For *very* long/large documents fetching all pages on load may cause quite bad performance, both memory and CPU wise. In order to at least slightly alleviate this, we can let the viewer treat these kind of documents[1] as if `disableAutoFetch` were set.
---
[1] One example of a really bad case is https://bugzilla.mozilla.org/show_bug.cgi?id=1588435, which this patch should at least help somewhat. In general, for these cases, we'd probably need to implement switching between `PDFViewer`/`PDFSinglePageViewer` (as already tracked on GitHub) and use the latter for these kind of long documents.
Unfortunately I forgot to test `gulp mozcentraldiff` with PR 11256, which is really bad since it will cause the en-US `chrome.properties` l10n file to be deleted at mozilla-central; sorry about breaking this!
In order to address this we'll have to re-add (only) the en-US `chrome.properties` l10n file, which is simple enough. Furthermore, since we're not doing any sort of build-specific parsing of the l10n files, we can just copy the en-US files as-is rather than having to run `gulp locale` during `gulp mozcentral`.
As can be seen in the API, there's a number of document loading Exception handlers which are both really simple and highly similar. Hence these are changed such that all the relevant Exceptions are sent via *one* message instead.
Furthermore, the patch also avoids unnecessarily re-creating `UnknownErrorException`s at the worker side and removes an unnecessary `bind` call.
This was done automatically, using the `npm audit fix` command.
*Please note:* There's two vulnerabilities left, however they couldn't be fixed automatically and may thus benefit from being handled separately.
Obviously this won't look exactly right, but considering that the PDF file doesn't bother embedding non-standard fonts this is the best that we can do here.
Originally only `skipPages` existed, but given that `firstPage`/`lastPage` has existed for a long time now using them whenever possible looks simpler overall.
- In the `ibwa-bad` case the sixteenth page contains corrupt/incomplete commands, but given that we're suppressing `Error`s by default now skipping hardly seems warranted any more.
- In the `geothermal.pdf` case the first page contains an unsupported ColourSpace, but again we're suppressing `Error`s by default now and skipping hardly seems warranted any more.
With the removal of the (standalone) Firefox building code in PR 9566 (a year and a half ago), these files are now completely unused in the GitHub repository[1].
Hence it doesn't really seem necessary to keep fetching them with `gulp importl10n`, and the existing files in the `l10n` folder can also be removed (thanks to version control, they're easy enough to restore should the need ever arise).
The patch also allows an additional simplification, for the `gulp locale` and `gulp mozcentral` commands, since it's now possible to stop writing `l10n` files to the `extensions/firefox/` folder and instead just copy them similar to other build targets.
---
[1] They're obviously still used in `mozilla-central`, for fallback messages displayed through `PdfStreamConverter.jsm`, but that doesn't make it necessary to keep them *here* as far as I'm concerned.
With the removal of the (standalone) Firefox building code in PR 9566 (a year and a half ago), these files are now completely unused.
Hence it doesn't really make sense to keep building them as part of `gulp locale`, and the existing files in the `l10n` folder can also be removed (thanks to version control, they're easy enough to restore should the need ever arise).
The issue has been open for years now, and has even been marked with `5-good-beginner-bug` for *months*, without any movement.
Considering just how simple the suggested solution is, I'm submitting this patch just to close out a long-standing issue.
The following is printed with every build, which gets kind of annoying when looking at the logs:
```
Replace Autoprefixer browsers option to Browserslist config.
Use browserslist key in package.json or .browserslistrc file.
Using browsers option cause some error. Browserslist config
can be used for Babel, Autoprefixer, postcss-normalize and other tools.
If you really need to use option, rename it to overrideBrowserslist.
Learn more at:
https://github.com/browserslist/browserslist#readmehttps://twitter.com/browserslist
```
Given how we're using Autoprefixer, with a number of build-specific configs, simply changing the option name seems like the easiest solution here.
(I'm also adding a couple of newlines at the `autoprefixer` call-sites, to aid readability.)
Obviously userAgent checks aren't that great, since it's very easy to spoof, but it probably doesn't hurt to attempt to extend this (since it's limited to `GENERIC` builds).
Besides, anyone using the default viewer can always set the `maxCanvasPixels` option to a value of their choosing.
This patch is making me somewhat worried about future regressions, since it's certainly easy to imagine this completely breaking certain kinds of corrupt/edited PDF documents while fixing others.[1]
Obviously it passes all existing reference tests (and even improves one), however compared to many other patches there's no telling how much it could break.
The only reason that I'm even submitting this patch, is because of the number of open issues that it would address.
Generally speaking though, the best course of action would probably be if `XRef.indexObjects` was re-written to be much more robust (since it currently feels somewhat hand-wavy in parts). E.g. by actually checking/validating more of the objects before committing to them.
---
[1] Especially given that it's reverting part of PR 5910, however in the case of issue 5909 it seems that other (more recent) changes have actually made that PR redundant.
Sometimes we also used `@return`, but `@returns` is what the JSDoc
documentation recommends. Even though `@return` works as an alias, it's
good to use the recommended syntax and to be consistent within the
project.
Sometimes we also used `@return` or `@returns`, but `@type` is what
the JSDoc documentation recommends. This also improves the documentation
because before this commit the types were not shown and now they are.
Rather than specifying certain build targets manually, it seems much more appropriate (and future-proof) to use the `SKIP_BABEL` build target instead.
Also, the patch adds a missing `/* eslint no-var: error */` line since I'm touch the file anyway and no code-changes were necessary for it.
As part of attempting to fix a number issues containing PDF documents with corrupt XRef tables, I'd like to improve the reference test-coverage slightly *first*.
Obviously this will increase the runtime of the tests a bit, however I'd rather "waste" resources on the bots instead of developer time fixing regressions which could have been avoided.
For badly generated PDF documents, with issue 6961 being one example, there's well over one hundred thousand function calls being made in total for just the *two* pages.
This handles the two different ways that fonts can be loaded, either by Name (which is the common case) or by Reference.
Furthermore, this also takes the `ignoreErrors` option into account when deciding whether to fallback or Error.
Finally, by creating a minimal but valid Font dictionary, there's no special-cases necessary in any of the font parsing code.
Co-authored-by: huzjakd <huzjakd@gmail.com>
Co-Authored-By: Jonas Jenwald <jonas.jenwald@gmail.com>
