--- ECHO_content/ECHO_xslt.py 2006/09/11 14:43:23 1.10 +++ ECHO_content/ECHO_xslt.py 2006/10/12 16:37:33 1.13 @@ -77,7 +77,7 @@ class getXML(Implicit): #print self.xsl - #print xsltproc.run(document) + #< xsltproc.run(document) tmp=xsltproc.run(document) self.result=tmp[0:] @@ -137,6 +137,7 @@ class ECHO_cache: if not fileStore: return None else: + return self.root[name].get(pn,None) @@ -252,7 +253,108 @@ class ECHO_xslt(ECHO_pageTemplate,ECHO_l fn=self.REQUEST['fn'] self.cache.deleteObject(fn) - def getPageLex(self,_pn="1",_caching=None): + + def createLinkNode(self,url,dom): + """createa a link node""" + txt=dom.createTextNode("") + node=dom.createElementNS("http://test.de","a") + node.setAttributeNS("http://test.de","href",url) + node.appendChild(txt) + return node + + def forwardLink(self,linkid,url,type="target",RESPONSE=None): + """forward to link""" + if RESPONSE: + RESPONSE.redirect(self.getLink(linkid,url,type=type)) + + else: + return self.getLink(linkid,url,type=type) + def getLink(self,linkid,url,type="target"): + """get target for linkid""" + dom=NonvalidatingReader.parseUri(url) + + masterurl=dom.xpath("//mpiwg:masterurl/@ref",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + slaveurl=dom.xpath("//mpiwg:slaveurl/@ref",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + + #check now if there are in the link file + + xp="//mpiwg:link[@id='%s']"%linkid + + if type=="target": + for link in dom.xpath(xp,explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'}): + fn=link.xpath("mpiwg:target/@filename",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + ref=link.xpath("mpiwg:target/@refid",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + + ref2=link.xpath("mpiwg:target/mpiwg:pagelink/@refid",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + selectionNodeIndex=link.xpath("mpiwg:target/mpiwg:pagelink/@selectionNodeIndex",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + + lstr=slaveurl+'fn='+fn+'&_id='+ref+'&_pagelink=%s///%s'%(ref2,selectionNodeIndex) + lstr+="&_links="+urllib.quote(url)+'&_linkid='+linkid+'&_linktype=target' + + else: + for link in dom.xpath(xp,explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'}): + fn=link.xpath("mpiwg:source/@filename",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + ref=link.xpath("mpiwg:source/@refid",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + + ref2=link.xpath("mpiwg:source/mpiwg:pagelink/@refid",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + selectionNodeIndex=link.xpath("mpiwg:source/mpiwg:pagelink/@selectionNodeIndex",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + + lstr=masterurl+'fn='+fn+'&_id='+ref+'&_pagelink=%s///%s'%(ref2,selectionNodeIndex) + lstr+="&_links="+urllib.quote(url)+'&_linkid='+linkid+'&_linktype=source' + return lstr + + def addLinks(self,txt,url="http://127.0.0.1:8080/HFQP/linkCreator/getCollectionXML?collection=commentary2"): + """add links to a page from xml linkfile""" + + dom=NonvalidatingReader.parseUri(url) + textDom=NonvalidatingReader.parseString(txt) + + #find ids in txt + ids=textDom.xpath("//*[@id]") + masterurl=dom.xpath("//mpiwg:masterurl/@ref",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + slaveurl=dom.xpath("//mpiwg:slaveurl/@ref",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + + #check now if there are in the link file + for textid in ids: + xp="//mpiwg:link[mpiwg:source/@refid='%s']"%textid.xpath("@id")[0].value + for link in dom.xpath(xp,explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'}): + fn=link.xpath("mpiwg:target/@filename",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + ref=link.xpath("mpiwg:target/@refid",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + + ref2=link.xpath("mpiwg:target/mpiwg:pagelink/@refid",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + selectionNodeIndex=link.xpath("mpiwg:target/mpiwg:pagelink/@selectionNodeIndex",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + linkid=link.xpath("@id")[0].value + lstr=slaveurl+'fn='+fn+'&_id='+ref+'&_pagelink=%s///%s'%(ref2,selectionNodeIndex) + lstr+="&_links="+urllib.quote(url)+'&_url='+url+'&_linkid='+linkid+'&_linktype=target' + node=self.createLinkNode(lstr,textDom) + textid.parentNode.insertBefore(node,textid) + + + xp="//mpiwg:link[mpiwg:target/@refid='%s']"%textid.xpath("@id")[0].value + for link in dom.xpath(xp,explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'}): + fn=link.xpath("mpiwg:source/@filename",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + ref=link.xpath("mpiwg:source/@refid",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + + ref2=link.xpath("mpiwg:source/mpiwg:pagelink/@refid",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + selectionNodeIndex=link.xpath("mpiwg:source/mpiwg:pagelink/@selectionNodeIndex",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'})[0].value + linkid=link.xpath("@id")[0].value + lstr=masterurl+'fn='+fn+'&_id='+ref+'&_pagelink=%s///%s'%(ref2,selectionNodeIndex) + lstr+="&_links="+urllib.quote(url)+'&_linkid='+linkid+'&_linktype=source' + + node=self.createLinkNode(lstr,textDom) + textid.parentNode.insertBefore(node,textid) + + + + strio = StringIO() + PrettyPrint(textDom,strio) + xmlstr = strio.getvalue() + + return xmlstr + + + + def getPageLex(self,_pn="1",_id=None,_caching=None,_links=None): """getpage mit lexikalischer analyse und xslt transform if _caching=yes dann wird die lwxikalisch analysierte seite in einem cache abgespeichert """ @@ -262,18 +364,21 @@ class ECHO_xslt(ECHO_pageTemplate,ECHO_l fn=self.REQUEST['fn'] - - fromCache=self.cache.retrieveObject(fn,_pn) - - if fromCache and _caching=="yes": - - txt = fromCache + if not _id: + + fromCache=self.cache.retrieveObject(fn,_pn) + + if fromCache and _caching=="yes": + + txt = fromCache + else: + txt=self.tagLex(nr=_pn) + + self.cache.storeObject(fn,_pn,txt[0:]) + else: - txt=self.tagLex(nr=_pn) - - self.cache.storeObject(fn,_pn,txt[0:]) - - + txt=self.tagLex(id=_id) + xsl=self.xslt() xsltproc=Processor() @@ -285,6 +390,14 @@ class ECHO_xslt(ECHO_pageTemplate,ECHO_l xsltproc.appendStylesheet(stylesheet) tmp=xsltproc.run(document) + if _links: + _links=urllib.unquote(_links) + tmp=self.addLinks(tmp,url=_links) + + #bugfix for digilib images which doesn't accept & + tmp=tmp.replace("&","&") + + return tmp[0:] def getTextInput(self): @@ -309,14 +422,14 @@ class ECHO_xslt(ECHO_pageTemplate,ECHO_l #return InputSource.InputSource(fact) #xmlt=urllib.urlopen(qs).read() - def getPage(self,_pn,REQUEST=None,_caching=None): + def getPage(self,_pn="-1",_id=None,REQUEST=None,_caching=None): """get a page from an xml""" if not _caching: _caching=self.caching pn=int(_pn)-1 - if pn<0: + if pn<0 and (not _id): if REQUEST: return "Sorry, pagenumbers have to be greater than 0" else: @@ -335,14 +448,33 @@ class ECHO_xslt(ECHO_pageTemplate,ECHO_l self.cache.storeObject(self.baseUri,"-1",txt) + dom=NonvalidatingReader.parseString(txt,self.baseUri) #pb should have a namespache - + pbs=dom.xpath("//mpiwg:pb",explicitNss={'mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'}) + if len(pbs)==0: # versuche nochmal ohne pbs=dom.xpath("//pb") + if _id: + #suche wieviele pb for der id + + + idpb=dom.xpath("//*[@id='%s']/preceding::node()/mpiwg:pb"%_id,explicitNss={'html':'http://test.de','mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'}) + if len(idpb)==0: + idpb=dom.xpath("//*[@id='%s']/preceding::node()/pb"%_id) + + if len(idpb)==0: + k=0 + for node in dom.xpath("//*[@id='%s']//preceding::node()"%_id,explicitNss={'html':'http://test.de','mpiwg':'http://www.mpiwg-berlin.mpg.de/namespace'}): + if getattr(node,'tagName',"")=="mpiwg:pb": + k+=1 + else: + k=len(idpb) + pn=k-1 #-1 wegen Seitenzahlzaehlung startet mit 0 + if pn > len(pbs): if REQUEST: return "Sorry, pagenumber %s does not exit"%(pn+1)