From 2e2a22a2f21e6b1acf7744f964a12ae5229b8b55 Mon Sep 17 00:00:00 2001
From: darinyu-coursera <dyu@coursera.org>
Date: Tue, 17 Nov 2015 10:57:06 -0800
Subject: [PATCH 1/3] allow new step to take in script_arguments

---
 dataduct/steps/load_reload_pk.py | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/dataduct/steps/load_reload_pk.py b/dataduct/steps/load_reload_pk.py
index b35b1c0..d96820f 100644
--- a/dataduct/steps/load_reload_pk.py
+++ b/dataduct/steps/load_reload_pk.py
@@ -21,8 +21,8 @@ class LoadReloadAndPrimaryKeyStep(ETLStep):
 
     def __init__(self, id, input_node, staging_table_definition,
                  production_table_definition, pipeline_name,
-                 analyze_table=True, non_transactional=False,
-                 log_to_s3=False, **kwargs):
+                 script_arguments=None, analyze_table=True,
+                 non_transactional=False, log_to_s3=False, **kwargs):
         """Constructor for the LoadReloadAndPrimaryKeyStep class
 
         Args:
@@ -39,7 +39,8 @@ def __init__(self, id, input_node, staging_table_definition,
         # they support all the parameters
         create_and_load_pipeline_object = self.create_and_load_redshift(
             table_definition=staging_table_definition,
-            input_node=input_node
+            input_node=input_node,
+            script_arguments=script_arguments
         )
 
         reload_pipeline_object = self.reload(
@@ -139,7 +140,10 @@ def reload(self, source, destination, depends_on,
         return reload_pipeline_object
 
 
-    def create_and_load_redshift(self, table_definition, input_node):
+    def create_and_load_redshift(self, table_definition,
+                                 input_node, script_arguments):
+        if not script_arguments:
+            script_arguments = list()
         table = self.get_table_from_def(table_definition)
 
         if isinstance(input_node, dict):

From d562a69ab682c0898760e9ecbdc46d20fc22982f Mon Sep 17 00:00:00 2001
From: darinyu-coursera <dyu@coursera.org>
Date: Tue, 17 Nov 2015 10:59:44 -0800
Subject: [PATCH 2/3] fix script_arguments override

---
 dataduct/steps/load_reload_pk.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/dataduct/steps/load_reload_pk.py b/dataduct/steps/load_reload_pk.py
index d96820f..5472ac5 100644
--- a/dataduct/steps/load_reload_pk.py
+++ b/dataduct/steps/load_reload_pk.py
@@ -151,10 +151,10 @@ def create_and_load_redshift(self, table_definition,
         else:
             input_paths = [input_node.path().uri]
 
-        script_arguments = [
+        script_arguments.extend([
             '--table_definition=%s' % table.sql().sql(),
             '--s3_input_paths'
-        ]
+        ])
         script_arguments.extend(input_paths)
 
         steps_path = os.path.abspath(os.path.dirname(__file__))

From b456b133c1cd49b3bc92f1c249e572e88420208d Mon Sep 17 00:00:00 2001
From: darinyu-coursera <dyu@coursera.org>
Date: Tue, 17 Nov 2015 11:49:56 -0800
Subject: [PATCH 3/3] Add documentation for new step

---
 docs/steps.rst | 40 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)

diff --git a/docs/steps.rst b/docs/steps.rst
index ff26a1e..e7e83f9 100644
--- a/docs/steps.rst
+++ b/docs/steps.rst
@@ -307,6 +307,46 @@ Example
     -   step_type: create-load-redshift
         table_definition: tables/dev.example_table.sql
 
+Load, Reload, Primary Key Check
+----------------------------------
+
+Combine ``create-load-redshift``, ``reload`` and ``primary-key-check`` into one single step.
+
+Properties
+^^^^^^^^^^
+
+-  ``staging_table_definition``: Intermidiate staging schema file for the table to be loaded into.
+   (Required)
+-  ``production_table_definition``: Production schema file for the table to be reloaded into.
+   (Required)
+-  ``script_arguments``: Arguments for the runner.
+
+   -  ``--max_error``: The maximum number of errors to be ignored during
+      the load. Usage: ``--max_error=5``
+   -  ``--replace_invalid_char``: Character the replace non-utf8
+      characters with. Usage: ``--replace_invalid_char='?'``
+   -  ``--no_escape``: If passed, does not escape special characters.
+      Usage: ``--no_escape``
+   -  ``--gzip``: If passed, compresses the output with gzip. Usage:
+      ``--gzip``
+   -  ``--command_options``: A custom SQL string as the options for the
+      copy command. Usage: ``--command_options="DELIMITER '\t'"``
+
+      -  Note: If ``--command_options`` is passed, script arguments
+         ``--max_error``, ``--replace_invalid_char``, ``--no_escape``,
+         and ``--gzip`` have no effect.
+
+Example
+^^^^^^^
+
+::
+
+    -   step_type: load-reload-pk
+        staging_table_definition: tables/staging.example_table.sql
+        production_table_definition: tables/dev.example_table.sql
+        script_arguments:
+        -   "--foo=bar"
+
 Upsert
 -------------------------