBasisEditor') @@ -1855,6 +2359,7 @@ class MPIWGProject(CatalogAware,Folder): splitted=[y.rstrip().lstrip() for y in splitted] for x in splitted: + x=re.sub(r"[^A-z ]","",x) if (not x==u'') and x in wert: return 1 return 0 @@ -1865,13 +2370,13 @@ class MPIWGProject(CatalogAware,Folder): pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','edit_MPIWGBasisNeu.zpt')).__of__(self) return pt() - + security.declareProtected('View management screens','editMPIWGBasisForm') def editMPIWGBasisForm(self): """editform""" pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','edit_MPIWGBasis.zpt')).__of__(self) return pt() - + security.declareProtected('View management screens','editMPIWGRelatedPublicationsForm') def editMPIWGRelatedPublicationsForm(self): """Edit related Publications""" pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','edit_MPIWGRelatedPublications.zpt')).__of__(self) @@ -1931,4 +2436,4 @@ def manage_addMPIWGProject(self,id,RESPO if RESPONSE is not None: - RESPONSE.redirect('manage_main') + RESPONSE.redirect('manage_main') \ No newline at end of file