Mercurial > repos > guerler > springsuite
annotate planemo/lib/python3.7/site-packages/boto/emr/step.py @ 0:d30785e31577 draft
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
| author | guerler | 
|---|---|
| date | Fri, 31 Jul 2020 00:18:57 -0400 | 
| parents | |
| children | 
| rev | line source | 
|---|---|
| 
0
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
1 # Copyright (c) 2010 Spotify AB | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
2 # Copyright (c) 2010-2011 Yelp | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
3 # | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
4 # Permission is hereby granted, free of charge, to any person obtaining a | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
5 # copy of this software and associated documentation files (the | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
6 # "Software"), to deal in the Software without restriction, including | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
7 # without limitation the rights to use, copy, modify, merge, publish, dis- | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
8 # tribute, sublicense, and/or sell copies of the Software, and to permit | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
9 # persons to whom the Software is furnished to do so, subject to the fol- | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
10 # lowing conditions: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
11 # | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
12 # The above copyright notice and this permission notice shall be included | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
13 # in all copies or substantial portions of the Software. | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
14 # | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
15 # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
16 # OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABIL- | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
17 # ITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
18 # SHALL THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
19 # WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
20 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
21 # IN THE SOFTWARE. | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
22 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
23 from boto.compat import six | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
24 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
25 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
26 class Step(object): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
27 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
28 Jobflow Step base class | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
29 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
30 def jar(self): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
31 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
32 :rtype: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
33 :return: URI to the jar | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
34 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
35 raise NotImplemented() | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
36 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
37 def args(self): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
38 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
39 :rtype: list(str) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
40 :return: List of arguments for the step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
41 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
42 raise NotImplemented() | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
43 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
44 def main_class(self): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
45 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
46 :rtype: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
47 :return: The main class name | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
48 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
49 raise NotImplemented() | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
50 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
51 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
52 class JarStep(Step): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
53 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
54 Custom jar step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
55 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
56 def __init__(self, name, jar, main_class=None, | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
57 action_on_failure='TERMINATE_JOB_FLOW', step_args=None): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
58 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
59 A elastic mapreduce step that executes a jar | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
60 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
61 :type name: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
62 :param name: The name of the step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
63 :type jar: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
64 :param jar: S3 URI to the Jar file | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
65 :type main_class: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
66 :param main_class: The class to execute in the jar | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
67 :type action_on_failure: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
68 :param action_on_failure: An action, defined in the EMR docs to | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
69 take on failure. | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
70 :type step_args: list(str) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
71 :param step_args: A list of arguments to pass to the step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
72 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
73 self.name = name | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
74 self._jar = jar | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
75 self._main_class = main_class | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
76 self.action_on_failure = action_on_failure | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
77 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
78 if isinstance(step_args, six.string_types): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
79 step_args = [step_args] | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
80 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
81 self.step_args = step_args | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
82 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
83 def jar(self): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
84 return self._jar | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
85 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
86 def args(self): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
87 args = [] | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
88 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
89 if self.step_args: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
90 args.extend(self.step_args) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
91 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
92 return args | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
93 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
94 def main_class(self): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
95 return self._main_class | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
96 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
97 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
98 class StreamingStep(Step): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
99 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
100 Hadoop streaming step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
101 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
102 def __init__(self, name, mapper, reducer=None, combiner=None, | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
103 action_on_failure='TERMINATE_JOB_FLOW', | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
104 cache_files=None, cache_archives=None, | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
105 step_args=None, input=None, output=None, | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
106 jar='/home/hadoop/contrib/streaming/hadoop-streaming.jar'): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
107 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
108 A hadoop streaming elastic mapreduce step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
109 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
110 :type name: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
111 :param name: The name of the step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
112 :type mapper: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
113 :param mapper: The mapper URI | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
114 :type reducer: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
115 :param reducer: The reducer URI | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
116 :type combiner: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
117 :param combiner: The combiner URI. Only works for Hadoop 0.20 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
118 and later! | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
119 :type action_on_failure: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
120 :param action_on_failure: An action, defined in the EMR docs to | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
121 take on failure. | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
122 :type cache_files: list(str) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
123 :param cache_files: A list of cache files to be bundled with the job | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
124 :type cache_archives: list(str) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
125 :param cache_archives: A list of jar archives to be bundled with | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
126 the job | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
127 :type step_args: list(str) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
128 :param step_args: A list of arguments to pass to the step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
129 :type input: str or a list of str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
130 :param input: The input uri | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
131 :type output: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
132 :param output: The output uri | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
133 :type jar: str | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
134 :param jar: The hadoop streaming jar. This can be either a local | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
135 path on the master node, or an s3:// URI. | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
136 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
137 self.name = name | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
138 self.mapper = mapper | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
139 self.reducer = reducer | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
140 self.combiner = combiner | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
141 self.action_on_failure = action_on_failure | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
142 self.cache_files = cache_files | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
143 self.cache_archives = cache_archives | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
144 self.input = input | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
145 self.output = output | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
146 self._jar = jar | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
147 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
148 if isinstance(step_args, six.string_types): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
149 step_args = [step_args] | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
150 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
151 self.step_args = step_args | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
152 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
153 def jar(self): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
154 return self._jar | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
155 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
156 def main_class(self): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
157 return None | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
158 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
159 def args(self): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
160 args = [] | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
161 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
162 # put extra args BEFORE -mapper and -reducer so that e.g. -libjar | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
163 # will work | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
164 if self.step_args: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
165 args.extend(self.step_args) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
166 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
167 args.extend(['-mapper', self.mapper]) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
168 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
169 if self.combiner: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
170 args.extend(['-combiner', self.combiner]) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
171 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
172 if self.reducer: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
173 args.extend(['-reducer', self.reducer]) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
174 else: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
175 args.extend(['-jobconf', 'mapred.reduce.tasks=0']) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
176 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
177 if self.input: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
178 if isinstance(self.input, list): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
179 for input in self.input: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
180 args.extend(('-input', input)) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
181 else: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
182 args.extend(('-input', self.input)) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
183 if self.output: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
184 args.extend(('-output', self.output)) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
185 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
186 if self.cache_files: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
187 for cache_file in self.cache_files: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
188 args.extend(('-cacheFile', cache_file)) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
189 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
190 if self.cache_archives: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
191 for cache_archive in self.cache_archives: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
192 args.extend(('-cacheArchive', cache_archive)) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
193 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
194 return args | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
195 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
196 def __repr__(self): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
197 return '%s.%s(name=%r, mapper=%r, reducer=%r, action_on_failure=%r, cache_files=%r, cache_archives=%r, step_args=%r, input=%r, output=%r, jar=%r)' % ( | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
198 self.__class__.__module__, self.__class__.__name__, | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
199 self.name, self.mapper, self.reducer, self.action_on_failure, | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
200 self.cache_files, self.cache_archives, self.step_args, | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
201 self.input, self.output, self._jar) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
202 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
203 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
204 class ScriptRunnerStep(JarStep): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
205 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
206 ScriptRunnerJar = 's3n://us-east-1.elasticmapreduce/libs/script-runner/script-runner.jar' | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
207 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
208 def __init__(self, name, **kw): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
209 super(ScriptRunnerStep, self).__init__(name, self.ScriptRunnerJar, **kw) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
210 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
211 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
212 class PigBase(ScriptRunnerStep): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
213 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
214 BaseArgs = ['s3n://us-east-1.elasticmapreduce/libs/pig/pig-script', | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
215 '--base-path', 's3n://us-east-1.elasticmapreduce/libs/pig/'] | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
216 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
217 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
218 class InstallPigStep(PigBase): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
219 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
220 Install pig on emr step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
221 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
222 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
223 InstallPigName = 'Install Pig' | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
224 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
225 def __init__(self, pig_versions='latest'): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
226 step_args = [] | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
227 step_args.extend(self.BaseArgs) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
228 step_args.extend(['--install-pig']) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
229 step_args.extend(['--pig-versions', pig_versions]) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
230 super(InstallPigStep, self).__init__(self.InstallPigName, step_args=step_args) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
231 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
232 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
233 class PigStep(PigBase): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
234 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
235 Pig script step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
236 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
237 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
238 def __init__(self, name, pig_file, pig_versions='latest', pig_args=[]): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
239 step_args = [] | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
240 step_args.extend(self.BaseArgs) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
241 step_args.extend(['--pig-versions', pig_versions]) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
242 step_args.extend(['--run-pig-script', '--args', '-f', pig_file]) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
243 step_args.extend(pig_args) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
244 super(PigStep, self).__init__(name, step_args=step_args) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
245 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
246 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
247 class HiveBase(ScriptRunnerStep): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
248 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
249 BaseArgs = ['s3n://us-east-1.elasticmapreduce/libs/hive/hive-script', | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
250 '--base-path', 's3n://us-east-1.elasticmapreduce/libs/hive/'] | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
251 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
252 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
253 class InstallHiveStep(HiveBase): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
254 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
255 Install Hive on EMR step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
256 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
257 InstallHiveName = 'Install Hive' | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
258 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
259 def __init__(self, hive_versions='latest', hive_site=None): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
260 step_args = [] | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
261 step_args.extend(self.BaseArgs) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
262 step_args.extend(['--install-hive']) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
263 step_args.extend(['--hive-versions', hive_versions]) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
264 if hive_site is not None: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
265 step_args.extend(['--hive-site=%s' % hive_site]) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
266 super(InstallHiveStep, self).__init__(self.InstallHiveName, | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
267 step_args=step_args) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
268 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
269 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
270 class HiveStep(HiveBase): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
271 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
272 Hive script step | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
273 """ | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
274 | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
275 def __init__(self, name, hive_file, hive_versions='latest', | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
276 hive_args=None): | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
277 step_args = [] | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
278 step_args.extend(self.BaseArgs) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
279 step_args.extend(['--hive-versions', hive_versions]) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
280 step_args.extend(['--run-hive-script', '--args', '-f', hive_file]) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
281 if hive_args is not None: | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
282 step_args.extend(hive_args) | 
| 
 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 
guerler 
parents:  
diff
changeset
 | 
283 super(HiveStep, self).__init__(name, step_args=step_args) | 
