--- documentViewer/MpdlXmlTextServer.py 2010/10/14 11:56:31 1.124 +++ documentViewer/MpdlXmlTextServer.py 2011/08/03 19:04:18 1.238.2.6 @@ -1,13 +1,63 @@ from OFS.SimpleItem import SimpleItem from Products.PageTemplates.PageTemplateFile import PageTemplateFile + from Ft.Xml import EMPTY_NAMESPACE, Parse +from Ft.Xml.Domlette import NonvalidatingReader +import Ft.Xml.Domlette +import cStringIO + +import xml.etree.ElementTree as ET -import sys +import re import logging import urllib -import documentViewer -from documentViewer import getTextFromNode, serializeNode + +from SrvTxtUtils import getInt, getText, getHttpData + +def serialize(node): + """returns a string containing an XML snippet of node""" + s = ET.tostring(node, 'UTF-8') + # snip off XML declaration + if s.startswith('') + return s[i+3:] + + return s + + +def getTextFromNode(node): + """get the cdata content of a node""" + if node is None: + return "" + # ET: +# text = node.text or "" +# for e in node: +# text += gettext(e) +# if e.tail: +# text += e.tail + + # 4Suite: + nodelist=node.childNodes + text = "" + for n in nodelist: + if n.nodeType == node.TEXT_NODE: + text = text + n.data + + return text + +def serializeNode(node, encoding="utf-8"): + """returns a string containing node as XML""" + #s = ET.tostring(node) + + # 4Suite: + stream = cStringIO.StringIO() + Ft.Xml.Domlette.Print(node, stream=stream, encoding=encoding) + s = stream.getvalue() + stream.close() + + return s + class MpdlXmlTextServer(SimpleItem): """TextServer implementation for MPDL-XML eXist server""" @@ -19,7 +69,7 @@ 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): """constructor""" self.id=id self.title=title @@ -31,21 +81,21 @@ class MpdlXmlTextServer(SimpleItem): def getHttpData(self, url, data=None): """returns result from url+data HTTP request""" - return documentViewer.getHttpData(url,data,timeout=self.timeout) + return getHttpData(url,data,timeout=self.timeout) def getServerData(self, method, data=None): """returns result from text server for method+data""" url = self.serverUrl+method - return documentViewer.getHttpData(url,data,timeout=self.timeout) + return getHttpData(url,data,timeout=self.timeout) - def getSearch(self, pn=1, pageinfo=None, docinfo=None, query=None, queryType=None, lemma=None): + # WTF: what does this really do? can it be integrated in getPage? + def getSearch(self, pageinfo=None, docinfo=None): """get search list""" + logging.debug("getSearch()") docpath = docinfo['textURLPath'] url = docinfo['url'] - logging.debug("documentViewer (gettoc) docpath: %s"%(docpath)) - logging.debug("documentViewer (gettoc) url: %s"%(url)) pagesize = pageinfo['queryPageSize'] - pn = pageinfo['searchPN'] + pn = pageinfo.get('searchPN',1) sn = pageinfo['sn'] highlightQuery = pageinfo['highlightQuery'] query =pageinfo['query'] @@ -53,14 +103,26 @@ class MpdlXmlTextServer(SimpleItem): viewMode= pageinfo['viewMode'] tocMode = pageinfo['tocMode'] characterNormalization = pageinfo['characterNormalization'] + #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&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,urllib.quote(highlightQuery))) - #page=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&highlightQuery=%s"%(docpath, 'text', queryType, query, pagesize, pn, sn, viewMode,highlightQuery) ,outputUnicode=False) - + data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&characterNormalization=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,characterNormalization, urllib.quote(highlightQuery))) pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url) pagedom = Parse(pagexml) + + """ + pagedivs = pagedom.xpath("//div[@class='queryResultHits']") + if (pagedivs == pagedom.xpath("//div[@class='queryResultHits']")): + if len(pagedivs)>0: + docinfo['queryResultHits'] = int(getTextFromNode(pagedivs[0])) + s = getTextFromNode(pagedivs[0]) + s1 = int(s)/10+1 + try: + docinfo['queryResultHits'] = int(s1) + logging.debug("SEARCH ENTRIES: %s"%(s1)) + except: + docinfo['queryResultHits'] = 0 + """ if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"): pagedivs = pagedom.xpath("//div[@class='queryResultPage']") if len(pagedivs)>0: @@ -72,8 +134,9 @@ class MpdlXmlTextServer(SimpleItem): href = hrefNode.nodeValue if href.startswith('page-fragment.xql'): 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'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN)) + pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN, 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']") @@ -86,13 +149,13 @@ class MpdlXmlTextServer(SimpleItem): href = hrefNode.nodeValue if href.startswith('page-fragment.xql'): 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'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN)) + pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,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"): @@ -104,186 +167,272 @@ class MpdlXmlTextServer(SimpleItem): hrefNode = l.getAttributeNodeNS(None, u"href") if hrefNode: href = hrefNode.nodeValue - hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s'%(viewMode,tocMode,tocPN,pn)) + hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&characterNormalization=%s'%(viewMode,tocMode,tocPN,pn,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 getNumPages(self, docinfo): - ""get list of pages from fulltext and put in docinfo"" - if 'numPages' in docinfo: - # already there - return docinfo - xquery = '//pb' - text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery)) - docinfo['numPages'] = text.count(" 0: + docinfo['numTextPages'] = np + if docinfo.get('numPages', 0) == 0: + # seems to be text-only - update page count + docinfo['numPages'] = np + pageinfo['end'] = min(pageinfo['end'], np) + pageinfo['numgroups'] = int(np / pageinfo['groupsize']) + if np % pageinfo['groupsize'] > 0: + pageinfo['numgroups'] += 1 + + return + + + def getTextPage(self, mode="text_dict", pn=1, docinfo=None, pageinfo=None): """returns single page from fulltext""" + logging.debug("getTextPage mode=%s, pn=%s"%(mode,pn)) + # check for cached text -- but this shouldn't be called twice + if pageinfo.has_key('textPage'): + logging.debug("getTextPage: using cached text") + return pageinfo['textPage'] + docpath = docinfo['textURLPath'] - path = docinfo['textURLPath'] + # just checking + if pageinfo['current'] != pn: + logging.warning("getTextPage: current!=pn!") + + # stuff for constructing full urls url = docinfo['url'] - name = docinfo['name'] - viewMode= pageinfo['viewMode'] - tocMode = pageinfo['tocMode'] - characterNormalization=pageinfo['characterNormalization'] - tocPN = pageinfo['tocPN'] - selfurl = self.absolute_url() + urlmode = docinfo['mode'] + sn = pageinfo.get('sn', None) + highlightQuery = pageinfo.get('highlightQuery', None) + tocMode = pageinfo.get('tocMode', None) + tocPN = pageinfo.get('tocPN',None) + characterNormalization = pageinfo.get('characterNormalization', None) + selfurl = docinfo['viewerUrl'] + if mode == "text_dict": + # text_dict is called textPollux in the backend textmode = "textPollux" else: textmode = mode - #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization)) + textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization) - if highlightQuery is not None: + if highlightQuery: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn) + # fetch the page pagexml = self.getServerData("page-fragment.xql",textParam) - #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False) + dom = ET.fromstring(pagexml) + # extract additional info + self.processPageInfo(dom, docinfo, pageinfo) + # page content is in
+ pagediv = None + # ElementTree 1.2 in Python 2.6 can't do div[@class='pageContent'] + # so we look at the second level divs + alldivs = dom.findall("div") + for div in alldivs: + dc = div.get('class') + # page content div + if dc == 'pageContent': + pagediv = div + break - pagedom = Parse(pagexml) # plain text mode if mode == "text": - # first div contains text - pagedivs = pagedom.xpath("/div") - if len(pagedivs) > 0: - pagenode = pagedivs[0] - links = pagenode.xpath("//a") + if pagediv: + links = pagediv.findall(".//a") for l in links: - hrefNode = l.getAttributeNodeNS(None, u"href") - 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)) - return serializeNode(pagenode) - if mode == "xml": - # first div contains text - pagedivs = pagedom.xpath("/div") - if len(pagedivs) > 0: - pagenode = pagedivs[0] - return serializeNode(pagenode) - if mode == "gis": - # first div contains text - pagedivs = pagedom.xpath("/div") - if len(pagedivs) > 0: - pagenode = pagedivs[0] - links =pagenode.xpath("//a") - for l in links: - hrefNode =l.getAttributeNodeNS(None, u"href") - if hrefNode: - href=hrefNode.nodeValue - if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'): - hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name) - l.setAttributeNS(None, 'target', '_blank') - return serializeNode(pagenode) - - if mode == "pureXml": - # first div contains text - pagedivs = pagedom.xpath("/div") - if len(pagedivs) > 0: - pagenode = pagedivs[0] - return serializeNode(pagenode) + href = l.get('href') + if href and href.startswith('#note-'): + href = href.replace('#note-',"?mode=%s&url=%s&viewMode=text&tocMode=%s&tocPN=%s&pn=%s#note-"%(urlmode,url,tocMode,tocPN,pn)) + l.set('href', href) + + return serialize(pagediv) + # text-with-links mode - if mode == "text_dict": - # first div contains text - pagedivs = pagedom.xpath("/div") - if len(pagedivs) > 0: - pagenode = pagedivs[0] + elif mode == "text_dict": + if pagediv: # check all a-tags - links = pagenode.xpath("//a") + links = pagediv.findall(".//a") for l in links: - hrefNode = l.getAttributeNodeNS(None, u"href") - if hrefNode: + href = l.get('href') + + if href: # 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) + l.set('href', 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.set('target', '_blank') + + 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) - 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.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql','%s/head_main_lemma'%selfurl)) + l.set('target', '_blank') + l.set('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.set('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)) - return serializeNode(pagenode) + l.set('href', href.replace('#note-',"?mode=%s&url=%s&viewMode=text_dict&tocMode=%s&tocPN=%s&pn=%s#note-"%(urlmode,url,tocMode,tocPN,pn))) + + return serialize(pagediv) + + # xml mode + elif mode == "xml": + if pagediv: + return serialize(pagediv) + + # pureXml mode + elif mode == "pureXml": + if pagediv: + return serialize(pagediv) + + # gis mode + elif mode == "gis": + name = docinfo['name'] + if pagediv: + # check all a-tags + links = pagediv.findall(".//a") + for l in links: + href = l.get('href') + if href: + if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'): + l.set('href', href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)) + l.set('target', '_blank') + + return serialize(pagediv) + return "no text here" - - def getTranslate(self, query=None, language=None): + + # WTF: is this needed? + def getOrigPages(self, docinfo=None, pageinfo=None): + logging.debug("CALLED: getOrigPages!") + if not pageinfo.has_key('pageNumberOrig'): + logging.warning("getOrigPages: not in pageinfo!") + return None + + return pageinfo['pageNumberOrig'] + + # WTF: is this needed? + def getOrigPagesNorm(self, docinfo=None, pageinfo=None): + logging.debug("CALLED: getOrigPagesNorm!") + if not pageinfo.has_key('pageNumberOrigNorm'): + logging.warning("getOrigPagesNorm: not in pageinfo!") + return None + + return pageinfo['pageNumberOrigNorm'] + + # TODO: should be getWordInfo + 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)) - #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query))) + data = self.getServerData("lt/wordInfo.xql","language="+str(language)+"&word="+urllib.quote(word)+"&output=html") return data + # WTF: what does this do? 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): + # WTF: what does this do? + 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 + # WTF: what does this do? + 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 + + # WTF: what does this do? def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1): - """number of""" + #number of docpath = docinfo['textURLPath'] pagesize = pageinfo['queryPageSize'] pn = pageinfo['searchPN'] @@ -293,19 +442,18 @@ class MpdlXmlTextServer(SimpleItem): tocDiv = None pagexml = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn)) - #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, query, pagesize, pn) ,outputUnicode=False) pagedom = Parse(pagexml) numdivs = pagedom.xpath("//div[@class='queryResultHits']") tocSearch = int(getTextFromNode(numdivs[0])) 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)) + """loads table of contents and stores XML in docinfo""" + logging.debug("getToc mode=%s"%mode) if mode == "none": - return docinfo + return docinfo + if 'tocSize_%s'%mode in docinfo: # cached toc return docinfo @@ -321,44 +469,81 @@ class MpdlXmlTextServer(SimpleItem): # number of entries in toc tocSize = 0 tocDiv = None - + # fetch full toc pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn)) - #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False) - # post-processing downloaded xml - pagedom = Parse(pagexml) - # get number of entries - numdivs = pagedom.xpath("//div[@class='queryResultHits']") - if len(numdivs) > 0: - tocSize = int(getTextFromNode(numdivs[0])) - docinfo['tocSize_%s'%mode] = tocSize + dom = ET.fromstring(pagexml) + # page content is in
+ pagediv = None + # ElementTree 1.2 in Python 2.6 can't do div[@class='queryResultPage'] + alldivs = dom.findall("div") + for div in alldivs: + dc = div.get('class') + # page content div + if dc == 'queryResultPage': + pagediv = div + + elif dc == 'queryResultHits': + docinfo['tocSize_%s'%mode] = getInt(div.text) + + if pagediv: + # store XML in docinfo + docinfo['tocXML_%s'%mode] = ET.tostring(pagediv, 'UTF-8') + return docinfo def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None): """returns single page from the table of contents""" - # TODO: this should use the cached TOC + logging.debug("getTocPage mode=%s, pn=%s"%(mode,pn)) if mode == "text": queryType = "toc" else: queryType = mode - docpath = docinfo['textURLPath'] - path = docinfo['textURLPath'] - pagesize = pageinfo['tocPageSize'] - pn = pageinfo['tocPN'] + + # check for cached TOC + if not docinfo.has_key('tocXML_%s'%mode): + self.getToc(mode=mode, docinfo=docinfo) + + tocxml = docinfo.get('tocXML_%s'%mode, None) + if not tocxml: + logging.error("getTocPage: unable to find tocXML") + return "No ToC" + + pagesize = int(pageinfo['tocPageSize']) url = docinfo['url'] - selfurl = self.absolute_url() + urlmode = docinfo['mode'] + selfurl = docinfo['viewerUrl'] viewMode= pageinfo['viewMode'] - characterNormalization = pageinfo ['characterNormalization'] tocMode = pageinfo['tocMode'] - tocPN = pageinfo['tocPN'] + tocPN = int(pageinfo['tocPN']) + pn = tocPN + + fulltoc = ET.fromstring(tocxml) - data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization)) - page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN)) - text = page.replace('mode=image','mode=texttool') - logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization)) - #logging.debug("documentViewer (characterNormalization) text: %s"%(text)) - return text + if fulltoc: + # paginate + start = (pn - 1) * pagesize * 2 + len = pagesize * 2 + del fulltoc[:start] + del fulltoc[len:] + tocdivs = fulltoc + + # check all a-tags + links = tocdivs.findall(".//a") + for l in links: + href = l.get('href') + if href: + # take pn from href + m = re.match(r'page-fragment\.xql.*pn=(\d+)', href) + if m is not None: + # and create new url + l.set('href', '%s?mode=%s&url=%s&viewMode=%s&pn=%s&tocMode=%s&tocPN=%s'%(selfurl, urlmode, url, viewMode, m.group(1), tocMode, tocPN)) + else: + logging.warning("getTocPage: Problem with link=%s"%href) + + return serialize(tocdivs) + - 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): """change settings""" self.title=title self.timeout = timeout @@ -372,12 +557,12 @@ 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') - - - + + \ No newline at end of file