refactor!: More refactoring and cleanup.

docs/api/table.md

@@ -359,7 +359,7 @@ const colValues = Array.from(table.values('colA'));
 <hr/><a id="data" href="#data">#</a>
 <em>table</em>.<b>data</b>() · [Source](https://github.com/uwdata/arquero/blob/master/src/table/table.js)
 
-Returns the internal table storage data structure: an object with column names for keys and column arrays for values.
+Returns the internal table storage data structure: an object with column names for keys and column arrays for values. This method returns the same structure used by the Table (not a copy) and its contents should not be modified.
 
 <hr/><a id="get" href="#get">#</a>
 <em>table</em>.<b>get</b>(<i>name</i>[, <i>row</i>]) · [Source](https://github.com/uwdata/arquero/blob/master/src/table/column-table.js)
@@ -428,7 +428,7 @@ Perform a table scan, invoking the provided *callback* function for each row of
 
 * *callback*: Function invoked for each row of the table. The callback is invoked with the following arguments:
   * *row*: The table row index.
-  * *data*: The backing table data store.
+  * *data*: The backing table data store (as returned by table [`data`](#data) method).
   * *stop*: A function to stop the scan early. The callback can invoke *stop()* to prevent future scan calls.
 * *order*: A boolean flag (default `false`), indicating if the table should be scanned in the order determined by [orderby](verbs#orderby). This argument has no effect if the table is unordered.
 

package.json

@@ -14,12 +14,12 @@
   ],
   "license": "BSD-3-Clause",
   "author": "Jeffrey Heer (http://idl.cs.washington.edu)",
-  "exports": "./src/index-node.js",
+  "exports": "./src/index.js",
   "unpkg": "dist/arquero.min.js",
   "jsdelivr": "dist/arquero.min.js",
   "types": "dist/types/index.d.ts",
   "browser": {
-    "./src/index-node.js": "./src/index.js"
+    "./src/index.js": "./src/index-browser.js"
   },
   "repository": {
     "type": "git",

rollup.config.js

@@ -2,14 +2,8 @@ import bundleSize from 'rollup-plugin-bundle-size';
 import { nodeResolve } from '@rollup/plugin-node-resolve';
 import terser from '@rollup/plugin-terser';
 
-function onwarn(warning, defaultHandler) {
-  if (warning.code !== 'CIRCULAR_DEPENDENCY') {
-    defaultHandler(warning);
-  }
-}
-
 const name = 'aq';
-const external = [ 'apache-arrow', 'node-fetch' ];
+const external = [ 'apache-arrow' ];
 const globals = { 'apache-arrow': 'Arrow' };
 const plugins = [
   bundleSize(),
@@ -18,10 +12,9 @@ const plugins = [
 
 export default [
   {
-    input: 'src/index.js',
+    input: 'src/index-browser.js',
     external,
     plugins,
-    onwarn,
     output: [
       {
         file: 'dist/arquero.js',

src/api.js

@@ -0,0 +1,32 @@
+// export internal class and method definitions
+export { BitSet } from './table/BitSet.js';
+export { Table } from './table/Table.js';
+export { ColumnTable } from './table/ColumnTable.js';
+export { default as Reducer } from './verbs/reduce/reducer.js';
+export { default as parse } from './expression/parse.js';
+export { default as walk_ast } from './expression/ast/walk.js';
+
+// public API
+export { seed } from './util/random.js';
+export { default as fromArrow } from './arrow/from-arrow.js';
+export { default as fromCSV } from './format/from-csv.js';
+export { default as fromFixed } from './format/from-fixed.js';
+export { default as fromJSON } from './format/from-json.js';
+export { default as toArrow } from './arrow/to-arrow.js';
+export { default as toArrowIPC } from './arrow/to-arrow-ipc.js';
+export { default as toCSV } from './format/to-csv.js';
+export { default as toHTML } from './format/to-html.js';
+export { default as toJSON } from './format/to-json.js';
+export { default as toMarkdown } from './format/to-markdown.js';
+export { default as bin } from './helpers/bin.js';
+export { default as escape } from './helpers/escape.js';
+export { default as desc } from './helpers/desc.js';
+export { default as field } from './helpers/field.js';
+export { default as frac } from './helpers/frac.js';
+export { default as names } from './helpers/names.js';
+export { default as rolling } from './helpers/rolling.js';
+export { all, endswith, matches, not, range, startswith } from './helpers/selection.js';
+export { default as agg } from './verbs/helpers/agg.js';
+export { default as op } from './op/op-api.js';
+export { addAggregateFunction, addFunction, addWindowFunction } from './op/register.js';
+export { table, from } from './table/index.js';

src/format/from-text-rows.js

@@ -44,6 +44,7 @@ export default function(next, names, options) {
     }
   }
 
+  /** @type {import('../table/types.js').ColumnData} */
   const columns = {};
   names.forEach((name, i) => columns[name] = values[i]);
   return new ColumnTable(columns, names);

src/index-browser.js

@@ -0,0 +1,2 @@
+export * from './api.js';
+export { load, loadArrow, loadCSV, loadFixed, loadJSON } from './format/load-url.js';

src/index.js

@@ -1,33 +1,2 @@
-// export internal class and method definitions
-export { BitSet } from './table/BitSet.js';
-export { Table } from './table/Table.js';
-export { ColumnTable } from './table/ColumnTable.js';
-export { default as Reducer } from './verbs/reduce/reducer.js';
-export { default as parse } from './expression/parse.js';
-export { default as walk_ast } from './expression/ast/walk.js';
-
-// public API
-export { seed } from './util/random.js';
-export { default as fromArrow } from './arrow/from-arrow.js';
-export { default as fromCSV } from './format/from-csv.js';
-export { default as fromFixed } from './format/from-fixed.js';
-export { default as fromJSON } from './format/from-json.js';
-export { load, loadArrow, loadCSV, loadFixed, loadJSON } from './format/load-url.js';
-export { default as toArrow } from './arrow/to-arrow.js';
-export { default as toArrowIPC } from './arrow/to-arrow-ipc.js';
-export { default as toCSV } from './format/to-csv.js';
-export { default as toHTML } from './format/to-html.js';
-export { default as toJSON } from './format/to-json.js';
-export { default as toMarkdown } from './format/to-markdown.js';
-export { default as bin } from './helpers/bin.js';
-export { default as escape } from './helpers/escape.js';
-export { default as desc } from './helpers/desc.js';
-export { default as field } from './helpers/field.js';
-export { default as frac } from './helpers/frac.js';
-export { default as names } from './helpers/names.js';
-export { default as rolling } from './helpers/rolling.js';
-export { all, endswith, matches, not, range, startswith } from './helpers/selection.js';
-export { default as agg } from './verbs/helpers/agg.js';
-export { default as op } from './op/op-api.js';
-export * from './op/register.js';
-export * from './table/index.js';
+export * from './api.js';
+export { load, loadArrow, loadCSV, loadFixed, loadJSON } from './format/load-file.js';

src/table/ColumnTable.js

@@ -54,7 +54,7 @@ export class ColumnTable extends Table {
    * @example table.assign(table1, table2)
    */
   assign(...tables) {
-    return assign(this, tables.flat());
+    return assign(this, ...tables);
   }
 
   /**
@@ -140,7 +140,7 @@ export class ColumnTable extends Table {
    * @example table.groupby({ key: d => d.colA + d.colB })
    */
   groupby(...keys) {
-    return groupby(this, keys.flat());
+    return groupby(this, ...keys);
   }
 
   /**
@@ -168,7 +168,7 @@ export class ColumnTable extends Table {
    * @example table.orderby(desc(d => d.a))
    */
   orderby(...keys) {
-    return orderby(this, keys.flat());
+    return orderby(this, ...keys);
   }
 
   /**
@@ -204,7 +204,7 @@ export class ColumnTable extends Table {
    * @example table.rename({ a: 'a2', b: 'b2' })
    */
   rename(...columns) {
-    return rename(this, columns.flat());
+    return rename(this, ...columns);
   }
 
   /**
@@ -273,7 +273,7 @@ export class ColumnTable extends Table {
    * @example table.select({ colA: 'newA', colB: 'newB' })
    */
   select(...columns) {
-    return select(this, columns.flat());
+    return select(this, ...columns);
   }
 
   /**
@@ -313,7 +313,7 @@ export class ColumnTable extends Table {
    * @example table.dedupe({ abs: d => op.abs(d.a) })
    */
   dedupe(...keys) {
-    return dedupe(this, keys.flat());
+    return dedupe(this, ...keys);
   }
 
   /**
@@ -460,7 +460,7 @@ export class ColumnTable extends Table {
    * @example table.lookup(other, ['key1', 'key2'], 'value1', 'value2')
    */
   lookup(other, on, ...values) {
-    return lookup(this, other, on, values.flat());
+    return lookup(this, other, on, ...values);
   }
 
   /**
@@ -730,7 +730,7 @@ export class ColumnTable extends Table {
    * @example table.concat([other1, other2])
    */
   concat(...tables) {
-    return concat(this, tables.flat());
+    return concat(this, ...tables);
   }
 
   /**
@@ -747,7 +747,7 @@ export class ColumnTable extends Table {
    * @example table.union([other1, other2])
    */
   union(...tables) {
-    return union(this, tables.flat());
+    return union(this, ...tables);
   }
 
   /**
@@ -763,7 +763,7 @@ export class ColumnTable extends Table {
    * @example table.intersect([other1, other2])
    */
   intersect(...tables) {
-    return intersect(this, tables.flat());
+    return intersect(this, ...tables);
   }
 
   /**
@@ -779,7 +779,7 @@ export class ColumnTable extends Table {
    * @example table.except([other1, other2])
    */
   except(...tables) {
-    return except(this, tables.flat());
+    return except(this, ...tables);
   }
 
   // -- Table Output Formats ------------------------------------------------

src/table/Table.js

@@ -14,8 +14,10 @@ export class Table {
 
   /**
    * Instantiate a Table instance.
-   * @param {object} columns An object mapping column names to values.
-   * @param {string[]} [names] An ordered list of column names.
+   * @param {import('./types.js').ColumnData} columns
+   *  An object mapping column names to values.
+   * @param {string[]} [names]
+   *  An ordered list of column names.
    * @param {import('./BitSet.js').BitSet} [filter]
    *  A filtering BitSet.
    * @param {import('./types.js').GroupBySpec} [group]
@@ -26,17 +28,29 @@ export class Table {
    *  An object mapping parameter names to values.
    */
   constructor(columns, names, filter, group, order, params) {
-    const data = { ...columns };
+    const data = Object.freeze({ ...columns });
     names = names ?? Object.keys(data);
     const nrows = names.length ? data[names[0]].length : 0;
+    /** @private */
     this._names = Object.freeze(names);
+    /** @private */
     this._data = data;
+    /** @private */
     this._total = nrows;
+    /** @private */
     this._nrows = filter?.count() ?? nrows;
+    /** @private */
     this._mask = filter ?? null;
+    /** @private */
     this._group = group ?? null;
+    /** @private */
     this._order = order ?? null;
+    /** @private */
     this._params = params;
+    /** @private */
+    this._index = null;
+    /** @private */
+    this._partitions = null;
   }
 
   /**
@@ -94,9 +108,10 @@ export class Table {
    */
   get [Symbol.toStringTag]() {
     if (!this._names) return 'Object'; // bail if called on prototype
-    const nr = this.numRows() + ' row' + (this.numRows() !== 1 ? 's' : '');
-    const nc = this.numCols() + ' col' + (this.numCols() !== 1 ? 's' : '');
-    return `Table: ${nc} x ${nr}`
+    const nr = this.numRows();
+    const nc = this.numCols();
+    const plural = v => v !== 1 ? 's' : '';
+    return `Table: ${nc} col${plural(nc)} x ${nr} row${plural(nr)}`
       + (this.isFiltered() ? ` (${this.totalRows()} backing)` : '')
       + (this.isGrouped() ? `, ${this._group.size} groups` : '')
       + (this.isOrdered() ? ', ordered' : '');
@@ -127,9 +142,9 @@ export class Table {
   }
 
   /**
-   * Returns the internal table storage data structure.
+   * Get the backing column data for this table.
    * @return {import('./types.js').ColumnData}
-   *  The backing table storage data structure.
+   *  Object of named column instances.
    */
   data() {
     return this._data;
@@ -233,15 +248,6 @@ export class Table {
     return this._names.indexOf(name);
   }
 
-  /**
-   * Get the backing set of columns for this table.
-   * @return {import('./types.js').ColumnData}
-   *  Object of named column instances.
-   */
-  columns() {
-    return this._data;
-  }
-
   /**
    * Get the column instance with the given name.
    * @param {string} name The column name.

src/table/columns-from.js

@@ -6,8 +6,12 @@ import isObject from '../util/is-object.js';
 import isRegExp from '../util/is-regexp.js';
 import isString from '../util/is-string.js';
 
+/**
+ * @return {import('./types.js').ColumnData}
+ */
 export function columnsFrom(values, names) {
   const raise = type => error(`Illegal argument type: ${type || typeof values}`);
+  // @ts-ignore
   return values instanceof Map ? fromKeyValuePairs(values.entries(), names)
     : isDate(values) ? raise('Date')
     : isRegExp(values) ? raise('RegExp')

src/table/index.js

@@ -17,6 +17,7 @@ import { columnsFrom } from './columns-from.js';
  */
 export function table(columns, names) {
   if (columns instanceof ColumnTable) return columns;
+  /** @type {import('./types.js').ColumnData} */
   const data = {};
   const keys = [];
   for (const [key, value] of entries(columns)) {

src/verbs/assign.js

@@ -1,13 +1,14 @@
 import { columnSet } from '../table/ColumnSet.js';
-import { table as newTable } from '../table/index.js';
+import { Table } from '../table/Table.js';
 import error from '../util/error.js';
 
-export function assign(table, others) {
+export function assign(table, ...others) {
+  others = others.flat();
   const nrows = table.numRows();
   const base = table.reify();
   const cols = columnSet(base).groupby(base.groups());
   others.forEach(input => {
-    input = newTable(input);
+    input = input instanceof Table ? input : new Table(input);
     if (input.numRows() !== nrows) error('Assign row counts do not match');
     input = input.reify();
     input.columnNames(name => cols.add(name, input.column(name)));

src/verbs/concat.js

@@ -1,7 +1,8 @@
 import { columnSet } from '../table/ColumnSet.js';
 import NULL from '../util/null.js';
 
-export function concat(table, others) {
+export function concat(table, ...others) {
+  others = others.flat();
   const trows = table.numRows();
   const nrows = trows + others.reduce((n, t) => n + t.numRows(), 0);
   if (trows === nrows) return table;

src/verbs/dedupe.js

@@ -1,7 +1,8 @@
 import { groupby } from './groupby.js';
 import { filter } from './filter.js';
 
-export function dedupe(table, keys = []) {
+export function dedupe(table, ...keys) {
+  keys = keys.flat();
   const gt = groupby(table, keys.length ? keys : table.columnNames());
   return filter(gt, 'row_number() === 1').ungroup().reify();
 }