This PR serializes font data into an ArrayBuffer
that is then transfered from the worker to the
main thread. It's more efficient than the current
solution which clones the "export data" object
which includes the font data as a Uint8Array.
It prepares us to switch to a SharedArrayBuffer
in the future, which would allow us to share
the font data with multiple agents, which would be
crucial for the upcoming "renderer" worker.
The size of the canvas has significant impact on the rendering
performance. If we are going to render a high-res detail
view on top of the full-page canvas, we can further
reduce the full-page canvas resolution to improve
rendering time without affecting the resolution seen by
the user.
Users will se the lower resolution when quickly scrolling around the
page, but it will then be replaced with the high-res
detail view.
This PR changes the way we store bounding boxes so that they use less
memory and can be more easily shared across threads in the future.
Instead of storing the bounding box and list of dependencies for each
operation that renders _something_, we now only store the bounding box
of _every_ operation and no dependencies list. The bounding box of
each operation covers the bounding box of all the operations affected
by it that render something. For example, the bounding box of a
`setFont` operation will be the bounding box of all the `showText`
operations that use that font.
This affects the debugging experience in pdfBug, since now the bounding
box of an operation may be larger than what it renders itself. To help
with this, now when hovering on an operation we also highlight (in red)
all its dependents. We highlight with white stripes operations that do
not affect any part of the page (i.e. with an empty bbox).
To save memory, we now save bounding box x/y coordinates as uint8
rather than float64. This effectively gives us a 256x256 uniform grid
that covers the page, which is high enough resolution for the usecase.
Bug 1978027 has been fixed upstream 10 days ago, so this integration
test can be enabled for Firefox too now that it passed with recent
Nightly versions.
Most places have a newline before/after `before{Each,All}`,
`after{Each,All}` and `it` to visually separate the blocks for clarity,
but in a handful of places this wasn't done. This commit removes the
inconsistencies so that the test code is formatted consistently.
The helper function was used in a number of places, but also a lot of
places contained the annotation selector string inline. This commit
makes sure that all places use `getAnnotationSelector` consistently to
make sure the annotation selector string is only defined in a single
place and to improve readability of the test code.
This test called `closeSinglePage` manually at the end of the test,
which is inconsistent with all other tests that call `closePages` in an
`afterEach` block. This commit fixes the difference for consistency.
Locally, on Arch Linux, this integration test permafails:
```
1) Text layer Text selection using selection carets doesn't jump when moving selection
Message:
second selection:
Expected '(frequently executed) bytecode sequences, records
them, and compiles them to fast native code. We call such a s' to roughly match /frequently .* We call such a se/s.
Stack:
at <Jasmine>
at UserContext.<anonymous> (file:///home/timvandermeij/Documenten/Ontwikkeling/pdf.js/Code/test/integration/text_layer_spec.mjs:521:12)
Message:
third selection:
Expected '(frequently executed) bytecode sequences, records
them, and compiles them to fast native code. We call such a s' to roughly match /frequently .* We call such a se/s.
Stack:
at <Jasmine>
at UserContext.<anonymous> (file:///home/timvandermeij/Documenten/Ontwikkeling/pdf.js/Code/test/integration/text_layer_spec.mjs:529:12
```
The exact selection can differ a bit per OS/browser. In this case the
last character was consistently not selected while on other platforms it
is, so this commit fixes the issue by relaxing the regex to not consider
the final character so that the test passes if the rest matches.
This commit is a first step towards #6419, and it can also help with
first compute which ops can affect what is visible in that part of
the page.
This commit adds logic to track operations with their respective
bounding boxes. Only operations that actually cause something to
be rendered have a bounding box and dependencies.
Consider the following example:
```
0. setFillRGBColor
1. beginText
2. showText "Hello"
3. endText
4. constructPath [...] -> eoFill
```
here we have three rendering operations: the showText op (2) and the
path (4). (2) depends on (0), (1) and (3), while (4) only depends on
(0). Both (2) and (4) have a bounding box.
This tracking happens when first rendering a PDF: we then use the
recorded information to optimize future partial renderings of a PDF, so
that we can skip operations that do not affected the PDF area on the
canvas.
All this logic only runs when the new `enableOptimizedPartialRendering`
preference, disabled by default, is enabled.
The bounding boxes and dependencies are also shown in the pdfBug
stepper. When hovering over a step now:
- it highlights the steps that they depend on
- it highlights on the PDF itself the bounding box
We don't need AI/ML features in the tests, so this should reduce CPU
usage by not having the inference process running. Moreover, it prevents
the following lines from being logged in the test output:
```
JavaScript error: resource://gre/actors/MLEngineParent.sys.mjs, line 509: Error: Unable to get the ML engine from Remote Settings.
JavaScript error: resource://gre/actors/MLEngineParent.sys.mjs, line 1279: TypeError: can't access property "postMessage", this[#port] is null
```
This is a temporary measure to reduce noise until #20136 is fixed. Note
that this shouldn't be an issue in terms of coverage because we still
run the test on Linux.
In Chrome protocol calls are faster than in Firefox and thus trigger in
quicker succession. This can cause intermittent failures because new
protocol calls can run before events triggered by the previous protocol
calls had a chance to be processed (essentially causing events to get
lost).
This commit fixes the issue by configuring Chrome with a protocol call
delay value that gives it a more similar execution speed as Firefox
(which also gives us more consistency between the two browser runs).
Note that this doesn't negatively impact the overall runtime of the
integration tests because Puppeteer already waits for a test to complete
in both browsers before continuing to the next one and Chrome
consistently was, and with this patch still slightly is, faster in
completing the tests.
The fixed -400px horizontal offset used by
scrollIntoView led to horizontal scroll only moving
part-way right on narrow screens. The highlights near
the right-edge remained party or completely off
screen.
This centres the highlighted match on any viewport width while
clamping the left margin to 20-400px. On very narrow screens
the scrollbar now moves all the way to the right instead of
stopping midway.
It's useful for users highlighting with NVDA.
They've to enable native selection and then selection some text.
In this case only a selectionchange is triggered once the selection is done.
