version bump 0.10.0: performance
- reworked IIFE - avoid Buffer read helpers (see https://github.com/joyent/node/issues/7809) - avoid nested functions in parse - eliminate consecutive nulls in test files (libreoffice) - travis + coveralls support - jscs linting
This commit is contained in:
parent
ab962d782e
commit
5c7df1f3db
6
.gitignore
vendored
Normal file
6
.gitignore
vendored
Normal file
@ -0,0 +1,6 @@
|
||||
test_files
|
||||
node_modules/
|
||||
*.xls
|
||||
misc/coverage.html
|
||||
prof.js
|
||||
v8.log
|
||||
6
.jscs.json
Normal file
6
.jscs.json
Normal file
@ -0,0 +1,6 @@
|
||||
{
|
||||
"requireCommaBeforeLineBreak": true,
|
||||
"disallowTrailingWhitespace": true,
|
||||
"disallowTrailingComma": true
|
||||
}
|
||||
|
||||
@ -1,8 +1,14 @@
|
||||
language: node_js
|
||||
node_js:
|
||||
- "0.11"
|
||||
- "0.10"
|
||||
- "0.8"
|
||||
before_install:
|
||||
- "npm install -g mocha"
|
||||
- "npm install blanket"
|
||||
- "npm install xlsjs"
|
||||
- "npm install coveralls mocha-lcov-reporter"
|
||||
before_script:
|
||||
- "make init"
|
||||
after_success:
|
||||
- "make coveralls-spin"
|
||||
|
||||
26
Makefile
26
Makefile
@ -1,16 +1,17 @@
|
||||
LIB=cfb
|
||||
DEPS=$(wildcard bits/*.js)
|
||||
DEPS=$(sort $(wildcard bits/*.js))
|
||||
TARGET=$(LIB).js
|
||||
|
||||
$(TARGET): $(DEPS)
|
||||
cat $^ > $@
|
||||
cat $^ | tr -d '\15\32' > $@
|
||||
|
||||
bits/31_version.js: package.json
|
||||
echo "this.version = '"`grep version package.json | awk '{gsub(/[^0-9a-z\.-]/,"",$$2); print $$2}'`"';" > $@
|
||||
echo "exports.version = '"`grep version package.json | awk '{gsub(/[^0-9a-z\.-]/,"",$$2); print $$2}'`"';" > $@
|
||||
|
||||
.PHONY: clean
|
||||
clean:
|
||||
rm -rf $(TARGET) ./test_files/
|
||||
rm -f $(TARGET)
|
||||
rm -rf ./test_files/
|
||||
|
||||
.PHONY: init
|
||||
init:
|
||||
@ -21,20 +22,31 @@ init:
|
||||
test mocha: test.js
|
||||
mocha -R spec
|
||||
|
||||
.PHONY: prof
|
||||
prof:
|
||||
cat misc/prof.js test.js > prof.js
|
||||
node --prof prof.js
|
||||
|
||||
.PHONY: lint
|
||||
lint: $(TARGET)
|
||||
jshint --show-non-errors $(TARGET)
|
||||
jscs $(TARGET)
|
||||
|
||||
.PHONY: cov
|
||||
.PHONY: cov cov-spin
|
||||
cov: misc/coverage.html
|
||||
cov-spin:
|
||||
make cov & bash misc/spin.sh $$!
|
||||
|
||||
misc/coverage.html: $(TARGET) test.js
|
||||
mocha --require blanket -R html-cov > misc/coverage.html
|
||||
mocha --require blanket -R html-cov > $@
|
||||
|
||||
.PHONY: coveralls
|
||||
.PHONY: coveralls coveralls-spin
|
||||
coveralls:
|
||||
mocha --require blanket --reporter mocha-lcov-reporter | ./node_modules/coveralls/bin/coveralls.js
|
||||
|
||||
coveralls-spin:
|
||||
make coveralls & bash misc/spin.sh $$!
|
||||
|
||||
.PHONY: dist
|
||||
dist: $(TARGET)
|
||||
cp $(TARGET) dist/
|
||||
|
||||
@ -75,7 +75,7 @@ The entry objects are available from `FullPathDir` and `FileIndex` elements of t
|
||||
container object.
|
||||
|
||||
- `.name` is the (case sensitive) internal name
|
||||
- `.type` is the type (`stream` for files, `storage` for dirs, `root` for root)
|
||||
- `.type` is the type (`2 (stream)` for files, `1 (storage)` for dirs, `5 (root)` for root)
|
||||
- `.content` is a Buffer/Array with the raw content
|
||||
- `.ct`/`.mt` are the creation and modification time (if provided in file)
|
||||
|
||||
@ -90,5 +90,9 @@ Writing is not supported. It is in the works, but it has not yet been released.
|
||||
This implementation is covered under Apache 2.0 license. It complies with the
|
||||
[Open Specifications Promise](http://www.microsoft.com/openspecifications/)
|
||||
|
||||
[](https://travis-ci.org/SheetJS/js-cfb)
|
||||
|
||||
[](https://coveralls.io/r/SheetJS/js-cfb?branch=master)
|
||||
|
||||
[](http://githalytics.com/SheetJS/js-cfb)
|
||||
|
||||
|
||||
125
bits/08_blob.js
125
bits/08_blob.js
@ -1,114 +1,81 @@
|
||||
var Base64 = (function(){
|
||||
var map = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=";
|
||||
return {
|
||||
/* (will need this for writing) encode: function(input, utf8) {
|
||||
var o = "";
|
||||
var c1, c2, c3, e1, e2, e3, e4;
|
||||
for(var i = 0; i < input.length; ) {
|
||||
c1 = input.charCodeAt(i++);
|
||||
c2 = input.charCodeAt(i++);
|
||||
c3 = input.charCodeAt(i++);
|
||||
e1 = c1 >> 2;
|
||||
e2 = (c1 & 3) << 4 | c2 >> 4;
|
||||
e3 = (c2 & 15) << 2 | c3 >> 6;
|
||||
e4 = c3 & 63;
|
||||
if (isNaN(c2)) { e3 = e4 = 64; }
|
||||
else if (isNaN(c3)) { e4 = 64; }
|
||||
o += map.charAt(e1) + map.charAt(e2) + map.charAt(e3) + map.charAt(e4);
|
||||
}
|
||||
return o;
|
||||
},*/
|
||||
decode: function(input, utf8) {
|
||||
decode: function(input) {
|
||||
var o = "";
|
||||
var c1, c2, c3;
|
||||
var e1, e2, e3, e4;
|
||||
input = input.replace(/[^A-Za-z0-9\+\/\=]/g, "");
|
||||
input = input.replace(/[^\w\+\/\=]/g, "");
|
||||
for(var i = 0; i < input.length;) {
|
||||
e1 = map.indexOf(input.charAt(i++));
|
||||
e2 = map.indexOf(input.charAt(i++));
|
||||
e3 = map.indexOf(input.charAt(i++));
|
||||
e4 = map.indexOf(input.charAt(i++));
|
||||
c1 = e1 << 2 | e2 >> 4;
|
||||
c2 = (e2 & 15) << 4 | e3 >> 2;
|
||||
c3 = (e3 & 3) << 6 | e4;
|
||||
c1 = (e1 << 2) | (e2 >> 4);
|
||||
o += String.fromCharCode(c1);
|
||||
if (e3 != 64) { o += String.fromCharCode(c2); }
|
||||
if (e4 != 64) { o += String.fromCharCode(c3); }
|
||||
|
||||
e3 = map.indexOf(input.charAt(i++));
|
||||
c2 = ((e2 & 15) << 4) | (e3 >> 2);
|
||||
if (e3 !== 64) { o += String.fromCharCode(c2); }
|
||||
|
||||
e4 = map.indexOf(input.charAt(i++));
|
||||
c3 = ((e3 & 3) << 6) | e4;
|
||||
if (e4 !== 64) { o += String.fromCharCode(c3); }
|
||||
}
|
||||
return o;
|
||||
}
|
||||
};
|
||||
})();
|
||||
|
||||
function s2a(s) {
|
||||
if(typeof Buffer !== 'undefined') return new Buffer(s, "binary");
|
||||
var w = s.split("").map(function(x){ return x.charCodeAt(0) & 0xff; });
|
||||
return w;
|
||||
}
|
||||
var chr0 = /\u0000/g, chr1 = /[\u0001-\u0006]/;
|
||||
|
||||
var s2a, _s2a;
|
||||
s2a = _s2a = function _s2a(s) { return s.split("").map(function(x){ return x.charCodeAt(0) & 0xff; }); };
|
||||
var __toBuffer, ___toBuffer;
|
||||
__toBuffer = ___toBuffer = function(bufs) { var x = []; for(var i = 0; i < bufs[0].length; ++i) { x.push.apply(x, bufs[0][i]); } return x; };
|
||||
var __utf16le, ___utf16le;
|
||||
__utf16le = ___utf16le = function(b,s,e) { var ss=[]; for(var i=s; i<e; i+=2) ss.push(String.fromCharCode(__readUInt16LE(b,i))); return ss.join("").replace(chr0,'').replace(chr1,'!'); };
|
||||
var __hexlify, ___hexlify;
|
||||
__hexlify = ___hexlify = function(b,s,l) { return b.slice(s,(s+l)).map(function(x){return (x<16?"0":"") + x.toString(16);}).join(""); };
|
||||
var bconcat = function(bufs) { return [].concat.apply([], bufs); };
|
||||
|
||||
|
||||
var __toBuffer;
|
||||
if(typeof Buffer !== "undefined") {
|
||||
Buffer.prototype.hexlify= function() { return this.toString('hex'); };
|
||||
Buffer.prototype.utf16le= function(s,e){return this.toString('utf16le',s,e).replace(/\u0000/,'').replace(/[\u0001-\u0006]/,'!');};
|
||||
Buffer.prototype.utf8 = function(s,e) { return this.toString('utf8',s,e); };
|
||||
__toBuffer = function(bufs) { return Buffer.concat(bufs[0]); };
|
||||
} else {
|
||||
__toBuffer = function(bufs) {
|
||||
var x = [];
|
||||
for(var i = 0; i != bufs[0].length; ++i) { x = x.concat(bufs[0][i]); }
|
||||
return x;
|
||||
__utf16le = function(b,s,e) {
|
||||
if(!Buffer.isBuffer(b)) return ___utf16le(b,s,e);
|
||||
return b.toString('utf16le',s,e).replace(chr0,'').replace(chr1,'!');
|
||||
};
|
||||
__hexlify = function(b,s,l) { return Buffer.isBuffer(b) ? b.toString('hex',s,s+l) : ___hexlify(b,s,l); };
|
||||
__toBuffer = function(bufs) { return (bufs[0].length > 0 && Buffer.isBuffer(bufs[0][0])) ? Buffer.concat(bufs[0]) : ___toBuffer(bufs);};
|
||||
s2a = function(s) { return Buffer(s, "binary"); };
|
||||
bconcat = function(bufs) { return Buffer.isBuffer(bufs[0]) ? Buffer.concat(bufs) : [].concat.apply([], bufs); };
|
||||
}
|
||||
|
||||
var __readUInt8 = function(b, idx) { return b.readUInt8 ? b.readUInt8(idx) : b[idx]; };
|
||||
var __readUInt16LE = function(b, idx) { return b.readUInt16LE ? b.readUInt16LE(idx) : b[idx+1]*(1<<8)+b[idx]; };
|
||||
var __readInt16LE = function(b, idx) { var u = __readUInt16LE(b,idx); if(!(u & 0x8000)) return u; return (0xffff - u + 1) * -1; };
|
||||
var __readUInt32LE = function(b, idx) { return b.readUInt32LE ? b.readUInt32LE(idx) : b[idx+3]*(1<<24)+b[idx+2]*(1<<16)+b[idx+1]*(1<<8)+b[idx]; };
|
||||
var __readInt32LE = function(b, idx) { if(b.readInt32LE) return b.readInt32LE(idx); var u = __readUInt32LE(b,idx); if(!(u & 0x80000000)) return u; return (0xffffffff - u + 1) * -1; };
|
||||
|
||||
var __hexlify = function(b) { return b.map(function(x){return (x<16?"0":"") + x.toString(16);}).join(""); };
|
||||
var __readUInt8 = function(b, idx) { return b[idx]; };
|
||||
var __readUInt16LE = function(b, idx) { return b[idx+1]*(1<<8)+b[idx]; };
|
||||
var __readInt16LE = function(b, idx) { var u = b[idx+1]*(1<<8)+b[idx]; return (u < 0x8000) ? u : (0xffff - u + 1) * -1; };
|
||||
var __readUInt32LE = function(b, idx) { return b[idx+3]*(1<<24)+(b[idx+2]<<16)+(b[idx+1]<<8)+b[idx]; };
|
||||
var __readInt32LE = function(b, idx) { return (b[idx+3]<<24)+(b[idx+2]<<16)+(b[idx+1]<<8)+b[idx]; };
|
||||
|
||||
var __utf16le = function(b,s,e) { if(b.utf16le) return b.utf16le(s,e); var ss=[]; for(var i=s; i<e; i+=2) ss.push(String.fromCharCode(__readUInt16LE(b,i))); return ss.join("").replace(/\u0000/,'').replace(/[\u0001-\u0006]/,'!'); };
|
||||
|
||||
var __utf8 = function(b,s,e) { if(b.utf8) return b.utf8(s,e); var ss=[]; for(var i=s; i<e; i++) ss.push(String.fromCharCode(__readUInt8(b,i))); return ss.join(""); };
|
||||
|
||||
function bconcat(bufs) { return (typeof Buffer !== 'undefined') ? Buffer.concat(bufs) : [].concat.apply([], bufs); }
|
||||
|
||||
/** Buffer helpers -- keep track of read location `.l` and move it */
|
||||
function ReadShift(size, t) {
|
||||
var o, oo=[], w, vv; t = t || 'u';
|
||||
var oI, oS, type = 0;
|
||||
switch(size) {
|
||||
case 1: o = __readUInt8(this, this.l); break;
|
||||
case 2: o=(t==='u' ? __readUInt16LE : __readInt16LE)(this, this.l); break;
|
||||
case 4: o = __readUInt32LE(this, this.l); break;
|
||||
case 8:
|
||||
case 16: o = this.toString('hex', this.l,this.l+size); break;
|
||||
|
||||
case 'utf8': size = t; o = __utf8(this, this.l, this.l + size); break;
|
||||
case 'utf16le': size=2*t; o = __utf16le(this, this.l, this.l + size); break;
|
||||
|
||||
case 'cstr': size = 0; o = "";
|
||||
while((w=__readUInt8(this, this.l + size++))!==0) oo.push(String.fromCharCode(w));
|
||||
o = oo.join(""); break;
|
||||
case 'wstr': size = 0; o = "";
|
||||
while((w=__readUInt16LE(this,this.l +size))!==0){oo.push(String.fromCharCode(w));size+=2;}
|
||||
size+=2; break;
|
||||
case 1: oI = __readUInt8(this, this.l); break;
|
||||
case 2: oI = (t !== 'i' ? __readUInt16LE : __readInt16LE)(this, this.l); break;
|
||||
case 4: oI = __readInt32LE(this, this.l); break;
|
||||
case 16: type = 2; oS = __hexlify(this, this.l, size);
|
||||
}
|
||||
this.l+=size; return o;
|
||||
this.l+=size; if(type === 0) return oI; return oS;
|
||||
}
|
||||
|
||||
function CheckField(hexstr, fld) {
|
||||
var b = this.slice(this.l, this.l+hexstr.length/2);
|
||||
var m = b.hexlify ? b.hexlify() : __hexlify(b);
|
||||
if(m !== hexstr) throw (fld||"") + 'Expected ' + hexstr + ' saw ' + m;
|
||||
this.l += hexstr.length/2;
|
||||
var m = __hexlify(this,this.l,hexstr.length>>1);
|
||||
if(m !== hexstr) throw fld + 'Expected ' + hexstr + ' saw ' + m;
|
||||
this.l += hexstr.length>>1;
|
||||
}
|
||||
|
||||
function prep_blob(blob, pos) {
|
||||
blob.read_shift = ReadShift.bind(blob);
|
||||
blob.l = pos;
|
||||
blob.read_shift = ReadShift;
|
||||
blob.chk = CheckField;
|
||||
blob.l = pos || 0;
|
||||
var read = ReadShift.bind(blob), chk = CheckField.bind(blob);
|
||||
return [read, chk];
|
||||
}
|
||||
|
||||
|
||||
@ -1,2 +1,3 @@
|
||||
/* [MS-CFB] v20130118 */
|
||||
var CFB = (function(){
|
||||
var CFB = (function _CFB(){
|
||||
var exports = {};
|
||||
|
||||
@ -1 +1 @@
|
||||
this.version = '0.9.1';
|
||||
exports.version = '0.10.0';
|
||||
|
||||
98
bits/40_parse.js
Normal file
98
bits/40_parse.js
Normal file
@ -0,0 +1,98 @@
|
||||
function parse(file) {
|
||||
var mver = 3; // major version
|
||||
var ssz = 512; // sector size
|
||||
var nmfs = 0; // number of mini FAT sectors
|
||||
var ndfs = 0; // number of DIFAT sectors
|
||||
var dir_start = 0; // first directory sector location
|
||||
var minifat_start = 0; // first mini FAT sector location
|
||||
var difat_start = 0; // first mini FAT sector location
|
||||
|
||||
var fat_addrs = []; // locations of FAT sectors
|
||||
|
||||
/* [MS-CFB] 2.2 Compound File Header */
|
||||
var blob = file.slice(0,512);
|
||||
prep_blob(blob, 0);
|
||||
|
||||
/* major version */
|
||||
mver = check_get_mver(blob);
|
||||
switch(mver) {
|
||||
case 3: ssz = 512; break; case 4: ssz = 4096; break;
|
||||
default: throw "Major Version: Expected 3 or 4 saw " + mver;
|
||||
}
|
||||
|
||||
/* reprocess header */
|
||||
if(ssz !== 512) { blob = file.slice(0,ssz); prep_blob(blob, 28 /* blob.l */); }
|
||||
/* Save header for final object */
|
||||
var header = file.slice(0,ssz);
|
||||
|
||||
check_shifts(blob, mver);
|
||||
|
||||
// Number of Directory Sectors
|
||||
var nds = blob.read_shift(4, 'i');
|
||||
if(mver === 3 && nds !== 0) throw '# Directory Sectors: Expected 0 saw ' + nds;
|
||||
|
||||
// Number of FAT Sectors
|
||||
//var nfs = blob.read_shift(4, 'i');
|
||||
blob.l += 4;
|
||||
|
||||
// First Directory Sector Location
|
||||
dir_start = blob.read_shift(4, 'i');
|
||||
|
||||
// Transaction Signature
|
||||
blob.l += 4;
|
||||
|
||||
// Mini Stream Cutoff Size
|
||||
blob.chk('00100000', 'Mini Stream Cutoff Size: ');
|
||||
|
||||
// First Mini FAT Sector Location
|
||||
minifat_start = blob.read_shift(4, 'i');
|
||||
|
||||
// Number of Mini FAT Sectors
|
||||
nmfs = blob.read_shift(4, 'i');
|
||||
|
||||
// First DIFAT sector location
|
||||
difat_start = blob.read_shift(4, 'i');
|
||||
|
||||
// Number of DIFAT Sectors
|
||||
ndfs = blob.read_shift(4, 'i');
|
||||
|
||||
// Grab FAT Sector Locations
|
||||
for(var q, j = 0; j < 109; ++j) { /* 109 = (512 - blob.l)>>>2; */
|
||||
q = blob.read_shift(4, 'i');
|
||||
if(q<0) break;
|
||||
fat_addrs[j] = q;
|
||||
}
|
||||
|
||||
/** Break the file up into sectors */
|
||||
var sectors = sectorify(file, ssz);
|
||||
|
||||
sleuth_fat(difat_start, ndfs, sectors, ssz, fat_addrs);
|
||||
|
||||
/** Chains */
|
||||
var sector_list = make_sector_list(sectors, dir_start, fat_addrs, ssz);
|
||||
|
||||
sector_list[dir_start].name = "!Directory";
|
||||
if(nmfs > 0 && minifat_start !== ENDOFCHAIN) sector_list[minifat_start].name = "!MiniFAT";
|
||||
sector_list[fat_addrs[0]].name = "!FAT";
|
||||
|
||||
/* [MS-CFB] 2.6.1 Compound File Directory Entry */
|
||||
var files = {}, Paths = [], FileIndex = [], FullPaths = [], FullPathDir = {};
|
||||
read_directory(dir_start, sector_list, sectors, Paths, nmfs, files, FileIndex);
|
||||
|
||||
build_full_paths(FileIndex, FullPathDir, FullPaths, Paths);
|
||||
|
||||
var root_name = Paths.shift();
|
||||
Paths.root = root_name;
|
||||
|
||||
/* [MS-CFB] 2.6.4 (Unicode 3.0.1 case conversion) */
|
||||
var find_path = make_find_path(FullPaths, Paths, FileIndex, files, root_name);
|
||||
|
||||
return {
|
||||
raw: {header: header, sectors: sectors},
|
||||
FileIndex: FileIndex,
|
||||
FullPaths: FullPaths,
|
||||
FullPathDir: FullPathDir,
|
||||
find: find_path
|
||||
};
|
||||
} // parse
|
||||
|
||||
@ -1,3 +0,0 @@
|
||||
function parse(file) {
|
||||
|
||||
|
||||
33
bits/41_mver.js
Normal file
33
bits/41_mver.js
Normal file
@ -0,0 +1,33 @@
|
||||
/* [MS-CFB] 2.2 Compound File Header -- read up to major version */
|
||||
function check_get_mver(blob) {
|
||||
// header signature 8
|
||||
blob.chk(HEADER_SIGNATURE, 'Header Signature: ');
|
||||
|
||||
// clsid 16
|
||||
blob.chk(HEADER_CLSID, 'CLSID: ');
|
||||
|
||||
// minor version 2
|
||||
blob.l += 2;
|
||||
|
||||
return blob.read_shift(2,'u');
|
||||
}
|
||||
function check_shifts(blob, mver) {
|
||||
var shift = 0x09;
|
||||
|
||||
// Byte Order
|
||||
blob.chk('feff', 'Byte Order: ');
|
||||
|
||||
// Sector Shift
|
||||
switch((shift = blob.read_shift(2))) {
|
||||
case 0x09: if(mver !== 3) throw 'MajorVersion/SectorShift Mismatch'; break;
|
||||
case 0x0c: if(mver !== 4) throw 'MajorVersion/SectorShift Mismatch'; break;
|
||||
default: throw 'Sector Shift: Expected 9 or 12 saw ' + shift;
|
||||
}
|
||||
|
||||
// Mini Sector Shift
|
||||
blob.chk('0600', 'Mini Sector Shift: ');
|
||||
|
||||
// Reserved
|
||||
blob.chk('000000000000', 'Reserved: ');
|
||||
}
|
||||
|
||||
@ -1,99 +0,0 @@
|
||||
var mver = 3; // major version
|
||||
var ssz = 512; // sector size
|
||||
var mssz = 64; // mini sector size
|
||||
var nds = 0; // number of directory sectors
|
||||
var nfs = 0; // number of FAT sectors
|
||||
var nmfs = 0; // number of mini FAT sectors
|
||||
var ndfs = 0; // number of DIFAT sectors
|
||||
var dir_start = 0; // first directory sector location
|
||||
var minifat_start = 0; // first mini FAT sector location
|
||||
var difat_start = 0; // first mini FAT sector location
|
||||
|
||||
var ms_cutoff_size = 4096; // mini stream cutoff size
|
||||
var minifat_store = 0; // first sector with minifat data
|
||||
var minifat_size = 0; // size of minifat data
|
||||
|
||||
var fat_addrs = []; // locations of FAT sectors
|
||||
|
||||
/* [MS-CFB] 2.2 Compound File Header */
|
||||
var blob = file.slice(0,512);
|
||||
prep_blob(blob);
|
||||
var read = ReadShift.bind(blob), chk = CheckField.bind(blob);
|
||||
var j = 0, q;
|
||||
|
||||
// header signature 8
|
||||
chk(HEADER_SIGNATURE, 'Header Signature: ');
|
||||
|
||||
// clsid 16
|
||||
chk(HEADER_CLSID, 'CLSID: ');
|
||||
|
||||
// minor version 2
|
||||
read(2);
|
||||
|
||||
// major version 3
|
||||
mver = read(2);
|
||||
switch(mver) {
|
||||
case 3: ssz = 512; break;
|
||||
case 4: ssz = 4096; break;
|
||||
default: throw "Major Version: Expected 3 or 4 saw " + mver;
|
||||
}
|
||||
|
||||
// reprocess header
|
||||
var pos = blob.l;
|
||||
blob = file.slice(0,ssz);
|
||||
prep_blob(blob,pos);
|
||||
read = ReadShift.bind(blob);
|
||||
chk = CheckField.bind(blob);
|
||||
var header = file.slice(0,ssz);
|
||||
|
||||
// Byte Order TODO
|
||||
chk('feff', 'Byte Order: ');
|
||||
|
||||
// Sector Shift
|
||||
switch((q = read(2))) {
|
||||
case 0x09: if(mver !== 3) throw 'MajorVersion/SectorShift Mismatch'; break;
|
||||
case 0x0c: if(mver !== 4) throw 'MajorVersion/SectorShift Mismatch'; break;
|
||||
default: throw 'Sector Shift: Expected 9 or 12 saw ' + q;
|
||||
}
|
||||
|
||||
// Mini Sector Shift
|
||||
chk('0600', 'Mini Sector Shift: ');
|
||||
|
||||
// Reserved
|
||||
chk('000000000000', 'Mini Sector Shift: ');
|
||||
|
||||
// Number of Directory Sectors
|
||||
nds = read(4);
|
||||
if(mver === 3 && nds !== 0) throw '# Directory Sectors: Expected 0 saw ' + nds;
|
||||
|
||||
// Number of FAT Sectors
|
||||
nfs = read(4);
|
||||
|
||||
// First Directory Sector Location
|
||||
dir_start = read(4);
|
||||
|
||||
// Transaction Signature TODO
|
||||
read(4);
|
||||
|
||||
// Mini Stream Cutoff Size TODO
|
||||
chk('00100000', 'Mini Stream Cutoff Size: ');
|
||||
|
||||
// First Mini FAT Sector Location
|
||||
minifat_start = read(4);
|
||||
|
||||
// Number of Mini FAT Sectors
|
||||
nmfs = read(4);
|
||||
|
||||
// First DIFAT sector location
|
||||
difat_start = read(4);
|
||||
|
||||
// Number of DIFAT Sectors
|
||||
ndfs = read(4);
|
||||
|
||||
// Grab FAT Sector Locations
|
||||
for(j = 0; blob.l != 512; ) {
|
||||
if((q = read(4))>=MAXREGSECT) break;
|
||||
fat_addrs[j++] = q;
|
||||
}
|
||||
|
||||
|
||||
9
bits/42_sectorify.js
Normal file
9
bits/42_sectorify.js
Normal file
@ -0,0 +1,9 @@
|
||||
/** Break the file up into sectors */
|
||||
function sectorify(file, ssz) {
|
||||
var nsectors = Math.ceil(file.length/ssz)-1;
|
||||
var sectors = new Array(nsectors);
|
||||
for(var i=1; i < nsectors; ++i) sectors[i-1] = file.slice(i*ssz,(i+1)*ssz);
|
||||
sectors[nsectors-1] = file.slice(nsectors*ssz);
|
||||
return sectors;
|
||||
}
|
||||
|
||||
41
bits/43_rbtree.js
Normal file
41
bits/43_rbtree.js
Normal file
@ -0,0 +1,41 @@
|
||||
/* [MS-CFB] 2.6.4 Red-Black Tree */
|
||||
function build_full_paths(FI, FPD, FP, Paths) {
|
||||
var i = 0, L = 0, R = 0, C = 0, j = 0, pl = Paths.length;
|
||||
var dad = new Array(pl), q = new Array(pl);
|
||||
|
||||
for(; i < pl; ++i) { dad[i]=q[i]=i; FP[i]=Paths[i]; }
|
||||
|
||||
for(; j < q.length; ++j) {
|
||||
i = q[j];
|
||||
L = FI[i].L; R = FI[i].R; C = FI[i].C;
|
||||
if(dad[i] === i) {
|
||||
if(L !== -1 /*NOSTREAM*/ && dad[L] !== L) dad[i] = dad[L];
|
||||
if(R !== -1 && dad[R] !== R) dad[i] = dad[R];
|
||||
}
|
||||
if(C !== -1 /*NOSTREAM*/) dad[C] = i;
|
||||
if(L !== -1) { dad[L] = dad[i]; q.push(L); }
|
||||
if(R !== -1) { dad[R] = dad[i]; q.push(R); }
|
||||
}
|
||||
for(i=1; i !== pl; ++i) if(dad[i] === i) {
|
||||
if(R !== -1 /*NOSTREAM*/ && dad[R] !== R) dad[i] = dad[R];
|
||||
else if(L !== -1 && dad[L] !== L) dad[i] = dad[L];
|
||||
}
|
||||
|
||||
for(i=1; i < pl; ++i) {
|
||||
if(FI[i].type === 0 /* unknown */) continue;
|
||||
j = dad[i];
|
||||
if(j === 0) FP[i] = FP[0] + "/" + FP[i];
|
||||
else while(j !== 0) {
|
||||
FP[i] = FP[j] + "/" + FP[i];
|
||||
j = dad[j];
|
||||
}
|
||||
dad[i] = 0;
|
||||
}
|
||||
|
||||
FP[0] += "/";
|
||||
for(i=1; i < pl; ++i) {
|
||||
if(FI[i].type !== 2 /* stream */) FP[i] += "/";
|
||||
FPD[FP[i]] = FI[i];
|
||||
}
|
||||
}
|
||||
|
||||
@ -1,51 +0,0 @@
|
||||
/** Break the file up into sectors */
|
||||
var nsectors = Math.ceil((file.length - ssz)/ssz);
|
||||
var sectors = [];
|
||||
for(var i=1; i != nsectors; ++i) sectors[i-1] = file.slice(i*ssz,(i+1)*ssz);
|
||||
sectors[nsectors-1] = file.slice(nsectors*ssz);
|
||||
|
||||
/** Chase down the rest of the DIFAT chain to build a comprehensive list
|
||||
DIFAT chains by storing the next sector number as the last 32 bytes */
|
||||
function sleuth_fat(idx, cnt) {
|
||||
if(idx === ENDOFCHAIN) {
|
||||
if(cnt !== 0) throw "DIFAT chain shorter than expected";
|
||||
return;
|
||||
}
|
||||
if(idx !== FREESECT) {
|
||||
var sector = sectors[idx];
|
||||
for(var i = 0; i != ssz/4-1; ++i) {
|
||||
if((q = __readUInt32LE(sector,i*4)) === ENDOFCHAIN) break;
|
||||
fat_addrs.push(q);
|
||||
}
|
||||
sleuth_fat(__readUInt32LE(sector,ssz-4),cnt - 1);
|
||||
}
|
||||
}
|
||||
sleuth_fat(difat_start, ndfs);
|
||||
|
||||
/** DONT CAT THE FAT! Just calculate where we need to go */
|
||||
function get_buffer(byte_addr, bytes) {
|
||||
var addr = fat_addrs[Math.floor(byte_addr*4/ssz)];
|
||||
if(ssz - (byte_addr*4 % ssz) < (bytes || 0)) throw "FAT boundary crossed: " + byte_addr + " "+bytes+" "+ssz;
|
||||
return sectors[addr].slice((byte_addr*4 % ssz));
|
||||
}
|
||||
|
||||
function get_buffer_u32(byte_addr) {
|
||||
return __readUInt32LE(get_buffer(byte_addr,4), 0);
|
||||
}
|
||||
|
||||
function get_next_sector(idx) { return get_buffer_u32(idx); }
|
||||
|
||||
/** Chains */
|
||||
var chkd = new Array(sectors.length), sector_list = [];
|
||||
var get_sector = function get_sector(k) { return sectors[k]; };
|
||||
for(i=0; i != sectors.length; ++i) {
|
||||
var buf = [], k = (i + dir_start) % sectors.length;
|
||||
if(chkd[k]) continue;
|
||||
for(j=k; j<=MAXREGSECT; buf.push(j),j=get_next_sector(j)) chkd[j] = true;
|
||||
sector_list[k] = {nodes: buf};
|
||||
sector_list[k].data = __toBuffer(Array(buf.map(get_sector)));
|
||||
}
|
||||
sector_list[dir_start].name = "!Directory";
|
||||
if(nmfs > 0 && minifat_start !== ENDOFCHAIN) sector_list[minifat_start].name = "!MiniFAT";
|
||||
sector_list[fat_addrs[0]].name = "!FAT";
|
||||
|
||||
17
bits/44_findpath.js
Normal file
17
bits/44_findpath.js
Normal file
@ -0,0 +1,17 @@
|
||||
/* [MS-CFB] 2.6.4 */
|
||||
function make_find_path(FullPaths, Paths, FileIndex, files, root_name) {
|
||||
var UCFullPaths = new Array(FullPaths.length);
|
||||
var UCPaths = new Array(Paths.length), i;
|
||||
for(i = 0; i < FullPaths.length; ++i) UCFullPaths[i] = FullPaths[i].toUpperCase();
|
||||
for(i = 0; i < Paths.length; ++i) UCPaths[i] = Paths[i].toUpperCase();
|
||||
return function find_path(path) {
|
||||
var k;
|
||||
if(path.charCodeAt(0) === 47 /* "/" */) { k=true; path = root_name + path; }
|
||||
else k = path.indexOf("/") !== -1;
|
||||
var UCPath = path.toUpperCase();
|
||||
var w = k === true ? UCFullPaths.indexOf(UCPath) : UCPaths.indexOf(UCPath);
|
||||
if(w === -1) return null;
|
||||
return k === true ? FileIndex[w] : files[Paths[w]];
|
||||
};
|
||||
}
|
||||
|
||||
@ -1,59 +0,0 @@
|
||||
/* [MS-CFB] 2.6.1 Compound File Directory Entry */
|
||||
var files = {}, Paths = [], FileIndex = [], FullPaths = [], FullPathDir = {};
|
||||
function read_directory(idx) {
|
||||
var blob, read, w;
|
||||
var sector = sector_list[idx].data;
|
||||
for(var i = 0; i != sector.length; i+= 128) {
|
||||
blob = sector.slice(i, i+128);
|
||||
prep_blob(blob, 64);
|
||||
read = ReadShift.bind(blob);
|
||||
var namelen = read(2);
|
||||
if(namelen === 0) return;
|
||||
var name = __utf16le(blob,0,namelen-(Paths.length?2:0)); // OLE
|
||||
Paths.push(name);
|
||||
var o = { name: name };
|
||||
o.type = EntryTypes[read(1)];
|
||||
o.color = read(1);
|
||||
o.left = read(4); if(o.left === NOSTREAM) delete o.left;
|
||||
o.right = read(4); if(o.right === NOSTREAM) delete o.right;
|
||||
o.child = read(4); if(o.child === NOSTREAM) delete o.child;
|
||||
o.clsid = read(16);
|
||||
o.state = read(4);
|
||||
var ctime = read(8); if(ctime != "0000000000000000") o.ctime = ctime;
|
||||
var mtime = read(8); if(mtime != "0000000000000000") o.mtime = mtime;
|
||||
o.start = read(4);
|
||||
o.size = read(4);
|
||||
if(o.type === 'root') { //root entry
|
||||
minifat_store = o.start;
|
||||
if(nmfs > 0 && minifat_store !== ENDOFCHAIN) sector_list[minifat_store].name = "!StreamData";
|
||||
minifat_size = o.size;
|
||||
} else if(o.size >= ms_cutoff_size) {
|
||||
o.storage = 'fat';
|
||||
if(!sector_list[o.start] && dir_start > 0) o.start = (o.start + dir_start) % sectors.length;
|
||||
sector_list[o.start].name = o.name;
|
||||
o.content = sector_list[o.start].data.slice(0,o.size);
|
||||
prep_blob(o.content);
|
||||
} else {
|
||||
o.storage = 'minifat';
|
||||
w = o.start * mssz;
|
||||
if(minifat_store !== ENDOFCHAIN && o.start !== ENDOFCHAIN) {
|
||||
o.content = sector_list[minifat_store].data.slice(w,w+o.size);
|
||||
prep_blob(o.content);
|
||||
}
|
||||
}
|
||||
if(o.ctime) {
|
||||
var ct = blob.slice(blob.l-24, blob.l-16);
|
||||
var c2 = (__readUInt32LE(ct,4)/1e7)*Math.pow(2,32)+__readUInt32LE(ct,0)/1e7;
|
||||
o.ct = new Date((c2 - 11644473600)*1000);
|
||||
}
|
||||
if(o.mtime) {
|
||||
var mt = blob.slice(blob.l-16, blob.l-8);
|
||||
var m2 = (__readUInt32LE(mt,4)/1e7)*Math.pow(2,32)+__readUInt32LE(mt,0)/1e7;
|
||||
o.mt = new Date((m2 - 11644473600)*1000);
|
||||
}
|
||||
files[name] = o;
|
||||
FileIndex.push(o);
|
||||
}
|
||||
}
|
||||
read_directory(dir_start);
|
||||
|
||||
@ -1,42 +0,0 @@
|
||||
/* [MS-CFB] 2.6.4 Red-Black Tree */
|
||||
function build_full_paths(Dir, pathobj, paths, patharr) {
|
||||
var i;
|
||||
var dad = new Array(patharr.length);
|
||||
|
||||
var q = new Array(patharr.length);
|
||||
|
||||
for(i=0; i != dad.length; ++i) { dad[i]=q[i]=i; paths[i]=patharr[i]; }
|
||||
|
||||
while(q.length > 0) {
|
||||
for(i = q[0]; typeof i !== "undefined"; i = q.shift()) {
|
||||
if(dad[i] === i) {
|
||||
if(Dir[i].left && dad[Dir[i].left] != Dir[i].left) dad[i] = dad[Dir[i].left];
|
||||
if(Dir[i].right && dad[Dir[i].right] != Dir[i].right) dad[i] = dad[Dir[i].right];
|
||||
}
|
||||
if(Dir[i].child) dad[Dir[i].child] = i;
|
||||
if(Dir[i].left) { dad[Dir[i].left] = dad[i]; q.push(Dir[i].left); }
|
||||
if(Dir[i].right) { dad[Dir[i].right] = dad[i]; q.push(Dir[i].right); }
|
||||
}
|
||||
for(i=1; i != dad.length; ++i) if(dad[i] === i) {
|
||||
if(Dir[i].right && dad[Dir[i].right] != Dir[i].right) dad[i] = dad[Dir[i].right];
|
||||
else if(Dir[i].left && dad[Dir[i].left] != Dir[i].left) dad[i] = dad[Dir[i].left];
|
||||
}
|
||||
}
|
||||
|
||||
for(i=1; i !== paths.length; ++i) {
|
||||
if(Dir[i].type === "unknown") continue;
|
||||
var j = dad[i];
|
||||
if(j === 0) paths[i] = paths[0] + "/" + paths[i];
|
||||
else while(j !== 0) {
|
||||
paths[i] = paths[j] + "/" + paths[i];
|
||||
j = dad[j];
|
||||
}
|
||||
dad[i] = 0;
|
||||
}
|
||||
|
||||
paths[0] += "/";
|
||||
for(i=1; i !== paths.length; ++i) if(Dir[i].type !== 'stream') paths[i] += "/";
|
||||
for(i=0; i !== paths.length; ++i) pathobj[paths[i]] = FileIndex[i];
|
||||
}
|
||||
build_full_paths(FileIndex, FullPathDir, FullPaths, Paths);
|
||||
|
||||
40
bits/45_readfat.js
Normal file
40
bits/45_readfat.js
Normal file
@ -0,0 +1,40 @@
|
||||
/** Chase down the rest of the DIFAT chain to build a comprehensive list
|
||||
DIFAT chains by storing the next sector number as the last 32 bytes */
|
||||
function sleuth_fat(idx, cnt, sectors, ssz, fat_addrs) {
|
||||
var q;
|
||||
if(idx === ENDOFCHAIN) {
|
||||
if(cnt !== 0) throw "DIFAT chain shorter than expected";
|
||||
} else if(idx !== -1 /*FREESECT*/) {
|
||||
var sector = sectors[idx], m = (ssz>>>2)-1;
|
||||
for(var i = 0; i < m; ++i) {
|
||||
if((q = __readInt32LE(sector,i*4)) === ENDOFCHAIN) break;
|
||||
fat_addrs.push(q);
|
||||
}
|
||||
sleuth_fat(__readInt32LE(sector,ssz-4),cnt - 1, sectors, ssz, fat_addrs);
|
||||
}
|
||||
}
|
||||
|
||||
/** Chase down the sector linked lists */
|
||||
function make_sector_list(sectors, dir_start, fat_addrs, ssz) {
|
||||
var sl = sectors.length, sector_list = new Array(sl);
|
||||
var chkd = new Array(sl), buf, buf_chain;
|
||||
var modulus = ssz - 1, i, j, k, jj;
|
||||
for(i=0; i < sl; ++i) {
|
||||
buf = [];
|
||||
k = (i + dir_start); if(k >= sl) k-=sl;
|
||||
if(chkd[k] === true) continue;
|
||||
buf_chain = [];
|
||||
for(j=k; j>=0;) {
|
||||
chkd[j] = true;
|
||||
buf[buf.length] = j;
|
||||
buf_chain.push(sectors[j]);
|
||||
var addr = fat_addrs[Math.floor(j*4/ssz)];
|
||||
jj = ((j*4) & modulus);
|
||||
if(ssz < 4 + jj) throw "FAT boundary crossed: " + j + " 4 "+ssz;
|
||||
j = __readInt32LE(sectors[addr], jj);
|
||||
}
|
||||
sector_list[k] = {nodes: buf, data:__toBuffer([buf_chain])};
|
||||
}
|
||||
return sector_list;
|
||||
}
|
||||
|
||||
@ -1,13 +0,0 @@
|
||||
var root_name = Paths.shift();
|
||||
Paths.root = root_name;
|
||||
|
||||
/* [MS-CFB] 2.6.4 (Unicode 3.0.1 case conversion) */
|
||||
function find_path(path) {
|
||||
if(path[0] === "/") path = root_name + path;
|
||||
var UCNames = (path.indexOf("/") !== -1 ? FullPaths : Paths).map(function(x) { return x.toUpperCase(); });
|
||||
var UCPath = path.toUpperCase();
|
||||
var w = UCNames.indexOf(UCPath);
|
||||
if(w === -1) return null;
|
||||
return path.indexOf("/") !== -1 ? FileIndex[w] : files[Paths[w]];
|
||||
}
|
||||
|
||||
59
bits/46_readdir.js
Normal file
59
bits/46_readdir.js
Normal file
@ -0,0 +1,59 @@
|
||||
/* [MS-CFB] 2.6.1 Compound File Directory Entry */
|
||||
function read_directory(dir_start, sector_list, sectors, Paths, nmfs, files, FileIndex) {
|
||||
var blob;
|
||||
var minifat_store = 0, pl = (Paths.length?2:0);
|
||||
var sector = sector_list[dir_start].data;
|
||||
var i = 0, namelen = 0, name, o, ctime, mtime;
|
||||
for(; i < sector.length; i+= 128) {
|
||||
blob = sector.slice(i, i+128);
|
||||
prep_blob(blob, 64);
|
||||
namelen = blob.read_shift(2);
|
||||
if(namelen === 0) continue;
|
||||
name = __utf16le(blob,0,namelen-pl);
|
||||
Paths.push(name);
|
||||
o = {
|
||||
name: name,
|
||||
type: blob.read_shift(1),
|
||||
color: blob.read_shift(1),
|
||||
L: blob.read_shift(4, 'i'),
|
||||
R: blob.read_shift(4, 'i'),
|
||||
C: blob.read_shift(4, 'i'),
|
||||
clsid: blob.read_shift(16),
|
||||
state: blob.read_shift(4, 'i')
|
||||
};
|
||||
ctime = blob.read_shift(2) + blob.read_shift(2) + blob.read_shift(2) + blob.read_shift(2);
|
||||
if(ctime !== 0) {
|
||||
o.ctime = ctime; o.ct = read_date(blob, blob.l-8);
|
||||
}
|
||||
mtime = blob.read_shift(2) + blob.read_shift(2) + blob.read_shift(2) + blob.read_shift(2);
|
||||
if(mtime !== 0) {
|
||||
o.mtime = mtime; o.mt = read_date(blob, blob.l-8);
|
||||
}
|
||||
o.start = blob.read_shift(4, 'i');
|
||||
o.size = blob.read_shift(4, 'i');
|
||||
if(o.type === 5) { /* root */
|
||||
minifat_store = o.start;
|
||||
if(nmfs > 0 && minifat_store !== ENDOFCHAIN) sector_list[minifat_store].name = "!StreamData";
|
||||
/*minifat_size = o.size;*/
|
||||
} else if(o.size >= 4096 /* MSCSZ */) {
|
||||
o.storage = 'fat';
|
||||
if(sector_list[o.start] === undefined) if((o.start+=dir_start)>=sectors.length) o.start-=sectors.length;
|
||||
sector_list[o.start].name = o.name;
|
||||
o.content = sector_list[o.start].data.slice(0,o.size);
|
||||
prep_blob(o.content, 0);
|
||||
} else {
|
||||
o.storage = 'minifat';
|
||||
if(minifat_store !== ENDOFCHAIN && o.start !== ENDOFCHAIN) {
|
||||
o.content = sector_list[minifat_store].data.slice(o.start*MSSZ,o.start*MSSZ+o.size);
|
||||
prep_blob(o.content, 0);
|
||||
}
|
||||
}
|
||||
files[name] = o;
|
||||
FileIndex.push(o);
|
||||
}
|
||||
}
|
||||
|
||||
function read_date(blob, offset) {
|
||||
return new Date(( ( (__readUInt32LE(blob,offset+4)/1e7)*Math.pow(2,32)+__readUInt32LE(blob,offset)/1e7 ) - 11644473600)*1000);
|
||||
}
|
||||
|
||||
@ -1,11 +0,0 @@
|
||||
var rval = {
|
||||
raw: {header: header, sectors: sectors},
|
||||
FileIndex: FileIndex,
|
||||
FullPaths: FullPaths,
|
||||
FullPathDir: FullPathDir,
|
||||
find: find_path
|
||||
};
|
||||
|
||||
return rval;
|
||||
} // parse
|
||||
|
||||
@ -1,17 +1,14 @@
|
||||
|
||||
var fs;
|
||||
function readFileSync(filename) {
|
||||
var fs = require('fs');
|
||||
var file = fs.readFileSync(filename);
|
||||
return parse(file);
|
||||
if(fs === undefined) fs = require('fs');
|
||||
return parse(fs.readFileSync(filename));
|
||||
}
|
||||
|
||||
function readSync(blob, options) {
|
||||
var o = options || {};
|
||||
switch((o.type || "base64")) {
|
||||
switch(options !== undefined && options.type !== undefined ? options.type : "base64") {
|
||||
case "file": return readFileSync(blob);
|
||||
case "base64": blob = Base64.decode(blob);
|
||||
/* falls through */
|
||||
case "binary": blob = s2a(blob); break;
|
||||
case "base64": return parse(s2a(Base64.decode(blob)));
|
||||
case "binary": return parse(s2a(blob));
|
||||
}
|
||||
return parse(blob);
|
||||
}
|
||||
|
||||
25
bits/75_consts.js
Normal file
25
bits/75_consts.js
Normal file
@ -0,0 +1,25 @@
|
||||
/** CFB Constants */
|
||||
var MSSZ = 64; /* Mini Sector Size = 1<<6 */
|
||||
//var MSCSZ = 4096; /* Mini Stream Cutoff Size */
|
||||
/* 2.1 Compound File Sector Numbers and Types */
|
||||
var ENDOFCHAIN = -2;
|
||||
/* 2.2 Compound File Header */
|
||||
var HEADER_SIGNATURE = 'd0cf11e0a1b11ae1';
|
||||
var HEADER_CLSID = '00000000000000000000000000000000';
|
||||
var consts = {
|
||||
/* 2.1 Compund File Sector Numbers and Types */
|
||||
MAXREGSECT: -6,
|
||||
DIFSECT: -4,
|
||||
FATSECT: -3,
|
||||
ENDOFCHAIN: ENDOFCHAIN,
|
||||
FREESECT: -1,
|
||||
/* 2.2 Compound File Header */
|
||||
HEADER_SIGNATURE: HEADER_SIGNATURE,
|
||||
HEADER_MINOR_VERSION: '3e00',
|
||||
MAXREGSID: -6,
|
||||
NOSTREAM: -1,
|
||||
HEADER_CLSID: HEADER_CLSID,
|
||||
/* 2.6.1 Compound File Directory Entry */
|
||||
EntryTypes: ['unknown','storage','stream','lockbytes','property','root']
|
||||
};
|
||||
|
||||
@ -1,3 +1,11 @@
|
||||
this.read = readSync;
|
||||
this.parse = parse;
|
||||
return this;
|
||||
exports.read = readSync;
|
||||
exports.parse = parse;
|
||||
exports.utils = {
|
||||
ReadShift: ReadShift,
|
||||
CheckField: CheckField,
|
||||
prep_blob: prep_blob,
|
||||
bconcat: bconcat,
|
||||
consts: consts
|
||||
};
|
||||
|
||||
return exports;
|
||||
|
||||
@ -1,18 +0,0 @@
|
||||
/** CFB Constants */
|
||||
{
|
||||
/* 2.1 Compund File Sector Numbers and Types */
|
||||
var MAXREGSECT = 0xFFFFFFFA;
|
||||
var DIFSECT = 0xFFFFFFFC;
|
||||
var FATSECT = 0xFFFFFFFD;
|
||||
var ENDOFCHAIN = 0xFFFFFFFE;
|
||||
var FREESECT = 0xFFFFFFFF;
|
||||
/* 2.2 Compound File Header */
|
||||
var HEADER_SIGNATURE = 'd0cf11e0a1b11ae1';
|
||||
var HEADER_MINOR_VERSION = '3e00';
|
||||
var MAXREGSID = 0xFFFFFFFA;
|
||||
var NOSTREAM = 0xFFFFFFFF;
|
||||
var HEADER_CLSID = '00000000000000000000000000000000';
|
||||
/* 2.6.1 Compound File Directory Entry */
|
||||
var EntryTypes = ['unknown','storage','stream','lockbytes','property','root'];
|
||||
}
|
||||
|
||||
@ -1,7 +0,0 @@
|
||||
var CFB_utils = {
|
||||
ReadShift: ReadShift,
|
||||
CheckField: CheckField,
|
||||
prep_blob: prep_blob,
|
||||
bconcat: bconcat
|
||||
};
|
||||
|
||||
@ -1,7 +1 @@
|
||||
if(typeof require !== 'undefined' && typeof exports !== 'undefined') {
|
||||
var fs = require('fs');
|
||||
exports.read = CFB.read;
|
||||
exports.parse = CFB.parse;
|
||||
exports.utils = CFB_utils;
|
||||
exports.version = CFB.version;
|
||||
}
|
||||
if(typeof require !== 'undefined' && typeof module !== 'undefined' && typeof DO_NOT_EXPORT_CFB === 'undefined') { module.exports = CFB; }
|
||||
|
||||
640
cfb.js
640
cfb.js
@ -5,450 +5,436 @@
|
||||
var Base64 = (function(){
|
||||
var map = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=";
|
||||
return {
|
||||
/* (will need this for writing) encode: function(input, utf8) {
|
||||
var o = "";
|
||||
var c1, c2, c3, e1, e2, e3, e4;
|
||||
for(var i = 0; i < input.length; ) {
|
||||
c1 = input.charCodeAt(i++);
|
||||
c2 = input.charCodeAt(i++);
|
||||
c3 = input.charCodeAt(i++);
|
||||
e1 = c1 >> 2;
|
||||
e2 = (c1 & 3) << 4 | c2 >> 4;
|
||||
e3 = (c2 & 15) << 2 | c3 >> 6;
|
||||
e4 = c3 & 63;
|
||||
if (isNaN(c2)) { e3 = e4 = 64; }
|
||||
else if (isNaN(c3)) { e4 = 64; }
|
||||
o += map.charAt(e1) + map.charAt(e2) + map.charAt(e3) + map.charAt(e4);
|
||||
}
|
||||
return o;
|
||||
},*/
|
||||
decode: function(input, utf8) {
|
||||
decode: function(input) {
|
||||
var o = "";
|
||||
var c1, c2, c3;
|
||||
var e1, e2, e3, e4;
|
||||
input = input.replace(/[^A-Za-z0-9\+\/\=]/g, "");
|
||||
input = input.replace(/[^\w\+\/\=]/g, "");
|
||||
for(var i = 0; i < input.length;) {
|
||||
e1 = map.indexOf(input.charAt(i++));
|
||||
e2 = map.indexOf(input.charAt(i++));
|
||||
e3 = map.indexOf(input.charAt(i++));
|
||||
e4 = map.indexOf(input.charAt(i++));
|
||||
c1 = e1 << 2 | e2 >> 4;
|
||||
c2 = (e2 & 15) << 4 | e3 >> 2;
|
||||
c3 = (e3 & 3) << 6 | e4;
|
||||
c1 = (e1 << 2) | (e2 >> 4);
|
||||
o += String.fromCharCode(c1);
|
||||
if (e3 != 64) { o += String.fromCharCode(c2); }
|
||||
if (e4 != 64) { o += String.fromCharCode(c3); }
|
||||
|
||||
e3 = map.indexOf(input.charAt(i++));
|
||||
c2 = ((e2 & 15) << 4) | (e3 >> 2);
|
||||
if (e3 !== 64) { o += String.fromCharCode(c2); }
|
||||
|
||||
e4 = map.indexOf(input.charAt(i++));
|
||||
c3 = ((e3 & 3) << 6) | e4;
|
||||
if (e4 !== 64) { o += String.fromCharCode(c3); }
|
||||
}
|
||||
return o;
|
||||
}
|
||||
};
|
||||
})();
|
||||
|
||||
function s2a(s) {
|
||||
if(typeof Buffer !== 'undefined') return new Buffer(s, "binary");
|
||||
var w = s.split("").map(function(x){ return x.charCodeAt(0) & 0xff; });
|
||||
return w;
|
||||
}
|
||||
var chr0 = /\u0000/g, chr1 = /[\u0001-\u0006]/;
|
||||
|
||||
var s2a, _s2a;
|
||||
s2a = _s2a = function _s2a(s) { return s.split("").map(function(x){ return x.charCodeAt(0) & 0xff; }); };
|
||||
var __toBuffer, ___toBuffer;
|
||||
__toBuffer = ___toBuffer = function(bufs) { var x = []; for(var i = 0; i < bufs[0].length; ++i) { x.push.apply(x, bufs[0][i]); } return x; };
|
||||
var __utf16le, ___utf16le;
|
||||
__utf16le = ___utf16le = function(b,s,e) { var ss=[]; for(var i=s; i<e; i+=2) ss.push(String.fromCharCode(__readUInt16LE(b,i))); return ss.join("").replace(chr0,'').replace(chr1,'!'); };
|
||||
var __hexlify, ___hexlify;
|
||||
__hexlify = ___hexlify = function(b,s,l) { return b.slice(s,(s+l)).map(function(x){return (x<16?"0":"") + x.toString(16);}).join(""); };
|
||||
var bconcat = function(bufs) { return [].concat.apply([], bufs); };
|
||||
|
||||
|
||||
var __toBuffer;
|
||||
if(typeof Buffer !== "undefined") {
|
||||
Buffer.prototype.hexlify= function() { return this.toString('hex'); };
|
||||
Buffer.prototype.utf16le= function(s,e){return this.toString('utf16le',s,e).replace(/\u0000/,'').replace(/[\u0001-\u0006]/,'!');};
|
||||
Buffer.prototype.utf8 = function(s,e) { return this.toString('utf8',s,e); };
|
||||
__toBuffer = function(bufs) { return Buffer.concat(bufs[0]); };
|
||||
} else {
|
||||
__toBuffer = function(bufs) {
|
||||
var x = [];
|
||||
for(var i = 0; i != bufs[0].length; ++i) { x = x.concat(bufs[0][i]); }
|
||||
return x;
|
||||
__utf16le = function(b,s,e) {
|
||||
if(!Buffer.isBuffer(b)) return ___utf16le(b,s,e);
|
||||
return b.toString('utf16le',s,e).replace(chr0,'').replace(chr1,'!');
|
||||
};
|
||||
__hexlify = function(b,s,l) { return Buffer.isBuffer(b) ? b.toString('hex',s,s+l) : ___hexlify(b,s,l); };
|
||||
__toBuffer = function(bufs) { return (bufs[0].length > 0 && Buffer.isBuffer(bufs[0][0])) ? Buffer.concat(bufs[0]) : ___toBuffer(bufs);};
|
||||
s2a = function(s) { return Buffer(s, "binary"); };
|
||||
bconcat = function(bufs) { return Buffer.isBuffer(bufs[0]) ? Buffer.concat(bufs) : [].concat.apply([], bufs); };
|
||||
}
|
||||
|
||||
var __readUInt8 = function(b, idx) { return b.readUInt8 ? b.readUInt8(idx) : b[idx]; };
|
||||
var __readUInt16LE = function(b, idx) { return b.readUInt16LE ? b.readUInt16LE(idx) : b[idx+1]*(1<<8)+b[idx]; };
|
||||
var __readInt16LE = function(b, idx) { var u = __readUInt16LE(b,idx); if(!(u & 0x8000)) return u; return (0xffff - u + 1) * -1; };
|
||||
var __readUInt32LE = function(b, idx) { return b.readUInt32LE ? b.readUInt32LE(idx) : b[idx+3]*(1<<24)+b[idx+2]*(1<<16)+b[idx+1]*(1<<8)+b[idx]; };
|
||||
var __readInt32LE = function(b, idx) { if(b.readInt32LE) return b.readInt32LE(idx); var u = __readUInt32LE(b,idx); if(!(u & 0x80000000)) return u; return (0xffffffff - u + 1) * -1; };
|
||||
|
||||
var __hexlify = function(b) { return b.map(function(x){return (x<16?"0":"") + x.toString(16);}).join(""); };
|
||||
var __readUInt8 = function(b, idx) { return b[idx]; };
|
||||
var __readUInt16LE = function(b, idx) { return b[idx+1]*(1<<8)+b[idx]; };
|
||||
var __readInt16LE = function(b, idx) { var u = b[idx+1]*(1<<8)+b[idx]; return (u < 0x8000) ? u : (0xffff - u + 1) * -1; };
|
||||
var __readUInt32LE = function(b, idx) { return b[idx+3]*(1<<24)+(b[idx+2]<<16)+(b[idx+1]<<8)+b[idx]; };
|
||||
var __readInt32LE = function(b, idx) { return (b[idx+3]<<24)+(b[idx+2]<<16)+(b[idx+1]<<8)+b[idx]; };
|
||||
|
||||
var __utf16le = function(b,s,e) { if(b.utf16le) return b.utf16le(s,e); var ss=[]; for(var i=s; i<e; i+=2) ss.push(String.fromCharCode(__readUInt16LE(b,i))); return ss.join("").replace(/\u0000/,'').replace(/[\u0001-\u0006]/,'!'); };
|
||||
|
||||
var __utf8 = function(b,s,e) { if(b.utf8) return b.utf8(s,e); var ss=[]; for(var i=s; i<e; i++) ss.push(String.fromCharCode(__readUInt8(b,i))); return ss.join(""); };
|
||||
|
||||
function bconcat(bufs) { return (typeof Buffer !== 'undefined') ? Buffer.concat(bufs) : [].concat.apply([], bufs); }
|
||||
|
||||
/** Buffer helpers -- keep track of read location `.l` and move it */
|
||||
function ReadShift(size, t) {
|
||||
var o, oo=[], w, vv; t = t || 'u';
|
||||
var oI, oS, type = 0;
|
||||
switch(size) {
|
||||
case 1: o = __readUInt8(this, this.l); break;
|
||||
case 2: o=(t==='u' ? __readUInt16LE : __readInt16LE)(this, this.l); break;
|
||||
case 4: o = __readUInt32LE(this, this.l); break;
|
||||
case 8:
|
||||
case 16: o = this.toString('hex', this.l,this.l+size); break;
|
||||
|
||||
case 'utf8': size = t; o = __utf8(this, this.l, this.l + size); break;
|
||||
case 'utf16le': size=2*t; o = __utf16le(this, this.l, this.l + size); break;
|
||||
|
||||
case 'cstr': size = 0; o = "";
|
||||
while((w=__readUInt8(this, this.l + size++))!==0) oo.push(String.fromCharCode(w));
|
||||
o = oo.join(""); break;
|
||||
case 'wstr': size = 0; o = "";
|
||||
while((w=__readUInt16LE(this,this.l +size))!==0){oo.push(String.fromCharCode(w));size+=2;}
|
||||
size+=2; break;
|
||||
case 1: oI = __readUInt8(this, this.l); break;
|
||||
case 2: oI = (t !== 'i' ? __readUInt16LE : __readInt16LE)(this, this.l); break;
|
||||
case 4: oI = __readInt32LE(this, this.l); break;
|
||||
case 16: type = 2; oS = __hexlify(this, this.l, size);
|
||||
}
|
||||
this.l+=size; return o;
|
||||
this.l+=size; if(type === 0) return oI; return oS;
|
||||
}
|
||||
|
||||
function CheckField(hexstr, fld) {
|
||||
var b = this.slice(this.l, this.l+hexstr.length/2);
|
||||
var m = b.hexlify ? b.hexlify() : __hexlify(b);
|
||||
if(m !== hexstr) throw (fld||"") + 'Expected ' + hexstr + ' saw ' + m;
|
||||
this.l += hexstr.length/2;
|
||||
var m = __hexlify(this,this.l,hexstr.length>>1);
|
||||
if(m !== hexstr) throw fld + 'Expected ' + hexstr + ' saw ' + m;
|
||||
this.l += hexstr.length>>1;
|
||||
}
|
||||
|
||||
function prep_blob(blob, pos) {
|
||||
blob.read_shift = ReadShift.bind(blob);
|
||||
blob.l = pos;
|
||||
blob.read_shift = ReadShift;
|
||||
blob.chk = CheckField;
|
||||
blob.l = pos || 0;
|
||||
var read = ReadShift.bind(blob), chk = CheckField.bind(blob);
|
||||
return [read, chk];
|
||||
}
|
||||
|
||||
/* [MS-CFB] v20130118 */
|
||||
var CFB = (function(){
|
||||
this.version = '0.9.1';
|
||||
var CFB = (function _CFB(){
|
||||
var exports = {};
|
||||
exports.version = '0.10.0';
|
||||
function parse(file) {
|
||||
|
||||
|
||||
var mver = 3; // major version
|
||||
var ssz = 512; // sector size
|
||||
var mssz = 64; // mini sector size
|
||||
var nds = 0; // number of directory sectors
|
||||
var nfs = 0; // number of FAT sectors
|
||||
var nmfs = 0; // number of mini FAT sectors
|
||||
var ndfs = 0; // number of DIFAT sectors
|
||||
var dir_start = 0; // first directory sector location
|
||||
var minifat_start = 0; // first mini FAT sector location
|
||||
var difat_start = 0; // first mini FAT sector location
|
||||
|
||||
var ms_cutoff_size = 4096; // mini stream cutoff size
|
||||
var minifat_store = 0; // first sector with minifat data
|
||||
var minifat_size = 0; // size of minifat data
|
||||
|
||||
var fat_addrs = []; // locations of FAT sectors
|
||||
|
||||
/* [MS-CFB] 2.2 Compound File Header */
|
||||
var blob = file.slice(0,512);
|
||||
prep_blob(blob);
|
||||
var read = ReadShift.bind(blob), chk = CheckField.bind(blob);
|
||||
var j = 0, q;
|
||||
prep_blob(blob, 0);
|
||||
|
||||
// header signature 8
|
||||
chk(HEADER_SIGNATURE, 'Header Signature: ');
|
||||
|
||||
// clsid 16
|
||||
chk(HEADER_CLSID, 'CLSID: ');
|
||||
|
||||
// minor version 2
|
||||
read(2);
|
||||
|
||||
// major version 3
|
||||
mver = read(2);
|
||||
/* major version */
|
||||
mver = check_get_mver(blob);
|
||||
switch(mver) {
|
||||
case 3: ssz = 512; break;
|
||||
case 4: ssz = 4096; break;
|
||||
case 3: ssz = 512; break; case 4: ssz = 4096; break;
|
||||
default: throw "Major Version: Expected 3 or 4 saw " + mver;
|
||||
}
|
||||
|
||||
// reprocess header
|
||||
var pos = blob.l;
|
||||
blob = file.slice(0,ssz);
|
||||
prep_blob(blob,pos);
|
||||
read = ReadShift.bind(blob);
|
||||
chk = CheckField.bind(blob);
|
||||
/* reprocess header */
|
||||
if(ssz !== 512) { blob = file.slice(0,ssz); prep_blob(blob, 28 /* blob.l */); }
|
||||
/* Save header for final object */
|
||||
var header = file.slice(0,ssz);
|
||||
|
||||
// Byte Order TODO
|
||||
chk('feff', 'Byte Order: ');
|
||||
|
||||
// Sector Shift
|
||||
switch((q = read(2))) {
|
||||
case 0x09: if(mver !== 3) throw 'MajorVersion/SectorShift Mismatch'; break;
|
||||
case 0x0c: if(mver !== 4) throw 'MajorVersion/SectorShift Mismatch'; break;
|
||||
default: throw 'Sector Shift: Expected 9 or 12 saw ' + q;
|
||||
}
|
||||
|
||||
// Mini Sector Shift
|
||||
chk('0600', 'Mini Sector Shift: ');
|
||||
|
||||
// Reserved
|
||||
chk('000000000000', 'Mini Sector Shift: ');
|
||||
check_shifts(blob, mver);
|
||||
|
||||
// Number of Directory Sectors
|
||||
nds = read(4);
|
||||
var nds = blob.read_shift(4, 'i');
|
||||
if(mver === 3 && nds !== 0) throw '# Directory Sectors: Expected 0 saw ' + nds;
|
||||
|
||||
// Number of FAT Sectors
|
||||
nfs = read(4);
|
||||
//var nfs = blob.read_shift(4, 'i');
|
||||
blob.l += 4;
|
||||
|
||||
// First Directory Sector Location
|
||||
dir_start = read(4);
|
||||
dir_start = blob.read_shift(4, 'i');
|
||||
|
||||
// Transaction Signature TODO
|
||||
read(4);
|
||||
// Transaction Signature
|
||||
blob.l += 4;
|
||||
|
||||
// Mini Stream Cutoff Size TODO
|
||||
chk('00100000', 'Mini Stream Cutoff Size: ');
|
||||
// Mini Stream Cutoff Size
|
||||
blob.chk('00100000', 'Mini Stream Cutoff Size: ');
|
||||
|
||||
// First Mini FAT Sector Location
|
||||
minifat_start = read(4);
|
||||
minifat_start = blob.read_shift(4, 'i');
|
||||
|
||||
// Number of Mini FAT Sectors
|
||||
nmfs = read(4);
|
||||
nmfs = blob.read_shift(4, 'i');
|
||||
|
||||
// First DIFAT sector location
|
||||
difat_start = read(4);
|
||||
difat_start = blob.read_shift(4, 'i');
|
||||
|
||||
// Number of DIFAT Sectors
|
||||
ndfs = read(4);
|
||||
ndfs = blob.read_shift(4, 'i');
|
||||
|
||||
// Grab FAT Sector Locations
|
||||
for(j = 0; blob.l != 512; ) {
|
||||
if((q = read(4))>=MAXREGSECT) break;
|
||||
fat_addrs[j++] = q;
|
||||
for(var q, j = 0; j < 109; ++j) { /* 109 = (512 - blob.l)>>>2; */
|
||||
q = blob.read_shift(4, 'i');
|
||||
if(q<0) break;
|
||||
fat_addrs[j] = q;
|
||||
}
|
||||
|
||||
|
||||
/** Break the file up into sectors */
|
||||
var nsectors = Math.ceil((file.length - ssz)/ssz);
|
||||
var sectors = [];
|
||||
for(var i=1; i != nsectors; ++i) sectors[i-1] = file.slice(i*ssz,(i+1)*ssz);
|
||||
sectors[nsectors-1] = file.slice(nsectors*ssz);
|
||||
var sectors = sectorify(file, ssz);
|
||||
|
||||
/** Chase down the rest of the DIFAT chain to build a comprehensive list
|
||||
DIFAT chains by storing the next sector number as the last 32 bytes */
|
||||
function sleuth_fat(idx, cnt) {
|
||||
if(idx === ENDOFCHAIN) {
|
||||
if(cnt !== 0) throw "DIFAT chain shorter than expected";
|
||||
return;
|
||||
}
|
||||
if(idx !== FREESECT) {
|
||||
var sector = sectors[idx];
|
||||
for(var i = 0; i != ssz/4-1; ++i) {
|
||||
if((q = __readUInt32LE(sector,i*4)) === ENDOFCHAIN) break;
|
||||
fat_addrs.push(q);
|
||||
}
|
||||
sleuth_fat(__readUInt32LE(sector,ssz-4),cnt - 1);
|
||||
}
|
||||
}
|
||||
sleuth_fat(difat_start, ndfs);
|
||||
|
||||
/** DONT CAT THE FAT! Just calculate where we need to go */
|
||||
function get_buffer(byte_addr, bytes) {
|
||||
var addr = fat_addrs[Math.floor(byte_addr*4/ssz)];
|
||||
if(ssz - (byte_addr*4 % ssz) < (bytes || 0)) throw "FAT boundary crossed: " + byte_addr + " "+bytes+" "+ssz;
|
||||
return sectors[addr].slice((byte_addr*4 % ssz));
|
||||
}
|
||||
|
||||
function get_buffer_u32(byte_addr) {
|
||||
return __readUInt32LE(get_buffer(byte_addr,4), 0);
|
||||
}
|
||||
|
||||
function get_next_sector(idx) { return get_buffer_u32(idx); }
|
||||
sleuth_fat(difat_start, ndfs, sectors, ssz, fat_addrs);
|
||||
|
||||
/** Chains */
|
||||
var chkd = new Array(sectors.length), sector_list = [];
|
||||
var get_sector = function get_sector(k) { return sectors[k]; };
|
||||
for(i=0; i != sectors.length; ++i) {
|
||||
var buf = [], k = (i + dir_start) % sectors.length;
|
||||
if(chkd[k]) continue;
|
||||
for(j=k; j<=MAXREGSECT; buf.push(j),j=get_next_sector(j)) chkd[j] = true;
|
||||
sector_list[k] = {nodes: buf};
|
||||
sector_list[k].data = __toBuffer(Array(buf.map(get_sector)));
|
||||
}
|
||||
var sector_list = make_sector_list(sectors, dir_start, fat_addrs, ssz);
|
||||
|
||||
sector_list[dir_start].name = "!Directory";
|
||||
if(nmfs > 0 && minifat_start !== ENDOFCHAIN) sector_list[minifat_start].name = "!MiniFAT";
|
||||
sector_list[fat_addrs[0]].name = "!FAT";
|
||||
|
||||
/* [MS-CFB] 2.6.1 Compound File Directory Entry */
|
||||
var files = {}, Paths = [], FileIndex = [], FullPaths = [], FullPathDir = {};
|
||||
function read_directory(idx) {
|
||||
var blob, read, w;
|
||||
var sector = sector_list[idx].data;
|
||||
for(var i = 0; i != sector.length; i+= 128) {
|
||||
blob = sector.slice(i, i+128);
|
||||
prep_blob(blob, 64);
|
||||
read = ReadShift.bind(blob);
|
||||
var namelen = read(2);
|
||||
if(namelen === 0) return;
|
||||
var name = __utf16le(blob,0,namelen-(Paths.length?2:0)); // OLE
|
||||
Paths.push(name);
|
||||
var o = { name: name };
|
||||
o.type = EntryTypes[read(1)];
|
||||
o.color = read(1);
|
||||
o.left = read(4); if(o.left === NOSTREAM) delete o.left;
|
||||
o.right = read(4); if(o.right === NOSTREAM) delete o.right;
|
||||
o.child = read(4); if(o.child === NOSTREAM) delete o.child;
|
||||
o.clsid = read(16);
|
||||
o.state = read(4);
|
||||
var ctime = read(8); if(ctime != "0000000000000000") o.ctime = ctime;
|
||||
var mtime = read(8); if(mtime != "0000000000000000") o.mtime = mtime;
|
||||
o.start = read(4);
|
||||
o.size = read(4);
|
||||
if(o.type === 'root') { //root entry
|
||||
minifat_store = o.start;
|
||||
if(nmfs > 0 && minifat_store !== ENDOFCHAIN) sector_list[minifat_store].name = "!StreamData";
|
||||
minifat_size = o.size;
|
||||
} else if(o.size >= ms_cutoff_size) {
|
||||
o.storage = 'fat';
|
||||
if(!sector_list[o.start] && dir_start > 0) o.start = (o.start + dir_start) % sectors.length;
|
||||
sector_list[o.start].name = o.name;
|
||||
o.content = sector_list[o.start].data.slice(0,o.size);
|
||||
prep_blob(o.content);
|
||||
} else {
|
||||
o.storage = 'minifat';
|
||||
w = o.start * mssz;
|
||||
if(minifat_store !== ENDOFCHAIN && o.start !== ENDOFCHAIN) {
|
||||
o.content = sector_list[minifat_store].data.slice(w,w+o.size);
|
||||
prep_blob(o.content);
|
||||
}
|
||||
}
|
||||
if(o.ctime) {
|
||||
var ct = blob.slice(blob.l-24, blob.l-16);
|
||||
var c2 = (__readUInt32LE(ct,4)/1e7)*Math.pow(2,32)+__readUInt32LE(ct,0)/1e7;
|
||||
o.ct = new Date((c2 - 11644473600)*1000);
|
||||
}
|
||||
if(o.mtime) {
|
||||
var mt = blob.slice(blob.l-16, blob.l-8);
|
||||
var m2 = (__readUInt32LE(mt,4)/1e7)*Math.pow(2,32)+__readUInt32LE(mt,0)/1e7;
|
||||
o.mt = new Date((m2 - 11644473600)*1000);
|
||||
}
|
||||
files[name] = o;
|
||||
FileIndex.push(o);
|
||||
}
|
||||
}
|
||||
read_directory(dir_start);
|
||||
read_directory(dir_start, sector_list, sectors, Paths, nmfs, files, FileIndex);
|
||||
|
||||
/* [MS-CFB] 2.6.4 Red-Black Tree */
|
||||
function build_full_paths(Dir, pathobj, paths, patharr) {
|
||||
var i;
|
||||
var dad = new Array(patharr.length);
|
||||
|
||||
var q = new Array(patharr.length);
|
||||
|
||||
for(i=0; i != dad.length; ++i) { dad[i]=q[i]=i; paths[i]=patharr[i]; }
|
||||
|
||||
while(q.length > 0) {
|
||||
for(i = q[0]; typeof i !== "undefined"; i = q.shift()) {
|
||||
if(dad[i] === i) {
|
||||
if(Dir[i].left && dad[Dir[i].left] != Dir[i].left) dad[i] = dad[Dir[i].left];
|
||||
if(Dir[i].right && dad[Dir[i].right] != Dir[i].right) dad[i] = dad[Dir[i].right];
|
||||
}
|
||||
if(Dir[i].child) dad[Dir[i].child] = i;
|
||||
if(Dir[i].left) { dad[Dir[i].left] = dad[i]; q.push(Dir[i].left); }
|
||||
if(Dir[i].right) { dad[Dir[i].right] = dad[i]; q.push(Dir[i].right); }
|
||||
}
|
||||
for(i=1; i != dad.length; ++i) if(dad[i] === i) {
|
||||
if(Dir[i].right && dad[Dir[i].right] != Dir[i].right) dad[i] = dad[Dir[i].right];
|
||||
else if(Dir[i].left && dad[Dir[i].left] != Dir[i].left) dad[i] = dad[Dir[i].left];
|
||||
}
|
||||
}
|
||||
|
||||
for(i=1; i !== paths.length; ++i) {
|
||||
if(Dir[i].type === "unknown") continue;
|
||||
var j = dad[i];
|
||||
if(j === 0) paths[i] = paths[0] + "/" + paths[i];
|
||||
else while(j !== 0) {
|
||||
paths[i] = paths[j] + "/" + paths[i];
|
||||
j = dad[j];
|
||||
}
|
||||
dad[i] = 0;
|
||||
}
|
||||
|
||||
paths[0] += "/";
|
||||
for(i=1; i !== paths.length; ++i) if(Dir[i].type !== 'stream') paths[i] += "/";
|
||||
for(i=0; i !== paths.length; ++i) pathobj[paths[i]] = FileIndex[i];
|
||||
}
|
||||
build_full_paths(FileIndex, FullPathDir, FullPaths, Paths);
|
||||
|
||||
var root_name = Paths.shift();
|
||||
Paths.root = root_name;
|
||||
|
||||
/* [MS-CFB] 2.6.4 (Unicode 3.0.1 case conversion) */
|
||||
function find_path(path) {
|
||||
if(path[0] === "/") path = root_name + path;
|
||||
var UCNames = (path.indexOf("/") !== -1 ? FullPaths : Paths).map(function(x) { return x.toUpperCase(); });
|
||||
var UCPath = path.toUpperCase();
|
||||
var w = UCNames.indexOf(UCPath);
|
||||
if(w === -1) return null;
|
||||
return path.indexOf("/") !== -1 ? FileIndex[w] : files[Paths[w]];
|
||||
}
|
||||
var find_path = make_find_path(FullPaths, Paths, FileIndex, files, root_name);
|
||||
|
||||
var rval = {
|
||||
return {
|
||||
raw: {header: header, sectors: sectors},
|
||||
FileIndex: FileIndex,
|
||||
FullPaths: FullPaths,
|
||||
FullPathDir: FullPathDir,
|
||||
find: find_path
|
||||
};
|
||||
|
||||
return rval;
|
||||
} // parse
|
||||
|
||||
/* [MS-CFB] 2.2 Compound File Header -- read up to major version */
|
||||
function check_get_mver(blob) {
|
||||
// header signature 8
|
||||
blob.chk(HEADER_SIGNATURE, 'Header Signature: ');
|
||||
|
||||
// clsid 16
|
||||
blob.chk(HEADER_CLSID, 'CLSID: ');
|
||||
|
||||
// minor version 2
|
||||
blob.l += 2;
|
||||
|
||||
return blob.read_shift(2,'u');
|
||||
}
|
||||
function check_shifts(blob, mver) {
|
||||
var shift = 0x09;
|
||||
|
||||
// Byte Order
|
||||
blob.chk('feff', 'Byte Order: ');
|
||||
|
||||
// Sector Shift
|
||||
switch((shift = blob.read_shift(2))) {
|
||||
case 0x09: if(mver !== 3) throw 'MajorVersion/SectorShift Mismatch'; break;
|
||||
case 0x0c: if(mver !== 4) throw 'MajorVersion/SectorShift Mismatch'; break;
|
||||
default: throw 'Sector Shift: Expected 9 or 12 saw ' + shift;
|
||||
}
|
||||
|
||||
// Mini Sector Shift
|
||||
blob.chk('0600', 'Mini Sector Shift: ');
|
||||
|
||||
// Reserved
|
||||
blob.chk('000000000000', 'Reserved: ');
|
||||
}
|
||||
|
||||
/** Break the file up into sectors */
|
||||
function sectorify(file, ssz) {
|
||||
var nsectors = Math.ceil(file.length/ssz)-1;
|
||||
var sectors = new Array(nsectors);
|
||||
for(var i=1; i < nsectors; ++i) sectors[i-1] = file.slice(i*ssz,(i+1)*ssz);
|
||||
sectors[nsectors-1] = file.slice(nsectors*ssz);
|
||||
return sectors;
|
||||
}
|
||||
|
||||
/* [MS-CFB] 2.6.4 Red-Black Tree */
|
||||
function build_full_paths(FI, FPD, FP, Paths) {
|
||||
var i = 0, L = 0, R = 0, C = 0, j = 0, pl = Paths.length;
|
||||
var dad = new Array(pl), q = new Array(pl);
|
||||
|
||||
for(; i < pl; ++i) { dad[i]=q[i]=i; FP[i]=Paths[i]; }
|
||||
|
||||
for(; j < q.length; ++j) {
|
||||
i = q[j];
|
||||
L = FI[i].L; R = FI[i].R; C = FI[i].C;
|
||||
if(dad[i] === i) {
|
||||
if(L !== -1 /*NOSTREAM*/ && dad[L] !== L) dad[i] = dad[L];
|
||||
if(R !== -1 && dad[R] !== R) dad[i] = dad[R];
|
||||
}
|
||||
if(C !== -1 /*NOSTREAM*/) dad[C] = i;
|
||||
if(L !== -1) { dad[L] = dad[i]; q.push(L); }
|
||||
if(R !== -1) { dad[R] = dad[i]; q.push(R); }
|
||||
}
|
||||
for(i=1; i !== pl; ++i) if(dad[i] === i) {
|
||||
if(R !== -1 /*NOSTREAM*/ && dad[R] !== R) dad[i] = dad[R];
|
||||
else if(L !== -1 && dad[L] !== L) dad[i] = dad[L];
|
||||
}
|
||||
|
||||
for(i=1; i < pl; ++i) {
|
||||
if(FI[i].type === 0 /* unknown */) continue;
|
||||
j = dad[i];
|
||||
if(j === 0) FP[i] = FP[0] + "/" + FP[i];
|
||||
else while(j !== 0) {
|
||||
FP[i] = FP[j] + "/" + FP[i];
|
||||
j = dad[j];
|
||||
}
|
||||
dad[i] = 0;
|
||||
}
|
||||
|
||||
FP[0] += "/";
|
||||
for(i=1; i < pl; ++i) {
|
||||
if(FI[i].type !== 2 /* stream */) FP[i] += "/";
|
||||
FPD[FP[i]] = FI[i];
|
||||
}
|
||||
}
|
||||
|
||||
/* [MS-CFB] 2.6.4 */
|
||||
function make_find_path(FullPaths, Paths, FileIndex, files, root_name) {
|
||||
var UCFullPaths = new Array(FullPaths.length);
|
||||
var UCPaths = new Array(Paths.length), i;
|
||||
for(i = 0; i < FullPaths.length; ++i) UCFullPaths[i] = FullPaths[i].toUpperCase();
|
||||
for(i = 0; i < Paths.length; ++i) UCPaths[i] = Paths[i].toUpperCase();
|
||||
return function find_path(path) {
|
||||
var k;
|
||||
if(path.charCodeAt(0) === 47 /* "/" */) { k=true; path = root_name + path; }
|
||||
else k = path.indexOf("/") !== -1;
|
||||
var UCPath = path.toUpperCase();
|
||||
var w = k === true ? UCFullPaths.indexOf(UCPath) : UCPaths.indexOf(UCPath);
|
||||
if(w === -1) return null;
|
||||
return k === true ? FileIndex[w] : files[Paths[w]];
|
||||
};
|
||||
}
|
||||
|
||||
/** Chase down the rest of the DIFAT chain to build a comprehensive list
|
||||
DIFAT chains by storing the next sector number as the last 32 bytes */
|
||||
function sleuth_fat(idx, cnt, sectors, ssz, fat_addrs) {
|
||||
var q;
|
||||
if(idx === ENDOFCHAIN) {
|
||||
if(cnt !== 0) throw "DIFAT chain shorter than expected";
|
||||
} else if(idx !== -1 /*FREESECT*/) {
|
||||
var sector = sectors[idx], m = (ssz>>>2)-1;
|
||||
for(var i = 0; i < m; ++i) {
|
||||
if((q = __readInt32LE(sector,i*4)) === ENDOFCHAIN) break;
|
||||
fat_addrs.push(q);
|
||||
}
|
||||
sleuth_fat(__readInt32LE(sector,ssz-4),cnt - 1, sectors, ssz, fat_addrs);
|
||||
}
|
||||
}
|
||||
|
||||
/** Chase down the sector linked lists */
|
||||
function make_sector_list(sectors, dir_start, fat_addrs, ssz) {
|
||||
var sl = sectors.length, sector_list = new Array(sl);
|
||||
var chkd = new Array(sl), buf, buf_chain;
|
||||
var modulus = ssz - 1, i, j, k, jj;
|
||||
for(i=0; i < sl; ++i) {
|
||||
buf = [];
|
||||
k = (i + dir_start); if(k >= sl) k-=sl;
|
||||
if(chkd[k] === true) continue;
|
||||
buf_chain = [];
|
||||
for(j=k; j>=0;) {
|
||||
chkd[j] = true;
|
||||
buf[buf.length] = j;
|
||||
buf_chain.push(sectors[j]);
|
||||
var addr = fat_addrs[Math.floor(j*4/ssz)];
|
||||
jj = ((j*4) & modulus);
|
||||
if(ssz < 4 + jj) throw "FAT boundary crossed: " + j + " 4 "+ssz;
|
||||
j = __readInt32LE(sectors[addr], jj);
|
||||
}
|
||||
sector_list[k] = {nodes: buf, data:__toBuffer([buf_chain])};
|
||||
}
|
||||
return sector_list;
|
||||
}
|
||||
|
||||
/* [MS-CFB] 2.6.1 Compound File Directory Entry */
|
||||
function read_directory(dir_start, sector_list, sectors, Paths, nmfs, files, FileIndex) {
|
||||
var blob;
|
||||
var minifat_store = 0, pl = (Paths.length?2:0);
|
||||
var sector = sector_list[dir_start].data;
|
||||
var i = 0, namelen = 0, name, o, ctime, mtime;
|
||||
for(; i < sector.length; i+= 128) {
|
||||
blob = sector.slice(i, i+128);
|
||||
prep_blob(blob, 64);
|
||||
namelen = blob.read_shift(2);
|
||||
if(namelen === 0) continue;
|
||||
name = __utf16le(blob,0,namelen-pl);
|
||||
Paths.push(name);
|
||||
o = {
|
||||
name: name,
|
||||
type: blob.read_shift(1),
|
||||
color: blob.read_shift(1),
|
||||
L: blob.read_shift(4, 'i'),
|
||||
R: blob.read_shift(4, 'i'),
|
||||
C: blob.read_shift(4, 'i'),
|
||||
clsid: blob.read_shift(16),
|
||||
state: blob.read_shift(4, 'i')
|
||||
};
|
||||
ctime = blob.read_shift(2) + blob.read_shift(2) + blob.read_shift(2) + blob.read_shift(2);
|
||||
if(ctime !== 0) {
|
||||
o.ctime = ctime; o.ct = read_date(blob, blob.l-8);
|
||||
}
|
||||
mtime = blob.read_shift(2) + blob.read_shift(2) + blob.read_shift(2) + blob.read_shift(2);
|
||||
if(mtime !== 0) {
|
||||
o.mtime = mtime; o.mt = read_date(blob, blob.l-8);
|
||||
}
|
||||
o.start = blob.read_shift(4, 'i');
|
||||
o.size = blob.read_shift(4, 'i');
|
||||
if(o.type === 5) { /* root */
|
||||
minifat_store = o.start;
|
||||
if(nmfs > 0 && minifat_store !== ENDOFCHAIN) sector_list[minifat_store].name = "!StreamData";
|
||||
/*minifat_size = o.size;*/
|
||||
} else if(o.size >= 4096 /* MSCSZ */) {
|
||||
o.storage = 'fat';
|
||||
if(sector_list[o.start] === undefined) if((o.start+=dir_start)>=sectors.length) o.start-=sectors.length;
|
||||
sector_list[o.start].name = o.name;
|
||||
o.content = sector_list[o.start].data.slice(0,o.size);
|
||||
prep_blob(o.content, 0);
|
||||
} else {
|
||||
o.storage = 'minifat';
|
||||
if(minifat_store !== ENDOFCHAIN && o.start !== ENDOFCHAIN) {
|
||||
o.content = sector_list[minifat_store].data.slice(o.start*MSSZ,o.start*MSSZ+o.size);
|
||||
prep_blob(o.content, 0);
|
||||
}
|
||||
}
|
||||
files[name] = o;
|
||||
FileIndex.push(o);
|
||||
}
|
||||
}
|
||||
|
||||
function read_date(blob, offset) {
|
||||
return new Date(( ( (__readUInt32LE(blob,offset+4)/1e7)*Math.pow(2,32)+__readUInt32LE(blob,offset)/1e7 ) - 11644473600)*1000);
|
||||
}
|
||||
|
||||
var fs;
|
||||
function readFileSync(filename) {
|
||||
var fs = require('fs');
|
||||
var file = fs.readFileSync(filename);
|
||||
return parse(file);
|
||||
if(fs === undefined) fs = require('fs');
|
||||
return parse(fs.readFileSync(filename));
|
||||
}
|
||||
|
||||
function readSync(blob, options) {
|
||||
var o = options || {};
|
||||
switch((o.type || "base64")) {
|
||||
switch(options !== undefined && options.type !== undefined ? options.type : "base64") {
|
||||
case "file": return readFileSync(blob);
|
||||
case "base64": blob = Base64.decode(blob);
|
||||
/* falls through */
|
||||
case "binary": blob = s2a(blob); break;
|
||||
case "base64": return parse(s2a(Base64.decode(blob)));
|
||||
case "binary": return parse(s2a(blob));
|
||||
}
|
||||
return parse(blob);
|
||||
}
|
||||
|
||||
this.read = readSync;
|
||||
this.parse = parse;
|
||||
return this;
|
||||
})();
|
||||
|
||||
/** CFB Constants */
|
||||
{
|
||||
var MSSZ = 64; /* Mini Sector Size = 1<<6 */
|
||||
//var MSCSZ = 4096; /* Mini Stream Cutoff Size */
|
||||
/* 2.1 Compound File Sector Numbers and Types */
|
||||
var ENDOFCHAIN = -2;
|
||||
/* 2.2 Compound File Header */
|
||||
var HEADER_SIGNATURE = 'd0cf11e0a1b11ae1';
|
||||
var HEADER_CLSID = '00000000000000000000000000000000';
|
||||
var consts = {
|
||||
/* 2.1 Compund File Sector Numbers and Types */
|
||||
var MAXREGSECT = 0xFFFFFFFA;
|
||||
var DIFSECT = 0xFFFFFFFC;
|
||||
var FATSECT = 0xFFFFFFFD;
|
||||
var ENDOFCHAIN = 0xFFFFFFFE;
|
||||
var FREESECT = 0xFFFFFFFF;
|
||||
MAXREGSECT: -6,
|
||||
DIFSECT: -4,
|
||||
FATSECT: -3,
|
||||
ENDOFCHAIN: ENDOFCHAIN,
|
||||
FREESECT: -1,
|
||||
/* 2.2 Compound File Header */
|
||||
var HEADER_SIGNATURE = 'd0cf11e0a1b11ae1';
|
||||
var HEADER_MINOR_VERSION = '3e00';
|
||||
var MAXREGSID = 0xFFFFFFFA;
|
||||
var NOSTREAM = 0xFFFFFFFF;
|
||||
var HEADER_CLSID = '00000000000000000000000000000000';
|
||||
HEADER_SIGNATURE: HEADER_SIGNATURE,
|
||||
HEADER_MINOR_VERSION: '3e00',
|
||||
MAXREGSID: -6,
|
||||
NOSTREAM: -1,
|
||||
HEADER_CLSID: HEADER_CLSID,
|
||||
/* 2.6.1 Compound File Directory Entry */
|
||||
var EntryTypes = ['unknown','storage','stream','lockbytes','property','root'];
|
||||
}
|
||||
EntryTypes: ['unknown','storage','stream','lockbytes','property','root']
|
||||
};
|
||||
|
||||
var CFB_utils = {
|
||||
exports.read = readSync;
|
||||
exports.parse = parse;
|
||||
exports.utils = {
|
||||
ReadShift: ReadShift,
|
||||
CheckField: CheckField,
|
||||
prep_blob: prep_blob,
|
||||
bconcat: bconcat
|
||||
bconcat: bconcat,
|
||||
consts: consts
|
||||
};
|
||||
|
||||
if(typeof require !== 'undefined' && typeof exports !== 'undefined') {
|
||||
var fs = require('fs');
|
||||
exports.read = CFB.read;
|
||||
exports.parse = CFB.parse;
|
||||
exports.utils = CFB_utils;
|
||||
exports.version = CFB.version;
|
||||
}
|
||||
return exports;
|
||||
})();
|
||||
|
||||
if(typeof require !== 'undefined' && typeof module !== 'undefined' && typeof DO_NOT_EXPORT_CFB === 'undefined') { module.exports = CFB; }
|
||||
|
||||
5
misc/prof.js
Normal file
5
misc/prof.js
Normal file
@ -0,0 +1,5 @@
|
||||
var describe = function(m,cb){if(cb) cb();};
|
||||
describe.skip = function(m,cb){};
|
||||
var it = function(m,cb){if(cb) cb();};
|
||||
it.skip = function(m,cb){};
|
||||
var before = function(cb){if(cb) cb();};
|
||||
14
misc/spin.sh
Executable file
14
misc/spin.sh
Executable file
@ -0,0 +1,14 @@
|
||||
#!/bin/bash
|
||||
# spin.sh -- show a spinner (for coverage test)
|
||||
# Copyright (C) 2014 SheetJS
|
||||
|
||||
wpid=$1
|
||||
delay=1
|
||||
str="|/-\\"
|
||||
while [ $(ps -a|awk '$1=='$wpid' {print $1}') ]; do
|
||||
t=${str#?}
|
||||
printf " [%c]" "$str"
|
||||
str=$t${str%"$t"}
|
||||
sleep $delay
|
||||
printf "\b\b\b\b"
|
||||
done
|
||||
21
package.json
21
package.json
@ -1,25 +1,32 @@
|
||||
{
|
||||
"name": "cfb",
|
||||
"version": "0.9.1",
|
||||
"author": "SheetJS",
|
||||
"version": "0.10.0",
|
||||
"author": "sheetjs",
|
||||
"description": "Compound File Binary File Format extractor",
|
||||
"keywords": [ "cfb", "compression", "office" ],
|
||||
"bin": {
|
||||
"cfb": "./bin/cfb"
|
||||
},
|
||||
"main": "./cfb",
|
||||
"repository": { "type":"git", "url":"git://github.com/SheetJS/js-cfb.git" },
|
||||
"scripts": {
|
||||
"test": "make test"
|
||||
},
|
||||
"dependencies": {
|
||||
"commander":""
|
||||
},
|
||||
"devDependencies": {
|
||||
"mocha":"",
|
||||
"xlsjs":"",
|
||||
"uglify-js":""
|
||||
},
|
||||
"repository": { "type":"git", "url":"git://github.com/SheetJS/js-cfb.git" },
|
||||
"scripts": {
|
||||
"pretest": "make init",
|
||||
"test": "make test"
|
||||
},
|
||||
"config": {
|
||||
"blanket": {
|
||||
"pattern": "cfb.js"
|
||||
}
|
||||
},
|
||||
"bugs": { "url": "https://github.com/SheetJS/js-cfb/issues" },
|
||||
"license": "Apache 2.0"
|
||||
"license": "Apache-2.0",
|
||||
"engines": { "node": ">=0.8" }
|
||||
}
|
||||
|
||||
12
test.js
12
test.js
@ -3,10 +3,11 @@ var CFB;
|
||||
var fs = require('fs');
|
||||
describe('source', function() { it('should load', function() { CFB = require('./'); }); });
|
||||
|
||||
var ffunc = function(x){return x.substr(-4)==".xls" && fails.indexOf(x) === -1;};
|
||||
var fails = fs.existsSync('./fails.lst') ? fs.readFileSync('./fails.lst', 'utf-8').split("\n") : [];
|
||||
var files = fs.readdirSync('test_files').filter(function(x){return x.substr(-4)==".xls" && fails.indexOf(x) === -1;});
|
||||
var f2011 = fs.readdirSync('test_files/2011').filter(function(x){return x.substr(-4)==".xls" && fails.indexOf(x) === -1;});
|
||||
var f2013 = fs.readdirSync('test_files/2013').filter(function(x){return x.substr(-4)==".xls" && fails.indexOf(x) === -1;});
|
||||
var files = fs.readdirSync('test_files').filter(ffunc);
|
||||
var f2011 = fs.readdirSync('test_files/2011').filter(ffunc);
|
||||
var f2013 = fs.readdirSync('test_files/2013').filter(ffunc);
|
||||
|
||||
var dir = "./test_files/";
|
||||
|
||||
@ -19,8 +20,6 @@ function parsetest(x, cfb) {
|
||||
if(!cfb.find('/Workbook') && !cfb.find('/Book')) throw new Error("Cannot find workbook for " + x);
|
||||
});
|
||||
});
|
||||
describe(x + ' should ', function() {
|
||||
});
|
||||
}
|
||||
|
||||
describe('should parse test files', function() {
|
||||
@ -53,4 +52,7 @@ describe('input formats', function() {
|
||||
it('should read base64 strings', function() {
|
||||
CFB.read(fs.readFileSync(dir + '/' + cp, 'base64'), {type: 'base64'});
|
||||
});
|
||||
it('should read buffers', function() {
|
||||
CFB.read(fs.readFileSync(dir + '/' + cp), {type: 'buffer'});
|
||||
});
|
||||
});
|
||||
|
||||
Loading…
Reference in New Issue
Block a user