Changeset 1619

Timestamp:

11/10/11 17:13:49 (13 years ago)

Author:

dietuyt

Message:

Several fixes to ensure validation against the relevant XSDs

Location:

metadata/trunk/toolkit/scripts

Files:

• dir2cmdicollection.py (modified) (11 diffs)
• lrt2cmdi.py (modified) (3 diffs)

metadata/trunk/toolkit/scripts/dir2cmdicollection.py

@@ -3 +3 @@
 # generates CMDI collection file hierarchy for collections of CMDI records
 # support and questions: Dieter Van Uytvanck <dietuyt@mpi.nl>
-# rework by Matej id@vronk.net : 
+# rework by Matej id@vronk.net :
 #       - already filling ResourceRef with handles read from the MdSelfLink of the mdrecords
 #   - also reading ProviderURL-file and filling as ID in the basic collection-profile
@@ -9 +9 @@
 
 # params:
-#  ref_flag := what to put into ResourceRef: 
-#                       'handle' := PID (read from MdSelfLink of the resources), 
+#  ref_flag := what to put into ResourceRef:
+#                       'handle' := PID (read from MdSelfLink of the resources),
 #                       'path' := the relative path of the record
 
@@ -29 +29 @@
         if not os.path.isdir(target_dir):
                 os.mkdir(target_dir)
-        
+
         if not simple_dir_structure:
-                root = os.getcwd()              
-                dirs = filter (os.path.isdir, os.listdir(root))         
+                root = os.getcwd()
+                dirs = filter (os.path.isdir, os.listdir(root))
                 dirs.remove(target_dir.replace("/",""))
                 #for root, dirs, files in os.walk(os.getcwd()):
-                print dirs              
-                for d in dirs:          
+                print dirs
+                for d in dirs:
                         rootList.append(generate_branch(root, [d], d))
-                        #startpath = os.getcwd()                
+                        #startpath = os.getcwd()
                         #for d in dirs:
                                 #if d == "0":
-                                
+
         else:
                 rootList.append(generate_branch(os.getcwd(), [""], "lrt_inventory"))
@@ -54 +54 @@
         #collectionName = os.path.relpath(root)
         collectionFile = "collection_%s.cmdi" % collectionName
-        
+
         dirs.sort()
-        collectionList = []     
+        collectionList = []
         for d in dirs:
                 fullpath = os.path.join(root, d)
@@ -63 +63 @@
                         if ".cmdi" in file:
                                 newFile = os.path.relpath(os.path.join(fullpath,file))
-                                collectionList.append(newFile)          
+                                collectionList.append(newFile)
         collid = writeCollection(collectionList, collectionFile, collectionName)
         print "genbranch:" + collid
@@ -74 +74 @@
 
         outstring = Template("""<?xml version="1.0" encoding="UTF-8"?>
-<CMD xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+<CMD xmlns="http://www.clarin.eu/cmd/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" CMDVersion="1.1"
     xsi:schemaLocation="http://www.clarin.eu/cmd http://catalog.clarin.eu/ds/ComponentRegistry/rest/registry/profiles/clarin.eu:cr1:p_1284723009187/xsd">
     <Header>
@@ -87 +87 @@
         <JournalFileProxyList/>
         <ResourceRelationList/>
-    </Resources>    
+    </Resources>
     <Components>
         <collection>
@@ -99 +99 @@
 
         resourceProxies = ""
-        collectionList.sort()           
+        collectionList.sort()
         if os.path.isfile(collectionName + "/ProviderURL"):
                 urlf = open(collectionName + "/ProviderURL", 'r')
                 url = urlf.readline()
-        else: 
+        else:
           url ="?"
         name = "OLAC: " + collectionName.replace("_", " ")
@@ -114 +114 @@
                                         if "<MdSelfLink>" in line:
                                                 #  WARNING! rocket science employed here !
-                                                idx = line.replace("<MdSelfLink>","").replace("</MdSelfLink>","").strip()                                               
+                                                idx = line.replace("<MdSelfLink>","").replace("</MdSelfLink>","").strip()
                                                 break
                 else:
                          idx = item
-                         
+
                 if ref_flag=="path":
                         # need to add '../' for the number of dirs in the target_dir, so paths match again.
@@ -124 +124 @@
                 else:
                         path = idx;
-                        
+
                 #idx = item.replace(".xml.cmdi","").replace("_", ":",1)[::-1].replace("_", ":",1)[::-1].replace("_", "-")
                 resourceProxies += "\n" + resourceTemplate.substitute(idname = idx.replace(".","_").replace("/","_").replace("\\","_").replace(":","_"), idx = path)
@@ -135 +135 @@
                 else:
                         collidx = collectionFile
-        print collidx 
+        print collidx
         outfile = outstring.substitute(date= datetime.datetime.now().strftime("%Y-%m-%d"), selflink=collidx, rp=resourceProxies,url=url, name=name)
         collectionFile = target_dir + collectionFile
-        f = open(collectionFile, 'w')   
+        f = open(collectionFile, 'w')
         f.write(outfile)
         f.close()
-        
+
         print collectionFile
         if ref_flag=="path":

metadata/trunk/toolkit/scripts/lrt2cmdi.py

@@ -40 +40 @@
 
     def removeEmptyNodes(self):
-        removeList = ["ResourceType", "BeginYearResourceCreation", "FinalizationYearResourceCreation", "Institute"]
+        # we added some elements so need to recalculate the parentmap
+        self.parentmap = dict((c, p) for p in self.xmlTree.getiterator() for c in p)
+
+        removeList = ["ResourceType", "BeginYearResourceCreation", "FinalizationYearResourceCreation", "Institute", \
+                      "DistributionType", "NonCommercialUsageOnly", "UsageReportRequired", "ModificationsRequireRedeposition", "WorkingLanguages", "Date"]
         for r in removeList:
             results = self.xmlTree.findall("//%s" % r)
@@ -161 +165 @@
                 <Size />
                 <Access />
-                <WorkingLanguages/>
+                <WorkingLanguages />
             </LrtLexiconDetails>'''
         partTree = ElementTree.fromstring(template)
@@ -175 +179 @@
         self.fillElement("//LrtLexiconDetails/Access", record["field_access_1"])
         self.addLanguages(isoList, record["field_working_languages_0"], 1, "//LrtLexiconDetails/WorkingLanguages")
-
 
     def addServiceDetails(self, record):