Combining library sets
You can install Python packages via pip or conda, and download a file as part of the library set by combining these options when you create the library set.
You can use the combination option as shown in the following example:
{
"application_details": {
"application": "/opt/ibm/customization-scripts/customize_instance_app.py",
"arguments": ["{\"library_set\":{\"action\":\"add\",\"name\":\"multi_library\",\"libraries\":{\"pip\":{\"python\":{\"packages\":[\"numpy\"]}}},\"script\":{\"source\":\"py_files\",\"params\":[\"script_params\"]}}}"],
"py-files": "cos://<bucket>.dev-cos/jar/customization_script.py",
"conf": {
"spark.hadoop.fs.cos.dev-cos.endpoint": "https://s3.direct.us-south.cloud-object-storage.appdomain.cloud",
"spark.hadoop.fs.cos.dev-cos.access.key": "<cos_access_key>",
"spark.hadoop.fs.cos.dev-cos.secret.key": "<cos_secret_key>"
}
}
}
To increase readability, the unescaped JSON for the arguments would look as follows:
{
"library_set": {
"action": "add",
"name": "multi_library",
"libraries": {
"pip": {
"python": {
"packages": ["numpy"]
}
}
},
"script": {
"source": "py_files",
"params": ["script_params"]
}
}
}