Il mer 11 apr 2018, 22:34 AnilKumar B <akumarb2...@gmail.com> ha scritto:
> Hi All, > > I am new to calcite. Need your help/inputs on usage of calcite to solve one > of our use case. > > We need to build a library to get the lineage graph of given query. Here > given query can be either Hive/ANSI SQL/Vertica queries etc. > > We can build this library by using their respective physical plans. But, > there are few issues as mentioned below > 1. We need develop the library specific to each source > 2. In most of the sources, there is a limitation on explain query for > *create > table* queries. > 3. To get the explain plan, we need to have all the tables (which we are > used in query) should be in the in their respective meta stores. So, it > might cause the issues with tmp tables from previous queries. > > So, to avoid above issues, we are trying to develop a generic library which > can work for any source. For that, we are trying to use calcite parser's > SQLNode to derive the lineage for given query. > > We started with ANSI SQL queries and as of now, it's going good. But any > inputs from community > on below will be great help for us. > 1. Does calcite is correct fit for our use case? > 2. How to parse the JDBC queries and get the lineage? > Not sure if I understand correctly, but you are looking for something that given an SQL statement and metadata about existing tables it returns one (or more) access plan ? In this case Calcite is very a good choice. I will be happy to provide some example if this is your case. Enrico > > Thanks & Regards, > B Anil Kumar. > -- -- Enrico Olivelli