Looking at this again, it struck me that added functionality in `Util.intersect` is probably more confusing than helpful in general; sorry about the churn in this code!
Based on the parameter name you'd probably expect it to only match when the intersection is `[0, 0, 0, 0]` and not when only one component is zero, hence the `skipEmpty` parameter thus feels too tightly coupled to the `Page.view` getter.
This is based on a real-world PDF file I encountered very recently[1], although I'm currently unable to recall where I saw it.
Note that different PDF viewers handle these sort of errors differently, with Adobe Reader outright failing to render the attached PDF file whereas PDFium mostly handles it "correctly".
The patch makes the following notable changes:
- Refactor the `cropBox` and `mediaBox` getters, on the `Page`, to reduce unnecessary duplication. (This will also help in the future, if support for extracting additional page bounding boxes are added to the API.)
- Ensure that the page bounding boxes, i.e. `cropBox` and `mediaBox`, are never empty to prevent issues/weirdness in the viewer.
- Ensure that the `view` getter on the `Page` will never return an empty intersection of the `cropBox` and `mediaBox`.
- Add an *optional* parameter to `Util.intersect`, to allow checking that the computed intersection isn't actually empty.
- Change `Util.intersect` to have consistent return types, since Arrays are of type `Object` and falling back to returning a `Boolean` thus seem strange.
---
[1] In that case I believe that only the `cropBox` was empty, but it seemed like a good idea to attempt to fix a bunch of related cases all at once.
The current code will only consider the `cropBox` and `mediaBox` as equal when they both point to the *same* underlying Array. In the case where a PDF file actually specifies both boxes independently, with the exact same values in each, the comparison will currently fail and lead to an unneeded intersection computation.
With the changes to the `StreamType`/`FontType` "enums" in PR 11029, one unfortunate result is that `getStats` now *always* returns empty Arrays. Something that everyone, myself included, apparently missed is that you obviously cannot index an Array with Strings :-)
I wrongly assumed that the unit-tests would catch any bugs, but they apparently suffered from the same issue as the code in `src/core/`.
Another possible option could perhaps be to use `Set`s, rather than objects, but that will require larger changes since `LoopbackPort` (in `src/display/api.js`) doesn't support them.
Firefox telemetry supports using string labels now. Convert our integers
that we used for categories to just use strings.
The upstream work will happen in:
https://bugzilla.mozilla.org/show_bug.cgi?id=1566882
There's a number of spots in the current code, and tests, where `cancel` methods are not called with appropriate arguments (leading to Promises not being rejected with Errors as intended).
In some cases the cancel `reason` is implicitly set to `undefined`, and in others the cancel `reason` is just a plain String. To address this inconsistency, the patch changes things such that cancelling is done with `AbortException`s everywhere instead.
Add a work-around, in `glyphlist.js`, for bad PDF generators which use a non-standard `/f_f` string in the `Encoding` dictionary when referring to the ff ligature (issue 11016)
This patch will not incur any (measurable) overhead, since the glyphlist is already quite long and one more entry won't really matter, which is important given that this sort of PDF corruption ought to be very rare.
Furthermore, this patch purposely does *not* add a bunch of similarly modified ligature names on pure speculation. Any similar additions, for other ligatures, should only be made if there's real-world examples of PDF files where that's actually necessary.
For very large and complex PDF files this will help performance slightly, since `EvaluatorPreprocessor.read` is called a lot during parsing in the worker.
This patch was tested using the PDF file from issue 2618, i.e. http://bugzilla-attachments.gnome.org/attachment.cgi?id=226471, using the following manifest file:
```
[
{ "id": "issue2618",
"file": "../web/pdfs/issue2618.pdf",
"md5": "",
"rounds": 200,
"type": "eq"
}
]
```
This gave the following results when comparing this patch against the `master` branch:
```
-- Grouped By browser, stat --
browser | stat | Count | Baseline(ms) | Current(ms) | +/- | % | Result(P<.05)
------- | ------------ | ----- | ------------ | ----------- | --- | ----- | -------------
Firefox | Overall | 200 | 3402 | 3358 | -43 | -1.28 | faster
Firefox | Page Request | 200 | 1 | 1 | 0 | 26.71 |
Firefox | Rendering | 200 | 3401 | 3357 | -44 | -1.28 | faster
```
For very large and complex PDF files this will help performance slightly, since `Parser.shift` is called *a lot* during parsing.
This patch was tested using the PDF file from issue 2618, i.e. http://bugzilla-attachments.gnome.org/attachment.cgi?id=226471 (with well over *four million* `Parser.shift` calls for just the one page), using the following manifest file:
```
[
{ "id": "issue2618",
"file": "../web/pdfs/issue2618.pdf",
"md5": "",
"rounds": 100,
"type": "eq"
}
]
```
This gave the following results when comparing this patch against the `master` branch:
```
-- Grouped By browser, stat --
browser | stat | Count | Baseline(ms) | Current(ms) | +/- | % | Result(P<.05)
------- | ------------ | ----- | ------------ | ----------- | --- | ----- | -------------
Firefox | Overall | 100 | 3386 | 3322 | -65 | -1.92 | faster
Firefox | Page Request | 100 | 1 | 1 | 0 | -8.08 |
Firefox | Rendering | 100 | 3385 | 3321 | -65 | -1.92 | faster
```
The way that this method handles documents without an `ID` entry in the Trailer dictionary feels overly complicated to me. Hence this patch adds `getByteRange` methods to the various Stream implementations[1], and utilize that rather than manually calling `ensureRange` when computing a fallback `fingerprint`.
---
[1] Note that `PDFDocument` is only ever initialized with either a `Stream` or a `ChunkedStream`, hence why the `DecodeStream.getByteRange` method isn't implemented.
The `finalize` helper function has only a *single* call-site, and furthermore it's just a one-liner too. Furthermore it's only ever called with a `Promise` as its argument, meaning that it's unnecessarily convoluted as well (i.e. the `Promise.resolve()` part shouldn't be necessary).
Hence this code can be both simplified *and* inlined at its only call-site instead.
Currently `wrapReason` is manually called at *every* `resolveOrReject` call-site, despite it being completely unnecessary unless there's an actual error being handled. This is obviously inefficient, and it's easy enough to avoid by having `resolveOrReject` handle this only when actually needed.
Note that, in the old code, there was a code-path which could prevent this from happening thus affecting future cleanup.
Furthermore, ensure that we'll always attempt to cleanup when handling the 'PageError' message, similar to the code in e.g. the `PDFPageProxy._renderPageChunk` method.
The `receivingOperatorList` property is currently tracked *twice* in the rendering code, both directly and inversely through the `intentState.operatorList.lastChunk` boolean. This type of double bookkeeping is never a good idea, since it's just too easy for the properties to accidentally fall out of sync.
In this case there's even a `cleanup`-related bug caused by this, which means that `PDFPageProxy._tryCleanup` will never be able to discard any data if there's an error on the worker-thread (as handled through the 'PageError' message).
Hence the simplest solution seems, at least to me, to update `PDFPageProxy._tryCleanup` to replace the `intentState.receivingOperatorList` check with a `!intentState.operatorList.lastChunk` check and completely remove the former property.
*Please note:* A a similar change was attempted in PR 5005, but it was subsequently backed out in PR 5069.
Unfortunately I don't think anyone ever tried to debug *exactly* why it didn't work, since it ought to have worked, and having re-tested this now I'm not able to reproduce the problem any more. However, given just how inefficient the current code is, with thousands of strictly unnecessary function calls for each `find` invocation, I'd really like to try fixing this again.
This reduces the total number of function calls, when reading the XRef table respectively when fetching uncompressed XRef entries.
Note in particular the `XRef.readXRefTable` method, where there're *two* back-to-back `isCmd` checks rather than just one.
A lot of the `new Parser()` call-sites look quite unwieldy/ugly as-is, with a bunch of somewhat randomly ordered arguments, which we can avoid by changing the constructor to accept an object instead. As an added bonus, this provides better documentation without having to add inline argument comments in the code.
See https://github.com/mozilla/eslint-plugin-no-unsanitized
Since we've generally never allowed e.g. `innerHTML`, which is enforced during review, there's only one linting failure with this patch. (Which is white-listed, according to the existing comment and the fact that it's test-only code.)
Since all other `IPDFStream` implementations live in their own files, it seems reasonable for these to do so as well.
Furthermore, converts all of the relevant code to ES6 classes and updates the interface definitions to mark a couple of methods `async`.
Given that `cleanupAfterRender` is already set for large images, when handling 'obj' messages, this patch *should* thus be safe in general (since otherwise there ought be existing bugs related to cleanup and printing).
The border `width` will instead fallback to the default value of `1`, rather than ignoring it altoghether, to also ensure that e.g. `LinkAnnotation`s become clickable as intended.
Fixes https://bugzilla.mozilla.org/show_bug.cgi?id=1552113
Usually when the worker is terminated it will also be completely destroyed/removed, which means that any global caches (such as the ones in `src/core/primitive.js`) should be automatically cleared in the process.
However, for certain ways of loading the `pdf.worker.js` file, e.g. passing in a re-usable worker to `getDocument`, using the `workerPort` functionality, or even disabling workers completely (even though this is never a good idea), the worker file may be kept in memory and these caches will not be cleared as expected.
Calling `someArray = []` will create a new Array, which seems completely unnecessary when it's sufficient to just call `someArray.length = 0` to achieve the same effect.
Even though I cannot imagine these particular cases having any noticeable performance impact, similar changes were made in `core/` code years ago since it's apparently more efficient memory wise.
The purpose of these caches is to reduce peak memory usage, by only ever having *a single* instance of a particular object.
However, as-is these caches are never cleared and they will thus remain until the worker is destroyed. This could very well have a negative effect on total memory usage, particularly for large/long documents, hence it seems to make sense to clear out these caches together with various other ones.
This is similar to the existing caching used to reduced the number of `Cmd` and `Name` objects.
With the `tracemonkey.pdf` file, this patch changes the number of `Ref` objects as follows (in the default viewer):
| | Loading the first page | Loading *all* the pages |
|----------|------------------------|-------------------------|
| `master` | 332 | 3265 |
| `patch` | 163 | 996 |
The specification states that `CreationDate` is only available for
markup annotations instead of for all annotation types.
Moreover, popup annotations are not markup annotations according to the
specification, so the creation date inheritance from the parent
annotation is also removed there (note that only the modification date
is used in e.g., the viewer).
This includes the information in the core and display layers. The
date parsing logic from the document properties is rewritten according
to the specification and now includes unit tests.
Moreover, missing unit tests for the color of a popup annotation have
been added.
Finally the styling of the popup is changed slightly to make the text a
bit smaller (it's currently quite large in comparison to other viewers)
and to make the drop shadow a bit more subtle. The former is done to be
able to easily include the modification date in the popup similar to how
other viewers do this.
Currently `handleColorN` will fallback to add a completely unparsed/unvalidated operator when no valid pattern was found. This is unfortunate, since it could very easily lead to a couple of different errors:
- `DataCloneError`s when attempting to send the data to the main-thread, e.g. when `args` is `Dict`/`Stream`.
- Errors in `getShadingPatternFromIR` on the main-thread, unless `args` just happens to have the expected format.
- Errors when actually attempting to render the pattern on the main-thread, since the `args` will most likely not have the expected format.
Hence it probably makes sense to error in `PartialEvaluator.handleColorN`, and having invalid patterns fail gracefully via the existing `ignoreErrors` code-paths instead.
It appears that this has been broken ever since PR 9089, which also introduced this code, since the `QueueOptimizer`/`NullOptimizer` choice was made based on the still undefined `this.intent` property.
Furthermore, fixing this also uncovered the fact that the `NullOptimizer.reset` method was missing.
First of all, while this simple approach appears to work OK in practice I'm not sure if it's the best way of addressing the problem (assuming that you even want to).
Second of all, while the solution implemented here only requires tracking/checking one new boolean in order for this to work, I'm nonetheless not entirely happy about this since it will add additional overhead (albeit *very* small) to the parsing of path operators in PDF documents just for a handful of *corrupt* ones.
This way we can avoid manually building a "document id" in multiple places in `evaluator.js`, and it also let's us avoid passing in an otherwise unnecessary `PDFManager` instance when creating a `PartialEvaluator`.
While PR 10714 did address the `disableRange=true` case, it also managed to "break" the `disableStream=true` case instead since the indeterminate loadingBar is now displayed when it shouldn't; sorry about that!
The solution is simple enough though, don't attempt to fallback to `_fullRequestReader.onProgress` when handling "incomplete" loading information.
Please see the specification, https://www.adobe.com/content/dam/acom/en/devnet/acrobat/pdfs/PDF32000_2008.pdf#M11.9.12864.1Heading.71.Viewer.Preferences
Furthermore, note that this patch *only* adds API support and unit-tests but does not attempt to integrate e.g. the `ViewerPreferences -> Direction` property into the viewer (which would be necessary to address issue 10736).
The reason for this is that it's not entirely clear to me exactly if/how that could be implemented; e.g. would it be as simple as setting the `dir` attribute on the `viewerContainer` DOM element, or will it be more complicated?
There's also the question of how the `ViewerPreferences -> Direction` value interacts with the `PageMode`, and this will generally require a fair bit of manual testing. Since the direction of the *entire* viewer depends on the browser locale, there's also a somewhat open question regarding what default value to use for different locales.
Finally, if the viewer supports `ViewerPreferences -> Direction` then I'm assuming that it will be necessary to allow users to override the default value, which will require (most likely) new `SecondaryToolbar` buttons and icons for those etc.
Hence this patch only lays the necessary foundation for eventually addressing issue 10736, but defers the actual implementation until later. (Time permitting, I'll try to look into the viewer part later.)
*Please note:* This patch purposely ignores `src/display/network.js`, since its support for progressive reading depends on the non-standard `moz-chunked-arraybuffer` responseType which is currently in the process of being removed.
The file `test/pdfs/annotation-caret-ink.pdf` is already available in
the repository as a reference test for this since I supplied it for
another patch that implemented ink annotations.
This mirrors the canvas implementation where we ignore these operators.
This avoids console spam regarding unimplemented operators we're not
interested in.
For the Tracemonkey paper, we're now down to one warning about tiling
patterns which is in fact a valid one.
In particular, this should reduce intermediate string creation by using
template strings and reduce variable lookup times by removing unneeded
variables and caching `this.current` in more places.
With PR 10675 having fixed the completely broken `disableRange=true` setting in the Firefox version of PDF.js, I couldn't help but noticing that loading progress is never reported properly in that case.
Currently loading progress is only reported for the `rangeProgress` chrome-event, which obviously isn't dispatched with `disableRange=true` set. However, the `progressiveRead` chrome-event includes loading progress as well, but this information isn't being used in any way.
Furthermore, the `PDFDataRangeTransport.onDataProgress` method wasn't able to handle "complete" loading information, and neither was `PDFDataTransportStream._onProgress` since that method would only ever attempt to report it through a RangeReader (which won't exist when `disableRange=true` is set).
Note how at https://mozilla.github.io/pdf.js/api/ it's being described as API docs, however `src/core/annotation.js` is not part of the public API.
Furthermore, given that the code residing in the `src/core/` folder is run in a worker-thread, it's not even accessible on the main-thread (since `postMessage` is being used to transfer the data).
Hence the different API methods simply returns a "proxy" to the underlying data, but not actually the same objects and data structures as in the worker-thread itself; thus it doesn't make a whole lot of sense to expose this in API docs as far as I'm concerned.
Finally, the patch fixes a small JSDoc related typo in `src/display/api.js` when referring to the `TextStyle` typedef.
The `moz-chunked-arraybuffer` responseType is a non-standard property, which has been subsumed by the Fetch API, and it's in the process of being removed from Firefox; please see https://bugzilla.mozilla.org/show_bug.cgi?id=1120171 and https://bugzilla.mozilla.org/show_bug.cgi?id=1411865
*Please note:* Rather than waiting for both `Fetch` *and* `ReadableStream` to be available in e.g. a Firefox ESR version (which is probably going to be 68 at the earliest), let's just decide that PDF.js release `2.1.266` will be the last one with `moz-chunked-arraybuffer` support and land this patch (since nothing should outright break without it anyway).
Note how `XRef.fetchUncompressed`, which is used *a lot* for most PDF documents, is calling the `makeSubStream` method without providing a `length` argument.
In practice this results in the `makeSubStream` method, on the `ChunkedStream` instance, calling the `ensureRange` method with `NaN` as the end position, thus resulting in no data being requested despite it possibly being necessary.
This may be quite bad, since in this particular case it will lead to a new `ChunkedStream` being created *and* also a new `Parser`/`Lexer` instance. Given that it's quite possible that even the very first `Parser.getObj` call could throw `MissingDataException`, this could thus lead to wasted time/resources (since re-parsing is necessary once the data finally arrives).
You obviously need to be very careful to not have `ChunkedStream.makeSubStream` accidentally requesting the *entire* file, hence its `this.end` property is of no use here, but it should be possible to at least check that the `start` of the data is present before any potentially expensive parsing occurs.
This transform resulted in an incorrectly positioned object when the
bounding box's upper-left corner did not start at (0,0), because
the translation was not reverted. This patch adds the missing transform.
The test file (tiling-pattern-box.pdf) is based on the PDF from #2825.
All but the first cube (including the PDF data) have been removed.
To trigger the bug that is fixed by this commit, I changed the BBox of
the first pattern from "[ 0 0 596 842]" to "[90 0 596 842]". Without
this patch, the dashed vertical line that intersects the corners at A
and E would disappear.
The new test file (tiling-pattern-large-steps.pdf) was manually created,
to have the following characteristics:
- Large xstep and ystep (90000)
- Page width is 4000 (which is larger than MAX_PATTERN_SIZE)
- Visually, the page consists of a red rectangle with a black border,
surrounded by a 50 unit white padding.
- Before patch: blurry; After patch: sharp
Fixes#6496Fixes#5698Fixes#1434Fixes#2825
Without this some fonts may incorrectly end up with matching `hash`es, thus breaking rendering since we'll not actually try to load/parse some of the fonts.
Note that `PartialEvaluator.preEvaluateFont` will return an empty string when no hash was computed. This will complete short-circuit the `fontAlias` comparison in `PartialEvaluator.loadFont`, since fonts which are totally different will then match if their `hash`es are empty.
Currently if trying to set `disableRange=true` in the built-in PDF Viewer in Firefox, either through `about:config` or via the URL hash, the PDF document will never load. It appears that this has been broken for a couple of years, without anyone noticing.
Obviously it's not a good idea to set `disableRange=true`, however it seems that this bug affects the PDF Viewer in Firefox even with default settings:
- In the case where `initialData` already contains the *entire* file, we're forced to dispatch a range request to re-fetch already available data just so that file loading may complete.
- (In the case where the data arrives, via streaming, before being specifically requested through `requestDataRange`, we're also forced to re-fetch data unnecessarily.) *This part was removed, to reduce the scope/risk of the patch somewhat.*
In the cases outlined above, we're having to re-fetch already available data thus potentially delaying loading/rendering of PDF files in Firefox (and wasting resources in the process).
This function is currently called with the `OperatorList` instance as its argument, hence I cannot think of any good reason for not just moving it into the `OperatorList` properly. (This will also help with other planned changes regarding the `ImageCache` functionality.)
By transfering `ArrayBuffer`s you can avoid having two copies of the same data, i.e. one copy on each of the worker/main-thread, for data that's used only *once* on the worker-thread.
Note how the code in [`PDFImage.createMask`](80135378ca/src/core/image.js (L284-L285)) goes to great lengths to actually enable tranfering of the image data. However in [`PartialEvaluator.buildPaintImageXObject`](80135378ca/src/core/evaluator.js (L336)) the `cached` property is always set to `true`, which disqualifies the image data from being transfered; see [`getTransfers`](80135378ca/src/core/operator_list.js (L552-L554)).
For most ImageMask data this patch won't matter, since images found in the `/Resources -> /XObject` dictionary will always be indexed by name. However for *inline* images which contains ImageMask data, where only "small" images are cached (in both `parser.js` and `evaluator.js`), the current code will result in some unnecessary memory usage.
This will further help reduce the amount of image data that's currently being held alive, by explicitly removing the `src` attribute.
Please note that this is mostly relevant for browsers which do not support `URL.createObjectURL`, or where `disableCreateObjectURL` was manually set by the user, since `blob:` URLs will be revoked (see the previous patch).
However, using `about:memory` (in Firefox) it does seem that this may also be generally helpful, given that calling `URL.revokeObjectURL` won't invalidate the image data itself (as far as I can tell).
Natively supported JPEG images are sent as-is, using a `blob:` or possibly a `data` URL, to the main-thread for loading/decoding.
However there's currently no attempt at releasing these resources, which are held alive by `blob:` URLs, which seems unfortunately given that images can be arbitrarily large.
As mentioned in https://developer.mozilla.org/en-US/docs/Web/API/URL/createObjectURL the lifetime of these URLs are tied to the document, hence they are not being removed when a page is cleaned-up/destroyed (e.g. when being removed from the `PDFPageViewBuffer` in the viewer).
This is easy to test with the help of `about:memory` (in Firefox), which clearly shows the number of `blob:` URLs becomming arbitrarily large *without* this patch. With this patch however the `blob:` URLs are immediately release upon clean-up as expected, and the memory consumption should thus be considerably reduced for long documents with (simple) JPEG images.
Note how `PDFDocumentProxy.destroy` is a nothing more than an alias for `PDFDocumentLoadingTask.destroy`. While removing the latter method would be a breaking API change, there's still room for at least some clean-up here.
The main changes in this patch are:
- Stop providing a `PDFDocumentLoadingTask` instance *separately* when creating a `PDFDocumentProxy`, since the loadingTask is already available through the `WorkerTransport` instance.
- Stop tracking the `PDFDocumentProxy` instance on the `WorkerTransport`, since that property is completely unused.
- Simplify the 'Multiple `getDocument` instances' unit-tests by only destroying *once*, rather than twice, for each document.
For Type3 fonts text-selection is often not that great, and there's a couple of heuristics used to try and improve things. This patch simple extends those heuristics a bit, and fixes a pre-existing "naive" array comparison, but this all feels a bit brittle to say the least.
The existing Type3 test-coverage isn't that great in general, and in particular Type3 `text` tests are few and far between, hence why this patch adds *two* different new `text` tests.
Notable changes:
- Remove the `return this;` from the `MurmurHash3_64.update` method, since it's completely unused and doesn't make a lot of sense.
- Remove the loop(s) from the `MurmurHash3_64.hexdigest` method, since creating a temporary array and then looping over it is wasteful given how simple this can be written with modern JavaScript.
Given that the function is (purposely) independent of the verbosity level and that its message is worded to only apply on the main-thread, there's no reason to duplicate this across the built `pdf.js`/`pdf.worker.js` files.
Currently for every single parsed/rendered page there's no less than *four* `Date.now()` calls being made on the worker-side. This seems totally unnecessary, since the result of these calls are, by default, not used for anything *unless* the verbosity level is set to `INFO`.
The default size of these canvases seem to be `300 x 150` (two orders of magnitude larger than the ones in PR 10597), which probably is sufficient enough to matter since there's one such canvas for each textLayer that's rendered in the viewer.
Also fixes the incorrect rejection reason, i.e. one using a string rather than an `Error`, in the `TextLayerRenderTask.cancel` method.
While this particular canvas may be small, there can still be an arbitrarily large number of them (one per page rendered), which can/will eventually add up memory wise. This can be easily avoided by using the `cachedCanvases` abstraction instead, which will ensure that the `isFontSubpixelAAEnabled` canvas is removed together with other temporary canvases in `CanvasGraphics.endDrawing`.
The `src/shared/util.js` file is being bundled into both the `pdf.js` and `pdf.worker.js` files, meaning that its code is by definition duplicated.
Some main-thread only utility functions have already been moved to a separate `src/display/display_utils.js` file, and this patch simply extends that concept to utility functions which are used *only* on the worker-thread.
Note in particular the `getInheritableProperty` function, which expects a `Dict` as input and thus *cannot* possibly ever be used on the main-thread.
This file (currently) contains not only DOM-specific helper functions/classes, but is used generally for various helper code relevant for main-thread functionality.
*Hopefully this patch makes sense, since I cannot claim to fully understand this function.*
With the changes made in PR 3354 *some* Type3 glyph outlines are no longer rendering correctly, since the final paths were being accidentally ignored.
The fact that Type3 fonts are not very common in PDF documents, and that most Type3 glyphs are unaffected by this regression, probably explains why this has gone unnoticed since 2013.
After PR 9340 all glyphs are now re-mapped to a Private Use Area (PUA) which means that if a font fails to load, for whatever reason[1], all glyphs in the font will now render as Unicode glyph outlines.
This obviously doesn't look good, to say the least, and might be seen as a "regression" since previously many glyphs were left in their original positions which provided a slightly better fallback[2].
Hence this patch, which implements a *general* fallback to the PDF.js built-in font renderer for fonts that fail to load (i.e. are rejected by the sanitizer). One caveat here is that this only works for the Font Loading API, since it's easy to handle errors in that case[3].
The solution implemented in this patch does *not* in any way delay the loading of valid fonts, which was the problem with my previous attempt at a solution, and will only require a bit of extra work/waiting for those fonts that actually fail to load.
*Please note:* This patch doesn't fix any of the underlying PDF.js font conversion bugs that's responsible for creating corrupt font files, however it does *improve* rendering in a number of cases; refer to this possibly incomplete list:
[Bug 1524888](https://bugzilla.mozilla.org/show_bug.cgi?id=1524888)
Issue 10175
Issue 10232
---
[1] Usually because the PDF.js font conversion code wasn't able to parse the font file correctly.
[2] Glyphs fell back to some default font, which while not accurate was more useful than the current state.
[3] Furthermore I'm not sure how to implement this generally, assuming that's even possible, and don't really have time/interest to look into it either.
- The only existing call-site, of this method, is never passing more than *one* font at a time anyway.
- As far as I can remember, this functionality has never actually been used (caveat: I didn't check the git history).
- This allows simplification of the method, especially by making use of the fact that it's now asynchronous.
- It should be just as easy to call `BaseFontLoader.bind` from within a loop, rather than having the loop in the method itself.
Currently all fonts are using the `_queueLoadingCallback` method to determine when they have been loaded[1]. However in most cases this is just adding unnecessary overhead, especially with `BaseFontLoader.bind` now being asynchronous, given how fonts are loaded:
- For fonts loaded using the Font Loading API, it's already possible to easily tell when a font has been loaded simply by checking the `loaded` promise on the FontFace object itself.
- For browsers, e.g. Firefox, which support synchronous font loading it's already assumed that fonts are immediately available.
Hence the `_queueLoadingCallback` method is moved into the `GenericFontLoader`, such that it's only utilized for fonts which are loaded using CSS.
---
[1] In the "fonts loaded using CSS" case, this is already a hack anyway as outlined in the comments.
pdf.js had a problem when copying characters on supplementary planes
(0xPPXXXX where PP is nonzero). This is because certain methods of
PartialEvaluator use classic String.fromCharCode instead of ES6's
String.fromCodePoint.
Despite the fact that readToUnicode method *tried* to parse out-of-UCS2
code points by parsing UTF-16BE, it was inadequate because
String.fromCharCode only supports UCS-2 range of Unicode.
Unsurprisingly IE11 doesn't support this, so a polyfill is needed since otherwise the sidebar can no longer be opened.
Also, simplifies the existing `classList.toggle` polyfill.
This polyfill is currently used in only *one* file, i.e. `src/display/api.js`, and only when trying to build a *fallback* `workerSrc` path.
Given that the global `workerSrc` should *always* be set[1] when using the PDF.js library[2], and that the fallback `workerSrc` should only be regarded as a best-effort solution anyway, there isn't a particularily strong reason to keep the compatibility code in my opinion.
---
[1] Other supported options include setting the global `workerPort`, or passing in a `PDFWorker` instance as part of the `getDocument` call.
[2] Which is clearly mentioned in the JSDocs in `src/display/worker_options.js`.
This piggybacks of the existing `cancel` functionality, to ensure that any pending operations are closed *and* that any temporary canvases are actually being removed.
Also simplifies `finishPaintTask` in `PDFPageView.draw` slightly, by converting it to an async function.
All objects in the PDF document follow this pattern:
```
0000000001 0 obj
<<
% Some content here...
>>
endobj
0000000002 0 obj
<<
% More content here...
endobj
```
Based on the discussion in https://bugzilla.mozilla.org/show_bug.cgi?id=1521413, this patch simply removes the `ReadableStream` polyfill completely from MOZCENTRAL builds.
With this patch, the size of the `gulp mozcentral` build target is thus further reduced (building on PR 10470):
| | `build/mozcentral`
|-------|-------------------
|master | 3 339 666
|patch | 3 209 572
With https://bugzilla.mozilla.org/show_bug.cgi?id=1505122 landing in Firefox 65, the native `ReadableStream` implementation is now enabled by default in Firefox.
Obviously it would be nice to simply stop bundling the polyfill in MOZCENTRAL builds altogether, however given that it's still possible to disable[1] `ReadableStream` this is probably not a good idea just yet.
Nonetheless, now that native support is available, it seems unnecessary (and wasteful) to keep bundling the polyfill twice[2] in MOZCENTRAL builds. Hence this patch, which contains a suggest approach for packing the polyfill in a *separate* file which is then *only* loaded if/when needed.
With this patch, the size of the `gulp mozcentral` build target is thus reduced accordingly:
| | `build/mozcentral`
|-------|-------------------
|master | 3 461 089
|patch | 3 340 268
Besides the PDF.js files taking up less space in Firefox this way, the additional benefit is that there's (by default) less code that needs to be loaded and parsed when the PDF Viewer is used which also cannot hurt.
---
[1] In `about:config`, by toggling the `javascript.options.streams` preference.
[2] Once in the `build/pdf.js` file, and once in the `build/pdf.worker.js` file.
In many cases in the code you don't actually care about the index itself, but rather just want to know if something exists in a String/Array or if a String starts in a particular way. With modern JavaScript functionality, it's thus possible to remove a number of existing `indexOf` cases.
This will allow the Metadata to be successfully extracted from the PDF file in issue 10395.
Furthermore, this patch also fixes a bug in `Metadata.get` which causes the method to return `null` rather than an empty string or zero (since either ought to be allowed).
The error was triggered for a particular set of metadata, where an end tag was encountered without the corresponding begin tag being present in the data.
(The patch also fixes a minor oversight, from a recent PR, in the `SimpleDOMNode.nextSibling` method.)
Given that the issue, as filed, is incomplete since no PDF file was provided for debugging, this patch is really the best that we can do here. *Please note:* This patch will *not* enable the Metadata to be successfully parsed, but it should at least prevent the errors.
This method creates quite a few intermediate strings on each call and
it's called often, even for smaller documents like the Tracemonkey
document. Scrolling from top to bottom in that document resulted in
14126 strings being created in this method. With this commit applied,
this is reduced to 2018 strings.
This method creates quite a few intermediate strings on each call and
it's called often, even for smaller documents like the Tracemonkey
document. Scrolling from top to bottom in that document resulted in
12936 strings being created in this method. With this commit applied,
this is reduced to 3610 strings.
The `toString` method always creates two string objects (for the 'R'
character and for the `num` concatenation) and in the worst case
creates three string objects (one more for the `gen` concatenation).
For the Tracemonkey paper alone, this resulted in 12000 string
objects when scrolling from the top to the bottom of the document.
Since this is a hot function, it's worth minimizing the number of string
objects, especially for large documents, to reduce peak memory usage.
This commit refactors the `toString` method to always create only one
string object.
For PDF documents with sufficiently broken XRef tables, it's usually quite obvious when you need to fallback to indexing the entire file. However, for certain kinds of corrupted PDF documents the XRef table will, for all intents and purposes, appear to be valid. It's not until you actually try to fetch various objects that things will start to break, which is the case in the referenced issues[1].
Since there's generally a real effort being in made PDF.js to load even corrupt PDF documents, this patch contains a suggested approach to attempt to do a bit more validation of the XRef table during the initial document loading phase.
Here the choice is made to attempt to load the *first* page, as a basic sanity check of the validity of the XRef table. Please note that attempting to load a more-or-less arbitrarily chosen object without any context of what it's supposed to be isn't a very useful, which is why this particular choice was made.
Obviously, just because the first page can be loaded successfully that doesn't guarantee that the *entire* XRef table is valid, however if even the first page fails to load you can be reasonably sure that the document is *not* valid[2].
Even though this patch won't cause any significant increase in the amount of parsing required during initial loading of the document[3], it will require loading of more data upfront which thus delays the initial `getDocument` call.
Whether or not this is a problem depends very much on what you actually measure, please consider the following examples:
```javascript
console.time('first');
getDocument(...).promise.then((pdfDocument) => {
console.timeEnd('first');
});
console.time('second');
getDocument(...).promise.then((pdfDocument) => {
pdfDocument.getPage(1).then((pdfPage) => { // Note: the API uses `pageNumber >= 1`, the Worker uses `pageIndex >= 0`.
console.timeEnd('second');
});
});
```
The first case is pretty much guaranteed to show a small regression, however the second case won't be affected at all since the Worker caches the result of `getPage` calls. Again, please remember that the second case is what matters for the standard PDF.js use-case which is why I'm hoping that this patch is deemed acceptable.
---
[1] In issue 7496, the problem is that the document is edited without the XRef table being correctly updated.
In issue 10326, the generator was sorting the XRef table according to the offsets rather than the objects.
[2] The idea of checking the first page in particular came from the "standard" use-case for the PDF.js library, i.e. the default viewer, where a failure to load the first page basically means that nothing will work; note how `{BaseViewer, PDFThumbnailViewer}.setDocument` depends completely on being able to fetch the *first* page.
[3] The only extra parsing is caused by, potentially, having to traverse *part* of the `Pages` tree to find the first page.
If, as PR 10368 suggests, more parameters should be added to `getViewport` I think that it would be a mistake to not change the signature *first* to avoid needlessly unwieldy call-sites.
To not break any existing code and third-party use-cases, this is obviously implemented with a deprecation warning *and* with a working fallback[1] for the old method signature.
---
[1] This is limited to `GENERIC` builds, which should be sufficient.
Note that the OpenAction dictionary may contain other information besides just a destination array, e.g. instructions for auto-printing[1].
Given first of all that an arbitrary `Dict` cannot be sent from the Worker (since cloning would fail), and second of all that the data obviously needs to be validated, this patch purposely only adds support for fetching a destination from the OpenAction entry[2].
---
[1] This information is, currently in PDF.js, being included through the `getJavaScript` API method.
[2] This significantly reduces the complexity of the implementation, which seems fine for now. If there's ever need for other kinds of OpenAction to be fetched, additional API methods could/should be implemented as necessary (could e.g. follow the `getOpenActionWhatever` naming scheme).
The custom entries, provided that they exist *and* that their types are safe to include, are exposed through a new `Custom` infoDict entry to clearly separate them from the standard ones.
Fixes 5970.
Fixes 10344.
Given that Signature (Widget) annotations are currently not supported, since they cannot be validated, simply ignoring the `fieldValue` seems OK for now considering that attempting to blindly include unparsed/unvalidated data isn't very useful.
Fixes 10347.
The intent of the code, based on existing comments, is to perform a binary search. However, because of what appears to be a typo in the code responsible for computing the current search index, this code is always checking *every* entry (albeit only at the "final" node) starting from the last one.
According to the specification, see https://www.adobe.com/content/dam/acom/en/devnet/acrobat/pdfs/PDF32000_2008.pdf#G6.2384179, the keys of NameTree/NumberTree should be ordered.
For corrupt PDF files, which violate this assumption, we thus need to fallback to an exhaustive search in order to e.g. find all destinations.
*Please note:* Given that this only implements a fallback for the "final" node of the Tree, there's obviously a risk that the patch isn't sufficient for dealing with all kinds of out-of-order corruption. However, this kind of problem should be rare in practice, and without a real-world test-case it's difficult to implement a completely general solution (and there's obviously a question if you'd even want to).
This changes all occurrences of `var` to `let`/`const` in this code, and updates the signature of the constructor to use object destructuring for better readability (and self documentation).
Also, `useRequestAnimationFrame` is changed to a parameter and the `typeof window` check is now done *once* rather than at every `_scheduleNext` call.
This changes all occurrences of `var` to `let`/`const` in this code, and updates the signatures of a couple of methods to use object destructuring.
Finally, when creating `InternalRenderTask` instances *only* the necessary parameter are now provided, since passing through the `RenderParameters` as-is seems completely unnecessary.
These interfaces are already used in different files, in both the `src/core/` and `src/display/` folders, and having them reside in their own file seems a lot clearer and is also similar to the existing viewer interfaces.
As part of moving the `interface` definitions, they're also converted to ES6 classes.
First of all, note how there's currently *two* methods for checking if a certain object exists, which seems completely unwarranted.
Furthermore, the rarely used `getData` method was removed and its only callsite changed to use a combination of `PDFObjects.{has, get}` instead.
Finally, the methods were rearranged slightly, to bring the most important ones (for an API user) to the top of the class.
Note how nowhere in the code `canvasInRendering.get()` is ever called, and that this structure is really only used to store references to `<canvas>` DOM elements.
The reason for this being a `WeakMap` is probably because at the time we weren't using `core-js` polyfills yet, and since there already existed a manually implemented `WeakMap` polyfill it was probably simpler to use that.
Please note that, given the lack of a runnable example, I'm not totally sure if this first of all is enough to *completely* address the issue as filed and second of all if we actually want this new behaviour.
`CMapCompressionType` makes a lot of sense to export, for anyone attempting to implement a custom `CMapReaderFactory`; fixes 10148.
`PermissionFlag` likewise needs to be exported, since otherwise the result of the `getPermissions` API method becomes difficult to interpret; follow-up to 10033.
*Please note:* I'm totally fine with this patch being rejected, and the issue closed as WONTFIX; however these changes should address the issue if that's desired.
From a conceptual point of view, reporting loading progress doesn't really make a lot of sense for PDF files opened by passing raw binary data directly to `getDocument` (since obviously *all* data was loaded).
This is compared to PDF files loaded via e.g. `XMLHttpRequest` or the Fetch API, where the entire PDF file isn't available from the start and knowing the loading progress makes total sense.
However I can certainly see why the current API could be considered inconsistent, which isn't great, since a registered `onProgress` callback will never be called for certain `getDocument` calls.
The simplest solution to this inconsistency thus seem to be to ensure that `onProgress` is always called when handling the `DataLoaded` message, since that will *always* be dispatched[1] from the worker-thread.
---
[1] Note that this isn't guaranteed to happen, since setting `disableAutoFetch = true` often prevents the *entire* file from ever loading. However, this isn't relevant for the issue at hand, and is a well-known consequence of using `disableAutoFetch = true`; note how the default viewer even has a specialized code-path for hiding the loadingBar.
*This should have been part of PR 10139.*
In the event that a user has attempted to manually load the worker file on the main-thread, but somehow failed to do that correctly, there's a possibility that `getMainThreadWorkerMessageHandler` could throw. Considering how/where that helper function is being called, an error could still prevent `PDFDocumentLoadingTask` from completing (regardless if it's being resolved/rejected).
With the way that the `getWorkerSrc()` helper function is implemented now, there's no longer a particularly strong reason for keeping the global `pdfjsFilePath` variable around.
With this patch the fallback `workerSrc` will thus, assuming is wasn't already set, be set to the "pdfjsFilePath" which simplifies the `getWorkerSrc()` function and reduces the amount of global state.
Finally, the global `workerSrc` variable was renamed to prevent shadowing.
This should, hopefully, cover all the possible ways[1] in which "fake workers" are loaded. Given the different code-paths, adding unit-tests might not be that simple.
Note that in order to make this work, the various `fakeWorkerFilesLoader` functions were converted to return `Promises`.
---
[1] Unfortunately there's lots of them, for various build targets and configurations.
This attempts to reduced the level of indirection, and the amount of code, when dispatching `fileattachmentannotation` events, by removing the `PDFLinkService.onFileAttachmentAnnotation` method and just accessing `PDFLinkService.eventBus` directly in the `FileAttachmentAnnotationElement` constructor.
Given that other properties, such as `externalLinkTarget`/`externalLinkRel`, are already being accessed directly this pattern seems fine here as well.
The `started` timestamp is completely usused, and the `end` timestamp is currently[1] being used essentially like a boolean value.
Hence this code can be simplified to use an actual boolean value instead, which avoids potentially hundreds (or even thousands) of unnecessary `Date.now()` calls.
---
[1] Looking briefly at the history of this code, I cannot tell if the timestamps themselves were ever used for anything (except for tracking "boolean" state).
The `Font.loading` property is only ever used *once* in the code, whereas `Font.missingFile` is more widely used. Furthermore the name `loading` feels, at least to me, slight less clear than `missingFile`. Finally, note that these two properties are the inverse of each other.
Currently there's only a single spot in the code-base where `JpegImage.getData` is called, however it nonetheless seem like a good idea to ensure during tests that the `isSourcePDF` parameter is correctly set. (Especially considering that the PDF use-cases will break without it.)
Additionally, in `JpegImage._getLinearizedBlockData`, the code can be made a tiny bit more efficient by checking the value of `isSourcePDF` *first* to avoid useless checks (for the default PDF use-cases).
There have been lots of problems with trying to map glyphs to their unicode
values. It's more reliable to just use the private use areas so the browser's
font renderer doesn't mess with the glyphs.
Using the private use area for all glyphs did highlight other issues that this
patch also had to fix:
* small private use area - Previously, only the BMP private use area was used
which can't map many glyphs. Now, the (much bigger) PUP 16 area can also be
used.
* glyph zero not shown - Browsers will not use the glyph from a font if it is
glyph id = 0. This issue was less prevalent when we mapped to unicode values
since the fallback font would be used. However, when using the private use
area, the glyph would not be drawn at all. This is illustrated in one of the
current test cases (issue #8234) where there's an "ä" glyph at position
zero. The PDF looked like it rendered correctly, but it was actually not
using the glyph from the font. To properly show the first glyph it is always
duplicated and appended to the glyphs and the maps are adjusted.
* supplementary characters - The private use area PUP 16 is 4 bytes, so
String.fromCodePoint must be used where we previously used
String.fromCharCode. This is actually an issue that should have been fixed
regardless of this patch.
* charset - Freetype fails to load fonts when the charset size doesn't match
number of glyphs in the font. We now write out a fake charset with the
correct length. This also brought up the issue that glyphs with seac/endchar
should only ever write a standard charset, but we now write a custom one.
To get around this the seac analysis is permanently enabled so those glyphs
are instead always drawn as two glyphs.
It's no longer necessary since https://bugzilla.mozilla.org/show_bug.cgi?id=1305963 is fixed quite some time ago.
While we're here, mark the `cachedGetSinglePixelWidth` member as being
private and use ES6 syntax in the `getSinglePixelWidth` method.
The purpose of this patch is to provide a better default behaviour when `JpegImage` is used to parse standalone JPEG images with CMYK colour spaces.
Since the issue that the patch concerns is somewhat of a special-case, the implementation utilizes the already existing decode support in an attempt to minimize the impact w.r.t. code size.
*Please note:* It's always possible for the user of `JpegImage` to control image inversion, and thus override the new behaviour, by simply passing a custom `decodeTransform` array upon initialization.
Ensure that the built `PdfJsDefaultPreferences.jsm` file won't be affected/touched during tree-wide ESLint rule changes in `mozilla-central` (PR 9571 follow-up)
Apparently there's some PDF generators, in this case the culprit is "Nooog Pdf Library / Nooog PStoPDF v1.5", that manage to mess up PDF creation enough that endstream[1] commands actually become truncated.
*Please note:* The solution implemented here isn't perfect, since it won't be able to cope with PDF files that contains a *mixture* of correct and truncated endstream commands.
However, considering that this particular mode of corruption *fortunately* doesn't seem very common[2], a slightly less complex solution ought to suffice for now.
Fixes 10004.
---
[1] Scanning through the PDF data to find endstream commands becomes necessary, in order to determine the stream length in cases where the `Length` entry of the (stream) dictionary is missing/incorrect.
[2] I cannot recall having seen any (previous) issues/bugs with "Missing endstream" errors.
Reduces the amount of boilerplate code when defining the the sub-classes.
Please note that a couple of the closures were kept, since it's not (yet) possible to include helper functions inside of `class`es.
This property is not only completely unused now, it never actually appears to have been used. Even though the memory savings, from not initializing these extra typed arrays, won't be significant in the grand scheme of things it still seems completely unnecessary to keep allocating this data.
As far as I can tell, the main reason for the existence of `defaultColor` seem to be for documentation purposes. Hence the code is changed into comments instead, to keep the information around (but without the unnecessary allocations).
For proof-of-concept, this patch converts a couple of `Promise` returning methods to use `async` instead.
Please note that the `generic` build, based on this patch, has been successfully testing in IE11 (i.e. the viewer loads and nothing is obviously broken).
Being able to use modern JavaScript features like `async`/`await` is a huge plus, but there's one (obvious) side-effect: The size of the built files will increase slightly (unless `SKIP_BABEL == true`). That's unavoidable, but seems like a small price to pay in the grand scheme of things.
Finally, note that the `chromium` build target was changed to no longer skip Babel translation, since the Chrome extension still supports version `49` of the browser (where native `async` support isn't available).
Not only is this method completely unused *now*, looking through the history of the code it never appears to have been used for anything either.
Years ago `mainXRefEntriesOffset` was included when creating `XRef` instances, however it wasn't actually used for anything (the parameter was never checked, nor assigned to a property on `XRef`).
If this method ever becomes useful (again) it's easy enough to restore it thanks to version control, but including dead code in the builds just seems wasteful.
Please note that while this *improves* issue 9984 slightly (and likely others too), it's not a complete solution.
The remaining issues are related to the, more general, problems with the existing heuristics related to attempting to combine separate text items.
One of the `QueueOptimizer` cases wasn't updated to use `Uint8ClampedArray`s, which leads to inconsistent image data on the API side (but no actual rendering bugs, as far as I can tell).
To prevent future errors, a non-production/test-only `assert` was added to ensure that the relevant image data only uses `Uint8ClampedArray`s.
This commit is the first step towards implementing parsing for the
appearance streams of annotations.
Co-authored-by: Jonas Jenwald <jonas.jenwald@gmail.com>
Co-authored-by: Tim van der Meij <timvandermeij@gmail.com>
The current font type/subtype detection code is quite inconsistent/unwieldy. In some cases it will simply assume that the font dictionary is correct, in others it will somewhat "arbitrarily" check the actual font file (more of these cases have been added over the years to fix specific bugs).
As is evident from e.g. issue 9949, the font type/subtype detection code is continuing to cause issues. In an attempt to get rid of these hacks once and for all, this patch instead re-factors the type/subtype detection to *always* parse the font file.
Please note that, as far as I can tell, we still appear to need to rely on the composite font detection based on the font dictionary. However, even if the composite/non-composite detection would get it wrong, that shouldn't really matter too much given that there's basically only two different code-paths (for "TrueType-like" vs "Type1-like" fonts).
The font in the PDF is marked as a CIDFontType0, but the font file is
actually a true type font. To fully address this issue we should really
peek into the font file and try to determine what it is. However, this
is the first case of this issue, so I think this solution is acceptable for
now.
Fixes a stupid oversight on my part, since /Filter may (obviously) contain an Array, which resulted in unnecessary console warning spam in perfectly valid PDF files.
Note that it still makes sense to check that /Filter is actually a Name, before attempting to access its `name` property, but the warning should definitely be removed.
According to the PDF specification, see https://www.adobe.com/content/dam/acom/en/devnet/acrobat/pdfs/PDF32000_2008.pdf#page=45
> When using the JPXDecode filter with image XObjects, the following changes to and constraints on some entries in the image dictionary shall apply (see 8.9.5, "Image Dictionaries" for details on these entries):
>
> - Width and Height shall match the corresponding width and height values in the JPEG2000 data.
>
> - . . .
Hence it seems reasonable to use the Width/Height of the image data *itself*, rather than the image dictionary when there's a mismatch. Given that JPEG 2000 images are already being parsed, in order to obtain basic parameters, the actual Width/Height is readily available in the `PDFImage` constructor.
Given that the code is currently assuming that the /Filter entry is a `Name`, it cannot hurt to actually ensure that's the case.
Also fixes an error message, for JPEG 2000 images with unsupported ColorSpaces, since `this.numComps` hasn't been initialized when it's accessed during the `throw new Error()` invocation.
[api-minor] Add an `IsLinearized` property to the `PDFDocument.documentInfo` getter, to allow accessing the linearization status through the API (via `PDFDocumentProxy.getMetadata`)
Since PDF.js already supports range requests and streaming, not to mention chunked rendering, attempting to use the `Linearization` dictionary in `PDFDocument.getPage` probably isn't going to improve performance in any noticeable way.
Nonetheless, when `Linearization` data is available, it will allow looking up the first Page *directly* without having to descend into the `Pages` tree to find the correct object.
With the `builtInCMapCache` being a simple Object, it unfortunately means that the `Catalog.cleanup` method isn't resetting it as intended.
By just replacing the `builtInCMapCache` with an empty Object, existing references to it will not actually be updated. The result is that e.g. `Page` instances still keeps references to, what should have been removed, CMap data.
To fix these problems, the `builtInCMapCache` is converted into a `Map` instead (since it can be easily reset).
There was a (somewhat) recent question on IRC about accessing the linearization status of a PDF document, and this patch contains a simple way to expose that through already existing API methods.
Please note that during setup/parsing in `PDFDocument` the linearization data is already being fetched and parsed, provided of course that it exists. Hence this patch will *not* cause any additional data to be loaded.
With this file now being a proper (ES6) module, it's no longer (technically) necessary for this structure to be lazily initialized. Considering its size, and simplicity, I therefore cannot see the harm in letting `DocumentInfoValidators` just be simple Object instead.
While I'm not aware of any bugs caused by the current code, it cannot hurt to add an `isDict` check in `PDFDocument.documentInfo` (since the current code assumes that `infoDict` being defined implies it also being a Dictionary).
Finally, the patch also converts a couple of `var` to `let`/`const`.
Note first of all that `PDFDocument` will be initialized with either a `Stream` or a `ChunkedStream`, and that both of these have `length` getters. Secondly, the `PDFDocument` constructor will assert that the `stream` has a non-zero (and positive) length. Hence there's no point in checking `stream.length` in the `linearization` getter.
For most other `DecodeStream` based streams, we'll attempt to estimate the minimum `buffer` length based on the raw stream data. The purpose of this is to avoid having to unnecessarily re-size the `buffer`, thus reducing the number of *intermediate* allocations necessary when decoding the stream data.
However, currently no such optimization is attempted for `StreamsSequenceStream`, and given that they can often be quite large that seems unfortunate. To improve this, at least somewhat, this patch utilizes the raw sizes of the `StreamsSequenceStream` sub-streams to estimate the minimum required `buffer` length.
Most likely this patch won't have a huge effect on memory consumption, however for pathological cases it should help reduce peak memory usage slightly.
One example is the PDF file in issue 2813, where currently the `StreamsSequenceStream` instances would grow their `buffer`s as `2 MiB -> 4 MiB -> 8 MiB -> 16 MiB -> 32 MiB`. With this patch, the same stream `buffers`s grow as `8 MiB -> 16 MiB -> 32 MiB`, thus avoiding a total of `12 MiB` of *intermediate* allocations (since there's two `StreamsSequenceStream` used, for rendering/text-extraction).
Without providing useful (custom) error messages for the `no-restricted-globals` rule, see https://eslint.org/docs/rules/no-restricted-globals, it's quite likely that the rule will be incorrectly disabled rather than the required globals being imported as intended.
To reduced duplication of the `no-restricted-globals` rule in multiple `.eslintrc` files, it's instead moved to the top-level `.eslintrc` file and disabled as needed on a folder/file basis outside of `/src` and `/web`.
With the new XML parser, see PR 9573, the referenced PDF file now causes `getMetadata` to fail when incomplete XML tags are encountered. This provides a simple, and hopefully generally useful, work-around that may also help prevent future bugs.
(Without being able to reproduce nor even understand the other (non XML) errors mentioned in issue 8884, I'd say that this patch is enough to close that one as fixed.)
Compared to all the other (static) methods in `Util`, the `toRoman` one looks slightly out of place. Even more so considering that `Util` is being exposed through `pdfjsLib`, where access to a Roman numerals conversion method doesn't make much sense.
*I was feeling bored; so this is a very quick, and somewhat naive, attempt at fixing the bug.*
The breaking error, i.e. `Error during font loading: invalid array length`, was thrown when attempting to re-size the `stack` to a *negative* length when parsing the CALL functions.
Fixes https://bugzilla.mozilla.org/show_bug.cgi?id=1473809.
PR 6141 changed `CompiledFont.compileGlyph` to, in the general case, return an Array. However, that PR apparenly forgot to update the no-glyph, empty-glyph, and endchar-glyph code-path and a String was still being (incorrectly) returned.
Given the way that `FontFaceObject.getPathGenerator` (on the API side) is implemented, this shouldn't have caused any bugs despite the Worker possible returning unexpected data.
This moves/exposes the `URL` polyfill similarily to the existing `ReadableStream` polyfill, rather than exposing it globally, to avoid interfering with any "outside" code.
Both the `URL` and `ReadableStream` polyfills are now exposed on the `pdfjsLib` object, such that they are accessible to the viewer components.
Furthermore, the `no-restricted-globals` ESLint rule is also enabled to prevent accidental usage of the native `URL`/`ReadableStream` implementations directly in the `src/` and `web/` folders; see also https://eslint.org/docs/rules/no-restricted-globals
Addresses the remaining TODO in https://github.com/mozilla/pdf.js/projects/6
Currently if `RenderTask.cancel` is called *immediately* after rendering was started, then by the time that `InternalRenderTask.initializeGraphics` is called rendering will already have been cancelled.
However, we're still inserting the canvas into the `canvasInRendering` map, thus breaking any future attempts at re-rendering using the same canvas. Considering that `InternalRenderTask.cancel` always removes the canvas from the map, I cannot imagine that we'd ever want to re-add it *after* rendering was cancelled (it was likely just a simple oversight in PR 8519).
Fixes 9456.
This wasn't included in PR 9245, since all the API options were still global at that time.
Writing the unit-tests also uncovered an issue with `getOperatorList` not starting the "Page Request" timer.
Incomplete path operators, in particular, can result in fairly chaotic rendering artifacts, as can be observed on page four of the referenced PDF file.
The initial (naive) solution that was attempted, was to simply throw a `FormatError` as soon as any invalid (i.e. too short) operator was found and rely on the existing `ignoreErrors` code-paths. However, doing so would have caused regressions in some files; see the existing `issue2391-1` test-case, which was promoted to an `eq` test to help prevent future bugs.
Hence this patch, which adds special handling for invalid path operators since those may cause quite bad rendering artifacts.
You could, in all fairness, argue that the patch is a handwavy solution and I wouldn't object. However, given that this only concerns *corrupt* PDF files, the way that PDF viewers (PDF.js included) try to gracefully deal with those could probably be described as a best-effort solution anyway.
This patch also adjusts the existing `warn`/`info` messages to print the command name according to the PDF specification, rather than an internal PDF.js enumeration value. The former should be much more useful for debugging purposes.
Fixes https://bugzilla.mozilla.org/show_bug.cgi?id=1443140.
This patch avoids choosing a (possible) 'trailer' dictionary that `XRef.parse` and/or the `Catalog` constructor/methods will reject anyway.
Since `XRef.indexObjects` is already parsing the entire PDF file, the extra dictionary look-ups added here shouldn't matter much. Besides, this is a fallback code-path that only applies to corrupt PDF files anyway.
Note that the `Catalog` constructor, and some of its methods, are already enforcing that the 'Root' dictionary is valid/well-formed. However, by doing additional validation already in `XRef.parse` there's a slightly larger chance that corrupt PDF files could be successfully parsed/rendered.
With the current code line-breaks are accepted not just after an operator, but after a decimal point as well. When looking at this again, the latter case seems prone to cause false positives and might also interfere with subsequent patches.
Hence this is code is adjusted to actually do what the original commit message says, and nothing more.
Please note that the standalone `pdf.image_decoders.js` file will be including the complete `src/shared/util.js` file, despite only using parts of it.[1] This was done *purposely*, to not negatively impact the readability/maintainability of the core PDF.js code.
Furthermore, to ensure that the compatibility is the same in the regular PDF.js library *and* in the the standalone image decoders, `src/shared/compatibility.js` was included as well.
To (hopefully) prevent future complaints about the size of the built `pdf.image_decoders.js` file, a few existing async-related polyfills are being skipped (since all of the image decoders are completely synchronous).
Obviously this required adding a couple of pre-processor statements, but given that these are all limited to "compatibility" code, I think this might be OK!?
---
[1] However, please note that previous commits moved `PageViewport` and `MessageHandler` out of `src/shared/util.js` which reduced its size.
The purpose of this patch is to hopefully provide *slightly* better user ergonomics, if/when the PDF.js image decoders are used standalone.
This implementation is (basically) reverting the changes in PR 9386, in conjunction with code from the `parse` method found at https://github.com/notmasteryet/jpgjs/blob/master/src/pdfjs.js
Obviously it's still not possible to render non-embedded fonts as paths, but in this way the rest of the page will at least be allowed to continue rendering.
*Please note:* Including the 14 standard fonts in PDF.js probably wouldn't be *that* difficult to implement. (I'm not a lawyer, but the fonts from PDFium could probably be used given their BSD license.)
However, the main blocker ought to be the total size of the necessary font data, since I cannot imagine people being OK with shipping ~5 MB of (additional) font data with Firefox. (Based on the reactions when the CMap files were added, and those are only ~1 MB in size.)
Since `ColorSpace` now depends on the native clamping of `Uint8ClampedArray`, this patch adds non-production/test-only `assert`s to enforce that the expected TypedArray is used for the output.
These `assert`s are purposely *not* included in PRODUCTION builds since that would break rendering completely, as opposed to "only" displaying some weird colours, when a `Uint8Array` was used. Furthermore, these are mostly added to help catch explicit developer errors when working with the `ColorSpace` and `PDFImage` code.
Since the tests (currently) run with the `pdf.worker.js` file built, i.e. with `PRODUCTION = true` set, there's no simple way to add e.g. `assert` calls for both non-production *and* test-only builds without also affecting PRODUCTION builds.
The built-in image decoders are already using `Uint8ClampedArray` when returning data, and this patch simply extends that to the rest of the image/colorspace code.
As far as I can tell, the only reason for using manual clamping/rounding in the first place was because TypedArrays used to be polyfilled (using regular arrays). And trying to polyfill the native clamping/rounding would probably have been had too much overhead, but given that TypedArray support is required in PDF.js version `2.0` that's no longer a concern.
*Please note:* Because of different rounding behaviour, basically `Math.round` in `Uint8ClampedArray` respectively `Math.floor` in the old code, there will be very slight movement in quite a few existing test-cases. However, the changes should be imperceivable to the naked eye, given that the absolute difference is *at most* `1` for each RGB component when comparing `master` and this patch (see also the updated expectation values in the unit-tests).
The built-in image decoders are already returning data as `Uint8ClampedArray`, and subsequently the JPEG/JBIG2/JPX streams are as well. However, for general streams we obviously don't want to force the use of `Uint8ClampedArray` unless an "Image" is actually being decoded.
Hence this patch, which adds a parameter that allows the caller of the `getBytes`/`peekBytes` methods to force a `Uint8ClampedArray` (rather than a `Uint8Array`) to be returned.
Not only is the `Util.loadScript` helper function unused on the Worker side, even trying to use it there would throw an Error (since `document` isn't defined/available in Workers).
Hence this helper function is moved, and its code modernized slightly by having it return a Promise rather than needing a callback function.
Finally, to reduced code duplication, the "new" loadScript function is exported and used in the viewer.
The various classes have `this._errored` and `this._reason` properties, where the first one is a boolean indicating if an error was encountered and the second one contains the actual `Error` (or `null` initially).
In practice this means that errors are basically tracked *twice*, rather than just once. This kind of double-bookkeeping is generally a bad idea, since it's quite easy for the properties to (accidentally) get into an inconsistent state whenever the relevant code is modified.
Rather than using a separate boolean, we can just as well check the "error" property directly (since `null` is falsy).
---
Somewhat unrelated to this patch, but `src/display/node_stream.js` is currently *not* handling errors in a consistent or even correct way; compared with `src/display/network.js` and `src/display/fetch_stream.js`.
Obviously using the `createResponseStatusError` utility function, from `src/display/network_utils.js`, might not make much sense in a Node.js environment. However at the *very* least it seems that `MissingPDFException`, or `UnknownErrorException` when one cannot tell that the PDF file is "missing", should be manually thrown.
As is, the API (i.e. `getDocument`) is not returning the *expected* errors when loading fails in Node.js environments (as evident from the `pending` API unit-test).
Since the old comment mentions a now unsupported browser, let's update it such that someone won't accidentally conclude that the code in question can be removed.
This special handling was added in PR 8567, but was made redundant in PR 8721 which stopped sending everything but the kitchen sink to the Worker side.
Since `PDFPageProxy` already provide getters for all the data returned by `GetPage` (in the Worker), there isn't any compelling reason for accessing the `pageInfo` directly on `PDFPageProxy`.
The patch also changes the `GetPage` handler, in `src/core/worker.js`, to use modern JavaScript features.
Since `PDFDocumentProxy` already provide getters for all the data returned by `GetDoc` (in the Worker), there isn't any compelling reason for accessing the `pdfInfo` directly on `PDFDocumentProxy`.
After PR 8617 the `PDFManagerReady` message handler function, in `src/display/api.js`, is now a no-op. Hence it seems completely unnecessary to keep sending this message from `src/core/worker.js`.
With native typed array support now being mandatory in PDF.js, since version 2.0, this probably isn't a huge problem even though the current code seems wrong (it was changed in PR 6571).
Note how in the `!(data instanceof Uint8Array)` case we're currently attempting to send `handler.send('test', 'main', false);` to the main-thread, which doesn't really make any sense since the signature of the method reads `send(actionName, data, transfers) {`.
Hence the data that's *actually* being sent here is `'main'`, with `false` as the transferList, which just seems weird. On the main-thread, this means that we're in this case checking `data && data.supportTypedArray`, where `data` contains the string `'main'` rather than being falsy. Since a string doesn't have a `supportTypedArray` property, that check still fails as expected but it doesn't seem great nonetheless.
Since all the built-in PDF.js image decoders now return their data as `Uint8ClampedArray`, for consistency `JpegDecode` on the main-thread should be doing the same thing; follow-up to PR 8778.
The signature of the `PDFWorker.fromPort` method, in addition to the `PDFWorker` constructor, was changed in PR 9480.
Hence it's probably a good idea to add a bit more validation to `PDFWorker.fromPort`, to ensure that it won't fail silently for an API consumer that updates to version 2.0 of the PDF.js library.
With version 2.0, native support for typed arrays is now a requirement for using the PDF.js library; see PR 9094 where the old polyfills were removed.
Hence the `isTypedArraysPresent` check, when setting up fake workers, no longer serves any purpose here and can thus be removed.
There's no good reason, as far as I can tell, to duplicate the functionality of the `LoopbackPort` in the unit-tests. The only difference between the implementations is that `LoopbackPort` mimics the (native) structured cloning, however that shouldn't matter here since the tests are only sending "simple" data (strings respectively arrays with numbers).
Furthermore the patch also changes `LoopbackPort` to default to using "structured cloning" and deferred invocation of the listeners, since native typed array support is now a requirement for using the PDF.js library.
The `MessageHandler` itself, and its assorted helper functions, are currently the single largest[1] piece of code in the `src/shared/util.js` file. By moving this code into its own file, `src/shared/util.js` thus becomes smaller and more manageable.
The `fontScale` property was added in PR 1531, see commit b312719d7e in particular, apparently for the sole purpose of supporting the "acroforms" example.
However, the `fontScale` property was never used anywhere else in the code-base, and after the modernization of the "acroforms" example in PR 8030 it's been completely unused.
Finally, note that there's also a (more suitably named) `scale` property on `PageViewport` instances, which contains the exact same information as the property being removed here.
I made some mistakes when trying to make the content_disposition.js
compatible with non-modern browsers (IE/Edge).
Notably, text decoding was usually skipped because of the inverted
logical check at the top of `textdecode`.
I verified that this new version works as expected, as follows:
1. Visit 55c71eb44e/test/
and get test-content-disposition.js
also get test-content-disposition.node.js if using Node.js,
or get test-content-disposition.html if you use a browser.
2. Modify `test-content-disposition.node.js` (or the HTML file) and
change `../extension/content-disposition.js` to `PDFJS-content_disposition.js`
3. Copy the `getFilenameFromContentDispositionHeader` function from
`content_disposition.js` (i.e. the file without the trailing exports)
and save it as `PDFJS-content_disposition.js`.
4. Run the tests (`node test-content-disposition.node.js` or by opening
`test-content-disposition.html` in a browser).
5. Confirm that there are no failures: "Finished all tests (0 failures)"
The code has a best-efforts fallback for Microsoft Edge, which lacks the
TextDecoder API. The fallback only supports the common UTF-8 encoding.
To simulate this in a test, modify `PDFJS-content_disposition.js` and
deliberately throw an error before `new TextDecoder`. There will be two
failures because we don't want to include too much code to support text
decoding for non-UTF-8 encodings in Edge
```
test-content-disposition.js:265 Assertion failed: Input: attachment; filename*=ISO-8859-1''%c3%a4
Expected: "ä"
Actual : "ä"
test-content-disposition.js:268 Assertion failed: Input: attachment; filename*=ISO-8859-1''%e2%82%ac
Expected: "€"
Actual : "€"
```
Please note that while the current code works, both in the viewer and the unit-tests, it can leave the `WorkerTransport._passwordCapability` Promise in a pending state.
In the `PasswordRequest` handler, in src/display/api.js, we're returning the Promise from a `capability` object (rather than just a "plain" Promise). While an error thrown anywhere within this handler was fortunately enough to propagate it to the Worker side, it won't cause the Promise (in `WorkerTransport._passwordCapability`) to actually be rejected.
Finally note that while we're now catching errors in the `PasswordRequest` handler, those errors are still propagated to the Worker side via the (now) rejected Promise and the existing `return this._passwordCapability.promise;` line.
This prevents warnings about uncaught Promises, with messages such as "Error: Worker was destroyed during onPassword callback", when running the unit-tests both in browsers *and* in Node.js/Travis.
This should provide a better out-of-the-box experience when using PDF.js in a Node.js environment, since it's missing native support for both `@font-face` and `Image`.
Please note that this change *only* affects the default values, hence it's still possible for an API consumer to override those values when calling `getDocument`.
Also, prevents "ReferenceError: document is not defined" errors, when running the unit-tests in Node.js/Travis.
This avoids the initialization of, potentially thousands of, unnecessary `Stream` objects, by getting the required number of bytes directly instead.
Given the special behaviour, when `length === 0`, of the `getBytes`/`skip` methods, it's also necessary to handle that particular case to prevent errors when encountering empty CharStrings.
*This is a final piece of clean-up of code that I recently wrote, after which I'm done :-)*
When the `getMainThreadWorkerMessageHandler` function was added, in PR 9385, it did so by basically introducing a `web/app.js` dependency in `src/display/api.js` through the `window.pdfjsNonProductionPdfWorker` property[1]. Even though this is limited to non-`PRODUCTION` mode, i.e. `gulp server`, it still seems unfortunate to have that sort of viewer dependency in the API code itself.
With the new, much nicer and shorter, names introduced in PR 9565 we can remove this non-`PRODUCTION` hack and just use `window.pdfjsWorker` in both the viewer and the API regardless of the build mode.
---
[1] It didn't seem correct to piggy-back on the `window.pdfjsDistBuildPdfWorker` property in non-`PRODUCTION` mode.
The current PageLabel dictionary validation code won't catch some (unlikely) forms of corruption. For example: a `Type`/`S` entry being `null`/`0`/empty string, a `P`/`St` entry being `null`/`0`.
Please note: I'm not aware of any bugs caused by the old code, but I've had this patch sitting locally for some time and figured it couldn't hurt to submit it.
The `getPageSizeInches` method was implemented on `PDFDocumentProxy`, which seems conceptually wrong since the size property isn't global to the document but rather specific to each page. Hence the method is moved into `PDFPageProxy`, as `get pageSizeInches` instead to address this.
Despite the fact that new API functionality was implemented, no unit-tests were added. To prevent issues later on, we should *always* ensure that new functionality has at least some test-coverage; something that this patch also takes care of.
The new `PDFDocumentProperties._parsePageSize` method seemed unnecessary convoluted. Furthermore, in the "no data provided"-case it even returned incorrect data (an array, rather than the expected object).
Finally, the fallback strings didn't actually agree with the `en-US` locale. This inconsistency doesn't look too great, and it's thus addressed here as well.
Chrome 60 and earlier does not include credentials (cookies) in requests
made with fetch, regardless of extension permissions. This was fixed in
61.0.3138.0 by
2e231cf052
This patch disables the fetch backend in all affected Chrome versions.
The browser detection is done by checking for a change that coincides
with the release of Chrome 61.
Test case:
1. Copy the `isChromeWithFetchCredentials` function from the patch.
2. Run it in the JS console of Chrome and verify the return value.
Verified results:
- 49.0.2623.75 - false (earliest supported version by us)
- 60.0.3112.90 - false (last major version affected by bug)
- 61.0.3163.100 - true (first major version without bug)
- 65.0.3325.146 - true (current stable)
Test case 2:
1. Build the extension (`gulp chromium`) and load it in Chrome.
2. Open the developer tools, and open any PDF file.
3. In the "Network tab" of the developer tools, look at "request type".
In Chrome 60-: Should be "xhr"
In Chrome 61+: Should be "fetch"
This function combines the logic of two separate methods into one.
The loop limit is also a good thing to have for the calls in
`src/core/annotation.js`.
Moreover, since this is important functionality, a set of unit tests and
documentation is added.
It's only used in two places in the class and those callsites can
directly get the information from the dictionary, which is more readable
and avoids an additional method call.
With options being moved from the global `PDFJS` object and into `getDocument`, a side-effect is that we're now passing in a fair number of useless parameters to the various transport/network streams.
Even though this doesn't *currently* cause any problems, it nonetheless seem like a good idea to explicitly provide the parameters that are actually necessary.
One additional complication with removing this option from the global `PDFJS` object, is that the viewer currently needs to check `disableAutoFetch` in a couple of places. To address this I'm thus proposing adding a getter in `PDFDocumentProxy`, to allow checking the *actually* used values for a particular `getDocument` invocation.
Unfortunately, as far as I can tell, we still need the ability to adjust certain API options depending on the browser environment in PDF.js version `2.0`. However, we should be able to separate this from the general compatibility code in the `src/shared/compatibility.js` file.
I don't understand why the previous way importing the polyfills didn't work, and I don't have time to try and figure it out, however this patch seems to fix things.
Fixes 9514.
Fixes 9516.
Compared to most other options currently/previously residing on the global `PDFJS` object, some of the Worker specific ones (e.g. `workerPort`/`workerSrc`) probably cannot be moved into options provided directly when initializing e.g. `PDFWorker`.
The reason is that in some cases, e.g. the Webpack examples, we try to provide Worker auto-configuration and I cannot see a good solution for that use-case if we completely remove the globally available Worker configuration.
However inline with previous patches for PDF.js version `2.0`, it does seem like a worthwhile goal to move away from storing options directly on the global `PDFJS` object, since that is a pattern we should avoid going forward. Especially since one of the (eventual) goals is to attempt to *completely* remove the global `PDFJS` object, and rely solely on exporting/importing the needed functionality.
By introducing the `GlobalWorkerOptions` we thus have larger flexibility in the future, if/when the global `PDFJS` object will be removed.
With PDF.js version `2.0` we'll only support browsers with built-in `TypedArray` functionality, hence there doesn't seem to be any good reason not to implement this now.
Fixes 4888.
In order to simplify things, the undocumented `enableStats` option was removed and `pdfBug` is now instead used to enabled general debugging *and* page request/rendering stats.
Considering that in the default viewer the `stats` was only used when debugging was also enabled, this simplification (code wise) definitely seem worthwhile to me.
This removes the `PDFJS.externalLinkTarget`/`PDFJS.externalLinkRel` dependency from the viewer components, but please note that as a *temporary* solution the default viewer still uses it.
This removes the `PDFJS.imageResourcesPath` dependency from the viewer components and the test-suite, but please note that as a *temporary* solution the default viewer still uses it.
Unfortunately, as far as I can tell, we still need the ability to adjust certain viewer options depending on the browser environment in PDF.js version `2.0`. However, we should be able to separate this from the general compatibility code in the `src/shared/compatibility.js` file.
The reason for the bug is that we're only computing a checksum of the image data itself, but completely ignore the inline dictionary. The latter is important, since in practice it's not uncommon for inline images to be identical but use e.g. different ColourSpaces.
There's obviously a couple of different ways that we could compute a hash/checksum of the dictionary.
Initially I tried using `MurmurHash3_64` to compute a hash of the keys/values in the dictionary. Unfortunately this approach turned out to be *way* too slow in practice, especially for PDF files with a huge number of inline images; in particular issue 2618 would regresses quite badly with this solution.
The solution that is instead implemented in this patch, is to compute a checksum of the dictionary contents. While this is a much simpler, not to mention a lot more efficient, solution there's one drawback associated with it:
If the contents of inline image dictionaries are ordered differently, they will not be considered equal with this approach which could thus lead to failures to cache repeated inline images. In practice this doesn't seem to be a problem in any of the PDF files I've tested, and generally I'd rather err on the side of *not* caching given that too aggressive caching can easily lead to rendering bugs.
One small, but somewhat annoying, complication is that by the time `Parser.makeInlineImage` is called, we no longer know the *exact* stream position where the inline image dictionary starts. Having access to that information is crucial here, and the easiest solution I could come up with is to track this in the current `Lexer` instance.[1]
With the patch, we're thus able to fix the referenced issues without incurring large regressions in problematic cases such as issue 2618.
Fixes 9398; also improves/fixes the `issue8823` reference test.
---
[1] Obviously I'd have preferred if this patch could be limited to `Parser.makeInlineImage`, without the need for this "hack", but I'm not sure what that'd look like here.
This rule is available from https://www.npmjs.com/package/eslint-plugin-mozilla, and is enforced in mozilla-central. Note that we have the necessary `Array`/`String` polyfills and that most cases have already been fixed, see PRs 9032 and 9434.
This rule is available from https://www.npmjs.com/package/eslint-plugin-mozilla, and is enforced in mozilla-central. Note that we have a polyfill for `ChildNode.remove()` and that most cases have already been fixed, see PRs 8056 and 8138.
Fallback to the built-in JPEG decoder when browser decoding fails, and attempt to handle JPEG images with DNL (Define Number of Lines) markers (issue 8614)
First of all, note how in both `fetch_stream.js` and `node_stream.js` we always overwrite the `this._contentLength` property even when the response headers doesn't actually contain any (valid) length information. This could thus result in the `length` parameter, as passed to the network stream, being completely ignored despite having no better information available.
Secondly, in `node_stream.js` the `this._isRangeSupported` property wasn't always updated correctly based on the response headers.
Please refer to the specification, found at https://www.w3.org/Graphics/JPEG/itu-t81.pdf#page=49
Given how the JPEG decoder is currently implemented, we need to know the value of the scanLines parameter (among others) *before* parsing of the SOS (Start of Scan) data begins.
Hence the best solution I could come up with here, is to re-parse the image in the *hopefully* rare case of JPEG images that include a DNL (Define Number of Lines) marker.
Fixes 8614.
This works by making `PartialEvaluator.buildPaintImageXObject` wait for the success/failure of `loadJpegStream` on the API side *before* parsing continues.
Please note that in practice, it should be quite rare for the browser to fail loading/decoding of a JPEG image. In the general case, it should thus not be completely surprising if even `src/core/jpg.js` will fail to decode the image.
Since `loadJpegStream` is only used at a *single* spot in the code-base, and given that it's very heavily tailored to the calling code (since it relies on the data structure of `PDFObjects`), this patch simply inlines the code in `src/display/api.js` instead.
This method currently requires a fair number of parameters, which creates quite unwieldy call-sites. When invoking `buildPaintImageXObject`, you have to remember not only which arguments to supply, but also the correct order, to prevent run-time errors.
This commit is the first step for extracting a base class for the
`AES128Cipher` and the `AES256Cipher` classes. The objective here is to
make code changes (not altering the logic) to make the implementations
as similar as possible as found by creating a diff of both classes.
In particular, we extract the key size and cycles of repetitions
constants since they are different for AES-128 and AES-256. Moreover, we
rename functions to be similar.
In the `AES256Cipher` class, there was an additional assignment to
`this` in the decryption function. However, this was unnecessary because
the assignment would also be done when the loop was exited.
In the JPEG images in the referenced PDF file, the DHT (Define Huffman Tables) segments contain more data than expected based on the length parameter.
Fixes 9425.
Despite this patch removing the `disableWorker` option itself, please note that we'll still fallback to loading the worker file(s) on the main-thread when running in environments without proper Web Worker support.
Furthermore it's still possible, even with this patch, to force the use of fake workers by manually loading the necessary file using a `<script>` tag on the main-thread.[1]
That way, the functionality of the now removed `SINGLE_FILE` build target and the resulting `build/pdf.combined.js` file can still be achieved simply by adding e.g. `<script src="build/pdf.worker.js"></script>` to the HTML (obviously with the path adjusted as needed).
Finally note that the `disableWorker` option is a performance footgun, and unfortunately many existing third-party examples actually use it without providing any sort of warning/justification.
---
[1] This approach is used in the default viewer, since certain kind of debugging may be easier if the code is running directly on the main-thread.
This method returns the currently used `workerSrc`, which thus allows obtaining the fallback `workerSrc` value (e.g. when the option wasn't set by the user).
Please note that this build target, and the resulting `build/pdf.combined.js` file, is equivalent to setting the `PDFJS.disableWorker` option to `true` which is a performance footgun.
Add comments with supported browser versions where missing.
Method:
- Use MDN compat tables if available.
- Otherwise test in Chrome (31+) otherwise.
(the Chrome Web Store does not update older versions of
Chrome, so probably nobody is interested in even older
versions, even though there is an existing comment for
Chrome<29 at `document.currentScript`).
It turns out that PR 9245 unfortunately broke benchmarking completely, sorry about that!
The bug is that we were attempting to reset the current instance of `StatTimer`, instead of creating a new one as was previously done. By resetting the current instance, the `StatTimer` data fetched in `test/driver.js` is now wiped out since it points to the *same* underlying object.
This re-use of a `StatTimer` instance was asked for during review, and unfortunately I didn't test this thoroughly enough before submitting the final version of the PR.[1]
---
[1] Note that while I did test the benchmarking scripts with that PR *before* initially submitting it, I did however forget to do that after addressing the review comments which might explain why this problem went unnoticed.
This patch updates the `IPDFStreamReader` interface and ensures that the interface/implementation of `network.js`, `fetch_stream.js`, `node_stream.js`, and `transport_stream.js` all match properly.
The unit-tests are also adjusted, to more closely replicate the actual behaviour of the various actual `IPDFStreamReader` implementations.
Finally, this patch adjusts the use of the Content-Disposition filename when setting the title in the viewer, and adds `PDFDocumentProperties` support as well.
The `fetch` API is only supported for http(s), even in Chrome extensions.
Because of this limitation, we should use the XMLHttpRequest API when the
requested URL is not a http(s) URL.
Fixes#9361
These were removed in PR 9170, since they were unused in the browsers that we'll support in PDF.js version `2.0`.
However looking at the output of Travis, where a subset of the unit-tests are run using Node.js, there's warnings about `btoa` being undefined. This doesn't appear to cause any errors, which probably explains why we didn't notice this before (despite PR 9201).
Please refer to the PDF specification, in particular http://www.adobe.com/content/dam/acom/en/devnet/acrobat/pdfs/PDF32000_2008.pdf#G7.3801570
> A colour space shall be specified in one of two ways:
> - Within a content stream, the CS or cs operator establishes the current colour space parameter in the graphics state. The operand shall always be name object, which either identifies one of the colour spaces that need no additional parameters (DeviceGray, DeviceRGB, DeviceCMYK, or some cases of Pattern) or shall be used as a key in the ColorSpace subdictionary of the current resource dictionary (see 7.8.3, "Resource Dictionaries"). In the latter case, the value of the dictionary entry in turn shall be a colour space array or name. A colour space array shall never be inline within a content stream.
>
> - Outside a content stream, certain objects, such as image XObjects, shall specify a colour space as an explicit parameter, often associated with the key ColorSpace. In this case, the colour space array or name shall always be defined directly as a PDF object, not by an entry in the ColorSpace resource subdictionary. This convention also applies when colour spaces are defined in terms of other colour spaces.
This is something that I noticed while attempting to debug https://bugzilla.mozilla.org/show_bug.cgi?id=1374945.
Just looking at the code, the `YRsiz` parameter seemed immediately wrong and the fact that every component used the *same* data also looked strange.
Comparing with the specification, see https://www.itu.int/rec/dologin_pub.asp?lang=e&id=T-REC-T.800-200208-S!!PDF-E&type=items#page=37, confirmed that this is indeed incorrect.
Note that I haven't got any example of a PDF file that is fixed by this patch, but that might be more luck than anything else. Manually checking a couple of files with included JPEG 2000 images, the `Csiz`/`XRsiz`/`YRsiz` parameters were `1` which could explain why this hasn't been an issue before.
Obviously we shouldn't generally make changes to `core` code without adding tests, but in this case I'm simply not sure how to obtain/create one. However, since the existing code doesn't make sense this patch could hopefully be deemed acceptable anyway.
Since multiple empty lines is virtually unused in the code-base, and the few cases that do exist look like "typos", let's enforce greater consistency here; please see https://eslint.org/docs/rules/no-multiple-empty-lines.
It is only used in a few places to handle prefixing style properties if
necessary. However, we used it only for `transform`, `transformOrigin`
and `borderRadius`, which according to Can I Use are supported natively
(unprefixed) in the browsers that PDF.js 2.0 supports. Therefore, we can
remove this class, which should help performance too since this avoids
extra function calls in parts of the code that are called often.
I've been looking into the remaining point in 8637 about blurry images, to see if we could perhaps improve the rendering quality slightly there. After quite a bit of debugging, it seems that the issue is limited to certain progressive JPEG images.
As mentioned previously, I've got no detailed knowledge of the JPEG format, but this patch does seem to improve things quite a bit for the images in question.
Squinting at https://searchfox.org/mozilla-central/rev/6c33dde6ca02b389c52e8db3d22494df8b916f33/media/libjpeg/jdphuff.c#492-639, it seems reasonable that we should take the sign of the data into account. Furthermore, looking at the specification in https://www.w3.org/Graphics/JPEG/itu-t81.pdf#page=118, the "F.2.4.3 Decoding the binary decision sequence for non-zero DC differences and AC coefficients" section even contains a description of this (even though I cannot claim to really understand the details).
The bug that this patch fixes is limited to the built-in JPEG decoder, and was unearthed by PR 9260. The underlying issue has existed since PR 6984, where the contents of this patch ought to have been included (if it weren't for the fact that we had no *easy* way to test `src/core/jpg.js` back then).
*Please note:* The slight movement in the reference test is a result of using the `src/core/jpg.js` decoder, rather than the native browser one.
This was an oversight in PR 9095, which unfortunately breaks rendering in some PDF files (e.g. the one from issue 6737).
It thus appears that we don't have any test-coverage for this code-path, and given the relative complexity of the PDF files affected by this bug I wasn't able to easily create a reduced test-case.
*Please note:* The linked test-case included in this patch is currently *not* rendered correctly (that'd be the PR 6606), but it at least gives us some test-coverage here.
Initially I just implemented the unit tests, but quickly found that they
were failing my expectation of having a size of 256 items. Some of them
did contain 256 items and some did not. I looked up various resources
and figured that they indeed all need to have 256 items. One of the good
resources is https://github.com/davidben/poppler/blob/master/poppler/FontEncodingTables.cc
Aside from some missing `notdef` (empty string) entries at the end of
the arrays, which I assume causes issues since it may cause
out-of-bounds array access which in JavaScript gives `undefined`, there
was a `notdef` entry missing in the `MacExpertEncoding`, causing the
entries after that to be shifted. This fix for this is similar to the
one in #8589.
The unit tests verify that, for known encoding names, the return value
is not only an array, but that it is also of the right length and
contains only strings.
The PDF file in the issue uses a number of *embedded* versions of Lucida fonts, but for some reason does *not* embed the LucidaSans-Demi font. According to https://en.wikipedia.org/wiki/Lucida#Usages that one should be bold, so we can at least improve rendering here (even though it won't look perfect).
Fixes 9291.
I recall being confused as to the purpose of the `encrypted` property all the way back when working on PR 4750.
Looking at the history, this property was added in PR 1698 when password support was added to the API/viewer. However, its only purpose seem to have been to facilitate the addition of a `isEncrypted` function in the API. That function never, as far as I can tell, saw any use and was unceremoniously removed in PR 4144.
Since we want to avoid sending all non-essential data early during initial document loading (e.g. PR 4750), it seems correct to get rid of the `encrypted` property. Especially since it hasn't even been exposed in the API for over three years, with no complaints that I'm aware of.
Finally note that the `encrypt` property on the `XRef` instance isn't tied to the code that's being removed here. Given that we're calling `PDFDocument.parse` during `createDocumentHandler` in the worker which, via `PDFDocument.setup`, calls `XRef.parse` where the `Encrypt` data (if it exists) is always parsed.
This patch refactors the searching for 'endobj', to try and find the next occurance of "obj" and then check if it was in fact an 'endobj' and continue searching otherwise.
This approach is used to avoid having to first find 'endobj', and then re-check the entire contents of the object and having to run (potentially expensive) regular expressions on arbitrary long strings.
Fixes 9105.
Please note that while this could be considered a regression in user-facing behaviour, I'm not convinced that it's really a regression as such since prior to PR 8912 the Metadata would fail to parse (with an XML error) and thus be ignored when setting the viewer title.
With the refactored Metadata parsing we're now able to parse this, which uncovered issues with a subset of broken Ghostscript Metadata that uses HTML character names.
Fixes https://bugzilla.mozilla.org/show_bug.cgi?id=1424938
It is quite confusing that the custom function is called `log2` while it
actually returns the ceiling value and handles zero and negative values
differently than the native function.
To resolve this, we add a comment that explains these differences and
make the function use the native `Math` functions internally instead of
using our own custom logic. To verify that the function does what we
expect, we add unit tests.
All browsers except for IE support `Math.log2` for quite a long time
already (see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Math/log2).
For IE, we use the core-js polyfill.
According to the microbenchmark at https://jsperf.com/log2-pdfjs/1,
using the native functions should also be faster, in my testing almost
six times as fast.
Note that no other image stream implements a special `getBytes` method, which makes `JpegStream` look somewhat odd.
I'm actually not sure what purpose this methods serves, since I successfully ran all tests locally with it commented out. Furhermore, I also ran tests with an added `if (length && length !== this.bufferLength) { throw new Error('length mismatch'); }` check, and didn't get a single test failure in that case either.
Looking at the history, it seems that this code originated back in PR 4528, but as far as I can tell there's no mention in either commit messages nor PR comments of why it was necessary to add a "special" `getBytes` function for the `JpegStream`.
My assumption is that there's a good reason why this method was added, e.g. to address a *specific* regression in one of the reference tests. However, I did check out commit 58f697f977 locally and ran tests with this method commented out, and there didn't seem to be any image-related failures in that case either!?
Hence I'm suggesting that we attempt to simplify this code slightly be removing this special `getBytes` method. However, please note that there's perhaps a *small* risk of regressions in an edge-case where we currently have insufficient test-coverage.
Unless the debugging tools (i.e. `PDFBug`) are enabled, or the `browsertest` is running, the `PDFPageProxy.stats` aren't actually used for anything.
Rather than initializing unnecessary `StatTimer` instances, we can simply re-use *one* dummy class (with static methods) for every page. Note that by using a dummy `StatTimer` in this way, rather than letting `PDFPageProxy.stats` be undefined, we don't need to guard *every* single stats collection callsite.
Since it wouldn't make much sense to attempt to use `PDFPageProxy.stats` when stat collection is disabled, it was instead changed to a "private" property (i.e. `PDFPageProxy._stats`) and a getter was added for accessing `PDFPageProxy.stats`. This getter will now return `null` when stat collection is disabled, making that case easy to handle.
For benchmarking purposes, the test-suite used to re-create the `StatTimer` after loading/rendering each page. However, modifying properties on various API code from the outside in this way seems very error-prone, and is an anti-pattern that we really should avoid at all cost. Hence the `PDFPageProxy.cleanup` method was modified to accept an optional parameter, which will take care of resetting `this.stats` when necessary, and `test/driver.js` was updated accordingly.
Finally, a tiny bit more validation was added on the viewer side, to ensure that all the code we're attempting to access is defined when handling `PDFPageProxy` stats.
There's a number of issues with the fonts in the referenced PDF file. First of all, they contain broken `ToUnicode` data (`NUL` bytes all over the place). However even if you skip those, the `ToUnicode` data appears to contain nothing but a `IdentityH` CMap which won't help provide a proper glyph mapping.
The real issue actually turns out to be that the PDF file uses the "Calibri" font[1], but doesn't include any font files. Since that one isn't a standard font, and uses a fairly different CID to GID map compared to the standard fonts, we're not able to render the file even remotely correct.
To work around this, I'm thus proposing that we include a (incomplete) glyph map for Calibri, and fallback to the standard Helvetica font. Obviously this isn't going to look perfect, but it's really the best that we can hope to achieve given that the PDF file is missing the necessary font data.
Finally, please note that none of the PDF readers I've tried (Adobe Reader, PDFium in Chrome) were able to extract the text (which isn't very surprising, given the broken `ToUnicode` data).
Fixes 9195.
---
[1] According to Wikipedia, see https://en.wikipedia.org/wiki/Calibri, Calibri is (primarily) a Windows font.
In some fonts, the included `ToUnicode` data is incomplete causing text-selection to not work properly. For simple fonts that contain encoding data, we can manually build a `ToUnicode` map to attempt to improve things.
Please note that since we're currently using the `ToUnicode` data during glyph mapping, in an attempt to avoid rendering regressions, I purposely didn't want to amend to original `ToUnicode` data for this text-selection edge-case.
Instead, I opted for the current solution, which will (hopefully) give slightly better text-extraction results in PDF file with incomplete `ToUnicode` data.
According to the PDF specification, see [section 9.10.2](http://www.adobe.com/content/dam/acom/en/devnet/acrobat/pdfs/PDF32000_2008.pdf#G8.1873172):
> A conforming reader can use these methods, in the priority given, to map a character code to a Unicode value.
> ...
Reading that paragraph literally, it doesn't seem too unreasonable to use *different* methods for different charcodes.
Fixes 8229.
