forked from sheetjs/sheetjs
434 lines
14 KiB
JavaScript
434 lines
14 KiB
JavaScript
function parse_ws_xml_dim(ws, s) {
|
|
var d = safe_decode_range(s);
|
|
if (d.s.r <= d.e.r && d.s.c <= d.e.c && d.s.r >= 0 && d.s.c >= 0) ws["!ref"] = encode_range(d);
|
|
}
|
|
var mergecregex = /<mergeCell ref="[A-Z0-9:]+"\s*\/>/g;
|
|
var sheetdataregex = /<(?:\w+:)?sheetData>([^\u2603]*)<\/(?:\w+:)?sheetData>/;
|
|
var hlinkregex = /<hyperlink[^>]*\/>/g;
|
|
var dimregex = /"(\w*:\w*)"/;
|
|
var colregex = /<col[^>]*\/>/g;
|
|
/* 18.3 Worksheets */
|
|
function parse_ws_xml(data, opts, rels) {
|
|
if (!data) return data;
|
|
/* 18.3.1.99 worksheet CT_Worksheet */
|
|
var s = {};
|
|
|
|
/* 18.3.1.35 dimension CT_SheetDimension ? */
|
|
var ridx = data.indexOf("<dimension");
|
|
if (ridx > 0) {
|
|
var ref = data.substr(ridx, 50).match(dimregex);
|
|
if (ref != null) parse_ws_xml_dim(s, ref[1]);
|
|
}
|
|
|
|
/* 18.3.1.55 mergeCells CT_MergeCells */
|
|
var mergecells = [];
|
|
if (data.indexOf("</mergeCells>") !== -1) {
|
|
var merges = data.match(mergecregex);
|
|
for (ridx = 0; ridx != merges.length; ++ridx)
|
|
mergecells[ridx] = safe_decode_range(merges[ridx].substr(merges[ridx].indexOf("\"") + 1));
|
|
}
|
|
|
|
/* 18.3.1.17 cols CT_Cols */
|
|
var columns = [];
|
|
if (opts.cellStyles && data.indexOf("</cols>") !== -1) {
|
|
/* 18.3.1.13 col CT_Col */
|
|
var cols = data.match(colregex);
|
|
parse_ws_xml_cols(columns, cols);
|
|
}
|
|
|
|
var refguess = {s: {r: 1000000, c: 1000000}, e: {r: 0, c: 0}};
|
|
|
|
/* 18.3.1.80 sheetData CT_SheetData ? */
|
|
var mtch = data.match(sheetdataregex);
|
|
if (mtch) parse_ws_xml_data(mtch[1], s, opts, refguess);
|
|
|
|
/* 18.3.1.48 hyperlinks CT_Hyperlinks */
|
|
if (data.indexOf("</hyperlinks>") !== -1) parse_ws_xml_hlinks(s, data.match(hlinkregex), rels);
|
|
|
|
if (!s["!ref"] && refguess.e.c >= refguess.s.c && refguess.e.r >= refguess.s.r) s["!ref"] = encode_range(refguess);
|
|
if (opts.sheetRows > 0 && s["!ref"]) {
|
|
var tmpref = safe_decode_range(s["!ref"]);
|
|
if (opts.sheetRows < +tmpref.e.r) {
|
|
tmpref.e.r = opts.sheetRows - 1;
|
|
if (tmpref.e.r > refguess.e.r) tmpref.e.r = refguess.e.r;
|
|
if (tmpref.e.r < tmpref.s.r) tmpref.s.r = tmpref.e.r;
|
|
if (tmpref.e.c > refguess.e.c) tmpref.e.c = refguess.e.c;
|
|
if (tmpref.e.c < tmpref.s.c) tmpref.s.c = tmpref.e.c;
|
|
s["!fullref"] = s["!ref"];
|
|
s["!ref"] = encode_range(tmpref);
|
|
}
|
|
}
|
|
if (mergecells.length > 0) s["!merges"] = mergecells;
|
|
if (columns.length > 0) s["!cols"] = columns;
|
|
return s;
|
|
}
|
|
|
|
function write_ws_xml_merges(merges) {
|
|
if (merges.length == 0) return "";
|
|
var o = '<mergeCells count="' + merges.length + '">';
|
|
for (var i = 0; i != merges.length; ++i) o += '<mergeCell ref="' + encode_range(merges[i]) + '"/>';
|
|
return o + '</mergeCells>';
|
|
}
|
|
|
|
function write_ws_xml_pagesetup(setup) {
|
|
var pageSetup = writextag('pageSetup', null, {
|
|
scale: setup.scale || '100',
|
|
orientation: setup.orientation || 'portrait',
|
|
horizontalDpi: setup.horizontalDpi || '4294967292',
|
|
verticalDpi: setup.verticalDpi || '4294967292'
|
|
})
|
|
return pageSetup;
|
|
}
|
|
|
|
|
|
function parse_ws_xml_hlinks(s, data, rels) {
|
|
for (var i = 0; i != data.length; ++i) {
|
|
var val = parsexmltag(data[i], true);
|
|
if (!val.ref) return;
|
|
var rel = rels ? rels['!id'][val.id] : null;
|
|
if (rel) {
|
|
val.Target = rel.Target;
|
|
if (val.location) val.Target += "#" + val.location;
|
|
val.Rel = rel;
|
|
} else {
|
|
val.Target = val.location;
|
|
rel = {Target: val.location, TargetMode: 'Internal'};
|
|
val.Rel = rel;
|
|
}
|
|
var rng = safe_decode_range(val.ref);
|
|
for (var R = rng.s.r; R <= rng.e.r; ++R) for (var C = rng.s.c; C <= rng.e.c; ++C) {
|
|
var addr = encode_cell({c: C, r: R});
|
|
if (!s[addr]) s[addr] = {t: "stub", v: undefined};
|
|
s[addr].l = val;
|
|
}
|
|
}
|
|
}
|
|
|
|
function parse_ws_xml_cols(columns, cols) {
|
|
var seencol = false;
|
|
for (var coli = 0; coli != cols.length; ++coli) {
|
|
var coll = parsexmltag(cols[coli], true);
|
|
var colm = parseInt(coll.min, 10) - 1, colM = parseInt(coll.max, 10) - 1;
|
|
delete coll.min;
|
|
delete 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;
|
|
}
|
|
}
|
|
|
|
function write_ws_xml_cols(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[o.length] = (writextag('col', null, p));
|
|
}
|
|
o[o.length] = "</cols>";
|
|
return o.join("");
|
|
}
|
|
|
|
function write_ws_xml_cell(cell, ref, ws, opts, idx, wb) {
|
|
if (cell.v === undefined && cell.s === undefined) return "";
|
|
var vv = "";
|
|
var oldt = cell.t, oldv = cell.v;
|
|
switch (cell.t) {
|
|
case 'b':
|
|
vv = cell.v ? "1" : "0";
|
|
break;
|
|
case 'n':
|
|
vv = '' + cell.v;
|
|
break;
|
|
case 'e':
|
|
vv = BErr[cell.v];
|
|
break;
|
|
case 'd':
|
|
if (opts.cellDates) vv = new Date(cell.v).toISOString();
|
|
else {
|
|
cell.t = 'n';
|
|
vv = '' + (cell.v = datenum(cell.v));
|
|
if (typeof cell.z === 'undefined') cell.z = SSF._table[14];
|
|
}
|
|
break;
|
|
default:
|
|
vv = cell.v;
|
|
break;
|
|
}
|
|
var v = writetag('v', escapexml(vv)), o = {r: ref};
|
|
/* TODO: cell style */
|
|
var os = get_cell_style(opts.cellXfs, cell, opts);
|
|
if (os !== 0) o.s = os;
|
|
switch (cell.t) {
|
|
case 'n':
|
|
break;
|
|
case 'd':
|
|
o.t = "d";
|
|
break;
|
|
case 'b':
|
|
o.t = "b";
|
|
break;
|
|
case 'e':
|
|
o.t = "e";
|
|
break;
|
|
default:
|
|
if (opts.bookSST) {
|
|
v = writetag('v', '' + get_sst_id(opts.Strings, cell.v));
|
|
o.t = "s";
|
|
break;
|
|
}
|
|
o.t = "str";
|
|
break;
|
|
}
|
|
if (cell.t != oldt) {
|
|
cell.t = oldt;
|
|
cell.v = oldv;
|
|
}
|
|
return writextag('c', v, o);
|
|
}
|
|
|
|
var parse_ws_xml_data = (function parse_ws_xml_data_factory() {
|
|
var cellregex = /<(?:\w+:)?c[ >]/, rowregex = /<\/(?:\w+:)?row>/;
|
|
var rregex = /r=["']([^"']*)["']/, isregex = /<is>([\S\s]*?)<\/is>/;
|
|
var match_v = matchtag("v"), match_f = matchtag("f");
|
|
|
|
return function parse_ws_xml_data(sdata, s, opts, guess) {
|
|
var ri = 0, x = "", cells = [], cref = [], idx = 0, i = 0, cc = 0, d = "", p;
|
|
var tag, tagr = 0, tagc = 0;
|
|
var sstr;
|
|
var fmtid = 0, fillid = 0, do_format = Array.isArray(styles.CellXf), cf;
|
|
for (var marr = sdata.split(rowregex), mt = 0, marrlen = marr.length; mt != marrlen; ++mt) {
|
|
x = marr[mt].trim();
|
|
var xlen = x.length;
|
|
if (xlen === 0) continue;
|
|
|
|
/* 18.3.1.73 row CT_Row */
|
|
for (ri = 0; ri < xlen; ++ri) if (x.charCodeAt(ri) === 62) break;
|
|
++ri;
|
|
tag = parsexmltag(x.substr(0, ri), true);
|
|
/* SpreadSheetGear uses implicit r/c */
|
|
tagr = typeof tag.r !== 'undefined' ? parseInt(tag.r, 10) : tagr + 1;
|
|
tagc = -1;
|
|
if (opts.sheetRows && opts.sheetRows < tagr) continue;
|
|
if (guess.s.r > tagr - 1) guess.s.r = tagr - 1;
|
|
if (guess.e.r < tagr - 1) guess.e.r = tagr - 1;
|
|
|
|
/* 18.3.1.4 c CT_Cell */
|
|
cells = x.substr(ri).split(cellregex);
|
|
for (ri = typeof tag.r === 'undefined' ? 0 : 1; ri != cells.length; ++ri) {
|
|
x = cells[ri].trim();
|
|
if (x.length === 0) continue;
|
|
cref = x.match(rregex);
|
|
idx = ri;
|
|
i = 0;
|
|
cc = 0;
|
|
x = "<c " + (x.substr(0, 1) == "<" ? ">" : "") + x;
|
|
if (cref !== null && cref.length === 2) {
|
|
idx = 0;
|
|
d = cref[1];
|
|
for (i = 0; i != d.length; ++i) {
|
|
if ((cc = d.charCodeAt(i) - 64) < 1 || cc > 26) break;
|
|
idx = 26 * idx + cc;
|
|
}
|
|
--idx;
|
|
tagc = idx;
|
|
} else ++tagc;
|
|
for (i = 0; i != x.length; ++i) if (x.charCodeAt(i) === 62) break;
|
|
++i;
|
|
tag = parsexmltag(x.substr(0, i), true);
|
|
if (!tag.r) tag.r = utils.encode_cell({r: tagr - 1, c: tagc});
|
|
d = x.substr(i);
|
|
p = {t: ""};
|
|
|
|
if ((cref = d.match(match_v)) !== null && cref[1] !== '') p.v = unescapexml(cref[1]);
|
|
if (opts.cellFormula && (cref = d.match(match_f)) !== null) p.f = unescapexml(cref[1]);
|
|
|
|
/* SCHEMA IS ACTUALLY INCORRECT HERE. IF A CELL HAS NO T, EMIT "" */
|
|
if (tag.t === undefined && tag.s === undefined && p.v === undefined) {
|
|
if (!opts.sheetStubs) continue;
|
|
p.t = "stub";
|
|
}
|
|
else p.t = tag.t || "n";
|
|
if (guess.s.c > idx) guess.s.c = idx;
|
|
if (guess.e.c < idx) guess.e.c = idx;
|
|
/* 18.18.11 t ST_CellType */
|
|
switch (p.t) {
|
|
case 'n':
|
|
p.v = parseFloat(p.v);
|
|
if (isNaN(p.v)) p.v = "" // we don't want NaN if p.v is null
|
|
break;
|
|
case 's':
|
|
// if (!p.hasOwnProperty('v')) continue;
|
|
sstr = strs[parseInt(p.v, 10)];
|
|
p.v = sstr.t;
|
|
p.r = sstr.r;
|
|
if (opts.cellHTML) p.h = sstr.h;
|
|
break;
|
|
case 'str':
|
|
p.t = "s";
|
|
p.v = (p.v != null) ? utf8read(p.v) : '';
|
|
if (opts.cellHTML) p.h = p.v;
|
|
break;
|
|
case 'inlineStr':
|
|
cref = d.match(isregex);
|
|
p.t = 's';
|
|
if (cref !== null) {
|
|
sstr = parse_si(cref[1]);
|
|
p.v = sstr.t;
|
|
} else p.v = "";
|
|
break; // inline string
|
|
case 'b':
|
|
p.v = parsexmlbool(p.v);
|
|
break;
|
|
case 'd':
|
|
if (!opts.cellDates) {
|
|
p.v = datenum(p.v);
|
|
p.t = 'n';
|
|
}
|
|
break;
|
|
/* error string in .v, number in .v */
|
|
case 'e':
|
|
p.w = p.v;
|
|
p.v = RBErr[p.v];
|
|
break;
|
|
}
|
|
/* formatting */
|
|
fmtid = fillid = 0;
|
|
if (do_format && tag.s !== undefined) {
|
|
cf = styles.CellXf[tag.s];
|
|
if (opts.cellStyles) {
|
|
p.s = get_cell_style_csf(cf)
|
|
}
|
|
if (cf != null) {
|
|
if (cf.numFmtId != null) fmtid = cf.numFmtId;
|
|
if (opts.cellStyles && cf.fillId != null) fillid = cf.fillId;
|
|
}
|
|
}
|
|
safe_format(p, fmtid, fillid, opts);
|
|
s[tag.r] = p;
|
|
}
|
|
}
|
|
};
|
|
})();
|
|
|
|
function write_ws_xml_data(ws, opts, idx, wb) {
|
|
var o = [], r = [], range = safe_decode_range(ws['!ref']), cell, ref, rr = "", cols = [], R, C;
|
|
for (C = range.s.c; C <= range.e.c; ++C) cols[C] = encode_col(C);
|
|
for (R = range.s.r; R <= range.e.r; ++R) {
|
|
r = [];
|
|
rr = encode_row(R);
|
|
for (C = range.s.c; C <= range.e.c; ++C) {
|
|
ref = cols[C] + rr;
|
|
if (ws[ref] === undefined) continue;
|
|
if ((cell = write_ws_xml_cell(ws[ref], ref, ws, opts, idx, wb)) != null) r.push(cell);
|
|
}
|
|
if (r.length > 0) o[o.length] = (writextag('row', r.join(""), {r: rr}));
|
|
}
|
|
return o.join("");
|
|
}
|
|
|
|
var WS_XML_ROOT = writextag('worksheet', null, {
|
|
'xmlns': XMLNS.main[0],
|
|
'xmlns:r': XMLNS.r
|
|
});
|
|
|
|
function write_ws_xml(idx, opts, wb) {
|
|
var o = [XML_HEADER, WS_XML_ROOT];
|
|
var s = wb.SheetNames[idx], sidx = 0, rdata = "";
|
|
var ws = wb.Sheets[s];
|
|
if (ws === undefined) ws = {};
|
|
var ref = ws['!ref'];
|
|
if (ref === undefined) ref = 'A1';
|
|
o[o.length] = (writextag('dimension', null, {'ref': ref}));
|
|
|
|
var kids = [];
|
|
if (ws['!freeze']) {
|
|
var pane = '';
|
|
pane = writextag('pane', null, ws['!freeze'])
|
|
kids.push(pane)
|
|
|
|
var selection = writextag('selection', null, {
|
|
pane: "topLeft"
|
|
})
|
|
kids.push(selection)
|
|
|
|
var selection = writextag('selection', null, {
|
|
pane: "bottomLeft"
|
|
})
|
|
kids.push(selection)
|
|
|
|
var selection = writextag('selection', null, {
|
|
pane: "bottomRight",
|
|
activeCell: ws['!freeze'],
|
|
sqref: ws['!freeze']
|
|
})
|
|
kids.push(selection)
|
|
}
|
|
|
|
|
|
//<selection pane="bottomRight" activeCell="A4" sqref="A4"/>
|
|
|
|
var sheetView = writextag('sheetView', kids.join('') || undefined, {
|
|
showGridLines: opts.showGridLines == false ? '0' : '1',
|
|
tabSelected: opts.tabSelected === undefined ? '0' : opts.tabSelected, // see issue #26, need to set WorkbookViews if this is set
|
|
workbookViewId: opts.workbookViewId === undefined ? '0' : opts.workbookViewId
|
|
});
|
|
o[o.length] = writextag('sheetViews', sheetView);
|
|
|
|
if (ws['!cols'] !== undefined && ws['!cols'].length > 0) o[o.length] = (write_ws_xml_cols(ws, ws['!cols']));
|
|
o[sidx = o.length] = '<sheetData/>';
|
|
if (ws['!ref'] !== undefined) {
|
|
rdata = write_ws_xml_data(ws, opts, idx, wb);
|
|
if (rdata.length > 0) o[o.length] = (rdata);
|
|
}
|
|
if (o.length > sidx + 1) {
|
|
o[o.length] = ('</sheetData>');
|
|
o[sidx] = o[sidx].replace("/>", ">");
|
|
}
|
|
|
|
if (ws['!merges'] !== undefined && ws['!merges'].length > 0) o[o.length] = (write_ws_xml_merges(ws['!merges']));
|
|
|
|
if (ws['!pageSetup'] !== undefined) o[o.length] = write_ws_xml_pagesetup(ws['!pageSetup']);
|
|
if (ws['!rowBreaks'] !== undefined) o[o.length] = write_ws_xml_row_breaks(ws['!rowBreaks']);
|
|
if (ws['!colBreaks'] !== undefined) o[o.length] = write_ws_xml_col_breaks(ws['!colBreaks']);
|
|
|
|
if (o.length > 2) {
|
|
o[o.length] = ('</worksheet>');
|
|
o[1] = o[1].replace("/>", ">");
|
|
}
|
|
return o.join("");
|
|
}
|
|
|
|
function write_ws_xml_row_breaks(breaks) {
|
|
var brk = [];
|
|
for (var i = 0; i < breaks.length; i++) {
|
|
var thisBreak = '' + (breaks[i]);
|
|
var nextBreak = '' + (breaks[i + 1] || '16383');
|
|
brk.push(writextag('brk', null, {id: thisBreak, max: nextBreak, man: '1'}))
|
|
}
|
|
return writextag('rowBreaks', brk.join(' '), {count: brk.length, manualBreakCount: brk.length})
|
|
}
|
|
function write_ws_xml_col_breaks(breaks) {
|
|
var brk = [];
|
|
for (var i = 0; i < breaks.length; i++) {
|
|
var thisBreak = '' + (breaks[i]);
|
|
var nextBreak = '' + (breaks[i + 1] || '1048575');
|
|
brk.push(writextag('brk', null, {id: thisBreak, max: nextBreak, man: '1'}))
|
|
}
|
|
return writextag('colBreaks', brk.join(' '), {count: brk.length, manualBreakCount: brk.length})
|
|
}
|