sheetjs/bits/22_xmlutils.js

290 lines
13 KiB
JavaScript
Raw Permalink Normal View History

var XML_HEADER = '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>\r\n';
var attregexg=/\s([^"\s?>\/]+)\s*=\s*((?:")([^"]*)(?:")|(?:')([^']*)(?:')|([^'">\s]+))/g;
var tagregex1=/<[\/\?]?[a-zA-Z0-9:_-]+(?:\s+[^"\s?<>\/]+\s*=\s*(?:"[^"]*"|'[^']*'|[^'"<>\s=]+))*\s*[\/\?]?>/mg, tagregex2 = /<[^<>]*>/g;
2022-03-20 01:54:41 +00:00
var tagregex = /*#__PURE__*/XML_HEADER.match(tagregex1) ? tagregex1 : tagregex2;
var nsregex=/<\w*:/, nsregex2 = /<(\/?)\w+:/;
function parsexmltag(tag/*:string*/, skip_root/*:?boolean*/, skip_LC/*:?boolean*/)/*:any*/ {
var z = ({}/*:any*/);
var eq = 0, c = 0;
for(; eq !== tag.length; ++eq) if((c = tag.charCodeAt(eq)) === 32 || c === 10 || c === 13) break;
if(!skip_root) z[0] = tag.slice(0, eq);
if(eq === tag.length) return z;
var m = tag.match(attregexg), j=0, v="", i=0, q="", cc="", quot = 1;
if(m) for(i = 0; i != m.length; ++i) {
cc = m[i].slice(1);
for(c=0; c != cc.length; ++c) if(cc.charCodeAt(c) === 61) break;
q = cc.slice(0,c).trim();
while(cc.charCodeAt(c+1) == 32) ++c;
quot = ((eq=cc.charCodeAt(c+1)) == 34 || eq == 39) ? 1 : 0;
v = cc.slice(c+1+quot, cc.length-quot);
for(j=0;j!=q.length;++j) if(q.charCodeAt(j) === 58) break;
if(j===q.length) {
if(q.indexOf("_") > 0) q = q.slice(0, q.indexOf("_")); // from ods
z[q] = v;
if(!skip_LC) z[q.toLowerCase()] = v;
}
else {
var k = (j===5 && q.slice(0,5)==="xmlns"?"xmlns":"")+q.slice(j+1);
if(z[k] && q.slice(j-3,j) == "ext") continue; // from ods
z[k] = v;
if(!skip_LC) z[k.toLowerCase()] = v;
}
}
return z;
}
function parsexmltagraw(tag/*:string*/, skip_root/*:?boolean*/, skip_LC/*:?boolean*/)/*:any*/ {
var z = ({}/*:any*/);
var eq = 0, c = 0;
for(; eq !== tag.length; ++eq) if((c = tag.charCodeAt(eq)) === 32 || c === 10 || c === 13) break;
if(!skip_root) z[0] = tag.slice(0, eq);
if(eq === tag.length) return z;
var m = tag.match(attregexg), j=0, v="", i=0, q="", cc="", quot = 1;
if(m) for(i = 0; i != m.length; ++i) {
cc = m[i].slice(1);
for(c=0; c != cc.length; ++c) if(cc.charCodeAt(c) === 61) break;
q = cc.slice(0,c).trim();
while(cc.charCodeAt(c+1) == 32) ++c;
quot = ((eq=cc.charCodeAt(c+1)) == 34 || eq == 39) ? 1 : 0;
v = cc.slice(c+1+quot, cc.length-quot);
if(q.indexOf("_") > 0) q = q.slice(0, q.indexOf("_")); // from ods
z[q] = v;
if(!skip_LC) z[q.toLowerCase()] = v;
}
return z;
}
function strip_ns(x/*:string*/)/*:string*/ { return x.replace(nsregex2, "<$1"); }
var encodings = {
'&quot;': '"',
'&apos;': "'",
'&gt;': '>',
'&lt;': '<',
'&amp;': '&'
};
2022-03-20 01:54:41 +00:00
var rencoding = /*#__PURE__*/evert(encodings);
//var rencstr = "&<>'\"".split("");
// TODO: CP remap (need to read file version to determine OS)
2022-03-12 14:05:57 +00:00
var unescapexml/*:StringConv*/ = /*#__PURE__*/(function() {
/* 22.4.2.4 bstr (Basic String) */
var encregex = /&(?:quot|apos|gt|lt|amp|#x?([\da-fA-F]+));/ig, coderegex = /_x([\da-fA-F]{4})_/g;
2022-04-20 17:31:11 +00:00
function raw_unescapexml(text/*:string*/)/*:string*/ {
var s = text + '', i = s.indexOf("<![CDATA[");
if(i == -1) return s.replace(encregex, function($$, $1) { return encodings[$$]||String.fromCharCode(parseInt($1,$$.indexOf("x")>-1?16:10))||$$; }).replace(coderegex,function(m,c) {return String.fromCharCode(parseInt(c,16));});
var j = s.indexOf("]]>");
2022-04-20 17:31:11 +00:00
return raw_unescapexml(s.slice(0, i)) + s.slice(i+9,j) + raw_unescapexml(s.slice(j+3));
}
return function unescapexml(text/*:string*/, xlsx/*:boolean*/) {
var out = raw_unescapexml(text);
return xlsx ? out.replace(/\r\n/g, "\n") : out;
};
})();
var decregex=/[&<>'"]/g, charegex = /[\u0000-\u0008\u000b-\u001f\uFFFE-\uFFFF]/g;
function escapexml(text/*:string*/)/*:string*/{
var s = text + '';
return s.replace(decregex, function(y) { return rencoding[y]; }).replace(charegex,function(s) { return "_x" + ("000"+s.charCodeAt(0).toString(16)).slice(-4) + "_";});
}
function escapexmltag(text/*:string*/)/*:string*/{ return escapexml(text).replace(/ /g,"_x0020_"); }
var htmlcharegex = /[\u0000-\u001f]/g;
function escapehtml(text/*:string*/)/*:string*/{
var s = text + '';
return s.replace(decregex, function(y) { return rencoding[y]; }).replace(/\n/g, "<br/>").replace(htmlcharegex,function(s) { return "&#x" + ("000"+s.charCodeAt(0).toString(16)).slice(-4) + ";"; });
}
function escapexlml(text/*:string*/)/*:string*/{
var s = text + '';
return s.replace(decregex, function(y) { return rencoding[y]; }).replace(htmlcharegex,function(s) { return "&#x" + (s.charCodeAt(0).toString(16)).toUpperCase() + ";"; });
}
/* TODO: handle codepages */
2022-03-12 14:05:57 +00:00
var xlml_fixstr/*:StringConv*/ = /*#__PURE__*/(function() {
var entregex = /&#(\d+);/g;
function entrepl($$/*:string*/,$1/*:string*/)/*:string*/ { return String.fromCharCode(parseInt($1,10)); }
return function xlml_fixstr(str/*:string*/)/*:string*/ { return str.replace(entregex,entrepl); };
})();
2022-03-16 03:18:09 +00:00
function xlml_unfixstr(str/*:string*/)/*:string*/ { return str.replace(/(\r\n|[\r\n])/g,"\&#10;"); }
/* note: xsd:boolean valid values: true / 1 / false / 0 */
function parsexmlbool(value/*:any*/)/*:boolean*/ {
switch(value) {
case 1: case true: case '1': case 'true': return true;
case 0: case false: case '0': case 'false': return false;
//default: throw new Error("Invalid xsd:boolean " + value);
}
return false;
}
2022-03-20 01:54:41 +00:00
function utf8reada(orig/*:string*/)/*:string*/ {
var out = "", i = 0, c = 0, d = 0, e = 0, f = 0, w = 0;
while (i < orig.length) {
c = orig.charCodeAt(i++);
if (c < 128) { out += String.fromCharCode(c); continue; }
d = orig.charCodeAt(i++);
if (c>191 && c<224) { f = ((c & 31) << 6); f |= (d & 63); out += String.fromCharCode(f); continue; }
e = orig.charCodeAt(i++);
if (c < 240) { out += String.fromCharCode(((c & 15) << 12) | ((d & 63) << 6) | (e & 63)); continue; }
f = orig.charCodeAt(i++);
w = (((c & 7) << 18) | ((d & 63) << 12) | ((e & 63) << 6) | (f & 63))-65536;
out += String.fromCharCode(0xD800 + ((w>>>10)&1023));
out += String.fromCharCode(0xDC00 + (w&1023));
}
return out;
2022-03-20 01:54:41 +00:00
}
function utf8readb(data) {
var out = new_raw_buf(2*data.length), w, i, j = 1, k = 0, ww=0, c;
for(i = 0; i < data.length; i+=j) {
j = 1;
if((c=data.charCodeAt(i)) < 128) w = c;
else if(c < 224) { w = (c&31)*64+(data.charCodeAt(i+1)&63); j=2; }
else if(c < 240) { w=(c&15)*4096+(data.charCodeAt(i+1)&63)*64+(data.charCodeAt(i+2)&63); j=3; }
else { j = 4;
w = (c & 7)*262144+(data.charCodeAt(i+1)&63)*4096+(data.charCodeAt(i+2)&63)*64+(data.charCodeAt(i+3)&63);
w -= 65536; ww = 0xD800 + ((w>>>10)&1023); w = 0xDC00 + (w&1023);
}
if(ww !== 0) { out[k++] = ww&255; out[k++] = ww>>>8; ww = 0; }
out[k++] = w%256; out[k++] = w>>>8;
}
return out.slice(0,k).toString('ucs2');
}
2022-03-20 01:54:41 +00:00
function utf8readc(data) { return Buffer_from(data, 'binary').toString('utf8'); }
var utf8corpus = "foo bar baz\u00e2\u0098\u0083\u00f0\u009f\u008d\u00a3";
var utf8read = has_buf && (/*#__PURE__*/utf8readc(utf8corpus) == /*#__PURE__*/utf8reada(utf8corpus) && utf8readc || /*#__PURE__*/utf8readb(utf8corpus) == /*#__PURE__*/utf8reada(utf8corpus) && utf8readb) || utf8reada;
var utf8write/*:StringConv*/ = has_buf ? function(data) { return Buffer_from(data, 'utf8').toString("binary"); } : function(orig/*:string*/)/*:string*/ {
var out/*:Array<string>*/ = [], i = 0, c = 0, d = 0;
while(i < orig.length) {
c = orig.charCodeAt(i++);
switch(true) {
case c < 128: out.push(String.fromCharCode(c)); break;
case c < 2048:
out.push(String.fromCharCode(192 + (c >> 6)));
out.push(String.fromCharCode(128 + (c & 63)));
break;
case c >= 55296 && c < 57344:
c -= 55296; d = orig.charCodeAt(i++) - 56320 + (c<<10);
out.push(String.fromCharCode(240 + ((d >>18) & 7)));
out.push(String.fromCharCode(144 + ((d >>12) & 63)));
out.push(String.fromCharCode(128 + ((d >> 6) & 63)));
out.push(String.fromCharCode(128 + (d & 63)));
break;
default:
out.push(String.fromCharCode(224 + (c >> 12)));
out.push(String.fromCharCode(128 + ((c >> 6) & 63)));
out.push(String.fromCharCode(128 + (c & 63)));
}
}
return out.join("");
};
2022-03-12 14:05:57 +00:00
var htmldecode/*:{(s:string):string}*/ = /*#__PURE__*/(function() {
var entities/*:Array<[RegExp, string]>*/ = [
['nbsp', ' '], ['middot', '·'],
['quot', '"'], ['apos', "'"], ['gt', '>'], ['lt', '<'], ['amp', '&']
].map(function(x/*:[string, string]*/) { return [new RegExp('&' + x[0] + ';', "ig"), x[1]]; });
return function htmldecode(str/*:string*/)/*:string*/ {
var o = str
// Remove new lines and spaces from start of content
.replace(/^[\t\n\r ]+/, "")
// Remove new lines and spaces from end of content
.replace(/(^|[^\t\n\r ])[\t\n\r ]+$/,"$1")
// Added line which removes any white space characters after and before html tags
.replace(/>\s+/g,">").replace(/\b\s+</g,"<")
// Replace remaining new lines and spaces with space
.replace(/[\t\n\r ]+/g, " ")
// Replace <br> tags with new lines
.replace(/<\s*[bB][rR]\s*\/?>/g,"\n")
// Strip HTML elements
.replace(/<[^<>]*>/g,"");
for(var i = 0; i < entities.length; ++i) o = o.replace(entities[i][0], entities[i][1]);
return o;
};
})();
var vtvregex = /<\/?(?:vt:)?variant>/g, vtmregex = /<(?:vt:)([^<"'>]*)>([\s\S]*)</;
function parseVector(data/*:string*/, opts)/*:Array<{v:string,t:string}>*/ {
var h = parsexmltag(data);
var matches/*:Array<string>*/ = str_match_xml_ns_g(data, h.baseType)||[];
var res/*:Array<any>*/ = [];
if(matches.length != h.size) {
if(opts.WTF) throw new Error("unexpected vector length " + matches.length + " != " + h.size);
return res;
}
matches.forEach(function(x/*:string*/) {
var v = x.replace(vtvregex,"").match(vtmregex);
if(v) res.push({v:utf8read(v[2]), t:v[1]});
});
return res;
}
var wtregex = /(^\s|\s$|\n)/;
function writetag(f/*:string*/,g/*:string*/)/*:string*/ { return '<' + f + (g.match(wtregex)?' xml:space="preserve"' : "") + '>' + g + '</' + f + '>'; }
function wxt_helper(h)/*:string*/ { return keys(h).map(function(k) { return " " + k + '="' + h[k] + '"';}).join(""); }
function writextag(f/*:string*/,g/*:?string*/,h) { return '<' + f + ((h != null) ? wxt_helper(h) : "") + ((g != null) ? (g.match(wtregex)?' xml:space="preserve"' : "") + '>' + g + '</' + f : "/") + '>';}
function write_w3cdtf(d/*:Date*/, t/*:?boolean*/)/*:string*/ { try { return d.toISOString().replace(/\.\d*/,""); } catch(e) { if(t) throw e; } return ""; }
function write_vt(s, xlsx/*:?boolean*/)/*:string*/ {
switch(typeof s) {
case 'string':
var o = writextag('vt:lpwstr', escapexml(s));
if(xlsx) o = o.replace(/&quot;/g, "_x0022_");
return o;
case 'number': return writextag((s|0)==s?'vt:i4':'vt:r8', escapexml(String(s)));
case 'boolean': return writextag('vt:bool',s?'true':'false');
}
if(s instanceof Date) return writextag('vt:filetime', write_w3cdtf(s));
throw new Error("Unable to serialize " + s);
}
function xlml_normalize(d)/*:string*/ {
if(has_buf &&/*::typeof Buffer !== "undefined" && d != null && d instanceof Buffer &&*/ Buffer.isBuffer(d)) return d.toString('utf8');
if(typeof d === 'string') return d;
/* duktape */
if(typeof Uint8Array !== 'undefined' && d instanceof Uint8Array) return utf8read(a2s(ab2a(d)));
throw new Error("Bad input format: expected Buffer or string");
}
/* UOS uses CJK in tags, ODS uses invalid XML */
var xlmlregex = /<([\/]?)([^\s?><!\/:"]*:|)([^\s?<>:\/"]+)(?:\s+[^<>=?"'\s]+="[^"]*?")*\s*[\/]?>/mg;
var XMLNS = ({
2022-03-12 14:05:57 +00:00
CORE_PROPS: 'http://schemas.openxmlformats.org/package/2006/metadata/core-properties',
CUST_PROPS: "http://schemas.openxmlformats.org/officeDocument/2006/custom-properties",
EXT_PROPS: "http://schemas.openxmlformats.org/officeDocument/2006/extended-properties",
CT: 'http://schemas.openxmlformats.org/package/2006/content-types',
RELS: 'http://schemas.openxmlformats.org/package/2006/relationships',
TCMNT: 'http://schemas.microsoft.com/office/spreadsheetml/2018/threadedcomments',
'dc': 'http://purl.org/dc/elements/1.1/',
'dcterms': 'http://purl.org/dc/terms/',
'dcmitype': 'http://purl.org/dc/dcmitype/',
'mx': 'http://schemas.microsoft.com/office/mac/excel/2008/main',
'r': 'http://schemas.openxmlformats.org/officeDocument/2006/relationships',
'sjs': 'http://schemas.openxmlformats.org/package/2006/sheetjs/core-properties',
'vt': 'http://schemas.openxmlformats.org/officeDocument/2006/docPropsVTypes',
'xsi': 'http://www.w3.org/2001/XMLSchema-instance',
'xsd': 'http://www.w3.org/2001/XMLSchema'
}/*:any*/);
2022-03-12 14:05:57 +00:00
var XMLNS_main = [
'http://schemas.openxmlformats.org/spreadsheetml/2006/main',
'http://purl.oclc.org/ooxml/spreadsheetml/main',
'http://schemas.microsoft.com/office/excel/2006/main',
'http://schemas.microsoft.com/office/excel/2006/2'
];
var XLMLNS = ({
'o': 'urn:schemas-microsoft-com:office:office',
'x': 'urn:schemas-microsoft-com:office:excel',
'ss': 'urn:schemas-microsoft-com:office:spreadsheet',
'dt': 'uuid:C2F41010-65B3-11d1-A29F-00AA00C14882',
'mv': 'http://macVmlSchemaUri',
'v': 'urn:schemas-microsoft-com:vml',
'html': 'http://www.w3.org/TR/REC-html40'
}/*:any*/);