From 7f3d5ae6d3e82b3a14129b219a210e794a927188 Mon Sep 17 00:00:00 2001 From: Artur Adib Date: Thu, 19 Jan 2012 14:19:19 -0500 Subject: [PATCH 01/10] Fix worker message, better error handling --- src/evaluator.js | 11 +++++++---- src/worker.js | 24 ++++++++++++++++++++---- 2 files changed, 27 insertions(+), 8 deletions(-) diff --git a/src/evaluator.js b/src/evaluator.js index 21530f42f..1c277e027 100644 --- a/src/evaluator.js +++ b/src/evaluator.js @@ -159,6 +159,10 @@ var PartialEvaluator = (function PartialEvaluatorClosure() { // a Stream in the main thread. if (translated.file) translated.file = translated.file.getBytes(); + if (translated.properties.file) { + translated.properties.file = + translated.properties.file.getBytes(); + } handler.send('obj', [ loadedName, @@ -779,12 +783,11 @@ var PartialEvaluator = (function PartialEvaluatorClosure() { dict: baseDict, properties: properties }; - } - - } + } // if (type.name == 'Type3') + } // if (!descriptor) // According to the spec if 'FontDescriptor' is declared, 'FirstChar', - // 'LastChar' and 'Widths' should exists too, but some PDF encoders seems + // 'LastChar' and 'Widths' should exist too, but some PDF encoders seem // to ignore this rule when a variant of a standart font is used. // TODO Fill the width array depending on which of the base font this is // a variant. diff --git a/src/worker.js b/src/worker.js index 4d9dd1bb6..468cce019 100644 --- a/src/worker.js +++ b/src/worker.js @@ -109,11 +109,27 @@ var WorkerMessageHandler = { // Pre compile the pdf page and fetch the fonts/images. IRQueue = page.getIRQueue(handler, dependency); } catch (e) { + var minimumStackMessage = + 'worker.js: while trying to getPage() and getIRQueue()'; + // Turn the error into an obj that can be serialized - e = { - message: typeof e === 'object' ? e.message : e, - stack: typeof e === 'object' ? e.stack : null - }; + if (typeof e === 'string') { + e = { + message: e, + stack: minimumStackMessage + }; + } else if (typeof e === 'object') { + e = { + message: e.message || e.toString(), + stack: e.stack || minimumStackMessage + }; + } else { + e = { + message: 'Unknown exception type: ' + (typeof e), + stack: minimumStackMessage + } + } + handler.send('page_error', { pageNum: pageNum, error: e From 38d28ecb2e81eaeeb319c76719878cb22f7b8ad5 Mon Sep 17 00:00:00 2001 From: Artur Adib Date: Thu, 19 Jan 2012 16:02:27 -0500 Subject: [PATCH 02/10] Improved error handling/message --- web/viewer.js | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/web/viewer.js b/web/viewer.js index ac3fbff0c..9f8b772ab 100644 --- a/web/viewer.js +++ b/web/viewer.js @@ -353,8 +353,14 @@ var PDFView = { if (moreInfo) { errorMoreInfo.value += 'Message: ' + moreInfo.message; - if (moreInfo.stack) + if (moreInfo.stack) { errorMoreInfo.value += '\n' + 'Stack: ' + moreInfo.stack; + } else { + if (moreInfo.filename) + errorMoreInfo.value += '\n' + 'File: ' + moreInfo.filename; + if (moreInfo.filename) + errorMoreInfo.value += '\n' + 'Line: ' + moreInfo.lineNumber; + } } errorMoreInfo.rows = errorMoreInfo.value.split('\n').length - 1; }, From 66eff7a5cb6fea0dbb8737bc40bcc3f7ee23c8c0 Mon Sep 17 00:00:00 2001 From: Artur Adib Date: Fri, 20 Jan 2012 14:55:52 -0500 Subject: [PATCH 03/10] more robust fontMatrix parsing, error checking --- src/canvas.js | 10 ++++++++++ src/fonts.js | 10 +++++++++- 2 files changed, 19 insertions(+), 1 deletion(-) diff --git a/src/canvas.js b/src/canvas.js index 5ef900861..6ec8076cf 100644 --- a/src/canvas.js +++ b/src/canvas.js @@ -551,6 +551,16 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { throw 'Can\'t find font for ' + fontRefName; } + // If any of the diagonal elements of a transformation matrix are null + // ctx.restore() will fail in FF. See bugzilla bug #719844. + if (fontObj.fontMatrix[0] === 0 || + fontObj.fontMatrix[3] === 0 ) { + warn('Invalid font matrix for font ' + fontRefName); + + // Fallback + fontObj.fontMatrix = IDENTITY_MATRIX; + } + var name = fontObj.loadedName || 'sans-serif'; this.current.font = fontObj; diff --git a/src/fonts.js b/src/fonts.js index f96c15458..96a11d1fa 100644 --- a/src/fonts.js +++ b/src/fonts.js @@ -2594,7 +2594,15 @@ var Type1Parser = function type1Parser() { while (str[index++] != ']') count++; - var array = str.substr(start, count).split(' '); + str = str.substr(start, count); + + // Trim + str = str.replace(/^\s+/, ''); + str = str.replace(/\s+$/, ''); + // Remove adjacent spaces + str = str.replace(/\s+/g, ' '); + + var array = str.split(' '); for (var i = 0, ii = array.length; i < ii; i++) array[i] = parseFloat(array[i] || 0); return array; From 357f4cc6659ca8ff2dfefcc2878416c11401265a Mon Sep 17 00:00:00 2001 From: Artur Adib Date: Fri, 20 Jan 2012 15:20:25 -0500 Subject: [PATCH 04/10] Clarifying variable role --- src/canvas.js | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/src/canvas.js b/src/canvas.js index 6ec8076cf..792a09c97 100644 --- a/src/canvas.js +++ b/src/canvas.js @@ -703,12 +703,12 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { if (textSelection) text.geom = this.getTextGeometry(); - var width = 0; + var x = 0; for (var i = 0; i < glyphsLength; ++i) { var glyph = glyphs[i]; if (glyph === null) { // word break - width += wordSpacing; + x += wordSpacing; continue; } @@ -719,28 +719,28 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { default: // other unsupported rendering modes case TextRenderingMode.FILL: case TextRenderingMode.FILL_ADD_TO_PATH: - ctx.fillText(char, width, 0); + ctx.fillText(char, x, 0); break; case TextRenderingMode.STROKE: case TextRenderingMode.STROKE_ADD_TO_PATH: - ctx.strokeText(char, width, 0); + ctx.strokeText(char, x, 0); break; case TextRenderingMode.FILL_STROKE: case TextRenderingMode.FILL_STROKE_ADD_TO_PATH: - ctx.fillText(char, width, 0); - ctx.strokeText(char, width, 0); + ctx.fillText(char, x, 0); + ctx.strokeText(char, x, 0); break; case TextRenderingMode.INVISIBLE: break; } - width += charWidth; + x += charWidth; text.str += glyph.unicode === ' ' ? '\u00A0' : glyph.unicode; text.length++; text.canvasWidth += charWidth; } - current.x += width * textHScale2; + current.x += x * textHScale2; ctx.restore(); } From 86de8aca3368f6aa63cd49ebf68cf0c6d4471e41 Mon Sep 17 00:00:00 2001 From: Artur Adib Date: Fri, 20 Jan 2012 16:25:06 -0500 Subject: [PATCH 05/10] Nit --- src/fonts.js | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/fonts.js b/src/fonts.js index 96a11d1fa..5d0114476 100644 --- a/src/fonts.js +++ b/src/fonts.js @@ -3603,7 +3603,7 @@ var Type2CFF = (function Type2CFFClosure() { dict['cidOperatorPresent'] = true; break; default: - TODO('interpret top dict key'); + TODO('interpret top dict key: ' + key); } } return dict; From 786cccf636a17c8a8644cf5529c0fb67e4c5a4a5 Mon Sep 17 00:00:00 2001 From: Artur Adib Date: Fri, 20 Jan 2012 18:41:01 -0500 Subject: [PATCH 06/10] setFont() supports negative size, closes #1049 --- src/canvas.js | 32 ++++++++++++++++++++++++-------- test/pdfs/issue1049.pdf.link | 1 + test/test_manifest.json | 7 +++++++ 3 files changed, 32 insertions(+), 8 deletions(-) create mode 100644 test/pdfs/issue1049.pdf.link diff --git a/src/canvas.js b/src/canvas.js index 792a09c97..b36772fe0 100644 --- a/src/canvas.js +++ b/src/canvas.js @@ -551,8 +551,8 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { throw 'Can\'t find font for ' + fontRefName; } - // If any of the diagonal elements of a transformation matrix are null - // ctx.restore() will fail in FF. See bugzilla bug #719844. + // A valid matrix needs all main diagonal elements to be non-zero + // This also ensures we bypass FF bugzilla bug #719844. if (fontObj.fontMatrix[0] === 0 || fontObj.fontMatrix[3] === 0 ) { warn('Invalid font matrix for font ' + fontRefName); @@ -563,9 +563,23 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { var name = fontObj.loadedName || 'sans-serif'; + // Clone fontMatrix so we can manipulate it without affecting original + this.current.fontMatrix = fontObj.fontMatrix.slice(0); + + // The spec for Tf (setFont) says that 'size' specifies the font 'scale', + // and in some docs this can be negative. We implement this in fontMatrix. + if (size < 0) { + size = -size; + this.current.fontMatrix[0] = -fontObj.fontMatrix[0]; + this.current.fontMatrix[3] = -fontObj.fontMatrix[3]; + } + this.current.font = fontObj; this.current.fontSize = size; + // Cache font matrix sign + this.current.fontMatrixXSign = this.current.fontMatrix[0] > 0 ? 1 : -1; + var name = fontObj.loadedName || 'sans-serif'; var bold = fontObj.black ? (fontObj.bold ? 'bolder' : 'bold') : (fontObj.bold ? 'bold' : 'normal'); @@ -605,7 +619,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { var ctx = this.ctx; var current = this.current; var textHScale = current.textHScale; - var fontMatrix = current.font.fontMatrix || IDENTITY_MATRIX; + var fontMatrix = current.fontMatrix || IDENTITY_MATRIX; ctx.transform.apply(ctx, current.textMatrix); ctx.scale(1, -1); @@ -639,7 +653,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { var charSpacing = current.charSpacing; var wordSpacing = current.wordSpacing; var textHScale = current.textHScale; - var fontMatrix = font.fontMatrix || IDENTITY_MATRIX; + var fontMatrix = current.fontMatrix || IDENTITY_MATRIX; var textHScale2 = textHScale * fontMatrix[0]; var glyphsLength = glyphs.length; var textLayer = this.textLayer; @@ -677,7 +691,8 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { this.restore(); var transformed = Util.applyTransform([glyph.width, 0], fontMatrix); - var width = transformed[0] * fontSize + charSpacing; + var width = transformed[0] * fontSize + + current.fontMatrixXSign * charSpacing; ctx.translate(width, 0); current.x += width * textHScale; @@ -708,12 +723,13 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { var glyph = glyphs[i]; if (glyph === null) { // word break - x += wordSpacing; + x += current.fontMatrixXSign * wordSpacing; continue; } var char = glyph.fontChar; - var charWidth = glyph.width * fontSize * 0.001 + charSpacing; + var charWidth = glyph.width * fontSize * 0.001 + + current.fontMatrixXSign * charSpacing; switch (textRenderingMode) { default: // other unsupported rendering modes @@ -756,7 +772,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { var fontSize = current.fontSize; var textHScale = current.textHScale; if (!font.coded) - textHScale *= (font.fontMatrix || IDENTITY_MATRIX)[0]; + textHScale *= (current.fontMatrix || IDENTITY_MATRIX)[0]; var arrLength = arr.length; var textLayer = this.textLayer; var text = {str: '', length: 0, canvasWidth: 0, geom: {}}; diff --git a/test/pdfs/issue1049.pdf.link b/test/pdfs/issue1049.pdf.link new file mode 100644 index 000000000..c486dda43 --- /dev/null +++ b/test/pdfs/issue1049.pdf.link @@ -0,0 +1 @@ +http://ernestinefont.com/wp-content/themes/iA3%201.2.1/assets/pdf/ErnestinePro-InfoGuide.pdf diff --git a/test/test_manifest.json b/test/test_manifest.json index 26ddceaf9..f934f3509 100644 --- a/test/test_manifest.json +++ b/test/test_manifest.json @@ -423,5 +423,12 @@ "rounds": 1, "link": false, "type": "eq" + }, + { "id": "issue1049", + "file": "pdfs/issue1049.pdf", + "md5": "15473fffcdde9fb8f3756a4cf1aab347", + "rounds": 1, + "link": true, + "type": "eq" } ] From bbdec90c0dc9aefe93b7a356a60c8176d6f2e3cd Mon Sep 17 00:00:00 2001 From: Artur Adib Date: Fri, 20 Jan 2012 18:44:51 -0500 Subject: [PATCH 07/10] Lint --- src/canvas.js | 2 +- src/worker.js | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/src/canvas.js b/src/canvas.js index b36772fe0..e6ba9f88d 100644 --- a/src/canvas.js +++ b/src/canvas.js @@ -554,7 +554,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { // A valid matrix needs all main diagonal elements to be non-zero // This also ensures we bypass FF bugzilla bug #719844. if (fontObj.fontMatrix[0] === 0 || - fontObj.fontMatrix[3] === 0 ) { + fontObj.fontMatrix[3] === 0) { warn('Invalid font matrix for font ' + fontRefName); // Fallback diff --git a/src/worker.js b/src/worker.js index 468cce019..3bf935f17 100644 --- a/src/worker.js +++ b/src/worker.js @@ -127,7 +127,7 @@ var WorkerMessageHandler = { e = { message: 'Unknown exception type: ' + (typeof e), stack: minimumStackMessage - } + }; } handler.send('page_error', { From 5c8753dcbac9e7870d1a66efbc5bb484efc07864 Mon Sep 17 00:00:00 2001 From: Artur Adib Date: Mon, 23 Jan 2012 15:23:09 -0500 Subject: [PATCH 08/10] Fixing regression errors, better logic --- src/canvas.js | 39 ++++++++++++++++++--------------------- src/util.js | 4 ++++ 2 files changed, 22 insertions(+), 21 deletions(-) diff --git a/src/canvas.js b/src/canvas.js index e6ba9f88d..0a73e034c 100644 --- a/src/canvas.js +++ b/src/canvas.js @@ -23,6 +23,7 @@ var CanvasExtraState = (function CanvasExtraStateClosure() { this.alphaIsShape = false; this.fontSize = 0; this.textMatrix = IDENTITY_MATRIX; + this.fontMatrix = IDENTITY_MATRIX; this.leading = 0; // Current point (in user coordinates) this.x = 0; @@ -546,40 +547,36 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { }, setFont: function canvasGraphicsSetFont(fontRefName, size) { var fontObj = this.objs.get(fontRefName).fontObj; + var current = this.current; - if (!fontObj) { + if (!fontObj) throw 'Can\'t find font for ' + fontRefName; - } + + // Slice-clone matrix so we can manipulate it without affecting original + if (fontObj.fontMatrix) + current.fontMatrix = fontObj.fontMatrix.slice(0); + else + current.fontMatrix = IDENTITY_MATRIX.slice(0); // A valid matrix needs all main diagonal elements to be non-zero // This also ensures we bypass FF bugzilla bug #719844. - if (fontObj.fontMatrix[0] === 0 || - fontObj.fontMatrix[3] === 0) { + if (current.fontMatrix[0] === 0 || + current.fontMatrix[3] === 0) { warn('Invalid font matrix for font ' + fontRefName); - - // Fallback - fontObj.fontMatrix = IDENTITY_MATRIX; } - var name = fontObj.loadedName || 'sans-serif'; - - // Clone fontMatrix so we can manipulate it without affecting original - this.current.fontMatrix = fontObj.fontMatrix.slice(0); - // The spec for Tf (setFont) says that 'size' specifies the font 'scale', - // and in some docs this can be negative. We implement this in fontMatrix. + // and in some docs this can be negative (inverted x-y axes). + // We implement this condition with fontMatrix. if (size < 0) { size = -size; - this.current.fontMatrix[0] = -fontObj.fontMatrix[0]; - this.current.fontMatrix[3] = -fontObj.fontMatrix[3]; + current.fontMatrix[0] *= -1; + current.fontMatrix[3] *= -1; } this.current.font = fontObj; this.current.fontSize = size; - // Cache font matrix sign - this.current.fontMatrixXSign = this.current.fontMatrix[0] > 0 ? 1 : -1; - var name = fontObj.loadedName || 'sans-serif'; var bold = fontObj.black ? (fontObj.bold ? 'bolder' : 'bold') : (fontObj.bold ? 'bold' : 'normal'); @@ -692,7 +689,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { var transformed = Util.applyTransform([glyph.width, 0], fontMatrix); var width = transformed[0] * fontSize + - current.fontMatrixXSign * charSpacing; + Util.sign(current.fontMatrix[0]) * charSpacing; ctx.translate(width, 0); current.x += width * textHScale; @@ -723,13 +720,13 @@ var CanvasGraphics = (function CanvasGraphicsClosure() { var glyph = glyphs[i]; if (glyph === null) { // word break - x += current.fontMatrixXSign * wordSpacing; + x += Util.sign(current.fontMatrix[0]) * wordSpacing; continue; } var char = glyph.fontChar; var charWidth = glyph.width * fontSize * 0.001 + - current.fontMatrixXSign * charSpacing; + Util.sign(current.fontMatrix[0]) * charSpacing; switch (textRenderingMode) { default: // other unsupported rendering modes diff --git a/src/util.js b/src/util.js index 99b422296..ce0daa17d 100644 --- a/src/util.js +++ b/src/util.js @@ -93,6 +93,10 @@ var Util = (function UtilClosure() { return [xt, yt]; }; + Util.sign = function sign(num) { + return num < 0 ? -1 : 1; + }; + return Util; })(); From b34c55cc3cdaecd7aad66a7fbf613f90637b48d7 Mon Sep 17 00:00:00 2001 From: Artur Adib Date: Mon, 23 Jan 2012 15:29:15 -0500 Subject: [PATCH 09/10] Use ES5 .trim() --- src/fonts.js | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/src/fonts.js b/src/fonts.js index 5d0114476..f68ccd9ff 100644 --- a/src/fonts.js +++ b/src/fonts.js @@ -2596,9 +2596,7 @@ var Type1Parser = function type1Parser() { str = str.substr(start, count); - // Trim - str = str.replace(/^\s+/, ''); - str = str.replace(/\s+$/, ''); + str = str.trim(); // Remove adjacent spaces str = str.replace(/\s+/g, ' '); From 30a01c5da6eeef910228c234619a26fdaf99f524 Mon Sep 17 00:00:00 2001 From: Artur Adib Date: Mon, 30 Jan 2012 09:24:49 -0500 Subject: [PATCH 10/10] addressing reviewer comments, bug fix --- src/evaluator.js | 4 ++-- web/viewer.js | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/src/evaluator.js b/src/evaluator.js index 1c277e027..e1e064f07 100644 --- a/src/evaluator.js +++ b/src/evaluator.js @@ -783,8 +783,8 @@ var PartialEvaluator = (function PartialEvaluatorClosure() { dict: baseDict, properties: properties }; - } // if (type.name == 'Type3') - } // if (!descriptor) + } + } // According to the spec if 'FontDescriptor' is declared, 'FirstChar', // 'LastChar' and 'Widths' should exist too, but some PDF encoders seem diff --git a/web/viewer.js b/web/viewer.js index 9f8b772ab..e0ab1cc90 100644 --- a/web/viewer.js +++ b/web/viewer.js @@ -358,7 +358,7 @@ var PDFView = { } else { if (moreInfo.filename) errorMoreInfo.value += '\n' + 'File: ' + moreInfo.filename; - if (moreInfo.filename) + if (moreInfo.lineNumber) errorMoreInfo.value += '\n' + 'Line: ' + moreInfo.lineNumber; } }