An OLAP engine for Google Apps Script
This module implements a relational pipeline tookit for manipulating Google Sheets using Apps Script.
The central object is a Table, which uses a column oriented storage model for holding Values. The values can be read from a sheet or provided as JavaScript data structures. Tables can be written out to target sheets when processing is complete.
To support pipelining, the Table class provides data streaming methods for relational and data cleaning operations. This lets you write SQL-like chains of operations:
Table.fromSheet(<source sheet>)
.select(<expressions>)
.unnest(<array column>)
.fill(<column>, <default>)
.where(<predicate>)
.equiJoin(<Table>, <matching keys>)
.groupby(<groups>, <aggregates>)
.orderby(<ordering>)
.limit(100)
.toSheet(<target sheet>);Pipelines avoid copying data wherever possible, so intermediate results can be cached without worrying about later processing.
Expressions are implemented as a simple tree of nodes, and any function can be used in a function node. Column reference expressions make no copies. Aggregates are a separate type of object, with initialize, update and finalize functions, and the arguments can be arbitraary expressions. Ordering is also expression-based.
Note that some of this functionality is not yet implemented, and the error checking is pretty much non-existent, but the architecture is based on 50 years of database theory and is hopefully easy to express data validation and sheet generation with.