Hi Team,
I'm using Databricks Labs LakeBridge and noticed an inconsistency between the analyze and transpile commands with respect to SSIS support.
Analyzer:
When I run the analyze command, Iām able to select SSIS as the source technology, and the tool works as expected:
databricks labs lakebridge analyze --source-directory "C:\path\to\ssis" --source-tech SSIS --report-file "C:\output\analysis_report.txt"
Transpiler:
However, when I try to transpile using:
databricks labs lakebridge transpile --input-source "C:\path\to\ssis" --source-dialect ssis --output-folder "C:\output"
I receive the following error:
ERROR [src/databricks/labs/lakebridge.transpile] ValueError: Invalid value for '--source-dialect': 'ssis' must be one of: datastage, informatica (desktop edition), informatica cloud, mssql, netezza, oracle, synapse, teradata
Additionally, when I run databricks labs lakebridge install-transpile, SSIS is not listed as a valid option for source dialect.
Attempted workaround:
I wanted to add SSIS, I tried using synapse as a source dialect in the transpiler command.
databricks labs lakebridge transpile --input-source "C:\path\to\ssis" --source-dialect synapse --output-folder "C:\output"
This command does not execute but generates .sql files. THIS SQL-based files are then passed into the next stage of the transpiler pipeline (e.g., notebook generation), which results in the following error:
JsonRpcInternalError: UnicodeDecodeError: 'utf-8' codec can't decode byte 0xa9 in position 14009: invalid start byte
ā however, the content of those files is actually HTML, not valid SQL. I manually verified the .sql files and confirmed they contain HTML structure (like a rendered page dump), likely because of parsing mismatch or unsupported handling of SSIS content using the Synapse dialect.
Questions:
Is SSIS currently unsupported as a valid --source-dialect for the transpile command, even though it's available in analyze?
Is there a recommended workaround for transpiling SQL logic extracted from SSIS packages?
Is the generation of HTML content in .sql files expected when using a mismatched dialect like Synapse?
Thanks in advance for your help!