[disclosr.git] / documents / datagov-export.py
--- a/documents/datagov-export.py
+++ b/documents/datagov-export.py
@@ -121,28 +121,31 @@
             # Collect the package metadata.
             pkg_name = doc.value['url'].replace("http://data.gov.au/dataset/",'').replace('/','');                                                                  _
             tags = []
-            if len(doc.value['metadata']["Keywords / Tags"]) > 0:
-                if hasattr(doc.value['metadata']["Keywords / Tags"], '__iter__'):
-                    tags = tags + doc.value['metadata']["Keywords / Tags"]
-                else:
-                    tags = tags + [doc.value['metadata']["Keywords / Tags"]]
-            if 'data.gov.au Category' in doc.value['metadata'].keys() and len(doc.value['metadata']['data.gov.au Category']) > 0:
-                if hasattr(doc.value['metadata']['data.gov.au Category'], '__iter__'):
-                    tags = tags + doc.value['metadata']['data.gov.au Category']
-                else:
-                    tags = tags + [doc.value['metadata']['data.gov.au Category']]
-            tags = [re.sub('[^a-zA-Z0-9-_.]', '', tag.replace('&', 'and')).lower() for tag in tags if tag]
-            print tags
-            package_entity = {
-                'name': pkg_name,
-                'title': doc.value['metadata']['DCTERMS.Title'],
-                'url': doc.value['metadata']['DCTERMS.Source.URI'],
-                'tags': tags, #tags are mandatory?
-                'author': doc.value['metadata']["DCTERMS.Creator"],
-                'maintainer': doc.value['metadata']["DCTERMS.Creator"],
-                'licence_id': get_licence_id(doc.value['metadata']['DCTERMS.License']),
-                'notes': html2text.html2text(doc.value['metadata']['Description']),
-            }
+            if doc.value['agencyID'] == "AGIMO":
+                if len(doc.value['metadata']["Keywords / Tags"]) > 0:
+                    if hasattr(doc.value['metadata']["Keywords / Tags"], '__iter__'):
+                        tags = tags + doc.value['metadata']["Keywords / Tags"]
+                    else:
+                        tags = tags + [doc.value['metadata']["Keywords / Tags"]]
+                if 'data.gov.au Category' in doc.value['metadata'].keys() and len(doc.value['metadata']['data.gov.au Category']) > 0:
+                    if hasattr(doc.value['metadata']['data.gov.au Category'], '__iter__'):
+                        tags = tags + doc.value['metadata']['data.gov.au Category']
+                    else:
+                        tags = tags + [doc.value['metadata']['data.gov.au Category']]
+                tags = [re.sub('[^a-zA-Z0-9-_.]', '', tag.replace('&', 'and')).lower() for tag in tags if tag]
+                print tags
+                package_entity = {
+                    'name': pkg_name,
+                    'title': doc.value['metadata']['DCTERMS.Title'],
+                    'url': doc.value['metadata']['DCTERMS.Source.URI'],
+                    'tags': tags, #tags are mandatory?
+                    'author': doc.value['metadata']["DCTERMS.Creator"],
+                    'maintainer': doc.value['metadata']["DCTERMS.Creator"],
+                    'licence_id': get_licence_id(doc.value['metadata']['DCTERMS.License']),
+                    'notes': html2text.html2text(doc.value['metadata']['Description']),
+                }
+            if doc.value['agencyID'] == "qld":
+                package_entity = doc.value['metadata']
 
             try:
                 print package_entity