There are three kinds of queries supported for keywords:
- field
- term
- term in field
Each file, and each row group within the file, has 3 bloom filters to handle these queries.
So something like:
{"user": {"name": "John", "tags": [{"type": "user"}, {"role": "admin"}]}}
Gets turned into queryable pairs of:
[{Path: "user.name", Values: ["John"]}, {Path: "user.tags.type", Values: ["user"]}, {Path: "user.tags.role", Values: ["admin"]}]
Then you can search for:
- any record that has "john" in it
- any record that has the "user.tags.type" key
- any record that has "user.tags.type"="user" and "user.tags.role"="admin"
Which bloom filters are used depends on how you build the query, but they test for whether a row matching the condition(s) is in the file/row group
- field
- term
- term in field
Each file, and each row group within the file, has 3 bloom filters to handle these queries.
So something like:
{"user": {"name": "John", "tags": [{"type": "user"}, {"role": "admin"}]}}
Gets turned into queryable pairs of:
[{Path: "user.name", Values: ["John"]}, {Path: "user.tags.type", Values: ["user"]}, {Path: "user.tags.role", Values: ["admin"]}]
Then you can search for:
- any record that has "john" in it
- any record that has the "user.tags.type" key
- any record that has "user.tags.type"="user" and "user.tags.role"="admin"
Which bloom filters are used depends on how you build the query, but they test for whether a row matching the condition(s) is in the file/row group