Whitespace in source column name rejected by copy activity

[Anonymous]

I am using a pipeline with a copy activity to get data from a SQL server. Since I loop through a list of tables, I get the mapping from a meta table that I prepare beforehand. However, if the source column name contains a space, I get the following error:

Failure happened on 'destination' side. ErrorCode=DeltaInvalidCharacterInColumnName,'Type=Microsoft.DataTransfer.Common.Shared.HybridDeliveryException,Message=Column name Location Code contains invalid characters. ",;{}()\n\t=" are not supported.,Source=Microsoft.DataTransfer.ClientLibrary,'

However, the failure is certainly NOT happening on the 'destination' side. The schema looks like this (it works neither with the encompassing backticks nor without them):

{
	"type": "TabularTranslator",
	"mappings": [
		{
			"source": { "name": "`timestamp`" },
			"sink": { "name": "timestamp" }
		},
		{
			"source": { "name": "`Location Code`" },
			"sink": { "name": "Location_Code" }
		},
	]
}

Since the space is in the name of the source column, I can't change it. So what do I do?

[priyankabis]

Isn't there a way or setting like we do in notebooks so that it can read columns with Spaces and save with _ instead od space

[tarektm7]

I was facing the same issue in a similar situation. 

The only workaround I could do is to use a notebook and get the data in a dataframe then write it into a delta table as below:

df.write.option('delta.columnMapping.mode', 'name').save(path='Tables/schemaname/tablename', format= 'delta', mode='overwrite')

[chrichards]

Hi - were you able to resolve this?  I'm encountering the same issue...

[Anonymous]

Hi @Anonymous 

You can try using square brackets as the source column name instead of backquotes. For example:

{
    "type": "TabularTranslator",
    "mappings": [
        {
            "source": { "name": "[timestamp]" },
            "sink": { "name": "timestamp" }
        },
        {
            "source": { "name": "[Location Code]" },
            "sink": { "name": "Location_Code" }
        }
    ]
}

Regards,

Nono Chen

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

[Anonymous]

Nice try, but it doesn't work, yielding the same error message.

Just in case, the input for the copy activity really looks like this:

{
    "source": {
        "type": "SqlServerSource",
        "queryTimeout": "02:00:00",
        "partitionOption": "None",
        "datasetSettings": {
            "annotations": [],
            "type": "SqlServerTable",
            "schema": [],
            "typeProperties": {
                "schema": "dbo",
                "table": "ABCDEFG",
                "database": "HIJKLMNOP"
            },
            "externalReferences": {
                "connection": "684354684651687654657684"
            }
        }
    },
    "sink": {
        "type": "LakehouseTableSink",
        "tableActionOption": "OverwriteSchema",
        "partitionOption": "None",
        "datasetSettings": {
            "annotations": [],
            "linkedService": {
                "name": "Sandbox",
                "properties": {
                    "annotations": [],
                    "type": "Lakehouse",
                    "typeProperties": {
                        "workspaceId": "XXX",
                        "artifactId": "XXX",
                        "rootFolder": "Tables"
                    }
                }
            },
            "type": "LakehouseTable",
            "schema": [],
            "typeProperties": {
                "table": "XXX"
            }
        }
    },
    "enableStaging": false,
    "translator": "{\n\t\"type\": \"TabularTranslator\",\n\t\"mappings\": [\n\t\t{\n\t\t\t\"source\": { \"name\": \"[timestamp]\" },\n\t\t\t\"sink\": { \"name\": \"timestamp\" }\n\t\t},\n\t\t{\n\t\t\t\"source\": { \"name\": \"[Location Code]\" },\n\t\t\t\"sink\": { \"name\": \"Location_Code\" },  *[here I removed the other columns]* 
 \n\t]\n}"
}

[Anonymous · Accepted solution]

Alright, I got it and it was trivial in hindsight: The translator above is a literal string. In order to pass it correctly as a parameter, you have to put "@json(mapping_string)" instead of what I did, which was simply "@mapping_string".