Generating example data for testing database queries
View Patent ↗Computer-implemented methods, modules and clients relate to expanded, pruned sample table for testing database queries against a base table. The expanded, pruned sample table is formed from the base table by a process of initial sampling, synthesis, and pruning.
1. A computer-implemented method of forming a sample table from a base table for testing semantics of a query to be run on the base table, comprising:
sampling the base table to form an initial sample table and forming an intermediate table of each operator from an operator tree of the query by passing the initial sample table through the operator tree;
inserting suggestion records into the intermediate table based on incompleteness in the intermediate table to form an expanded sample table;
passing the expanded sample table through the operator tree to form an expanded intermediate table;
pruning the expanded sample table to form a final sample table, wherein said pruning comprises:
grouping equivalence classes of equivalent records in the expanded intermediate table;
retaining a set of expanded sample table records whose corresponding set of intermediate records includes at least one record from all equivalence classes while being as small as possible; and
classifying expanded sample table records and expanded intermediate table records into lineage groups, wherein the lineage groups each correspond to a set of records that derive from each other as they flow through the operator tree.
2. The computer-implemented method of claim 1 , wherein the step of synthesizing employs suggestion records inserted into the intermediate table to construct the expanded sample table.
3. The computer-implemented method of claim 2 , wherein the suggestion records propagated up the operator tree to a leaf operator of the tree where they and the intermediate table are use to form the expanded sample table from the base table.
4. The computer-implemented method of claim 2 , wherein the suggestion record is listing of a field paired with either a concrete value or a token denoting that any value is acceptable.
5. The computer-implemented method of claim 2 , wherein the structure of the suggestion record varies depending on the operator involved.
6. The computer-implemented method of claim 1 , wherein the equivalence classes are formed differently for different operators.
7. A computer-implemented method of forming a sample table from a base table for testing semantics of a query to be run on the base table, comprising:
sampling the base table to form an initial sample table and forming an intermediate table of each operator from an operator tree of the query by passing the initial sample table through the operator tree; and
pruning the initial sample table to form a final sample table, wherein said pruning comprises:
grouping equivalence classes of equivalent records in the intermediate table;
retaining a set of sample table records whose corresponding set of intermediate records includes at least one record from all equivalence classes while being as small as possible; and
classifying sample table records and intermediate table records into lineage groups, wherein the lineage groups each correspond to a set of records that derive from each other as they flow through the operator tree.
8. The computer-implemented method of claim 7 , wherein the equivalence classes are formed differently for different operators.
9. A computer-implemented module for interfacing with a database server and a database client to form a sample database from a base table, comprising:
an input element configured to receive a query comprising a set of operators that form an operator tree;
a database sampling element configured to sample the base table to form an initial sample table;
a processing element configured to form an intermediate table by passing the initial sample table through the operator tree, to synthesize an expanded sample table from the base table based on incompleteness in the intermediate table of each operator, and to prune the expanded sample table to form a final sample table, wherein the pruning proceeds by grouping equivalence classes of equivalent records in an expanded intermediate table, formed by passing the expanded sample table through the operator tree, retaining a set of expanded sample table records whose corresponding set of intermediate records includes at least one record from all equivalent classes while being as small as possible, and classifying expanded sample table records and expanded intermediate table records into lineage groups, wherein the lineage groups each correspond to a set of records that derive from each other as they flow through the operator tree.
10. The computer-implemented module of claim 9 , wherein synthesizing employs suggestion records inserted into the intermediate table to construct the expanded sample table.
11. The computer-implemented module of claim 10 , wherein the suggestion records propagated up the operator tree to a leaf operator of the tree where they and the intermediate table are use to form the expanded sample table from the base table.
12. The computer-implemented module of claim 10 , wherein the suggestion record is listing of a field paired with either a concrete value or a token denoting that any value is acceptable.
13. The computer-implemented module of claim 10 , wherein the structure of the suggestion record varies depending on the operator involved.
14. The computer-implemented module of claim 9 , wherein the equivalence classes are formed differently for different operators.