[UPDATE] refactor to use strided arrays.

TODO.md

@@ -215,7 +215,16 @@ TODO
 		-	row/col names
 			-	only getters!!
 	- 	See [link](http://pandas.pydata.org/pandas-docs/dev/comparison_with_sql.html#select)
-89. 
+89. instead of `data` --> `toArray`?
+90. allow column-order (order option)
+	-	this has deep implications for the rest of the module, with possible sig increased complexity
+91. deep copy TODOs only applies to generic array values; if known dtype, then deep copy is not necessary
+	-	will probably need to use code generation to accommodate all the various array types
+92. Creating new linear arrays
+	-	may want to create a routine for zero arrays for pre-initialization to avoid creating a sparse array
+		-	use `push`, rather than `new`
+		- 	this routine should replace all `new` instances in the code when creating new strided arrays
+93. 
 
 
 

lib/addCols.js

@@ -51,8 +51,12 @@ var isSize = require( 'validate.io-size' ),
 */
 function addCols( arr, options ) {
 	/* jshint validthis:true */
-	var nRows = this._nRows,
+	var data = this._data,
+		nRows = this._nRows,
+		nCols = this._nCols,
 		chash = this._chash,
+		rs = this._rStride,
+		cs = this._cStride,
 		opts = {},
 		hash,
 		names,
@@ -98,21 +102,17 @@ function addCols( arr, options ) {
 		if ( idx.length !== len ) {
 			throw new Error( 'addCols()::invalid option. Number of column indices must equal the number of added columns.' );
 		}
-		N = this._nCols + len - 1;
+		N = nCols + len - 1;
 		if ( !hasMax( idx, N ) ) {
 			throw new RangeError( 'addCols()::invalid option. A column index cannot exceed the total number of columns.' );
 		}
 	}
-	// Add the columns to the data array...
 	this._nCols += len;
 	if ( idx ) {
 		N = this._nCols;
-		d = new Array( nRows );
-		for ( i = 0; i < nRows; i++ ) {
-			d[ i ] = new Array( N );
-		}
-		n = new Array( N );
-		hash = {};
+
+		// Build a new data array...
+		d = new Array( nRows*N );
 		k = 0;
 		for ( j = 0; j < N; j++ ) {
 			flg = false;
@@ -123,16 +123,33 @@ function addCols( arr, options ) {
 				}
 			}
 			if ( flg ) {
-				name = names[ i ];
-				// TODO: deep copy
 				col = arr[ i ];
 				for ( i = 0; i < nRows; i++ ) {
-					d[ i ][ j ] = col[ i ];
+					// TODO: deep copy
+					d[ i*N + j*cs ] = col[ i ];
 				}
 			} else {
 				for ( i = 0; i < nRows; i++ ) {
-					d[ i ][ j ] = this._data[ i ][ k ];
+					d[ i*N + j*cs ] = data[ i*rs + k*cs ];
+				}
+				k++;
+			}
+		}
+		// Build a new column names array and an associated hash...
+		n = new Array( N );
+		hash = {};
+		k = 0;
+		for ( j = 0; j < N; j++ ) {
+			flg = false;
+			for ( i = 0; i < len; i++ ) {
+				if ( idx[ i ] === j ) {
+					flg = true;
+					break;
 				}
+			}
+			if ( flg ) {
+				name = names[ i ];
+			} else {
 				name = this._colnames[ k ];
 				k++;
 			}
@@ -143,27 +160,41 @@ function addCols( arr, options ) {
 				hash[ name ]= [ j ];
 			}
 		}
-		this._data = d;
 		this._colnames = n;
 		this._chash = hash;
 	} else {
-		for ( i = 0; i < len; i++ ) {
-			name = names[ i ];
+		N = this._nCols;
+		d = new Array( nRows*N );
 
-			// TODO: deep copy
-			col = arr[ i ];
-			for ( j = 0; j < nRows; j++ ) {
-				this._data[ j ].push( col[ j ] );
+		// Build a new data array...
+		k = 0;
+		for ( j = 0; j < N; j++ ) {
+			if ( j < nCols ) {
+				for ( i = 0; i < nRows; i++ ) {
+					d[ i*N + j*cs ] = data[ i*rs + j*cs ];
+				}
+			} else {
+				col = arr[ k ];
+				for ( i = 0; i < nRows; i++ ) {
+					// TODO: deep copy
+					d[ i*N + j*cs ] = col[ i ];
+				}
+				k++;
 			}
+		}
+		// Append the new column names to the existing column name array and update the column name hash...
+		for ( i = 0; i < len; i++ ) {
+			name = names[ i ];
 			this._colnames.push( name );
-			N = this._colnames.length;
 			if ( chash.hasOwnProperty( name ) ) {
-				chash[ name ].push( N );
+				chash[ name ].push( nCols+i );
 			} else {
-				chash[ name ] = [ N ];
+				chash[ name ] = [ nCols+i ];
 			}
 		}
 	}
+	this._data = d;
+	this._rStride = this._nCols;
 	return this;
 } // end FUNCTION addCols()
 

lib/addRows.js

@@ -52,12 +52,17 @@ var isSize = require( 'validate.io-size' ),
 */
 function addRows( arr, options ) {
 	/* jshint validthis:true */
-	var nCols = this._nCols,
+	var data = this._data,
+		nRows = this._nRows,
+		nCols = this._nCols,
 		rhash = this._rhash,
+		rs = this._rStride,
+		cs = this._cStride,
 		opts = {},
 		hash,
 		names,
 		name,
+		row,
 		idx,
 		len,
 		flg,
@@ -98,16 +103,40 @@ function addRows( arr, options ) {
 		if ( idx.length !== len ) {
 			throw new Error( 'addRows()::invalid option. Number of row indices must equal the number of added rows.' );
 		}
-		N = this._nRows + len - 1;
+		N = nRows + len - 1;
 		if ( !hasMax( idx, N ) ) {
 			throw new RangeError( 'addRows()::invalid option. A row index cannot exceed the total number of rows.' );
 		}
 	}
-	// Add the rows to the data array...
 	this._nRows += len;
 	if ( idx ) {
 		N = this._nRows;
-		d = new Array( N );
+
+		// Build a new data array...
+		d = new Array( N*nCols );
+		k = 0;
+		for ( i = 0; i < N; i++ ) {
+			flg = false;
+			for ( j = 0; j < len; j++ ) {
+				if ( idx[ j ] === i ) {
+					flg = true;
+					break;
+				}
+			}
+			if ( flg ) {
+				row = arr[ j ];
+				for ( j = 0; j < nCols; j++ ) {
+					// TODO: deep copy
+					d[ i*rs + j*cs ] = row[ j ];
+				}
+			} else {
+				for ( j = 0; j < nCols; j++ ) {
+					d[ i*rs + j*cs ] = data[ k*rs + j*cs ];
+				}
+				k++;
+			}
+		}
+		// Build a new row names array and an associated hash...
 		n = new Array( N );
 		hash = {};
 		k = 0;
@@ -120,11 +149,8 @@ function addRows( arr, options ) {
 				}
 			}
 			if ( flg ) {
-				// TODO: deep copy
-				d[ i ] = arr[ j ].slice();
 				name = names[ j ];
 			} else {
-				d[ i ] = this._data[ k ];
 				name = this._rownames[ k ];
 				k++;
 			}
@@ -139,18 +165,22 @@ function addRows( arr, options ) {
 		this._rownames = n;
 		this._rhash = hash;
 	} else {
+		// Append the row data to the existing data array...
+		for ( i = 0; i < len; i++ ) {
+			row = arr[ i ];
+			for ( j = 0; j < nCols; j++ ) {
+				// TODO: deep copy
+				data.push( row[ j ] );
+			}
+		}
+		// Append the new row names to the existing row name array and update the row name hash...
 		for ( i = 0; i < len; i++ ) {
 			name = names[ i ];
-
-			// TODO: deep copy
-			this._data.push( arr[ i ].slice() );
-
 			this._rownames.push( name );
-			N = this._rownames.length;
 			if ( rhash.hasOwnProperty( name ) ) {
-				rhash[ name ].push( N );
+				rhash[ name ].push( nRows+i );
 			} else {
-				rhash[ name ] = [ N ];
+				rhash[ name ] = [ nRows+i ];
 			}
 		}
 	}

lib/ctor.js

@@ -55,9 +55,11 @@ function DataFrame( arr, options ) {
 		names,
 		name,
 		data,
-		len,
-		N,
-		i;
+		nRows,
+		nCols,
+		row,
+		i, j;
+
 	if ( !nArgs ) {
 		throw new Error( 'DataFrame()::insufficient input arguments. Must provide a data array.' );
 	} else if ( nArgs === 2 ) {
@@ -70,60 +72,64 @@ function DataFrame( arr, options ) {
 	if ( !isArray( arr ) ) {
 		throw new TypeError( 'DataFrame()::invalid input argument. Must provide a data array.' );
 	}
-	len = arr.length;
+	nRows = arr.length;
 	// [1] Dirty check...
 	if ( !isArray( arr[ 0 ] ) ) {
 		throw new TypeError( 'DataFrame()::invalid input argument. Data must be an array of arrays.' );
 	}
 	// [2] All first-level arrays must be equal length...
-	N = arr[ 0 ].length;
-	if ( !isSize( arr, [ null, N ] ) ) {
+	nCols = arr[ 0 ].length;
+	if ( !isSize( arr, [ null, nCols ] ) ) {
 		throw new Error( 'DataFrame()::invalid input argument. Input data array must contain equal length arrays.' );
 	}
 
-	// TODO: deep copy; this is a shallow copy. Need to acct for Boolean, Date, RegExp, Number, circular references, etc.
-	data = new Array( len );
+	// TODO: allow for dtype; e.g., Uint8Array, etc
 
-	// [3] Copy the input data...
-	for ( i = 0; i < len; i++ ) {
-		data[ i ] = arr[ i ].slice();
+	// [3] Copy the input data into a 1D array...
+	data = new Array( nRows*nCols );
+	for ( i = 0; i < nRows; i++ ) {
+		row = arr[ i ];
+		for ( j = 0; j < nCols; j++ ) {
+			// TODO: deep copy. Need to acct for Boolean, Date, RegExp, Number, String, circular refs, etc.
+			data[ i*nCols + j ] = row[ j ];
+		}
 	}
 	// [4] Number of row names must equal number of rows...
 	if ( !opts.hasOwnProperty( 'rownames' ) ) {
-		this._rownames = new Array( len );
-		for ( i = 0; i < len; i++ ) {
+		this._rownames = new Array( nRows );
+		for ( i = 0; i < nRows; i++ ) {
 			this._rownames[ i ] = i.toString();
 		}
 	} else {
 		names = opts.rownames;
 		if ( !isStringArray( names ) ) {
 			throw new TypeError( 'DataFrame()::invalid option. Row names must be an array of strings.' );
 		}
-		if ( names.length !== len ) {
+		if ( names.length !== nRows ) {
 			throw new Error( 'DataFrame()::invalid option. Number of row names must equal the number of rows.' );
 		}
 		this._rownames = names.slice();
 	}
 	// [5] Number of column names must equal number of columns...
 	if ( !opts.hasOwnProperty( 'colnames' ) ) {
-		this._colnames = new Array( N );
-		for ( i = 0; i < N; i++ ) {
+		this._colnames = new Array( nCols );
+		for ( i = 0; i < nCols; i++ ) {
 			this._colnames[ i ] = i.toString();
 		}
 	} else {
 		names = opts.colnames;
 		if ( !isStringArray( names ) ) {
 			throw new TypeError( 'DataFrame()::invalid option. Column names must be an array.' );
 		}
-		if ( opts.colnames.length !== N ) {
+		if ( opts.colnames.length !== nCols ) {
 			throw new Error( 'DataFrame()::invalid option. Number of column names does not equal the number of columns.' );
 		}
 		this._colnames = names.slice();
 	}
 	// [5] Create lookup tables for row and column names...
 	this._rhash = {};
 	names = this._rownames;
-	for ( i = 0; i < names.length; i++ ) {
+	for ( i = 0; i < nRows; i++ ) {
 		name = names[ i ];
 		if ( this._rhash.hasOwnProperty( name ) ) {
 			this._rhash[ name ].push( i );
@@ -133,7 +139,7 @@ function DataFrame( arr, options ) {
 	}
 	this._chash = {};
 	names = this._colnames;
-	for ( i = 0; i < names.length; i++ ) {
+	for ( i = 0; i < nCols; i++ ) {
 		name = names[ i ];
 		if ( this._chash.hasOwnProperty( name ) ) {
 			this._chash[ name ].push( i );
@@ -142,10 +148,12 @@ function DataFrame( arr, options ) {
 		}
 	}
 	this._data = data;
-	this._nRows = len;
-	this._nCols = N;
-	this._rid = len;
-	this._cid = N;
+	this._nRows = nRows;
+	this._nCols = nCols;
+	this._rStride = nCols;
+	this._cStride = 1;
+	this._rid = nRows;
+	this._cid = nCols;
 	return this;
 } // end FUNCTION DataFrame()