--- documentViewer/MpdlXmlTextServer.py 2011/01/13 13:45:38 1.223 +++ documentViewer/MpdlXmlTextServer.py 2011/04/28 09:58:39 1.235 @@ -3,6 +3,7 @@ from OFS.SimpleItem import SimpleItem from Products.PageTemplates.PageTemplateFile import PageTemplateFile from Ft.Xml import EMPTY_NAMESPACE, Parse +import md5 import sys import logging import urllib @@ -19,7 +20,9 @@ class MpdlXmlTextServer(SimpleItem): manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals()) - def __init__(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40): + def __init__(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40): + #def __init__(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/", serverName=None, timeout=40): + """constructor""" self.id=id self.title=title @@ -38,7 +41,7 @@ class MpdlXmlTextServer(SimpleItem): url = self.serverUrl+method return documentViewer.getHttpData(url,data,timeout=self.timeout) - def getSearch(self, pageinfo=None, docinfo=None, lemma=None): + def getSearch(self, pageinfo=None, docinfo=None): """get search list""" docpath = docinfo['textURLPath'] url = docinfo['url'] @@ -54,9 +57,7 @@ class MpdlXmlTextServer(SimpleItem): optionToggle = pageinfo['optionToggle'] tocPN = pageinfo['tocPN'] selfurl = self.absolute_url() - data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&characterNormalization=%s&optionToggle=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,characterNormalization,optionToggle ,urllib.quote(highlightQuery))) - pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url) pagedom = Parse(pagexml) @@ -86,6 +87,7 @@ class MpdlXmlTextServer(SimpleItem): selfurl = self.absolute_url() pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle,characterNormalization)) hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) + #logging.debug("PUREXML :%s"%(serializeNode(pagenode))) return serializeNode(pagenode) if (queryType=="fulltextMorph"): pagedivs = pagedom.xpath("//div[@class='queryResult']") @@ -101,10 +103,10 @@ class MpdlXmlTextServer(SimpleItem): pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,optionToggle,characterNormalization)) hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) if href.startswith('../lt/lemma.xql'): - hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma_New'%(selfurl)) + hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_query'%(selfurl)) l.setAttributeNS(None, 'target', '_blank') l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") - l.setAttributeNS(None, 'onClick', 'popupWin.focus();') + l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']") return serializeNode(pagenode) if (queryType=="ftIndex")or(queryType=="ftIndexMorph"): @@ -118,18 +120,17 @@ class MpdlXmlTextServer(SimpleItem): href = hrefNode.nodeValue hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&optionToggle=%s&characterNormalization=%s'%(viewMode,tocMode,tocPN,pn,optionToggle,characterNormalization)) if href.startswith('../lt/lex.xql'): - hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%selfurl) + hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_lex'%selfurl) l.setAttributeNS(None, 'target', '_blank') l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") - l.setAttributeNS(None, 'onClick', 'popupWin.focus();') + l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') if href.startswith('../lt/lemma.xql'): - hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%selfurl) + hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%(selfurl)) l.setAttributeNS(None, 'target', '_blank') l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") - l.setAttributeNS(None, 'onClick', 'popupWin.focus();') + l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') return serializeNode(pagenode) return "no text here" - def getGisPlaces(self, docinfo=None, pageinfo=None): """ Show all Gis Places of whole Page""" @@ -151,7 +152,7 @@ class MpdlXmlTextServer(SimpleItem): href= hrefNode.nodeValue hrefList.append(href) myList = ",".join(hrefList) - logging.debug("getGisPlaces :%s"%(myList)) + #logging.debug("getGisPlaces :%s"%(myList)) return myList def getAllGisPlaces (self, docinfo=None, pageinfo=None): @@ -172,21 +173,21 @@ class MpdlXmlTextServer(SimpleItem): href= hrefNode.nodeValue hrefList.append(href) myList = ",".join(hrefList) - logging.debug("getALLGisPlaces :%s"%(myList)) + #logging.debug("getALLGisPlaces :%s"%(myList)) return myList - def getTextPage(self, docinfo=None, pageinfo=None): + def getTextPage(self, mode="text_dict", pn=1, docinfo=None, pageinfo=None): """returns single page from fulltext""" docpath = docinfo['textURLPath'] path = docinfo['textURLPath'] url = docinfo.get('url',None) name = docinfo.get('name',None) pn =pageinfo['current'] - viewMode= pageinfo['viewMode'] sn = pageinfo['sn'] + optionToggle =pageinfo ['optionToggle'] highlightQuery = pageinfo['highlightQuery'] - mode = pageinfo ['viewMode'] + #mode = pageinfo ['viewMode'] tocMode = pageinfo['tocMode'] characterNormalization=pageinfo['characterNormalization'] tocPN = pageinfo['tocPN'] @@ -196,20 +197,26 @@ class MpdlXmlTextServer(SimpleItem): else: textmode = mode - textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization) + textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s&optionToggle=%s"%(docpath,textmode,pn,characterNormalization,optionToggle) if highlightQuery is not None: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn) pagexml = self.getServerData("page-fragment.xql",textParam) dom = Parse(pagexml) - #original Pages - pagedivs = dom.xpath("//div[@class='pageNumberOrig']") - if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"): + pagedivs = dom.xpath("//div[@class='pageNumberOrig']") + """if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"): if len(pagedivs)>0: docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0]) - logging.debug("ORIGINAL PAGES: %s"%(docinfo['pageNumberOrig'])) - + logging.debug("ORIGINAL PAGE: %s"%(docinfo['pageNumberOrig'])) + + #original Pages Norm + pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']") + if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"): + if len(pagedivs)>0: + docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0]) + logging.debug("ORIGINAL PAGE NORM: %s"%(docinfo['pageNumberOrigNorm'])) + """ #figureEntries pagedivs = dom.xpath("//div[@class='countFigureEntries']") if pagedivs == dom.xpath("//div[@class='countFigureEntries']"): @@ -219,14 +226,12 @@ class MpdlXmlTextServer(SimpleItem): if s=='0': try: docinfo['countFigureEntries'] = int(s) - logging.debug("FIGURE ENTRIES: %s"%(s)) except: docinfo['countFigureEntries'] = 0 else: s1 = int(s)/30+1 try: docinfo['countFigureEntries'] = int(s1) - logging.debug("FIGURE ENTRIES: %s"%(s1)) except: docinfo['countFigureEntries'] = 0 @@ -238,7 +243,6 @@ class MpdlXmlTextServer(SimpleItem): s = getTextFromNode(pagedivs[0]) try: docinfo['countPlaces'] = int(s) - logging.debug("PLACES HERE: %s"%(s)) except: docinfo['countPlaces'] = 0 @@ -251,14 +255,12 @@ class MpdlXmlTextServer(SimpleItem): if s=='0': try: docinfo['countTocEntries'] = int(s) - logging.debug("TEXT ENTRIES: %s"%(s)) except: docinfo['countTocEntries'] = 0 else: s1 = int(s)/30+1 try: docinfo['countTocEntries'] = int(s1) - logging.debug("TEXT ENTRIES: %s"%(s1)) except: docinfo['countTocEntries'] = 0 @@ -271,7 +273,7 @@ class MpdlXmlTextServer(SimpleItem): try: docinfo['numPages'] = int(s) - logging.debug("PAGE NUMBER: %s"%(s)) + #logging.debug("PAGE NUMBER: %s"%(s)) np = docinfo['numPages'] pageinfo['end'] = min(pageinfo['end'], np) @@ -288,6 +290,7 @@ class MpdlXmlTextServer(SimpleItem): docinfo['countPlaces'] = 0 docinfo['countTocEntries'] = 0 docinfo['numPages'] = 0 + docinfo['pageNumberOrigNorm'] = 0 #return docinfo # plain text mode @@ -302,7 +305,7 @@ class MpdlXmlTextServer(SimpleItem): if hrefNode: href= hrefNode.nodeValue if href.startswith('#note-'): - hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn)) + hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=text&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn)) return serializeNode(pagenode) if mode == "xml": # first div contains text @@ -334,6 +337,7 @@ class MpdlXmlTextServer(SimpleItem): # text-with-links mode if mode == "text_dict": # first div contains text + #mode = pageinfo ['viewMode'] pagedivs = dom.xpath("/div") if len(pagedivs) > 0: pagenode = pagedivs[0] @@ -344,42 +348,73 @@ class MpdlXmlTextServer(SimpleItem): if hrefNode: # is link with href href = hrefNode.nodeValue - if href.startswith('lt/lex.xql'): + if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql'): # is pollux link selfurl = self.absolute_url() # change href - hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl) + hrefNode.nodeValue = href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql','%s/head_main_voc'%selfurl) # add target l.setAttributeNS(None, 'target', '_blank') - l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;") - l.setAttributeNS(None, 'onClick', 'popupWin.focus();') - if href.startswith('lt/lemma.xql'): + l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;") + l.setAttributeNS(None, "onDblclick", "popupWin.focus();") + #window.open("this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=yes, scrollbars=1'"); return false;") + + if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql'): selfurl = self.absolute_url() - hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl) + hrefNode.nodeValue = href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql','%s/head_main_lemma'%selfurl) l.setAttributeNS(None, 'target', '_blank') - l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;") - l.setAttributeNS(None, 'onClick', 'popupWin.focus();') + l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;") + l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') + if href.startswith('#note-'): - hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn)) + hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=text_dict&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,tocMode,tocPN,pn)) return serializeNode(pagenode) return "no text here" + + def getOrigPages(self, docinfo=None, pageinfo=None): + docpath = docinfo['textURLPath'] + pn =pageinfo['current'] + selfurl = self.absolute_url() + pagexml = self.getServerData("page-fragment.xql","document=%s&pn=%s"%(docpath, pn)) + dom = Parse(pagexml) + pagedivs = dom.xpath("//div[@class='pageNumberOrig']") + if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"): + if len(pagedivs)>0: + docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0]) + return docinfo['pageNumberOrig'] + + def getOrigPagesNorm(self, docinfo=None, pageinfo=None): + docpath = docinfo['textURLPath'] + pn =pageinfo['current'] + selfurl = self.absolute_url() + pagexml = self.getServerData("page-fragment.xql","document=%s&pn=%s"%(docpath, pn)) + dom = Parse(pagexml) + pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']") + if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"): + if len(pagedivs)>0: + docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0]) + return docinfo['pageNumberOrigNorm'] - def getTranslate(self, query=None, language=None): + + def getTranslate(self, word=None, language=None): """translate into another languages""" - data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query)) + data = self.getServerData("lt/wordInfo.xql","language="+str(language)+"&word="+urllib.quote(word)+"&output=html") #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query))) return data def getLemma(self, lemma=None, language=None): """simular words lemma """ - data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma)) - #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma))) + data = self.getServerData("lt/lemma.xql","language="+str(language)+"&lemma="+urllib.quote(lemma)+"&output=html") return data - def getLemmaNew(self, query=None, language=None): + def getLemmaQuery(self, query=None, language=None): """simular words lemma """ - data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query)) - #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query))) + data = self.getServerData("lt/lemma.xql","language="+str(language)+"&query="+urllib.quote(query)+"&output=html") + return data + + def getLex(self, query=None, language=None): + #simular words lemma + data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query)) return data def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1, optionToggle=None): @@ -396,14 +431,11 @@ class MpdlXmlTextServer(SimpleItem): pagedom = Parse(pagexml) numdivs = pagedom.xpath("//div[@class='queryResultHits']") tocSearch = int(getTextFromNode(numdivs[0])) - logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch)) tc=int((tocSearch/10)+1) - logging.debug("documentViewer (gettoc) tc: %s"%(tc)) return tc def getToc(self, mode="text", docinfo=None): """loads table of contents and stores in docinfo""" - logging.debug("documentViewer (gettoc) mode: %s"%(mode)) if mode == "none": return docinfo if 'tocSize_%s'%mode in docinfo: @@ -455,11 +487,10 @@ class MpdlXmlTextServer(SimpleItem): data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=regPlusNorm&optionToggle=1"%(docpath,queryType, pagesize, pn)) page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&optionToggle=1'%(selfurl,url, viewMode, tocMode, tocPN)) text = page.replace('mode=image','mode=texttool') - #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization)) - return text - def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): + def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): + #def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/",timeout=40,RESPONSE=None): """change settings""" self.title=title self.timeout = timeout @@ -473,12 +504,10 @@ def manage_addMpdlXmlTextServerForm(self pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self) return pt() -def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): +def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): +#def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/",timeout=40,RESPONSE=None): """add zogiimage""" newObj = MpdlXmlTextServer(id,title,serverUrl,timeout) self.Destination()._setObject(id, newObj) if RESPONSE is not None: - RESPONSE.redirect('manage_main') - - - + RESPONSE.redirect('manage_main') \ No newline at end of file