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.
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.
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.
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 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.
Since this patch will now treat (some) `NUL` bytes as "ASCII", the number of `followingBytes` checked are thus increased to (hopefully) reduce the risk of introducing new false positives.
Fixes 8823.
This is a trivial follow-up to PR 5383, and it's a bit strange that this has been wrong since late 2014 without anyone noticing (maybe because inline images aren't too common).
So, apparently code works better if you actually spell correctly, who knew ;-)
Fixes 8613.
Please note that the `glyphlist.js` and `unicode.js` files are converted to CommonJS modules instead, since Babel cannot handle files that large and they are thus excluded from transpilation.
Please see http://eslint.org/docs/rules/object-shorthand.
For the most part, these changes are of the search-and-replace kind, and the previously enabled `no-undef` rule should complement the tests in helping ensure that no stupid errors crept into to the patch.
*Please note:* The rendering of the PDF file in issue 8061 first regressed in PR 7039, and then PR 7493 exacerbated the problem even further by causing an infinite loop.
In this particular case, when errors were encountered inside of the `Lexer.getObject` method *itself*, we didn't advance the stream position. This thus caused an inifinite loop in `parseCMap`, since the exact same character was then parsed over and over again.
Fixes 8061.
Given the nature of `EOF` and `isEOF`, it seems to me that they really ought to be placed in `core/primitives.js` instead.
In general, it doesn't seem great to have to depend on the entire `core/parser.js` file for such simple primitives/helper functions.
In particular, while `core/ps_parser.js` is completely separate from `core/parser.js` with regards to its function, it still depends on the latter for just *one* primitive.
Note that compared to e.g. PR 7389, this will not reduce the number of dependencies for `core/ps_parser`, however the new dependency IMHO makes more sense.
*Please note that most of the necessary code adjustments were made in PR 7890.*
ESLint has a number of advantageous properties, compared to JSHint. Among those are:
- The ability to find subtle bugs, thanks to more rules (e.g. PR 7881).
- Much more customizable in general, and many rules allow fine-tuned behaviour rather than the just the on/off rules in JSHint.
- Many more rules that can help developers avoid bugs, and a lot of rules that can be used to enforce a consistent coding style. The latter should be particularily useful for new contributors (and reduce the amount of stylistic review comments necessary).
- The ability to easily specify exactly what rules to use/not to use, as opposed to JSHint which has a default set. *Note:* in future JSHint version some of the rules we depend on will be removed, according to warnings in http://jshint.com/docs/options/, so we wouldn't be able to update without losing lint coverage.
- More easily disable one, or more, rules temporarily. In JSHint this requires using a numeric code, which isn't very user friendly, whereas in ESLint the rule name is simply used instead.
By default there's no rules enabled in ESLint, but there are some default rule sets available. However, to prevent linting failures if we update ESLint in the future, it seemed easier to just explicitly specify what rules we want.
Obviously this makes the ESLint config file somewhat bigger than the old JSHint config file, but given how rarely that one has been updated over the years I don't think that matters too much.
I've tried, to the best of my ability, to ensure that we enable the same rules for ESLint that we had for JSHint. Furthermore, I've also enabled a number of rules that seemed to make sense, both to catch possible errors *and* various style guide violations.
Despite the ESLint README claiming that it's slower that JSHint, https://github.com/eslint/eslint#how-does-eslint-performance-compare-to-jshint, locally this patch actually reduces the runtime for `gulp` lint (by approximately 20-25%).
A couple of stylistic rules that would have been nice to enable, but where our code currently differs to much to make it feasible:
- `comma-dangle`, controls trailing commas in Objects and Arrays (among others).
- `object-curly-spacing`, controls spacing inside of Objects.
- `spaced-comment`, used to enforce spaces after `//` and `/*. (This is made difficult by the fact that there's still some usage of the old preprocessor left.)
Rules that I indend to look into possibly enabling in follow-ups, if it seems to make sense: `no-else-return`, `no-lonely-if`, `brace-style` with the `allowSingleLine` parameter removed.
Useful links:
- http://eslint.org/docs/user-guide/configuring
- http://eslint.org/docs/rules/
I've not actually, thus far, come across a PDF file that this patch fixes. However, given the string of recent patches that has fixed issues with indirect objects in arrays, I think that it makes sense to proactively avoid any issues in this code.
In `Parser_filter` the `DecodeParms` data is fetched and passed to `Parser_makeFilter`, where we also make sure that a `Ref` is resolved to a direct object.
We can thus pass this along to the various image `Stream` constructors, to avoid the current situation where we lookup/resolve data that is already available.
Note also that we currently do *not* handle the case where `DecodeParms` is an Array entirely correct in the various image `Stream`s, and this patch fixes that for free.
For PDF files with multiple `/Filter`s, where the `/Length` entry is zero, we fail to render the file correctly. The reason is that `maybeLength` is `null` for the every filter except the first, and `!maybeLength` is thus truthy.
Hence it seems that we should completely ignore the `/Length` entry and also explicitly check `maybeLength === 0`.
Note that I've not (yet) come across a PDF file with this issue in the wild, but given all the stupid things PDF generators do I wouldn't be surprised if such a file actually exists. In order to prevent a possible future bug, I'm submitting this patch which includes a hand-edited PDF file that we currently cannot render correctly (but e.g. Adobe Reader can).
Instead of having `Parser_getObj` fail unconditionally for the referenced PDF file, this patch attempts to let searching for the main trailer continue even if there are errors.
Fixes https://bugzilla.mozilla.org/show_bug.cgi?id=1250079.
Currently the `isSpace` utility function is a member of `Lexer`, which seems suboptimal, given that it's placed in `core/parser.js`. In practice, this means that in a number of `core/*.js` files we thus have an *otherwise* completely unnecessary dependency on `core/parser.js` for a one-line function.
Instead, this patch moves `isSpace` into `shared/util.js` which seems more appropriate for this kind of utility function. Not to mention that since all the affected `core/*.js` files already depends on `shared/util.js`, this doesn't incur any more file dependencies.
*This is a regression from PR 3424.*
The PDF file in the referenced issue is using `Type3` fonts. In one of those, the `/CharProcs` dictionary contains an entry with the name `/#`. Before the changes to `Lexer_getName` in PR 3424, we were allowing certain invalid `Name` patterns containing the NUMBER SIGN (#).
It's unfortunate that this has been broken for close to two and a half years before the bug surfaced, but it should at least indicate that this is not a widespread issue.
Fixes 6692.
This code was added in PR 1214, but was made obsolete by PRs 1488/1493. Prior to the latter ones, `Dict_get` retured the raw objects. However, afterwards (and currently) `Dict_get` now resolves indirect objects, which makes `Parser_fetchIfRef` redundant.
*Potential risks with this patch:*
This patch passes all tests locally, but there's a *small* possibility that it could break some weird PDF files.
In the current code, wrapping `Dict_get` inside `Parser_fetchIfRef` will potentially mean two back-to-back call of `XRef_fetch`, if a reference points directly to another reference. I'm not sure if this can actually happen in practice, and I'd think that if that were the case we'd already have run into it elsewhere in the code-base, given that `Parser` is the only place where we try to "double" resolve references.
Having a warning here would have meant that issue 6360 could have been solved in approximately five minutes, instead of an hour. To avoid that happening again, this patch adds a warning whenever we treat a stream as empty.
The problem with the PDF files in the issue, besides the obviously broken XRef tables which we're able to recover from, is that many/most of the streams have Dictionaries where the `Length` entry is set to `0`. This causes us to return `NullStream`, instead of the appropriate one in `Parser_makeFilter`.
Fixes 6360.
Basic mathematics would suggest that a double negative should always become positive, but it appears that Adobe Reader simply ignores that case. Hence I think that it makes sense for us to do the same.
Fixes 6218.