add pipeline step to push to git (#387)

* add push to git

* small fixes

* work around .after()

* format

code_search/docker/ks/launch_search_index_creator_job.sh

@@ -16,12 +16,18 @@ ksEnvName="pipeline"
 component="search-index-creator"
 
 usage() {
-	echo "Usage: launch_search_index_creator_job.sh --workingDir=<working dir> --workflowId=<workflow id invoking the container>
-	--dataDir=<data dir> --timeout=<timeout>  --namespace=<kubernetes namespace>  --cluster=<cluster to deploy job to> "
+	echo "Usage: launch_search_index_creator_job.sh
+	--workflowId=<workflow id invoking the container>
+	--indexFile=<index file>
+	--lookupFile=<lookup file>
+	--dataDir=<data dir>
+	--timeout=<timeout>
+	--namespace=<kubernetes namespace>
+	--cluster=<cluster to deploy job to>"
 }
 
 # List of required parameters
-names=(workingDir workflowId dataDir namespace cluster)
+names=(workflowId indexFile lookupFile dataDir namespace cluster)
 
 source "${DIR}/parse_arguments.sh"
 source "${DIR}/initialize_kubectl.sh"
@@ -29,8 +35,8 @@ source "${DIR}/initialize_kubectl.sh"
 # Apply parameters
 ks param set ${component} dataDir ${dataDir} --env ${ksEnvName}
 ks param set ${component} jobNameSuffix ${workflowId} --env ${ksEnvName}
-ks param set ${component} lookupFile ${workingDir}/code-embeddings-index/embedding-to-info.csv --env ${ksEnvName}
-ks param set ${component} indexFile ${workingDir}/code-embeddings-index/embeddings.index --env ${ksEnvName}
+ks param set ${component} lookupFile ${lookupFile} --env ${ksEnvName}
+ks param set ${component} indexFile ${indexFile} --env ${ksEnvName}
 
 ks show ${ksEnvName} -c "${component}"
 ks apply ${ksEnvName} -c "${component}"

code_search/docker/ks/submit_code_embeddings_job.sh

@@ -20,9 +20,16 @@ numWorkers=5
 workerMachineType=n1-highcpu-32
 
 usage() {
-	echo "Usage: submit_code_embeddings_job.sh --workflowId=<workflow id invoking the container> --modelDir=<directory contains the model>
-	--dataDir=<data dir> --numWorkers=<num of workers> --project=<project> --targetDataset=<target BQ dataset>
-	--workerMachineType=<worker machine type> --workingDir=<working dir> --cluster=<cluster to deploy job to>"
+	echo "Usage: submit_code_embeddings_job.sh
+	--workflowId=<workflow id invoking the container>
+	--modelDir=<directory contains the model>
+	--dataDir=<data dir>
+	--numWorkers=<num of workers>
+	--project=<project>
+	--targetDataset=<target BQ dataset>
+	--workerMachineType=<worker machine type>
+	--workingDir=<working dir>
+	--cluster=<cluster to deploy job to>"
 }
 
 # List of required parameters

code_search/docker/ks/update_index.sh

@@ -10,13 +10,20 @@ DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" > /dev/null && pwd)"
 branch=master
 
 usage() {
-	echo "Usage: update_index.sh --branch=<base branch> --appDir=<ksonnet app dir>
-	--gitRepo=<github repo with Argo CD hooked up> --env=<ksonnet environment> --indexFile=<index file>
-	--lookupFile=<lookup file> --workflowId=<workflow id invoking the container>"
+	echo "Usage: update_index.sh
+	--baseGitRepo=<base git repo name>
+	--baseBranch=<base branch>
+	--appDir=<ksonnet app dir>
+	--forkGitRepo=<github repo with Argo CD hooked up>
+	--env=<ksonnet environment>
+	--indexFile=<index file>
+	--lookupFile=<lookup file>
+	--workflowId=<workflow id invoking the container>
+	--botEmail=<email account of the bot that send the PR>"
 }
 
 # List of required parameters
-names=(appDir gitRepo env lookupFile indexFile workflowId)
+names=(baseGitRepo baseBranch appDir forkGitRepo env indexFile lookupFile workflowId botEmail)
 
 source "${DIR}/parse_arguments.sh"
 
@@ -29,9 +36,13 @@ if [ -z ${dryrun} ]; then
 fi
 
 
-git config --global user.email pipeline@localhost
-git clone -b ${branch} https://${GITHUB_TOKEN}@github.com/${gitRepo}.git repo && cd repo/${appDir}
+git config --global user.email ${botEmail}
+git clone https://${GITHUB_TOKEN}@github.com/${forkGitRepo}.git repo && cd repo/${appDir}
 git config credential.helper store
+git remote add upstream https://github.com/${baseGitRepo}.git
+git fetch upstream
+git merge upstream/${baseBranch} master
+
 git checkout -b ${workflowId}
 ks param set --env=${env} search-index-server indexFile ${indexFile}
 ks param set --env=${env} search-index-server lookupFile ${lookupFile}
@@ -51,7 +62,7 @@ EOF
 # Create a pull request
 if (! ${dryrun}); then
     git push origin ${workflowId}
-    hub pull-request --base=${gitRepo}:${branch} -F ${FILE}
+    hub pull-request --base=${baseGitRepo}:${baseBranch} -F ${FILE}
 else
 	echo "dry run; not committing to git."
 fi

code_search/pipeline/index_update_pipeline.py

@@ -64,7 +64,7 @@ def dataflow_function_embedding_op(
         num_workers: int, working_dir: str, step_name='dataflow_function_embedding'):
   return default_gcp_op(
     name=step_name,
-    image='gcr.io/kubeflow-examples/code-search-ks:v20181127-08f8c05-dirty-19ca4c',
+    image='gcr.io/kubeflow-examples/code-search/ks:v20181130-b807843',
     command=['/usr/local/src/submit_code_embeddings_job.sh'],
     arguments=[
       "--workflowId=%s" % workflow_id,
@@ -81,14 +81,16 @@ def dataflow_function_embedding_op(
 
 
 def search_index_creator_op(
-        working_dir: str, data_dir: str, workflow_id: str, cluster_name: str, namespace: str):
+        index_file: str, lookup_file: str, data_dir: str,
+        workflow_id: str, cluster_name: str, namespace: str):
   return dsl.ContainerOp(
     # use component name as step name
     name='search_index_creator',
-    image='gcr.io/kubeflow-examples/code-search-ks:v20181127-08f8c05-dirty-19ca4c',
+    image='gcr.io/kubeflow-examples/code-search/ks:v20181130-b807843',
     command=['/usr/local/src/launch_search_index_creator_job.sh'],
     arguments=[
-      '--workingDir=%s' % working_dir,
+      '--indexFile=%s' % index_file,
+      '--lookupFile=%s' % lookup_file,
       '--dataDir=%s' % data_dir,
       '--workflowId=%s' % workflow_id,
       '--cluster=%s' % cluster_name,
@@ -97,29 +99,85 @@ def search_index_creator_op(
   )
 
 
+def update_index_op(
+        base_git_repo: str, base_branch: str, app_dir: str, fork_git_repo: str,
+        index_file: str, lookup_file: str, workflow_id: str, bot_email: str):
+  return (
+    dsl.ContainerOp(
+      name='update_index',
+      image='gcr.io/kubeflow-examples/code-search/ks:v20181130-b807843',
+      command=['/usr/local/src/update_index.sh'],
+      arguments=[
+        '--baseGitRepo=%s' % base_git_repo,
+        '--baseBranch=%s' % base_branch,
+        '--appDir=%s' % app_dir,
+        '--forkGitRepo=%s' % fork_git_repo,
+        '--env=%s' % 'pipeline',
+        '--indexFile=%s' % index_file,
+        '--lookupFile=%s' % lookup_file,
+        '--workflowId=%s' % workflow_id,
+        '--botEmail=%s' % bot_email,
+      ],
+    )
+    .add_volume(
+      k8s_client.V1Volume(
+        name='github-access-token',
+        secret=k8s_client.V1SecretVolumeSource(
+          secret_name='github-access-token'
+        )
+      )
+    )
+    .add_env_variable(
+      k8s_client.V1EnvVar(
+        name='GITHUB_TOKEN',
+        value_from=k8s_client.V1EnvVarSource(
+          secret_key_ref=k8s_client.V1SecretKeySelector(
+            name='github-access-token',
+            key='token',
+          )
+        )
+      )
+    )
+  )
+
+
 # The pipeline definition
 @dsl.pipeline(
   name='function_embedding',
   description='Example function embedding pipeline'
 )
 def function_embedding_update(
-    project,
-    working_dir,
-    saved_model_dir,
-    cluster_name,
-    namespace,
-    target_dataset=dsl.PipelineParam(name='target-dataset', value='code_search'),
-    worker_machine_type=dsl.PipelineParam(name='worker-machine-type', value='n1-highcpu-32'),
-    num_workers=dsl.PipelineParam(name='num-workers', value=5)):
+    project='code-search-demo',
+    cluster_name='cs-demo-1103',
+    namespace='kubeflow',
+    working_dir='gs://code-search-demo/pipeline',
+    data_dir='gs://code-search-demo/20181104/data',
+    saved_model_dir='gs://code-search-demo/models/20181107-dist-sync-gpu/export/1541712907/',
+    target_dataset='code_search',
+    worker_machine_type='n1-highcpu-32',
+    function_embedding_num_workers=5,
+    base_git_repo='kubeflow/examples',
+    base_branch='master',
+    app_dir='code_search/ks-web-app',
+    fork_git_repo='IronPan/examples',
+    bot_email='kf.sample.bot@gmail.com'):
   workflow_name = '{{workflow.name}}'
   working_dir = '%s/%s' % (working_dir, workflow_name)
-  data_dir = '%s/data' % working_dir
+  lookup_file = '%s/code-embeddings-index/embedding-to-info.csv' % working_dir
+  index_file = '%s/code-embeddings-index/embeddings.index'% working_dir
+
   function_embedding = dataflow_function_embedding_op(
                             project, cluster_name, target_dataset, data_dir,
-                            saved_model_dir,
-                            workflow_name, worker_machine_type, num_workers, working_dir)
-  search_index_creator_op(
-      working_dir, data_dir, workflow_name, cluster_name, namespace).after(function_embedding)
+                            saved_model_dir, workflow_name, worker_machine_type,
+                            function_embedding_num_workers, working_dir)
+
+  search_index_creator = search_index_creator_op(
+    index_file, lookup_file, data_dir, workflow_name, cluster_name, namespace)
+  search_index_creator.after(function_embedding)
+  update_index_op(
+      base_git_repo, base_branch, app_dir, fork_git_repo,
+      index_file, lookup_file, workflow_name, bot_email)\
+    .after(search_index_creator)
 
 
 if __name__ == '__main__':