version bump 0.1.1: DIF/SYLK write

- DIF write support
- DIF interpret dates
- SYLK write
- socialcalc write
- DBF trims strings
- DBF handles dateNF flag
- included cp 1257 1258 10079 10081
- socialcalc roundtrips special chars (fixes #4 h/t @audreyt)

README.md

@@ -9,14 +9,14 @@ File format support for known spreadsheet formats:
 |:-------------------------------------------------------------|:-----:|:-----:|
 | **Excel Supported Text Formats**                             |:-----:|:-----:|
 | Delimiter-Separated Values (CSV/TSV/DSV)                     |  :o:  |       |
-| Data Interchange Format (DIF)                                |  :o:  |       |
-| Symbolic Link (SYLK/SLK)                                     |  :o:  |       |
+| Data Interchange Format (DIF)                                |  :o:  |  :o:  |
+| Symbolic Link (SYLK/SLK)                                     |  :o:  |  :o:  |
 | Space-Delimited Text (PRN)                                   |  :o:  |       |
 | UTF-16 Unicode Text (TXT)                                    |  :o:  |       |
 | **Other Workbook/Worksheet Formats**                         |:-----:|:-----:|
 | dBASE II/III/IV / Visual FoxPro (DBF)                        |  :o:  |       |
 | **Other Output Formats**                                     |:-----:|:-----:|
-| SocialCalc                                                   |  :o:  |       |
+| SocialCalc                                                   |  :o:  |  :o:  |
 
 js-harb follows [Common Spreadsheet Format](https://github.com/sheetjs/js-xlsx).
 The objects can be used in conjunction with readers and writers from `js-xlsx`
@@ -25,17 +25,47 @@ and other libraries.
 
 ## Installation
 
-In [nodejs](https://www.npmjs.org/package/harb):
+With [npm](https://www.npmjs.org/package/harb):
 
 ```bash
 $ npm install harb
 ```
 
-## Usage
+## Interface
 
-This module provides support for [j](https://www.npmjs.org/package/j). For usage
-information, consult [the xlsx module](http://git.io/xlsx) as they use the same
-interface and style.
+`HARB` is the exposed variable in the browser and the exported node variable
+
+`HARB.version` is the version of the library (added by the build script).
+
+### Parsing functions
+
+`HARB.read(data, read_opts)` attempts to parse `data`.
+
+`HARB.readFile(filename, read_opts)` attempts to read `filename` and parse.
+
+### Utilities
+
+Utilities are available in the `HARB.utils` object:
+
+Exporting:
+
+- `sheet_to_socialcalc` converts a worksheet object to socialcalc format.
+
+The [utilities from js-xlsx](https://github.com/sheetjs/js-xlsx/#utilities) work
+with the workbook/worksheet objects from js-harb:
+
+- `sheet_to_json` converts a worksheet object to an array of JSON objects.
+  `sheet_to_row_object_array` is an alias that will be removed in the future.
+- `sheet_to_csv` generates delimiter-separated-values output.
+- `sheet_to_formulae` generates a list of the formulae (with value fallbacks).
+
+## Parsing Options
+
+The exported `read` and `readFile` functions accept an options argument:
+
+| Option Name | Default | Description                                          |
+| :---------- | ------: | :--------------------------------------------------- |
+| dateNF      | ""      | override the date format                             |
 
 ## File Formats
 

bin/harb.njs

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-/* harb.js (C) 2014-present SheetJS -- http://sheetjs.com */
+/* harb.js (C) 2014-present  SheetJS -- http://sheetjs.com */
 var n = "harb";
 /* vim: set ts=2 ft=javascript: */
 var X = require('../');
@@ -10,13 +10,18 @@ program
 	.usage('[options] <file> [sheetname]')
 	.option('-f, --file <file>', 'use specified workbook')
 	.option('-s, --sheet <sheet>', 'print specified sheet (default first sheet)')
-	.option('-p, --password <pw>', 'if file is encrypted, try with specified pw')
+	.option('-N, --sheet-index <idx>', 'use specified sheet index (0-based)')
 	.option('-l, --list-sheets', 'list sheet names and exit')
 	.option('-o, --output <file>', 'output to specified file')
+
 	.option('-S, --formulae', 'print formulae')
 	.option('-j, --json', 'emit formatted JSON (all fields text)')
 	.option('-J, --raw-js', 'emit raw JS object (raw numbers)')
 	.option('-A, --arrays', 'emit rows as JS objects (raw numbers)')
+	.option('-D, --dif', 'emit data interchange format (dif)')
+	.option('-K, --sylk', 'emit symbolic link (sylk)')
+	.option('-E, --socialcalc', 'emit socialcalc')
+
 	.option('-F, --field-sep <sep>', 'CSV field separator', ",")
 	.option('-R, --row-sep <sep>', 'CSV row separator', "\n")
 	.option('-n, --sheet-rows <num>', 'Number of rows to process (0=all rows)')
@@ -55,8 +60,7 @@ if(!fs.existsSync(filename)) {
 var opts = {}, wb/*:?Workbook*/;
 if(program.listSheets) opts.bookSheets = true;
 if(program.sheetRows) opts.sheetRows = program.sheetRows;
-if(program.password) opts.password = program.password;
-if(program.formulae) opts.cellFormula = true;
+else if(program.formulae) opts.cellFormula = true;
 else opts.cellFormula = false;
 
 if(program.all) {
@@ -81,12 +85,16 @@ if(program.read) process.exit(0);
 
 /*::   if(wb) { */
 if(program.listSheets) {
-	console.log(wb.SheetNames.join("\n"));
+	console.log((wb.SheetNames||[]).join("\n"));
 	process.exit(0);
 }
 
+
 var target_sheet = sheetname || '';
-if(target_sheet === '') target_sheet = wb.SheetNames[0];
+if(target_sheet === '') {
+	if(program.sheetIndex < (wb.SheetNames||[]).length) target_sheet = wb.SheetNames[program.sheetIndex];
+	else target_sheet = (wb.SheetNames||[""])[0];
+}
 
 var ws;
 try {
@@ -106,10 +114,13 @@ function get_xlsx() {
 
 var oo = "";
 if(!program.quiet) console.error(target_sheet);
-if(program.formulae) oo = X.utils.get_formulae(ws).join("\n");
+if(program.formulae) oo = get_xlsx().utils.get_formulae(ws).join("\n");
 else if(program.json) oo = JSON.stringify(get_xlsx().utils.sheet_to_row_object_array(ws));
 else if(program.rawJs) oo = JSON.stringify(get_xlsx().utils.sheet_to_row_object_array(ws,{raw:true}));
 else if(program.arrays) oo = JSON.stringify(get_xlsx().utils.sheet_to_row_object_array(ws,{raw:true, header:1}));
+else if(program.dif) oo = X.utils.sheet_to_dif(ws);
+else if(program.sylk) oo = X.utils.sheet_to_sylk(ws);
+else if(program.socialcalc) oo = X.utils.sheet_to_socialcalc(ws);
 else oo = get_xlsx().utils.make_csv(ws, {FS:program.fieldSep, RS:program.rowSep});
 
 if(program.output) fs.writeFileSync(program.output, oo);

bits/01_version.js

@@ -1 +1 @@
-HARB.version = '0.1.0';
+HARB.version = '0.1.1';

bits/22_helpers.js

@@ -17,5 +17,9 @@ function numdate(v/*:number*/)/*:Date*/ {
 	return val;
 }
 
-var sheet_to_workbook = function (sheet/*:Worksheet*/)/*:Workbook*/ { return {SheetNames: ['Sheet1'], Sheets: {Sheet1: sheet}};	};
+function sheet_to_workbook(sheet/*:Worksheet*/, opts)/*:Workbook*/ {
+	var n = opts && opts.sheet ? opts.sheet : "Sheet1";
+	var sheets = {}; sheets[n] = sheet;
+	return { SheetNames: [n], Sheets: sheets };
+}
 

bits/30_aoatos.js

@@ -4,18 +4,20 @@ function aoa_to_sheet(data/*:AOA*/, opts)/*:Worksheet*/ {
 	var range/*:Range*/ = ({s: {c:10000000, r:10000000}, e: {c:0, r:0 }}/*:any*/);
 	for(var R = 0; R != data.length; ++R) {
 		for(var C = 0; C != data[R].length; ++C) {
+			if(data[R][C] == null) continue;
+			var cell/*:Cell*/ = ({v: data[R][C] }/*:any*/);
 			if(range.s.r > R) range.s.r = R;
 			if(range.s.c > C) range.s.c = C;
 			if(range.e.r < R) range.e.r = R;
 			if(range.e.c < C) range.e.c = C;
-			var cell/*:Cell*/ = ({v: data[R][C] }/*:any*/);
-			if(cell.v == null) continue;
 			var cell_ref = encode_cell(({c:C,r:R}/*:any*/));
 			if(typeof cell.v === 'number') cell.t = 'n';
 			else if(typeof cell.v === 'boolean') cell.t = 'b';
 			else if(cell.v instanceof Date) {
-				cell.t = 'n'; cell.z = o.dateNF || SSF._table[14];
+				cell.t = 'n';
+				cell.z = o.dateNF || SSF._table[14];
 				cell.v = datenum(cell.v);
+				cell.w = SSF.format(cell.z, cell.v);
 			}
 			else cell.t = 's';
 			ws[cell_ref] = cell;

bits/50_csv.js

@@ -2,7 +2,7 @@ var bpopts = {
 	dynamicTyping: true,
 	keepEmptyRows: true
 };
-var csv_to_aoa = function (str) {
+function csv_to_aoa(str/*:string*/, opts)/*:AOA*/ {
 	var data = babyParse.parse(str, bpopts).data;
 	for(var R=0; R != data.length; ++R) for(var C=0; C != data[R].length; ++C) {
 		var d = data[R][C];
@@ -11,9 +11,9 @@ var csv_to_aoa = function (str) {
 		else if(d === 'FALSE') data[R][C] = false;
 	}
 	return data;
-};
+}
 
-var csv_to_sheet = function (str) { return aoa_to_sheet(csv_to_aoa(str)); };
+function csv_to_sheet(str/*:string*/, opts)/*:Worksheet*/ { return aoa_to_sheet(csv_to_aoa(str, opts), opts); }
 
-var csv_to_workbook = function (str) { return sheet_to_workbook(csv_to_sheet(str)); };
+function csv_to_workbook(str/*:string*/, opts)/*:Workbook*/ { return sheet_to_workbook(csv_to_sheet(str, opts), opts); }
 

bits/55_prn.js

@@ -1,12 +1,12 @@
-function set_text_arr(data/*:string*/, arr/*:Array<Array<any> >*/, R/*:number*/, C/*:number*/) {
+function set_text_arr(data/*:string*/, arr/*:AOA*/, R/*:number*/, C/*:number*/) {
 	if(data === 'TRUE') arr[R][C] = true;
 	else if(data === 'FALSE') arr[R][C] = false;
 	else if(+data == +data) arr[R][C] = +data;
 	else if(data !== "") arr[R][C] = data;
 }
 
-var prn_to_aoa = function(f) {
-	var arr = [];
+function prn_to_aoa(f/*:string*/, opts)/*:AOA*/ {
+	var arr/*:AOA*/ = ([]/*:any*/);
 	if(!f || f.length === 0) return arr;
 	var lines = f.split(/[\r\n]/);
 	var L = lines.length - 1;
@@ -27,9 +27,9 @@ var prn_to_aoa = function(f) {
 			set_text_arr(lines[R].slice(start+(C-1)*10,start+C*10).trim(),arr,R,C);
 	}
 	return arr;
-};
+}
 
-var prn_to_sheet = function (str) { return aoa_to_sheet(prn_to_aoa(str)); };
+function prn_to_sheet(str/*:string*/, opts)/*:Worksheet*/ { return aoa_to_sheet(prn_to_aoa(str, opts), opts); }
 
-var prn_to_workbook = function (str) { return sheet_to_workbook(prn_to_sheet(str)); };
+function prn_to_workbook(str/*:string*/, opts)/*:Workbook*/ { return sheet_to_workbook(prn_to_sheet(str, opts), opts); }
 

bits/60_dif.js

@@ -1,4 +1,4 @@
-var dif_to_aoa = function (str) {
+function dif_to_aoa(str/*:string*/, opts)/*:AOA*/ {
 	var records = str.split('\n'), R = -1, C = -1, ri = 0, arr = [];
 	for (; ri !== records.length; ++ri) {
 		if (records[ri].trim() === 'BOT') { arr[++R] = []; C = 0; continue; }
@@ -16,6 +16,7 @@ var dif_to_aoa = function (str) {
 				if(data === 'TRUE') arr[R][C] = true;
 				else if(data === 'FALSE') arr[R][C] = false;
 				else if(+value == +value) arr[R][C] = +value;
+				else if(!isNaN(new Date(value).getDate())) arr[R][C] = new Date(value);
 				else arr[R][C] = value;
 				++C; break;
 			case 1:
@@ -26,9 +27,48 @@ var dif_to_aoa = function (str) {
 		if (data === 'EOD') break;
 	}
 	return arr;
-};
+}
 
-var dif_to_sheet = function (str) { return aoa_to_sheet(dif_to_aoa(str)); };
+function dif_to_sheet(str/*:string*/, opts)/*:Worksheet*/ { return aoa_to_sheet(dif_to_aoa(str, opts), opts); }
 
-var dif_to_workbook = function (str) { return sheet_to_workbook(dif_to_sheet(str)); };
+function dif_to_workbook(str/*:string*/, opts)/*:Workbook*/ { return sheet_to_workbook(dif_to_sheet(str, opts), opts); }
 
+
+
+var sheet_to_dif = (function() {
+	var push_field = function pf(o/*:Array<string>*/, topic/*:string*/, v/*:number*/, n/*:number*/, s/*:string*/) {
+		o.push(topic);
+		o.push(v + "," + n);
+		o.push('"' + s.replace(/"/g,'""') + '"');
+	};
+	var push_value = function po(o/*:Array<string>*/, type/*:number*/, v/*:number*/, s/*:string*/) {
+		o.push(type + "," + v);
+		o.push(type == 1 ? '"' + s.replace(/"/g,'""') + '"' : s);
+	};
+	return function sheet_to_dif(ws/*:Worksheet*/, opts/*:?any*/)/*:string*/ {
+		var o/*:Array<string>*/ = [];
+		var r = decode_range(ws['!ref']), cell/*:Cell*/;
+		push_field(o, "TABLE", 0, 1, "sheetjs");
+		push_field(o, "VECTORS", 0, r.e.r - r.s.r + 1,"");
+		push_field(o, "TUPLES", 0, r.e.c - r.s.c + 1,"");
+		push_field(o, "DATA", 0, 0,"");
+		for(var R = r.s.r; R <= r.e.r; ++R) {
+			push_value(o, -1, 0, "BOT");
+			for(var C = r.s.c; C <= r.e.c; ++C) {
+				var coord = encode_cell({r:R,c:C});
+				if(!(cell = ws[coord]) || cell.v == null) { push_value(o, 1, 0, ""); continue;}
+				switch(cell.t) {
+					case 'n': push_value(o, 0, (cell.w || cell.v), "V"); break;
+					case 'b': push_value(o, 0, cell.v ? 1 : 0, cell.v ? "TRUE" : "FALSE"); break;
+					case 's': push_value(o, 1, 0, cell.v); break;
+					default: push_value(o, 1, 0, "");
+				}
+			}
+		}
+		push_value(o, -1, 0, "EOD");
+		var RS = "\r\n";
+		var oo = o.join(RS);
+		//while((oo.length & 0x7F) != 0) oo += "\0";
+		return oo;
+	};
+})();

bits/65_slk.js

@@ -1,5 +1,5 @@
 /* TODO: find an actual specification */
-var sylk_to_aoa = function(str) {
+function sylk_to_aoa(str/*:string*/, opts)/*:AOA*/ {
 	var records = str.split(/[\n\r]+/), R = -1, C = -1, ri = 0, rj = 0, arr = [];
 	var formats = [];
 	var next_cell_format = null;
@@ -36,9 +36,36 @@ var sylk_to_aoa = function(str) {
 		}
 	}
 	return arr;
-};
+}
 
-var sylk_to_sheet = function(str) { return aoa_to_sheet(sylk_to_aoa(str)); };
+function sylk_to_sheet(str/*:string*/, opts)/*:Worksheet*/ { return aoa_to_sheet(sylk_to_aoa(str, opts), opts); }
 
-var sylk_to_workbook = function (str) { return sheet_to_workbook(sylk_to_sheet(str)); };
+function sylk_to_workbook(str/*:string*/, opts)/*:Workbook*/ { return sheet_to_workbook(sylk_to_sheet(str, opts), opts); }
 
+function write_ws_cell_sylk(cell/*:Cell*/, ws/*:Worksheet*/, R/*:number*/, C/*:number*/, opts)/*:string*/ {
+	var o = "C;Y" + (R+1) + ";X" + (C+1) + ";K";
+	switch(cell.t) {
+		case 'n': o += cell.v; break;
+		case 'b': o += cell.v ? "TRUE" : "FALSE"; break;
+		case 'e': o += cell.w || cell.v; break;
+		case 'd': o += '"' + (cell.w || cell.v) + '"'; break;
+		case 's': o += '"' + cell.v.replace(/"/g,"") + '"'; break;
+	}
+	return o;
+}
+
+function sheet_to_sylk(ws/*:Worksheet*/, opts/*:?any*/)/*:string*/ {
+	var preamble/*:Array<string>*/ = ["ID;PWXL;N;E"], o/*:Array<string>*/ = [];
+	preamble.push("P;PGeneral");
+	var r = decode_range(ws['!ref']), cell/*:Cell*/;
+	for(var R = r.s.r; R <= r.e.r; ++R) {
+		for(var C = r.s.c; C <= r.e.c; ++C) {
+			var coord = encode_cell({r:R,c:C});
+			if(!(cell = ws[coord]) || cell.v == null) continue;
+			o.push(write_ws_cell_sylk(cell, ws, R, C, opts));
+		}
+	}
+	preamble.push("F;P0;DG0G8;M255");
+	var RS = "\r\n";
+	return preamble.join(RS) + RS + o.join(RS) + RS + "E" + RS;
+}