[PYTHON] It may be a problem to use Japanese for folder names and notebook names in Databricks

Conclusion first

If you use ** Japanese for the folder name or Notebook name **, use dbutils.notebook.run to use another There is a case that an error occurs when calling Notebook

What do you mean

I have a notebook with the following folder structure

/Users/xxx＠yyy.jp
 |-MyNotebook
 |-My notebook
 |-MyNotebookCaller
 |-MyNotebook Caller
 |-test
   |-MyNotebook
   |-MyNotebookCaller

Of these, if the following cases apply, calling another Notebook using dbutils.notebook.run failed.

-** Caller ** If you use Japanese in the name of your Notebook -/Users/[email protected]/MyNotebook Caller -** Caller ** When Japanese is used for the name of the storage folder of Notebook -/Users/[email protected]/test/MyNotebookCaller

In the following cases, the call was successful without any problem.

--When Japanese is used for the name of the called notebook -/Users/[email protected]/My Notebook --When Japanese is used for the name of the storage folder of the called Notebook -/Users/[email protected]/test/MyNotebook

Verification

Description of each notebook

It is a simple process of calling by passing parameters from MyNotebookCaller or MyNotebook Caller to MyNotebook, and printing the received parameters in MyNotebook.

/Users/xxx＠yyy.jp/MyNotebook

dbutils.widgets.text("param1", "111")
dbutils.widgets.text("param2", "222")

print("param1:{},param2:{}".format(dbutils.widgets.get("param1"), dbutils.widgets.get("param2")))

/Users/[email protected]/My Notebook

#/Users/xxx＠yyy.jp/Same as My Notebook

/Users/xxx＠yyy.jp/MyNotebookCaller

#Cmd1 Call MyNotebook in the same folder
dbutils.notebook.run(
  "./MyNotebook",
  60,
  {
    "param1": "val1",
    "param2": "val2"
  }
)

#Cmd2 Call My Notebook in the same folder
dbutils.notebook.run(
  "./My notebook",
  60,
  {
    "param1": "val1",
    "param2": "val2"
  }
)

#Call My Notebook in the Cmd3 test folder
dbutils.notebook.run(
  "./test/MyNotebook",
  60,
  {
    "param1": "val1",
    "param2": "val2"
  }
)

/Users/[email protected]/MyNotebook Caller

#/Users/xxx＠yyy.jp/Same as MyNotebookCaller

/Users/[email protected]/Test/MyNotebook

#/Users/xxx＠yyy.jp/Same as My Notebook

/Users/[email protected]/Test/MyNotebookCaller

#Cmd1 Call MyNotebook in the same folder
dbutils.notebook.run(
  "./MyNotebook",
  60,
  {
    "param1": "val1",
    "param2": "val2"
  }
)

#Cmd2 Call MyNotebook in the folder one level above
dbutils.notebook.run(
  "../MyNotebook",
  60,
  {
    "param1": "val1",
    "param2": "val2"
  }
)

Verification (1) When Japanese is used for the name of the calling Notebook

Try calling MyNotebook in the same folder from /Users/[email protected]/MyNotebook Caller

Cmd1 Call MyNotebook in the same folder

dbutils.notebook.run(
  "./MyNotebook",
  60,
  {
    "param1": "val1",
    "param2": "val2"
  }
)

As a result, I got a WorkflowException. It seems that an error was returned by using Japanese, which is a character other than Latin characters (ASCII character set).

com.databricks.WorkflowException: com.databricks.common.client.DatabricksServiceHttpClientException: INVALID_PARAMETER_VALUE: Only Latin1 (ASCII) characters are currently supported. Any international characters must be removed or replaced in workflow_context

Even if I call MyNotebook in the test folder,

Cmd2 Call MyNotebook in the test folder

dbutils.notebook.run(
  "./test/MyNotebook",
  60,
  {
    "param1": "val1",
    "param2": "val2"
  }
)

This also caused a WorkflowException

com.databricks.WorkflowException: com.databricks.common.client.DatabricksServiceHttpClientException: INVALID_PARAMETER_VALUE: Only Latin1 (ASCII) characters are currently supported. Any international characters must be removed or replaced in workflow_context

Verification (2) When Japanese is used for the name of the storage folder of the calling Notebook

Try calling MyNotebook in the same folder from /Users/[email protected]/test/MyNotebookCaller

Cmd1 Call MyNotebook in the same folder

dbutils.notebook.run(
  "./MyNotebook",
  60,
  {
    "param1": "val1",
    "param2": "val2"
  }
)

This also resulted in a WorkflowException

com.databricks.WorkflowException: com.databricks.common.client.DatabricksServiceHttpClientException: INVALID_PARAMETER_VALUE: Only Latin1 (ASCII) characters are currently supported. Any international characters must be removed or replaced in workflow_context

Verification ③ When Japanese is used for the name of the called notebook

Try calling /Users/[email protected]/MyNotebook from /Users/[email protected]/MyNotebookCaller

Cmd2 Call My Notebook in the same folder

dbutils.notebook.run(
  "./My notebook",
  60,
  {
    "param1": "val1",
    "param2": "val2"
  }
)

The process ended normally even though Japanese was used for the called notebook name. It seems that the passed parameters are also output properly.

param1:val1,param2:val2

Verification ④ When Japanese is used for the name of the storage folder of the called notebook

Try calling /Users/[email protected]/test/MyNotebook from /Users/[email protected]/MyNotebookCaller

Call My Notebook in the Cmd3 test folder

dbutils.notebook.run(
  "./test/MyNotebook",
  60,
  {
    "param1": "val1",
    "param2": "val2"
  }
)

The process ended normally even though Japanese was used for the folder name of the called notebook. It seems that the passed parameters are also output properly.

param1:val1,param2:val2

Summary

Be careful when using Japanese for folder names and notebook names