Using multiple '*' patterns when loading into BigQuery won't work

Question

We're trying to use a glob pattern when loading into BigQuery, for example:

gs://<bucket_name>/Network*Impressions_12345_20150201*

We have both "..NetworkImpressions_.." and "..NetworkBackfillImpressions_.." in our bucket, so we use the first '*' to scoop up both types of files. But BQ borks with:

"Not found: URI gs://backup-gdfp-7415/Network*Impressions_232503_20150101_20*"

The files definitely exist. If we remove the first '*' it works fine (and when we explicitly specify both types).

Here's a job id for a failed load job where we are trying to use the pattern: job_LXNGEAeWsaU9HyFgcCCJMHu8YtY

I would have thought this should be possible with BigQuery?

Answer 1

From the documentation for load job configuration sourceUris parameter:

[Required] The fully-qualified URIs that point to your data in Google Cloud Storage. Wildcard names are only supported when they appear at the end of the URI.