--- documentViewer/documentViewer.py 2010/08/24 13:27:07 1.97 +++ documentViewer/documentViewer.py 2010/10/18 12:11:34 1.142 @@ -5,6 +5,7 @@ from Products.PageTemplates.PageTemplate from AccessControl import ClassSecurityInfo from AccessControl import getSecurityManager from Globals import package_home +from Products.zogiLib.zogiLib import browserCheck from Ft.Xml import EMPTY_NAMESPACE, Parse import Ft.Xml.Domlette @@ -16,6 +17,8 @@ import logging import math import urlparse import cStringIO +import re + def logger(txt,method,txt2): """logging""" @@ -48,7 +51,17 @@ def serializeNode(node, encoding='utf-8' buf.close() return s - +def getBrowserType(self): + """get browser type object""" + if self.REQUEST.SESSION.has_key('browserType'): + return self.REQUEST.SESSION['browserType'] + else: + bt = browserCheck(self) + self.REQUEST.SESSION.set('browserType', bt) + logging.debug("documentViewer (BROWSER TYPE) bt %s"%bt) + return bt + + def getParentDir(path): """returns pathname shortened by one""" return '/'.join(path.split('/')[0:-1]) @@ -172,11 +185,27 @@ class documentViewer(Folder): def getSearch(self, **args): """get search""" return self.template.fulltextclient.getSearch(**args) - + + def getGisPlaces(self, **args): + """get gis places""" + return self.template.fulltextclient.getGisPlaces(**args) + + def getAllGisPlaces(self, **args): + """get all gis places """ + return self.template.fulltextclient.getAllGisPlaces(**args) + + def getOrigPages(self, **args): + """get original page number """ + return self.template.fulltextclient.getOrigPages(**args) + def getNumPages(self, docinfo): """get numpages""" return self.template.fulltextclient.getNumPages(docinfo) - + + def getNumTextPages(self, docinfo): + """get numpages text""" + return self.template.fulltextclient.getNumTextPages(docinfo) + def getTranslate(self, **args): """get translate""" return self.template.fulltextclient.getTranslate(**args) @@ -214,7 +243,9 @@ class documentViewer(Folder): self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" docinfo = self.getDocinfo(mode=mode,url=url) - pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo) + #pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo) + pageinfo = self.getPageinfo(start=start,current=pn, docinfo=docinfo) + ''' ZDES ''' pt = getattr(self.template, 'thumbs_main_rss') if viewMode=="auto": # automodus gewaehlt @@ -259,7 +290,7 @@ class documentViewer(Folder): else: viewMode="images" - pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo,viewMode=viewMode,tocMode=tocMode) + 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)) @@ -390,10 +421,20 @@ class documentViewer(Folder): return docinfo - - def getIndexMeta(self, url): - """returns dom of index.meta document at url""" - dom = None + def getIndexMetaPath(self,url): + """gib nur den Pfad zurueck""" + regexp = re.compile(r".*(experimental|permanent)/(.*)") + regpath = regexp.match(url) + if (regpath==None): + return "" + logging.debug("(getDomFromIndexMeta): URLXAXA: %s"%regpath.group(2)) + return ("/mpiwg/online/"+regpath.group(1)+"/"+regpath.group(2)) + + + + def getIndexMetaUrl(self,url): + """returns utr of index.meta document at url""" + metaUrl = None if url.startswith("http://"): # real URL @@ -404,8 +445,15 @@ class documentViewer(Folder): metaUrl=server+url.replace("/mpiwg/online","") if not metaUrl.endswith("index.meta"): metaUrl += "/index.meta" + + return metaUrl + + def getDomFromIndexMeta(self, url): + """get dom from index meta""" + dom = None + metaUrl = self.getIndexMetaUrl(url) - logging.debug("(getIndexMeta): METAURL: %s"%metaUrl) + logging.debug("(getDomFromIndexMeta): METAURL: %s"%metaUrl) txt=getHttpData(metaUrl) if txt is None: raise IOError("Unable to read index meta from %s"%(url)) @@ -445,7 +493,7 @@ class documentViewer(Folder): if dom is None: for x in range(cut): path=getParentDir(path) - dom = self.getIndexMeta(path) + dom = self.getDomFromIndexMeta(path) acctype = dom.xpath("//access-conditions/access/@type") if acctype and (len(acctype)>0): @@ -467,7 +515,9 @@ class documentViewer(Folder): if dom is None: for x in range(cut): path=getParentDir(path) - dom = self.getIndexMeta(path) + dom = self.getDomFromIndexMeta(path) + + docinfo['indexMetaPath']=self.getIndexMetaPath(path); logging.debug("documentViewer (getbibinfofromindexmeta cutted) path: %s"%(path)) # put in all raw bib fields as dict "bib" @@ -489,6 +539,8 @@ class documentViewer(Folder): bibtype=bibtype.replace("-"," ") # wrong typesiin index meta "-" instead of " " (not wrong! ROC) docinfo['bib_type'] = bibtype bibmap=metaData.generateMappingForType(bibtype) + logging.debug("documentViewer (getbibinfofromindexmeta) bibmap:"+repr(bibmap)) + logging.debug("documentViewer (getbibinfofromindexmeta) bibtype:"+repr(bibtype)) # if there is no mapping bibmap is empty (mapping sometimes has empty fields) if len(bibmap) > 0 and len(bibmap['author'][0]) > 0: try: @@ -508,6 +560,20 @@ class documentViewer(Folder): return docinfo + + def getNameFromIndexMeta(self,path,docinfo=None,dom=None,cut=0): + """gets name info from the index.meta file at path or given by dom""" + if docinfo is None: + docinfo = {} + + if dom is None: + for x in range(cut): + path=getParentDir(path) + dom = self.getDomFromIndexMeta(path) + + docinfo['name']=getTextFromNode(dom.xpath("/resource/name")[0]) + logging.debug("documentViewer docinfo[name] %s"%docinfo['name']) + return docinfo def getDocinfoFromTextTool(self, url, dom=None, docinfo=None): """parse texttool tag in index meta""" @@ -517,7 +583,7 @@ class documentViewer(Folder): if docinfo.get('lang', None) is None: docinfo['lang'] = '' # default keine Sprache gesetzt if dom is None: - dom = self.getIndexMeta(url) + dom = self.getDomFromIndexMeta(url) archivePath = None archiveName = None @@ -591,10 +657,12 @@ class documentViewer(Folder): docinfo['textURLPath'] = textUrl if not docinfo['imagePath']: # text-only, no page images - docinfo = self.getNumPages(docinfo) + docinfo = self.getNumTextPages(docinfo) presentationUrls = dom.xpath("//texttool/presentation") docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag + docinfo = self.getNameFromIndexMeta(url, docinfo=docinfo, dom=dom) + #docinfo =self.getOrigPages(docinfo=docinfo) if presentationUrls and (len(presentationUrls) > 0): # ueberschreibe diese durch presentation informationen # presentation url ergiebt sich ersetzen von index.meta in der url der fuer die Metadaten @@ -675,6 +743,7 @@ class documentViewer(Folder): raise ValueError("Unknown mode %s! Has to be one of 'texttool','imagepath','filepath'."%(mode)) logging.debug("documentViewer (getdocinfo) docinfo: %s"%docinfo) + #logging.debug("documentViewer (getdocinfo) docinfo: %s"%) self.REQUEST.SESSION['docinfo'] = docinfo return docinfo @@ -682,6 +751,8 @@ class documentViewer(Folder): """returns pageinfo with the given parameters""" pageinfo = {} current = getInt(current) + + #pageinfo ['originalPage'] = originalPage pageinfo['current'] = current rows = int(rows or self.thumbrows) pageinfo['rows'] = rows @@ -701,8 +772,8 @@ class documentViewer(Folder): pageinfo['numgroups'] += 1 pageinfo['viewMode'] = viewMode pageinfo['tocMode'] = tocMode - pageinfo['characterNormalization'] =characterNormalization - #pageinfo['characterNormalization'] = self.REQUEST.get('characterNormalization',' ') + #pageinfo['characterNormalization'] =characterNormalization + pageinfo['characterNormalization'] = self.REQUEST.get('characterNormalization',' ') pageinfo['query'] = self.REQUEST.get('query',' ') pageinfo['queryType'] = self.REQUEST.get('queryType',' ') pageinfo['querySearch'] =self.REQUEST.get('querySearch', 'fulltext') @@ -714,6 +785,8 @@ class documentViewer(Folder): toc = int (pageinfo['tocPN']) pageinfo['textPages'] =int (toc) + + if 'tocSize_%s'%tocMode in docinfo: tocSize = int(docinfo['tocSize_%s'%tocMode]) tocPageSize = int(pageinfo['tocPageSize'])