xsheetjs/bits/85_parsezip.js
SheetJS af3df44633 version bump 0.11.5: "string" type
- proper JS string input / output type
- bower main now uses full version (fixes #820 h/t @newmesiss)
- DOM parse directly acts on innerHTML (see #779 h/t @danxfisher)
- unicode core props and ext props (fixes #822 h/t @fureweb-com)
- shim update for IE10/11
- test refresh and flow checks
2017-09-30 02:18:11 -04:00

213 lines
7.8 KiB
JavaScript

function get_sheet_type(n/*:string*/)/*:string*/ {
if(RELS.WS.indexOf(n) > -1) return "sheet";
if(RELS.CS && n == RELS.CS) return "chart";
if(RELS.DS && n == RELS.DS) return "dialog";
if(RELS.MS && n == RELS.MS) return "macro";
return (n && n.length) ? n : "sheet";
}
function safe_parse_wbrels(wbrels, sheets) {
if(!wbrels) return 0;
try {
wbrels = sheets.map(function pwbr(w) { if(!w.id) w.id = w.strRelID; return [w.name, wbrels['!id'][w.id].Target, get_sheet_type(wbrels['!id'][w.id].Type)]; });
} catch(e) { return null; }
return !wbrels || wbrels.length === 0 ? null : wbrels;
}
function safe_parse_sheet(zip, path/*:string*/, relsPath/*:string*/, sheet, sheetRels, sheets, stype/*:string*/, opts, wb, themes, styles) {
try {
sheetRels[sheet]=parse_rels(getzipstr(zip, relsPath, true), path);
var data = getzipdata(zip, path);
switch(stype) {
case 'sheet': sheets[sheet]=parse_ws(data, path, opts,sheetRels[sheet], wb, themes, styles); break;
case 'chart':
var cs = parse_cs(data, path, opts,sheetRels[sheet], wb, themes, styles);
sheets[sheet] = cs;
if(!cs || !cs['!chart']) break;
var dfile = resolve_path(cs['!chart'].Target, path);
var drelsp = get_rels_path(dfile);
var draw = parse_drawing(getzipstr(zip, dfile, true), parse_rels(getzipstr(zip,drelsp,true), dfile));
var chartp = resolve_path(draw, dfile);
var crelsp = get_rels_path(chartp);
cs = parse_chart(getzipstr(zip, chartp, true), chartp, opts, parse_rels(getzipstr(zip, crelsp,true), chartp), wb, cs);
break;
case 'macro': sheets[sheet]=parse_ms(data, path, opts,sheetRels[sheet], wb, themes, styles); break;
case 'dialog': sheets[sheet]=parse_ds(data, path, opts,sheetRels[sheet], wb, themes, styles); break;
}
} catch(e) { if(opts.WTF) throw e; }
}
var nodirs = function nodirs(x/*:string*/)/*:boolean*/{return x.slice(-1) != '/';};
function strip_front_slash(x/*:string*/)/*:string*/ { return x.charAt(0) == '/' ? x.slice(1) : x; }
function parse_zip(zip/*:ZIP*/, opts/*:?ParseOpts*/)/*:Workbook*/ {
make_ssf(SSF);
opts = opts || {};
fix_read_opts(opts);
reset_cp();
/* OpenDocument Part 3 Section 2.2.1 OpenDocument Package */
if(safegetzipfile(zip, 'META-INF/manifest.xml')) return parse_ods(zip, opts);
/* UOC */
if(safegetzipfile(zip, 'objectdata.xml')) return parse_ods(zip, opts);
var entries = keys(zip.files).filter(nodirs).sort();
var dir = parse_ct((getzipstr(zip, '[Content_Types].xml')/*:?any*/), opts);
var xlsb = false;
var sheets, binname;
if(dir.workbooks.length === 0) {
binname = "xl/workbook.xml";
if(getzipdata(zip,binname, true)) dir.workbooks.push(binname);
}
if(dir.workbooks.length === 0) {
binname = "xl/workbook.bin";
if(!getzipdata(zip,binname,true)) throw new Error("Could not find workbook");
dir.workbooks.push(binname);
xlsb = true;
}
if(dir.workbooks[0].slice(-3) == "bin") xlsb = true;
if(xlsb) set_cp(1200);
var themes = ({}/*:any*/);
var styles = ({}/*:any*/);
if(!opts.bookSheets && !opts.bookProps) {
strs = [];
if(dir.sst) strs=parse_sst(getzipdata(zip, strip_front_slash(dir.sst)), dir.sst, opts);
if(opts.cellStyles && dir.themes.length) themes = parse_theme(getzipstr(zip, dir.themes[0].replace(/^\//,''), true)||"",dir.themes[0], opts);
if(dir.style) styles = parse_sty(getzipdata(zip, strip_front_slash(dir.style)), dir.style, themes, opts);
}
var externbooks = dir.links.map(function(link) {
return parse_xlink(getzipdata(zip, strip_front_slash(link)), link, opts);
});
var wb = parse_wb(getzipdata(zip, strip_front_slash(dir.workbooks[0])), dir.workbooks[0], opts);
var props = {}, propdata = "";
if(dir.coreprops.length) {
propdata = getzipdata(zip, strip_front_slash(dir.coreprops[0]), true);
if(propdata) props = parse_core_props(propdata);
if(dir.extprops.length !== 0) {
propdata = getzipdata(zip, strip_front_slash(dir.extprops[0]), true);
if(propdata) parse_ext_props(propdata, props, opts);
}
}
var custprops = {};
if(!opts.bookSheets || opts.bookProps) {
if (dir.custprops.length !== 0) {
propdata = getzipstr(zip, strip_front_slash(dir.custprops[0]), true);
if(propdata) custprops = parse_cust_props(propdata, opts);
}
}
var out = ({}/*:any*/);
if(opts.bookSheets || opts.bookProps) {
if(wb.Sheets) sheets = wb.Sheets.map(function pluck(x){ return x.name; });
else if(props.Worksheets && props.SheetNames.length > 0) sheets=props.SheetNames;
if(opts.bookProps) { out.Props = props; out.Custprops = custprops; }
if(opts.bookSheets && typeof sheets !== 'undefined') out.SheetNames = sheets;
if(opts.bookSheets ? out.SheetNames : opts.bookProps) return out;
}
sheets = {};
var deps = {};
if(opts.bookDeps && dir.calcchain) deps=parse_cc(getzipdata(zip, strip_front_slash(dir.calcchain)),dir.calcchain,opts);
var i=0;
var sheetRels = ({}/*:any*/);
var path, relsPath;
{
var wbsheets = wb.Sheets;
props.Worksheets = wbsheets.length;
props.SheetNames = [];
for(var j = 0; j != wbsheets.length; ++j) {
props.SheetNames[j] = wbsheets[j].name;
}
}
var wbext = xlsb ? "bin" : "xml";
var wbrelsfile = 'xl/_rels/workbook.' + wbext + '.rels';
var wbrels = parse_rels(getzipstr(zip, wbrelsfile, true), wbrelsfile);
if(wbrels) wbrels = safe_parse_wbrels(wbrels, wb.Sheets);
/* Numbers iOS hack */
var nmode = (getzipdata(zip,"xl/worksheets/sheet.xml",true))?1:0;
for(i = 0; i != props.Worksheets; ++i) {
var stype = "sheet";
if(wbrels && wbrels[i]) {
path = 'xl/' + (wbrels[i][1]).replace(/[\/]?xl\//, "");
stype = wbrels[i][2];
} else {
path = 'xl/worksheets/sheet'+(i+1-nmode)+"." + wbext;
path = path.replace(/sheet0\./,"sheet.");
}
relsPath = path.replace(/^(.*)(\/)([^\/]*)$/, "$1/_rels/$3.rels");
safe_parse_sheet(zip, path, relsPath, props.SheetNames[i], sheetRels, sheets, stype, opts, wb, themes, styles);
}
if(dir.comments) parse_comments(zip, dir.comments, sheets, sheetRels, opts);
out = ({
Directory: dir,
Workbook: wb,
Props: props,
Custprops: custprops,
Deps: deps,
Sheets: sheets,
SheetNames: props.SheetNames,
Strings: strs,
Styles: styles,
Themes: themes,
SSF: SSF.get_table()
}/*:any*/);
if(opts.bookFiles) {
out.keys = entries;
out.files = zip.files;
}
if(opts.bookVBA) {
if(dir.vba.length > 0) out.vbaraw = getzipdata(zip,strip_front_slash(dir.vba[0]),true);
else if(dir.defaults && dir.defaults.bin === 'application/vnd.ms-office.vbaProject') out.vbaraw = getzipdata(zip,'xl/vbaProject.bin',true);
}
return out;
}
/* references to [MS-OFFCRYPTO] */
function parse_xlsxcfb(cfb, opts/*:?ParseOpts*/)/*:Workbook*/ {
var f = 'Version';
var data = CFB.find(cfb, f);
if(!data || !data.content) throw new Error("ECMA-376 Encrypted file missing " + f);
var version = parse_DataSpaceVersionInfo(data.content);
/* 2.3.4.1 */
f = 'DataSpaceMap';
data = CFB.find(cfb, f);
if(!data || !data.content) throw new Error("ECMA-376 Encrypted file missing " + f);
var dsm = parse_DataSpaceMap(data.content);
if(dsm.length != 1 || dsm[0].comps.length != 1 || dsm[0].comps[0].t != 0 || dsm[0].name != "StrongEncryptionDataSpace" || dsm[0].comps[0].v != "EncryptedPackage")
throw new Error("ECMA-376 Encrypted file bad " + f);
f = 'StrongEncryptionDataSpace';
data = CFB.find(cfb, f);
if(!data || !data.content) throw new Error("ECMA-376 Encrypted file missing " + f);
var seds = parse_DataSpaceDefinition(data.content);
if(seds.length != 1 || seds[0] != "StrongEncryptionTransform")
throw new Error("ECMA-376 Encrypted file bad " + f);
/* 2.3.4.3 */
f = '!Primary';
data = CFB.find(cfb, f);
if(!data || !data.content) throw new Error("ECMA-376 Encrypted file missing " + f);
var hdr = parse_Primary(data.content);
f = 'EncryptionInfo';
data = CFB.find(cfb, f);
if(!data || !data.content) throw new Error("ECMA-376 Encrypted file missing " + f);
var einfo = parse_EncryptionInfo(data.content);
throw new Error("File is password-protected");
}