Getting a compilation error when running flux query with backslashes

[taylanisikdemir]

I'm using influxdb go client to run flux queries. I am having trouble filtering by a tag whose value is a windows file path with backslashes.

Query:

"from(bucket: \"7446077d\")
|> range(start: 2020-07-19T01:18:45.417Z, stop: 2020-07-19T01:19:05.417Z)
|> filter(fn: (r) => r._measurement == \"agent.anomaly2\")
|> filter(fn: (r) => r._field == \"value\")
|> filter(fn: (r) => r.source == \"C:\\Program Files\\Data\\a.log\")"

Error:
invalid (): compilation failed: expected RPAREN, got EOF

If I replace all backslashes with double backslashes it works fine.

"from(bucket: \"7446077d\")
|> range(start: 2020-07-19T01:01:43.336Z, stop: 2020-07-19T01:02:03.336Z)
|> filter(fn: (r) => r._measurement == \"agent.anomaly2\")
|> filter(fn: (r) => r._field == \"value\")
|> filter(fn: (r) => r.source == \"C:\\\\Program Files\\\\Data\\\\a.log\")"

Questions:

1. Is this expected?
2. Is my workaround safe?
3. Are there other characters we should be escaping to avoid similar query compilation issues?

Note: I'm using an older version of go client (from April) but I am able to repro this issue with latest version v1.4.0

[sranka]

Yes, it is expected, a backslash is escaped in a flux string and then also in a go string. You can use golang's raw string literal to have less backslashes in a go string:

`from(bucket: "7446077d")
|> range(start: 2020-07-19T01:18:45.417Z, stop: 2020-07-19T01:19:05.417Z)
|> filter(fn: (r) => r._measurement == "agent.anomaly2")
|> filter(fn: (r) => r._field == "value")
|> filter(fn: (r) => r.source == "C:\\Program Files\\Data\\a.log")`

[taylanisikdemir]

The example queries I shared are generated with raw string literals by the way. It still requires escaping the backslashes twice. Thanks for confirming the backslashes are escaped in a flux string as well as in a go string.

[vlastahajek]

Escaping special chars will be handled by the Query parameters feature, once implemented. The issue for this is #146.