Changeset 905

Timestamp:

11/15/10 14:49:32 (14 years ago)

Author:

dietuyt

Message:

Small fixes to lrt2cmdi and added a -simpledir option to
dir2cmdicollection.py for the LRT inventory (just processes all cmdi
files in the current directory)

Location:

metadata/trunk/toolkit/scripts

Files:

• dir2cmdicollection.py (modified) (2 diffs)
• lrt2cmdi.py (modified) (4 diffs)

metadata/trunk/toolkit/scripts/dir2cmdicollection.py

@@ -13 +13 @@
 #                       'path' := the relative path of the record
 
-import os, datetime
+import os, datetime, sys
 from string import Template
 
@@ -19 +19 @@
 ref_flag = "path" # "handle" | "path"
 
+
 def main():
+        simple_dir_structure = False
+        if len(sys.argv) > 1:
+                if sys.argv[1] == "-simpledir":
+                        simple_dir_structure = True
+
         rootList = []
         if not os.path.isdir(target_dir):
                 os.mkdir(target_dir)
-        for root, dirs, files in os.walk(os.getcwd()):
-                startpath = os.getcwd()         
-                for d in dirs:
-                        if d == "0":
-                                rootList.append(generate_branch(root, dirs))
+        
+        if not simple_dir_structure:
+                for root, dirs, files in os.walk(os.getcwd()):
+                        startpath = os.getcwd()         
+                        for d in dirs:
+                                if d == "0":
+                                        rootList.append(generate_branch(root, dirs))
+        else:
+                rootList.append(generate_branch(os.getcwd(), [""], "lrt_inventory"))
         writeCollection(rootList, "collection_root.cmdi", "olac-root")
-                
+
 def generate_branch(root, dirs):
         collectionName = os.path.relpath(root)
+        return generate_branch(root, dirs, collectionName)
+
+def generate_branch(root, dirs, collectionName):
+        #collectionName = os.path.relpath(root)
         collectionFile = "collection_%s.cmdi" % collectionName
         

metadata/trunk/toolkit/scripts/lrt2cmdi.py

@@ -17 +17 @@
 
     def fillElement(self, xpath, value):
+        #print "fill %s with %s" % (xpath, value)
         self.xmlTree.find(xpath).text = value.strip() 
         
     def fillMultipleElement(self, elementname, xpath, values):
         # fill in the already existing element
-        if values[0]:
+        #print values
+        if (values[0]):
+            #print "first one", values[0]
+            #print values[0]
             self.fillElement(xpath, values[0])
+            #print "fill %s with %s" % (xpath, values[0])
+        #print
         
         element = self.xmlTree.find(xpath)
@@ -31 +37 @@
         for value in values[1:]:
             if value:
-                # create new sibling of xpath (elementname) = value            
+                # create new sibling of xpath (elementname) = value
+                #print value
+                #print "next one", value
                 position += 1
                 newElement = ElementTree.Element(elementname)
@@ -56 +64 @@
                 uniqueItems = set(items) # filter out double items
                 items = [i for i in uniqueItems] # convert set back to a list
+                #print items
+                 
                 self.fillMultipleElement("Institute", "//LrtCommon/Institute", items)
     
@@ -193 +203 @@
             
 def loadInfo():
-    #csvFile = csv.reader(urllib.urlopen("http://www.clarin.eu/export_resources").readlines())
-    csvFile = csv.reader(urllib.urlopen("resources.csv").readlines())  
+    csvFile = csv.reader(urllib.urlopen("http://www.clarin.eu/export_resources").readlines())
+    #csvFile = csv.reader(urllib.urlopen("resources.csv").readlines())  
     linenr = 0
     newDict = dict()