Pip packages are not installed for Iceberg source plugin with Hive type

[usmanovbf]

Describe the bug
Pip packages are not installed for Iceberg source plugin with hive type

To Reproduce
Steps to reproduce the behavior:

1. Create Iceberg ingestion source by the example below but without type: hive
2. Click Save & Run
3. Get an error about required type field
4. Update the recipe with type: hive
5. Click Save & Run button
6. See the errors (logs are below):
   1. ModuleNotFoundError: No module named 'thrift'
   2. pyiceberg.exceptions.NotInstalledError: Apache Hive support not installed: pip install 'pyiceberg[hive]'

Expected behavior
All pypi packages 'pyiceberg[hive]' thrift should be installed properly

Solution

Execute pip install every time before execution of recipe

Screenshots

Desktop (please complete the following information):

• OS: MacOS Sonoma arm64
• Browser Chrome
• Version 122.0.6261.112

Additional context

1. Recipe:

source:
    type: iceberg
    config:
        env: PROD
        catalog:
            name: iceberg-catalog
            type: hive
            config:
                uri: 'https://hostname1:9083'
                s3.endpoint: 'https://hostname2'
                s3.access-key-id: '${secret1}'
                s3.secret-access-key: '${secret2}'
        table_pattern:
            allow:
                - 'test.*'
        profiling:
            enabled: false

3. Error logs: exec-urn_li_dataHubExecutionRequest_1d2b870e-81e9-477a-8869-39505a9f2b3d.log
4. Even adding Extra Pip Libraries does not help

9. Datahub version 0.12.1 10. As I see, it is not fixed in 0.13.0 from 0.12.1 https://github.com/datahub-project/datahub/commits/v0.12.1/metadata-ingestion/src/datahub/ingestion/source/iceberg

[hsheth2]

@usmanovbf would you be open to sending a PR for this?

[usmanovbf]

@hsheth2 sorry, I have no time for now. Hope you or your teammate will find some time to fix it

[igorvoltaic]

might be related to #10289