xlsx upate from steve.

Reference: /n/atom/patch/applied/xlsxupdate
Date: Tue Mar 25 17:17:07 CET 2014
Signed-off-by: quanstro@quanstro.net

--- /sys/src/cmd/opc/xlsx/excel2txt.c	Tue Mar 25 17:16:55 2014
+++ /sys/src/cmd/opc/xlsx/excel2txt.c	Tue Mar 25 17:16:55 2014
@@ -1,3 +1,9 @@
+/*
+ * Bugs: we somtimes see zero width columns which we take to be as wide as you like.
+ * we should have an auto mode to ignore what the file says, do 2 passes, measure
+ * the width of the columns and layout accordingly.
+ */
+
 #include <u.h>
 #include <libc.h>
 #include <bio.h>
@@ -7,7 +13,6 @@
 
 enum { Widefield = 40 };		/* wrap fields longer than this in tbl mode */
 
-char *Currency = "£";			/* currency symbol */
 int Epoch1904 = 1;				/* disable "as broken as Lotus-123" mode (yes really) */
 
 static char *Strtype[] = { "numeric", "inline", "shared", "boolean", "string", "error", "date" };
@@ -31,7 +36,7 @@
 	l = strlen(str);
 	if(Trunc && l > *remainp -1){
 		if(Doquote)
-			Bprint(bp, "%.*q…", *remainp -2, str);
+			Bprint(bp, "%.*q…", *remainp -4, str);
 		else
 			Bprint(bp, "%.*s…", *remainp -2, str);
 	}
@@ -121,8 +126,8 @@
 static int
 cellwidth(Elem *ep, int type, int style)
 {
-	int width, first, id;
-	char *fmt, buf[1024];
+	int width, first, numfmt;
+	char buf[1024];
 
 	width = 0;
 	first = 1;
@@ -141,12 +146,8 @@
 				break;
 			case Numeric:
 			case Date:
-				id = style2numid(style);
-				if(fmtnum(buf, sizeof(buf), id, ep->pcdata, type) == -1){
-					fmt = numid2fmtstr(id);
-					fprint(2, "%s: %d '%s' numfmt unknown\n", argv0, id, fmt);
-					strcpy(buf, "unknon format");
-				}
+				numfmt = style2numfmt(style);
+				fmtnum(buf, sizeof(buf), numfmt, ep->pcdata, type);
 				width += strlen(buf);
 				break;
 			case String:
@@ -171,15 +172,17 @@
 colwidth(int col)
 {
 	if(col < 0 || col >= Ncols)
-		return Defwidth;			/* default width */
+		return Defwidth;
+	if(Colwidth[col -1] < 1)
+		return Defwidth;
 	return Colwidth[col -1];		/* -1 as column indices start at 1 */
 }
 
 static int
 rd_c(Biobuf *bp, Elem *ep, int type, int style, int col)
 {
-	int id, first, remain, strwid, colwid;
-	char *fmt, buf[1024];
+	int remain, strwid, colwid, numfmt;
+	char buf[1024];
 
 	colwid = colwidth(col) -1;			/* -1 to ensures there a space between columns */
 	remain = colwid;
@@ -191,12 +194,7 @@
 	if(Tbl && strwid > Widefield)
 		Bprint(bp, "T{\n");
 
-
-	first = 1;
 	for(; ep; ep = ep->next){
-		if(! first)
-			Bprint(bp, " ");
-		first = 0;
 
 		if(strcmp(ep->name, "is") == 0 && type == Inline && ep->child)
 			rd_inlinestr(bp, ep->child, &remain);
@@ -208,12 +206,8 @@
 				break;
 			case Numeric:
 			case Date:
-				id = style2numid(style);
-				if(fmtnum(buf, sizeof(buf), id, ep->pcdata, type) < 0){
-					fmt = numid2fmtstr(id);
-					fprint(2, "%s: %d '%s' numfmt unknown\n", argv0, id, fmt);
-					strcpy(buf, "unknon format");
-				}
+				numfmt = style2numfmt(style);
+				fmtnum(buf, sizeof(buf), numfmt, ep->pcdata, type);
 				prnt(bp, buf, &remain);
 				break;
 			case String:
@@ -414,7 +408,6 @@
 	fprint(2, "  -s n       select sheet number to print\n");
 	fprint(2, "  -t         truncate long cells to column width\n");
 	fprint(2, "  -T         generate tbl(1) input\n");
-	fprint(2, "  -C x       set currency symbol to x\n");
 	exits("usage");
 }
 
@@ -462,9 +455,6 @@
 		Tbl = 1;
 		Delim = "\t";
 		break;
-	case 'C':
-		Currency = EARGF(usage());
-		break;
 	default:
 		usage();
 	}ARGEND;
@@ -490,6 +480,7 @@
 		if(dmpsty)
 			dumpstyles();
 	}
+
 	if((xp = parsefile("%s/xl/workbook.xml", argv[0])) != nil){
 		if((ep = xmllook(xp->root, "/workbook/workbookPr", nil, nil)) != nil)
 			if((v = xmlvalue(ep, "date1904")) != nil)
--- /sys/src/cmd/opc/xlsx/fmtnum.c	Tue Mar 25 17:16:55 2014
+++ /sys/src/cmd/opc/xlsx/fmtnum.c	Tue Mar 25 17:16:55 2014
@@ -4,9 +4,6 @@
 #include <xml.h>
 #include "xlsx.h"
 
-/* for number format syntax see http://www.ozgrid.com/Excel/CustomFormats.htm */
-
-
 static char *Months[] = {
 	"Jan", "Feb", "Mar", "Apr", "May", "Jun",
 	"Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
@@ -137,8 +134,8 @@
 	return localtime((long)t);
 }
 
-int
-fmtnum(char *buf, int len, int id, char *str, int type)
+void
+fmtnum(char *buf, int len, int numfmt, char *str, int type)
 {
 	Tm *tm;
 	double num, err;
@@ -149,8 +146,10 @@
 		tm = isotime(str);
 	else
 		tm = exceltime(num);
+	if(tm == nil)
+		return;
 
-	switch(id){
+	switch(numfmt){
 	case 0:	 	// General
 		snprint(buf, len, "%g", num);
 		break;
@@ -161,19 +160,19 @@
 		snprint(buf, len, "%4.2f", num);
 		break;
 	case 3:	 	// #,##0
-		snprint(buf, len, "%.0f", num);
+		snprint(buf, len, "%g", num);
 		break;
 	case 4:	 	// #,##0.00
-		snprint(buf, len, "%.2f", num);
+		snprint(buf, len, "%g", num);
 		break;
 	case 9:	 	// 0%
-		snprint(buf, len, "%.0f%%", num * 100.0);
+		snprint(buf, len, "%.0f%%", num);
 		break;
 	case 10:	 // 0.00%
-		snprint(buf, len, "%6.2f%%", num * 100.0);
+		snprint(buf, len, "%6.2f%%", num);
 		break;
 	case 11:	// 0.00E+00
-		snprint(buf, len, "%.2e", num);
+		snprint(buf, len, "%e", num);
 		break;
 	case 12:	// # ?/?
 		i = (int)num; num -= i;
@@ -218,12 +217,12 @@
 		snprint(buf, len, "%d:%02d", tm->hour, tm->min);
 		break;
 	case 21:	// h:mm:ss
-		snprint(buf, len, "%d:%02d:%02d", tm->hour, tm->min, tm->sec);
+		snprint(buf, len, "%d:%02d%02d", tm->hour, tm->min, tm->sec);
 		break;
 	case 22:	// m/d/yy h:mm
 		snprint(buf, len, "%d/%s/%02d %d:%02d", tm->mday, Months[tm->mon], tm->year % 100, tm->hour, tm->min);
 		/*
-		 * We don't use the "proper" form as it is Locale specific and we want to sidestep that horror.
+		 * We don't use this form as it is Locale specific and we want to sidestep that horror.
 		 *		snprint(buf, len, "%d/%d/%02d %d:%02d", tm->mon+1, tm->mday, tm->year % 100, tm->hour, tm->min);
 		 */
 		break;
@@ -231,17 +230,17 @@
 		if(num == 0)
 			snprint(buf, len, "0");
 		else if (num < 0)
-			snprint(buf, len, "(%.0f)", num);
+			snprint(buf, len, "(%f)", num);
 		else
-			snprint(buf, len, "%.0f", num);
+			snprint(buf, len, "%f", num);
 		break;
 	case 38:	// #,##0 ;[Red](#,##0)
 		if(num == 0)
 			snprint(buf, len, "0");
 		else if (num < 0)
-			snprint(buf, len, "(%.0f)", num);
+			snprint(buf, len, "(%f)", num);
 		else
-			snprint(buf, len, "%.0f", num);
+			snprint(buf, len, "%f", num);
 		break;
 	case 39:	// #,##0.00;(#,##0.00)
 		if(num == 0)
@@ -255,19 +254,10 @@
 		if(num == 0)
 			snprint(buf, len, "0");
 		else if (num < 0)
-			snprint(buf, len, "(%.2f)", num);
-		else
-			snprint(buf, len, "%.2f", num);
-		break;
-	case 44:	// _("$"* #,##0.00_);_("$"* \(#,##0.00\);_("$"* "-"??_);_(@_)  <Not published>
-		if(num == 0)
-			snprint(buf, len, "%s 0", Currency);
-		else if (num < 0)
-			snprint(buf, len, "(%s %.2f)", Currency, num);
+			snprint(buf, len, "(%.0f)", num);
 		else
-			snprint(buf, len, "%s %.2f", Currency, num);
+			snprint(buf, len, "%.f", num);
 		break;
-
 	case 45:	// mm:ss
 		snprint(buf, len, "%02d:%02d", tm->min, tm->sec);
 		break;
@@ -284,37 +274,19 @@
 		h = num / 60; num -= h * 60;
 		m = num / 60; num -= m * 60;
 		s = num / 60;
-			snprint(buf, len, "%02d:%02d.0", m, s);
+		if(h)
+			snprint(buf, len, "%d:%02d:%02d.0", h, m, s);
+		else
+			snprint(buf, len, ":%02d:%02d.0", m, s);
 		break;
 	case 48:	// ##0.0
-		snprint(buf, len, "%.1f", num);
+		snprint(buf, len, "%f", num);
 		break;
 	case 49:	// @
 		snprint(buf, len, "%s", str);
 		break;
-
-#ifdef Coraid
-	/*
-	 * special custom numfmts for coraid's Spreadsheets,
-	 * these will almost definitely be very, very wrong
-	 * for anyone else, you have been warned.
-	 */
-	case 164:	// "$"#,##0.00
-		snprint(buf, len, "$ %.2f", num);
-		break;
-	case 165:	// "$"#,##0.0000
-		snprint(buf, len, "$ %.4f", num);
-		break;
-	case 166:	// "$"#,##0.0000_);[Red]\("$"#,##0.0000\)
-		if(num >= 0)
-			snprint(buf, len, "$ %.4f", num);
-		else
-			snprint(buf, len, "($ %.4f)", num);
-		break;
-#endif
-
 	default:
-		return -1;
+		snprint(buf, len, "'%s' '%d' <unknown numfmt>", str, numfmt);
+		break;
 	}
-	return 0;
 }
--- /sys/src/cmd/opc/xlsx/styles.c	Tue Mar 25 17:16:55 2014
+++ /sys/src/cmd/opc/xlsx/styles.c	Tue Mar 25 17:16:55 2014
@@ -6,33 +6,25 @@
 
 typedef struct Numfmt Numfmt;
 struct Numfmt {
+	int num;
 	int style;
-	int id;
+	char *fmt;
 	Numfmt *left;
 	Numfmt *right;
 };
 
-typedef struct Fmtstr Fmtstr;
-struct Fmtstr {
-	int id;
-	char *fmt;
-	Fmtstr *left;
-	Fmtstr *right;
-};
-
-static Numfmt *Numroot;
-static Fmtstr *Fmtroot;
+static Numfmt *Root;
 
 static char *
-lookfmtstr(Fmtstr *fs, int id)
+lookfmtstr(Numfmt *nf, int style)
 {
-	if(fs == nil)
+	if(nf == nil)
 		return nil;
-	if(fs->id < id)
-		return lookfmtstr(fs->left, id);
-	if(fs->id > id)
-		return lookfmtstr(fs->right, id);
-	return fs->fmt;
+	if(nf->style < style)
+		return lookfmtstr(nf->left, style);
+	if(nf->style > style)
+		return lookfmtstr(nf->right, style);
+	return nf->fmt;
 }
 
 static int
@@ -44,108 +36,104 @@
 		return looknumfmt(nf->left, style);
 	if(nf->style > style)
 		return looknumfmt(nf->right, style);
-	return nf->id;
+	return nf->num;
 }
 
 int
-style2numid(int style)
+style2numfmt(int style)
 {
-	return looknumfmt(Numroot, style);
+	return looknumfmt(Root, style);
 }
 
 char *
-numid2fmtstr(int id)
+style2fmtstr(int style)
 {
-	return lookfmtstr(Fmtroot, id);
-}
+	char *s;
+	static char buf[16];
 
+	if((s = lookfmtstr(Root, style)) == nil){
+		snprint(buf, sizeof(buf), "<%d>", style);
+		return buf;
+	}
+	return s;
+}
 
 
 static Numfmt *
-addnum(Numfmt *nf, int style, int id)
+add(Numfmt *nf, int style, int num)
 {
 	if(nf == nil){
 		nf = malloc(sizeof(Numfmt));
 		if(nf == nil)
-			sysfatal("No memory for Numfmt\n");
+			sysfatal("no memory for Numfmt\n");
 		nf->style = style;
-		nf->id = id;
+		nf->num = num;
+		nf->fmt = nil;
 		nf->left = nf->right = nil;
 		return nf;
 	}
 
 	if(nf->style < style)
-		nf->left = addnum(nf->left, style, id);
+		nf->left = add(nf->left, style, num);
 	if(nf->style > style)
-		nf->right = addnum(nf->right, style, id);
+		nf->right = add(nf->right, style, num);
 	return nf;
 }
 
-static Fmtstr *
-addfmt(Fmtstr *fs, int id, char *fmt)
+static Numfmt *
+upd(Numfmt *nf, int num, char *fmt)
 {
 
-	if(fs == nil){
-		fs = malloc(sizeof(Fmtstr));
-		if(fs == nil)
-			sysfatal("No memory for Fmtstr\n");
-		fs->id = id;
-		fs->fmt = strdup(fmt);
-		if(fs->fmt == nil)
-			sysfatal("No memory for fmt\n");
-		fs->left = fs->right = nil;
-		return fs;
+	if(nf == nil)
+		sysfatal("numfmt=%d not known, bad style format\n", num);
+	if(nf->num == num){
+		nf->fmt = strdup(fmt);
+		if(nf->fmt == nil)
+			sysfatal("No memory for numfmt\n");
 	}
-	if(fs->id < id)
-		fs->left = addfmt(fs->left, id, fmt);
 	else
-	if(fs->id > id)
-		fs->right = addfmt(fs->right, id, fmt);
-	return fs;
+	if(nf->num < num)
+		nf->left = upd(nf->left, num, fmt);
+	else
+	if(nf->num > num)
+		nf->right = upd(nf->right, num, fmt);
+	return nf;
 }
 
 static void
 dumpsty(Numfmt *nf)
 {
-	char *fmt;
-
 	if(nf == nil)
 		return;
 	dumpsty(nf->right);
-	if((fmt = lookfmtstr(Fmtroot, nf->id)) == nil)
-		fmt = "<none>";
-	fprint(2, "%-6d %-6d %q\n", nf->style, nf->id, fmt);
+	fprint(2, "%-6d %-6d %q\n", nf->num, nf->style, nf->fmt);
 	dumpsty(nf->left);
 }
 
 void
 dumpstyles(void)
 {
-	fprint(2, "%-6s %-6s %q\n", "styleid", "numid", "fmtstr");
-	dumpsty(Numroot);
+	dumpsty(Root);
 }
 
 static void
 numfmts(Elem *ep)
 {
-	int id;
+	int num;
 	char *fmt, *v;
 
-	id = -1;
+	num = -1;
 	fmt = nil;
 	for(; ep; ep = ep->next){
-		if(strcmp(ep->name, "numFmt") == 0){
+		if(strcmp(ep->name, "numFmt") == 0)
 			if((v = xmlvalue(ep, "numFmtId")) != nil)
-				id = atoi(v);
+				num = atoi(v);
+		if(strcmp(ep->name, "numFmtId") == 0)
 			if((v = xmlvalue(ep, "formatCode")) != nil)
 				fmt = v;
-		}
-		if(id < 164)	/* seems these are builtin to excel */
-			continue;
-
-		if(id != -1 && fmt != nil)
-			Fmtroot = addfmt(Fmtroot, id, fmt);
-		id = -1;
+		if(num != -1 && fmt != nil)
+			Root = upd(Root, num, fmt);
+		num = -1;
 		fmt = nil;
 	}
 }
@@ -153,20 +141,19 @@
 static void
 cellxfs(Elem *ep)
 {
-	int enab, style;
+	int style;
 	char *v;
 
 	style = 0;
 	for(; ep; ep = ep->next)
 		if(strcmp(ep->name, "xf") == 0){
-			enab = 0;
-			if((v = xmlvalue(ep, "applyNumberFormat")) != nil)
-				enab = atoi(v);
-			if(enab && (v = xmlvalue(ep, "numFmtId")) != nil)
-				Numroot = addnum(Numroot, style, atoi(v));
+			if((v = xmlvalue(ep, "numFmtId")) != nil)
+				Root = add(Root, style, atoi(v));
 			style++;
 		}
 }
+
+
 
 void
 rd_styles(Elem *base)
--- /sys/src/cmd/opc/xlsx/xlsx.h	Tue Mar 25 17:16:55 2014
+++ /sys/src/cmd/opc/xlsx/xlsx.h	Tue Mar 25 17:16:55 2014
@@ -1,5 +1,4 @@
 extern int Epoch1904;		/* disable "as broken as Lotus-123" mode (yes really) */
-extern char *Currency;		/* Currency symbol */
 
 enum {
 	Numeric,				/* type of field */
@@ -12,7 +11,7 @@
 };
 
 /* fmtnum.c */
-int fmtnum(char *buf, int len, int id, char *str, int type);
+void fmtnum(char *buf, int len, int numfmt, char *str, int type);
 
 /* strings.c */
 char *lookstring(int idx);
@@ -26,7 +25,7 @@
 void dumpstrings(void);
 
 /* styles.c */
-int style2numid(int style);
-char *numid2fmtstr(int id);
-void dumpstyles(void);
+int style2numfmt(int style);
+char *style2fmtstr(int style);
 void rd_styles(Elem *base);
+void dumpstyles(void);
--- /sys/man/1/xlsx2txt	Tue Mar 25 17:16:56 2014
+++ /sys/man/1/xlsx2txt	Tue Mar 25 17:16:56 2014
@@ -9,8 +9,6 @@
 ] [
 .BI -c range
 ] [
-.BI -C symb
-] [
 .BI -d str
 ] [
 .B -q
@@ -61,9 +59,6 @@
 similar to standard plan9 character class specification. Any combination
 of comma seperated numbers and number ranges (two numbers
 seperated by a dash) may be specified.
-.IP "-C symb"
-.br
-Currency symbol for financial number formats, defaults to UK pounds (£).
 .IP "-d str"
 .br
 Set the field delimiter to the string