Parquet file with Snappy compression on ADSL Gen 2

[jnh3_powerbi]

We have files in our Azure Data Lake Storage Gen 2 storage account that are parquet files with Snappy compression (very common with Apache Spark). I don't see any menu option for reading those, so after searching around I tried the following with Power Query M:

let
    Source = AzureStorage.DataLake("https://oursa.dfs.core.windows.net/ourfiledir/"),
    ContentColumnTable = Table.SelectColumns(Source,{"Content"}),
    DecompressedContentTable = Table.TransformColumns(ContentColumnTable,{"Content",(x) => Binary.Decompress(x,Compression.Snappy)})
in
    DecompressedContentTable

While the first 2 statements seem to work OK, when I add the decompression line I get the following error:

Expression.Error: Invalid compression type.
Details:
2

However, I got that compression type directly from the documentation: Compression.Snappy - PowerQuery M | Microsoft Docs

Is there something wrong with the code, or does Power BI not implement all of the PowerQuery functionality? 

[Syndicate_Admin · Accepted solution]

Hi @jnh3_powerbi 

Currently Binary.Decompress only supports Compression.GZip and Compression.Deflate types. It doesn't support Compression.Snappy. See Binary.Decompress - PowerQuery M | Microsoft Docs

From Parquet format - Azure Data Factory & Azure Synapse, I see that gzip type is supported. Is it possible to compress the files to gzip in ADLS Gen 2?

Additionally, Parquet connector is currently available in Power Query Desktop. Maybe you can try this connector. 

Best Regards,
Community Support Team _ Jing

[v-jingzhang]

Hi @jnh3_powerbi 

Currently Binary.Decompress only supports Compression.GZip and Compression.Deflate types. It doesn't support Compression.Snappy. See Binary.Decompress - PowerQuery M

From Parquet format - Azure Data Factory & Azure Synapse, I see that gzip type is supported. Is it possible to compress the files to gzip in ADLS Gen 2?

Additionally, Parquet connector is currently available in Power Query Desktop. Maybe you can try this connector. 

This article may be helpful: Power BI reading Parquet from a Data Lake - Simple Talk (red-gate.com)

Best Regards,
Community Support Team _ Jing

If this post helps, please Accept it as Solution to help other members find it.