--- documentViewer/documentViewer.py 2006/04/06 17:13:43 1.3 +++ documentViewer/documentViewer.py 2006/04/10 19:51:50 1.7 @@ -14,10 +14,19 @@ from Ft.Xml import EMPTY_NAMESPACE import Ft.Xml.XPath import os.path +import sys import cgi import urllib import zLOG +def getInt(number, default=0): + """returns always an int (0 in case of problems)""" + try: + return int(number) + except: + return default + + def getTextFromNode(nodename): nodelist=nodename.childNodes rc = "" @@ -34,25 +43,6 @@ def urlopen(url): ret=urllib.urlopen(url) socket.setdefaulttimeout(5) return ret - -def getParamFromDigilib(path,param): - """gibt param von dlInfo aus""" - imageUrl=genericDigilib+"/dirInfo-xml.jsp?mo=dir&fn="+path - - zLOG.LOG("documentViewer (getparamfromdigilib)", zLOG.INFO, "dirInfo (%s) from %s"%(param,imageUrl)) - - try: - dom = NonvalidatingReader.parseUri(imageUrl) - except: - return None - - params=dom.xpath("//dir/%s"%param) - zLOG.LOG("documentViewer (getparamfromdigilib)", zLOG.INFO, "dirInfo:%s"%params) - - if params: - return getTextFromNode(params[0]) - - ## @@ -79,20 +69,23 @@ class documentViewer(Folder): changeDocumentViewerForm = PageTemplateFile('zpt/changeDocumentViewer', globals()) - def __init__(self,id,imageViewerUrl,title="",digilibBaseUrl=None): + def __init__(self,id,imageViewerUrl,title="",digilibBaseUrl=None,thumbcols=2,thumbrows=10): """init document viewer""" self.id=id self.title=title self.imageViewerUrl=imageViewerUrl - self.digilibBaseUrl = digilibBaseUrl - if not self.digilibBaseUrl: + if not digilibBaseUrl: self.digilibBaseUrl = self.findDigilibUrl() + else: + self.digilibBaseUrl = digilibBaseUrl + self.thumbcols = thumbcols + self.thumbrows = thumbrows # add template folder so we can always use template.something self.manage_addFolder('template') security.declareProtected('View','index_html') - def index_html(self,mode,url,start=0,pn=1): + def index_html(self,mode,url,start=None,pn=1): ''' view it @param mode: defines which type of document is behind url @@ -100,85 +93,71 @@ class documentViewer(Folder): ''' zLOG.LOG("documentViewer (index)", zLOG.INFO, "mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn)) - print "dlbaseurl:", self.digilibBaseUrl if not hasattr(self, 'template'): # create template folder if it doesn't exist - print "no template folder -- creating" self.manage_addFolder('template') if not self.digilibBaseUrl: self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" - print "dlbaseurl:", self.digilibBaseUrl - - docinfo = self.getDocinfo(mode=mode, url=url) - pageinfo = self.getPageinfo(start=start,current=pn) + docinfo = self.getDocinfo(mode=mode,url=url) + pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo) pt = getattr(self.template, 'viewer_main') return pt(docinfo=docinfo,pageinfo=pageinfo) - def imageLink(self,nr): - """link hinter den images""" - paramsTmp=cgi.parse_qs(self.REQUEST['QUERY_STRING']) - params={} - for x in paramsTmp.iteritems(): - params[x[0]]=x[1][0] + def getLink(self,param=None,val=None): + """link to documentviewer with parameter param set to val""" + params=cgi.parse_qs(self.REQUEST['QUERY_STRING']) + if param is not None: + if val is None: + if params.has_key(param): + del params[param] + else: + params[param] = [str(val)] + + ps = "&".join(["%s=%s"%(k,urllib.quote(v[0])) for (k, v) in params.items()]) + url=self.REQUEST['URL']+"?"+ps + #url=self.REQUEST['URL']+"?"+urllib.urlencode(params, doseq=True) + return url + - params['pn']=nr - newUrl=self.REQUEST['URL']+"?"+urllib.urlencode(params) - return newUrl - def getStyle(self, idx, selected, style=""): - """returns a string with the given style + 'sel' if path == selected.""" + """returns a string with the given style and append 'sel' if path == selected.""" #zLOG.LOG("documentViewer (getstyle)", zLOG.INFO, "idx: %s selected: %s style: %s"%(idx,selected,style)) if idx == selected: return style + 'sel' else: return style - - - def thumbruler(self,cols,rows,start,maximum): - """ruler for thumbs""" - ret="" - paramsTmp=cgi.parse_qs(self.REQUEST['QUERY_STRING']) - params={} - for x in paramsTmp.iteritems(): - - if not x[0]=="start": - params[x[0]]=x[1][0] - - newUrlSelect=self.REQUEST['URL']+"?"+urllib.urlencode(params) - if start>0: - newStart=max(start-cols*rows,0) - params['start']=newStart - newUrl=self.REQUEST['URL']+"?"+urllib.urlencode(params) - ret+="""prev"""%newUrl - - - ret+="""" - if startnext"""%newUrl - return ret + def getDirinfoFromDigilib(self,path,docinfo=None): + """gibt param von dlInfo aus""" + if docinfo is None: + docinfo = {} + + imageUrl=self.digilibBaseUrl+"/dirInfo-xml.jsp?mo=dir&fn="+path + zLOG.LOG("documentViewer (getparamfromdigilib)", zLOG.INFO, "dirInfo from %s"%(imageUrl)) + try: + dom = NonvalidatingReader.parseUri(imageUrl) + except: + zLOG.LOG("documentViewer (getparamfromdigilib)", zLOG.ERROR, "error reading %s"%(imageUrl)) + raise IOError("Unable to get dirinfo from %s"%(imageUrl)) + + params=dom.xpath("//dir/size") + zLOG.LOG("documentViewer (getparamfromdigilib)", zLOG.INFO, "dirInfo:size"%params) + if params: + docinfo['numPages'] = getTextFromNode(params[0]) + else: + docinfo['numPages'] = 0 + + return docinfo + + def getBibinfoFromIndexMeta(self,path,docinfo=None,dom=None): """gets bibliographical info from the index.meta file at url or given by dom""" zLOG.LOG("documentViewer (getbibinfofromindexmeta)", zLOG.INFO,"path: %s"%(path)) @@ -186,9 +165,8 @@ class documentViewer(Folder): if docinfo is None: docinfo = {} - metaData=self.metadata.main.meta.bib if dom is None: - server="http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary/servlet/Texter?fn=" + server=self.digilibBaseUrl+"/servlet/Texter?fn=" path="/".join(path.split("/")[0:-1]) metaUrl=server+path+"/index.meta" try: @@ -196,17 +174,19 @@ class documentViewer(Folder): except: return docinfo - type=dom.xpath("//bib/@type") - if type and (len(type)>0): - type=type[0].value + metaData=self.metadata.main.meta.bib + bibtype=dom.xpath("//bib/@type") + if bibtype and (len(bibtype)>0): + bibtype=bibtype[0].value else: - type="generic" - type=type.replace("-"," ")# wrong typesiin index meta "-" instead of " " - hash=metaData.generateMappingForType(type) - - docinfo['author']=getTextFromNode(dom.xpath("//bib/%s"%hash['author'][0])[0]) - docinfo['title']=getTextFromNode(dom.xpath("//bib/%s"%hash['title'][0])[0]) - docinfo['year']=getTextFromNode(dom.xpath("//bib/%s"%hash['year'][0])[0]) + bibtype="generic" + bibtype=bibtype.replace("-"," ") # wrong typesiin index meta "-" instead of " " (not wrong! ROC) + bibmap=metaData.generateMappingForType(bibtype) + print "bibmap: ", bibmap, " for: ", bibtype + if len(bibmap) > 0 and len(bibmap['author'][0]) > 0: + docinfo['author']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['author'][0])[0]) + docinfo['title']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['title'][0])[0]) + docinfo['year']=getTextFromNode(dom.xpath("//bib/%s"%bibmap['year'][0])[0]) return docinfo @@ -221,17 +201,15 @@ class documentViewer(Folder): dom = NonvalidatingReader.parseUri(url) except: zLOG.LOG("documentViewer (parseUrlTexttool)", zLOG.INFO,"%s (%s)"%sys.exc_info()[0:2]) - return docinfo + raise IOError("Unable to get texttool info from %s"%(url)) archivePaths=dom.xpath("//resource/archive-path") - if archivePaths and (len(archivePaths)>0): archivePath=getTextFromNode(archivePaths[0]) else: archivePath=None images=dom.xpath("//texttool/image") - if images and (len(images)>0): image=getTextFromNode(images[0]) else: @@ -240,21 +218,19 @@ class documentViewer(Folder): if image and archivePath: image=os.path.join(archivePath,image) image=image.replace("/mpiwg/online",'') - pt=getParamFromDigilib(image,'size') + docinfo=self.getDirinfoFromDigilib(image,docinfo=docinfo) docinfo['imagePath'] = image - docinfo['numberOfPages'] = pt + docinfo['imageURL'] = self.digilibBaseUrl+"/servlet/Scaler?fn="+image viewerUrls=dom.xpath("//texttool/digiliburlprefix") - if viewerUrls and (len(viewerUrls)>0): viewerUrl=getTextFromNode(viewerUrls[0]) - docinfo['imageURL'] = viewerURL + docinfo['viewerURL'] = viewerUrl textUrls=dom.xpath("//texttool/text") - if textUrls and (len(textUrls)>0): textUrl=getTextFromNode(textUrls[0]) - docinfo['textURL'] = textURL + docinfo['textURL'] = textUrl docinfo = self.getBibinfoFromIndexMeta(url,docinfo=docinfo,dom=dom) return docinfo @@ -265,11 +241,10 @@ class documentViewer(Folder): zLOG.LOG("documentViewer (getdocinfofromimagepath)", zLOG.INFO,"path: %s"%(path)) if docinfo is None: docinfo = {} - docinfo['imagePath'] = path path=path.replace("/mpiwg/online","") - pt=getParamFromDigilib(path,'size') - docinfo['numberOfPages'] = pt - imageUrl=genericDigilib+"/servlet/Scaler?fn=%s"%path + docinfo['imagePath'] = path + docinfo=self.getDirinfoFromDigilib(path,docinfo=docinfo) + imageUrl=self.digilibBaseUrl+"/servlet/Scaler?fn="+path docinfo['imageURL'] = imageUrl docinfo = self.getBibinfoFromIndexMeta(path,docinfo=docinfo) @@ -299,11 +274,27 @@ class documentViewer(Folder): return docinfo - def getPageinfo(self, start, current): + def getPageinfo(self, current, start=None, rows=None, cols=None, docinfo=None): """returns pageinfo with the given parameters""" pageinfo = {} - pageinfo['start'] = start + current = getInt(current) pageinfo['current'] = current + rows = int(rows or self.thumbrows) + pageinfo['rows'] = rows + cols = int(cols or self.thumbcols) + pageinfo['cols'] = cols + grpsize = cols * rows + pageinfo['groupsize'] = grpsize + start = getInt(start, default=(int(current / grpsize) * grpsize +1)) + pageinfo['start'] = start + pageinfo['end'] = start + grpsize + if docinfo is not None: + np = int(docinfo['numPages']) + pageinfo['end'] = min(pageinfo['end'], np) + pageinfo['numgroups'] = int(np / grpsize) + if np % grpsize > 0: + pageinfo['numgroups'] += 1 + return pageinfo def text(self,mode,url,pn): @@ -360,22 +351,6 @@ class documentViewer(Folder): # return list # - def image(self,mode,url,pn): - """give image out""" - if mode=="texttool": #index.meta with texttool information - (viewerUrl,imagepath,textpath)=parseUrlTextTool(url) - if not viewerUrl: - viewerUrl=self.imageViewerUrl - url=viewerUrl+"pn=%s&fn=%s"%(pn,imagepath[0]) - ret="""