Allow set type to receive re as field name (#26)

* Allow set type to receive re as field name

* v0.0.18

PROCESSORS.md

@@ -288,7 +288,7 @@ def set_type(name, resources=-1, **options):
     pass
 ```
 
-- `name` - the name of the field to modify
+- `name` - the name of the field to modify (or a regular expression to match multiple fields)
 - `resources`
   - A name of a resource to operate on
   - A regular expression matching resource names

dataflows/VERSION

@@ -1 +1 @@
-0.0.17
+0.0.18

dataflows/processors/set_type.py

@@ -1,3 +1,5 @@
+import re
+
 from ..helpers.resource_matcher import ResourceMatcher
 from .. import DataStreamProcessor, schema_validator
 
@@ -6,7 +8,7 @@ class set_type(DataStreamProcessor):
 
     def __init__(self, name, resources=-1, **options):
         super(set_type, self).__init__()
-        self.name = name
+        self.name = re.compile(name)
         self.options = options
         self.resources = resources
 
@@ -24,9 +26,8 @@ def process_datapackage(self, dp):
         for res in dp.descriptor['resources']:
             if self.matcher.match(res['name']):
                 for field in res['schema']['fields']:
-                    if field['name'] == self.name:
+                    if self.name.match(field['name']):
                         field.update(self.options)
                         added = True
-                        break
         assert added, 'Failed to find field {} in schema'.format(self.name)
         return dp

tests/test_lib.py

@@ -387,15 +387,15 @@ def test_set_type_resources():
     f = Flow(
         [dict(a=str(i)) for i in range(10)],
         [dict(b=str(i)) for i in range(10)],
-        [dict(c=str(i)) for i in range(10)],
+        [dict(c='0_' + str(i)) for i in range(10)],
         set_type('a', resources='res_[1]', type='integer'),
         set_type('b', resources=['res_2'], type='integer'),
-        set_type('c', resources=-1, type='integer'),
+        set_type('[cd]', resources=-1, type='number', groupChar='_'),
         validate()
     )
     results, dp, stats = f.results()
     print(dp.descriptor) 
     assert results[0][1]['a'] == 1
     assert results[1][3]['b'] == 3
-    assert results[2][8]['c'] == 8
+    assert results[2][8]['c'] == 8.0