Mercurial > hg > MetaDataProvider

comparison MetaData.py @ 0:9f9d9be26e53

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
first checkin in Mercurial (see history in SVN)
author casties
date Mon, 25 Jul 2011 16:50:48 +0200
parents
children e4bae49e657b
comparison
equal deleted inserted replaced
-1:000000000000 0:9f9d9be26e53
1 from OFS.Folder import Folder
2 from Products.PageTemplates.PageTemplateFile import PageTemplateFile
3 from Globals import package_home
4 from AccessControl import ClassSecurityInfo
5 import os.path
6 import urllib
7 import logging
8 import urlparse
9
10 # TODO: which xml toolkit?
11 import amara
12 import xml.sax.saxutils
13 import xml.dom.minidom
14 import xml.etree.ElementTree as ET
15
16
17 # TODO: do we need this?
18 #from Products.OSA_system2 import OSAS_helpers
19 #from Products.OSA_system2.OSAS_metadata import OSAS_Metadata,OSAS_MetadataMapping
20
21 from OSAS_metadata import OSAS_Metadata, OSAS_MetadataMapping
22
23
24 from SrvTxtUtils import getHttpData, getText
25
26
27 # TODO: get rid of this
28 def getTextFromNode(nodelist):
29 """gibt text aus nodelist"""
30 rc = ""
31 for node in nodelist:
32 if node.nodeType == node.TEXT_NODE:
33 rc = rc + node.data
34 return rc
35
36
37 def normalizeBibtype(bt):
38 """returns normalised bib type for looking up mappings"""
39 bt = bt.strip().replace(' ', '-').lower()
40 return bt
41
42 def toString(list):
43 ret=u""
44
45 for l in list:
46 ret+=unicode(l)
47
48 return ret
49
50 def dcMetaDataToHash(mdSet):
51 """Convenience Function for creates a hash from the DCMetadataset
52 @param mdSet: String containing DCMetadata informmation
53 currently only in the format getDCMetadata of this module"""
54
55 NSS = {
56 'rdf': 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
57 'dc': 'http://dublincore.org/documents/dcmi-namespace/',
58 'owl':"http://www.w3.org/2002/07/owl#",
59 'rdfs':"http://www.w3.org/2000/01/rdf-schema#"
60 }
61 ret={}
62 import StringIO
63 import sys
64 buffer= StringIO.StringIO(mdSet)
65 try:
66 md = amara.parse(buffer,prefixes=NSS)
67 except:
68 logging.error("Error: %s (%s)"%(sys.exc_info()[0],sys.exc_info()[1]))
69
70 ret["error"]=mdSet
71 return ret
72
73 ret["title"] = toString(md.xml_xpath("//dc:title/text()"))
74 ret["creator"] =toString(md.xml_xpath("//dc:creator/text()"))
75 ret["date"] = toString(md.xml_xpath("//dc:date/text()"))
76
77 return ret
78
79 def getBibdataFromDom(dom):
80 """returns dict with all elements from bib-tag"""
81 bibinfo = {}
82 bib = dom.find(".//meta/bib")
83 if bib is not None:
84 # put type in @type
85 type = bib.get('type')
86 bibinfo['@type'] = normalizedBibtype(type)
87 # put all subelements in dict
88 for e in bib:
89 bibinfo[e.tag] = getText(e)
90
91 return bibinfo
92
93
94
95
96 class MetaData(OSAS_Metadata):
97 """provides basic methods for managing metadata structures"""
98 meta_type='MetaData'
99 security=ClassSecurityInfo()
100 manage_options = Folder.manage_options+(
101 {'label':'Main Config','action':'changeMetadataForm'},
102 {'label':'Import XML Schema','action':'importMetaDataExportXML'},
103 {'label':'Select Fields for Display','action':'indicateDisplayFieldsForm'},
104 )
105
106 def __init__(self,id,shortDescription='',description='',fields=''):
107 """initialize a new instance"""
108 self.id = id
109 self.shortDescription =shortDescription #label fuer link auf add page
110 self.description=description #description of the method for link page
111 self.fieldList=fields.split(",")[0:]
112 self.metaDataServerUrl="" # muss mit change metadata gesetzt werden
113
114
115 def correctPath(self,path,remove=None,prefix=None,cut=0):
116 """convinience method um einen pfad zu veraendern"""
117
118 if remove is not None:
119 path=path.replace(remove,'')
120 if prefix is not None:
121 path=os.path.join(prefix,path)
122
123 if cut>0:
124 splitted=path.split("/")
125 path="/".join(splitted[0:len(splitted)-cut])
126 return path
127
128 def importMetaDataExportXML(self,importFile=None,RESPONSE=None):
129 """imports metadata from the metadataexportxml file"""
130
131 if importFile is None:
132 pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','importMetaDataExportXML.zpt')).__of__(self)
133 return pt()
134
135 dom=xml.dom.minidom.parse(importFile)
136 self.createMappingFromDom(dom.getElementsByTagName("metadataExport")[0])
137
138 if RESPONSE is not None:
139 RESPONSE.redirect('manage_main')
140
141
142 def createMappingFromDom(self,metadatanode,metadata=None):
143 """erzeuge ein Mapping aus dem der metadatanode des xmlformats, metadata ist ein metadataobject"""
144
145 if metadata is None:
146 metadata=self
147
148 nodes=metadatanode.childNodes
149
150 #erster schritt: anlegen der fieldlist
151 for node in nodes:
152 logging.debug("node: %s"%node.tagName)
153 if node.tagName=="set":
154 set=node
155 if set.getAttribute('name')=='generic':
156 list=[]
157 for entry in set.getElementsByTagName('entry'):
158 list.append(entry.getAttribute('genericName'))
159 metadata.fieldList=list[0:]
160
161 else:
162 id=set.getAttribute('name').encode('utf-8')
163 list=[]
164 argList={}
165 for entry in set.getElementsByTagName('entry'):
166 genericName=entry.getAttribute('genericName')
167 tag=entry.getAttribute('tag')
168 label=entry.getAttribute('label')
169 description=getTextFromNode(entry.childNodes) #TODO: clean
170 argList[genericName]=(tag,label,description)
171 metadata._setObject(id,MetaDataMapping(id,id,argList))
172
173 elif node.tagName=="metadata":
174 mn=node
175 name=mn.getAttribute('name').encode('utf-8')
176 metadata._setObject(name,MetaData(name,name))
177 mdObj=getattr(metadata,name)
178 mdObj.createMappingFromDom(mn)
179
180
181 def getMDFromPathOrUrl(self,path):
182 parsedurl = urlparse.urlparse(path)
183 if parsedurl[0] != "":
184 # has schema (e.g. http)
185 url=path
186 else:
187 # path only
188 if path.endswith("index.meta"):
189 url =self.metaDataServerUrl%path
190 else:
191 url=os.path.join(self.metaDataServerUrl%path,'index.meta')
192
193 #logging.debug("get Metadata: %s"%url)
194 md = getHttpData(url)
195 return md
196
197 def getDCFormatted(self,path):
198 """get the metadata as dc set"""
199 logging.debug("getDCFormatted(path=%s)"%path)
200 namespace={ 'mpiwg': "http://www.mpiwg-berlin.mpg.de/ns/mpiwg"}
201 namespaceUsed=False
202
203 md = self.getMDFromPathOrUrl(path)
204 logging.debug("MD in XML"+md)
205 im = amara.parse(md, prefixes=namespace)
206
207 typePaths=im.xml_xpath('//bib/@type')
208 archimedes=False
209
210 if len(typePaths)<1:
211 typePaths=im.xml_xpath('//meta/archimedes') # sinderfall fuer veraltete index.meta files vom typ archimedes
212 if len(typePaths)>0:
213 type = "archimedes"
214 archimedes=True
215 else:
216 typePaths=im.xml_xpath('//mpiwg:bib/@type')
217 if len(typePaths)<1:
218 return ""
219 else:
220 namespaceUsed=True
221
222 type=unicode(typePaths[0])
223 else:
224 type=unicode(typePaths[0])
225 logging.info("got type:"+type)
226 try:
227 mapping=getattr(self.main.meta.bib,type.lower(),None)
228 except:
229 logging.error("getMetaDataFromServer no mapping for type: %s"%type)
230 return ""
231
232 try:
233 dcMapping=getattr(self.main.meta.bib,"dc",None)
234 except:
235 logging.error("getMetaDataFromServer no dc in meta/bib")
236 return ""
237
238 mds=mapping.generateMappingHash() # Hole das Mapping generisches Feld --> Feld im entsprechenden Typ
239 dcMds=dcMapping.generateMappingHash()
240
241 mdHash=[]
242 logging.debug("Value: %s"%repr(mds))
243
244 for key,valueTriple in mds.items():
245 value=valueTriple[0]
246 logging.debug("Value: %s"%repr(value))
247 logging.debug("Key: %s"%repr(key))
248 if value!="":
249 if not archimedes:
250 if namespaceUsed:
251 try:
252 v = im.xml_xpath('//mpiwg:bib/mpiwg:%s/text()'%value)
253 except:
254 logging.error('cannot do: //mpiwg:bib/mpiwg:%s/text()'%value)
255 else:
256 v = im.xml_xpath('//bib/%s/text()'%value)
257 else:
258 v = im.xml_xpath('//archimedes/%s/text()'%value)
259 if len(v) > 0:
260 dc=dcMds[key][0]
261
262 if (dc !="") and (value !=""):
263 logging.debug("%s--> : %s"%(repr(value),dc))
264 mdHash.append([dc,unicode(v[0])])
265
266 ret="""<bib xmlns:dc="http://dublincore.org/documents/dcmi-namespace/"> """
267 ret+="<dc:type>%s</dc:type>"%type
268 for md in mdHash:
269
270 ret+="""<dc:%s>%s</dc:%s>"""%(md[0],xml.sax.saxutils.escape(md[1]),md[0])
271 ret+="</bib>"
272 return ret
273
274
275 def getStdMappedHash(self, bibdata):
276 """returns dict with metadata from bibdata mapped according to standard mapping"""
277 mdHash={}
278 bibtype = bibdata['@type']
279 # get mapping from main/meta/bib
280 try:
281 mapping=getattr(self.main.meta.bib, bibtype.lower())
282 except:
283 logging.error("getStdMappedHash: no mapping for type: %s"%bibtype)
284 return mdHash
285
286 mds = mapping.generateMappingHash() # Hole das Mapping generisches Feld --> Feld im entsprechenden Typ
287
288 for field in mds.keys():
289 # get mapped field name
290 mf = mds[field][0]
291 if not mf:
292 continue
293 logging.debug("mapping: %s = %s"%(field,mf))
294 mdHash[field] = bibdata.get(mf, '')
295
296 return mdHash
297
298
299 def getFormatted(self, template, path=None, dom=None, bibdata=None):
300 """returns string with document data formatted according to template.
301 gets data from server or dom or pre-parsed bibdata."""
302 logging.debug("getFormatted(template=%s)"%(template))
303
304 if dom is None and bibdata is None:
305 # get from server
306 md = self.getMDFromPathOrUrl(path.replace("/mpiwg/online",""))
307 #logging.debug("md:"+md)
308 #dom = amara.parse(md)
309 dom = ET.fromstring(md)
310
311 # get contents of bib tag
312 if bibdata is None:
313 bibdata = getBibdataFromDom(dom)
314
315 bibtype = bibdata['@type']
316
317 # get template
318 tp=getattr(self,"%s_%s"%(template, bibtype.lower()), None)
319 if tp is None:
320 logging.warning("getFormatted: no template for: %s_%s"%(template, bibtype))
321 # try generic
322 tp=getattr(self,"%s_generic"%(template), None)
323 if tp is None:
324 logging.error("getFormatted: no generic template either: %s"%(template))
325 return ""
326
327 # put mapped data in mdHash
328 mdHash = self.getStdMappedHash(bibdata)
329
330 return tp(stdmd=mdHash, md=bibdata)
331
332
333 def getFormattedMetaData(self, path=None, dom=None, bibdata=None):
334 """get the metadafrom server"""
335 logging.debug("getFormattedMetaData(path=%s)"%path)
336 return self.getFormatted('metadata_template', path=path, dom=dom, bibdata=bibdata)
337
338 def getFormattedMetaDataShort(self, path=None, dom=None, bibdata=None):
339 """get the metadafrom server"""
340 logging.debug("getFormattedMetaDataShort(path=%s)"%path)
341 return self.getFormatted('metadata_template', path=path, dom=dom, bibdata=bibdata)
342
343 def getFormattedMetaDataExtended(self,path=None, dom=None, bibdata=None):
344 """get the metadafrom server"""
345 logging.debug("getFormattedMetaDataExtended(path=%s)"%path)
346 return self.getFormatted('metadata_extended_template', path=path, dom=dom, bibdata=bibdata)
347
348 def getFormattedLabel(self,path=None, dom=None, bibdata=None):
349 """get the metadafrom server"""
350 logging.debug("getFormattedLabel(%s)"%path)
351 return self.getFormatted('label_template', path=path, dom=dom, bibdata=bibdata)
352
353 def getFormattedMetaDataShortFromServer(self,path):
354 """get the metadafrom server"""
355 logging.debug("getFormattedMetaDataShortFromServer(path=%s)"%path)
356 return self.getFormatted('metadata_template', path)
357
358 def getFormattedMetaDataExtendedFromServer(self,path):
359 """get the metadafrom server"""
360 logging.debug("getFormattedMetaDataExtendedFromServer(path=%s)"%path)
361 return self.getFormatted('metadata_extended_template', path)
362
363 def getFormattedLabelFromServer(self,path):
364 """get the metadafrom server"""
365 logging.debug("getFormattedLabelFromServer(%s)"%path)
366 return self.getFormatted('label_template', path)
367
368
369 security.declarePublic('changeMetadataForm')
370 def changeMetadataForm(self):
371 """Main configuration"""
372 pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','changeMetadata.zpt')).__of__(self)
373 return pt()
374
375 security.declarePublic('changeMetadata')
376 def changeMetadata(self,shortDescription,description,fields,metaDataServerUrl,RESPONSE=None):
377 """Change Metadata"""
378 self.shortDescription=shortDescription
379 self.description=description
380 self.fieldList=fields.split(",")[0:]
381 self.metaDataServerUrl=metaDataServerUrl
382 if RESPONSE is not None:
383 RESPONSE.redirect('manage_main')
384
385
386 def manage_addMetaDataForm(self):
387 """interface for adding the OSAS_add_Metadata"""
388 pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','addMetadataForm.zpt')).__of__(self)
389 return pt()
390
391 def manage_addMetaData(self,id,shortDescription,description,fields,RESPONSE=None):
392 """a metadata objekt"""
393 newObj=MetaData(id,shortDescription,description,fields)
394 self.Destination()._setObject(id,newObj)
395 if RESPONSE is not None:
396 RESPONSE.redirect('manage_main')
397
398 class MetaDataMapping(OSAS_MetadataMapping):
399 meta_type="MetadataMapping"
400
401 def manage_addMetaDataMappingForm(self):
402 """interface for adding the OSAS_root"""
403 pt=PageTemplateFile(os.path.join(package_home(globals()),'zpt','addMetadataMappingForm.zpt')).__of__(self)
404
405 return pt()
406
407 def manage_addMetaDataMapping(self,idOfObject,titleOfObject,RESPONSE=None):
408 """add the OSAS_root"""
409
410 argList={}
411 for arg in self.fieldList:
412 if not (arg in ['idOfObject','titleOfObject']):
413 argList[arg]=(self.REQUEST.form[arg],self.REQUEST.form['label_'+arg],self.REQUEST.form['explanation_'+arg],self.REQUEST.form['status_'+arg],self.REQUEST.form['values_'+arg])
414
415 newObj=MetaDataMapping(idOfObject,titleOfObject,argList)
416 self._setObject(idOfObject,newObj)
417 if RESPONSE is not None:
418 RESPONSE.redirect('manage_main')
419