--- documentViewer/documentViewer.py 2010/06/18 19:41:28 1.65.2.2 +++ documentViewer/documentViewer.py 2010/06/01 14:31:26 1.66 @@ -2,20 +2,31 @@ from OFS.Folder import Folder from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate from Products.PageTemplates.PageTemplateFile import PageTemplateFile +from Products.PythonScripts.standard import url_quote from AccessControl import ClassSecurityInfo from AccessControl import getSecurityManager from Globals import package_home +from Ft.Xml.Domlette import NonvalidatingReader +from Ft.Xml.Domlette import PrettyPrint, Print from Ft.Xml import EMPTY_NAMESPACE, Parse -import Ft.Xml.Domlette + +from xml.dom.minidom import parse, parseString + + + +import Ft.Xml.XPath +import cStringIO +import xmlrpclib import os.path import sys +import cgi import urllib -import urllib2 import logging import math + import urlparse -import cStringIO +from types import * def logger(txt,method,txt2): """logging""" @@ -43,7 +54,7 @@ def getTextFromNode(nodename): def serializeNode(node, encoding='utf-8'): """returns a string containing node as XML""" buf = cStringIO.StringIO() - Ft.Xml.Domlette.Print(node, stream=buf, encoding=encoding) + Print(node, stream=buf, encoding=encoding) s = buf.getvalue() buf.close() return s @@ -54,49 +65,14 @@ def getParentDir(path): return '/'.join(path.split('/')[0:-1]) -def getHttpData(url, data=None, num_tries=3, timeout=10): - """returns result from url+data HTTP request""" - # we do GET (by appending data to url) - if isinstance(data, str) or isinstance(data, unicode): - # if data is string then append - url = "%s?%s"%(url,data) - elif isinstance(data, dict) or isinstance(data, list) or isinstance(data, tuple): - # urlencode - url = "%s?%s"%(url,urllib.urlencode(data)) - - response = None - errmsg = None - for cnt in range(num_tries): - try: - logging.debug("getHttpData(#%s %ss) url=%s"%(cnt+1,timeout,url)) - if sys.version_info < (2, 6): - # set timeout on socket -- ugly :-( - import socket - socket.setdefaulttimeout(float(timeout)) - response = urllib2.urlopen(url) - else: - response = urllib2.urlopen(url,timeout=float(timeout)) - # check result? - break - except urllib2.HTTPError, e: - logging.error("getHttpData: HTTP error(%s): %s"%(e.code,e)) - errmsg = str(e) - # stop trying - break - except urllib2.URLError, e: - logging.error("getHttpData: URLLIB error(%s): %s"%(e.reason,e)) - errmsg = str(e) - # stop trying - #break - - if response is not None: - data = response.read() - response.close() - return data - - raise IOError("ERROR fetching HTTP data from %s: %s"%(url,errmsg)) - #return None +import socket +def urlopen(url,timeout=2): + """urlopen mit timeout""" + socket.setdefaulttimeout(timeout) + ret=urllib.urlopen(url) + socket.setdefaulttimeout(5) + return ret ## @@ -104,6 +80,8 @@ def getHttpData(url, data=None, num_trie ## class documentViewer(Folder): """document viewer""" + #textViewerUrl="http://127.0.0.1:8080/HFQP/testXSLT/getPage?" + meta_type="Document viewer" security=ClassSecurityInfo() @@ -123,8 +101,7 @@ class documentViewer(Folder): head_main = PageTemplateFile('zpt/head_main', globals()) docuviewer_css = PageTemplateFile('css/docuviewer.css', globals()) info_xml = PageTemplateFile('zpt/info_xml', globals()) - - + thumbs_main_rss = PageTemplateFile('zpt/thumbs_main_rss', globals()) security.declareProtected('View management screens','changeDocumentViewerForm') changeDocumentViewerForm = PageTemplateFile('zpt/changeDocumentViewer', globals()) @@ -144,12 +121,12 @@ class documentViewer(Folder): #self['template'] = templateFolder # Zope-2.12 style self._setObject('template',templateFolder) # old style try: - import MpdlXmlTextServer - textServer = MpdlXmlTextServer.MpdlXmlTextServer(id='fulltextclient',serverName=textServerName) + from Products.XMLRpcTools.XMLRpcTools import XMLRpcServerProxy + xmlRpcClient = XMLRpcServerProxy(id='fulltextclient', serverUrl=textServerName, use_xmlrpc=False) #templateFolder['fulltextclient'] = xmlRpcClient - templateFolder._setObject('fulltextclient',textServer) + templateFolder._setObject('fulltextclient',xmlRpcClient) except Exception, e: - logging.error("Unable to create MpdlXmlTextServer for fulltextclient: "+str(e)) + logging.error("Unable to create XMLRpcTools for fulltextclient: "+str(e)) try: from Products.zogiLib.zogiLib import zogiLib zogilib = zogiLib(id="zogilib", title="zogilib for docuviewer", dlServerURL=imageScalerUrl, layout="book") @@ -158,41 +135,7 @@ class documentViewer(Folder): except Exception, e: logging.error("Unable to create zogiLib for zogilib: "+str(e)) - - # proxy text server methods to fulltextclient - def getTextPage(self, **args): - """get page""" - return self.template.fulltextclient.getTextPage(**args) - - def getQuery(self, **args): - """get query""" - return self.template.fulltextclient.getQuery(**args) - - def getSearch(self, **args): - """get search""" - return self.template.fulltextclient.getSearch(**args) - - def getNumPages(self, docinfo): - """get numpages""" - return self.template.fulltextclient.getNumPages(docinfo) - - def getTranslate(self, **args): - """get translate""" - return self.template.fulltextclient.getTranslate(**args) - - def getLemma(self, **args): - """get lemma""" - return self.template.fulltextclient.getLemma(**args) - - def getToc(self, **args): - """get toc""" - return self.template.fulltextclient.getToc(**args) - - def getTocPage(self, **args): - """get tocpage""" - return self.template.fulltextclient.getTocPage(**args) - security.declareProtected('View','thumbs_rss') def thumbs_rss(self,mode,url,viewMode="auto",start=None,pn=1): ''' @@ -217,7 +160,7 @@ class documentViewer(Folder): pt = getattr(self.template, 'thumbs_main_rss') if viewMode=="auto": # automodus gewaehlt - if docinfo.has_key("textURL") or docinfo.has_key('textURLPath'): #texturl gesetzt und textViewer konfiguriert + if docinfo.get("textURL",'') and self.textViewerUrl: #texturl gesetzt und textViewer konfiguriert viewMode="text" else: viewMode="images" @@ -243,22 +186,23 @@ class documentViewer(Folder): return "ERROR: template folder missing!" if not getattr(self, 'digilibBaseUrl', None): - self.digilibBaseUrl = self.findDigilibUrl() or "http://digilib.mpiwg-berlin.mpg.de/digitallibrary" + self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" docinfo = self.getDocinfo(mode=mode,url=url) + if tocMode != "thumbs": # get table of contents docinfo = self.getToc(mode=tocMode, docinfo=docinfo) + pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo,viewMode=viewMode,tocMode=tocMode) + if viewMode=="auto": # automodus gewaehlt - if docinfo.has_key('textURL') or docinfo.has_key('textURLPath'): #texturl gesetzt und textViewer konfiguriert - viewMode="text_dict" + if docinfo.get("textURL",''): #texturl gesetzt und textViewer konfiguriert + viewMode="text" else: viewMode="images" - pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo,viewMode=viewMode,tocMode=tocMode) - pt = getattr(self.template, 'viewer_main') return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode,mk=self.generateMarks(mk)) @@ -266,8 +210,8 @@ class documentViewer(Folder): ret="" if mk is None: return "" - if not isinstance(mk, list): - mk=[mk] + if type(mk) is not ListType: + mk=[mk] for m in mk: ret+="mk=%s"%m return ret @@ -277,10 +221,6 @@ class documentViewer(Folder): """try to get the digilib URL from zogilib""" url = self.template.zogilib.getDLBaseUrl() return url - - def getDocumentViewerURL(self): - """returns the URL of this instance""" - return self.absolute_url() def getStyle(self, idx, selected, style=""): """returns a string with the given style and append 'sel' if path == selected.""" @@ -305,8 +245,7 @@ class documentViewer(Folder): params["url"] = getParentDir(params["url"]) # quote values and assemble into query string - #ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()]) - ps = urllib.urlencode(params) + ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()]) url=self.REQUEST['URL1']+"?"+ps return url @@ -321,7 +260,7 @@ class documentViewer(Folder): params[param] = str(val) # quote values and assemble into query string - logging.debug("XYXXXXX: %s"%repr(params.items())) + logging.info("XYXXXXX: %s"%repr(params.items())) ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()]) url=self.REQUEST['URL1']+"?"+ps return url @@ -340,9 +279,9 @@ class documentViewer(Folder): def isAccessible(self, docinfo): """returns if access to the resource is granted""" access = docinfo.get('accessType', None) - logging.debug("documentViewer (accessOK) access type %s"%access) + logger("documentViewer (accessOK)", logging.INFO, "access type %s"%access) if access is not None and access == 'free': - logging.debug("documentViewer (accessOK) access is free") + logger("documentViewer (accessOK)", logging.INFO, "access is free") return True elif access is None or access in self.authgroups: # only local access -- only logged in users @@ -353,12 +292,13 @@ class documentViewer(Folder): else: return False - logging.debug("documentViewer (accessOK) unknown access type %s"%access) + logger("documentViewer (accessOK)", logging.INFO, "unknown access type %s"%access) return False def getDirinfoFromDigilib(self,path,docinfo=None,cut=0): """gibt param von dlInfo aus""" + num_retries = 3 if docinfo is None: docinfo = {} @@ -368,15 +308,21 @@ class documentViewer(Folder): infoUrl=self.digilibBaseUrl+"/dirInfo-xml.jsp?mo=dir&fn="+path - logging.debug("documentViewer (getparamfromdigilib) dirInfo from %s"%(infoUrl)) + logger("documentViewer (getparamfromdigilib)", logging.INFO, "dirInfo from %s"%(infoUrl)) - txt = getHttpData(infoUrl) - if txt is None: + for cnt in range(num_retries): + try: + # dom = NonvalidatingReader.parseUri(imageUrl) + txt=urllib.urlopen(infoUrl).read() + dom = Parse(txt) + break + except: + logger("documentViewer (getdirinfofromdigilib)", logging.ERROR, "error reading %s (try %d)"%(infoUrl,cnt)) + else: raise IOError("Unable to get dir-info from %s"%(infoUrl)) - - dom = Parse(txt) + sizes=dom.xpath("//dir/size") - logging.debug("documentViewer (getparamfromdigilib) dirInfo:size"%sizes) + logger("documentViewer (getparamfromdigilib)", logging.INFO, "dirInfo:size"%sizes) if sizes: docinfo['numPages'] = int(getTextFromNode(sizes[0])) @@ -390,6 +336,7 @@ class documentViewer(Folder): def getIndexMeta(self, url): """returns dom of index.meta document at url""" + num_retries = 3 dom = None metaUrl = None if url.startswith("http://"): @@ -401,17 +348,25 @@ class documentViewer(Folder): metaUrl=server+url.replace("/mpiwg/online","") if not metaUrl.endswith("index.meta"): metaUrl += "/index.meta" + logging.debug("METAURL: %s"%metaUrl) + for cnt in range(num_retries): + try: + # patch dirk encoding fehler treten dann nicht mehr auf + # dom = NonvalidatingReader.parseUri(metaUrl) + txt=urllib.urlopen(metaUrl).read() + dom = Parse(txt) + break + except: + logger("ERROR documentViewer (getIndexMeta)", logging.INFO,"%s (%s)"%sys.exc_info()[0:2]) - logging.debug("(getIndexMeta): METAURL: %s"%metaUrl) - txt=getHttpData(metaUrl) - if txt is None: + if dom is None: raise IOError("Unable to read index meta from %s"%(url)) - - dom = Parse(txt) + return dom def getPresentationInfoXML(self, url): """returns dom of info.xml document at url""" + num_retries = 3 dom = None metaUrl = None if url.startswith("http://"): @@ -422,17 +377,25 @@ class documentViewer(Folder): server=self.digilibBaseUrl+"/servlet/Texter?fn=" metaUrl=server+url.replace("/mpiwg/online","") - txt=getHttpData(metaUrl) - if txt is None: + for cnt in range(num_retries): + try: + # patch dirk encoding fehler treten dann nicht mehr auf + # dom = NonvalidatingReader.parseUri(metaUrl) + txt=urllib.urlopen(metaUrl).read() + dom = Parse(txt) + break + except: + logger("ERROR documentViewer (getPresentationInfoXML)", logging.INFO,"%s (%s)"%sys.exc_info()[0:2]) + + if dom is None: raise IOError("Unable to read infoXMLfrom %s"%(url)) - - dom = Parse(txt) + return dom def getAuthinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0): """gets authorization info from the index.meta file at path or given by dom""" - logging.debug("documentViewer (getauthinfofromindexmeta) path: %s"%(path)) + logger("documentViewer (getauthinfofromindexmeta)", logging.INFO,"path: %s"%(path)) access = None @@ -508,7 +471,7 @@ class documentViewer(Folder): def getDocinfoFromTextTool(self, url, dom=None, docinfo=None): """parse texttool tag in index meta""" - logging.debug("documentViewer (getdocinfofromtexttool) url: %s" % (url)) + logger("documentViewer (getdocinfofromtexttool)", logging.INFO, "url: %s" % (url)) if docinfo is None: docinfo = {} if docinfo.get('lang', None) is None: @@ -523,7 +486,7 @@ class documentViewer(Folder): if archiveNames and (len(archiveNames) > 0): archiveName = getTextFromNode(archiveNames[0]) else: - logging.warning("documentViewer (getdocinfofromtexttool) resource/name missing in: %s" % (url)) + logger("documentViewer (getdocinfofromtexttool)", logging.WARNING, "resource/name missing in: %s" % (url)) archivePaths = dom.xpath("//resource/archive-path") if archivePaths and (len(archivePaths) > 0): @@ -535,7 +498,7 @@ class documentViewer(Folder): archivePath += "/" + archiveName else: # try to get archive-path from url - logging.warning("documentViewer (getdocinfofromtexttool) resource/archive-path missing in: %s" % (url)) + logger("documentViewer (getdocinfofromtexttool)", logging.WARNING, "resource/archive-path missing in: %s" % (url)) if (not url.startswith('http')): archivePath = url.replace('index.meta', '') @@ -568,8 +531,7 @@ class documentViewer(Folder): if viewerUrls and (len(viewerUrls) > 0): viewerUrl = getTextFromNode(viewerUrls[0]) docinfo['viewerURL'] = viewerUrl - - # old style text URL + textUrls = dom.xpath("//texttool/text") if textUrls and (len(textUrls) > 0): textUrl = getTextFromNode(textUrls[0]) @@ -581,14 +543,13 @@ class documentViewer(Folder): docinfo['textURL'] = textUrl - # new style text-url-path textUrls = dom.xpath("//texttool/text-url-path") if textUrls and (len(textUrls) > 0): textUrl = getTextFromNode(textUrls[0]) docinfo['textURLPath'] = textUrl if not docinfo['imagePath']: # text-only, no page images - docinfo = self.getNumPages(docinfo) + docinfo = self.getNumPages(docinfo) #im moment einfach auf eins setzen, navigation ueber die thumbs geht natuerlich nicht presentationUrls = dom.xpath("//texttool/presentation") docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag @@ -629,7 +590,7 @@ class documentViewer(Folder): def getDocinfoFromImagePath(self,path,docinfo=None,cut=0): """path ist the path to the images it assumes that the index.meta file is one level higher.""" - logging.debug("documentViewer (getdocinfofromimagepath) path: %s"%(path)) + logger("documentViewer (getdocinfofromimagepath)", logging.INFO,"path: %s"%(path)) if docinfo is None: docinfo = {} path=path.replace("/mpiwg/online","") @@ -639,7 +600,7 @@ class documentViewer(Folder): pathorig=path for x in range(cut): path=getParentDir(path) - logging.debug("documentViewer (getdocinfofromimagepath) PATH:"+path) + logging.error("PATH:"+path) imageUrl=self.digilibBaseUrl+"/servlet/Scaler?fn="+path docinfo['imageURL'] = imageUrl @@ -651,13 +612,13 @@ class documentViewer(Folder): def getDocinfo(self, mode, url): """returns docinfo depending on mode""" - logging.debug("documentViewer (getdocinfo) mode: %s, url: %s"%(mode,url)) + logger("documentViewer (getdocinfo)", logging.INFO,"mode: %s, url: %s"%(mode,url)) # look for cached docinfo in session if self.REQUEST.SESSION.has_key('docinfo'): docinfo = self.REQUEST.SESSION['docinfo'] # check if its still current if docinfo is not None and docinfo.get('mode') == mode and docinfo.get('url') == url: - logging.debug("documentViewer (getdocinfo) docinfo in session: %s"%docinfo) + logger("documentViewer (getdocinfo)", logging.INFO,"docinfo in session: %s"%docinfo) return docinfo # new docinfo docinfo = {'mode': mode, 'url': url} @@ -668,13 +629,14 @@ class documentViewer(Folder): elif mode=="filepath": docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo,cut=1) else: - logging.error("documentViewer (getdocinfo) unknown mode: %s!"%mode) + logger("documentViewer (getdocinfo)", logging.ERROR,"unknown mode!") raise ValueError("Unknown mode %s! Has to be one of 'texttool','imagepath','filepath'."%(mode)) - logging.debug("documentViewer (getdocinfo) docinfo: %s"%docinfo) + logger("documentViewer (getdocinfo)", logging.INFO,"docinfo: %s"%docinfo) self.REQUEST.SESSION['docinfo'] = docinfo return docinfo - + + def getPageinfo(self, current, start=None, rows=None, cols=None, docinfo=None, viewMode=None, tocMode=None): """returns pageinfo with the given parameters""" pageinfo = {} @@ -695,7 +657,9 @@ class documentViewer(Folder): pageinfo['end'] = min(pageinfo['end'], np) pageinfo['numgroups'] = int(np / grpsize) if np % grpsize > 0: - pageinfo['numgroups'] += 1 + pageinfo['numgroups'] += 1 + + pageinfo['viewMode'] = viewMode pageinfo['tocMode'] = tocMode pageinfo['query'] = self.REQUEST.get('query',' ') @@ -712,17 +676,293 @@ class documentViewer(Folder): if 'tocSize_%s'%tocMode in docinfo: tocSize = int(docinfo['tocSize_%s'%tocMode]) tocPageSize = int(pageinfo['tocPageSize']) - # cached toc + # cached toc + if tocSize%tocPageSize>0: tocPages=tocSize/tocPageSize+1 else: tocPages=tocSize/tocPageSize - pageinfo['tocPN'] = min (tocPages,toc) + pageinfo['tocPN'] = min (tocPages,toc) + pageinfo['searchPN'] =self.REQUEST.get('searchPN','1') pageinfo['sn'] =self.REQUEST.get('sn','') + return pageinfo + + def getSearch(self, pn=1, pageinfo=None, docinfo=None, query=None, queryType=None): + """get search list""" + 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'] + sn = pageinfo['sn'] + highlightQuery = pageinfo['highlightQuery'] + query =pageinfo['query'] + queryType =pageinfo['queryType'] + viewMode= pageinfo['viewMode'] + tocMode = pageinfo['tocMode'] + tocPN = pageinfo['tocPN'] + selfurl = self.absolute_url() + 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) + pagexml = page.replace('?document=%s'%str(docpath),'?url=%s'%url) + pagedom = Parse(pagexml) + if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"): + pagedivs = pagedom.xpath("//div[@class='queryResultPage']") + 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('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,query,pagesize,pn,tocMode,pn,tocPN)) + hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) + return serializeNode(pagenode) + + if (queryType=="fulltextMorph"): + pagedivs = pagedom.xpath("//div[@class='queryResult']") + 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('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,query,pagesize,pn,tocMode,pn,tocPN)) + 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'%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, 'onDblclick', 'popupWin.focus();') + pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']") + return serializeNode(pagenode) + + if (queryType=="ftIndex")or(queryType=="ftIndexMorph"): + pagedivs= pagedom.xpath("//div[@class='queryResultPage']") + 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 + hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s'%(viewMode,tocMode,tocPN,pn)) + + if href.startswith('../lt/lex.xql'): + hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%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, 'onDblclick', 'popupWin.focus();') + if href.startswith('../lt/lemma.xql'): + 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, 'onDblclick', 'popupWin.focus();') + return serializeNode(pagenode) + return "no text here" + + def getNumPages(self,docinfo=None): + """get list of pages from fulltext and put in docinfo""" + xquery = '//pb' + text = self.template.fulltextclient.eval("/mpdl/interface/xquery.xql", "document=%s&xquery=%s"%(docinfo['textURLPath'],xquery)) + # TODO: better processing of the page list. do we need the info somewhere else also? + docinfo['numPages'] = text.count(" 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('#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 == "pureXml": + # first div contains text + pagedivs = pagedom.xpath("/div") + if len(pagedivs) > 0: + pagenode = pagedivs[0] + return serializeNode(pagenode) + # text-with-links mode + if mode == "text_dict": + # first div contains text + pagedivs = pagedom.xpath("/div") + if len(pagedivs) > 0: + pagenode = pagedivs[0] + # check all a-tags + links = pagenode.xpath("//a") + for l in links: + hrefNode = l.getAttributeNodeNS(None, u"href") + if hrefNode: + # is link with href + href = hrefNode.nodeValue + if href.startswith('lt/lex.xql'): + # is pollux link + selfurl = self.absolute_url() + # change href + hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/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, 'onDblclick', 'popupWin.focus();') + + if href.startswith('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, '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) + + return "no text here" + + def getTranslate(self, query=None, language=None): + """translate into another languages""" + pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query))) + return pagexml + + def getLemma(self, lemma=None, language=None): + """simular words lemma """ + pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma))) + #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","lemma=%s&language=%s"%(lemma,language),outputUnicode=False) + return pagexml + + def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1): + """number of""" + docpath = docinfo['textURLPath'] + pagesize = pageinfo['queryPageSize'] + pn = pageinfo['searchPN'] + query =pageinfo['query'] + queryType =pageinfo['queryType'] + + tocSearch = 0 + tocDiv = None + 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)) + if 'tocSize_%s'%mode in docinfo: + # cached toc + return docinfo + docpath = docinfo['textURLPath'] + # we need to set a result set size + pagesize = 1000 + pn = 1 + if mode == "text": + queryType = "toc" + else: + queryType = mode + # number of entries in toc + tocSize = 0 + tocDiv = None + 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])) + # div contains text + #pagedivs = pagedom.xpath("//div[@class='queryResultPage']") + #if len(pagedivs) > 0: + # tocDiv = pagedivs[0] + + docinfo['tocSize_%s'%mode] = tocSize + #docinfo['tocDiv_%s'%mode] = tocDiv + 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 + if mode == "text": + queryType = "toc" + else: + queryType = mode + docpath = docinfo['textURLPath'] + path = docinfo['textURLPath'] + #logging.debug("documentViewer (gettoc) pathNomer: %s"%(pathNomer)) + pagesize = pageinfo['tocPageSize'] + pn = pageinfo['tocPN'] + url = docinfo['url'] + selfurl = self.absolute_url() + viewMode= pageinfo['viewMode'] + tocMode = pageinfo['tocMode'] + tocPN = pageinfo['tocPN'] + + pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn), outputUnicode=False) + page = pagexml.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') + return text + # post-processing downloaded xml + #pagedom = Parse(text) + # div contains text + #pagedivs = pagedom.xpath("//div[@class='queryResultPage']") + #if len(pagedivs) > 0: + # pagenode = pagedivs[0] + # return serializeNode(pagenode) + #else: + # return "No TOC!" -def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None): + def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None): """init document viewer""" self.title=title self.digilibBaseUrl = digilibBaseUrl @@ -731,6 +971,8 @@ def changeDocumentViewer(self,title="",d self.authgroups = [s.strip().lower() for s in authgroups.split(',')] if RESPONSE is not None: RESPONSE.redirect('manage_main') + + def manage_AddDocumentViewerForm(self): """add the viewer form""" @@ -745,7 +987,10 @@ def manage_AddDocumentViewer(self,id,ima if RESPONSE is not None: RESPONSE.redirect('manage_main') + +## ## DocumentViewerTemplate class +## class DocumentViewerTemplate(ZopePageTemplate): """Template for document viewer""" meta_type="DocumentViewer Template"