From e086cf36f4d9d7f021b218de77edefde60ab5214 Mon Sep 17 00:00:00 2001
From: Yury Delendik <ydelendik@mozilla.com>
Date: Thu, 13 Sep 2012 08:09:46 -0700
Subject: [PATCH] Refactor text extraction / font loading logic
---
src/canvas.js | 14 ++---
src/evaluator.js | 131 ++++++++++++++++++++++-------------------------
src/fonts.js | 72 +++++++++-----------------
web/viewer.js | 25 ++++-----
4 files changed, 98 insertions(+), 144 deletions(-)
diff --git a/src/canvas.js b/src/canvas.js
index e5a204cb7..cca2c9212 100644
--- a/src/canvas.js
+++ b/src/canvas.js
@@ -782,15 +782,9 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
x += charWidth;
var glyphUnicode = glyph.unicode === ' ' ? '\u00A0' : glyph.unicode;
- var glyphUnicodeLength = glyphUnicode.length;
- //reverse an arabic ligature
- if (glyphUnicodeLength > 1 &&
- isRTLRangeFor(glyphUnicode.charCodeAt(0))) {
- for (var ii = glyphUnicodeLength - 1; ii >= 0; ii--)
- text.str += glyphUnicode[ii];
- } else
- text.str += glyphUnicode;
- text.length += glyphUnicodeLength;
+ if (glyphUnicode in NormalizedUnicodes)
+ glyphUnicode = NormalizedUnicodes[glyphUnicode];
+ text.str += reverseIfRtl(glyphUnicode);
text.canvasWidth += charWidth;
}
current.x += x * textHScale2;
@@ -842,7 +836,6 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
var numFakeSpaces = Math.round(-e / text.geom.spaceWidth);
if (numFakeSpaces > 0) {
text.str += '\u00A0';
- text.length++;
}
}
}
@@ -856,7 +849,6 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
text.str += shownText.str;
}
text.canvasWidth += shownText.canvasWidth;
- text.length += shownText.length;
}
} else {
error('TJ array element ' + e + ' is not string or num');
diff --git a/src/evaluator.js b/src/evaluator.js
index 704091149..59fec61a6 100644
--- a/src/evaluator.js
+++ b/src/evaluator.js
@@ -26,6 +26,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
this.handler = handler;
this.uniquePrefix = uniquePrefix;
this.objIdCounter = 0;
+ this.fontIdCounter = 0;
}
var OP_MAP = {
@@ -138,6 +139,35 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
};
PartialEvaluator.prototype = {
+ loadFont: function PartialEvaluator_loadFont(fontName, font, xref,
+ resources, dependency) {
+ var fontRes = resources.get('Font');
+
+ assert(fontRes, 'fontRes not available');
+
+ font = xref.fetchIfRef(font) || fontRes.get(fontName);
+ assertWellFormed(isDict(font));
+
+ ++this.fontIdCounter;
+ var loadedName = font.loadedName;
+ if (!loadedName) {
+ // keep track of each font we translated so the caller can
+ // load them asynchronously before calling display on a page
+ loadedName = 'font_' + this.uniquePrefix + this.fontIdCounter;
+ font.loadedName = loadedName;
+
+ var translated;
+ try {
+ translated = this.translateFont(font, xref, resources,
+ dependency);
+ } catch (e) {
+ translated = { error: e };
+ }
+ font.translated = translated;
+ }
+ return font;
+ },
+
getOperatorList: function PartialEvaluator_getOperatorList(stream,
resources,
dependency,
@@ -160,45 +190,35 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
}
function handleSetFont(fontName, font) {
- var loadedName = null;
+ font = self.loadFont(fontName, font, xref, resources, dependency);
- var fontRes = resources.get('Font');
+ var loadedName = font.loadedName;
+ if (!font.sent) {
+ var data = font.translated;
+ if (data.loadCharProcs) {
+ delete data.loadCharProcs;
- assert(fontRes, 'fontRes not available');
-
- font = xref.fetchIfRef(font) || fontRes.get(fontName);
- assertWellFormed(isDict(font));
-
- ++self.objIdCounter;
- if (!font.loadedName) {
- var translated = self.translateFont(font, xref, resources,
- dependency);
- if (translated) {
- // keep track of each font we translated so the caller can
- // load them asynchronously before calling display on a page
- loadedName = 'font_' + uniquePrefix + self.objIdCounter;
- translated.properties.loadedName = loadedName;
- font.loadedName = loadedName;
- font.translated = translated;
-
- var data;
- try {
- var fontObj = new Font(translated.name,
- translated.file,
- translated.properties);
- data = fontObj.export();
- } catch (e) {
- data = { error: e };
+ var charProcs = font.get('CharProcs').getAll();
+ var fontResources = font.get('Resources') || resources;
+ var charProcOperatorList = {};
+ for (var key in charProcs) {
+ var glyphStream = charProcs[key];
+ charProcOperatorList[key] =
+ self.getOperatorList(glyphStream, fontResources, dependency);
}
-
- handler.send('obj', [
- loadedName,
- 'Font',
- data
- ]);
+ data.charProcOperatorList = charProcOperatorList;
}
+
+ if (data instanceof Font)
+ data = data.export();
+
+ handler.send('obj', [
+ loadedName,
+ 'Font',
+ data
+ ]);
+ font.sent = true;
}
- loadedName = loadedName || font.loadedName;
// Ensure the font is ready before the font is set
// and later on used for drawing.
@@ -491,20 +511,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
var xref = this.xref;
function handleSetFont(fontName, fontRef) {
- var fontRes = resources.get('Font');
-
- // TODO: TOASK: Is it possible to get here? If so, what does
- // args[0].name should be like???
- assert(fontRes, 'fontRes not available');
-
- fontRes = xref.fetchIfRef(fontRes);
- fontRef = fontRef || fontRes.get(fontName);
- var font = xref.fetchIfRef(fontRef), tra;
- assertWellFormed(isDict(font));
- if (!font.translated) {
- font.translated = self.translateFont(font, xref, resources);
- }
- return font;
+ return self.loadFont(fontName, fontRef, xref, resources, null);
}
resources = xref.fetchIfRef(resources) || new Dict();
@@ -546,7 +553,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
break;
} // switch
if (chunk !== '') {
- text += fontCharsToUnicode(chunk, font.translated.properties);
+ text += fontCharsToUnicode(chunk, font.translated);
chunk = '';
}
@@ -853,7 +860,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
// - get the FontDescriptor from the descendant font
var df = dict.get('DescendantFonts');
if (!df)
- return null;
+ error('Descendant fonts are not specified');
dict = isArray(df) ? xref.fetchIfRef(df[0]) : df;
@@ -876,7 +883,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
// This case is here for compatibility.
var baseFontName = dict.get('BaseFont');
if (!isName(baseFontName))
- return null;
+ error('Base font is not specified');
// Using base font name as a font name.
baseFontName = baseFontName.name.replace(/[,_]/g, '-');
@@ -899,11 +906,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
};
this.extractDataStructures(dict, dict, xref, properties);
- return {
- name: baseFontName,
- dict: baseDict,
- properties: properties
- };
+ return new Font(baseFontName, null, properties);
}
}
@@ -939,6 +942,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
file: fontFile,
length1: length1,
length2: length2,
+ loadedName: baseDict.loadedName,
composite: composite,
wideChars: composite,
fixedPitch: false,
@@ -959,22 +963,9 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
if (type.name === 'Type3') {
properties.coded = true;
- var charProcs = dict.get('CharProcs').getAll();
- var fontResources = dict.get('Resources') || resources;
- properties.charProcOperatorList = {};
- for (var key in charProcs) {
- var glyphStream = charProcs[key];
- properties.charProcOperatorList[key] =
- this.getOperatorList(glyphStream, fontResources, dependency);
- }
}
- return {
- name: fontName.name,
- dict: baseDict,
- file: fontFile,
- properties: properties
- };
+ return new Font(fontName.name, fontFile, properties);
}
};
diff --git a/src/fonts.js b/src/fonts.js
index 820ada8a0..7707bffb6 100644
--- a/src/fonts.js
+++ b/src/fonts.js
@@ -1464,54 +1464,30 @@ var NormalizedUnicodes = {
'\uFE4F': '\u005F'
};
-function fontCharsToUnicode(charCodes, fontProperties) {
- var toUnicode = fontProperties.toUnicode;
- var composite = fontProperties.composite;
- var encoding, differences, cidToUnicode;
- var result = '';
- if (composite) {
- cidToUnicode = fontProperties.cidToUnicode;
- for (var i = 0, ii = charCodes.length; i < ii; i += 2) {
- var charCode = (charCodes.charCodeAt(i) << 8) |
- charCodes.charCodeAt(i + 1);
- if (toUnicode && charCode in toUnicode) {
- var unicode = toUnicode[charCode];
- result += typeof unicode !== 'number' ? unicode :
- String.fromCharCode(unicode);
- continue;
- }
- result += String.fromCharCode(!cidToUnicode ? charCode :
- cidToUnicode[charCode] || charCode);
- }
- } else {
- differences = fontProperties.differences;
- encoding = fontProperties.baseEncoding;
- for (var i = 0, ii = charCodes.length; i < ii; i++) {
- var charCode = charCodes.charCodeAt(i);
- var unicode;
- if (toUnicode && charCode in toUnicode) {
- var unicode = toUnicode[charCode];
- result += typeof unicode !== 'number' ? unicode :
- String.fromCharCode(unicode);
- continue;
- }
+function reverseIfRtl(chars) {
+ var charsLength = chars.length;
+ //reverse an arabic ligature
+ if (charsLength <= 1 || !isRTLRangeFor(chars.charCodeAt(0)))
+ return chars;
- var glyphName = charCode in differences ? differences[charCode] :
- encoding[charCode];
- if (glyphName in GlyphsUnicode) {
- result += String.fromCharCode(GlyphsUnicode[glyphName]);
- continue;
- }
- result += String.fromCharCode(charCode);
- }
- }
- // normalizing the unicode characters
- for (var i = 0, ii = result.length; i < ii; i++) {
- if (!(result[i] in NormalizedUnicodes))
+ var s = '';
+ for (var ii = charsLength - 1; ii >= 0; ii--)
+ s += chars[ii];
+ return s;
+}
+
+function fontCharsToUnicode(charCodes, font) {
+ var glyphs = font.charsToGlyphs(charCodes);
+ var result = '';
+ for (var i = 0, ii = glyphs.length; i < ii; i++) {
+ var glyph = glyphs[i];
+ if (!glyph)
continue;
- result = result.substring(0, i) + NormalizedUnicodes[result[i]] +
- result.substring(i + 1);
- ii = result.length;
+
+ var glyphUnicode = glyph.unicode;
+ if (glyphUnicode in NormalizedUnicodes)
+ glyphUnicode = NormalizedUnicodes[glyphUnicode];
+ result += reverseIfRtl(glyphUnicode);
}
return result;
}
@@ -1536,8 +1512,9 @@ var Font = (function FontClosure() {
}
this.name = name;
+ this.loadedName = properties.loadedName;
this.coded = properties.coded;
- this.charProcOperatorList = properties.charProcOperatorList;
+ this.loadCharProcs = properties.coded;
this.sizes = [];
var names = name.split('+');
@@ -1641,7 +1618,6 @@ var Font = (function FontClosure() {
this.widthMultiplier = !properties.fontMatrix ? 1.0 :
1.0 / properties.fontMatrix[0];
this.encoding = properties.baseEncoding;
- this.loadedName = properties.loadedName;
this.loading = true;
};
diff --git a/web/viewer.js b/web/viewer.js
index 2fe30debe..bf2c44012 100644
--- a/web/viewer.js
+++ b/web/viewer.js
@@ -1844,22 +1844,18 @@ var TextLayerBuilder = function textLayerBuilder(textLayerDiv) {
return;
}
var textDiv = textDivs.shift();
- if (textDiv.dataset.textLength > 0) {
- textLayerDiv.appendChild(textDiv);
+ textLayerDiv.appendChild(textDiv);
- if (textDiv.dataset.textLength > 1) { // avoid div by zero
- // Adjust div width to match canvas text
+ ctx.font = textDiv.style.fontSize + ' sans-serif';
+ var width = ctx.measureText(textDiv.textContent).width;
- ctx.font = textDiv.style.fontSize + ' sans-serif';
- var width = ctx.measureText(textDiv.textContent).width;
+ if (width > 0) {
+ var textScale = textDiv.dataset.canvasWidth / width;
- var textScale = textDiv.dataset.canvasWidth / width;
-
- CustomStyle.setProp('transform' , textDiv,
- 'scale(' + textScale + ', 1)');
- CustomStyle.setProp('transformOrigin' , textDiv, '0% 0%');
- }
- } // textLength > 0
+ CustomStyle.setProp('transform' , textDiv,
+ 'scale(' + textScale + ', 1)');
+ CustomStyle.setProp('transformOrigin' , textDiv, '0% 0%');
+ }
}
renderTimer = setInterval(renderTextLayer, renderInterval);
@@ -1899,7 +1895,6 @@ var TextLayerBuilder = function textLayerBuilder(textLayerDiv) {
textDiv.style.top = (text.geom.y - fontHeight) + 'px';
textDiv.textContent = PDFJS.bidi(text, -1);
textDiv.dir = text.direction;
- textDiv.dataset.textLength = text.length;
this.textDivs.push(textDiv);
};
};
@@ -2068,7 +2063,7 @@ document.addEventListener('DOMContentLoaded', function webViewerLoad(evt) {
});
document.getElementById('searchTermsInput').addEventListener('keydown',
- function() {
+ function(event) {
if (event.keyCode == 13) {
PDFView.search();
}