Make FileFormat respect missing values (#129)

* Make FileFormat respect missing values

* Fixed the implementation

* Fixed tests

dataflows/helpers/iterable_loader.py

@@ -29,12 +29,12 @@ def field_type(self, values):
         for value in values:
             if isinstance(value, str):
                 types.add('string')
+            elif isinstance(value, bool):
+                types.add('boolean')
             elif isinstance(value, int):
                 types.add('integer')
             elif isinstance(value, (float, decimal.Decimal)):
                 types.add('number')
-            elif isinstance(value, bool):
-                types.add('boolean')
             elif isinstance(value, list):
                 types.add('array')
             elif isinstance(value, dict):

dataflows/processors/dumpers/file_formats.py

@@ -34,6 +34,7 @@ def __init__(self, writer, schema, temporal_format_property=None):
         self.headers = [f.name for f in schema.fields]
         self.fields = dict((f.name, f) for f in schema.fields)
         self.temporal_format_property = temporal_format_property
+        self.missing_values = schema.descriptor.get('missingValues', [])
 
         # Set fields' serializers
         for field in schema.fields:
@@ -62,6 +63,10 @@ def __transform_row(self, row):
     def __transform_value(self, value, field):
         if value is None:
             return self.NULL_VALUE
+        # It supports a `tableschema`'s mode of perserving missing values
+        # https://github.com/frictionlessdata/tableschema-py#experimental
+        if value in self.missing_values:
+            return value
         return field.descriptor['serializer'](value)
 
     def write_transformed_row(self, *_):