SQL for NoSQL Databases: Deja Vu (Part 2)

Providing a query interface for querying " schema-less " JSON documents using SQL as declarative query language [1] is finally gaining momentum (e.g. [4], [5]). The approach proposed in [2] called JSON SQL, however, represents an alternative approach as it adheres to the following principles: • extends the SQL 92 grammar ([3]) without changing the existing grammar rules, but by strictly adding new rules or extending existing rules • supports a syntax for specifying query results as documents or tables (!) with the results as documents precisely matching the stored documents' structure • defines a clear semantics based on the relational model • implements the complete set of JSON data types, including JSON true, JSON false and JSON null In JSON SQL no assumption is made about the JSON document structure except that it must comply with the JSON standard. Documents are grouped into collections (which imply neither a specific semantics nor specific constraints). A few examples are shown and explained next with more detail to be found in [2]: select {a, b} from exampleColl return a JSON document for each JSON document in exampleColl with properties " a " and/or " b " if present, if none is present, return the empty document select a, b from exampleColl return a table with two columns with the value of property " a " and/or the value of property " b " populated if present select {a AS a.b} from exampleColl project property " a " but relocate it to " a ". " b " in the result document select {a AS a.b.c, b AS a.b} from exampleColl illegal as " a ". " b " would remove " a ". " b ". " c " from the result select {*} from exampleColl where a.x.y = true return all documents completely where the value of property " a ". " x ". " y " is of literal JSON true select {*} from exampleColl where a = 'true' return all documents where the value of property " a " equals to String " true " select {*} from exampleColl where " true " < 'true' return all documents where the value of property " true " is less than the String " true ". select * from exampleColl where a.b = c.[3] return all properties as columns where the value of property " a ". " b " is …