--- ECHO_content/ECHO_collection.py 2004/08/05 12:19:30 1.157 +++ ECHO_content/ECHO_collection.py 2004/08/17 15:24:38 1.164 @@ -48,6 +48,45 @@ import xml.dom.minidom from ECHO_graphicalOverview import javaHandler,javaScriptMain import ECHO_helpers +def donatus(txt2): + import xmlrpclib + + server = xmlrpclib.ServerProxy("http://archimedes.fas.harvard.edu/cgi-bin/donatus-rpc") + + txt=txt2.encode('utf-8') + bin=xmlrpclib.Binary(txt) + + + + ret=server.donatus.analyze(bin) + + + return ret['morphData'].data + + +def donatusVariant2Lemma(morphData): + """creates hash variant -> morphdata""" + ret={} + dom=xml.dom.minidom.parseString(morphData) + lemmas=dom.getElementsByTagName('lemma') + for lemma in lemmas: + variants=lemma.getElementsByTagName('variant') + for variant in variants: + atr=variant.getAttribute('form') + if ret.has_key(atr): + ret[atr].append=lemma.getAttribute('form') + else: + ret[atr]=[lemma.getAttribute('form')] + + return ret + +#regexp for extracting elements from xml +patternTXT=r"<\s*txt.*?>(.*?)" +regexpTXT = re.compile(patternTXT, re.IGNORECASE + re.DOTALL) +patternPage=r"<\s*page.*?>(.*?)" +regexpPage = re.compile(patternPage, re.IGNORECASE + re.DOTALL) + + def ECHO_rerenderLinksMD(self,obj=None,types=['title','label']): """Rerender all Links""" ret="" @@ -274,6 +313,7 @@ class BrowserCheck: def writeMetadata(url,metadict,project=None,startpage=None,xslt=None,thumbtemplate=None,topbar=None,digiLibTemplate=None,xmlfrag=None,digiliburlprefix=None): """Einlesen der Metadaten und und erstellen des geaenderten XML file""" + def updateTextToolNode(tag,value): #print dom,tag,value metanode=dom.getElementsByTagName('texttool')[0] @@ -302,7 +342,7 @@ def writeMetadata(url,metadict,project=N else: try: geturl="" - for line in urllib.urlopen(url).readlines(): + for line in ECHO_helpers.urlopen(url).readlines(): geturl=geturl+line @@ -384,7 +424,7 @@ def readMetadata(url): metadict={} try: geturl="" - for line in urllib.urlopen(url).readlines(): + for line in ECHO_helpers.urlopen(url).readlines(): geturl=geturl+line @@ -576,7 +616,171 @@ def manage_addECHO_layoutTemplate(self, REQUEST.RESPONSE.redirect(u+'/manage_main') return '' +class ECHO_fullText(ZopePageTemplate): + """echo fulltext in xml""" + + meta_type="ECHO_fullText" + + def donatusVariant2Lemma(self,nr='1'): + """analyze by donatus""" + return donatusVariant2Lemma(donatus(self.lemmatize(nr))) + + def tagLex(self,nr="1"): + """generate Links""" + global retLex + global toggle + toggle=0 + retLex="" + + lemmatized=self.lemmatize(nr)[0:] + #print "ho",repr(lemmatized) + variants=donatusVariant2Lemma(donatus(lemmatized)) + + def createTag(name,attrs): + global toggle + + if name=="w": + toggle=1 + return "" + else: + tag="<" + tag+=name + for attr in attrs.keys(): + tag+=""" %s="%s" """%(attr,attrs[attr]) + tag+=">" + return tag + + def createData(data): + global toggle + astring="""%s """ + if toggle: # tag war ein w + toggle=0 + if variants.has_key(data): + return astring%(variants[data][0],data) + else: + return astring%(data,data) + + + + # 3 handler functions + def start_element(name, attrs): + global retLex + print name + retLex+=createTag(name,attrs) + def end_element(name): + global retLex + if not name=="w": + retLex+=""%(name.encode('utf-8')) + + + def char_data(data): + global retLex + if data: + try: + retLex+=createData(data) + except: + """no""" + + p = xml.parsers.expat.ParserCreate() + + p.StartElementHandler = start_element + p.EndElementHandler = end_element + p.CharacterDataHandler = char_data + + p.Parse(lemmatized.encode('utf-8'),1) + print repr(lemmatized.encode('utf-8')) + + return retLex + + + def lemmatize(self,nr='1',lang="de"): + """lemmatize""" + global ret + ret="" + + def createTag(name,attrs): + tag="<" + tag+=name + for attr in attrs.keys(): + tag+=""" %s="%s" """%(attr,attrs[attr]) + tag+=">" + return tag + + def insertW(str): + splitted=str.split() + wordlist=["%s"%split for split in splitted] + return string.join(wordlist,'\n') + + # 3 handler functions + def start_element(name, attrs): + global ret + ret+=createTag(name,attrs) + def end_element(name): + global ret + ret+=""%(name.encode('utf-8')) + + def char_data(data): + global ret + ret+=insertW(data) + + p = xml.parsers.expat.ParserCreate() + + p.StartElementHandler = start_element + p.EndElementHandler = end_element + p.CharacterDataHandler = char_data + + p.Parse(self.getPage(nr), 1) + txt=""" +
%s
+ """ + ret=txt%(lang,ret) + + return ret + + def getPage(self,nr='1'): + """get page n""" + dom=xml.dom.minidom.parseString(self()) + pages=dom.getElementsByTagName('page') + + return pages[int(nr)-1].toxml('utf-8') + +# Product registration and Add support +manage_addECHO_fullTextForm = PageTemplateFile( + 'zpt/AddECHO_fullText.zpt', globals()) + +from urllib import quote + +def manage_addECHO_fullText(self, id, title=None, text=None, + REQUEST=None, submit=None): + "Add a Page Template with optional file content." + + id = str(id) + if REQUEST is None: + self._setObject(id, ECHO_fullText(id, text)) + ob = getattr(self, id) + if title: + ob.pt_setTitle(title) + return ob + else: + file = REQUEST.form.get('file') + headers = getattr(file, 'headers', None) + if headers is None or not file.filename: + zpt = ECHO_fullText(id) + else: + zpt = ECHO_fullText(id, file, headers.get('content_type')) + + self._setObject(id, zpt) + + try: + u = self.DestinationURL() + except AttributeError: + u = REQUEST['URL1'] + + if submit == " Add and Edit ": + u = "%s/%s" % (u, quote(id)) + REQUEST.RESPONSE.redirect(u+'/manage_main') + return '' class ECHO_resource(Folder,Persistent): """ECHO Ressource""" @@ -680,7 +884,7 @@ class ECHO_resource(Folder,Persistent): try: geturl="" - for line in urllib.urlopen(url).readlines(): + for line in ECHO_helpers.urlopen(url).readlines(): geturl=geturl+line @@ -710,10 +914,10 @@ class ECHO_resource(Folder,Persistent): params="accessright=%s"%accessright - #print urllib.urlopen(self.absolute_url()+'/setAccessRightXML'+'?'+params).read() + #print ECHO_helpers.urlopen(self.absolute_url()+'/setAccessRightXML'+'?'+params).read() - urllib.urlopen('http://xserve02.mpiwg-berlin.mpg.de:18880/echo_nav/storage/downloadExternalXML?index_meta_url=%s&xml_url=%s'%(self.metalink,self.absolute_url()+'/setAccessRightXML'+urllib.quote('?'+params))).read() + ECHO_helpers.urlopen('http://xserve02.mpiwg-berlin.mpg.de:18880/echo_nav/storage/downloadExternalXML?index_meta_url=%s&xml_url=%s'%(self.metalink,self.absolute_url()+'/setAccessRightXML'+urllib.quote('?'+params))).read() if RESPONSE is not None: @@ -726,7 +930,7 @@ class ECHO_resource(Folder,Persistent): try: geturl="" - for line in urllib.urlopen(url).readlines(): + for line in ECHO_helpers.urlopen(url).readlines(): geturl=geturl+line @@ -802,7 +1006,7 @@ class ECHO_resource(Folder,Persistent): params="startpage=%s"%startpage - urllib.urlopen('http://xserve02.mpiwg-berlin.mpg.de:18880/echo_nav/storage/downloadExternalXML?index_meta_url=%s&xml_url=%s'%(self.metalink,self.absolute_url()+'/newMetaXML'+urllib.quote('?'+params))).read() + ECHO_helpers.urlopen('http://xserve02.mpiwg-berlin.mpg.de:18880/echo_nav/storage/downloadExternalXML?index_meta_url=%s&xml_url=%s'%(self.metalink,self.absolute_url()+'/newMetaXML'+urllib.quote('?'+params))).read() path=self.metalink @@ -815,10 +1019,11 @@ class ECHO_resource(Folder,Persistent): path=re.sub('http://foxridge.rz-berlin.mpg.de:8080','',path) # falls foxridge als server path=re.sub('http://content.mpiwg-berlin.mpg.de','',path) # falls content als server path=re.sub('http://foxridge.rz-berlin.mpg.de','',path) # falls foxridge als server + path=re.sub('http://vision.rz-berlin.mpg.de','',path) # falls vision als server path=re.sub('/index.meta','',path) - urllib.urlopen("http://nausikaa2.rz-berlin.mpg.de:86/cgi-bin/toc/admin/reg.cgi?path=%s"%path).readlines() + ECHO_helpers.urlopen("http://nausikaa2.rz-berlin.mpg.de:86/cgi-bin/toc/admin/reg.cgi?path=%s"%path).readlines() if RESPONSE is not None: RESPONSE.redirect('manage_main') @@ -858,7 +1063,7 @@ class ECHO_resource(Folder,Persistent): params="project=%s&xslt=%s&thumbtemplate=%s&topbar=%s&digiLibTemplate=%s&digiliburlprefix=%s"%(project,xslt,thumbtemplate,topbar,digiLibTemplate,digiliburlprefix) - urllib.urlopen('http://echo.mpiwg-berlin.mpg.de/echo_nav/storage/downloadExternalXML?index_meta_url=%s&xml_url=%s'%(self.metalink,self.absolute_url()+'/newMetaXML'+urllib.quote('?'+params))).read() + ECHO_helpers.urlopen('http://echo.mpiwg-berlin.mpg.de/echo_nav/storage/downloadExternalXML?index_meta_url=%s&xml_url=%s'%(self.metalink,self.absolute_url()+'/newMetaXML'+urllib.quote('?'+params))).read() #print self.absolute_url()+'/newMetaXML'+urllib.quote'?'+params) # hack Pfad auf die Dokumente @@ -878,7 +1083,11 @@ class ECHO_resource(Folder,Persistent): path=re.sub('http://foxridge.rz-berlin.mpg.de:8080','',path) # falls foxridge als server path=re.sub('http://foxridge.rz-berlin.mpg.de','',path) # falls foxridge als server path=re.sub('http://content.mpiwg-berlin.mpg.de','',path) # falls content als server - return urllib.urlopen("http://nausikaa2.rz-berlin.mpg.de:86/cgi-bin/toc/admin/reg.cgi?path=%s"%path).readlines() + + path=re.sub('http://vision.rz-berlin.mpg.de','',path) # falls vision als server + + return ECHO_helpers.urlopen("http://nausikaa2.rz-berlin.mpg.de:86/cgi-bin/toc/admin/reg.cgi?path=%s"%path).readlines() + if RESPONSE is not None: RESPONSE.redirect('manage_main') @@ -913,7 +1122,7 @@ class ECHO_resource(Folder,Persistent): def getFullTextXML(self,noredirect=None): """getFullTextXML; gives the FullText as an XML Document, and if somthing goes wrong.""" try: - fh=urllib.urlopen(self.metalink) + fh=ECHO_helpers.urlopen(self.metalink) dom=xml.dom.minidom.parse(fh) texttools=dom.getElementsByTagName('texttool') text=texttools[0].getElementsByTagName('text') @@ -936,7 +1145,7 @@ class ECHO_resource(Folder,Persistent): def getImageView(self,noredirect=None): """getImages; give Imageviewr and if somthing goes wrong.""" try: - fh=urllib.urlopen(self.metalink) + fh=ECHO_helpers.urlopen(self.metalink) dom=xml.dom.minidom.parse(fh) texttools=dom.getElementsByTagName('texttool') text=texttools[0].getElementsByTagName('image') @@ -997,7 +1206,7 @@ class ECHO_resource(Folder,Persistent): return ret def getInstitutionsHTML(self): - """gibt Liste der fšrdernden Institutionen aus""" + """gibt Liste der fšrdernden Institutionen aus""" if hasattr(self,'support'): obj=self.support @@ -1041,7 +1250,7 @@ class ECHO_resource(Folder,Persistent): def getCoords(self): - """gibt coordinaten als String zurück und löscht zugleich einträge die keine Koordinaten sind, letzteres zur korrektur der Eingabe der alten version""" + """gibt coordinaten als String zurück und löscht zugleich einträge die keine Koordinaten sind, letzteres zur korrektur der Eingabe der alten version""" retList=[] if hasattr(self,'coords'): for x in self.coords: @@ -1119,7 +1328,7 @@ class ECHO_resource(Folder,Persistent): def changeECHO_resource_main(self,metalink,link,title,label,description,contentType,renderingType,weight,resourceID,RESPONSE=None): - """Änderung der Properties""" + """Änderung der Properties""" self.resourceID=resourceID self.title=title self.label=label @@ -1137,7 +1346,7 @@ class ECHO_resource(Folder,Persistent): def changeECHO_resource_coords(self,coords,viewClassification,RESPONSE=None): - """Änderung der Properties - coords""" + """Änderung der Properties - coords""" if type(coords)==StringType: coords=[coords] @@ -1154,7 +1363,7 @@ class ECHO_resource(Folder,Persistent): RESPONSE.redirect('manage_main') def changeECHO_resource_credits(self,credits,responsible,copyrightType,RESPONSE=None): - """Änderung der Properties""" + """Änderung der Properties""" self.credits=credits self.responsible=responsible self.copyrightType=copyrightType @@ -1185,7 +1394,7 @@ class ECHO_resource(Folder,Persistent): except: """nothing""" - urllib.urlopen('http://xserve02.mpiwg-berlin.mpg.de:18880/echo_nav/storage/downloadExternalXML?index_meta_url=%s&xml_url=%s'%(self.metalink,self.absolute_url()+'/newMetaXML')).read() + ECHO_helpers.urlopen('http://xserve02.mpiwg-berlin.mpg.de:18880/echo_nav/storage/downloadExternalXML?index_meta_url=%s&xml_url=%s'%(self.metalink,self.absolute_url()+'/newMetaXML')).read() if RESPONSE is not None: @@ -1222,7 +1431,7 @@ class ECHO_resource(Folder,Persistent): return writeMetadata(self.metalink,self.metaDataHash) def changeECHO_resource(self,metalink,link,title,label,description,contentType,responsible,weight,viewClassification="",coords=None,credits=None,RESPONSE=None): - """Änderung der Properties""" + """Änderung der Properties""" try: coordsnew=[ string.split(x,",") for x in coords] @@ -1494,7 +1703,7 @@ class ECHO_resource(Folder,Persistent): if sp: return sp[1]() - #prüfen ob irgendwo ein template + #prüfen ob irgendwo ein template if hasattr(self,'startpage_index_template'): return self.startpage_index_template() @@ -1608,7 +1817,7 @@ class ECHO_externalLink(Folder): def changeECHO_externalLink(self,link,title,label,description,contentType,responsible,weight,coords=None,credits=None,RESPONSE=None): - """Änderung der Properties""" + """Änderung der Properties""" try: coordsnew=[ string.split(x,",") for x in coords] except: @@ -1722,7 +1931,8 @@ class ECHO_collection(Folder, Persistent return ret - + + def changeLabels(self): """change form""" pt=PageTemplateFile('Products/ECHO_content/zpt/changeLabelsForm').__of__(self) @@ -1735,6 +1945,17 @@ class ECHO_collection(Folder, Persistent pt.content_type="text/html" return pt() + def changeWeights(self): + """change form""" + pt=PageTemplateFile('Products/ECHO_content/zpt/changeWeightForm').__of__(self) + pt.content_type="text/html" + return pt() + + def changeMetaDataLinks(self): + """change form""" + pt=PageTemplateFile('Products/ECHO_content/zpt/changeMetaDataLinkForm').__of__(self) + pt.content_type="text/html" + return pt() def changeAccessRightsCollection(self): """change""" ret="" @@ -1750,7 +1971,22 @@ class ECHO_collection(Folder, Persistent """not""" return ret - def changeLabelsInCollection(self): + def changeMetaDataLinkInCollection(self): + """change all lables of a collection""" + ret="" + argv=self.REQUEST.form + + resources=self.ZopeFind(self,obj_metatypes=['ECHO_resource']) + for resource in resources: + + try: + ret+=resource[1].getId()+" "+argv[resource[1].getId()]+"
" + resource[1].metalink=argv[resource[1].getId()][0:] + except: + """not""" + return ret + + def changeMetaDataLinkInCollection(self): """change all lables of a collection""" ret="" argv=self.REQUEST.form @@ -1760,7 +1996,22 @@ class ECHO_collection(Folder, Persistent try: ret+=resource[1].getId()+" "+argv[resource[1].getId()]+"
" - resource[1].label=argv[resource[1].getId()][0:] + resource[1].metalink=argv[resource[1].getId()][0:] + except: + """not""" + return ret + + def changeWeightsInCollection(self): + """change all lables of a collection""" + ret="" + argv=self.REQUEST.form + + resources=self.ZopeFind(self,obj_metatypes=['ECHO_pageTemplate','ECHO_resource','ECHO_collection','ECHO_link','ECHO_externalLink']) + for resource in resources: + + try: + ret+=resource[1].getId()+" "+argv[resource[1].getId()]+"
" + resource[1].weight=argv[resource[1].getId()][0:] except: """not""" return ret @@ -1852,7 +2103,7 @@ class ECHO_collection(Folder, Persistent def isSelectedViewerTemplateSet(self,obj,id): - """is ausgewählt""" + """is ausgewählt""" if self.REQUEST['viewerTemplateSet']==id: return 1 @@ -2216,6 +2467,7 @@ class ECHO_collection(Folder, Persistent {'label':'Main Config','action':'ECHO_collection_config'}, {'label':'Change Labels','action':'changeLabels'}, {'label':'Change Titles','action':'changeTitles'}, + {'label':'Change Weights','action':'changeWeights'}, {'label':'Rerender Labels and Titles','action':'ECHO_rerenderLinksMDWarning'}, {'label':'Graphics','action':'ECHO_graphicEntry'}, {'label':'create resources from XML','action':'createRessourcesFromXMLForm'}, @@ -2511,7 +2763,7 @@ class ECHO_group(ECHO_collection): return pt() def changeECHO_group(self,title,label,description,contentType,responsible,weight,secondaryLink,secondaryLinkTitle,credits=None,sortfield="weight",coords=None,RESPONSE=None,imageTag="",bgcolour="",logo=""): - """Änderung der Properties""" + """Änderung der Properties""" self.secondaryLink=secondaryLink self.secondaryLinkTitle=secondaryLinkTitle @@ -2688,11 +2940,33 @@ class ECHO_root(Folder,Persistent,Implic security=ClassSecurityInfo() meta_type="ECHO_root" + + def changeWeightsInCollection(self): + """change all lables of a collection""" + ret="" + argv=self.REQUEST.form + + resources=self.ZopeFind(self,obj_metatypes=['ECHO_resource','ECHO_collection','ECHO_link','ECHO_externalLink']) + for resource in resources: + + try: + ret+=resource[1].getId()+" "+argv[resource[1].getId()]+"
" + resource[1].weight=argv[resource[1].getId()][0:] + except: + """not""" + return ret + def changeWeights(self): + """change form""" + pt=PageTemplateFile('Products/ECHO_content/zpt/changeWeightForm').__of__(self) + pt.content_type="text/html" + return pt() + getSubCols = ECHO_helpers.getSubCols manage_options=Folder.manage_options+( {'label':'Main Config','action':'ECHO_copyright_configForm'}, {'label':'Reload Metadata','action':'reloadMetaDataFromStorageWarning'}, + {'label':'Change Weights','action':'changeWeights'}, ) @@ -2740,7 +3014,7 @@ class ECHO_root(Folder,Persistent,Implic def showContent(self,path): """return content/html""" - return urllib.urlopen(path+"/content_html").read() + return ECHO_helpers.urlopen(path+"/content_html").read() def getImageViewers(self): """images""" @@ -2749,7 +3023,7 @@ class ECHO_root(Folder,Persistent,Implic def getBibTag(self,tag,content): - """get field tag für index-meta-generation""" + """get field tag für index-meta-generation""" if not content or content=="": return "" ret="<%s>"%tag @@ -2887,7 +3161,7 @@ class ECHO_root(Folder,Persistent,Implic """nothing""" return retStr - mediaTypes=['all','text','images','sound','video'] # konfigurierbar in neuer version über echoroot?? + mediaTypes=['all','text','images','sound','video'] # konfigurierbar in neuer version über echoroot?? def mediaTypeSelector_HTML(self,selected=None): """give type selector""" @@ -3029,18 +3303,20 @@ class ECHO_root(Folder,Persistent,Implic link.setAttribute("href",self.aq_parent.absolute_url()+"/vlp_coll?id="+ref) newxml=dom.toxml('utf-8') - match=re.sub(r'<\?xml version=\"1.0\" encoding=\"utf-8\"\?>\n','',newxml) - match=re.sub(r'','',match) - - return match + + retStr=regexpTXT.search(newxml) + + return retStr.group(1) + return "" - def xml2html(self,str): + def xml2html(self,str,quote="yes"): """link2html fuer VLP muss hier noch raus""" + print str if str: - - str=re.sub("\&","&",str) + if quote=="yes2": + str=re.sub("\&","&",str) dom=xml.dom.minidom.parseString(str) links=dom.getElementsByTagName("link") @@ -3051,7 +3327,12 @@ class ECHO_root(Folder,Persistent,Implic if self.checkRef(ref): link.setAttribute("href",self.aq_parent.absolute_url()+"/vlp_coll?id="+ref) - return dom.toxml('utf-8') + str= dom.toxml() + + #print link.toxml('utf-8') + retStr=regexpPage.search(str) + + return retStr.group(1) return "" def checkRef(self,ref): @@ -3487,15 +3768,15 @@ def manage_addECHO_linkList(self, id, la class ECHO_support(Folder): - """gefšrdert durch""" + """gefšrdert durch""" meta_type="ECHO_support" def __init__(self,id,institutions=None): """init""" - ##self.title=title self.id=id + self.title='' self.institutions=toList(institutions) def getSupporter(self): @@ -3517,7 +3798,6 @@ class ECHO_support(Folder): def ECHO_support_config(self,institutions=None,RESPONSE=None): """change""" - ##self.title=title self.institutions=toList(institutions) if RESPONSE is not None: @@ -3529,7 +3809,7 @@ def manage_addECHO_supportForm(self): pt=PageTemplateFile('Products/ECHO_content/zpt/AddECHO_support.zpt').__of__(self) return pt() -def manage_addECHO_support(self, id,title,institutions=None,RESPONSE=None): +def manage_addECHO_support(self, id,institutions=None,RESPONSE=None): """add the copyright""" meta_type="ECHO_support" @@ -3619,4 +3899,4 @@ def manage_addECHO_institution(self, id, try: url=self.DestinationURL() except: url=REQUEST['URL1'] REQUEST.RESPONSE.redirect('%s/manage_main' % url) - return id \ No newline at end of file + return id