Add the script fix_harvesters_collections.py.

scripts/fix_harvesters_collections.py

+# -*- coding: utf-8 -*-
+""" NAME
+        fix_harvesters_collections
+
+    SYNOPSIS
+        fix_harvesters_collections [options]
+
+    DESCRIPTION
+        Up to version 0.9.6.5, the field collections can contains a list
+        of collection separated by a comma. This rule is in conflict when
+        defining a collection using author name since author name can contain
+        also comma.
+
+        In version 0.9.6.6, it has been decided that the collection field
+        contains the definition for only one collection.
+
+        This script allow to migrate from 0.9.6.5 to 0.9.6.6 approach.
+
+    OPTIONS
+
+    EXAMPLE
+
+        > cd ...limbra/scripts
+        > run -S test_limbra script fix_harvesters_collections.py
+        > run loop fix_harvesters_collections.py
+        > ...
+
+    AUTHOR
+        R. Le Gac -- Oct 2017
+
+"""
+if __name__ == "__main__":
+
+    import sys
+
+    # scan the harvesters table
+    for row in db(db.harvesters).iterselect():
+
+        collections = row.collections
+        if row.host == "cds.cern.ch" and "," in collections:
+
+            print "\n\tsplit:", collections
+            data = row.as_dict()
+            del data["id"]
+
+            # insert new harvesters
+            for elt in collections.split(","):
+                print "\t\tinsert →", elt.strip()
+                data["collections"] = elt.strip()
+                db.harvesters.insert(**data)
+
+            # delete old harvesters
+            print "\t\tdelete →", row.id
+            del db.harvesters[row.id]
+
+    # commit change
+    rep = raw_input("Commit database changes [y/N]: ")
+    if rep == 'y':
+        db.commit()
+
+    # exit gently
+    sys.exit(0)