Those methods are:
* `Value.coerce_to_receiver`: coerce to object for receivers
* `Value.as_callable`: as_object for callables
* `Activation.superclass_object`: get superclass of currently-called object
* `Activation.resolve_class`: resolve scope value, unwrap `ClassObject`, and error out if either step fails
This accounts for ~80% of coercion-related errors in `Activation`.
PNGs with 16-bit samples were not being normalized to 8-bit, screwing up their
appearance.
Add the `STRIP_16` transformation using `Transformations::normalize_to_color8()`,
which is equivalent to `Transformations::EXPAND | Transformations::STRIP_16`.
Also warn in case `png` outputs a color type other than `Rgb` and `Rgba`.
Due to the `EXPAND` transformation, this may be only `Grayscale` or
`GrayscaleAlpha`. `Indexed` is converted to either `Rgb` or `Rgba`.
Fixes#6662.
* Remove LineStyle::color, instead using fill_style with
FillStyle::Color to indicate solid color.
* Store `flags` in the struct instead of separate bools/values.
* Add getters/setters for ease of use.
* Add builder-style methods for setting LineStyle properties.
* Fix misnamed ALLOW_CLOSE flag to NO_CLOSE.
This feature stopped building with the bump to symphonia 0.5, which
added a `Sync` bound to its traits.
* Add `Sync` bounds to our own internal audio traits to match.
* nellymoser::Decoder was also tweaked to add a Sync bound.
* Lock the nellymoser dependency to a specific git commit.
Unify the previous 3 creation methods to a single `XmlNode::new`.
This allows supporting arbitrary `nodeType` values passed to the
`XMLNode` constructor.
Integer math was used when calculating a sound's start/endpoints,
because it was assumed that the sound sample rate was always an
even divisor of 44100Hz. However, some third party tools can embed
MP3s with other samples rates, such as #6569 which has a 16000Hz
MP3. This could also occur for dynamically loaded MP3s. This
results in the sound starting at an incorrect position.
Use floating point math to ensure the correct position is
caluclated.
Fixes#6569.
Remove the preload step that would pre-create the shapes for each
morph shape ratio on SWF load. Instead, lazily crate the shapes
when they are needed.
Loading unknown data is not considered an error on Flash; Both
`onLoadProgress` and `onLoadComplete` events are dispatched. But,
`onLoadProgress` reports 0 bytes loaded.
This re-uses the logic we have for handling AVM1's `ExternalInterface`.
For now, serialization/deserialization of non-array objects is
left unimplemented.
When decoding PNG/GIF data, convert the image to premultiplied
alpha so that it plays nicely with the renderer.
This applies to both dynamically loaded images and to PNG/GIF data
embedded inside a DefineBitsJPEG tag. This is in contrast to
DefineBitsLossless and DefineBitsJPEG tags w/ alpha, which are
already premultiplied in the SWF.
Also remove unnecessary clamps now that Rust defines casts as
saturating.
Fixes#6559.
These are *not* methods, because we cannot borrow both the update context and the loader at the same time. At least not without making loaders free-standing objects (`GcCell`s), which I don't want to do yet.
I (Michael R. Welsh) assign to Ruffle LLC all rights, title, and
interest to copyrights of my personal contributions to Ruffle,
effective March 26, 2022.
The difference between element nodes and text nodes is very minor.
So instead of representing them by two distinct enum members, make
`XmlNodeData` a single unified struct that can represent both. A new
`node_type` field is introduced, in order to still distinguish
between element and text nodes. Also, Ruffle made some incorrect
assumptions, which are now corrected, including:
* Nodes can have any arbitrary `u8` type. This is resolved by the
introduction of the `node_type` field which is a `u8`.
* Text nodes can have children. This is resolved by simply not checking
for text nodes in `append_child` etc.
Make it a thin abstraction layer over either the `futures` or `wasm-bindgen-futures`
crates, as already done in `render/wgpu/src/uniform_buffer.rs`,
instead of a hand-made single-thread executor.
Ideally this would also be usable on desktop, but I didn't manage to
get `LocalPool` working with `winit` (it needs to post a task to the
`EventLoopProxy` as a wake procedure).
In both `FrameLabel` and `Scene`, we define multiple
'public property / private slot' pairs.
The public property has a getter which delegates to the private
property. There is no setter for the property, ensuring that
the private slot can only be modified from within Ruffle itself.
This PR adds a macro `define_indirect_properties` to abstract over
this pattern. Currently, it only supports the read-only property
pattern - however, it could be extended in the future to generate
a setter that invokes a caller-provided callback function.
This needs to be a macro (rather than a method) so that we can
generate a function with the property name hard-coded into it.
Using a closure that references an upvar will not work, since
`Method::from_builtin` requires a function pointer.
Document roots (a.k.a. `XML` objects) are very similar to regular
element nodes (a.k.a. `XMLNode` objects). The primary difference is
that `XML` objects return `null` for their `nodeName`. But this can
be changed too; Setting `xml.nodeName = "someName"` will make `XML`
objects behave much like `XMLNode`. Moreover, many checks in Ruffle
that refuse to operate on document roots were wrong, and actually
these should be accepted as normal element nodes.
Besides the functional corrections, this also simplifies the code.
This also slightly changes behavior: Previously `SwfMovie::from_data`
errors were propagated as `FetchError` and `display_root_movie_download_failed_message`
was called, offering the user to sidestep CORS by opening the SWF
in a new tab. But that wouldn't help, obviously, because no network
error is involved. Now, these errors are propagated as `InvalidSwf`,
and `display_root_movie_download_failed_message` is not called.
`core` already depends on the `instant` crate which abstracts
`std::instant::Instant` and polyfills it on Web. Use it to replace
`NavigatorBackend::time_since_launch` in order to make `NavigatorBackend`
a little smaller and more simple.
Previously there were 3 implementations of `LocaleBackend`:
`DesktopLocaleBackend`, `WebLocaleBackend` and `NullLocaleBackend`.
While `DesktopLocaleBackend`, `WebLocaleBackend` were identical,
`NullLocaleBackend` always returned a fixed date/time for tests
determinism.
Unify them in a single file, and use `cfg!(test)` and a new dedicated
`deterministic` feature to decide whether to mock date/time or not.
This should not cause any behavioral changes.
Resolving `_levelN` had some inconsistencies with Flash:
1. `_flash` can be a prefix too.
2. The level ID parsing cannot fail; non-digit characters are ignored,
the value wraps around at `i32::MAX`, and negative values are valid.
This logic is relevant also for the `GetUrl` and `GetUrl2` opcodes.
For now only add TODOs for this.
This PR adds the following class stubs:
* 'flash.display.Loader'
* 'flash.net.URLRequest'
* 'flash.ui.Keyboard'
* 'flash.utils.Timer'
These are needed for 'This is the Only Level TOO' (though we'll need
actual implementations to get this game past the loading screen).
Previously, `isFinite()` with no arguments on SWF<7 incorrectly
returned `true`, as `undefined` coerced to `f64` is `0.0`, which
is finite. Fix this by not defaulting to `undefined`, similarly to
the `isNaN()` implementation.
Instead of returning a `Result` which is anyway always handled with
a `log::warn!()`, simply `log::warn!()` in place of errors. This
removes the last 3 remaining `Error` enum members besides `InvalidXml`.
The 2 existing usages of `remove_node` always operated on a child and
its parent: One iterates over all of its children and removes each one,
and the other explicitly grabs the parent of a given node.
As a simplification, `remove_node` operates only on a child node,
without the need for the parent node in addition; it grabs the parent
from the child by itself. As such, it's non-failable.
Text nodes are guaranteed to not be parents, as `adopt_child`
refuses to adopt children into them. So instead of returning an
`Err(Error::TextNodeCantHaveChildren)` in case of a text node parent,
mark those code paths as `unreachable!()`. This makes `orphan_child`
non-failable.
The `json` crate seems unmaintained, and recently also causes compile
errors with stable Rust 1.59.0. On the other hand, `serde_json` is
very maintained and more popular.
However, from some reason a cyclic package dependency has introduced
by this change. For now use a workaround from: https://github.com/tkaitchuck/aHash/issues/95#issuecomment-903560879
This is basically a revert of 61298b2be3.
`SharedObject`s used to be saved as JSON in Ruffle, but since #4238
they're saved in AMF to match Flash's behavior. The legacy JSON
deserialization remained for backwards-compatibility, but from what it
seems, it has never worked; cd1cde1708
changed `LocalStorageBackend` to store base64-encoded strings instead
of plain ones. Therefore, Ruffle attempts to base64-decode old JSON
data, and unsurprisingly fails.
In addition, this removes 1 out of 2 usages of the unmaintained `json`
crate, which recently also causes compile errors with stable Rust 1.59.0.
The only use of `last_parse_error` was in the `XML.prototype.status`
property, where it was converted into a number. Avoid storing it by
storing just the number.
Revert some of e50aea864b for an even
better approach - Remove `XmlNodeObject::empty_node` entirely by
making `XmlNodeObject::from_xml_node` a suitable alternative. That is,
being able to accept a custom `proto` like before.
Also, make it return an `XmlNodeObject` instead of an `Object`, and
add a few `.into()` where needed.
* Don't use `quick_xml::Writer` for formatting the XML, being much
more simple.
* Return `WString` instead of `String`, reducing `to_utf8_lossy()`
calls except when the string needs to be escaped (attribute values
and text contents).
As `XmlDocument` and `XmlObject` had 1-to-1 relation, and `XmlDocument`
is already tightly coupled with AVM1, there's no good reason for them
being separate objects.
This brings us one step closer towards an XML implementation hosted
completely in AVM1.
A future PR will merge `XmlNode` into `XmlNodeObject` in a similar
manner.
Instead of storing shared pointers to `Avm1ConstructorRegistry` in
`MovieLibrary`, access the `PropertyMap` directly, without an extra
abstraction.
Also, move the constructor registries to `Avm1`, for better
encapsulation.
Instead call `XmlNode::script_object`, which internally calls
`XmlNode::introduce_script_object`. This is a preparation for changing
the signature of `XmlNodeObject::from_xml_node`.
Currently it is not directly possible to configure lints for the
entire workspace via TOML, which forced us to repeat `#![allow]`
blocks in each crate.
embark pointed out this workaround to configure lints at the
workspace level via RUSTFLAGS:
https://github.com/EmbarkStudios/rust-ecosystem/issues/22#issuecomment-947011395
Remove the common `#![allow]` blocks and switch to this method for
global lint config.
Temporarily allow `needless_borrow` lint, buggy pending this fix:
https://github.com/rust-lang/rust-clippy/pull/8355
Previously we called `toString` when concatenating a string to an
Object. However, Flash actually has more complex behavior, usually
calling both `valueOf` and `toString`. This is loosely based on
ToPrimitive/DefaultValue with no type hint in the ECMAScript spec.
* Call `valueOf`.
* If the result isn't a primitive, call `toString`.
* If the result still isn't primitive, return `"[type Object]"`.
* For Date objects in SWFv6 and higher, call `toString`.
* If the result isn't a primitive, call `toString` (AVM1 bug?)
* If it still isn't primitive, return `"[type Object]"`.
This also rearranges some things about how we construct events, because `MouseEvent` has different defaults from `Event`. When we finally support parameter metadata on methods we should remove that code.
We also remove the `value_of` code on `EventObject` as that was a mistake. Events don't do anything special in there and I misinterpreted the test results the first time around.
This requires adding another notion of mouse-release events to `ClipEvent`. We now have four:
* `MouseUp` - the mouse was released, any object on the render list can handle this event ("anycast" event)
* `MouseUpInside` - the mouse was released inside this display object, only the mouse-picked target of the event can handle it
* `Release` - the mouse was released inside the last clicked display object
* `ReleaseOutside` - the mouse was released outside the last clicked display object
For those keeping score at home, in AVM2, the valid progression of events is either...
* On the same object, `mouseDown`, `mouseUp`, and `click`
* On one object, `mouseDown`, then some mouse movement that takes the cursor out of the first object, then on another object `mouseUp`, and then finally the first object gets `releaseOutside`.
`MouseDown`/`MouseUp` are effectively broadcasts; they hit every movie clip that can accept them until one of them has a handler for it. AVM2 instead wants events that only apply to specific mouse-picked display objects, which means we need to use the Player-tracked events `Press`, `Release`, and `ReleaseOutside`. The only problem is that we also need to emit a `mouseUp` event on both `Release` and `ReleaseOutside`.
SWFv5 always calls `Object.valueOf` at least once and sometimes
twice in the Equals2 op, even when comparing two Objects.
For example, `Object(1) == Object(1)` is true in SWFv5 but false
in SWFv6.
Consolidate several cases and fix some issues:
* Object-to-primitive comparison always goes through `valueOf`.
* `Object(undefined) == undefined` is true; this will coerce
to a bare object with no `valueOf`, resulting in
`undefined==undefined`.
* `{valueOf:function() { return NaN; }} == NaN` is true.
When creating a scope for a closure, any `with` scopes were being
filtered out, but this was incorrect; `with` scopes are still on
the scope chain when the function is called.
Flash ignores mismatched end tags (i.e. end tags with a missing/different
corresponding start tag). `quick-xml` checks end tag mismatches by
default, but it cannot recover after encountering one.
Commit 7e20543578 already disabled
`quick-xml`'s check, but that caused mismatched `Event::End` to be
handled, which may empty `format_stack` and later panic on
`format_stack.last().unwrap()`.
Thus, check for mismatched end tags ourselves, in a similar manner
of `quick-xml`, but in a recoverable way.
* Have `DefineFunction` and `DefineFunction2` go through the same
code path by implementing `From<DefineFunction>` for
`DefineFunction2`.
* Change `register` to a `Option<NonZeroU8>` for size optimization.
* Add `function::Param` to store param info instead of a tuple.
Use a struct for all variants of `avm1::Action`.
This makes the style more consistent instead of using a mix of
struct and tuple variants, and allows the data to be easily passed
around.
Handle strings, numbers and DisplayObject targets (not just MovieClips).
To support non-MovieClip targets, turn `clip.as_movie_clip().unwrap()`
to `if let Some(mc) = clip.as_movie_clip()` in `Loader`.