Sakurai/pdf.js
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge pull request #53 from vingtetun/master

Browse Source 
Merge TTF font rewriting to pass the font sanitization pass.
This commit is contained in:
Andreas Gal 2011-06-22 11:42:06 -07:00
commit 5585c6a5fe
2 changed files with 95 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

102
fonts.js
View File

@ -103,7 +103,7 @@ var Font = (function () {
// If the font is to be ignored, register it like an already loaded font // If the font is to be ignored, register it like an already loaded font
// to avoid the cost of waiting for it be be loaded by the platform. // to avoid the cost of waiting for it be be loaded by the platform.
if (properties.ignore || properties.type == "TrueType" || kDisableFonts) { if (properties.ignore || kDisableFonts) {
Fonts[name] = { Fonts[name] = {
data: file, data: file,
loading: false, loading: false,
@ -242,7 +242,7 @@ var Font = (function () {
return ranges; return ranges;
}; };
function createCMAPTable(glyphs) { function createCMapTable(glyphs) {
var ranges = getRanges(glyphs); var ranges = getRanges(glyphs);
var headerSize = (12 * 2 + (ranges.length * 4 * 2)); var headerSize = (12 * 2 + (ranges.length * 4 * 2));
@ -274,7 +274,7 @@ var Font = (function () {
var bias = 0; var bias = 0;
for (var i = 0; i < segCount - 1; i++) { for (var i = 0; i < segCount - 1; i++) {
var range = ranges[i]; var range = ranges[i];
var start = range[0]; var start = range[0];
var end = range[1]; var end = range[1];
var delta = (((start - 1) - bias) ^ 0xffff) + 1; var delta = (((start - 1) - bias) ^ 0xffff) + 1;
bias += (end - start + 1); bias += (end - start + 1);
@ -284,8 +284,8 @@ var Font = (function () {
idDeltas += string16(delta); idDeltas += string16(delta);
idRangeOffsets += string16(0); idRangeOffsets += string16(0);
for (var j = start; j <= end; j++) for (var j = 0; j < range.length; j++)
glyphsIds += String.fromCharCode(j); glyphsIds += String.fromCharCode(range[j]);
} }
startCount += "\xFF\xFF"; startCount += "\xFF\xFF";
@ -368,11 +368,11 @@ var Font = (function () {
var length = FontsUtils.bytesToInteger(file.getBytes(4)); var length = FontsUtils.bytesToInteger(file.getBytes(4));
// Read the table associated data // Read the table associated data
var currentPosition = file.pos; var previousPosition = file.pos;
file.pos = file.start + offset; file.pos = file.start ? file.start : 0;
file.skip(offset);
var data = file.getBytes(length); var data = file.getBytes(length);
file.pos = currentPosition; file.pos = previousPosition;
return { return {
tag: tag, tag: tag,
@ -393,6 +393,76 @@ var Font = (function () {
} }
}; };
function replaceCMapTable(cmap, font, properties) {
var version = FontsUtils.bytesToInteger(font.getBytes(2));
var numTables = FontsUtils.bytesToInteger(font.getBytes(2));
for (var i = 0; i < numTables; i++) {
var platformID = FontsUtils.bytesToInteger(font.getBytes(2));
var encodingID = FontsUtils.bytesToInteger(font.getBytes(2));
var offset = FontsUtils.bytesToInteger(font.getBytes(4));
var format = FontsUtils.bytesToInteger(font.getBytes(2));
var length = FontsUtils.bytesToInteger(font.getBytes(2));
var language = FontsUtils.bytesToInteger(font.getBytes(2));
if ((format == 0 && numTables == 1) ||
(format == 6 && numTables == 1 && !properties.encoding.empty)) {
// Format 0 alone is not allowed by the sanitizer so let's rewrite
// that to a 3-1-4 Unicode BMP table
var charset = properties.charset;
var glyphs = [];
for (var j = 0; j < charset.length; j++) {
glyphs.push({
unicode: GlyphsUnicode[charset[j]] || 0
});
}
cmap.data = createCMapTable(glyphs);
} else if (format == 6 && numTables == 1) {
// Format 6 is a 2-bytes dense mapping, which means the font data
// lives glue together even if they are pretty far in the unicode
// table. (This looks weird, so I can have missed something), this
// works on Linux but seems to fails on Mac so let's rewrite the
// cmap table to a 3-1-4 style
var firstCode = FontsUtils.bytesToInteger(font.getBytes(2));
var entryCount = FontsUtils.bytesToInteger(font.getBytes(2));
var glyphs = [];
var min = 0xffff, max = 0;
for (var j = 0; j < entryCount; j++) {
var charcode = FontsUtils.bytesToInteger(font.getBytes(2));
glyphs.push(charcode);
if (charcode < min)
min = charcode;
if (charcode > max)
max = charcode;
}
// Since Format 6 is a dense array, check for gaps
for (var j = min; j < max; j++) {
if (glyphs.indexOf(j) == -1)
glyphs.push(j);
}
for (var j = 0; j < glyphs.length; j++)
glyphs[j] = { unicode: glyphs[j] + firstCode };
var ranges= getRanges(glyphs);
assert(ranges.length == 1, "Got " + ranges.length + " ranges in a dense array");
var encoding = properties.encoding;
var denseRange = ranges[0];
var start = denseRange[0];
var end = denseRange[1];
var index = firstCode;
for (var j = start; j <= end; j++)
encoding[index++] = glyphs[j - firstCode - 1].unicode;
cmap.data = createCMapTable(glyphs);
}
}
};
// Check that required tables are present // Check that required tables are present
var requiredTables = [ "OS/2", "cmap", "head", "hhea", var requiredTables = [ "OS/2", "cmap", "head", "hhea",
"hmtx", "maxp", "name", "post" ]; "hmtx", "maxp", "name", "post" ];
@ -448,18 +518,8 @@ var Font = (function () {
data: OS2 data: OS2
}); });
// If the font is missing a OS/2 table it's could be an old mac font
// without a 3-1-4 Unicode BMP table, so let's rewrite it.
var charset = properties.charset;
var glyphs = [];
for (var i = 0; i < charset.length; i++) {
glyphs.push({
unicode: GlyphsUnicode[charset[i]]
});
}
// Replace the old CMAP table with a shiny new one // Replace the old CMAP table with a shiny new one
cmap.data = createCMAPTable(glyphs); replaceCMapTable(cmap, font, properties);
// Rewrite the 'post' table if needed // Rewrite the 'post' table if needed
if (!post) { if (!post) {
@ -599,7 +659,7 @@ var Font = (function () {
var charstrings = font.getOrderedCharStrings(properties.glyphs); var charstrings = font.getOrderedCharStrings(properties.glyphs);
/** CMAP */ /** CMAP */
cmap = createCMAPTable(charstrings); cmap = createCMapTable(charstrings);
createTableEntry(otf, offsets, "cmap", cmap); createTableEntry(otf, offsets, "cmap", cmap);
/** HEAD */ /** HEAD */

22
pdf.js
View File

@ -2362,7 +2362,7 @@ var CanvasGraphics = (function() {
// Fonts with an embedded cmap but without any assignment in // Fonts with an embedded cmap but without any assignment in
// it are not yet supported, so ask the fonts loader to ignore // it are not yet supported, so ask the fonts loader to ignore
// them to not pay a stupid one sec latence. // them to not pay a stupid one sec latence.
var ignoreFont = true; var ignoreFont = false;
var encodingMap = {}; var encodingMap = {};
var charset = []; var charset = [];
@ -2406,20 +2406,24 @@ var CanvasGraphics = (function() {
} }
} }
} else if (fontDict.has("ToUnicode")) { } else if (fontDict.has("ToUnicode")) {
encodingMap = {empty: true};
var cmapObj = xref.fetchIfRef(fontDict.get("ToUnicode")); var cmapObj = xref.fetchIfRef(fontDict.get("ToUnicode"));
if (IsName(cmapObj)) { if (IsName(cmapObj)) {
error("ToUnicode file cmap translation not implemented"); error("ToUnicode file cmap translation not implemented");
} else if (IsStream(cmapObj)) { } else if (IsStream(cmapObj)) {
var encoding = Encodings["WinAnsiEncoding"]; var encoding = Encodings["WinAnsiEncoding"];
var firstChar = xref.fetchIfRef(fontDict.get("FirstChar")); var firstChar = xref.fetchIfRef(fontDict.get("FirstChar"));
for (var i = firstChar; i < encoding.length; i++)
encodingMap[i] = new Name(encoding[i]);
var tokens = []; var tokens = [];
var token = ""; var token = "";
var buffer = cmapObj.ensureBuffer ? cmapObj.ensureBuffer() : cmapObj; var length = cmapObj.length;
var cmap = cmapObj.getBytes(buffer.byteLength); if (cmapObj instanceof FlateStream) {
cmapObj.readBlock();
length = cmapObj.bufferLength;
}
var cmap = cmapObj.getBytes(length);
for (var i =0; i < cmap.length; i++) { for (var i =0; i < cmap.length; i++) {
var byte = cmap[i]; var byte = cmap[i];
if (byte == 0x20 || byte == 0x0A || byte == 0x3C || byte == 0x3E) { if (byte == 0x20 || byte == 0x0A || byte == 0x3C || byte == 0x3E) {
@ -2446,8 +2450,10 @@ var CanvasGraphics = (function() {
var code = parseInt("0x" + tokens[j+2]); var code = parseInt("0x" + tokens[j+2]);
for (var k = startRange; k <= endRange; k++) { for (var k = startRange; k <= endRange; k++) {
encodingMap[k] = GlyphsUnicode[encoding[code]]; // The encoding mapping table will be filled
charset.push(encoding[code++]); // later during the building phase
//encodingMap[k] = GlyphsUnicode[encoding[code]];
charset.push(encoding[code++] || ".notdef");
} }
} }
break; break;
@ -2752,7 +2758,7 @@ var CanvasGraphics = (function() {
} }
this.current.fontSize = size; this.current.fontSize = size;
this.ctx.font = this.current.fontSize +'px "' + fontName + '"'; this.ctx.font = this.current.fontSize +'px "' + fontName + '", Symbol';
}, },
setTextRenderingMode: function(mode) { setTextRenderingMode: function(mode) {
TODO("text rendering mode"); TODO("text rendering mode");