[hailctl] batch submit fixes

build.yaml

@@ -1027,6 +1027,7 @@ steps:
       export HAIL_DOCTEST_DATA_DIR=$(realpath ./data)
       export HAIL_TEST_STORAGE_URI={{ global.test_storage_uri }}/{{ token }}
       export PYSPARK_SUBMIT_ARGS="--driver-memory 6g pyspark-shell"
+
       python3 -m pytest \
               -Werror:::hail -Werror:::hailtop -Werror::ResourceWarning \
               --log-cli-level=INFO \
@@ -1037,6 +1038,7 @@ steps:
               --durations=50 \
               --ignore=test/hailtop/batch/ \
               --ignore=test/hailtop/inter_cloud \
+              --ignore=test/hailtop/hailctl/batch \
               --timeout=120 \
               test
     inputs:
@@ -2995,7 +2997,8 @@ steps:
               --instafail \
               --durations=50 \
               --timeout=360 \
-              /io/test/hailtop/batch/
+              /io/test/hailtop/batch/ /io/test/hailtop/hailctl/batch
+
     inputs:
       - from: /repo/hail/python/pytest.ini
         to: /io/pytest.ini
@@ -3060,9 +3063,7 @@ steps:
       BATCH_ID=$(hailctl batch submit simple_hail.py --name=test-hailctl-batch-submit --files=foo -o json | jq '.id')
       STATUS=$(hailctl batch wait -o json $BATCH_ID)
       STATE=$(echo $STATUS | jq -jr '.state')
-      if [ "$STATE" == "success" ]; then
-          exit 0;
-      else
+      if [ "$STATE" != "success" ]; then
           echo $STATUS;
           exit 1;
       fi
@@ -3081,21 +3082,19 @@ steps:
       BATCH_ID=$(hailctl batch submit --name=test-hailctl-batch-submit --files=foo -o json hail_with_args.py 100 | jq '.id')
       STATUS=$(hailctl batch wait -o json $BATCH_ID)
       STATE=$(echo $STATUS | jq -jr '.state')
-      if [ "$STATE" == "success" ]; then
-          exit 0;
-      else
+      if [ "$STATE" != "success" ]; then
           echo $STATUS;
           exit 1;
       fi
 
-      cat >file.sh <<EOF
+      cat > file.sh << 'EOF'
       set -ex
 
-      cat foo
+      cat foo/baz.txt
       echo "Hello World!"
       EOF
 
-      BATCH_ID=$(hailctl batch submit --name=test-hailctl-batch-submit --files=foo -o json file.sh | jq '.id')
+      BATCH_ID=$(hailctl batch submit --name=test-hailctl-batch-submit --files=foo -o json --image-name ubuntu:22.04 file.sh | jq '.id')
       STATUS=$(hailctl batch wait -o json $BATCH_ID)
       STATE=$(echo $STATUS | jq -jr '.state')
       if [ "$STATE" == "success" ]; then
@@ -3114,12 +3113,10 @@ steps:
       echo "Hello World! $1 $2"
       EOF
 
-      BATCH_ID=$(hailctl batch submit --name=test-hailctl-batch-submit --files=foo -o json file-with-args.sh abc 123 | jq '.id')
+      BATCH_ID=$(hailctl batch submit --name=test-hailctl-batch-submit --files=foo -o json --image-name ubuntu:22.04 file-with-args.sh abc 123 | jq '.id')
       STATUS=$(hailctl batch wait -o json $BATCH_ID)
       STATE=$(echo $STATUS | jq -jr '.state')
-      if [ "$STATE" == "success" ]; then
-          exit 0;
-      else
+      if [ "$STATE" != "success" ]; then
           echo $STATUS;
           exit 1;
       fi
@@ -3978,8 +3975,6 @@ steps:
     dependsOn:
       - ci_utils_image
       - default_ns
-    scopes:
-      - deploy
   - kind: runImage
     name: test_gcp_ar_cleanup_policies
     resources:

hail/python/hailtop/hailctl/batch/cli.py

@@ -1,15 +1,14 @@
 import asyncio
-import json
 from enum import Enum
 from typing import Annotated as Ann
 from typing import Any, Dict, List, Optional, cast
 
+import orjson
 import typer
 from typer import Argument as Arg
 from typer import Option as Opt
 
 from . import billing, list_batches
-from . import submit as _submit
 from .batch_cli_utils import (
     ExtendedOutputFormat,
     ExtendedOutputFormatOption,
@@ -131,7 +130,7 @@ def wait(
         quiet = quiet or output != StructuredFormatPlusText.TEXT
         out = batch.wait(disable_progress_bar=quiet)
         if output == StructuredFormatPlusText.JSON:
-            print(json.dumps(out))
+            print(orjson.dumps(out).decode('utf-8'))
         else:
             print(out)
 
@@ -168,16 +167,40 @@ def submit(
     name: Ann[str, Opt(help='The name of the batch.')] = '',
     image_name: Ann[Optional[str], Opt(help='Name of Docker image for the job (default: hailgenetics/hail)')] = None,
     output: StructuredFormatPlusTextOption = StructuredFormatPlusText.TEXT,
+    wait: Ann[bool, Opt(help='Wait for the batch to complete.')] = False,
 ):
     """Submit a batch with a single job that runs SCRIPT with the arguments ARGUMENTS.
 
     If you wish to pass option-like arguments you should use "--". For example:
+    $ hailctl batch submit --image-name docker.io/image my_script.py -- some-argument --animal dog
 
+    Copy a local file into the working directory of the job:
+    $ hailctl batch submit --image-name docker.io/image my_script.py --files a-file -- some-argument --animal dog
 
+    Copy a local file into a particular directory in the job:
+    $ hailctl batch submit --image-name docker.io/image my_script.py --files a-file:/foo/bar/ -- some-argument --animal dog
 
-    $ hailctl batch submit --image-name docker.io/image my_script.py -- some-argument --animal dog
+    Copy a local directory to the directory /foo/bar/a-directory in the job:
+    $ hailctl batch submit --image-name docker.io/image my_script.py --files a-directory:/foo/bar/ -- some-argument --animal dog
+
+    Copy a local file or a directory to a specific location in the job:
+    $ hailctl batch submit --image-name docker.io/image my_script.py --files a/local/path:/foo/bar -- some-argument --animal dog
+
+    Copy a local directory to a specific location in the job:
+    $ hailctl batch submit --image-name docker.io/image my_script.py --files a-file:/foo/bar -- some-argument --animal dog
+
+    Notes
+    -----
+    Copying a local directory to the root directory in the job is not supported (example: ``--files my-local-dir/:/``).
     """
-    asyncio.run(_submit.submit(name, image_name, files or [], output, script, [*(arguments or []), *ctx.args]))
+    from .submit import HailctlBatchSubmitError  # pylint: disable=import-outside-toplevel
+    from .submit import submit as _submit  # pylint: disable=import-outside-toplevel
+
+    try:
+        asyncio.run(_submit(name, image_name, files or [], output, script, [*(arguments or []), *ctx.args], wait))
+    except HailctlBatchSubmitError as err:
+        print(err.message)
+        raise typer.Exit(err.exit_code)
 
 
 @app.command('init', help='Initialize a Hail Batch environment.')