Mercurial > hg > MetaDataProvider
changeset 28:40508e672841
richer metadata for bib. still not sufficient though.
| author | casties |
|---|---|
| date | Mon, 20 Aug 2012 20:14:45 +0200 |
| parents | a0d273542509 |
| children | b3428e281ee2 |
| files | MetaData.py MetaDataFolder.py |
| diffstat | 2 files changed, 19 insertions(+), 134 deletions(-) [+] |
line wrap: on
line diff
--- a/MetaData.py Wed Aug 15 17:07:37 2012 +0200 +++ b/MetaData.py Mon Aug 20 20:14:45 2012 +0200 @@ -181,11 +181,18 @@ if self.mappingSelectAttribute: # put type in @type - attr = data.get('@attr', None) + if all: + attr = data[0].get('@attr', None) + else: + attr = data.get('@attr', None) + if attr: type = attr.get(self.mappingSelectAttribute, None) if type is not None: - data['@type'] = normalizeFieldName(type) + if all: + data[0]['@type'] = normalizeFieldName(type) + else: + data['@type'] = normalizeFieldName(type) return data @@ -277,7 +284,7 @@ dc = fields[bk].get('dcmap', None) if dc: # add value - if dcData.get('dc',None): + if dcData.get('dc', None): # key exists - append dcData[dc] += '/' + data[bk] else: @@ -285,7 +292,7 @@ return dcData - def getFormatted(self, template, path=None, dom=None, data=None, allFields=False): + def getFormatted(self, template, path=None, dom=None, data=None, xdata=None, allFields=False): """returns string with document data formatted according to template. gets data from server or dom or pre-parsed data.""" logging.debug("getFormatted(template=%s)"%(template)) @@ -318,7 +325,7 @@ fields = {} #logging.debug(tp) - return tp(mdmap=fields, md=data) + return tp(mdmap=fields, md=data, mdx=xdata) def correctPath(self,path,remove=None,prefix=None,cut=0):
--- a/MetaDataFolder.py Wed Aug 15 17:07:37 2012 +0200 +++ b/MetaDataFolder.py Mon Aug 20 20:14:45 2012 +0200 @@ -20,20 +20,6 @@ return bt -def OLDgetBibdataFromDom(dom): - """returns dict with all elements from bib-tag""" - bibinfo = {} - bib = dom.find(".//meta/bib") - if bib is not None: - # put type in @type - type = bib.get('type') - bibinfo['@type'] = normalizeBibField(type) - # put all subelements in dict - for e in bib: - bibinfo[normalizeBibField(e.tag)] = getText(e) - - return bibinfo - def toString(list): ret=u"" @@ -42,36 +28,6 @@ return ret -def dcMetaDataToHash(mdSet): - """Convenience Function for creates a hash from the DCMetadataset - @param mdSet: String containing DCMetadata informmation - currently only in the format getDCMetadata of this module""" - - NSS = { - 'rdf': 'http://www.w3.org/1999/02/22-rdf-syntax-ns#', - 'dc': 'http://dublincore.org/documents/dcmi-namespace/', - 'owl':"http://www.w3.org/2002/07/owl#", - 'rdfs':"http://www.w3.org/2000/01/rdf-schema#" - } - ret={} - import StringIO - import sys - buffer= StringIO.StringIO(mdSet) - try: - md = amara.parse(buffer,prefixes=NSS) - except: - logging.error("Error: %s (%s)"%(sys.exc_info()[0],sys.exc_info()[1])) - - ret["error"]=mdSet - return ret - - ret["title"] = toString(md.xml_xpath("//dc:title/text()")) - ret["creator"] =toString(md.xml_xpath("//dc:creator/text()")) - ret["date"] = toString(md.xml_xpath("//dc:date/text()")) - - return ret - - class MetaDataFolder(Folder): """provides methods for managing complete metadata structures""" meta_type='MetaDataFolder' @@ -231,9 +187,9 @@ return None - def getBibData(self, path=None, dom=None): + def getBibData(self, path=None, dom=None, all=False, recursive=0): """returns contents of bib tag as dict""" - return self.resource.meta.bib.getData(path=path, dom=dom) + return self.resource.meta.bib.getData(path=path, dom=dom, all=all, recursive=recursive) def getBibMapping(self, bibtype): """returns MetaDataMapping for resource/meta/bib of bibtype""" @@ -251,15 +207,15 @@ """returns dict with DC keys and data from bibdata""" return self.resource.meta.bib.getDCMappedData(bibdata) - def getBibFormattedMetaData(self, path=None, dom=None, bibdata=None): + def getBibFormattedMetaData(self, path=None, dom=None, bibdata=None, bibxdata=None): """get the metadafrom server""" logging.debug("getBibFormattedMetaData(path=%s)"%path) - return self.resource.meta.bib.getFormatted('metadata_template', path=path, dom=dom, data=bibdata) + return self.resource.meta.bib.getFormatted('metadata_template', path=path, dom=dom, data=bibdata, xdata=bibxdata) - def getBibFormattedMetaDataExtended(self,path=None, dom=None, bibdata=None): + def getBibFormattedMetaDataExtended(self,path=None, dom=None, bibdata=None, bibxdata=None): """get the metadafrom server""" logging.debug("getBibFormattedMetaDataExtended(path=%s)"%path) - return self.resource.meta.bib.getFormatted('metadata_extended_template', path=path, dom=dom, data=bibdata, allFields=True) + return self.resource.meta.bib.getFormatted('metadata_extended_template', path=path, dom=dom, data=bibdata, xdata=bibxdata, allFields=True) def getBibFormattedLabel(self,path=None, dom=None, bibdata=None): """get the metadafrom server""" @@ -272,85 +228,7 @@ getFormattedMetaDataExtended = getBibFormattedMetaDataExtended getFormattedLabel = getBibFormattedLabel - - def OLDgetDCFormatted(self,path): - """get the metadata as dc set""" - logging.debug("getDCFormatted(path=%s)"%path) - namespace={ 'mpiwg': "http://www.mpiwg-berlin.mpg.de/ns/mpiwg"} - namespaceUsed=False - - md = self.getMDFromPathOrUrl(path) - logging.debug("MD in XML"+md) - im = amara.parse(md, prefixes=namespace) - - typePaths=im.xml_xpath('//bib/@type') - archimedes=False - - if len(typePaths)<1: - typePaths=im.xml_xpath('//meta/archimedes') # sinderfall fuer veraltete index.meta files vom typ archimedes - if len(typePaths)>0: - type = "archimedes" - archimedes=True - else: - typePaths=im.xml_xpath('//mpiwg:bib/@type') - if len(typePaths)<1: - return "" - else: - namespaceUsed=True - - type=unicode(typePaths[0]) - else: - type=unicode(typePaths[0]) - logging.info("got type:"+type) - try: - mapping=getattr(self.main.meta.bib,type.lower(),None) - except: - logging.error("getMetaDataFromServer no mapping for type: %s"%type) - return "" - - try: - dcMapping=getattr(self.main.meta.bib,"dc",None) - except: - logging.error("getMetaDataFromServer no dc in meta/bib") - return "" - - mds=mapping.generateMappingHash() # Hole das Mapping generisches Feld --> Feld im entsprechenden Typ - dcMds=dcMapping.generateMappingHash() - - mdHash=[] - logging.debug("Value: %s"%repr(mds)) - - for key,valueTriple in mds.items(): - value=valueTriple[0] - logging.debug("Value: %s"%repr(value)) - logging.debug("Key: %s"%repr(key)) - if value!="": - if not archimedes: - if namespaceUsed: - try: - v = im.xml_xpath('//mpiwg:bib/mpiwg:%s/text()'%value) - except: - logging.error('cannot do: //mpiwg:bib/mpiwg:%s/text()'%value) - else: - v = im.xml_xpath('//bib/%s/text()'%value) - else: - v = im.xml_xpath('//archimedes/%s/text()'%value) - if len(v) > 0: - dc=dcMds[key][0] - - if (dc !="") and (value !=""): - logging.debug("%s--> : %s"%(repr(value),dc)) - mdHash.append([dc,unicode(v[0])]) - - ret="""<bib xmlns:dc="http://dublincore.org/documents/dcmi-namespace/"> """ - ret+="<dc:type>%s</dc:type>"%type - for md in mdHash: - - ret+="""<dc:%s>%s</dc:%s>"""%(md[0],xml.sax.saxutils.escape(md[1]),md[0]) - ret+="</bib>" - return ret - - + changeMetaDataFolderForm = PageTemplateFile('zpt/changeMetaDataFolder',globals()) security.declarePublic('changeMetaDataFolder')