liangsheng8708
/
ruffle
mirror of https://mirror.ghproxy.com/https://github.com/ruffle-rs/ruffle.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
ruffle/core/src/display_object.rs

2819 lines
105 KiB
Rust
Raw Blame History

use crate::avm1::{
ActivationIdentifier as Avm1ActivationIdentifier, Object as Avm1Object, TObject as Avm1TObject,
Value as Avm1Value,
};
use crate::avm2::{
Activation as Avm2Activation, Avm2, Error as Avm2Error, EventObject as Avm2EventObject,
Multiname as Avm2Multiname, Object as Avm2Object, TObject as Avm2TObject, Value as Avm2Value,
};
use crate::context::{RenderContext, UpdateContext};
use crate::drawing::Drawing;
use crate::prelude::*;
use crate::string::{AvmString, WString};
use crate::tag_utils::SwfMovie;
use crate::types::{Degrees, Percent};
use crate::vminterface::Instantiator;
use bitflags::bitflags;
use gc_arena::{Collect, Mutation};
use ruffle_macros::enum_trait_object;
use ruffle_render::pixel_bender::PixelBenderShaderHandle;
use ruffle_render::transform::{Transform, TransformStack};
use std::cell::{Ref, RefMut};
use std::fmt::Debug;
use std::hash::Hash;
use std::sync::Arc;
use swf::{ColorTransform, Fixed8};
mod avm1_button;
mod avm2_button;
mod bitmap;
mod container;
mod edit_text;
mod graphic;
mod interactive;
mod loader_display;
mod morph_shape;
mod movie_clip;
mod stage;
mod text;
mod video;
use crate::avm1::Activation;
use crate::display_object::bitmap::BitmapWeak;
pub use crate::display_object::container::{
dispatch_added_event_only, dispatch_added_to_stage_event_only, DisplayObjectContainer,
TDisplayObjectContainer,
};
pub use avm1_button::{Avm1Button, ButtonState, ButtonTracking};
pub use avm2_button::Avm2Button;
pub use bitmap::{Bitmap, BitmapClass};
pub use edit_text::{AutoSizeMode, EditText, TextSelection};
pub use graphic::Graphic;
pub use interactive::{Avm2MousePick, InteractiveObject, TInteractiveObject};
pub use loader_display::LoaderDisplay;
pub use morph_shape::MorphShape;
pub use movie_clip::{MovieClip, MovieClipWeak, Scene};
use ruffle_render::backend::{BitmapCacheEntry, RenderBackend};
use ruffle_render::bitmap::{BitmapHandle, BitmapInfo, PixelSnapping};
use ruffle_render::blend::ExtendedBlendMode;
use ruffle_render::commands::{CommandHandler, CommandList, RenderBlendMode};
use ruffle_render::filters::Filter;
pub use stage::{Stage, StageAlign, StageDisplayState, StageScaleMode, WindowMode};
pub use text::Text;
pub use video::Video;
use self::loader_display::LoaderDisplayWeak;
/// If a `DisplayObject` is marked `cacheAsBitmap` (via tag or AS),
/// this struct keeps the information required to uphold that cache.
/// A cached Display Object must have its bitmap invalidated when
/// any "visual" change happens, which can include:
/// - Changing the rotation
/// - Changing the scale
/// - Changing the alpha
/// - Changing the color transform
/// - Any "visual" change to children, **including** position changes
///
/// Position changes to the cached Display Object does not regenerate the cache,
/// allowing Display Objects to move freely without being regenerated.
///
/// Flash isn't very good at always recognising when it should be invalidated,
/// and there's cases such as changing the blend mode which don't always trigger it.
///
#[derive(Clone, Debug, Default)]
pub struct BitmapCache {
/// The `Matrix.a` value that was last used with this cache
matrix_a: f32,
/// The `Matrix.b` value that was last used with this cache
matrix_b: f32,
/// The `Matrix.c` value that was last used with this cache
matrix_c: f32,
/// The `Matrix.d` value that was last used with this cache
matrix_d: f32,
/// The width of the original bitmap, pre-filters
source_width: u16,
/// The height of the original bitmap, pre-filters
source_height: u16,
/// The offset used to draw the final bitmap (i.e. if a filter increases the size)
draw_offset: Point<i32>,
/// The current contents of the cache, if any. Values are post-filters.
bitmap: Option<BitmapInfo>,
/// Whether we warned that this bitmap was too large to be cached
warned_for_oversize: bool,
}
impl BitmapCache {
/// Forcefully make this BitmapCache invalid and require regeneration.
/// This should be used for changes that aren't automatically detected, such as children.
pub fn make_dirty(&mut self) {
// Setting the old transform to something invalid is a cheap way of making it invalid,
// without reserving an extra field for.
self.matrix_a = f32::NAN;
}
fn is_dirty(&self, other: &Matrix, source_width: u16, source_height: u16) -> bool {
self.matrix_a != other.a
|| self.matrix_b != other.b
|| self.matrix_c != other.c
|| self.matrix_d != other.d
|| self.source_width != source_width
|| self.source_height != source_height
|| self.bitmap.is_none()
}
/// Clears any dirtiness and ensure there's an appropriately sized texture allocated
#[allow(clippy::too_many_arguments)]
fn update(
&mut self,
renderer: &mut dyn RenderBackend,
matrix: Matrix,
source_width: u16,
source_height: u16,
actual_width: u16,
actual_height: u16,
draw_offset: Point<i32>,
swf_version: u8,
) {
self.matrix_a = matrix.a;
self.matrix_b = matrix.b;
self.matrix_c = matrix.c;
self.matrix_d = matrix.d;
self.source_width = source_width;
self.source_height = source_height;
self.draw_offset = draw_offset;
if let Some(current) = &mut self.bitmap {
if current.width == actual_width && current.height == actual_height {
return; // No need to resize it
}
}
let acceptable_size = if swf_version > 9 {
let total = actual_width as u32 * actual_height as u32;
actual_width < 8191 && actual_height < 8191 && total < 16777215
} else {
actual_width < 2880 && actual_height < 2880
};
if renderer.is_offscreen_supported()
&& actual_width > 0
&& actual_height > 0
&& acceptable_size
{
let handle = renderer.create_empty_texture(actual_width as u32, actual_height as u32);
self.bitmap = handle.ok().map(|handle| BitmapInfo {
width: actual_width,
height: actual_height,
handle,
});
} else {
self.bitmap = None;
}
}
/// Explicitly clears the cached value and drops any resources.
/// This should only be used in situations where you can't render to the cache and it needs to be
/// temporarily disabled.
fn clear(&mut self) {
self.bitmap = None;
}
fn handle(&self) -> Option<BitmapHandle> {
self.bitmap.as_ref().map(|b| b.handle.clone())
}
}
#[derive(Clone, Collect)]
#[collect(no_drop)]
pub struct DisplayObjectBase<'gc> {
parent: Option<DisplayObject<'gc>>,
place_frame: u16,
depth: Depth,
#[collect(require_static)]
transform: Transform,
name: Option<AvmString<'gc>>,
#[collect(require_static)]
filters: Vec<Filter>,
clip_depth: Depth,
// Cached transform properties `_xscale`, `_yscale`, `_rotation`.
// These are expensive to calculate, so they will be calculated and cached
// when AS requests one of these properties.
#[collect(require_static)]
rotation: Degrees,
#[collect(require_static)]
scale_x: Percent,
#[collect(require_static)]
scale_y: Percent,
skew: f64,
/// The next display object in order of execution.
///
/// `None` in an AVM2 movie.
next_avm1_clip: Option<DisplayObject<'gc>>,
/// The sound transform of sounds playing via this display object.
#[collect(require_static)]
sound_transform: SoundTransform,
/// The display object that we are being masked by.
masker: Option<DisplayObject<'gc>>,
/// The display object we are currently masking.
maskee: Option<DisplayObject<'gc>>,
meta_data: Option<Avm2Object<'gc>>,
/// The blend mode used when rendering this display object.
/// Values other than the default `BlendMode::Normal` implicitly cause cache-as-bitmap behavior.
#[collect(require_static)]
blend_mode: ExtendedBlendMode,
#[collect(require_static)]
blend_shader: Option<PixelBenderShaderHandle>,
/// The opaque background color of this display object.
/// The bounding box of the display object will be filled with the given color. This also
/// triggers cache-as-bitmap behavior. Only solid backgrounds are supported; the alpha channel
/// is ignored.
#[collect(require_static)]
opaque_background: Option<Color>,
/// Bit flags for various display object properties.
#[collect(require_static)]
flags: DisplayObjectFlags,
/// The 'internal' scroll rect used for rendering and methods like 'localToGlobal'.
/// This is updated from 'pre_render'
#[collect(require_static)]
scroll_rect: Option<Rectangle<Twips>>,
/// The 'next' scroll rect, which we will copy to 'scroll_rect' from 'pre_render'.
/// This is used by the ActionScript 'DisplayObject.scrollRect' getter, which sees
/// changes immediately (without needing wait for a render)
#[collect(require_static)]
next_scroll_rect: Rectangle<Twips>,
/// Rectangle used for 9-slice scaling (`DisplayObject.scale9grid`).
#[collect(require_static)]
scaling_grid: Rectangle<Twips>,
/// If this Display Object should cacheAsBitmap - and if so, the cache itself.
/// None means not cached, Some means cached.
#[collect(require_static)]
cache: Option<BitmapCache>,
}
impl<'gc> Default for DisplayObjectBase<'gc> {
fn default() -> Self {
Self {
parent: Default::default(),
place_frame: Default::default(),
depth: Default::default(),
transform: Default::default(),
name: None,
filters: Default::default(),
clip_depth: Default::default(),
rotation: Degrees::from_radians(0.0),
scale_x: Percent::from_unit(1.0),
scale_y: Percent::from_unit(1.0),
skew: 0.0,
next_avm1_clip: None,
masker: None,
maskee: None,
meta_data: None,
sound_transform: Default::default(),
blend_mode: Default::default(),
blend_shader: None,
opaque_background: Default::default(),
flags: DisplayObjectFlags::VISIBLE,
scroll_rect: None,
next_scroll_rect: Default::default(),
scaling_grid: Default::default(),
cache: None,
}
}
}
impl<'gc> DisplayObjectBase<'gc> {
/// Reset all properties that would be adjusted by a movie load.
fn reset_for_movie_load(&mut self) {
let flags_to_keep = self.flags & DisplayObjectFlags::LOCK_ROOT;
self.flags = flags_to_keep | DisplayObjectFlags::VISIBLE;
}
fn depth(&self) -> Depth {
self.depth
}
fn set_depth(&mut self, depth: Depth) {
self.depth = depth;
}
fn place_frame(&self) -> u16 {
self.place_frame
}
fn set_place_frame(&mut self, frame: u16) {
self.place_frame = frame;
}
fn transform(&self) -> &Transform {
&self.transform
}
pub fn matrix(&self) -> &Matrix {
&self.transform.matrix
}
pub fn matrix_mut(&mut self) -> &mut Matrix {
&mut self.transform.matrix
}
pub fn set_matrix(&mut self, matrix: Matrix) {
self.transform.matrix = matrix;
self.set_scale_rotation_cached(false);
}
pub fn color_transform(&self) -> &ColorTransform {
&self.transform.color_transform
}
pub fn color_transform_mut(&mut self) -> &mut ColorTransform {
&mut self.transform.color_transform
}
pub fn set_color_transform(&mut self, color_transform: ColorTransform) {
self.transform.color_transform = color_transform;
}
fn x(&self) -> Twips {
self.transform.matrix.tx
}
fn set_x(&mut self, x: Twips) -> bool {
let changed = self.transform.matrix.tx != x;
self.set_transformed_by_script(true);
self.transform.matrix.tx = x;
changed
}
fn y(&self) -> Twips {
self.transform.matrix.ty
}
fn set_y(&mut self, y: Twips) -> bool {
let changed = self.transform.matrix.ty != y;
self.set_transformed_by_script(true);
self.transform.matrix.ty = y;
changed
}
/// Caches the scale and rotation factors for this display object, if necessary.
/// Calculating these requires heavy trig ops, so we only do it when `_xscale`, `_yscale` or
/// `_rotation` is accessed.
fn cache_scale_rotation(&mut self) {
if !self.scale_rotation_cached() {
let (a, b, c, d) = (
f64::from(self.transform.matrix.a),
f64::from(self.transform.matrix.b),
f64::from(self.transform.matrix.c),
f64::from(self.transform.matrix.d),
);
// If this object's transform matrix is:
// [[a c tx]
// [b d ty]]
// After transformation, the X-axis and Y-axis will turn into the column vectors x' = <a, b> and y' = <c, d>.
// We derive the scale, rotation, and skew values from these transformed axes.
// The skew value is not exposed by ActionScript, but is remembered internally.
// xscale = len(x')
// yscale = len(y')
// rotation = atan2(b, a) (the rotation of x' from the normal x-axis).
// skew = atan2(-c, d) - atan2(b, a) (the signed difference between y' and x' rotation)
// This can produce some surprising results due to the overlap between flipping/rotation/skewing.
// For example, in Flash, using Modify->Transform->Flip Horizontal and then tracing _xscale, _yscale, and _rotation
// will output 100, 100, and 180. (a horizontal flip could also be a 180 degree skew followed by 180 degree rotation!)
let rotation_x = f64::atan2(b, a);
let rotation_y = f64::atan2(-c, d);
let scale_x = f64::sqrt(a * a + b * b);
let scale_y = f64::sqrt(c * c + d * d);
self.rotation = Degrees::from_radians(rotation_x);
self.scale_x = Percent::from_unit(scale_x);
self.scale_y = Percent::from_unit(scale_y);
self.skew = rotation_y - rotation_x;
}
}
fn rotation(&mut self) -> Degrees {
self.cache_scale_rotation();
self.rotation
}
fn set_rotation(&mut self, degrees: Degrees) -> bool {
self.set_transformed_by_script(true);
self.cache_scale_rotation();
let changed = self.rotation != degrees;
self.rotation = degrees;
// FIXME - this isn't quite correct. In Flash player,
// trying to set rotation to NaN does nothing if the current
// matrix 'b' and 'd' terms are both zero. However, if one
// of those terms is non-zero, then the entire matrix gets
// modified in a way that depends on its starting values.
// I haven't been able to figure out how to reproduce those
// values, so for now, we never modify the matrix if the
// rotation is NaN. Hopefully, there are no SWFs depending
// on the weird behavior when b or d is non-zero.
if degrees.into_radians().is_nan() {
return changed;
}
let cos_x = f64::cos(degrees.into_radians());
let sin_x = f64::sin(degrees.into_radians());
let cos_y = f64::cos(degrees.into_radians() + self.skew);
let sin_y = f64::sin(degrees.into_radians() + self.skew);
let matrix = &mut self.transform.matrix;
matrix.a = (self.scale_x.unit() * cos_x) as f32;
matrix.b = (self.scale_x.unit() * sin_x) as f32;
matrix.c = (self.scale_y.unit() * -sin_y) as f32;
matrix.d = (self.scale_y.unit() * cos_y) as f32;
changed
}
fn scale_x(&mut self) -> Percent {
self.cache_scale_rotation();
self.scale_x
}
fn set_scale_x(&mut self, mut value: Percent) -> bool {
let changed = self.scale_x != value;
self.set_transformed_by_script(true);
self.cache_scale_rotation();
self.scale_x = value;
// Note - in order to match Flash's behavior, the 'scale_x' field is set to NaN
// (which gets reported back to ActionScript), but we treat it as 0 for
// the purposes of updating the matrix
if value.percent().is_nan() {
value = 0.0.into();
}
// Similarly, a rotation of `NaN` can be reported to ActionScript, but we
// treat it as 0.0 when calculating the matrix
let mut rot = self.rotation.into_radians();
if rot.is_nan() {
rot = 0.0;
}
let cos = f64::cos(rot);
let sin = f64::sin(rot);
let matrix = &mut self.transform.matrix;
matrix.a = (cos * value.unit()) as f32;
matrix.b = (sin * value.unit()) as f32;
changed
}
fn scale_y(&mut self) -> Percent {
self.cache_scale_rotation();
self.scale_y
}
fn set_scale_y(&mut self, mut value: Percent) -> bool {
let changed = self.scale_y != value;
self.set_transformed_by_script(true);
self.cache_scale_rotation();
self.scale_y = value;
// Note - in order to match Flash's behavior, the 'scale_y' field is set to NaN
// (which gets reported back to ActionScript), but we treat it as 0 for
// the purposes of updating the matrix
if value.percent().is_nan() {
value = 0.0.into();
}
// Similarly, a rotation of `NaN` can be reported to ActionScript, but we
// treat it as 0.0 when calculating the matrix
let mut rot = self.rotation.into_radians();
if rot.is_nan() {
rot = 0.0;
}
let cos = f64::cos(rot + self.skew);
let sin = f64::sin(rot + self.skew);
let matrix = &mut self.transform.matrix;
matrix.c = (-sin * value.unit()) as f32;
matrix.d = (cos * value.unit()) as f32;
changed
}
fn name(&self) -> Option<AvmString<'gc>> {
self.name
}
fn set_name(&mut self, name: AvmString<'gc>) {
self.name = Some(name);
}
fn filters(&self) -> Vec<Filter> {
self.filters.clone()
}
fn set_filters(&mut self, filters: Vec<Filter>) -> bool {
if filters != self.filters {
self.filters = filters;
self.recheck_cache_as_bitmap();
true
} else {
false
}
}
fn alpha(&self) -> f64 {
f64::from(self.color_transform().a_multiply)
}
fn set_alpha(&mut self, value: f64) -> bool {
let changed = self.alpha() != value;
self.set_transformed_by_script(true);
self.color_transform_mut().a_multiply = Fixed8::from_f64(value);
changed
}
fn clip_depth(&self) -> Depth {
self.clip_depth
}
fn set_clip_depth(&mut self, depth: Depth) {
self.clip_depth = depth;
}
fn parent(&self) -> Option<DisplayObject<'gc>> {
self.parent
}
/// You should almost always use `DisplayObject.set_parent` instead, which
/// properly handles 'orphan' movie clips
fn set_parent_ignoring_orphan_list(&mut self, parent: Option<DisplayObject<'gc>>) {
self.parent = parent;
}
fn next_avm1_clip(&self) -> Option<DisplayObject<'gc>> {
self.next_avm1_clip
}
fn set_next_avm1_clip(&mut self, node: Option<DisplayObject<'gc>>) {
self.next_avm1_clip = node;
}
fn avm1_removed(&self) -> bool {
self.flags.contains(DisplayObjectFlags::AVM1_REMOVED)
}
fn avm1_pending_removal(&self) -> bool {
self.flags
.contains(DisplayObjectFlags::AVM1_PENDING_REMOVAL)
}
pub fn should_skip_next_enter_frame(&self) -> bool {
self.flags
.contains(DisplayObjectFlags::SKIP_NEXT_ENTER_FRAME)
}
pub fn set_skip_next_enter_frame(&mut self, skip: bool) {
self.flags
.set(DisplayObjectFlags::SKIP_NEXT_ENTER_FRAME, skip);
}
fn set_avm1_removed(&mut self, value: bool) {
self.flags.set(DisplayObjectFlags::AVM1_REMOVED, value);
}
fn set_avm1_pending_removal(&mut self, value: bool) {
self.flags
.set(DisplayObjectFlags::AVM1_PENDING_REMOVAL, value);
}
fn scale_rotation_cached(&self) -> bool {
self.flags
.contains(DisplayObjectFlags::SCALE_ROTATION_CACHED)
}
fn set_scale_rotation_cached(&mut self, set_flag: bool) {
if set_flag {
self.flags |= DisplayObjectFlags::SCALE_ROTATION_CACHED;
} else {
self.flags -= DisplayObjectFlags::SCALE_ROTATION_CACHED;
}
}
pub fn sound_transform(&self) -> &SoundTransform {
&self.sound_transform
}
pub fn set_sound_transform(&mut self, sound_transform: SoundTransform) {
self.sound_transform = sound_transform;
}
fn visible(&self) -> bool {
self.flags.contains(DisplayObjectFlags::VISIBLE)
}
fn set_visible(&mut self, value: bool) -> bool {
let changed = self.visible() != value;
self.flags.set(DisplayObjectFlags::VISIBLE, value);
changed
}
fn blend_mode(&self) -> ExtendedBlendMode {
self.blend_mode
}
fn set_blend_mode(&mut self, value: ExtendedBlendMode) -> bool {
let changed = self.blend_mode != value;
self.blend_mode = value;
changed
}
fn blend_shader(&self) -> Option<PixelBenderShaderHandle> {
self.blend_shader.clone()
}
fn set_blend_shader(&mut self, value: Option<PixelBenderShaderHandle>) {
self.blend_shader = value;
}
/// The opaque background color of this display object.
/// The bounding box of the display object will be filled with this color.
fn opaque_background(&self) -> Option<Color> {
self.opaque_background
}
/// The opaque background color of this display object.
/// The bounding box of the display object will be filled with the given color. This also
/// triggers cache-as-bitmap behavior. Only solid backgrounds are supported; the alpha channel
/// is ignored.
fn set_opaque_background(&mut self, value: Option<Color>) -> bool {
let value = value.map(|mut color| {
color.a = 255;
color
});
let changed = self.opaque_background != value;
self.opaque_background = value;
changed
}
fn is_root(&self) -> bool {
self.flags.contains(DisplayObjectFlags::IS_ROOT)
}
fn set_is_root(&mut self, value: bool) {
self.flags.set(DisplayObjectFlags::IS_ROOT, value);
}
fn lock_root(&self) -> bool {
self.flags.contains(DisplayObjectFlags::LOCK_ROOT)
}
fn set_lock_root(&mut self, value: bool) {
self.flags.set(DisplayObjectFlags::LOCK_ROOT, value);
}
fn transformed_by_script(&self) -> bool {
self.flags
.contains(DisplayObjectFlags::TRANSFORMED_BY_SCRIPT)
}
fn set_transformed_by_script(&mut self, value: bool) {
self.flags
.set(DisplayObjectFlags::TRANSFORMED_BY_SCRIPT, value);
}
fn placed_by_script(&self) -> bool {
self.flags.contains(DisplayObjectFlags::PLACED_BY_SCRIPT)
}
fn set_placed_by_script(&mut self, value: bool) {
self.flags.set(DisplayObjectFlags::PLACED_BY_SCRIPT, value);
}
fn is_bitmap_cached_preference(&self) -> bool {
self.flags.contains(DisplayObjectFlags::CACHE_AS_BITMAP)
}
fn set_bitmap_cached_preference(&mut self, value: bool) {
self.flags.set(DisplayObjectFlags::CACHE_AS_BITMAP, value);
self.recheck_cache_as_bitmap();
}
fn bitmap_cache_mut(&mut self) -> Option<&mut BitmapCache> {
self.cache.as_mut()
}
/// Invalidates a cached bitmap, if it exists.
/// This may only be called once per frame - the first call will return true, regardless of
/// if there was a cache.
/// Any subsequent calls will return false, indicating that you do not need to invalidate the ancestors.
/// This is reset during rendering.
fn invalidate_cached_bitmap(&mut self) -> bool {
if self.flags.contains(DisplayObjectFlags::CACHE_INVALIDATED) {
return false;
}
if let Some(cache) = &mut self.cache {
cache.make_dirty();
}
self.flags.insert(DisplayObjectFlags::CACHE_INVALIDATED);
true
}
fn clear_invalidate_flag(&mut self) {
self.flags.remove(DisplayObjectFlags::CACHE_INVALIDATED);
}
fn recheck_cache_as_bitmap(&mut self) {
let should_cache = self.is_bitmap_cached_preference() || !self.filters.is_empty();
if should_cache && self.cache.is_none() {
self.cache = Some(Default::default());
} else if !should_cache && self.cache.is_some() {
self.cache = None;
}
}
fn instantiated_by_timeline(&self) -> bool {
self.flags
.contains(DisplayObjectFlags::INSTANTIATED_BY_TIMELINE)
}
fn set_instantiated_by_timeline(&mut self, value: bool) {
self.flags
.set(DisplayObjectFlags::INSTANTIATED_BY_TIMELINE, value);
}
fn has_scroll_rect(&self) -> bool {
self.flags.contains(DisplayObjectFlags::HAS_SCROLL_RECT)
}
fn set_has_scroll_rect(&mut self, value: bool) {
self.flags.set(DisplayObjectFlags::HAS_SCROLL_RECT, value);
}
fn has_explicit_name(&self) -> bool {
self.flags.contains(DisplayObjectFlags::HAS_EXPLICIT_NAME)
}
fn set_has_explicit_name(&mut self, value: bool) {
self.flags.set(DisplayObjectFlags::HAS_EXPLICIT_NAME, value);
}
fn masker(&self) -> Option<DisplayObject<'gc>> {
self.masker
}
fn set_masker(&mut self, node: Option<DisplayObject<'gc>>) {
self.masker = node;
}
fn maskee(&self) -> Option<DisplayObject<'gc>> {
self.maskee
}
fn set_maskee(&mut self, node: Option<DisplayObject<'gc>>) {
self.maskee = node;
}
fn meta_data(&self) -> Option<Avm2Object<'gc>> {
self.meta_data
}
fn set_meta_data(&mut self, value: Avm2Object<'gc>) {
self.meta_data = Some(value);
}
}
struct DrawCacheInfo {
handle: BitmapHandle,
dirty: bool,
base_transform: Transform,
bounds: Rectangle<Twips>,
draw_offset: Point<i32>,
filters: Vec<Filter>,
}
pub fn render_base<'gc>(this: DisplayObject<'gc>, context: &mut RenderContext<'_, 'gc>) {
if this.maskee().is_some() {
return;
}
context.transform_stack.push(this.base().transform());
let blend_mode = this.blend_mode();
let original_commands = if blend_mode != ExtendedBlendMode::Normal {
Some(std::mem::take(&mut context.commands))
} else {
None
};
let cache_info = if context.use_bitmap_cache && this.is_bitmap_cached() {
let mut cache_info: Option<DrawCacheInfo> = None;
let base_transform = context.transform_stack.transform();
let bounds: Rectangle<Twips> = this.render_bounds_with_transform(
&base_transform.matrix,
false, // we want to do the filter growth for this object ourselves, to know the offsets
&context.stage.view_matrix(),
);
let name = this.name();
let mut filters: Vec<Filter> = this.filters();
let swf_version = this.swf_version();
filters.retain(|f| !f.impotent());
if let Some(cache) = this.base_mut(context.gc_context).bitmap_cache_mut() {
let width = bounds.width().to_pixels().ceil().max(0.0);
let height = bounds.height().to_pixels().ceil().max(0.0);
if width <= u16::MAX as f64 && height <= u16::MAX as f64 {
let width = width as u16;
let height = height as u16;
let mut filter_rect = Rectangle {
x_min: Twips::ZERO,
x_max: Twips::from_pixels_i32(width as i32),
y_min: Twips::ZERO,
y_max: Twips::from_pixels_i32(height as i32),
};
let stage_matrix = context.stage.view_matrix();
for filter in &mut filters {
// Scaling is done by *stage view matrix* only, nothing in-between
filter.scale(stage_matrix.a, stage_matrix.d);
filter_rect = filter.calculate_dest_rect(filter_rect);
}
let filter_rect = Rectangle {
x_min: filter_rect.x_min.to_pixels().floor() as i32,
x_max: filter_rect.x_max.to_pixels().ceil() as i32,
y_min: filter_rect.y_min.to_pixels().floor() as i32,
y_max: filter_rect.y_max.to_pixels().ceil() as i32,
};
let draw_offset = Point::new(filter_rect.x_min, filter_rect.y_min);
if cache.is_dirty(&base_transform.matrix, width, height) {
cache.update(
context.renderer,
base_transform.matrix,
width,
height,
filter_rect.width() as u16,
filter_rect.height() as u16,
draw_offset,
swf_version,
);
cache_info = cache.handle().map(|handle| DrawCacheInfo {
handle,
dirty: true,
base_transform,
bounds,
draw_offset,
filters,
});
} else {
cache_info = cache.handle().map(|handle| DrawCacheInfo {
handle,
dirty: false,
base_transform,
bounds,
draw_offset,
filters,
});
}
} else {
if !cache.warned_for_oversize {
tracing::warn!(
"Skipping cacheAsBitmap for incredibly large object {:?} ({width} x {height})",
name
);
cache.warned_for_oversize = true;
}
cache.clear();
cache_info = None;
}
}
cache_info
} else {
None
};
// We can't hold `cache` (which will hold `base`), so this is split up
if let Some(cache_info) = cache_info {
// In order to render an object to a texture, we need to draw its entire bounds.
// Calculate the offset from tx/ty in order to accommodate any drawings that extend the bounds
// negatively
let offset_x = cache_info.bounds.x_min - cache_info.base_transform.matrix.tx
+ Twips::from_pixels_i32(cache_info.draw_offset.x);
let offset_y = cache_info.bounds.y_min - cache_info.base_transform.matrix.ty
+ Twips::from_pixels_i32(cache_info.draw_offset.y);
if cache_info.dirty {
let mut transform_stack = TransformStack::new();
transform_stack.push(&Transform {
color_transform: Default::default(),
matrix: Matrix {
tx: -offset_x,
ty: -offset_y,
..cache_info.base_transform.matrix
},
});
let mut offscreen_context = RenderContext {
renderer: context.renderer,
commands: CommandList::new(),
cache_draws: context.cache_draws,
gc_context: context.gc_context,
library: context.library,
transform_stack: &mut transform_stack,
is_offscreen: true,
use_bitmap_cache: true,
stage: context.stage,
};
this.render_self(&mut offscreen_context);
offscreen_context.cache_draws.push(BitmapCacheEntry {
handle: cache_info.handle.clone(),
commands: offscreen_context.commands,
clear: this.opaque_background().unwrap_or_default(),
filters: cache_info.filters,
});
}
// When rendering it back, ensure we're only keeping the translation - scale/rotation is within the image already
apply_standard_mask_and_scroll(this, context, |context| {
context.commands.render_bitmap(
cache_info.handle,
Transform {
matrix: Matrix {
tx: context.transform_stack.transform().matrix.tx + offset_x,
ty: context.transform_stack.transform().matrix.ty + offset_y,
..Default::default()
},
color_transform: cache_info.base_transform.color_transform,
},
true,
PixelSnapping::Always, // cacheAsBitmap forces pixel snapping
)
});
} else {
if let Some(background) = this.opaque_background() {
// This is intended for use with cacheAsBitmap, but can be set for non-cached objects too
// It wants the entire bounding box to be cleared before any draws happen
let bounds: Rectangle<Twips> = this.render_bounds_with_transform(
&context.transform_stack.transform().matrix,
true,
&context.stage.view_matrix(),
);
context.commands.draw_rect(
background,
Matrix::create_box(
bounds.width().to_pixels() as f32,
bounds.height().to_pixels() as f32,
bounds.x_min,
bounds.y_min,
),
);
}
apply_standard_mask_and_scroll(this, context, |context| this.render_self(context));
}
if let Some(original_commands) = original_commands {
let sub_commands = std::mem::replace(&mut context.commands, original_commands);
let render_blend_mode = if let ExtendedBlendMode::Shader = blend_mode {
// Note - Flash appears to let you set `dobj.blendMode = BlendMode.SHADER` without
// having `dobj.blendShader` result, but the resulting rendered displayobject
// seems to be corrupted. For now, let's panic, and see if any swfs actually
// rely on this behavior.
RenderBlendMode::Shader(this.blend_shader().expect("Missing blend shader"))
} else {
RenderBlendMode::Builtin(blend_mode.try_into().unwrap())
};
context.commands.blend(sub_commands, render_blend_mode);
}
context.transform_stack.pop();
}
/// This applies the **standard** method of `mask` and `scrollRect`.
///
/// It uses the stencil buffer so that any pixel drawn in the mask will allow the inner contents to show.
/// This is what is used for most cases, except for cacheAsBitmap-on-cacheAsBitmap.
pub fn apply_standard_mask_and_scroll<'gc, F>(
this: DisplayObject<'gc>,
context: &mut RenderContext<'_, 'gc>,
draw: F,
) where
F: FnOnce(&mut RenderContext<'_, 'gc>),
{
let scroll_rect_matrix = if let Some(rect) = this.scroll_rect() {
let cur_transform = context.transform_stack.transform();
// The matrix we use for actually drawing a rectangle for cropping purposes
// Note that we do *not* apply the translation yet
Some(
cur_transform.matrix
* Matrix::scale(
rect.width().to_pixels() as f32,
rect.height().to_pixels() as f32,
),
)
} else {
None
};
if let Some(rect) = this.scroll_rect() {
// Translate everything that we render (including DisplayObject.mask)
context.transform_stack.push(&Transform {
matrix: Matrix::translate(-rect.x_min, -rect.y_min),
color_transform: Default::default(),
});
}
let mask = this.masker();
let mut mask_transform = ruffle_render::transform::Transform::default();
if let Some(m) = mask {
mask_transform.matrix = this.global_to_local_matrix().unwrap_or_default();
mask_transform.matrix *= m.local_to_global_matrix();
context.commands.push_mask();
context.transform_stack.push(&mask_transform);
m.render_self(context);
context.transform_stack.pop();
context.commands.activate_mask();
}
// There are two parts to 'DisplayObject.scrollRect':
// a scroll effect (translation), and a crop effect.
// This scroll is implementing by applying a translation matrix
// when we defined 'scroll_rect_matrix'.
// The crop is implemented as a rectangular mask using the height
// and width provided by 'scrollRect'.
// Note that this mask is applied *in addition to* a mask defined
// with 'DisplayObject.mask'. We will end up rendering content that
// lies in the intersection of the scroll rect and DisplayObject.mask,
// which is exactly the behavior that we want.
if let Some(rect_mat) = scroll_rect_matrix {
context.commands.push_mask();
// The color doesn't matter, as this is a mask.
context.commands.draw_rect(Color::WHITE, rect_mat);
context.commands.activate_mask();
}
draw(context);
if let Some(rect_mat) = scroll_rect_matrix {
// Draw the rectangle again after deactivating the mask,
// to reset the stencil buffer.
context.commands.deactivate_mask();
context.commands.draw_rect(Color::WHITE, rect_mat);
context.commands.pop_mask();
}
if let Some(m) = mask {
context.commands.deactivate_mask();
context.transform_stack.push(&mask_transform);
m.render_self(context);
context.transform_stack.pop();
context.commands.pop_mask();
}
if scroll_rect_matrix.is_some() {
// Remove the translation that we pushed
context.transform_stack.pop();
}
}
#[enum_trait_object(
#[derive(Clone, Collect, Debug, Copy)]
#[collect(no_drop)]
pub enum DisplayObject<'gc> {
Stage(Stage<'gc>),
Bitmap(Bitmap<'gc>),
Avm1Button(Avm1Button<'gc>),
Avm2Button(Avm2Button<'gc>),
EditText(EditText<'gc>),
Graphic(Graphic<'gc>),
MorphShape(MorphShape<'gc>),
MovieClip(MovieClip<'gc>),
Text(Text<'gc>),
Video(Video<'gc>),
LoaderDisplay(LoaderDisplay<'gc>)
}
)]
pub trait TDisplayObject<'gc>:
'gc + Clone + Copy + Collect + Debug + Into<DisplayObject<'gc>>
{
fn base<'a>(&'a self) -> Ref<'a, DisplayObjectBase<'gc>>;
fn base_mut<'a>(&'a self, mc: &Mutation<'gc>) -> RefMut<'a, DisplayObjectBase<'gc>>;
/// The `SCALE_ROTATION_CACHED` flag should only be set in SWFv5+.
/// So scaling/rotation values always have to get recalculated from the matrix in SWFv4.
fn set_scale_rotation_cached(&self, gc_context: &Mutation<'gc>) {
if self.swf_version() >= 5 {
self.base_mut(gc_context).set_scale_rotation_cached(true);
}
}
fn id(&self) -> CharacterId;
fn depth(&self) -> Depth {
self.base().depth()
}
fn set_depth(&self, gc_context: &Mutation<'gc>, depth: Depth) {
self.base_mut(gc_context).set_depth(depth)
}
/// The untransformed inherent bounding box of this object.
/// These bounds do **not** include child DisplayObjects.
/// To get the bounds including children, use `bounds`, `local_bounds`, or `world_bounds`.
///
/// Implementors must override this method.
/// Leaf DisplayObjects should return their bounds.
/// Composite DisplayObjects that only contain children should return `&Default::default()`
fn self_bounds(&self) -> Rectangle<Twips>;
/// The untransformed bounding box of this object including children.
fn bounds(&self) -> Rectangle<Twips> {
self.bounds_with_transform(&Matrix::default())
}
/// The local bounding box of this object including children, in its parent's coordinate system.
fn local_bounds(&self) -> Rectangle<Twips> {
self.bounds_with_transform(self.base().matrix())
}
/// The world bounding box of this object including children, relative to the stage.
fn world_bounds(&self) -> Rectangle<Twips> {
self.bounds_with_transform(&self.local_to_global_matrix())
}
/// Bounds used for drawing debug rects and picking objects.
fn debug_rect_bounds(&self) -> Rectangle<Twips> {
// Make the rect at least as big as highlight bounds to ensure that anything
// interactive is also highlighted even if not included in world bounds.
let highlight_bounds = self
.as_interactive()
.map(|int| int.highlight_bounds())
.unwrap_or_default();
self.world_bounds().union(&highlight_bounds)
}
/// Gets the bounds of this object and all children, transformed by a given matrix.
/// This function recurses down and transforms the AABB each child before adding
/// it to the bounding box. This gives a tighter AABB then if we simply transformed
/// the overall AABB.
fn bounds_with_transform(&self, matrix: &Matrix) -> Rectangle<Twips> {
// A scroll rect completely overrides an object's bounds,
// and can even grow the bounding box to be larger than the actual content
if let Some(scroll_rect) = self.scroll_rect() {
return *matrix
* Rectangle {
x_min: Twips::ZERO,
y_min: Twips::ZERO,
x_max: scroll_rect.width(),
y_max: scroll_rect.height(),
};
}
let mut bounds = *matrix * self.self_bounds();
if let Some(ctr) = self.as_container() {
for child in ctr.iter_render_list() {
let matrix = *matrix * *child.base().matrix();
bounds = bounds.union(&child.bounds_with_transform(&matrix));
}
}
bounds
}
/// Gets the **render bounds** of this object and all its children.
/// This differs from the bounds that are exposed to Flash, in two main ways:
/// - It may be larger if filters are applied which will increase the size of what's shown
/// - It does not respect scroll rects
fn render_bounds_with_transform(
&self,
matrix: &Matrix,
include_own_filters: bool,
view_matrix: &Matrix,
) -> Rectangle<Twips> {
let mut bounds = *matrix * self.self_bounds();
if let Some(ctr) = self.as_container() {
for child in ctr.iter_render_list() {
let matrix = *matrix * *child.base().matrix();
bounds =
bounds.union(&child.render_bounds_with_transform(&matrix, true, view_matrix));
}
}
if include_own_filters {
let filters = self.filters();
for mut filter in filters {
filter.scale(view_matrix.a, view_matrix.d);
bounds = filter.calculate_dest_rect(bounds);
}
}
bounds
}
fn place_frame(&self) -> u16 {
self.base().place_frame()
}
fn set_place_frame(&self, gc_context: &Mutation<'gc>, frame: u16) {
self.base_mut(gc_context).set_place_frame(frame)
}
/// Sets the matrix of this object.
/// This does NOT invalidate the cache, as it's often used with other operations.
/// It is the callers responsibility to do so.
fn set_matrix(&self, gc_context: &Mutation<'gc>, matrix: Matrix) {
self.base_mut(gc_context).set_matrix(matrix);
}
/// Sets the color transform of this object.
/// This does NOT invalidate the cache, as it's often used with other operations.
/// It is the callers responsibility to do so.
fn set_color_transform(&self, gc_context: &Mutation<'gc>, color_transform: ColorTransform) {
self.base_mut(gc_context)
.set_color_transform(color_transform)
}
/// Should only be used to implement 'Transform.concatenatedMatrix'
fn local_to_global_matrix_without_own_scroll_rect(&self) -> Matrix {
let mut node = self.parent();
let mut matrix = *self.base().matrix();
while let Some(display_object) = node {
// We want to transform to Stage-local coordinates,
// so do *not* apply the Stage's matrix
if display_object.as_stage().is_some() {
break;
}
if let Some(rect) = display_object.scroll_rect() {
matrix = Matrix::translate(-rect.x_min, -rect.y_min) * matrix;
}
matrix = *display_object.base().matrix() * matrix;
node = display_object.parent();
}
matrix
}
/// Returns the matrix for transforming from this object's local space to global stage space.
fn local_to_global_matrix(&self) -> Matrix {
let mut matrix = Matrix::IDENTITY;
if let Some(rect) = self.scroll_rect() {
matrix = Matrix::translate(-rect.x_min, -rect.y_min) * matrix;
}
self.local_to_global_matrix_without_own_scroll_rect() * matrix
}
/// Returns the matrix for transforming from global stage to this object's local space.
/// `None` is returned if the object has zero scale.
fn global_to_local_matrix(&self) -> Option<Matrix> {
self.local_to_global_matrix().inverse()
}
/// Converts a local position to a global stage position
fn local_to_global(&self, local: Point<Twips>) -> Point<Twips> {
self.local_to_global_matrix() * local
}
/// Converts a local position on the stage to a local position on this display object
/// Returns `None` if the object has zero scale.
fn global_to_local(&self, global: Point<Twips>) -> Option<Point<Twips>> {
self.global_to_local_matrix().map(|matrix| matrix * global)
}
/// Converts the mouse position on the stage to a local position on this display object.
/// If the object has zero scale, then the stage `TWIPS_TO_PIXELS` matrix will be used.
/// This matches Flash's behavior for `mouseX`/`mouseY` on an object with zero scale.
fn local_mouse_position(&self, context: &UpdateContext<'_, 'gc>) -> Point<Twips> {
let stage = context.stage;
let pixel_ratio = stage.view_matrix().a;
let virtual_to_device = Matrix::scale(pixel_ratio, pixel_ratio);
// Get mouse pos in global device pixels
let global_twips = *context.mouse_position;
let global_device_twips = virtual_to_device * global_twips;
let global_device_pixels = Matrix::TWIPS_TO_PIXELS * global_device_twips;
// Make transformation matrix
let local_twips_to_global_twips = self.local_to_global_matrix();
let twips_to_device_pixels = virtual_to_device * Matrix::TWIPS_TO_PIXELS;
let local_twips_to_global_device_pixels =
twips_to_device_pixels * local_twips_to_global_twips;
let global_device_pixels_to_local_twips = local_twips_to_global_device_pixels
.inverse()
.unwrap_or(Matrix::IDENTITY);
// Get local mouse position in twips
global_device_pixels_to_local_twips * global_device_pixels
}
/// The `x` position in pixels of this display object in local space.
/// Returned by the `_x`/`x` ActionScript properties.
fn x(&self) -> Twips {
self.base().x()
}
/// Sets the `x` position in pixels of this display object in local space.
/// Set by the `_x`/`x` ActionScript properties.
/// This invalidates any ancestors cacheAsBitmap automatically.
fn set_x(&self, gc_context: &Mutation<'gc>, x: Twips) {
if self.base_mut(gc_context).set_x(x) {
if let Some(parent) = self.parent() {
// Self-transform changes are automatically handled,
// we only want to inform ancestors to avoid unnecessary invalidations for tx/ty
parent.invalidate_cached_bitmap(gc_context);
}
}
}
/// The `y` position in pixels of this display object in local space.
/// Returned by the `_y`/`y` ActionScript properties.
fn y(&self) -> Twips {
self.base().y()
}
/// Sets the `y` position in pixels of this display object in local space.
/// Set by the `_y`/`y` ActionScript properties.
/// This invalidates any ancestors cacheAsBitmap automatically.
fn set_y(&self, gc_context: &Mutation<'gc>, y: Twips) {
if self.base_mut(gc_context).set_y(y) {
if let Some(parent) = self.parent() {
// Self-transform changes are automatically handled,
// we only want to inform ancestors to avoid unnecessary invalidations for tx/ty
parent.invalidate_cached_bitmap(gc_context);
}
}
}
/// The rotation in degrees this display object in local space.
/// Returned by the `_rotation`/`rotation` ActionScript properties.
fn rotation(&self, gc_context: &Mutation<'gc>) -> Degrees {
let degrees = self.base_mut(gc_context).rotation();
self.set_scale_rotation_cached(gc_context);
degrees
}
/// Sets the rotation in degrees this display object in local space.
/// Set by the `_rotation`/`rotation` ActionScript properties.
/// This invalidates any ancestors cacheAsBitmap automatically.
fn set_rotation(&self, gc_context: &Mutation<'gc>, radians: Degrees) {
if self.base_mut(gc_context).set_rotation(radians) {
self.set_scale_rotation_cached(gc_context);
if let Some(parent) = self.parent() {
// Self-transform changes are automatically handled,
// we only want to inform ancestors to avoid unnecessary invalidations for tx/ty
parent.invalidate_cached_bitmap(gc_context);
}
}
}
/// The X axis scale for this display object in local space.
/// Returned by the `_xscale`/`scaleX` ActionScript properties.
fn scale_x(&self, gc_context: &Mutation<'gc>) -> Percent {
let percent = self.base_mut(gc_context).scale_x();
self.set_scale_rotation_cached(gc_context);
percent
}
/// Sets the X axis scale for this display object in local space.
/// Set by the `_xscale`/`scaleX` ActionScript properties.
/// This invalidates any ancestors cacheAsBitmap automatically.
fn set_scale_x(&self, gc_context: &Mutation<'gc>, value: Percent) {
if self.base_mut(gc_context).set_scale_x(value) {
self.set_scale_rotation_cached(gc_context);
if let Some(parent) = self.parent() {
// Self-transform changes are automatically handled,
// we only want to inform ancestors to avoid unnecessary invalidations for tx/ty
parent.invalidate_cached_bitmap(gc_context);
}
}
}
/// The Y axis scale for this display object in local space.
/// Returned by the `_yscale`/`scaleY` ActionScript properties.
fn scale_y(&self, gc_context: &Mutation<'gc>) -> Percent {
let percent = self.base_mut(gc_context).scale_y();
self.set_scale_rotation_cached(gc_context);
percent
}
/// Sets the Y axis scale for this display object in local space.
/// Returned by the `_yscale`/`scaleY` ActionScript properties.
/// This invalidates any ancestors cacheAsBitmap automatically.
fn set_scale_y(&self, gc_context: &Mutation<'gc>, value: Percent) {
if self.base_mut(gc_context).set_scale_y(value) {
self.set_scale_rotation_cached(gc_context);
if let Some(parent) = self.parent() {
// Self-transform changes are automatically handled,
// we only want to inform ancestors to avoid unnecessary invalidations for tx/ty
parent.invalidate_cached_bitmap(gc_context);
}
}
}
/// Gets the pixel width of the AABB containing this display object in local space.
/// Returned by the ActionScript `_width`/`width` properties.
fn width(&self) -> f64 {
self.local_bounds().width().to_pixels()
}
/// Sets the pixel width of this display object in local space.
/// The width is based on the AABB of the object.
/// Set by the ActionScript `_width`/`width` properties.
/// This does odd things on rotated clips to match the behavior of Flash.
fn set_width(&self, context: &mut UpdateContext<'_, 'gc>, value: f64) {
let gc_context = context.gc_context;
let object_bounds = self.bounds();
let object_width = object_bounds.width().to_pixels();
let object_height = object_bounds.height().to_pixels();
let aspect_ratio = object_height / object_width;
let (target_scale_x, target_scale_y) = if object_width != 0.0 {
(value / object_width, value / object_height)
} else {
(0.0, 0.0)
};
// No idea about the derivation of this -- figured it out via lots of trial and error.
// It has to do with the length of the sides A, B of an AABB enclosing the object's OBB with sides a, b:
// A = sin(t) * a + cos(t) * b
// B = cos(t) * a + sin(t) * b
let prev_scale_x = self.scale_x(gc_context).unit();
let prev_scale_y = self.scale_y(gc_context).unit();
let rotation = self.rotation(gc_context);
let cos = f64::abs(f64::cos(rotation.into_radians()));
let sin = f64::abs(f64::sin(rotation.into_radians()));
let mut new_scale_x = aspect_ratio * (cos * target_scale_x + sin * target_scale_y)
/ ((cos + aspect_ratio * sin) * (aspect_ratio * cos + sin));
let mut new_scale_y =
(sin * prev_scale_x + aspect_ratio * cos * prev_scale_y) / (aspect_ratio * cos + sin);
if !new_scale_x.is_finite() {
new_scale_x = 0.0;
}
if !new_scale_y.is_finite() {
new_scale_y = 0.0;
}
self.set_scale_x(gc_context, Percent::from_unit(new_scale_x));
self.set_scale_y(gc_context, Percent::from_unit(new_scale_y));
}
/// Gets the pixel height of the AABB containing this display object in local space.
/// Returned by the ActionScript `_height`/`height` properties.
fn height(&self) -> f64 {
self.local_bounds().height().to_pixels()
}
/// Sets the pixel height of this display object in local space.
/// Set by the ActionScript `_height`/`height` properties.
/// This does odd things on rotated clips to match the behavior of Flash.
fn set_height(&self, context: &mut UpdateContext<'_, 'gc>, value: f64) {
let gc_context = context.gc_context;
let object_bounds = self.bounds();
let object_width = object_bounds.width().to_pixels();
let object_height = object_bounds.height().to_pixels();
let aspect_ratio = object_width / object_height;
let (target_scale_x, target_scale_y) = if object_height != 0.0 {
(value / object_width, value / object_height)
} else {
(0.0, 0.0)
};
// No idea about the derivation of this -- figured it out via lots of trial and error.
// It has to do with the length of the sides A, B of an AABB enclosing the object's OBB with sides a, b:
// A = sin(t) * a + cos(t) * b
// B = cos(t) * a + sin(t) * b
let prev_scale_x = self.scale_x(gc_context).unit();
let prev_scale_y = self.scale_y(gc_context).unit();
let rotation = self.rotation(gc_context);
let cos = f64::abs(f64::cos(rotation.into_radians()));
let sin = f64::abs(f64::sin(rotation.into_radians()));
let mut new_scale_x =
(aspect_ratio * cos * prev_scale_x + sin * prev_scale_y) / (aspect_ratio * cos + sin);
let mut new_scale_y = aspect_ratio * (sin * target_scale_x + cos * target_scale_y)
/ ((cos + aspect_ratio * sin) * (aspect_ratio * cos + sin));
if !new_scale_x.is_finite() {
new_scale_x = 0.0;
}
if !new_scale_y.is_finite() {
new_scale_y = 0.0;
}
self.set_scale_x(gc_context, Percent::from_unit(new_scale_x));
self.set_scale_y(gc_context, Percent::from_unit(new_scale_y));
}
/// The opacity of this display object.
/// 1 is fully opaque.
/// Returned by the `_alpha`/`alpha` ActionScript properties.
fn alpha(&self) -> f64 {
self.base().alpha()
}
/// Sets the opacity of this display object.
/// 1 is fully opaque.
/// Set by the `_alpha`/`alpha` ActionScript properties.
/// This invalidates any cacheAsBitmap automatically.
fn set_alpha(&self, gc_context: &Mutation<'gc>, value: f64) {
if self.base_mut(gc_context).set_alpha(value) {
if let Some(parent) = self.parent() {
// Self-transform changes are automatically handled
parent.invalidate_cached_bitmap(gc_context);
}
}
}
fn name(&self) -> AvmString<'gc> {
self.base().name().unwrap_or_default()
}
fn name_optional(&self) -> Option<AvmString<'gc>> {
self.base().name()
}
fn set_name(&self, gc_context: &Mutation<'gc>, name: AvmString<'gc>) {
self.base_mut(gc_context).set_name(name)
}
fn filters(&self) -> Vec<Filter> {
self.base().filters()
}
fn set_filters(&self, gc_context: &Mutation<'gc>, filters: Vec<Filter>) {
if self.base_mut(gc_context).set_filters(filters) {
self.invalidate_cached_bitmap(gc_context);
}
}
/// Returns the dot-syntax path to this display object, e.g. `_level0.foo.clip`
fn path(&self) -> WString {
if let Some(parent) = self.avm1_parent() {
let mut path = parent.path();
path.push_byte(b'.');
path.push_str(&self.name());
path
} else {
WString::from_utf8_owned(format!("_level{}", self.depth()))
}
}
/// Returns the Flash 4 slash-syntax path to this display object, e.g. `/foo/clip`.
/// Returned by the `_target` property in AVM1.
fn slash_path(&self) -> WString {
fn build_slash_path(object: DisplayObject<'_>) -> WString {
if let Some(parent) = object.avm1_parent() {
let mut path = build_slash_path(parent);
path.push_byte(b'/');
path.push_str(&object.name());
path
} else {
let level = object.depth();
if level == 0 {
// _level0 does not append its name in slash syntax.
WString::new()
} else {
// Other levels do append their name.
WString::from_utf8_owned(format!("_level{level}"))
}
}
}
if self.avm1_parent().is_some() {
build_slash_path((*self).into())
} else {
// _target of _level0 should just be '/'.
WString::from_unit(b'/'.into())
}
}
fn clip_depth(&self) -> Depth {
self.base().clip_depth()
}
fn set_clip_depth(&self, gc_context: &Mutation<'gc>, depth: Depth) {
self.base_mut(gc_context).set_clip_depth(depth);
}
/// Retrieve the parent of this display object.
///
/// This version of the function merely exposes the display object parent,
/// without any further filtering.
fn parent(&self) -> Option<DisplayObject<'gc>> {
self.base().parent()
}
/// Set the parent of this display object.
fn set_parent(&self, context: &mut UpdateContext<'_, 'gc>, parent: Option<DisplayObject<'gc>>) {
let had_parent = self.parent().is_some();
self.base_mut(context.gc_context)
.set_parent_ignoring_orphan_list(parent);
let has_parent = self.parent().is_some();
let parent_removed = had_parent && !has_parent;
if parent_removed {
if let Some(int) = self.as_interactive() {
int.drop_focus(context);
}
self.on_parent_removed(context);
}
}
/// This method is called when the parent is removed.
/// It may be overwritten to inject some implementation-specific behavior.
fn on_parent_removed(&self, _context: &mut UpdateContext<'_, 'gc>) {}
/// Retrieve the parent of this display object.
///
/// This version of the function implements the concept of parenthood as
/// seen in AVM1. Notably, it disallows access to the `Stage` and to
/// non-AVM1 DisplayObjects; for an unfiltered concept of parent,
/// use the `parent` method.
fn avm1_parent(&self) -> Option<DisplayObject<'gc>> {
self.parent()
.filter(|p| p.as_stage().is_none())
.filter(|p| !p.movie().is_action_script_3())
}
/// Retrieve the parent of this display object.
///
/// This version of the function implements the concept of parenthood as
/// seen in AVM2. Notably, it disallows access to non-container parents.
fn avm2_parent(&self) -> Option<DisplayObject<'gc>> {
self.parent().filter(|p| p.as_container().is_some())
}
fn next_avm1_clip(&self) -> Option<DisplayObject<'gc>> {
self.base().next_avm1_clip()
}
fn set_next_avm1_clip(&self, gc_context: &Mutation<'gc>, node: Option<DisplayObject<'gc>>) {
self.base_mut(gc_context).set_next_avm1_clip(node);
}
fn masker(&self) -> Option<DisplayObject<'gc>> {
self.base().masker()
}
fn set_masker(
&self,
gc_context: &Mutation<'gc>,
node: Option<DisplayObject<'gc>>,
remove_old_link: bool,
) {
if remove_old_link {
let old_masker = self.base().masker();
if let Some(old_masker) = old_masker {
old_masker.set_maskee(gc_context, None, false);
}
if let Some(parent) = self.parent() {
// Masks are natively handled by cacheAsBitmap - don't invalidate self, only parents
parent.invalidate_cached_bitmap(gc_context);
}
}
self.base_mut(gc_context).set_masker(node);
}
fn maskee(&self) -> Option<DisplayObject<'gc>> {
self.base().maskee()
}
fn set_maskee(
&self,
gc_context: &Mutation<'gc>,
node: Option<DisplayObject<'gc>>,
remove_old_link: bool,
) {
if remove_old_link {
let old_maskee = self.base().maskee();
if let Some(old_maskee) = old_maskee {
old_maskee.set_masker(gc_context, None, false);
}
self.invalidate_cached_bitmap(gc_context);
}
self.base_mut(gc_context).set_maskee(node);
}
fn scroll_rect(&self) -> Option<Rectangle<Twips>> {
self.base().scroll_rect.clone()
}
fn next_scroll_rect(&self) -> Rectangle<Twips> {
self.base().next_scroll_rect.clone()
}
fn set_next_scroll_rect(&self, gc_context: &Mutation<'gc>, rectangle: Rectangle<Twips>) {
self.base_mut(gc_context).next_scroll_rect = rectangle;
// Scroll rect is natively handled by cacheAsBitmap - don't invalidate self, only parents
if let Some(parent) = self.parent() {
parent.invalidate_cached_bitmap(gc_context);
}
}
fn scaling_grid(&self) -> Rectangle<Twips> {
self.base().scaling_grid.clone()
}
fn set_scaling_grid(&self, gc_context: &Mutation<'gc>, rect: Rectangle<Twips>) {
self.base_mut(gc_context).scaling_grid = rect;
}
/// Whether this object has been removed. Only applies to AVM1.
fn avm1_removed(&self) -> bool {
self.base().avm1_removed()
}
// Sets whether this object has been removed. Only applies to AVM1
fn set_avm1_removed(&self, gc_context: &Mutation<'gc>, value: bool) {
self.base_mut(gc_context).set_avm1_removed(value)
}
/// Is this object waiting to be removed on the start of the next frame
fn avm1_pending_removal(&self) -> bool {
self.base().avm1_pending_removal()
}
fn set_avm1_pending_removal(&self, gc_context: &Mutation<'gc>, value: bool) {
self.base_mut(gc_context).set_avm1_pending_removal(value)
}
/// Whether this display object is visible.
/// Invisible objects are not rendered, but otherwise continue to exist normally.
/// Returned by the `_visible`/`visible` ActionScript properties.
fn visible(&self) -> bool {
self.base().visible()
}
/// Sets whether this display object will be visible.
/// Invisible objects are not rendered, but otherwise continue to exist normally.
/// Returned by the `_visible`/`visible` ActionScript properties.
fn set_visible(&self, context: &mut UpdateContext<'_, 'gc>, value: bool) {
if self.base_mut(context.gc()).set_visible(value) {
if let Some(parent) = self.parent() {
// We don't need to invalidate ourselves, we're just toggling if the bitmap is rendered.
parent.invalidate_cached_bitmap(context.gc());
}
}
if !value {
if let Some(int) = self.as_interactive() {
// The focus is dropped when it's made invisible.
int.drop_focus(context);
}
}
}
fn meta_data(&self) -> Option<Avm2Object<'gc>> {
self.base().meta_data()
}
fn set_meta_data(&self, gc_context: &Mutation<'gc>, value: Avm2Object<'gc>) {
self.base_mut(gc_context).set_meta_data(value);
}
/// The blend mode used when rendering this display object.
/// Values other than the default `BlendMode::Normal` implicitly cause cache-as-bitmap behavior.
fn blend_mode(&self) -> ExtendedBlendMode {
self.base().blend_mode()
}
/// Sets the blend mode used when rendering this display object.
/// Values other than the default `BlendMode::Normal` implicitly cause cache-as-bitmap behavior.
fn set_blend_mode(&self, gc_context: &Mutation<'gc>, value: ExtendedBlendMode) {
if self.base_mut(gc_context).set_blend_mode(value) {
if let Some(parent) = self.parent() {
// We don't need to invalidate ourselves, we're just toggling how the bitmap is rendered.
// Note that Flash does not always invalidate on changing the blend mode;
// but that's a bug we don't need to copy :)
parent.invalidate_cached_bitmap(gc_context);
}
}
}
fn blend_shader(&self) -> Option<PixelBenderShaderHandle> {
self.base().blend_shader()
}
fn set_blend_shader(&self, gc_context: &Mutation<'gc>, value: Option<PixelBenderShaderHandle>) {
self.base_mut(gc_context).set_blend_shader(value);
self.set_blend_mode(gc_context, ExtendedBlendMode::Shader);
}
/// The opaque background color of this display object.
fn opaque_background(&self) -> Option<Color> {
self.base().opaque_background()
}
/// Sets the opaque background color of this display object.
/// The bounding box of the display object will be filled with the given color. This also
/// triggers cache-as-bitmap behavior. Only solid backgrounds are supported; the alpha channel
/// is ignored.
fn set_opaque_background(&self, gc_context: &Mutation<'gc>, value: Option<Color>) {
if self.base_mut(gc_context).set_opaque_background(value) {
self.invalidate_cached_bitmap(gc_context);
}
}
/// Whether this display object represents the root of loaded content.
fn is_root(&self) -> bool {
self.base().is_root()
}
/// Sets whether this display object represents the root of loaded content.
fn set_is_root(&self, gc_context: &Mutation<'gc>, value: bool) {
self.base_mut(gc_context).set_is_root(value);
}
/// The sound transform for sounds played inside this display object.
fn set_sound_transform(
&self,
context: &mut UpdateContext<'_, 'gc>,
sound_transform: SoundTransform,
) {
self.base_mut(context.gc_context)
.set_sound_transform(sound_transform);
context.set_sound_transforms_dirty();
}
/// Whether this display object is used as the _root of itself and its children.
/// Returned by the `_lockroot` ActionScript property.
fn lock_root(&self) -> bool {
self.base().lock_root()
}
/// Sets whether this display object is used as the _root of itself and its children.
/// Returned by the `_lockroot` ActionScript property.
fn set_lock_root(&self, gc_context: &Mutation<'gc>, value: bool) {
self.base_mut(gc_context).set_lock_root(value);
}
/// Whether this display object has been transformed by ActionScript.
/// When this flag is set, changes from SWF `PlaceObject` tags are ignored.
fn transformed_by_script(&self) -> bool {
self.base().transformed_by_script()
}
/// Sets whether this display object has been transformed by ActionScript.
/// When this flag is set, changes from SWF `PlaceObject` tags are ignored.
fn set_transformed_by_script(&self, gc_context: &Mutation<'gc>, value: bool) {
self.base_mut(gc_context).set_transformed_by_script(value)
}
/// Whether this display object prefers to be cached into a bitmap rendering.
/// This is the PlaceObject `cacheAsBitmap` flag - and may be overridden if filters are applied.
/// Consider `is_bitmap_cached` for if a bitmap cache is actually in use.
fn is_bitmap_cached_preference(&self) -> bool {
self.base().is_bitmap_cached_preference()
}
/// Whether this display object is using a bitmap cache, whether by preference or necessity.
fn is_bitmap_cached(&self) -> bool {
self.base().cache.is_some()
}
/// Explicitly sets the preference of this display object to be cached into a bitmap rendering.
/// Note that the object will still be bitmap cached if a filter is active.
fn set_bitmap_cached_preference(&self, gc_context: &Mutation<'gc>, value: bool) {
self.base_mut(gc_context)
.set_bitmap_cached_preference(value)
}
/// Whether this display object has a scroll rectangle applied.
fn has_scroll_rect(&self) -> bool {
self.base().has_scroll_rect()
}
/// Sets whether this display object has a scroll rectangle applied.
fn set_has_scroll_rect(&self, gc_context: &Mutation<'gc>, value: bool) {
self.base_mut(gc_context).set_has_scroll_rect(value)
}
/// Whether this display object has been created by ActionScript 3.
/// When this flag is set, changes from SWF `RemoveObject` tags are
/// ignored.
fn placed_by_script(&self) -> bool {
self.base().placed_by_script()
}
/// Sets whether this display object has been created by ActionScript 3.
/// When this flag is set, changes from SWF `RemoveObject` tags are
/// ignored.
fn set_placed_by_script(&self, gc_context: &Mutation<'gc>, value: bool) {
self.base_mut(gc_context).set_placed_by_script(value)
}
/// Whether this display object has been instantiated by the timeline.
/// When this flag is set, attempts to change the object's name from AVM2
/// throw an exception.
fn instantiated_by_timeline(&self) -> bool {
self.base().instantiated_by_timeline()
}
/// Sets whether this display object has been instantiated by the timeline.
/// When this flag is set, attempts to change the object's name from AVM2
/// throw an exception.
fn set_instantiated_by_timeline(&self, gc_context: &Mutation<'gc>, value: bool) {
self.base_mut(gc_context)
.set_instantiated_by_timeline(value);
}
/// Whether this display object was placed by a SWF tag with an explicit
/// name.
///
/// When this flag is set, the object will attempt to set a dynamic property
/// on the parent with the same name as itself.
fn has_explicit_name(&self) -> bool {
self.base().has_explicit_name()
}
/// Sets whether this display object was placed by a SWF tag with an
/// explicit name.
///
/// When this flag is set, the object will attempt to set a dynamic property
/// on the parent with the same name as itself.
fn set_has_explicit_name(&self, gc_context: &Mutation<'gc>, value: bool) {
self.base_mut(gc_context).set_has_explicit_name(value);
}
fn state(&self) -> Option<ButtonState> {
None
}
fn set_state(self, _context: &mut UpdateContext<'_, 'gc>, _state: ButtonState) {}
/// Run any start-of-frame actions for this display object.
///
/// When fired on `Stage`, this also emits the AVM2 `enterFrame` broadcast.
fn enter_frame(&self, _context: &mut UpdateContext<'_, 'gc>) {}
/// Construct all display objects that the timeline indicates should exist
/// this frame, and their children.
///
/// This function should ensure the following, from the point of view of
/// downstream VMs:
///
/// 1. That the object itself has been allocated, if not constructed
/// 2. That newly created children have been instantiated and are present
/// as properties on the class
fn construct_frame(&self, _context: &mut UpdateContext<'_, 'gc>) {}
/// To be called when an AVM2 display object has finished being constructed.
///
/// This function must be called once and ONLY once, after the object's
/// AVM2 side has been constructed. Typically, this is in construct_frame,
/// unless your object needs to construct itself earlier or later. When
/// this function is called on the child, it will fire its add events and,
/// if possible, set a named property on the parent matching the name of
/// the object.
///
/// This still needs to be called for objects placed by AVM2, since we
/// need to stop the underlying MovieClip if the constructed class
/// does not extend MovieClip.
///
/// Since we construct AVM2 display objects after they are allocated and
/// placed on the render list, these steps have to be done by the child
/// object to signal to its parent that it was added.
#[inline(never)]
fn on_construction_complete(&self, context: &mut UpdateContext<'_, 'gc>) {
let placed_by_script = self.placed_by_script();
self.fire_added_events(context);
// Check `self.placed_by_script()` before we fire events, since those
// events might `placed_by_script`
if !placed_by_script {
self.set_on_parent_field(context);
}
if let Some(movie) = self.as_movie_clip() {
let obj = movie
.object2()
.as_object()
.expect("MovieClip object should have been constructed");
let movieclip_class = context.avm2.classes().movieclip.inner_class_definition();
// It's possible to have a DefineSprite tag with multiple frames, but have
// the corresponding `SymbolClass` *not* extend `MovieClip` (e.g. extending `Sprite` directly.)
// When this occurs, Flash Player will run the first frame, and immediately stop.
// However, Flash Player runs frames for the root movie clip, even if it doesn't extend `MovieClip`.
if !obj.is_of_type(movieclip_class) && !movie.is_root() {
movie.stop(context);
}
movie.set_initialized(context.gc_context);
}
}
fn fire_added_events(&self, context: &mut UpdateContext<'_, 'gc>) {
if !self.placed_by_script() {
// Since we construct AVM2 display objects after they are
// allocated and placed on the render list, we have to emit all
// events after this point.
//
// Children added to buttons by the timeline do not emit events.
if self.parent().and_then(|p| p.as_avm2_button()).is_none() {
dispatch_added_event_only((*self).into(), context);
if self.avm2_stage(context).is_some() {
dispatch_added_to_stage_event_only((*self).into(), context);
}
}
}
}
fn set_on_parent_field(&self, context: &mut UpdateContext<'_, 'gc>) {
//TODO: Don't report missing property errors.
//TODO: Don't attempt to set properties if object was placed without a name.
if self.has_explicit_name() {
if let Some(Avm2Value::Object(p)) = self.parent().map(|p| p.object2()) {
if let Avm2Value::Object(c) = self.object2() {
let domain = context
.library
.library_for_movie(self.movie())
.unwrap()
.avm2_domain();
let mut activation = Avm2Activation::from_domain(context.reborrow(), domain);
let name =
Avm2Multiname::new(activation.avm2().find_public_namespace(), self.name());
if let Err(e) = p.init_property(&name, c.into(), &mut activation) {
tracing::error!(
"Got error when setting AVM2 child named \"{}\": {}",
&self.name(),
e
);
}
}
}
}
}
/// Execute all other timeline actions on this object.
fn run_frame_avm1(&self, _context: &mut UpdateContext<'_, 'gc>) {}
/// Emit a `frameConstructed` event on this DisplayObject and any children it
/// may have.
fn frame_constructed(&self, context: &mut UpdateContext<'_, 'gc>) {
let frame_constructed_evt =
Avm2EventObject::bare_default_event(context, "frameConstructed");
let dobject_constr = context.avm2.classes().display_object;
Avm2::broadcast_event(context, frame_constructed_evt, dobject_constr);
}
/// Run any frame scripts (if they exist and this object needs to run them).
fn run_frame_scripts(self, context: &mut UpdateContext<'_, 'gc>) {
if let Some(container) = self.as_container() {
for child in container.iter_render_list() {
child.run_frame_scripts(context);
}
}
}
/// Emit an `exitFrame` broadcast event.
fn exit_frame(&self, context: &mut UpdateContext<'_, 'gc>) {
let exit_frame_evt = Avm2EventObject::bare_default_event(context, "exitFrame");
let dobject_constr = context.avm2.classes().display_object;
Avm2::broadcast_event(context, exit_frame_evt, dobject_constr);
self.on_exit_frame(context);
}
fn on_exit_frame(&self, context: &mut UpdateContext<'_, 'gc>) {
if let Some(container) = self.as_container() {
for child in container.iter_render_list() {
child.on_exit_frame(context);
}
}
}
/// Called before the child is about to be rendered.
/// Note that this happens even if the child is invisible
/// (as long as the child is still on a render list)
fn pre_render(&self, context: &mut RenderContext<'_, 'gc>) {
let mut this = self.base_mut(context.gc_context);
this.clear_invalidate_flag();
this.scroll_rect = this
.has_scroll_rect()
.then(|| this.next_scroll_rect.clone());
}
fn render_self(&self, _context: &mut RenderContext<'_, 'gc>) {}
fn render(&self, context: &mut RenderContext<'_, 'gc>) {
render_base((*self).into(), context)
}
#[cfg(not(feature = "avm_debug"))]
fn display_render_tree(&self, _depth: usize) {}
#[cfg(feature = "avm_debug")]
fn display_render_tree(&self, depth: usize) {
let mut self_str = &*format!("{self:?}");
if let Some(end_char) = self_str.find(|c: char| !c.is_ascii_alphanumeric()) {
self_str = &self_str[..end_char];
}
let bounds = self.world_bounds();
let mut classname = "".to_string();
if let Some(o) = self.object2().as_object() {
classname = format!("{:?}", o.base().debug_class_name());
}
println!(
"{} rel({},{}) abs({},{}) {} {} {} id={} depth={}",
" ".repeat(depth),
self.x(),
self.y(),
bounds.x_min.to_pixels(),
bounds.y_min.to_pixels(),
classname,
self.name(),
self_str,
self.id(),
depth
);
if let Some(ctr) = self.as_container() {
ctr.recurse_render_tree(depth + 1);
}
}
fn avm1_unload(&self, context: &mut UpdateContext<'_, 'gc>) {
// Unload children.
if let Some(ctr) = self.as_container() {
for child in ctr.iter_render_list() {
child.avm1_unload(context);
}
}
if let Some(node) = self.maskee() {
node.set_masker(context.gc_context, None, true);
} else if let Some(node) = self.masker() {
node.set_maskee(context.gc_context, None, true);
}
// Unregister any text field variable bindings, and replace them on the unbound list.
if let Avm1Value::Object(object) = self.object() {
if let Some(stage_object) = object.as_stage_object() {
stage_object.unregister_text_field_bindings(context);
}
}
context
.audio_manager
.stop_sounds_with_display_object(context.audio, (*self).into());
self.set_avm1_removed(context.gc_context, true);
}
fn as_stage(&self) -> Option<Stage<'gc>> {
None
}
fn as_avm1_button(&self) -> Option<Avm1Button<'gc>> {
None
}
fn as_avm2_button(&self) -> Option<Avm2Button<'gc>> {
None
}
fn as_movie_clip(&self) -> Option<MovieClip<'gc>> {
None
}
fn as_edit_text(&self) -> Option<EditText<'gc>> {
None
}
fn as_morph_shape(&self) -> Option<MorphShape<'gc>> {
None
}
fn as_container(self) -> Option<DisplayObjectContainer<'gc>> {
None
}
fn as_video(self) -> Option<Video<'gc>> {
None
}
fn as_drawing(&self, _gc_context: &Mutation<'gc>) -> Option<RefMut<'_, Drawing>> {
None
}
fn as_bitmap(self) -> Option<Bitmap<'gc>> {
None
}
fn as_interactive(self) -> Option<InteractiveObject<'gc>> {
None
}
fn apply_place_object(
&self,
context: &mut UpdateContext<'_, 'gc>,
place_object: &swf::PlaceObject,
) {
// PlaceObject tags only apply if this object has not been dynamically moved by AS code.
if !self.transformed_by_script() {
if let Some(matrix) = place_object.matrix {
self.set_matrix(context.gc_context, matrix.into());
if let Some(parent) = self.parent() {
// Self-transform changes are automatically handled,
// we only want to inform ancestors to avoid unnecessary invalidations for tx/ty
parent.invalidate_cached_bitmap(context.gc_context);
}
}
if let Some(color_transform) = &place_object.color_transform {
self.set_color_transform(context.gc_context, *color_transform);
if let Some(parent) = self.parent() {
parent.invalidate_cached_bitmap(context.gc_context);
}
}
if let Some(ratio) = place_object.ratio {
if let Some(mut morph_shape) = self.as_morph_shape() {
morph_shape.set_ratio(context.gc_context, ratio);
} else if let Some(video) = self.as_video() {
video.seek(context, ratio.into());
}
}
if let Some(is_bitmap_cached) = place_object.is_bitmap_cached {
self.set_bitmap_cached_preference(context.gc_context, is_bitmap_cached);
}
if let Some(blend_mode) = place_object.blend_mode {
self.set_blend_mode(context.gc_context, blend_mode.into());
}
if self.swf_version() >= 11 {
if let Some(visible) = place_object.is_visible {
self.set_visible(context, visible);
}
if let Some(mut color) = place_object.background_color {
let color = if color.a > 0 {
// Force opaque background to have no transpranecy.
color.a = 255;
Some(color)
} else {
None
};
self.set_opaque_background(context.gc_context, color);
}
}
if let Some(filters) = &place_object.filters {
self.set_filters(
context.gc_context,
filters.iter().map(Filter::from).collect(),
);
}
// Purposely omitted properties:
// name, clip_depth, clip_actions
// These properties are only set on initial placement in `MovieClip::instantiate_child`
// and can not be modified by subsequent PlaceObject tags.
}
}
/// Called when this object should be replaced by a PlaceObject tag.
fn replace_with(&self, _context: &mut UpdateContext<'_, 'gc>, _id: CharacterId) {
// Noop for most symbols; only shapes can replace their innards with another Graphic.
}
fn object(&self) -> Avm1Value<'gc> {
Avm1Value::Undefined // TODO: Implement for every type and delete this fallback.
}
fn object2(&self) -> Avm2Value<'gc> {
Avm2Value::Undefined // TODO: See above. Also, unconstructed objects should return null.
}
fn set_object2(&self, _context: &mut UpdateContext<'_, 'gc>, _to: Avm2Object<'gc>) {}
/// Tests if a given stage position point intersects with the world bounds of this object.
fn hit_test_bounds(&self, point: Point<Twips>) -> bool {
self.world_bounds().contains(point)
}
/// Tests if a given object's world bounds intersects with the world bounds
/// of this object.
fn hit_test_object(&self, other: DisplayObject<'gc>) -> bool {
self.world_bounds().intersects(&other.world_bounds())
}
/// Tests if a given stage position point intersects within this object, considering the art.
fn hit_test_shape(
&self,
_context: &mut UpdateContext<'_, 'gc>,
point: Point<Twips>,
options: HitTestOptions,
) -> bool {
// Default to using bounding box.
(!options.contains(HitTestOptions::SKIP_INVISIBLE) || self.visible())
&& self.hit_test_bounds(point)
}
fn post_instantiation(
&self,
context: &mut UpdateContext<'_, 'gc>,
_init_object: Option<Avm1Object<'gc>>,
_instantiated_by: Instantiator,
run_frame: bool,
) {
if run_frame && !self.movie().is_action_script_3() {
self.run_frame_avm1(context);
}
}
/// Return the version of the SWF that created this movie clip.
fn swf_version(&self) -> u8 {
self.movie().version()
}
/// Return the SWF that defines this display object.
fn movie(&self) -> Arc<SwfMovie>;
fn loader_info(&self) -> Option<Avm2Object<'gc>> {
None
}
fn instantiate(&self, gc_context: &Mutation<'gc>) -> DisplayObject<'gc>;
fn as_ptr(&self) -> *const DisplayObjectPtr;
/// Whether this object can be used as a mask.
/// If this returns false and this object is used as a mask, the mask will not be applied.
/// This is used by movie clips to disable the mask when there are no children, for example.
fn allow_as_mask(&self) -> bool {
true
}
/// Obtain the top-most non-Stage parent of the display tree hierarchy.
///
/// This function implements the AVM1 concept of root clips. For the AVM2
/// version, see `avm2_root`.
fn avm1_root(&self) -> DisplayObject<'gc> {
let mut root = (*self).into();
loop {
if root.lock_root() {
break;
}
if let Some(parent) = root.avm1_parent() {
if !parent.movie().is_action_script_3() {
root = parent;
} else {
// We've traversed upwards into a loader AVM2 movie, so break.
break;
}
} else {
break;
}
}
root
}
/// `avm1_root`, but disregards _lockroot
fn avm1_root_no_lock(&self) -> DisplayObject<'gc> {
let mut root = (*self).into();
while let Some(parent) = root.avm1_parent() {
if !parent.movie().is_action_script_3() {
root = parent;
} else {
// We've traversed upwards into a loader AVM2 movie, so break.
break;
}
}
root
}
/// Obtain the top-most Stage or LoaderDisplay object of the display tree hierarchy, for use in mixed AVM.
fn avm1_stage(&self) -> DisplayObject<'gc> {
let mut root = (*self).into();
loop {
if let Some(parent) = root.parent() {
if matches!(
parent,
DisplayObject::LoaderDisplay(_) | DisplayObject::Stage(_)
) {
return parent;
}
root = parent;
} else {
return root;
}
}
}
/// Obtain the top-most non-Stage parent of the display tree hierarchy, if
/// a suitable object exists.
///
/// This function implements the AVM2 concept of root clips. For the AVM1
/// version, see `avm1_root`.
fn avm2_root(&self) -> Option<DisplayObject<'gc>> {
let mut parent = Some((*self).into());
while let Some(p) = parent {
if p.is_root() {
return parent;
}
if let Some(p_parent) = p.parent() {
if !p_parent.movie().is_action_script_3() {
// We've traversed upwards into a loader AVM1 movie, so return the current parent.
return parent;
}
}
parent = p.parent();
}
None
}
/// Obtain the root of the display tree hierarchy, if a suitable object
/// exists.
///
/// This implements the AVM2 concept of `stage`. Notably, it deliberately
/// will fail to locate the current player's stage for objects that are not
/// rooted to the DisplayObject hierarchy correctly. If you just want to
/// access the current player's stage, grab it from the context.
fn avm2_stage(&self, _context: &UpdateContext<'_, 'gc>) -> Option<DisplayObject<'gc>> {
let mut parent = Some((*self).into());
while let Some(p) = parent {
if p.as_stage().is_some() {
return parent;
}
parent = p.parent();
}
None
}
/// Determine if this display object is currently on the stage.
fn is_on_stage(self, context: &UpdateContext<'_, 'gc>) -> bool {
let mut ancestor = self.avm2_parent();
while let Some(parent) = ancestor {
if parent.avm2_parent().is_some() {
ancestor = parent.avm2_parent();
} else {
break;
}
}
let ancestor = ancestor.unwrap_or_else(|| self.into());
DisplayObject::ptr_eq(ancestor, context.stage.into())
}
/// Assigns a default instance name `instanceN` to this object.
fn set_default_instance_name(&self, context: &mut UpdateContext<'_, 'gc>) {
if self.base().name().is_none() {
let name = format!("instance{}", *context.instance_counter);
self.set_name(
context.gc_context,
AvmString::new_utf8(context.gc_context, name),
);
*context.instance_counter = context.instance_counter.wrapping_add(1);
}
}
/// Assigns a default root name to this object.
///
/// The default root names change based on the AVM configuration of the
/// clip; AVM2 clips get `rootN` while AVM1 clips get blank strings.
fn set_default_root_name(&self, context: &mut UpdateContext<'_, 'gc>) {
if self.movie().is_action_script_3() {
let name = AvmString::new_utf8(context.gc_context, format!("root{}", self.depth() + 1));
self.set_name(context.gc_context, name);
} else {
self.set_name(context.gc_context, Default::default());
}
}
fn bind_text_field_variables(&self, activation: &mut Activation<'_, 'gc>) {
// Check all unbound text fields to see if they apply to this object.
// TODO: Replace with `Vec::drain_filter` when stable.
let mut i = 0;
let mut len = activation.context.unbound_text_fields.len();
while i < len {
if activation.context.unbound_text_fields[i]
.try_bind_text_field_variable(activation, false)
{
activation.context.unbound_text_fields.swap_remove(i);
len -= 1;
} else {
i += 1;
}
}
}
/// Inform this object and its ancestors that it has visually changed and must be redrawn.
/// If this object or any ancestor is marked as cacheAsBitmap, it will invalidate that cache.
fn invalidate_cached_bitmap(&self, mc: &Mutation<'gc>) {
if self.base_mut(mc).invalidate_cached_bitmap() {
// Don't inform ancestors if we've already done so this frame
if let Some(parent) = self.parent() {
parent.invalidate_cached_bitmap(mc);
}
}
}
/// Retrieve a named property from the AVM1 object.
///
/// This is required as some boolean properties in AVM1 can in fact hold any value.
fn get_avm1_boolean_property<F>(
self,
context: &mut UpdateContext<'_, 'gc>,
name: &'static str,
default: F,
) -> bool
where
F: FnOnce(&mut UpdateContext<'_, 'gc>) -> bool,
{
if let Avm1Value::Object(object) = self.object() {
let mut activation = Activation::from_nothing(
context.reborrow(),
Avm1ActivationIdentifier::root("[AVM1 Boolean Property]"),
self.avm1_root(),
);
if let Ok(value) = object.get(name, &mut activation) {
match value {
Avm1Value::Undefined => default(&mut activation.context),
_ => value.as_bool(activation.swf_version()),
}
} else {
default(&mut activation.context)
}
} else {
false
}
}
fn set_avm1_property(
self,
context: &mut UpdateContext<'_, 'gc>,
name: &'static str,
value: Avm1Value<'gc>,
) {
if let Avm1Value::Object(object) = self.object() {
let mut activation = Activation::from_nothing(
context.reborrow(),
Avm1ActivationIdentifier::root("[AVM1 Property Set]"),
self.avm1_root(),
);
let _ = object.set(name, value, &mut activation);
}
}
}
pub enum DisplayObjectPtr {}
impl<'gc> DisplayObject<'gc> {
pub fn ptr_eq(a: DisplayObject<'gc>, b: DisplayObject<'gc>) -> bool {
a.as_ptr() == b.as_ptr()
}
pub fn option_ptr_eq(a: Option<DisplayObject<'gc>>, b: Option<DisplayObject<'gc>>) -> bool {
a.map(|o| o.as_ptr()) == b.map(|o| o.as_ptr())
}
pub fn downgrade(self) -> DisplayObjectWeak<'gc> {
match self {
DisplayObject::MovieClip(mc) => DisplayObjectWeak::MovieClip(mc.downgrade()),
DisplayObject::LoaderDisplay(l) => DisplayObjectWeak::LoaderDisplay(l.downgrade()),
DisplayObject::Bitmap(b) => DisplayObjectWeak::Bitmap(b.downgrade()),
_ => panic!("Downgrade not yet implemented for {:?}", self),
}
}
}
bitflags! {
/// Bit flags used by `DisplayObject`.
#[derive(Clone, Copy)]
struct DisplayObjectFlags: u16 {
/// Whether this object has been removed from the display list.
/// Necessary in AVM1 to throw away queued actions from removed movie clips.
const AVM1_REMOVED = 1 << 0;
/// If this object is visible (`_visible` property).
const VISIBLE = 1 << 1;
/// Whether the `_xscale`, `_yscale` and `_rotation` of the object have been calculated and cached.
const SCALE_ROTATION_CACHED = 1 << 2;
/// Whether this object has been transformed by ActionScript.
/// When this flag is set, changes from SWF `PlaceObject` tags are ignored.
const TRANSFORMED_BY_SCRIPT = 1 << 3;
/// Whether this object has been placed in a container by ActionScript 3.
/// When this flag is set, changes from SWF `RemoveObject` tags are ignored.
const PLACED_BY_SCRIPT = 1 << 4;
/// Whether this object has been instantiated by a SWF tag.
/// When this flag is set, attempts to change the object's name from AVM2 throw an exception.
const INSTANTIATED_BY_TIMELINE = 1 << 5;
/// Whether this object is a "root", the top-most display object of a loaded SWF or Bitmap.
/// Used by `MovieClip.getBytesLoaded` in AVM1 and `DisplayObject.root` in AVM2.
const IS_ROOT = 1 << 6;
/// Whether this object has `_lockroot` set to true, in which case
/// it becomes the _root of itself and of any children
const LOCK_ROOT = 1 << 7;
/// Whether this object will be cached to bitmap.
const CACHE_AS_BITMAP = 1 << 8;
/// Whether this object has a scroll rectangle applied.
const HAS_SCROLL_RECT = 1 << 9;
/// Whether this object has an explicit name.
const HAS_EXPLICIT_NAME = 1 << 10;
/// Flag set when we should skip running our next 'enterFrame'
/// for ourself and our children.
/// This is set for objects constructed from ActionScript,
/// which are observed to lag behind objects placed by the timeline
/// (even if they are both placed in the same frame)
const SKIP_NEXT_ENTER_FRAME = 1 << 11;
/// If this object has already had `invalidate_cached_bitmap` called this frame
const CACHE_INVALIDATED = 1 << 12;
/// If this AVM1 object is pending removal (will be removed on the next frame).
const AVM1_PENDING_REMOVAL = 1 << 13;
}
}
bitflags! {
/// Defines how hit testing should be performed.
/// Used for mouse picking and ActionScript's hitTestClip functions.
#[derive(Clone, Copy)]
pub struct HitTestOptions: u8 {
/// Ignore objects used as masks (setMask / clipDepth).
const SKIP_MASK = 1 << 0;
/// Ignore objects with the ActionScript's visibility flag turned off.
const SKIP_INVISIBLE = 1 << 1;
/// The options used for `hitTest` calls in ActionScript.
const AVM_HIT_TEST = Self::SKIP_MASK.bits();
/// The options used for mouse picking, such as clicking on buttons.
const MOUSE_PICK = Self::SKIP_MASK.bits() | Self::SKIP_INVISIBLE.bits();
}
}
/// Represents the sound transform of sounds played inside a Flash MovieClip.
/// Every value is a percentage (0-100), but out of range values are allowed.
/// In AVM1, this is returned by `Sound.getTransform`.
/// In AVM2, this is returned by `Sprite.soundTransform`.
#[derive(Debug, PartialEq, Eq, Clone)]
pub struct SoundTransform {
pub volume: i32,
pub left_to_left: i32,
pub left_to_right: i32,
pub right_to_left: i32,
pub right_to_right: i32,
}
impl SoundTransform {
pub const MAX_VOLUME: i32 = 100;
/// Applies another SoundTransform on top of this SoundTransform.
pub fn concat(&mut self, other: &SoundTransform) {
const MAX_VOLUME: i64 = SoundTransform::MAX_VOLUME as i64;
// It seems like Flash masks the results below to 30-bit integers:
// * Negative values are equivalent to their absolute value (their sign bit is unset).
// * Specifically, 0x40000000, -0x40000000 and -0x80000000 are equivalent to zero.
const MASK: i32 = (1 << 30) - 1;
self.volume = (i64::from(self.volume) * i64::from(other.volume) / MAX_VOLUME) as i32 & MASK;
// This is a 2x2 matrix multiply between the transforms.
// Done with integer math to match Flash behavior.
let ll0: i64 = self.left_to_left.into();
let lr0: i64 = self.left_to_right.into();
let rl0: i64 = self.right_to_left.into();
let rr0: i64 = self.right_to_right.into();
let ll1: i64 = other.left_to_left.into();
let lr1: i64 = other.left_to_right.into();
let rl1: i64 = other.right_to_left.into();
let rr1: i64 = other.right_to_right.into();
self.left_to_left = ((ll0 * ll1 + rl0 * lr1) / MAX_VOLUME) as i32 & MASK;
self.left_to_right = ((lr0 * ll1 + rr0 * lr1) / MAX_VOLUME) as i32 & MASK;
self.right_to_left = ((ll0 * rl1 + rl0 * rr1) / MAX_VOLUME) as i32 & MASK;
self.right_to_right = ((lr0 * rl1 + rr0 * rr1) / MAX_VOLUME) as i32 & MASK;
}
/// Returns the pan of this transform.
/// -100 is full left and 100 is full right.
/// This matches the behavior of AVM1 `Sound.getPan()`
pub fn pan(&self) -> i32 {
// It's not clear why Flash has the weird `abs` behavior, but this
// matches the values that Flash returns (see `sound` regression test).
if self.left_to_left != Self::MAX_VOLUME {
Self::MAX_VOLUME - self.left_to_left.abs()
} else {
self.right_to_right.abs() - Self::MAX_VOLUME
}
}
/// Sets this transform of this pan.
/// -100 is full left and 100 is full right.
/// This matches the behavior of AVM1 `Sound.setPan()`.
pub fn set_pan(&mut self, pan: i32) {
if pan >= 0 {
self.left_to_left = Self::MAX_VOLUME - pan;
self.right_to_right = Self::MAX_VOLUME;
} else {
self.left_to_left = Self::MAX_VOLUME;
self.right_to_right = Self::MAX_VOLUME + pan;
}
self.left_to_right = 0;
self.right_to_left = 0;
}
pub fn from_avm2_object<'gc>(
activation: &mut Avm2Activation<'_, 'gc>,
as3_st: Avm2Object<'gc>,
) -> Result<Self, Avm2Error<'gc>> {
Ok(SoundTransform {
left_to_left: (as3_st
.get_public_property("leftToLeft", activation)?
.coerce_to_number(activation)?
* 100.0) as i32,
left_to_right: (as3_st
.get_public_property("leftToRight", activation)?
.coerce_to_number(activation)?
* 100.0) as i32,
right_to_left: (as3_st
.get_public_property("rightToLeft", activation)?
.coerce_to_number(activation)?
* 100.0) as i32,
right_to_right: (as3_st
.get_public_property("rightToRight", activation)?
.coerce_to_number(activation)?
* 100.0) as i32,
volume: (as3_st
.get_public_property("volume", activation)?
.coerce_to_number(activation)?
* 100.0) as i32,
})
}
pub fn into_avm2_object<'gc>(
self,
activation: &mut Avm2Activation<'_, 'gc>,
) -> Result<Avm2Object<'gc>, Avm2Error<'gc>> {
let as3_st = activation
.avm2()
.classes()
.soundtransform
.construct(activation, &[])?;
as3_st.set_public_property(
"leftToLeft",
(self.left_to_left as f64 / 100.0).into(),
activation,
)?;
as3_st.set_public_property(
"leftToRight",
(self.left_to_right as f64 / 100.0).into(),
activation,
)?;
as3_st.set_public_property(
"rightToLeft",
(self.right_to_left as f64 / 100.0).into(),
activation,
)?;
as3_st.set_public_property(
"rightToRight",
(self.right_to_right as f64 / 100.0).into(),
activation,
)?;
as3_st.set_public_property("volume", (self.volume as f64 / 100.0).into(), activation)?;
Ok(as3_st)
}
}
impl Default for SoundTransform {
fn default() -> Self {
Self {
volume: 100,
left_to_left: 100,
left_to_right: 0,
right_to_left: 0,
right_to_right: 100,
}
}
}
/// A version of `DisplayObject` that holds weak pointers.
/// Currently, this is only used by orphan handling, so we only
/// need two variants. If other use cases arise, feel free
/// to add more variants.
#[derive(Copy, Clone, Collect)]
#[collect(no_drop)]
pub enum DisplayObjectWeak<'gc> {
MovieClip(MovieClipWeak<'gc>),
LoaderDisplay(LoaderDisplayWeak<'gc>),
Bitmap(BitmapWeak<'gc>),
}
impl<'gc> DisplayObjectWeak<'gc> {
pub fn as_ptr(&self) -> *const DisplayObjectPtr {
match self {
DisplayObjectWeak::MovieClip(mc) => mc.as_ptr(),
DisplayObjectWeak::LoaderDisplay(ld) => ld.as_ptr(),
DisplayObjectWeak::Bitmap(b) => b.as_ptr(),
}
}
pub fn upgrade(&self, mc: &Mutation<'gc>) -> Option<DisplayObject<'gc>> {
match self {
DisplayObjectWeak::MovieClip(movie) => movie.upgrade(mc).map(|m| m.into()),
DisplayObjectWeak::LoaderDisplay(ld) => ld.upgrade(mc).map(|ld| ld.into()),
DisplayObjectWeak::Bitmap(b) => b.upgrade(mc).map(|ld| ld.into()),
}
}
}
Reference in New Issue View Git Blame Copy Permalink