version bump 0.7.5: more performance

- eliminated functional constructs in hot functions
- format try-catch block extracted into new function
- cpexcel + codepage updated to 1.2.0
- more efficient (and correct) clean implementation of RGB/HSL/tint algorithms
- xlsx binary --all option enables every extra formatting and saving option
- column widths parsed and saved (requires cellStyles:true)

Author: SheetJSDev

bin/xlsx.njs

@@ -22,6 +22,7 @@ program
 	.option('-n, --sheet-rows <num>', 'Number of rows to process (0=all rows)')
 	.option('--no-sst', 'do not generate sst')
 	.option('--perf', 'do not generate output')
+	.option('--all', 'parse everything; XLS[XMB] write as much as possible')
 	.option('--dev', 'development mode')
 	.option('--read', 'read but do not print out contents')
 	.option('-q, --quiet', 'quiet mode');
@@ -62,6 +63,12 @@ if(program.xlsx || program.xlsm || program.xlsb) {
 else if(program.formulae);
 else opts.cellFormula = false;
 
+if(program.all) {
+	opts.cellFormula = true;
+	opts.cellNF = true;
+	opts.cellStyles = true;
+}
+
 if(program.dev) {
 	X.verbose = 2;
 	opts.WTF = true;

bits/01_version.js

@@ -1 +1 @@
-XLSX.version = '0.7.4';
+XLSX.version = '0.7.5';

bits/30_jsutils.js renamed to bits/20_jsutils.js

@@ -6,11 +6,16 @@ function parsexmltag(tag) {
 	var words = tag.split(/\s+/);
 	var z = {'0': words[0]};
 	if(words.length === 1) return z;
-	(tag.match(attregexg) || []).map(function(x){
-		var y=x.match(attregex);
-		y[1] = y[1].replace(/xmlns:/,"xmlns");
-		z[y[1].replace(/^[a-zA-Z]*:/,"")] = y[2].substr(1,y[2].length-2);
-	});
+	var m = tag.match(attregexg), y, j, w, i;
+	if(m) for(i = 0; i != m.length; ++i) {
+		y = m[i].match(attregex);
+		if((j=y[1].indexOf(":")) === -1) z[y[1]] = y[2].substr(1,y[2].length-2);
+		else {
+			if(y[1].substr(0,6) === "xmlns:") w = "xmlns"+y[1].substr(6);
+			else w = y[1].substr(j+1);
+			z[w] = y[2].substr(1,y[2].length-2);
+		}
+	}
 	return z;
 }
 
@@ -27,7 +32,7 @@ var rencstr = "&<>'\"".split("");
 // TODO: CP remap (need to read file version to determine OS)
 function unescapexml(text){
 	var s = text + '';
-	s = s.replace(/&quot;/g, '"').replace(/&apos;/g, "'").replace(/&gt;/g, ">").replace(/&lt;/g, "<").replace(/&amp;/g, "&");
+	s = s.replace(/&[a-z]*;/g, function($$) { return encodings[$$]; });
 	return s.replace(/_x([0-9a-fA-F]*)_/g,function(m,c) {return _chr(parseInt(c,16));});
 }
 function escapexml(text){

bits/35_ziputils.js renamed to bits/21_ziputils.js

@@ -0,0 +1,61 @@
+var hex2RGB = function(h) {return h.substr(h[0]==="#"?1:0,6).match(/../g).map(function(x) { return parseInt(x,16); });};
+var rgb2Hex = function(rgb) {
+	for(var i=0,o=1; i!=3; ++i) o = o*256 + (rgb[i]>255?255:rgb[i]<0?0:rgb[i]);
+	return o.toString(16).toUpperCase().substr(1);
+};
+
+var rgb2HSL = function(rgb) {
+	var R = rgb[0]/255, G = rgb[1]/255, B=rgb[2]/255;
+	var M = Math.max(R, G, B), m = Math.min(R, G, B), C = M - m;
+	if(C === 0) return [0, 0, R];
+
+	var H6 = 0, S = 0, L2 = (M + m);
+	S = C / (L2 > 1 ? 2 - L2 : L2);
+	switch(M){
+		case R: H6 = ((G - B) / C + 6)%6; break;
+		case G: H6 = ((B - R) / C + 2); break;
+		case B: H6 = ((R - G) / C + 4); break;
+	}
+	return [H6 / 6, S, L2 / 2];
+};
+
+var hsl2RGB = function(hsl){
+	var H = hsl[0], S = hsl[1], L = hsl[2];
+	var C = S * 2 * (L < 0.5 ? L : 1 - L), m = L - C/2;
+	var rgb = [m,m,m], h6 = 6*H;
+
+	var X;
+	if(S !== 0) switch(h6|0) {
+		case 0: case 6: X = C * h6; rgb[0] += C; rgb[1] += X; break;
+		case 1: X = C * (2 - h6);   rgb[0] += X; rgb[1] += C; break;
+		case 2: X = C * (h6 - 2);   rgb[1] += C; rgb[2] += X; break;
+		case 3: X = C * (4 - h6);   rgb[1] += X; rgb[2] += C; break;
+		case 4: X = C * (h6 - 4);   rgb[2] += C; rgb[0] += X; break;
+		case 5: X = C * (6 - h6);   rgb[2] += X; rgb[0] += C; break;
+	}
+	for(var i = 0; i != 3; ++i) rgb[i] = Math.round(rgb[i]*255);
+	return rgb;
+};
+
+/* 18.8.3 bgColor tint algorithm */
+function rgb_tint(hex, tint) {
+	if(tint === 0) return hex;
+	var hsl = rgb2HSL(hex2RGB(hex));
+	if (tint < 0) hsl[2] = hsl[2] * (1 + tint);
+	else hsl[2] = 1 - (1 - hsl[2]) * (1 - tint);
+	return rgb2Hex(hsl2RGB(hsl));
+}
+
+/* 18.3.1.13 width calculations */
+var DEF_MDW = 7, MAX_MDW = 15, MIN_MDW = 1, MDW = DEF_MDW;
+function width2px(width) { return (( width + ((128/MDW)|0)/256 )* MDW )|0; }
+function px2char(px) { return (((px - 5)/MDW * 100 + 0.5)|0)/100; }
+function char2width(chr) { return (((chr * MDW + 5)/MDW*256)|0)/256; }
+function cycle_width(collw) { return char2width(px2char(width2px(collw))); }
+function find_mdw(collw, coll) {
+	if(cycle_width(collw) != collw) {
+		for(MDW=DEF_MDW; MDW>MIN_MDW; --MDW) if(cycle_width(collw) === collw) break;
+		if(MDW === MIN_MDW) for(MDW=DEF_MDW+1; MDW<MAX_MDW; ++MDW) if(cycle_width(collw) === collw) break;
+		if(MDW === MAX_MDW) MDW = DEF_MDW;
+	}
+}

bits/36_xmlutils.js renamed to bits/22_xmlutils.js

@@ -0,0 +1,41 @@
+var strs = {}; // shared strings
+var _ssfopts = {}; // spreadsheet formatting options
+
+RELS.WS = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/worksheet";
+
+function get_sst_id(sst, str) {
+	for(var i = 0; i != sst.length; ++i) if(sst[i].t === str) { sst.Count ++; return i; }
+	sst[sst.length] = {t:str}; sst.Count ++; sst.Unique ++; return sst.length-1;
+}
+
+function get_cell_style(styles, cell, opts) {
+	var z = opts.revssf[cell.z||"General"];
+	for(var i = 0; i != styles.length; ++i) if(styles[i].numFmtId === z) return i;
+	styles[styles.length] = {
+		numFmtId:z,
+		fontId:0,
+		fillId:0,
+		borderId:0,
+		xfId:0,
+		applyNumberFormat:1
+	};
+	return styles.length-1;
+}
+
+function safe_format(p, fmtid, fillid, opts) {
+	try {
+		p.w = SSF.format(fmtid,p.v,_ssfopts);
+		if(opts.cellNF) p.z = SSF._table[fmtid];
+	} catch(e) { if(opts.WTF) throw e; }
+	if(fillid) try {
+		p.s = styles.Fills[fillid];
+		if (p.s.fgColor && p.s.fgColor.theme) {
+			p.s.fgColor.rgb = rgb_tint(themes.themeElements.clrScheme[p.s.fgColor.theme].rgb, p.s.fgColor.tint || 0);
+			if(opts.WTF) p.s.fgColor.raw_rgb = themes.themeElements.clrScheme[p.s.fgColor.theme].rgb;
+		}
+		if (p.s.bgColor && p.s.bgColor.theme) {
+			p.s.bgColor.rgb = rgb_tint(themes.themeElements.clrScheme[p.s.bgColor.theme].rgb, p.s.bgColor.tint || 0);
+			if(opts.WTF) p.s.bgColor.raw_rgb = themes.themeElements.clrScheme[p.s.bgColor.theme].rgb;
+		}
+	} catch(e) { if(opts.WTF) throw e; }
+}

bits/37_binutils.js renamed to bits/23_binutils.js

@@ -17,6 +17,27 @@ function parse_ws_xml(data, opts, rels) {
 		});
 	}
 
+	/* 18.3.1.17 cols CT_Cols */
+	var columns = [];
+	if(opts.cellStyles && data.match(/<\/cols>/)) {
+		/* 18.3.1.13 col CT_Col */
+		var cols = data.match(/<col[^>]*\/>/g);
+		var seencol = false;
+		for(var coli = 0; coli != cols.length; ++coli) {
+			var coll = parsexmltag(cols[coli]);
+			delete coll[0];
+			var colm=Number(coll.min)-1, colM=Number(coll.max)-1;
+			delete coll.min, coll.max;
+			if(!seencol && coll.width) { seencol = true; find_mdw(+coll.width, coll); }
+			if(coll.width) {
+				coll.wpx = width2px(+coll.width);
+				coll.wch = px2char(coll.wpx);
+				coll.MDW = MDW;
+			}
+			while(colm <= colM) columns[colm++] = coll;
+		}
+	}
+
 	var refguess = {s: {r:1000000, c:1000000}, e: {r:0, c:0} };
 	var sidx = 0;
 
@@ -34,7 +55,8 @@ function parse_ws_xml(data, opts, rels) {
 		if(refguess.e.r < row.r - 1) refguess.e.r = row.r - 1;
 		/* 18.3.1.4 c CT_Cell */
 		var cells = x.substr(x.indexOf('>')+1).split(/<(?:\w+:)?c /);
-		for(var ix = 0, c=cells[0]; ix != cells.length; ++ix,c=cells[ix]) {
+		for(var ix = 0, c=cells[0]; ix != cells.length; ++ix) {
+			c = cells[ix];
 			if(c === "" || c.trim() === "") continue;
 			var cref = c.match(/r=["']([^"']*)["']/), idx = ix;
 			c = "<c " + c;
@@ -85,21 +107,9 @@ function parse_ws_xml(data, opts, rels) {
 				if(cf && cf.numFmtId) fmtid = cf.numFmtId;
 				if(opts.cellStyles && cf && cf.fillId) fillid = cf.fillId;
 			}
-			try {
-				p.w = SSF.format(fmtid,p.v,_ssfopts);
-				if(opts.cellNF) p.z = SSF._table[fmtid];
-				if(fillid) {
-					p.s = styles.Fills[fillid];
-					if (p.s.fgColor && p.s.fgColor.theme) {
-						p.s.fgColor.rgb = rgb_tint(themes.themeElements.clrScheme[p.s.fgColor.theme].rgb, p.s.fgColor.tint || 0);
-					}
-					if (p.s.bgColor && p.s.bgColor.theme) {
-						p.s.bgColor.rgb = rgb_tint(themes.themeElements.clrScheme[p.s.bgColor.theme].rgb, p.s.bgColor.tint || 0);
-					}
-				}
-			} catch(e) { if(opts.WTF) throw e; }
+			safe_format(p, fmtid, fillid, opts);
 			s[cell.r] = p;
-		};
+		}
 	}
 
 	/* 18.3.1.48 hyperlinks CT_Hyperlinks */
@@ -134,9 +144,11 @@ function parse_ws_xml(data, opts, rels) {
 		}
 	}
 	if(mergecells.length > 0) s["!merges"] = mergecells;
+	if(columns.length > 0) s["!cols"] = columns;
 	return s;
 }
 
+
 var WS_XML_ROOT = writextag('worksheet', null, {
 	'xmlns': XMLNS.main[0],
 	'xmlns:r': XMLNS.r
@@ -176,12 +188,28 @@ var write_ws_xml_data = function(ws, opts, idx, wb) {
 	return o.join("");
 };
 
+var write_ws_cols = function(ws, cols) {
+	var o = ["<cols>"], col, width;
+	for(var i = 0; i != cols.length; ++i) {
+		if(!(col = cols[i])) continue;
+		var p = {min:i+1,max:i+1};
+		/* wch (chars), wpx (pixels) */
+		width = -1;
+		if(col.wpx) width = px2char(col.wpx);
+		else if(col.wch) width = col.wch;
+		if(width > -1) { p.width = char2width(width); p.customWidth= 1; }
+		o.push(writextag('col', null, p));
+	}
+	o.push("</cols>");
+	return o.join("");
+};
+
 var write_ws_xml = function(idx, opts, wb) {
 	var o = [], s = wb.SheetNames[idx], ws = wb.Sheets[s] || {}, sidx = 0, rdata = "";
 	o.push(XML_HEADER);
 	o.push(WS_XML_ROOT);
 	o.push(writextag('dimension', null, {'ref': ws['!ref'] || 'A1'}));
-
+	if((ws['!cols']||[]).length > 0) o.push(write_ws_cols(ws, ws['!cols'])); 
 	sidx = o.length;
 	o.push(writextag('sheetData', null));
 	if(ws['!ref']) rdata = write_ws_xml_data(ws, opts, idx, wb);

bits/38_hoppers.js renamed to bits/24_hoppers.js

@@ -176,10 +176,7 @@ var parse_ws_bin = function(data, opts, rels) {
 					case 'str': p.v = utf8read(val[1]); break;
 				}
 				if(opts.cellFormula && val.length > 3) p.f = val[3];
-				if((cf = styles.CellXf[val[0].iStyleRef])) try {
-					p.w = SSF.format(cf.ifmt,p.v,_ssfopts);
-					if(opts.cellNF) p.z = SSF._table[cf.ifmt];
-				} catch(e) { if(opts.WTF) throw e; }
+				if((cf = styles.CellXf[val[0].iStyleRef])) safe_format(p,cf.ifmt,null,opts);
 				s[encode_cell({c:val[0].c,r:row.r})] = p;
 				if(refguess.s.r > row.r) refguess.s.r = row.r;
 				if(refguess.s.c > val[0].c) refguess.s.c = val[0].c;