Merge b31274f into c755e7d

cartoframes/data/observatory/__init__.py

@@ -3,7 +3,7 @@
 from .catalog import Catalog
 from .category import Category
 from .country import Country
-from .dataset import Dataset
+from .dataset import CatalogDataset
 from .geography import Geography
 from .provider import Provider
 from .variable import Variable
@@ -13,7 +13,7 @@
     'Catalog',
     'Category',
     'Country',
-    'Dataset',
+    'CatalogDataset',
     'Geography',
     'Provider',
     'Variable'

cartoframes/data/observatory/catalog.py

@@ -4,7 +4,7 @@
 from .category import Category
 from .country import Country
 from .geography import Geography
-from .dataset import Dataset
+from .dataset import CatalogDataset
 from .repository.constants import COUNTRY_FILTER, CATEGORY_FILTER, GEOGRAPHY_FILTER
 
 
@@ -45,7 +45,7 @@ def datasets(self):
 
         """
 
-        return Dataset.get_all(self.filters)
+        return CatalogDataset.get_all(self.filters)
 
     @property
     def geographies(self):
@@ -128,4 +128,4 @@ def purchased_datasets(self, credentials=None):
 
         """
 
-        return Dataset.get_all(self.filters, credentials)
+        return CatalogDataset.get_all(self.filters, credentials)

cartoframes/data/observatory/dataset.py

@@ -8,7 +8,7 @@
 from .repository.constants import DATASET_FILTER
 
 
-class Dataset(CatalogEntity):
+class CatalogDataset(CatalogEntity):
     entity_repo = get_dataset_repo()
 
     @property

cartoframes/data/observatory/entity.py

@@ -21,6 +21,7 @@ class CatalogEntity(ABC):
 
     id_field = 'id'
     entity_repo = None
+    export_excluded_fields = ['summary_jsonb']
 
     def __init__(self, data):
         self.data = data
@@ -52,7 +53,7 @@ def to_series(self):
         return pd.Series(self.data)
 
     def to_dict(self):
-        return self.data
+        return {key: value for key, value in self.data.items() if key not in self.export_excluded_fields}
 
     def __eq__(self, other):
         return self.data == other.data

cartoframes/data/observatory/repository/dataset_repo.py

@@ -24,8 +24,8 @@ def get_all(self, filters=None, credentials=None):
 
     @classmethod
     def _get_entity_class(cls):
-        from cartoframes.data.observatory.dataset import Dataset
-        return Dataset
+        from cartoframes.data.observatory.dataset import CatalogDataset
+        return CatalogDataset
 
     def _get_rows(self, filters=None):
         return self.client.get_datasets(filters)

cartoframes/data/observatory/variable.py

@@ -6,6 +6,9 @@
 from .repository.constants import VARIABLE_FILTER
 
 
+_DESCRIPTION_LENGTH_LIMIT = 20
+
+
 class Variable(CatalogEntity):
 
     entity_repo = get_variable_repo()
@@ -64,3 +67,12 @@ def schema_name(self):
     def dataset_name(self):
         _, _, dataset, _ = self.id.split('.')
         return dataset
+
+    def __repr__(self):
+        descr = self.description
+
+        if len(descr) > _DESCRIPTION_LENGTH_LIMIT:
+            descr = descr[0:_DESCRIPTION_LENGTH_LIMIT] + '...'
+
+        return "<{classname}('{entity_id}','{descr}')>"\
+               .format(classname=self.__class__.__name__, entity_id=self._get_print_id(), descr=descr)

test/data/observatory/examples.py

@@ -1,5 +1,5 @@
 from cartoframes.data.observatory.variable import Variable
-from cartoframes.data.observatory.dataset import Dataset
+from cartoframes.data.observatory.dataset import CatalogDataset
 from cartoframes.data.observatory.category import Category
 from cartoframes.data.observatory.geography import Geography
 from cartoframes.data.observatory.country import Country
@@ -93,15 +93,15 @@
     'is_public_data': False,
     'summary_jsonb': {}
 }
-test_dataset1 = Dataset(db_dataset1)
-test_dataset2 = Dataset(db_dataset2)
+test_dataset1 = CatalogDataset(db_dataset1)
+test_dataset2 = CatalogDataset(db_dataset2)
 test_datasets = CatalogList([test_dataset1, test_dataset2])
 
 db_variable1 = {
     'id': 'carto-do.variable.var1',
     'slug': 'var1',
     'name': 'Population',
-    'description': 'The number of people within each geography',
+    'description': 'Number of people',
     'column_name': 'pop',
     'db_type': 'Numeric',
     'dataset_id': 'dataset1',

test/data/observatory/repository/test_dataset_repo.py

@@ -1,7 +1,7 @@
 import unittest
 
 from cartoframes.auth import Credentials
-from cartoframes.data.observatory.dataset import Dataset
+from cartoframes.data.observatory.dataset import CatalogDataset
 
 from cartoframes.exceptions import DiscoveryException
 from cartoframes.data.observatory.entity import CatalogList
@@ -176,7 +176,7 @@ def test_missing_fields_are_mapped_as_None(self, mocked_repo):
         mocked_repo.return_value = [{'id': 'dataset1'}]
         repo = DatasetRepository()
 
-        expected_datasets = CatalogList([Dataset({
+        expected_datasets = CatalogList([CatalogDataset({
             'id': 'dataset1',
             'slug': None,
             'name': None,

test/data/observatory/test_catalog.py

@@ -5,7 +5,7 @@
 from cartoframes.data.observatory.geography import Geography
 from cartoframes.data.observatory.country import Country
 from cartoframes.data.observatory.category import Category
-from cartoframes.data.observatory.dataset import Dataset
+from cartoframes.data.observatory.dataset import CatalogDataset
 from cartoframes.data.observatory.catalog import Catalog
 from cartoframes.data.observatory.repository.geography_repo import GeographyRepository
 from .examples import test_country2, test_country1, test_category1, test_category2, test_dataset1, test_dataset2, \
@@ -45,7 +45,7 @@ def test_categories(self, mocked_categories):
         # Then
         assert categories == expected_categories
 
-    @patch.object(Dataset, 'get_all')
+    @patch.object(CatalogDataset, 'get_all')
     def test_datasets(self, mocked_datasets):
         # Given
         expected_datasets = [test_dataset1, test_dataset2]
@@ -84,7 +84,7 @@ def test_filters_on_categories(self, mocked_categories):
         mocked_categories.called_once_with({'country_id': 'usa'})
         assert categories == test_categories
 
-    @patch.object(Dataset, 'get_all')
+    @patch.object(CatalogDataset, 'get_all')
     def test_filters_on_datasets(self, mocked_datasets):
         # Given
         mocked_datasets.return_value = test_datasets
@@ -110,7 +110,7 @@ def test_filters_on_geographies(self, mocked_geographies):
         mocked_geographies.called_once_with({'country_id': 'usa', 'category_id': 'demographics'})
         assert geographies == test_geographies
 
-    @patch.object(Dataset, 'get_all')
+    @patch.object(CatalogDataset, 'get_all')
     def test_all_filters(self, mocked_datasets):
         # Given
         mocked_datasets.return_value = test_datasets
@@ -128,7 +128,7 @@ def test_all_filters(self, mocked_datasets):
 
         assert datasets == test_datasets
 
-    @patch.object(Dataset, 'get_all')
+    @patch.object(CatalogDataset, 'get_all')
     @patch.object(GeographyRepository, 'get_by_id')
     def test_geography_filter_by_slug(self, mocked_repo, mocked_datasets):
         # Given
@@ -145,7 +145,7 @@ def test_geography_filter_by_slug(self, mocked_repo, mocked_datasets):
         mocked_datasets.assert_called_once_with({'geography_id': test_geography1.id})
         assert datasets == test_datasets
 
-    @patch.object(Dataset, 'get_all')
+    @patch.object(CatalogDataset, 'get_all')
     def test_purchased_datasets(self, mocked_purchased_datasets):
         # Given
         expected_datasets = [test_dataset1, test_dataset2]

test/data/observatory/test_category.py

@@ -104,7 +104,7 @@ def test_category_is_exported_as_dict(self):
         assert isinstance(category_dict, dict)
         assert category_dict == db_category1
 
-    def test_category_is_represented_with_id(self):
+    def test_category_is_represented_with_classname_and_id(self):
         # Given
         category = Category(db_category1)
 
@@ -137,7 +137,7 @@ def test_get_all(self, mocked_repo):
         assert isinstance(categories, CatalogList)
         assert categories == test_categories
 
-    def test_category_list_is_printed_with_classname(self):
+    def test_category_list_is_printed_with_classname_and_ids(self):
         # Given
         categories = CatalogList([test_category1, test_category2])
 
@@ -148,7 +148,7 @@ def test_category_list_is_printed_with_classname(self):
         assert categories_str == "[<Category('{id1}')>, <Category('{id2}')>]" \
                                  .format(id1=db_category1['id'], id2=db_category2['id'])
 
-    def test_category_list_is_represented_with_ids(self):
+    def test_category_list_is_represented_with_classname_and_ids(self):
         # Given
         categories = CatalogList([test_category1, test_category2])
 

test/data/observatory/test_country.py

@@ -117,7 +117,7 @@ def test_country_is_exported_as_dict(self):
         assert isinstance(country_dict, dict)
         assert country_dict == db_country1
 
-    def test_country_is_represented_with_id(self):
+    def test_country_is_represented_with_classname_and_id(self):
         # Given
         country = Country(db_country1)
 
@@ -150,7 +150,7 @@ def test_get_all_countries(self, mocked_repo):
         assert isinstance(countries, CatalogList)
         assert countries == test_countries
 
-    def test_country_list_is_printed_with_classname(self):
+    def test_country_list_is_printed_with_classname_and_ids(self):
         # Given
         countries = CatalogList([test_country1, test_country2])
 
@@ -161,7 +161,7 @@ def test_country_list_is_printed_with_classname(self):
         assert countries_str == "[<Country('{id1}')>, <Country('{id2}')>]" \
                                 .format(id1=db_country1['id'], id2=db_country2['id'])
 
-    def test_country_list_is_represented_with_ids(self):
+    def test_country_list_is_represented_with_classname_and_ids(self):
         # Given
         countries = CatalogList([test_country1, test_country2])