Add MapFields transformation

bonobo/__init__.py

@@ -29,6 +29,7 @@ from bonobo._api import (
     LdjsonReader,
     LdjsonWriter,
     Limit,
+    MapFields,
     OrderFields,
     PickleReader,
     PickleWriter,

bonobo/_api.py

@@ -149,6 +149,7 @@ api.register_group(
     LdjsonReader,
     LdjsonWriter,
     Limit,
+    MapFields,
     OrderFields,
     PickleReader,
     PickleWriter,

bonobo/errors.py

@@ -48,6 +48,10 @@ class UnrecoverableTypeError(UnrecoverableError, TypeError):
     pass
 
 
+class UnrecoverableAttributeError(UnrecoverableError, AttributeError):
+    pass
+
+
 class UnrecoverableValueError(UnrecoverableError, ValueError):
     pass
 

bonobo/nodes/basics.py

@@ -10,6 +10,7 @@ from bonobo.config import Configurable, Method, Option, use_context, use_no_inpu
 from bonobo.config.functools import transformation_factory
 from bonobo.config.processors import ContextProcessor, use_context_processor
 from bonobo.constants import NOT_MODIFIED
+from bonobo.errors import UnrecoverableAttributeError
 from bonobo.util.objects import ValueHolder
 from bonobo.util.term import CLEAR_EOL
 
@@ -18,6 +19,7 @@ __all__ = [
     "Format",
     "Limit",
     "OrderFields",
+    "MapFields",
     "PrettyPrinter",
     "Rename",
     "SetFields",
@@ -314,6 +316,40 @@ def Format(**formats):
     return _Format
 
 
+@transformation_factory
+def MapFields(function, key=True):
+    """
+    Transformation factory that maps `function` on the values of a row.
+    It can be applied either to
+    1. all columns (`key=True`),
+    2. no column (`key=False`), or
+    3. a subset of columns by passing a callable, which takes column name and returns `bool`
+     (same as the parameter `function` in `filter`).
+
+    :param function: callable
+    :param key: bool or callable
+    :return: callable
+    """
+    @use_raw_input
+    def _MapFields(bag):
+        try:
+            factory = type(bag)._make
+        except AttributeError as e:
+            raise UnrecoverableAttributeError('This transformation works only on objects with named'
+                                              ' fields (namedtuple, BagType, ...).') from e
+
+        if callable(key):
+            return factory(
+                function(value) if key(key_) else value for key_, value in zip(bag._fields, bag)
+            )
+        elif key:
+            return factory(function(value) for value in bag)
+        else:
+            return NOT_MODIFIED
+
+    return _MapFields
+
+
 def _count(self, context):
     counter = yield ValueHolder(0)
     context.send(counter.get())