version 1.101, 2010/10/05 12:25:32
|
version 1.113, 2010/10/08 14:02:24
|
Line 401 class documentViewer(Folder):
|
Line 401 class documentViewer(Folder):
|
regpath = regexp.match(url) |
regpath = regexp.match(url) |
if (regpath==None): |
if (regpath==None): |
return "" |
return "" |
|
logging.debug("(getDomFromIndexMeta): URLXAXA: %s"%regpath.group(2)) |
return ("/mpiwg/online/"+regpath.group(1)+"/"+regpath.group(2)) |
return ("/mpiwg/online/"+regpath.group(1)+"/"+regpath.group(2)) |
|
|
|
|
|
|
def getIndexMetaUrl(self,url): |
def getIndexMetaUrl(self,url): |
"""returns utr of index.meta document at url""" |
"""returns utr of index.meta document at url""" |
|
|
Line 533 class documentViewer(Folder):
|
Line 535 class documentViewer(Folder):
|
return docinfo |
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['indexMetaPath']=self.getIndexMetaPath(path); |
|
|
|
#result= dom.xpath("//result/resultPage") |
|
#docinfo['numPages']=int(getTextFromNode(result[0])) |
|
|
|
#result =dom.xpath("//name") |
|
docinfo['name']=getTextFromNode(dom.xpath("//name")[0]) |
|
logging.debug("documentViewer docinfo[name] %s"%docinfo[name]) |
|
|
|
#logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype) |
|
return docinfo |
|
|
def getDocinfoFromTextTool(self, url, dom=None, docinfo=None): |
def getDocinfoFromTextTool(self, url, dom=None, docinfo=None): |
"""parse texttool tag in index meta""" |
"""parse texttool tag in index meta""" |
logging.debug("documentViewer (getdocinfofromtexttool) url: %s" % (url)) |
logging.debug("documentViewer (getdocinfofromtexttool) url: %s" % (url)) |
Line 615 class documentViewer(Folder):
|
Line 639 class documentViewer(Folder):
|
docinfo['textURLPath'] = textUrl |
docinfo['textURLPath'] = textUrl |
if not docinfo['imagePath']: |
if not docinfo['imagePath']: |
# text-only, no page images |
# text-only, no page images |
docinfo = self.getTextPages(docinfo) |
docinfo = self.getNumTextPages(docinfo) |
|
|
presentationUrls = dom.xpath("//texttool/presentation") |
presentationUrls = dom.xpath("//texttool/presentation") |
docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag |
docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag |
|
#docinfo = self.getNameFromIndexMeta(url, docinfo=docinfo, dom=dom) |
|
|
if presentationUrls and (len(presentationUrls) > 0): # ueberschreibe diese durch presentation informationen |
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 |
# presentation url ergiebt sich ersetzen von index.meta in der url der fuer die Metadaten |