raqb/src/optimizers/set-collapse-by-columns.js

"use strict";

const NoChange = require("./util/no-change");
const deriveNode = require("../derive-node");
const operations = require("../operations");
const typeOf = require("../type-of");
const unreachable = require("../unreachable");
const concat = require("../concat");

const uniqueByPredicate = require("../unique-by-predicate");

// FIXME: Support for foreign column names

function uniqueColumns(columns) {
	return uniqueByPredicate(columns, (column) => column.name);
}

/*
valid columns when collapsing:
- columns that appear in the collapseBy column list, within or without a hierarchical wrapper
- any column that is wrapped in an aggregrate function of some sort
*/

module.exports = {
	name: "set-collapse-by-columns",
	category: [ "normalization" ],
	visitors: {
		collapseBy: (node, { setState }) => {
			setState("isCollapsing", true);
			return NoChange;
		},
		columnName: (node, { setState }) => {
			setState("columnSeen", node);
			return NoChange;
		},
		// FIXME: Think of a generic way to express "only match columns under this specific child property"
		collapseByColumns: (node, { registerStateHandler, setState, defer }) => {
			let columns = [];

			registerStateHandler("columnSeen", (node) => {
				columns.push(node);
			});

			return defer(() => {
				setState("setCollapsedColumns", columns);
				return NoChange;
			});
		},
		addColumns: (node, { setState }) => {
			setState("setAddColumns", node.columns);
			return NoChange;
		},
		onlyColumns: (node, { setState }) => {
			setState("setOnlyColumns", node.columns);
			return NoChange;
		},
		aggregrateFunction: (node, { registerStateHandler }) => {
			// FIXME: Also report isCollapsing here, due to aggregrate function use, but make sure that the error describes this as the (possible) cause
			return NoChange;
		},
		compute: (node, { setState }) => {
			setState("computeSeen", node);
			return NoChange;
		},
		select: (node, { registerStateHandler, defer }) => {
			let isCollapsing;
			let onlyColumns = [];
			let addColumns = [];
			let computes = [];
			let collapsedColumns;

			registerStateHandler("isCollapsing", (value) => {
				isCollapsing = isCollapsing || value;
			});

			registerStateHandler("setCollapsedColumns", (columns) => {
				if (collapsedColumns == null) {
					collapsedColumns = columns;
				} else {
					throw new Error(`You can currently only specify a single 'collapseBy' clause. Please file an issue if you have a reason to need more than one!`);
				}
			});

			registerStateHandler("setOnlyColumns", (columns) => {
				onlyColumns = onlyColumns.concat(columns);
			});

			registerStateHandler("setAddColumns", (columns) => {
				addColumns = addColumns.concat(columns);
			});

			registerStateHandler("computeSeen", (node) => {
				computes.push(node);
			});

			return defer(() => {
				if (isCollapsing) {
					if (addColumns.length > 0) {
						let extraColumnNames = addColumns.map((column) => column.name);

						throw new Error(`You tried to add extra columns (${extraColumnNames.join(", ")}) in your query, but this is not possible when using collapseBy. See [FIXME: link] for more information, and how to solve this.`);
					} else if (onlyColumns.length > 0) {
						// NOTE: This can happen either because the user specified an onlyColumns clause, *or* because a previous run of this optimizer did so!
						let uniqueSelectedColumns = uniqueColumns(onlyColumns);
						let collapsedColumnNames = collapsedColumns.map((column) => column.name);

						let invalidColumnSelection = uniqueSelectedColumns.filter((node) => {
							let isAggregrateComputation = typeOf(node) === "alias" && typeOf(node.expression) === "aggregrateFunction";
							let isCollapsedColumn = typeOf(node) === "columnName" && collapsedColumnNames.includes(node.name);

							let isValid = isAggregrateComputation || isCollapsedColumn;

							return !isValid;
						});

						// FIXME: We can probably optimize this by marking the optimizer-created onlyColumns as inherently-valid, via some sort of node metadata mechanism

						if (invalidColumnSelection.length > 0) {
							let invalidColumnNames = invalidColumnSelection.map((column) => {
								let columnType = typeOf(column);

								if (columnType === "columnName") {
									return column.name;
								} else if (columnType === "alias") {
									// FIXME: Show alias target instead of column name here?
									return column.column.name;
								} else {
									return unreachable(`Encountered '${columnType}' node in invalid columns`);
								}
							});

							throw new Error(`You tried to include one or more columns in your query (${invalidColumnNames.join(", ")}), that are not used in a collapseBy clause or aggregrate function. See [FIXME: link] for more information.`);
						} else {
							return NoChange;
						}
					} else {
						let computeAliases = computes.map((node) => {
							return operations.alias(node.column, node.expression);
						});

						return deriveNode(node, {
							clauses: node.clauses.concat([
								operations.onlyColumns(concat([
									collapsedColumns,
									computeAliases
								]))
							])
						});
					}
				}
			});
		},
	}
};

// FIXME: a ConsumeNode marker, like RemoveNode but it does not invalidate that node's state... may need to actually make it a reference, so that a parent node can decide whether to consume that node. Basically passing a "consume this node" function as a state value, that correctly internally triggers the optimizer infrastructure to change the tree as a result.
// FIXME: Consume the compute nodes, and have an optimizer that removes empty computeMultiple nodes