Add new action to infer a tabular resource schema

.github/workflows/test.yml

@@ -58,9 +58,10 @@ jobs:
       run: |
         ckan -c test.ini db init
     - name: Run tests
-      run: pytest --ckan-ini=test.ini --cov=ckanext.validation --cov-report=xml --cov-append --disable-warnings ckanext/validation/tests -vv
+      #      run: pytest --ckan-ini=test.ini --cov=ckanext.validation --cov-report=xml --cov-append --disable-warnings ckanext/validation/tests -vv
+      run: pytest --ckan-ini=test.ini --disable-warnings ckanext/validation/tests -vv
 
-    - name: Upload coverage report to codecov
-      uses: codecov/codecov-action@v1
-      with:
-        file: ./coverage.xml
+        #- name: Upload coverage report to codecov
+        #  uses: codecov/codecov-action@v1
+        #  with:
+        #    file: ./coverage.xml

ckanext/validation/logic.py

@@ -5,6 +5,7 @@
 import json
 
 from sqlalchemy.orm.exc import NoResultFound
+from frictionless import system, Resource
 
 import ckan.plugins as plugins
 import ckan.lib.uploader as uploader
@@ -174,6 +175,42 @@ def resource_validation_show(context, data_dict):
 
     return _validation_dictize(validation)
 
+def resource_table_schema_infer(context, data_dict):
+    '''
+    Use frictionless framework to infer a resource schema
+    '''
+
+    t.check_access('resource_create', context, data_dict)
+
+    t.get_or_bust(data_dict, 'resource_id')
+
+    store_schema = data_dict.get('store_schema', True)
+
+    resource = t.get_action('resource_show')(
+        {}, {u'id': data_dict['resource_id']})
+
+    source = None
+    if resource.get('url_type') == 'upload':
+        upload = uploader.get_resource_uploader(resource)
+        if isinstance(upload, uploader.ResourceUpload):
+            source = upload.get_path(resource['id'])
+
+    if not source:
+        source = resource['url']
+
+    with system.use_context(trusted=True):
+        # TODO: check for valid formats
+        fric_resource = Resource({'path': source, 'format': resource.get('format', 'csv').lower()})
+        fric_resource.infer()
+        resource['schema'] = fric_resource.schema.to_json()
+
+        # TODO: check for exception
+        if store_schema:
+            t.get_action('resource_update')(
+                context, resource)
+
+        return {u'schema': fric_resource.schema.to_dict()}
+
 
 def resource_validation_delete(context, data_dict):
     u'''

ckanext/validation/plugin/__init__.py

@@ -6,6 +6,7 @@
 
 from werkzeug.datastructures import FileStorage as FlaskFileStorage
 import ckan.plugins as p
+import ckan.lib.uploader as uploader
 import ckantoolkit as t
 
 from ckanext.validation import settings
@@ -17,6 +18,7 @@
     auth_resource_validation_delete, auth_resource_validation_run_batch,
     resource_create as custom_resource_create,
     resource_update as custom_resource_update,
+    resource_table_schema_infer
 )
 from ckanext.validation.helpers import (
     get_validation_badge,
@@ -34,6 +36,7 @@
     get_create_mode_from_config,
     get_update_mode_from_config,
 )
+
 from ckanext.validation.interfaces import IDataValidation
 from ckanext.validation import blueprints, cli
 
@@ -89,6 +92,7 @@ def get_actions(self):
             u'resource_validation_run_batch': resource_validation_run_batch,
             u'resource_create': custom_resource_create,
             u'resource_update': custom_resource_update,
+            u'resource_table_schema_infer': resource_table_schema_infer
         }
 
         return new_actions