Added support for schema constraints (#55)

* updated jsontableschema version * implemented all constraints except unique * implemented unique constraint check * updated to jsontableschema-v0.8.2 * fixed linting
frictionlessdata · Oct 29, 2016 · 66c5aea · 66c5aea
1 parent 08f8255
commit 66c5aea
Show file tree

Hide file tree

Showing 10 changed files with 104 additions and 20 deletions.
diff --git a/goodtables/checks/body/constraints/enumerable_constraint.py b/goodtables/checks/body/constraints/enumerable_constraint.py
@@ -11,7 +11,16 @@
 
 @check('enumerable-constraint')
 def enumerable_constraint(errors, columns, row_number, state=None):
-    # https://github.com/frictionlessdata/goodtables-py/issues/116
     for column in columns:
         if len(column) == 4:
-            pass
+            valid = column['field'].test_value(column['value'], constraint='enum')
+            if not valid:
+                # Add error
+                message = 'Row %s has enum constraint violation in column %s'
+                message = message % (row_number, column['number'])
+                errors.append({
+                    'code': 'enumerable-constraint',
+                    'message': message,
+                    'row-number': row_number,
+                    'column-number': column['number'],
+                })
diff --git a/goodtables/checks/body/constraints/maximum_constraint.py b/goodtables/checks/body/constraints/maximum_constraint.py
@@ -11,7 +11,16 @@
 
 @check('maximum-constraint')
 def maximum_constraint(errors, columns, row_number, state=None):
-    # https://github.com/frictionlessdata/goodtables-py/issues/116
     for column in columns:
         if len(column) == 4:
-            pass
+            valid = column['field'].test_value(column['value'], constraint='maximum')
+            if not valid:
+                # Add error
+                message = 'Row %s has maximum constraint violation in column %s'
+                message = message % (row_number, column['number'])
+                errors.append({
+                    'code': 'maximum-constraint',
+                    'message': message,
+                    'row-number': row_number,
+                    'column-number': column['number'],
+                })
diff --git a/goodtables/checks/body/constraints/maximum_length_constraint.py b/goodtables/checks/body/constraints/maximum_length_constraint.py
@@ -11,7 +11,16 @@
 
 @check('maximum-length-constraint')
 def maximum_length_constraint(errors, columns, row_number, state=None):
-    # https://github.com/frictionlessdata/goodtables-py/issues/116
     for column in columns:
         if len(column) == 4:
-            pass
+            valid = column['field'].test_value(column['value'], constraint='maxLength')
+            if not valid:
+                # Add error
+                message = 'Row %s has maxLength constraint violation in column %s'
+                message = message % (row_number, column['number'])
+                errors.append({
+                    'code': 'maximum-length-constraint',
+                    'message': message,
+                    'row-number': row_number,
+                    'column-number': column['number'],
+                })
diff --git a/goodtables/checks/body/constraints/minimum_constraint.py b/goodtables/checks/body/constraints/minimum_constraint.py
@@ -11,7 +11,16 @@
 
 @check('minimum-constraint')
 def minimum_constraint(errors, columns, row_number, state=None):
-    # https://github.com/frictionlessdata/goodtables-py/issues/116
     for column in columns:
         if len(column) == 4:
-            pass
+            valid = column['field'].test_value(column['value'], constraint='minimum')
+            if not valid:
+                # Add error
+                message = 'Row %s has minimum constraint violation in column %s'
+                message = message % (row_number, column['number'])
+                errors.append({
+                    'code': 'minimum-constraint',
+                    'message': message,
+                    'row-number': row_number,
+                    'column-number': column['number'],
+                })
diff --git a/goodtables/checks/body/constraints/minimum_length_constraint.py b/goodtables/checks/body/constraints/minimum_length_constraint.py
@@ -11,7 +11,16 @@
 
 @check('minimum-length-constraint')
 def minimum_length_constraint(errors, columns, row_number, state=None):
-    # https://github.com/frictionlessdata/goodtables-py/issues/116
     for column in columns:
         if len(column) == 4:
-            pass
+            valid = column['field'].test_value(column['value'], constraint='minLength')
+            if not valid:
+                # Add error
+                message = 'Row %s has minLength constraint violation in column %s'
+                message = message % (row_number, column['number'])
+                errors.append({
+                    'code': 'minimum-length-constraint',
+                    'message': message,
+                    'row-number': row_number,
+                    'column-number': column['number'],
+                })
diff --git a/goodtables/checks/body/constraints/pattern_constraint.py b/goodtables/checks/body/constraints/pattern_constraint.py
@@ -4,14 +4,26 @@
 from __future__ import absolute_import
 from __future__ import unicode_literals
 
+from copy import copy
 from ....register import check
 
 
 # Module API
 
 @check('pattern-constraint')
 def pattern_constraint(errors, columns, row_number, state=None):
-    # https://github.com/frictionlessdata/goodtables-py/issues/116
-    for column in columns:
+    for column in copy(columns):
         if len(column) == 4:
-            pass
+            valid = column['field'].test_value(column['value'], constraint='pattern')
+            if not valid:
+                # Add error
+                message = 'Row %s has pattern constraint violation in column %s'
+                message = message % (row_number, column['number'])
+                errors.append({
+                    'code': 'pattern-constraint',
+                    'message': message,
+                    'row-number': row_number,
+                    'column-number': column['number'],
+                })
+                # Remove column
+                columns.remove(column)
diff --git a/goodtables/checks/body/constraints/required_constraint.py b/goodtables/checks/body/constraints/required_constraint.py
@@ -4,14 +4,26 @@
 from __future__ import absolute_import
 from __future__ import unicode_literals
 
+from copy import copy
 from ....register import check
 
 
 # Module API
 
 @check('required-constraint')
 def required_constraint(errors, columns, row_number, state=None):
-    # https://github.com/frictionlessdata/goodtables-py/issues/116
-    for column in columns:
+    for column in copy(columns):
         if len(column) == 4:
-            pass
+            valid = column['field'].test_value(column['value'], constraint='required')
+            if not valid:
+                # Add error
+                message = 'Row %s has required constraint violation in column %s'
+                message = message % (row_number, column['number'])
+                errors.append({
+                    'code': 'required-constraint',
+                    'message': message,
+                    'row-number': row_number,
+                    'column-number': column['number'],
+                })
+                # Remove column
+                columns.remove(column)
diff --git a/goodtables/checks/body/constraints/unique_constraint.py b/goodtables/checks/body/constraints/unique_constraint.py
@@ -11,7 +11,21 @@
 
 @check('unique-constraint')
 def unique_constraint(errors, columns, row_number, state):
-    # https://github.com/frictionlessdata/goodtables-py/issues/116
+    rindexes = state.setdefault('cache', {})
     for column in columns:
         if len(column) == 4:
-            pass
+            if column['field'].constraints.get('unique'):
+                rindex = rindexes.setdefault(column['number'], {})
+                references = rindex.setdefault(column['value'], [])
+                if references:
+                    # Add error
+                    mesrows = ', '.join(map(str, references + [row_number]))
+                    message = 'Rows %s has unique constraint violation in column %s'
+                    message = message % (mesrows, column['number'])
+                    errors.append({
+                        'code': 'unique-constraint',
+                        'message': message,
+                        'row-number': row_number,
+                        'column-number': column['number'],
+                    })
+                references.append(row_number)
diff --git a/goodtables/checks/body/non_castable_value.py b/goodtables/checks/body/non_castable_value.py
@@ -16,7 +16,8 @@ def non_castable_value(errors, columns, row_number, state=None):
     for column in copy(columns):
         if len(column) == 4:
             try:
-                column['value'] = column['field'].cast_value(column['value'])
+                column['value'] = column['field'].cast_value(
+                    column['value'], skip_constraints=True)
             except jsontableschema.exceptions.JsonTableSchemaException:
                 # Add error
                 message = 'Row %s has non castable value in column %s'

diff --git a/setup.py b/setup.py
@@ -25,7 +25,7 @@ def read(*paths):
     'click>=6.6,<7.0a',
     'requests>=2.10,<3.0a',
     'tabulator>=0.10,<1.0a',
-    'jsontableschema>=0.7,<1.0a',
+    'jsontableschema>=0.8.2,<1.0a',
     'datapackage>=0.8,<1.0a',
 ]
 TESTS_REQUIRE = [