summaryrefslogtreecommitdiff
path: root/testing/resources
AgeCommit message (Collapse)Author
2017-06-13Allow zero length streams when parsing.chromium/3130Dan Sinclair
It's possible to create a stream of length 0 in a PDF document. Currently the code will early exit and return a nullptr. This causes issues when you want to print the given PDF as the FPDF_ImportPages code ends up only generating up to the zero length object. This CL allows creating streams with length 0 and updates the PDF saving code to output a blank stream. Bug: chromium:732380 Change-Id: I44182ba4aaac7c51284b002ba01bbc34b6bcf9e0 Reviewed-on: https://pdfium-review.googlesource.com/6490 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org>
2017-06-01Basic APIs and tests for extracting annotationsJane Liu
1. Added API for extracting annotation properties: * Added testing flag "--annot" that outputs the annotation properties into a .txt file. * Added two embedder tests covering all the API functions. Bug=pdfium:737 Change-Id: I95943a9b2b3d5d431bc8a74a31b27b4f4b521026 Reviewed-on: https://pdfium-review.googlesource.com/6092 Commit-Queue: Jane Liu <janeliulwq@google.com> Reviewed-by: Lei Zhang <thestig@chromium.org> Reviewed-by: dsinclair <dsinclair@chromium.org>
2017-06-01Be less trusting of MaybeLocal<> return types from V8 To* methods.Tom Sepez
Calling ToLocalChecked() will crash otherwise. Bug: 707673 Change-Id: I66a5b36d8cf1710a725e30c2d14a195d08ef25a4 Reviewed-on: https://pdfium-review.googlesource.com/6130 Reviewed-by: dsinclair <dsinclair@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
2017-06-01Fix content generation to only generate dirty page objects.wileyrya
BUG=pdfium:717 R=dsinclair@chromium.org,thestig@chromium.org Change-Id: I7e0e6fd301d40f9b5341d40cf11167b7748af243 Reviewed-on: https://pdfium-review.googlesource.com/6071 Reviewed-by: dsinclair <dsinclair@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org>
2017-05-18Add pixel test for bug 113910Nicolas Pena
Bug: chromium:113910 Change-Id: I52d712828a0adde02b1e867421ee06f1ca8be1e2 Reviewed-on: https://pdfium-review.googlesource.com/5636 Reviewed-by: Tom Sepez <tsepez@chromium.org> Commit-Queue: Nicolás Peña <npm@chromium.org>
2017-05-18Adding fm2js embedder testsDan Sinclair
This Cl adds the basis of the test framework to execute the javascript produced by CXFA_FM2JSContext and verify the results are correct. Change-Id: Ie46625b7e27ca0808e9cc41fdc00b7c0a212837d Reviewed-on: https://pdfium-review.googlesource.com/5651 Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
2017-05-18Add kerning test for bug 528103Nicolas Pena
Bug: chromium:528103 Change-Id: I990c829af8c6a9f481a34fc0e25de6814bbab34a Reviewed-on: https://pdfium-review.googlesource.com/5633 Reviewed-by: dsinclair <dsinclair@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Nicolás Peña <npm@chromium.org>
2017-05-18Fix a situation where images are not properly rendered.Lei Zhang
This regressed in commit e21fe98. When the image's bpc is a multiple of 8, there exists a colorspace, and there is a Decode parameter, the image data source was incorrectly pointing to a data structure that only contained black pixels. BUG=chromium:718762 Change-Id: I5d3fa739e41726b4ed1ebc16465e17f83fff9f8d Reviewed-on: https://pdfium-review.googlesource.com/5333 Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
2017-05-17Reland: Small fix in CPDF_TrueTypeFont loadNicolas Pena
The ToUnicode map should not be ignored when it exists. Doing so can cause a charcode to be assigned an incorrect glyph index, and will result in garbled text. Previously, some bots failed with 'unable to open' the .png file. Bug: chromium:665467 Change-Id: I435a73647eadcc3ba37bb0120f3b5cee381ae7a3 Reviewed-on: https://pdfium-review.googlesource.com/5610 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Nicolás Peña <npm@chromium.org>
2017-05-17Revert "Small fix in CPDF_TrueTypeFont load"Nicolás Peña
This reverts commit dde95d8be9bc2817e34429fc38ee6d89d6d5ab75. Reason for revert: the test added is flaky Original change's description: > Small fix in CPDF_TrueTypeFont load > > The ToUnicode map should not be ignored when it exists. Doing so can cause a > charcode to be assigned an incorrect glyph index, and will result in garbled > text. > > Bug: chromium:665467 > Change-Id: I21c1bf560a0731d974191d4189ea730ef9868334 > Reviewed-on: https://pdfium-review.googlesource.com/5512 > Reviewed-by: Lei Zhang <thestig@chromium.org> > Commit-Queue: Nicolás Peña <npm@chromium.org> > TBR=thestig@chromium.org,tsepez@chromium.org,dsinclair@chromium.org,npm@chromium.org NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true Bug: chromium:665467 Change-Id: I704a34f326d31018061bcfd857fb25f7e4ee4cc2 Reviewed-on: https://pdfium-review.googlesource.com/5493 Reviewed-by: Nicolás Peña <npm@chromium.org> Commit-Queue: Nicolás Peña <npm@chromium.org>
2017-05-16Small fix in CPDF_TrueTypeFont loadNicolas Pena
The ToUnicode map should not be ignored when it exists. Doing so can cause a charcode to be assigned an incorrect glyph index, and will result in garbled text. Bug: chromium:665467 Change-Id: I21c1bf560a0731d974191d4189ea730ef9868334 Reviewed-on: https://pdfium-review.googlesource.com/5512 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Nicolás Peña <npm@chromium.org>
2017-05-11Add missing CheckUnsupportedErrorDan Sinclair
The FPDF_LoadDocument call was missing the CheckUnSupportedError so, if the document contained unsuppoted information the user would not be notified. This brings the method in line with the other loading methods. Change-Id: I308b25335a228eb02c51562f9caf91cda9193b73 Reviewed-on: https://pdfium-review.googlesource.com/5336 Reviewed-by: Tom Sepez <tsepez@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org>
2017-04-29Fix rotationschromium/3086chromium/3085rbpotter
Normalize rotations read from PDF documents. Make FPDFPage_GetRotation always return a value 0 to 3 as specified in fpdf_edit.h instead of returning the page rotation / 90 (page rotation may be negative for some PDFs). BUG=chromium:713197 Change-Id: Ie477803f7d298b777a3ace89b21cfda8b7f6808b Reviewed-on: https://pdfium-review.googlesource.com/4532 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
2017-04-25Add test for bug 714187Dan Sinclair
This CL adds a minimized test case for https://crbug.com/714187 in order to keep it from regressing in the future. Bug: chromium:714187 Change-Id: I913f380c85a57621424d82165393b1616c2f6a9a Reviewed-on: https://pdfium-review.googlesource.com/4491 Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
2017-04-20Do not insert null node if ParseAsXDPPacket_Config failsTom Sepez
Doing so trips an assert in the debug build but is harmless otherwise. This appears to be wrong since the beginning of time. Bug: 709793 Change-Id: I179f6944a3f7b2a89efa65a16078f6f58448ba4b Reviewed-on: https://pdfium-review.googlesource.com/4390 Reviewed-by: dsinclair <dsinclair@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
2017-04-13Add embeddertest for form text rendering and savingNicolas Pena
This CL adds an embeddertest that adds text to a textfield and saves it. It also adds a new 'charcode' option for .evt files in pdfium_test. Change-Id: I14fbf50e2b1d5ae0bdc68d1dd25dc4f889c49bfb Reviewed-on: https://pdfium-review.googlesource.com/4150 Commit-Queue: Nicolás Peña <npm@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org> Reviewed-by: dsinclair <dsinclair@chromium.org>
2017-03-30Upgrade to FreeType 2.7.1.Lei Zhang
BUG=pdfium:601 Change-Id: I07756cd208cd2221802ff2d331f316b6618a41e0 Reviewed-on: https://pdfium-review.googlesource.com/3120 Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: dsinclair <dsinclair@chromium.org>
2017-03-27Add Mac 10.12 pixel expectationsNicolas Pena
Bug: pdfium:626 Change-Id: I93a0d02c26d1a44c0fca666cfd09df32ecc1a23d Reviewed-on: https://pdfium-review.googlesource.com/3222 Commit-Queue: Nicolás Peña <npm@chromium.org> Reviewed-by: dsinclair <dsinclair@chromium.org>
2017-03-23Fix two CloneNonCycle issuesWei Li
CloneNonCycle() tries to detect cyclic object references without copying them. There are two issues: -- for elements in an array or a dictionary, they should be able to refer to the same object, which are not cyclic; -- for cyclic referenced elements in an array or a dictionary, do not clone the element at all. Having nullptr or <key, nullptr> as an element, like we did before, might cause crash when the element being accessed. BUG=chromium:701860 Change-Id: Id0304accde76ed06fa5ce640994c7628359600fb Reviewed-on: https://pdfium-review.googlesource.com/3156 Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: dsinclair <dsinclair@chromium.org>
2017-03-17Handle web links across lineschromium/3045Wei Li
When a web link has a hyphen at the end of line, we consider it to be continued to the next line. For example, "http://www.abc.com/my-\r\ntest" should be extracted as "http://www.abc.com/my-test". BUG=pdfium:650 Change-Id: I64a93d9c66faf2be0abdaf8cfe8ee496c435d0ca Reviewed-on: https://pdfium-review.googlesource.com/3092 Commit-Queue: Wei Li <weili@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
2017-03-15Add array_buffer JS test.Tom Sepez
JS Array Buffers are the first candidate to be allocated from PartitionAlloc when it becomes available, so add test first. Presently, we will return as large an array buffer as the system can handle; this is generally a bad idea so limit them to 256MB and test that we handle failure. Change-Id: I205745a7938d69eb32ac883b90824f2f9e584ec7 Reviewed-on: https://pdfium-review.googlesource.com/3065 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
2017-02-27Fix uninitialized memory read in CJS_Object::GetEmbedObject()Tom Sepez
The expected way to create native PDFium objects for JS is via the NewFxDynamicObject() call in C++, but that doesn't mean that the corresponding constructors won't be called from JS. In that case, the internal fields will be uninitialized, and subsequent method calls may try to use them. Add a constructor callback for all PDFium objects that nulls out these fields (shame that v8 doesn't do this by default, but probably saves some cycles). Then ensure that we check for this possibility in all the places it might turn up. Conversely, if we've just gotten a successful return from NewFxDynamicObject(), we know the CJS_Object/EmbedObj are good, so avoid checking there. BUG=695826 Change-Id: Iadad644c4af937def967ddc83daac1dad7544d69 Reviewed-on: https://pdfium-review.googlesource.com/2839 Reviewed-by: dsinclair <dsinclair@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
2017-01-18Bad indexing in CPDF_Document::FindPageIndex when page tree corrupt.tsepez
Moving to std::vector from the more forgiving CFX_ArrayTemplate revealed the dubious page tree traversal, which depends on the correctness of the /Count entries to properly summarize the total descendants under a given node. The only "correct" thing to do is to throw away these counts as parsed, and re-compute them, perhaps in CountPages(). But I'm not willing to do that since it may break unknown documents in the wild. Pass out-params as pointers while we're at it. BUG=680376 Review-Url: https://codereview.chromium.org/2636403003
2017-01-12Don't put timers with ID == 0 into the global timer map.tsepez
A return of ID == 0 from the embedder means the timer was not created (see public/fpdf_formfill.h), although few embedders actually conform to this convention. Firing a timer with ID == 0 will thus do nothing since there can't be such a timer in the map. BUG=679649 Review-Url: https://codereview.chromium.org/2626863003
2017-01-12Custom toString() methods may delete annots.tsepez
In this case, we observe the destruction of the object, but have unfortunately saved a pointer to it in a local variable. BUG=679643 Review-Url: https://codereview.chromium.org/2628233002
2017-01-11Annotation deleted while retrieving it in JStsepez
Widgets as returned from GetWidgets() can pop out of existence unexpectedly, so always return observed pointers. This extends the same pattern used elsewhere in the file to all occurrences. BUG=679642 Review-Url: https://codereview.chromium.org/2624933002
2016-12-19Relax the EncryptMetadata check.chromium/2957tsepez
BUG=pdfium:644 Review-Url: https://codereview.chromium.org/2581873002
2016-11-23Add APIs for limited use of document tagged code.thestig
BUG=pdfium:568 Review-Url: https://codereview.chromium.org/2519343002
2016-11-23Add API for getting page labels.thestig
BUG=pdfium:479 Review-Url: https://codereview.chromium.org/2521843003
2016-11-21Add a regression test for rasterizing PDFs.thestig
BUG=chromium:667012 Review-Url: https://codereview.chromium.org/2508203007
2016-11-11Fix unique ptrs in fpdfppo.cpptsepez
There's a path out that deletes a pointer whose ownership was passed off earlier. This will get simpler once more APIs take unique_ptr. BUG=664284 Review-Url: https://codereview.chromium.org/2495003006
2016-11-04Implement FPDF_VIEWERREF_GetName() API.chromium/2910thestig
This is a generic API function to retrieve any viewer preference of type name. BUG=pdfium:414 Review-Url: https://codereview.chromium.org/2475923003
2016-10-18Add a test case for bug 494057.thestig
The expectation is set incorrectly to allow the test to pass. BUG=chromium:494057 Review-Url: https://codereview.chromium.org/2430583002
2016-09-15Use either /RECT or /QuadPoints for annotation coordinates, depending on /APtonikitoo
On Acrobat, if "/AP" is present on a text markup definition, the coordinates used to draw the annotation come from "/Rect values, whereas if "/AP" is not defined, the array defined in /QuadPoints is used to grab the annotation coordinates from. PDFium, on the other hand, uses "/Rect" regardless of presence or absence of "/AP". CL fixes PDFium to work similarly to Acrobat, in this case. TEST=testing/resources/pixel/bug_585_*.in BUG=pdfium:585 Review-Url: https://codereview.chromium.org/2289293005
2016-09-15Use ToUnicode mapping even when unicode is 0.npm
CPDF_Font::UnicodeFromCharcode returns 0 only if ToUnicode map maps the charcode to 0. CPDF_SimpleFont::UnicodeFromCharcode and CPDF_CID_Font:: UnicodeFromCharCode return 0 only if the call to CPDF_Font returns 0. In other cases, these methods return an empty string. So when processing text, a 0 return from the method should not be replaced with the charcode. BUG=pdfium:583 Review-Url: https://codereview.chromium.org/2342073002
2016-08-29Fix the test case added in https://codereview.chromium.org/2277063003/tonikitoo
In [1], it was made a mistake in the way the test case testing/resources/pixel/bug_492.pdf was generated. This CL aims at fixing this mistake by: 1- keep making use of the new pdfium_test capability introduced by [1], 2- add a proper .in file for the test case to generate its respective .pdf file. [1] https://codereview.chromium.org/2277063003/ BUG=pdfium:492 Review-Url: https://codereview.chromium.org/2286023002
2016-08-26Add support to Document::getAnnots methodtonikitoo
Although notably, the parameters handling support is not complete, CL intends to be the first step towards a more complete implementation of this API. TEST=testing/resources/javascript/bug_492_1.in BUG=pdfium:492 Review-Url: https://codereview.chromium.org/2281273002
2016-08-26Extend pdfium_test capability so that more Javascript can be executedchromium/2841tonikitoo
In [1], the lack of support of pdfium_test to some application level hooks was felt. More specifically, the lack of implementation of the hook FFI_GetPage, called when 'this.getAnnot()' is executed in an Acrobar JS context, makes it non-trivial to JS texts that manipulate PDF annotations. [1] https://codereview.chromium.org/2265313002/ Here is the failing call stack in pdfium_test: 0 ::RenderPdf (samples/pdfium_test.cc) 1 ::FORM_DoDocumentOpenAction (fpdfsdk/fpdfformfill.cpp) 2 CPDFSDK_Document::ProcOpenAction (fpdfsdk/fsdk_mgr.cpp) 3 CPDFSDK_ActionHandler::DoAction_DocOpen (fpdfsdk/fsdk_actionhandler.cpp) <----v8----> 4 Document::getAnnot (fpdfsdk/javascript/Document.cpp) 5 CPDFSDK_Document::GetPageView (fpdfsdk/fsdk_mgr.cpp) 6 CPDFDoc_Environment::FFI_GetPage (fpdfsdk/include/fsdk_mgr.h) (frame 6 returns nullptr, and getAnnot call in frame 4 bails) CL extends pdfium_test app with a FFI_GetPage hook implementation. Basically what FFI_GetPage does is returning a FPDF_PAGE instance. In case of pdfium_test, FPDF_PAGE instances were only created on demand when the page was going to get rendered, and then discarded. Since FFI_GetPage can be called by JS before pages are rendered, CL moved the page creation code into a helper function, and cached the FPDF_PAGE instances created in a map, so it does not recreate them needlessly. BUG=pdfium:492 Review-Url: https://codereview.chromium.org/2277063003
2016-08-19Stub out Document::syncAnnotScan method.chromium/2834tonikitoo
The PDF specification [1] says: " syncAnnotScan guarantees that all annotations will be scanned by the time this method returns. (..) Normally a background task runs that examine every page and looks for annotations during idle times. " The statement details specifically how Acrobat implements this method. Although, neither the method itself nor the background scanner task are implemented in PDFium (as of today, Ago/2016), not having ::syncAnnotScan at least stubbed out can be considered harmfull since its absence makes JS acrobat scripts silently fail when it has a call to it. Given that, and following a stub-out pattern present in other methods including ::addAnnot and ::addField, CL provides a stubbed out implementation of Document::syncAnnotScan. [1] http://www.adobe.com/content/dam/Adobe/en/devnet/acrobat/pdfs/js_api_reference.pdf BUG=pdfium:492 Review-Url: https://codereview.chromium.org/2265553002
2016-08-18Add initial Document::getAnnot supportchromium/2833tonikitoo
CL implements the first step in order to support Annotations manipulation in PDFium: Document::getAnnot. The method takes two arguments, an integer (page number) and a string (annotation name). When called, it iterates over the annotations on the given page number, searching for the one whose name matches the string in the second parameter. If found, then an Annot instance (see Annot.cpp/g added by this CL), is bound to a Javascript object and returned. With the use cases described in bug [1] as an initial test case, CL adds support to the following Annotation object properties: - hidden - name - type Idea is to keep evolving the implementation with more methods and properties in follow up CLs. [1] https://bugs.chromium.org/p/pdfium/issues/detail?id=492 BUG=pdfium:492 Review-Url: https://codereview.chromium.org/2260663002
2016-08-10Make Document's 'info' property readonlytonikitoo
As per the PDF specification in [1], page 103, the 'info' property of the Document object is readonly. [1] http://partners.adobe.com/public/developer/en/acrobat/sdk/5186AcroJS.pdf Review-Url: https://codereview.chromium.org/2235883003
2016-08-08Add support to Document::gotoNamedDest method.tonikitoo
Patch implements the Document's API gotoNamedDest, which is part of the PDF specification [1], page 129, with the following (short) description: "Use this method to go to a named destination within the PDF document". [1] http://partners.adobe.com/public/developer/en/acrobat/sdk/5186AcroJS.pdf "Named destination" is a common concept in the PDF world. It can be used together with PDF's Links, Annotations, Bookmarks and OpenActions, as well as an action per se, in case "this.gotoNamedDest" is called directly. Note that the implementation makes use of the existing hook CPDFDoc_Environment::FFI_DoGoToAction, which ends up calling out the embedder to actually handle it. In case of Chromium, for instance, it calls PDFiumEngine::Form_DoGoToAction which only handles for now the "page" property of the "destination". Other properties, including zoom level, and scroll position are ignored for the moment. BUG=pdfium:492 Review-Url: https://codereview.chromium.org/2221823003
2016-08-08Add support to Document::URL property getter.tonikitoo
As per the PDF specification at [1] " This property specifies the document's URL. ". IE/Acrobat supports it, and getting it implemented would be one step forward in order to support Acrobat JS script as the one in [2]. [1] http://partners.adobe.com/public/developer/en/acrobat/sdk/5186AcroJS.pdf [2] https://bugs.chromium.org/p/pdfium/issues/detail?id=492 BUG=492 Review-Url: https://codereview.chromium.org/2219183002
2016-08-05Remove another potential stale CJS_Timer usagetsepez
Fix memory ownership model for PDFium timers. The |app| class owns the CJS_Timer as part of its vector<unique_ptr> to them. The CJS_Timer "owns" its slot in the global ID to timer map, and removes itself when it is destroyed. Nothing else deletes from the global map. Deleting from the global map is accompanied by a callback to the embedder to clear its resources. Next, the proper way to remove a CJS_Timer is by going through the app, and having the app erase its unique ptr, which then deletes the CJS_Timer, which in turn cleans up the global map. Provide a CJS_Timer::Cancel static method to do this conveniently. There is a alternate path to the CJS_timer via JS and its CJS_TimerObj. CJS_TimerObj owns a TimerObj that currently points to the CJS_Timer. If the timer fires, and cleans itself up, this can go stale. Make the TimerObj maintain a weak reference via global timer ID rather than a direct pointer to the CJS_Timer, so that if the timer fires and is destroyed, future attempts to cancel find nothing. There is another path, where if the JS timer object is GC'd, then we just clean up its CJS_TimerObj without touching the actual CJS_Timers. We could make this match the spec by calling into the new cancel routine as described above, but it seems weird to have a timer depend on whether a gc happened or not. A subsequent CL will rename these objects to more closely match the conventions used by the other JS wrappers. BUG=634716 Review-Url: https://codereview.chromium.org/2221513002
2016-08-04Fix issue when firing TimerProc() destroys timerchromium/2820tsepez
We must look the timer up a second time since the callback may have released it. BUG=634394 Review-Url: https://codereview.chromium.org/2214003003
2016-08-04Beef up timer cancellation teststsepez
Adds more questionable invocations of ClearTimeOut(). Also, checking that nothing happened is fragile. Log at least one thing to show that the code ran. Review-Url: https://codereview.chromium.org/2218473002
2016-08-03Add test for bug 620428 (setinterval cancellation)tsepez
While we're at it, beef up existing test for non-cancellation. In turn, fix test harness to implement intervals properly. In turn, fix public documentation to be clearer about timers. Also rename a few identifiers that sounded "off". Review-Url: https://codereview.chromium.org/2211513002
2016-08-02Fix Jbig2 document context creation by checking proper pointerweili
The pointer a unique_ptr contains should be checked instead of the pointer of the unique_ptr itself. BUG=chromium:631912 Review-Url: https://codereview.chromium.org/2205573004
2016-05-23Distinguish between user and owner passwords.thestig
BUG=pdfium:496 Review-Url: https://codereview.chromium.org/2005653002
2016-05-11Revert "Reland of relax a couple checks to allow certain non-standard PDF ↵weili
files. (patchset #1 id:1 of https://codereview.chromium.org/1946693002/ )" This reverts commit a031357eaab7c934ac03717968cf78ff556c819b. The reason to revert it is that some malformed or maliciously crafted PDF files may cause crashes. BUG=610973 Review-Url: https://codereview.chromium.org/1971013002