如何向Hive/Beeline传递多个HQL文件并在单会话中执行？

Great question! When you run hive -f with multiple files, Hive indeed only processes the first one because the -f flag expects a single script file. Here are a few reliable ways to execute multiple HQL files in the same Hive session:

Method 1: Use Shell Concatenation with Pipe #

You can concatenate the content of all your HQL files and pipe it directly to the Hive CLI. This ensures all commands run in the same session:

cat file_1.hql file_2.hql | hive

This works because the Hive CLI reads commands from standard input when no -f or -e flag is specified. Just make sure each HQL statement in your files ends with a semicolon (;) so Hive knows where one command ends and the next begins.

Method 2: Create a Master HQL Script with source Commands #

Another clean approach is to create a master script that references your individual HQL files using Hive's built-in source command. This lets you control the execution order and even add extra logic if needed.

1. Create a file named master_script.hql with the following content:

-- Execute file_1 first
source /path/to/file_1.hql;

-- Then execute file_2
source /path/to/file_2.hql;

(Replace /path/to/ with the actual path to your HQL files, or use relative paths if they're in the same directory.)

2. Run the master script with:

hive -f master_script.hql

The source command tells Hive to load and execute the content of the specified file within the current session, so all tables, variables, or session configurations from file_1.hql will persist when running file_2.hql.

Method 3: Use hive -e with Concatenated Content #

If you prefer a one-liner without creating a master file, you can pass the concatenated content directly via the -e flag (which accepts a string of HQL commands):

hive -e "$(cat file_1.hql file_2.hql)"

Note that this might not be ideal for very large HQL files, as it can hit shell command-length limits, but it works well for smaller scripts.

内容的提问来源于stack exchange，提问作者Sano babu