This avoids unnecessary duplication of many images, thus reducing the size of PDF.js image resources slightly.
Note that since the images should only be flipped horizontally, this required specifying the horizontal/vertical scaling separately for the hiDPI-images.
This functionality was only added to the default viewer for backwards compatibility and to support the various PDF viewer tests in mozilla-central, with the intention to eventually remove it completely.
While the different mozilla-central tests cannot be *easily* converted from DOM events, it's however possible to limit that functionality to only MOZCENTRAL builds *and* when tests are running.
Rather than depending of the re-dispatching of internal events to the DOM, the default viewer can instead be used in e.g. the following way:
```javascript
document.addEventListener("webviewerloaded", function() {
PDFViewerApplication.initializedPromise.then(function() {
// The viewer has now been initialized, and its properties can be accessed.
PDFViewerApplication.eventBus.on("pagerendered", function(event) {
console.log("Has rendered page number: " + event.pageNumber);
});
});
});
```
The correct/intended way of working with the "viewer components" is by providing an `EventBus` instance upon initialization, and the `getGlobalEventBus` was only added for backwards compatibility.
Note, for example, that using `getGlobalEventBus` doesn't really work at all well with a use-case where there's *multiple* `PDFViewer` instances on a one page, since it may then be difficult/impossible to tell which viewer a particular event originated from.
All of the "viewer components" examples have been previously updated, such that there's no longer any code/examples which relies on the now removed `getGlobalEventBus` functionality.
Given the way that "classes" were previously implemented in PDF.js, using regular functions and closures, there's a fair number of false positives when the `no-shadow` ESLint rule was enabled.
Note that while *some* of these `eslint-disable` statements can be removed if/when the relevant code is converted to proper `class`es, we'll probably never be able to get rid of all of them given our naming/coding conventions (however I don't really see this being a problem).
Given that none of the relevant options are marked as optional in the code/JSDocs, and that the `PDFDocumentProperties` class is specific to the default viewer (and not exposed as part of the viewer components), there's no good reason as far as I can tell for these checks.
The `BaseViewer.setDocument` method in particular is necessary for rendering to start when the viewer loads, hence you obviously want that to happen as soon as possible and without any unnecessary delays.
Unfortunately *some* API calls need to be done before that, note existing comments, however the `ViewHistory` initialization (and subsequent fetching of data) in particular can be moved slightly without any adverse effects.
As part of testing I've used logging with `performance.now()` inserted in various parts of this code, and there's *obviously* no discernible changes between `master` and this patch for e.g. rendering starting in the viewer.
*Note:* The vast majority of this patch is simple indentation changes, which were forced by Prettier (and done automatically with `gulp lint --fix`).
- Ensure that `database.files` actually contains an Array, rather than some arbitrary data.
- Only try to lookup an existing entry when the `database` existed on load, since there's obviously nothing to find when `database.files = []` was set (this case is very common in the MOZCENTRAL build since `sessionStorage` is being used there).
This property has never been documented and/or *intentionally* exposed through the API, instead the `PDFPageProxy.pageNumber` property is the documented/intended API to use here.
Hence pageIndex is changed to a "private" property on `PDFPageProxy` instances, and internal API functionality is also updated to *consistently* use `this._pageIndex` rather than a mix of formats.
The viewer doesn't currently support executing of any JavaScript, as found in some PDF documents, for security reasons. (Although there's a small hack to at least provide basic support for automatic printing on document load, without running scripts.)
However, in the event that the browser doesn't support printing we're not run *any* of this code. In particular that means that we're also not displaying the "Warning: JavaScript is not supported" message, which seems strange since JavaScript found in a PDF document can really contain *anything* (and not only printing instructions).
It thus seem reasonable, as far as I'm concerned, to always display the JavaScript-warning *even* if printing happens to be unsupported.
*This is part of a series of patches that will try to split PR 11566 into smaller chunks, to make reviewing more feasible.*
Once all the code has been fixed, we'll be able to eventually enable the ESLint `no-shadow` rule; see https://eslint.org/docs/rules/no-shadow
While Firefox originally used transition effects for browser UI toolbar buttons, that was removed years ago in https://bugzilla.mozilla.org/show_bug.cgi?id=1393057
Since the PDF.js viewer toolbar transitions were likely based on the Firefox ones, it seems reasonable that these transition effects are removed from PDF.js as well. Besides removing a bunch of CSS, this also makes the toolbar feel ever so slightly more "snappy" without these delays on mouse interaction.
(In order to make it more feasible to modernize/improve the viewer UI, trying to clean-up/simplify existing rules iteratively seems like the most reasonble way to make any progress here w.r.t. being able to test/review things.)
For reasons that I don't even pretend to understand, the `textLayerFactory` property is determined for *every single* page in the PDF document.
Given that the `TextLayerMode` should be consistent for *all* pages in a document, we obviously could/should define `textLayerFactory` just once instead.
There's a couple of issues with this functionality:
- The respective `PromiseCapability` instances are not being reset, in `BaseViewer._resetView`, when the document is closed which is inconsistent with all other state.
- While the default viewer depends on these promises, and they thus ought to be considered part of e.g. the `PDFViewer` API-surface, they're not really defined in a particularily user-visible way (being that they're attached to the `BaseViewer` instance *inline* in `BaseViewer.setDocument`).
- There's some internal `BaseViewer` state, e.g. `BaseViewer._pageViewsReady`, which is tracked manually and could instead be tracked indirectly via the relevant `PromiseCapability`, thus reducing the need to track state *twice* since that's always best to avoid.
*Please note:* In the existing implementation, these promises are not defined *until* the `BaseViewer.setDocument` method has been called.
While it would've been simple to lift that restriction in this patch, I'm purposely choosing *not* to do so since this ensures that any Promise handlers added inside of `BaseViewer.setDocument` are always invoked *before* any external ones (and keeping that behaviour seems generally reasonable).
This patch deprecates the existing `getOpenActionDestination` API method, in favor of a better and more general `getOpenAction` method instead. (For now JavaScript actions, related to printing, are still handled as before.)
By clearly separating "regular" Print actions from the JavaScript handling, it's thus possible to get rid of the somewhat annoying and strictly incorrect warning when the viewer loads.
It occured to me that similar to the `getGlobalEventBus` function, it's probably a good idea to *also* notify users of the fact that `eventBusDispatchToDOM` is now deprecated.
Rather than depending of the re-dispatching of internal events to the DOM, the default viewer can instead be used in e.g. the following way:
```javascript
document.addEventListener("webviewerloaded", function() {
PDFViewerApplication.initializedPromise.then(function() {
// The viewer has now been initialized, and its properties can be accessed.
PDFViewerApplication.eventBus.on("pagerendered", function(event) {
console.log("Has rendered page number: " + event.pageNumber);
});
});
});
```
I overlooked this in PR 11631; sorry about that!
Also, ensure that `EventBus` instances *always* track "external" events using a boolean regardless of the actual option value.
This changes the dropdown icon from being set using the `background` CSS property, to being set with `::after` which is *similar* to all the other toolbar button icons (which use `::before`).
Also tweaks the dropdown `background-color` on `:hover` slightly, since the other changes made it too light.
Since the goal has always been, essentially since the `EventBus` abstraction was added, to remove all dispatching of DOM events[1] from the viewer components this patch tries to address one thing that came up when updating the examples:
The DOM events are always dispatched last, and it's thus guaranteed that all internal event listeners have been invoked first.
However, there's no such guarantees with the general `EventBus` functionality and the order in which event listeners are invoked is *not* specified. With the promotion of the `EventBus` in the examples, over DOM events, it seems like a good idea to at least *try* to keep this ordering invariant[2] intact.
Obviously this won't prevent anyone from manually calling the new *internal* viewer component methods on the `EventBus`, but hopefully that won't be too common since any existing third-party code would obviously use the `on`/`off` methods and that all of the examples shows the *correct* usage (which should be similarily documented on the "Third party viewer usage" Wiki-page).
---
[1] Looking at the various Firefox-tests, I'm not sure that it'll be possible to (easily) re-write all of them to not rely on DOM events (since getting access to `PDFViewerApplication` might be generally difficult/messy depending on scopes).
In any case, even if technically feasible, it would most likely add *a lot* of complication that may not be desireable in the various Firefox-tests. All-in-all, I'd be fine with keeping the DOM events only for the `MOZCENTRAL` target and gated on `Cu.isInAutomation` (or similar) rather than a preference.
[2] I wouldn't expect any *real* bugs in a custom implementation, simply based on event ordering, but it nonetheless seem like a good idea if any "external" events are still handled last.
To avoid outright breaking third-party usages of the "viewer components" the `getGlobalEventBus` functionality is left intact, but a deprecation message is printed if the function is invoked.
The various examples are updated to *explicitly* initialize an `EventBus` instance, and provide that when initializing the relevant viewer components.
This complements the existing `PDFViewerApplication.initialized` boolean property, and may be helpful for custom implementations of the default viewer. This will thus provide users of the default viewer an alternative to setting the preference to dispatch events to the DOM (and listen for the "localized" event), since they can instead use:
```javascript
document.addEventListener("webviewerloaded", function() {
PDFViewerApplication.initializedPromise.then(function() {
// The viewer has now been initialized.
})
});
```
Note that in order to avoid manually tracking the initialization state *twice*, this implementation purposely uses the `PromiseCapability` functionality to handle both `PDFViewerApplication.initialized` and `PDFViewerApplication.initializedPromise` internally.
This patch contains some *much* needed clean-up of, and improvements to, this old code thus addressing a number of issues:
- Set more reasonable *default* values for the widths, in `web/viewer.css`, since the current ones are actually too small even for the (default) `en-US` locale.
This obviously result in a slightly larger zoom dropdown width for many locales, but the more consistent UI does look good to me.
- Stop setting the `min-width`/`max-width` and just use `width` instead.
- Set an explicit `height` of the zoom dropdown, in an attempt to get Google Chrome to display it with the same height as the toolbar buttons.
- Remove additional `Element.setAttribute("style", ...)` usage.
- Actually check *all* of the predefined l10n strings, since the old implementation (implicitly) assumed that the currently selected one was the longest (note e.g. the `ja-JP` locale where one string is considerably longer than the rest).
- Stop invalidating the DOM multiple times when doing the measurements. This was achieved by using a temporary in-memory `canvas`, and we now only need to query the DOM once in order to get the current font properties of the zoom dropdown.
This should hopefully be useful in environments where restrictive CSPs are in effect.
In most cases the replacement is entirely straighforward, and there's only a couple of special cases:
- For the `src/display/font_loader.js` and `web/pdf_outline_viewer.js `cases, since the elements aren't appended to the document yet, it shouldn't matter if the style properties are set one-by-one rather than all at once.
- For the `web/debugger.js` case, there's really no need to set the `padding` inline at all and the definition was simply moved to `web/viewer.css` instead.
*Please note:* There's still *a single* case left, in `web/toolbar.js` for setting the width of the zoom dropdown, which is left intact for now.
The reasons are that this particular case shouldn't matter for users of the general PDF.js library, and that it'd make a lot more sense to just try and re-factor that very old code anyway (thus fixing the `setAttribute` usage in the process).
Having thought *briefly* about using `css-vars-ponyfill`, I'm no longer convinced that it'd be a good idea. The reason is that if we actually want to properly support CSS variables, then that functionality should be available in *all* of our CSS files.
Note in particular the `pdf_viewer.css` file that's built as part of the `COMPONENTS` target, in which case I really cannot see how a rewrite-at-the-client solution would ever be guaranteed to always work correctly and without accidentally touching other CSS in the surrounding application.
All-in-all, simply re-writing the CSS variables at build-time seems much easier and is thus the approach taken in this patch; courtesy of https://github.com/MadLittleMods/postcss-css-variables
By using its `preserve` option, the built files will thus include *both* a fallback and a modern `var(...)` format[1]. As a proof-of-concept this patch removes a couple of manually added fallback values, and converts an additional sidebar related property to use a CSS variable.
---
[1] Comparing the `master` branch with this patch, when using `gulp generic`, produces the following diff for the built `web/viewer.css` file:
```diff
@@ -408,6 +408,7 @@
:root {
--sidebar-width: 200px;
+ --sidebar-transition-duration: 200ms;
}
* {
@@ -550,27 +551,28 @@
position: absolute;
top: 32px;
bottom: 0;
- width: 200px; /* Here, and elsewhere below, keep the constant value for compatibility
- with older browsers that lack support for CSS variables. */
+ width: 200px;
width: var(--sidebar-width);
visibility: hidden;
z-index: 100;
border-top: 1px solid rgba(51, 51, 51, 1);
-webkit-transition-duration: 200ms;
transition-duration: 200ms;
+ -webkit-transition-duration: var(--sidebar-transition-duration);
+ transition-duration: var(--sidebar-transition-duration);
-webkit-transition-timing-function: ease;
transition-timing-function: ease;
}
html[dir='ltr'] #sidebarContainer {
-webkit-transition-property: left;
transition-property: left;
- left: -200px;
+ left: calc(-1 * 200px);
left: calc(-1 * var(--sidebar-width));
}
html[dir='rtl'] #sidebarContainer {
-webkit-transition-property: right;
transition-property: right;
- right: -200px;
+ right: calc(-1 * 200px);
right: calc(-1 * var(--sidebar-width));
}
@@ -640,6 +642,8 @@
#viewerContainer:not(.pdfPresentationMode) {
-webkit-transition-duration: 200ms;
transition-duration: 200ms;
+ -webkit-transition-duration: var(--sidebar-transition-duration);
+ transition-duration: var(--sidebar-transition-duration);
-webkit-transition-timing-function: ease;
transition-timing-function: ease;
}
```
Since this rule is now enforced in the entire `/web` folder, there's no good reason for it to remain disabled in this file. Most likely, it's added to reduce the number of linting errors when we started enforcing block-scoped variables.
There's no particular reason for using the PDF.js helper function `createObjectURL`[1] in the debugger, instead of the native `URL.createObjectURL` directly, for a couple of reasons:
- The relevant code-path only applies to fonts loaded with the Font Loading API, and this isn't supported in IE anyway.
- The debugger can, since quite some time, not even be loaded in IE any more.
- General support for IE is now limited, and there's no guaratee that everything actually works.
---
[1] It provides a fallback for browsers with broken `Blob` support, which as usual means Internet Explorer :-P
Please find additional details about the ESLint rule at https://eslint.org/docs/rules/prefer-const
With the recent introduction of Prettier this sort of mass enabling of ESLint rules becomes a lot easier, since the code will be automatically reformatted as necessary to account for e.g. changed line lengths.
Note that this patch is generated automatically, by using the ESLint `--fix` argument, and will thus require some additional clean-up (which is done separately).
The debugging hash parameters[1] are intended to facilitate access to various tools/settings in PRODUCTION builds, protected by the `pdfBugEnabled` preference.
At this point, the remaining debugging hash parameters are mainly intended to allow access to the `PDFBug` tools and/or to quickly toggle certain larger features.
The "useOnlyCssZoom" functionality doesn't really seem to fit in with the rest of these hash parameters, since:
- This is, comparatively speaking, a minor viewer-specific feature.
- The zooming implementation will (almost) always fallback to CSS-only zooming, for any document, once the canvases becomes large enough. Hence, the majority of the CSS zooming feature can still be tested *directly* in any build of the viewer.
- After the initial implementation, years ago, the CSS-only zooming code in question hasn't changed much (or even at all), i.e. it doesn't seem like an active development target.[2]
- If the "useOnlyCssZoom" functionality was added today, it's unlikely that a hash parameter would've been added.
- Last, but not least, there's also a `useOnlyCssZoom` preference hence toggling this functionality shouldn't be too difficult (e.g. if someone needs to hack on it).
All in all, I'm thus suggesting that we remove the "useOnlyCssZoom" hash parameter.
---
[1] Originally these hash parameters could be used directly in any build, which was bad since it would allow any link to potentially disable functionality and/or reduce performance.
[2] If it had seen active development over the years, I'd be *much* more inclined to keep the hash parameter.
For people running e.g. Firefox with the `pdfBugEnabled` preference set, to allow quick access to debugging tools, this method will obviously run for every opened PDF file. However, in most cases the URL hash is empty and we can thus skip most of the parsing and simply return early instead.
This removes a couple of, thanks to preceeding code, unnecessary `typeof PDFJSDev` checks, and also fixes a couple of incorrectly implemented (my fault) checks intended for `TESTING` builds.
After PR 9566, which removed all of the old Firefox extension code, the `FIREFOX` build flag is no longer used for anything.
It thus seems to me that it should be removed, for a couple of reasons:
- It's simply dead code now, which only serves to add confusion when looking at the `PDFJSDev` calls.
- It used to be that `MOZCENTRAL` and `FIREFOX` was *almost* always used together. However, ever since PR 9566 there's obviously been no effort put into keeping the `FIREFOX` build flags up to date.
- In the event that a new, Webextension based, Firefox addon is created in the future you'd still need to audit all `MOZCENTRAL` (and possibly `CHROME`) build flags to see what'd make sense for the addon.
While only the `MOZCENTRAL` builds will actually do anything meaningful with the telemetry data, none of the code in question actually runs *at all* in e.g. development mode.[1]
This seems bad since it essentially means that this code is completely untested, despite being quite important for the built-in Firefox PDF viewer, and this thus ought to be fixed.
In this case, the explanation for the current state of the code should be "for historical reasons". Before the viewer was split into the current components and before the pre-processor was improved, back when all code resided in the `web/viewer.js` file, the telemetry reporting was done with *direct* `FirefoxCom` calls. However, with the dummy `DefaultExternalServices.reportTelemetry` method there's nothing actually preventing attempting to report telemetry in any type of build.
NOTE: By running this code in GENERIC builds as well, in addition to just locally, the *viewer* part of telemetry reporting becomes tested e.g. in preview builds too which should help with reviewing.
---
[1] When fixing bug 1606566, I had to edit the relevant `PDFJSDev` checks to be able to actually test the changes locally.
With https://bugzilla.mozilla.org/show_bug.cgi?id=844349 now being fixed in Firefox, the textLayer will now actually stay hidden as intended regardless of the browser settings.
Hence it should no longer be necessary to display the fallback bar, nor print a warning in the console, for documents which contains a textLayer.
Besides removing the `supportsDocumentColors` methods in the default viewer, we can also remove a now unused l10n string.
This rule is already enabled in mozilla-central, and helps avoid some confusing formatting, see https://searchfox.org/mozilla-central/rev/9e45d74b956be046e5021a746b0c8912f1c27318/tools/lint/eslint/eslint-plugin-mozilla/lib/configs/recommended.js#209-210
With the recent introduction of Prettier some of the existing nested ternary statements became even more difficult to read, since any possibly helpful indentation was removed.
This particular ESLint rule wasn't entirely straightforward to enable, and I do recognize that there's a certain amount of subjectivity in the changes being made. Generally, the changes in this patch fall into three categories:
- Cases where a value is only clamped to a certain range (the easiest ones to update).
- Cases where the values involved are "simple", such as Numbers and Strings, which are re-factored to initialize the variable with the *default* value and only update it when necessary by using `if`/`else if` statements.
- Cases with more complex and/or larger values, such as TypedArrays, which are re-factored to let the variable be (implicitly) undefined and where all values are then set through `if`/`else if`/`else` statements.
Please find additional details about the ESLint rule at https://eslint.org/docs/rules/no-nested-ternary
As described in the issue, having a DOM element with `id=page2` (or any other number) will automatically cause that element to become linkable through the URL hash. That's currently leading to some confusing and outright wrong behaviour, since it obviously only works for pages that have been loaded and rendered.
For PDF documents the only officially supported way to reference a particular page through the URL hash is using the `#page=2` format, which also works for all pages regardless if they're loaded or not.
As far as I can tell there's nothing in the PDF.js default viewer that actually depends on the page/thumbnail `id` at this point in time, hence why I believe that this removal ought to be safe.
Just as a pre-caution this patch adds an `aria-label` to the page canvas, similar to the thumbnail canvas/image, to at least keep this information in the DOM.
In order to eventually get rid of SystemJS and start using native `import`s instead, we'll need to provide "complete" file identifiers since otherwise there'll be MIME type errors when attempting to use `import`.
When working on PR 11463 I couldn't help thinking that the `Array.prototype.some` callback function, used when determining the generator, was somewhat difficult to read with its partly unused and strangely named parameters.
This is not only useful to have one format for consistency, but also to
be able to quickly search for colors for e.g., finding duplicates or
when tweaking the CSS for custom deployments.
This covers the handful of cases that the `--fix` command couldn't deal with, and the changes aren't just fixing the linting errors but attempt to slightly improve the relevant code.
Please find additional details about the ESLint rule at https://eslint.org/docs/rules/prefer-const
Note that this patch is generated automatically, by using the ESLint `--fix` argument, and will thus require some additional clean-up (which is done separately).
Rather than having a copy of this regular expression in the `test/unit/api_spec.js` file, with a comment about keeping it up-to-date with the code in the viewer (note the incorrect file reference as well), we can just import it instead to simplify all of this.
This patch makes the follow changes:
- Remove no longer necessary inline `// eslint-disable-...` comments.
- Fix `// eslint-disable-...` comments that Prettier moved down, thus causing new linting errors.
- Concatenate strings which now fit on just one line.
- Fix comments that are now too long.
- Finally, and most importantly, adjust comments that Prettier moved down, since the new positions often is confusing or outright wrong.
Note that Prettier, purposely, has only limited [configuration options](https://prettier.io/docs/en/options.html). The configuration file is based on [the one in `mozilla central`](https://searchfox.org/mozilla-central/source/.prettierrc) with just a few additions (to avoid future breakage if the defaults ever changes).
Prettier is being used for a couple of reasons:
- To be consistent with `mozilla-central`, where Prettier is already in use across the tree.
- To ensure a *consistent* coding style everywhere, which is automatically enforced during linting (since Prettier is used as an ESLint plugin). This thus ends "all" formatting disussions once and for all, removing the need for review comments on most stylistic matters.
Many ESLint options are now redundant, and I've tried my best to remove all the now unnecessary options (but I may have missed some).
Note also that since Prettier considers the `printWidth` option as a guide, rather than a hard rule, this patch resorts to a small hack in the ESLint config to ensure that *comments* won't become too long.
*Please note:* This patch is generated automatically, by appending the `--fix` argument to the ESLint call used in the `gulp lint` task. It will thus require some additional clean-up, which will be done in a *separate* commit.
(On a more personal note, I'll readily admit that some of the changes Prettier makes are *extremely* ugly. However, in the name of consistency we'll probably have to live with that.)
Apparently Ghostscript can, in some cases, generate/include `Metadata` with incorrectly encoded characters.[1] This results in the viewer title looking wrong, which we thus attempt to avoid by falling back to the `Info` entry instead.
*Please note:* Obviously it would be better if this was fixed in the `Metadata` parser instead, rather than using a viewer work-around, but I'm just not sure how or even *if* that could actually be done given that the `Metadata` stream contains no trace of the *original* character.
Fixes https://bugzilla.mozilla.org/show_bug.cgi?id=1605526
---
[1] The problematic characters are from the Specials Unicode block, see https://en.wikipedia.org/wiki/Specials_(Unicode_block)
With these changes we'll always set the `pdfTitle` to the `Info` dictionary entry *first* (assuming it exists and isn't empty), before attempting to override it with the `Metadata` stream entry (assuming it exists, is non-empty *and* valid).
There should be no functional changes with this patch, but it will simplify the following patch somewhat.
This patch reduces some duplication, by moving *all* fake worker loader code into the `setupFakeWorkerGlobal` function. Furthermore, the functions are simplified further by using `async`/`await` where appropriate.
Looking at the `parseCurrentHash` function again it's now difficult for me to understand *what* I was thinking, since having a helper function that needs to be manually passed a `linkService` reference just looks weird.
