Note first of all how the `PDFDocumentProxy.getJSActions` method in the API caches the result, which makes repeated lookups cheap enough to not really be an issue.
Secondly, with the previous patch, we're now only dispatching "pageopen"/"pageclose"-events when there's actually a sandbox that listens for them.
All-in-all, with these changes we can thus simplify the default-viewer "pageopen"-event handler a fair bit.
The "pageopen"/"pageclose"-events are only necessary if, and only if, there's actually a sandbox to dispatch the events in. Hence we shouldn't dispatch those events unconditionally, as soon as `enableScripting` is set, but rather initialize that functionality only when needed.
Furthermore, in `web/app.js`, there's currently a bug since we're attempting to *manually* simulate a "pageopen"-event for a page that may not actually have been rendered at the time. With the modified `BaseViewer.initializeScriptingEvents` method, we'll now dispatch a correct "pageopen"-event here.
There's built-in ESLint rule, see `sort-imports`, to ensure that all `import`-statements are sorted alphabetically, since that often helps with readability.
Unfortunately there's no corresponding rule to sort `export`-statements alphabetically, however there's an ESLint plugin which does this; please see https://www.npmjs.com/package/eslint-plugin-sort-exports
The only downside here is that it's not automatically fixable, but the re-ordering is a one-time "cost" and the plugin will help maintain a *consistent* ordering of `export`-statements in the future.
*Note:* To reduce the possibility of introducing any errors here, the re-ordering was done by simply selecting the relevant lines and then using the built-in sort-functionality of my editor.
This implementation is inspired by the behaviour in (recent versions of) Adobe Reader, since it leads to reasonably simple and straightforward code as far as I'm concerned.
*Specifically:* We'll only consider *one* destination per page when finding/highlighting the current outline item, which is similar to e.g. Adobe Reader, and we choose the *first* outline item at the *lowest* level of the outline tree.
Given that this functionality requires not only parsing of the `outline`, but looking up *all* of the destinations in the document, this feature can when initialized have a non-trivial performance overhead for larger PDF documents.
In an attempt to reduce the performance impact, the following steps are taken here:
- The "find current outline item"-functionality will only be enabled once *one* page has rendered and *all* the pages have been loaded[1], to prevent it interfering with data regular fetching/parsing early on during document loading and viewer initialization.
- With the exception of a couple of small and simple `eventBus`-listeners, in `PDFOutlineViewer`, this new functionality is initialized *lazily* the first time that the user clicks on the `currentOutlineItem`-button.
- The entire "find current outline item"-functionality is disabled when `disableAutoFetch = true` is set, since it can easily lead to the setting becoming essentially pointless[2] by triggering *a lot* of data fetching from a relatively minor viewer-feature.
- Fetch the destinations *individually*, since that's generally more efficient than using `PDFDocumentProxy.getDestinations` to fetch them all at once. Despite making the overall parsing code *more* asynchronous, and leading to a lot more main/worker-thread message passing, in practice this seems faster for larger documents.
Finally, we'll now always highlight an outline item that the user manually clicked on, since only highlighting when the new "find current outline item"-functionality is used seemed inconsistent.
---
[1] Keep in mind that the `outline` itself already isn't fetched/parsed until at least *one* page has been rendered in the viewer.
[2] And also quite slow, since it can take a fair amount of time to fetch all of the necessary `destinations` data when `disableAutoFetch = true` is set.
Ensure that `PDFViewerApplication._contentLength` is always updated with the *correct* length, as returned by `PDFDocumentProxy.getDownloadInfo`, and only let the `PDFViewerApplication._initializeMetadata` method overwrite if it's not already been set.
Finally, in `PDFViewerApplication._initializeJavaScript`, the fallback `_contentLength` handling is now moved to just after the fallback `documentInfo` handling, such that all the fallback code is in one place within the method.
Note how the `onerror` functionality is not being used in the GENERIC `DownloadManager`, since we have no way of knowing if downloading succeeded.
Hence this functionality is only *possibly* useful in MOZCENTRAL builds, however as outlined in the existing comments it's unlikely to be helpful in practice. Generally speaking, if downloading failed once in [`PdfStreamConverter.jsm`](https://searchfox.org/mozilla-central/rev/809ac3660845fef6faf18ec210232fdadc0f1ad9/toolkit/components/pdfjs/content/PdfStreamConverter.jsm#294-406) it seems very likely that it would fail again; all-in-all I'm thus suggesting that we just remove the `onerror` functionality altogether here.
Currently this code is duplicated no less than three times in the `web/app.js` file, and by introducing a helper method we can avoid unnecessary repetition.
This method currently accepts a callback-function, which does feel a bit old fashioned now. At the time that this code was introduced, native Promises didn't exist yet and there's a custom Promise-implementation used instead.
However, today with Promises and async/await being used *a lot* it seems reasonable to change `DefaultExternalServices.fallback` to an `async` method instead such that the callback-function can be removed.
This follows the same principle as the `once` option that exists in the native `addEventListener` method, and will thus automatically remove an `EventBus` listener when it's invoked; see https://developer.mozilla.org/en-US/docs/Web/API/EventTarget/addEventListener#Parameters
Finally, this patch also tweaks some the existing `EventBus`-code to use modern features such as optional chaining and logical assignment operators.
Given that we already have a `PresentationModeState`-enumeration, we should use that with the "presentationmodechanged" event rather than including separate properties. Note that this new behaviour, of including an enumeration-value in the event, is consistent with lots of other existing viewer-events.
To hopefully avoid issues in custom implementations of the default viewer, any attempt to access the removed properties will now throw.
Similar to e.g. the "locale" option, this in *only* done for those build-targets where the "sandboxBundleSrc" is actually defined.
With these changes we can remove an `AppOptions` dependency from the `web/generic_scripting.js` file, thus limiting *direct* `AppOptions` usage in the default viewer files.
Given that the `dispatchEventInSandbox` method (on the scripting-classes) is asynchronous, there's a very real risk that the events won't be dispatched/handled until *after* their associated functionality has actually run (with the "Will..." events being particularily susceptible to this issue).
To reduce the likelihood of that happening, we can simply `await` the `dispatchEventInSandbox` calls as necessary. A couple of methods are now marked as `async` to support these changes, however that shouldn't be a problem as far as I can tell.
*Please note:* Given that the browser "beforeprint"/"afterprint" events are *synchronous*, we unfortunately cannot await the `WillPrint`/`DidPrint` event dispatching. To fix this properly the web-platform would need support for asynchronous printing, and we'll thus have to hope that things work correctly anyway.
Note that currently the `DidSave` event is not *guaranteed* to actually be dispatched if there's any errors during saving, which is easily fixed by simply moving it to occur in the `finally`-handler in `PDFViewerApplication.save` method.
For the `WillPrint`/`DidPrint` events, things are unfortunately more complicated. Currently these events will *only* be dispatched iff the printing request comes from within the viewer itself (e.g. by the user clicking on the "Print" toolbar button), however printing can be triggered in a few additional ways:
- In the GENERIC viewer:
- By the <kbd>Ctrl</kbd>+<kbd>P</kbd> keyboard shortcut.
- In the MOZCENTRAL viewer, i.e. the Firefox built-in viewer:
- By the <kbd>Ctrl</kbd>+<kbd>P</kbd> keyboard shortcut.
- By the "Print" item, as found in either the Firefox "Hamburger menu" or in the browser-window menu.
In either of the cases described above, no `WillPrint`/`DidPrint` events will be dispatched. In order to *guarantee* that things work in the general case, we thus have to move the `dispatchEventInSandbox` calls to the "beforeprint"/"afterprint" event handlers instead.
Rather than calling `getJavaScript` in the API and then ignoring the result, when "enableScripting" is set, it should be more efficient/faster to simply skip it altogether instead.
Finally, the `setTimeout` call at the end of `PDFViewerApplication._initializeAutoPrint` is removed, since it doesn't seem necessary any more as far as I can tell.[1]
Note that when this functionality was originally added, back in PR 2839, it seems that `pagesPromise` simply waited for the `getPage` calls of *all* pages to resolve. Today, on the other hand, the viewer fetches *and* renders the first page *before* doing the remaining `getPage` calls, and only afterwards is `pagesPromise` resolved. Hence it's not really clear why we now need to delay printing even further with a `setTimeout` call.
---
[1] The patch was tested with the following documents: https://github.com/mozilla/pdf.js/blob/master/test/pdfs/bug1001080.pdf and https://github.com/mozilla/pdf.js/blob/master/test/pdfs/issue6106.pdf
These callbacks should not be necessary *before* the document has been initialized. Furthermore, move the functionality to a new helper-method since `PDFViewerApplication.load` is already quite large.
Given that this relies on accessing properties on the `PDFDocumentProxy`-instance, it seems more appropriate for this code to live in `PDFViewerApplication`.
It seems that the timeout is way too short in practice, since this new integration-test failed *intermittently* already in PR 12702 (which is where the test was added).
The ideal solution here would be to simply await an event, dispatched by the viewer, however that unfortunately doesn't appear to be supported by Puppeteer.
Instead, the solution implemented here is to add a new method in `PDFViewerApplication` which Puppeteer can query to check if the scripting/sandbox has been fully initialized.
There's really no point, as far as I can tell, to attempt to dispatch an event in a non-existent sandbox. Generally speaking, even trying to do this *could* possibly even lead to errors in some cases.
Furthermore, utilize optional chaining to simplify some `dispatchEventInSandbox` calls throughout the viewer.
Finally, replace superfluous `return` statements with `break` in the switch-statement in the `updateFromSandbox` event-handler.
For DOM events all event names are lower-case, and the newly added PDF.js scripting-events thus "stick out" quite a bit. Even more so, considering that our internal `eventBus`-events follow the same naming convention.
Hence this patch, which changes the "updateFromSandbox"/"dispatchEventInSandbox" events to be lower-case instead.
Furthermore, using DOM events for communication *within* the PDF.js code itself (i.e. between code in `web/app.js` and `src/display/annotation_layer.js/`) feels *really* out of place.
That's exactly the reason that we have the `EventBus` abstraction, since it allowed us to remove prior use of DOM events, and this patch thus re-factors the code to make use of the `EventBus` instead for scripting-related events.
Obviously for events targeting a *specific element* using DOM events is still fine, but the "updatefromsandbox"/"dispatcheventinsandbox" ones should be using the `EventBus` internally.
*Drive-by change:* Use the `BaseViewer.currentScaleValue` setter unconditionally in `PDFViewerApplication._initializeJavaScript`, since it accepts either a string or a number.
- Actually remove the `isDown` property when destroying the scripting-instance.
- Mark all `mouseState` usage as "private" in the various classes.
- Ensure that the `AnnotationLayer` actually treats the parameter as properly *optional*, the same way that the viewer components do.
- For now remove the `mouseState` parameter from the `PDFPageView` class, and keep it only on the `BaseViewer`, since it's questionable if all of the scripting-functionality will work all that well without e.g. a full `BaseViewer`.
- Append the `mouseState` to the JSDoc for the `AnnotationElement` class, and just move its definition into the base-`AnnotationElement` class.
* the goal is to execute actions like Open or OpenAction
* can be tested with issue6106.pdf (auto-print)
* once #12701 is merged, we can add page actions
* move set/clear|Timeout/Interval and crackURL code in pdf.js
* remove the "backdoor" in the proxy (used to dispatch event) and so return the dispatch function in the initializer
* remove listeners if an error occured during sandbox initialization
* add support for alert and prompt in the sandbox
* add a function to eval in the global scope
Given that the GENERIC default viewer supports opening more than one document, and that a unique scripting-instance is now used for each document, the changes made in this patch seem appropriate.
Since the `close` method has become quite large, this small re-factoring shouldn't hurt (and may also be useful with future changes to the `_initializeJavaScript` method).
I was actually quite surprised to find that, despite the various `scripting`-getters implementing `destroySandbox` methods, there were no attempts at actually cleaning-up either the "sandbox" or removing the globally registered event listeners.
This patch also changes the method to skip *all* data fetching when "enableScripting" isn't active. Finally, simplifies some event-data accesses in the "updateFromSandbox" listener.
Another possible option here could be to use the `contentLength`, when it exists, and then using e.g. a custom event to always update the "filesize" in the sandbox "after the fact" with the result of the `getDownloadInfo`-call.
We can easily avoid unnecessary API-calls here, since most of the time the `metadata` will already be available here. In the *rare* case that it's not available, we can simply wait for the existing `getMetadata`-call to resolve.
This will be useful in the following patch, and note that there's also an old issue (see 5765) which asked for such an event. However, given that the use-case wasn't *clearly* specified, and that we didn't have an internal use for it at the time it wasn't implemented.
Also, ensure that all of the metadata-related properties are actually reset when the document is closed.
Given that we already include the "Content-Disposition"-header filename, when it exists, it shouldn't hurt to also include the information from the "Content-Length"-header.
For PDF documents opened via a URL, which should be a very common way for the PDF.js library to be used, this will[1] thus provide a way of getting the PDF filesize without having to wait for the `getDownloadInfo`-promise to resolve[2].
With these API improvements, we can also simplify the filesize handling in the `PDFDocumentProperties` class.
---
[1] Assuming that the server is correctly configured, of course.
[2] Since that's not *guaranteed* to happen in general, with e.g. `disableAutoFetch = true` set.
Given that it's generally faster to call *one* function and have it loop through an object, rather than looping through an object and calling a function for every iteration, this patch will reduce the total time spent in `PDFViewerApplication._readPreferences` ever so slightly.
Also, over time we've been adding more and more preferences, rather than removing them, so using the new `AppOptions.setAll` method should be generally beneficial as well.
While the effect of these changes is quite small, it does reduces the time it takes for the preferences to be fully initialized. Given the amount of asynchronous code during viewer initialization, every bit of time that we can save should thus help.
Especially considering the recently added `viewerCssTheme` preference, which needs to be read very early to reduce the risk of the viewer UI "flashing" visibly as the theme changes, I figured that a couple of small patches reducing the time spend reading preferences cannot hurt.
Given that only two debugging hash parameters (i.e. `disableWorker` and `pdfBug`) will make this method asynchronous, we can avoid what's most of the time is an unnecessary `Promise.all` invocation.
While this does work pretty well in my quick testing, it's *very much* a hack since as far as I can tell there's no support in the CSS specification for using e.g. a CSS variable to override a `@media (prefers-color-scheme: dark) {...}` block.
The solution implemented here is thus to *edit* the viewer CSS, by either removing the entire `@media ...` block in light-mode or by ensuring that its rules become *unconditionally* applied in dark-mode.
To simplify the overall implementation, since all of this does seem like somewhat of an edge-case, the `viewerCssTheme` preference will *only* be read during viewer initialization. (Similar to many other existing preferences, a reload is thus required when changing it.)
* When no actions then set it to null instead of empty object
* Even if a field has no actions, it needs to listen to events from the sandbox in order to be updated if an action changes something in it.
Note that a number of these cases are covered by existing unit-tests, and a few others only matter for the development/build scripts.
Furthermore, I've also tried to the best of my ability to test each case *manually* to hopefully further reduce the likelihood of this patch introducing any bugs.
Please find additional details about the ESLint rule at https://eslint.org/docs/rules/no-useless-escape
