Merge branch 'ogdch-79-optional-tag-munging'

README.rst

@@ -206,6 +206,12 @@ field. The currently supported configuration options are:
     present in the local CKAN. Setting it to 'create' will make an attempt to
     create the organizations by copying the details from the remote CKAN.
 
+*   clean_tags: By default, tags are not stripped of accent characters, spaces and
+    capital letters for display. If this option is set to True, accent characters
+    will be replaced by their ascii equivalents, capital letters replaced by
+    lower-case ones, and spaces replaced with dashes. Setting this option to False
+    gives the same effect as leaving it unset.
+
 Here is an example of a configuration object (the one that must be entered in
 the configuration field)::
 

ckanext/harvest/harvesters/base.py

@@ -20,6 +20,7 @@ from ckanext.harvest.model import HarvestJob, HarvestObject, HarvestGatherError,
 from ckan.plugins.core import SingletonPlugin, implements
 from ckanext.harvest.interfaces import IHarvester
 
+
 log = logging.getLogger(__name__)
 
 
@@ -150,10 +151,11 @@ class HarvesterBase(SingletonPlugin):
                 'ignore_auth': True,
             }
 
-            tags = package_dict.get('tags', [])
+            if self.config and self.config.get('clean_tags', False):
-            tags = [munge_tag(t) for t in tags]
+                tags = package_dict.get('tags', [])
-            tags = list(set(tags))
+                tags = [munge_tag(t) for t in tags]
-            package_dict['tags'] = tags
+                tags = list(set(tags))
+                package_dict['tags'] = tags
 
             # Check if package exists
             data_dict = {}