DBF write encoding (fixes #2781 h/t @ZJS248)

This commit is contained in:
SheetJS 2022-09-06 00:01:57 -04:00
parent df0e7b5f25
commit 26cbfe37be
3 changed files with 20 additions and 8 deletions

@ -181,21 +181,29 @@ function WriteShift(t/*:number*/, val/*:string|number*/, f/*:?string*/)/*:any*/
/*:: if(typeof val !== 'string') throw new Error("unreachable"); */
for(i = 0; i != val.length; ++i) __writeUInt16LE(this, val.charCodeAt(i), this.l + 2 * i);
size = 2 * val.length;
} else if(f === 'sbcs') {
} else if(f === 'sbcs' || f == 'cpstr') {
if(typeof $cptable !== 'undefined' && current_ansi == 874) {
/* TODO: use tables directly, don't encode */
/*:: if(typeof val !== "string") throw new Error("unreachable"); */
for(i = 0; i != val.length; ++i) {
var cppayload = $cptable.utils.encode(current_ansi, val.charAt(i));
this[this.l + i] = cppayload[0];
var cpp = $cptable.utils.encode(current_ansi, val.charAt(i));
this[this.l + i] = cpp[0];
}
size = val.length;
} else if(typeof $cptable !== 'undefined' && f == 'cpstr') {
var cpp = $cptable.utils.encode(current_ansi, val);
/* replace null bytes with _ when relevant */
if(cpp.length == val.length) for(i = 0; i < val.length; ++i) if(cpp[i] == 0 && val.charCodeAt(i) != 0) cpp[i] = 0x5F;
if(cpp.length == 2 * val.length) for(i = 0; i < val.length; ++i) if(cpp[2*i] == 0 && cpp[2*i+1] == 0 && val.charCodeAt(i) != 0) cpp[2*i] = 0x5F;
for(i = 0; i < cpp.length; ++i) this[this.l + i] = cpp[i];
size = cpp.length;
} else {
/*:: if(typeof val !== 'string') throw new Error("unreachable"); */
val = val.replace(/[^\x00-\x7F]/g, "_");
/*:: if(typeof val !== 'string') throw new Error("unreachable"); */
for(i = 0; i != val.length; ++i) this[this.l + i] = (val.charCodeAt(i) & 0xFF);
size = val.length;
}
size = val.length;
} else if(f === 'hex') {
for(; i < t; ++i) {
/*:: if(typeof val !== "string") throw new Error("unreachable"); */

@ -282,7 +282,8 @@ function sheet_to_dbf(ws/*:Worksheet*/, opts/*:WriteOpts*/) {
case 'object': _guess = col[j] instanceof Date ? 'D' : 'C'; break;
default: _guess = 'C';
}
maxlen = Math.max(maxlen, String(col[j]).length);
/* TODO: cache the values instead of encoding twice */
maxlen = Math.max(maxlen, (typeof $cptable !== "undefined" && typeof col[j] == "string" ? $cptable.utils.encode(current_ansi, col[j]): String(col[j])).length);
guess = guess && guess != _guess ? 'C' : _guess;
//if(guess == 'C') break;
}
@ -352,9 +353,11 @@ function sheet_to_dbf(ws/*:Worksheet*/, opts/*:WriteOpts*/) {
rout.write_shift(2, ("00"+data[i][j].getDate()).slice(-2), "sbcs");
} break;
case 'C':
var _l = rout.l;
var _s = String(data[i][j] != null ? data[i][j] : "").slice(0, colwidths[j]);
rout.write_shift(1, _s, "sbcs");
for(hcnt=0; hcnt < colwidths[j]-_s.length; ++hcnt) rout.write_shift(1, 0x20); break;
rout.write_shift(1, _s, "cpstr");
_l += colwidths[j] - rout.l;
for(hcnt=0; hcnt < _l; ++hcnt) rout.write_shift(1, 0x20); break;
}
}
// data
@ -598,6 +601,7 @@ var SYLK = /*#__PURE__*/(function() {
}
function sheet_to_sylk(ws/*:Worksheet*/, opts/*:?any*/, wb/*:?WorkBook*/)/*:string*/ {
/* TODO: codepage */
var preamble/*:Array<string>*/ = ["ID;PSheetJS;N;E"], o/*:Array<string>*/ = [];
var r = safe_decode_range(ws['!ref']), cell/*:Cell*/;
var dense = Array.isArray(ws);

@ -55,7 +55,7 @@ function write_comments_xml(data/*::, opts*/) {
d[1].forEach(function(c) {
if(c.a) lastauthor = iauthor.indexOf(escapexml(c.a));
if(c.T) ++tcnt;
ts.push(writetag("t", c.t == null ? "" : escapexml(c.t)));
ts.push(c.t == null ? "" : escapexml(c.t));
});
if(tcnt === 0) {
d[1].forEach(function(c) {