1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
//! This module provides capabilities for managing a cache of rendered glyphs in
//! GPU memory, with the goal of minimisng the size and frequency of glyph
//! uploads to GPU memory from the CPU.
//!
//! This module is optional, and not compiled by default. To use it enable the
//! `gpu_cache` feature in your Cargo.toml.
//!
//! Typical applications that render directly with hardware graphics APIs (e.g.
//! games) need text rendering. There is not yet a performant solution for high
//! quality text rendering directly on the GPU that isn't experimental research
//! work. Quality is often critical for legibility, so many applications use
//! text or individual characters that have been rendered on the CPU. This is
//! done either ahead-of-time, giving a fixed set of fonts, characters, and
//! sizes that can be used at runtime, or dynamically as text is required. This
//! latter scenario is more flexible and the focus of this module.
//!
//! To minimise the CPU load and texture upload bandwidth saturation, recently
//! used glyphs should be cached on the GPU for use by future frames. This
//! module provides a mechanism for maintaining such a cache in the form of a
//! single packed 2D GPU texture. When a rendered glyph is requested, it is
//! either retrieved from its location in the texture if it is present or room
//! is made in the cache (if necessary), the CPU renders the glyph then it is
//! uploaded into a gap in the texture to be available for GPU rendering. This
//! cache uses a Least Recently Used (LRU) cache eviction scheme - glyphs in the
//! cache that have not been used recently are as a rule of thumb not likely to
//! be used again soon, so they are the best candidates for eviction to make
//! room for required glyphs.
//!
//! The API for the cache does not assume a particular graphics API. The
//! intended usage is to queue up glyphs that need to be present for the current
//! frame using `Cache::queue_glyph`, update the cache to ensure that the queued
//! glyphs are present using `Cache::cache_queued` (providing a function for
//! uploading pixel data), then when it's time to render call `Cache::rect_for`
//! to get the UV coordinates in the cache texture for each glyph. For a
//! concrete use case see the `gpu_cache` example.
//!
//! Cache dimensions are immutable. If you need to change the dimensions of the
//! cache texture (e.g. due to high cache pressure), rebuild a new `Cache`.
//! Either from scratch or with `CacheBuilder::rebuild`.
//!
//! # Example
//!
//! ```
//! # use rusttype::{Font, gpu_cache::Cache, point, Scale};
//! # use std::error::Error;
//! # fn example() -> Result<(), Box<dyn Error>> {
//! # let font_data: &[u8] = include_bytes!("../dev/fonts/dejavu/DejaVuSansMono.ttf");
//! # let font: Font<'static> = Font::from_bytes(font_data)?;
//! # let glyph = font.glyph('a').scaled(Scale::uniform(25.0)).positioned(point(0.0, 0.0));
//! # let glyph2 = glyph.clone();
//! # fn update_gpu_texture(_: rusttype::Rect<u32>, _: &[u8]) {};
//! // Build a default Cache.
//! let mut cache = Cache::builder().build();
//!
//! // Queue some positioned glyphs needed for the next frame.
//! cache.queue_glyph(0, glyph);
//!
//! // Cache all queued glyphs somewhere in the cache texture.
//! // If new glyph data has been drawn the closure is called to upload
//! // the pixel data to GPU memory.
//! cache.cache_queued(|region, data| update_gpu_texture(region, data))?;
//!
//! # let glyph = glyph2;
//! // Lookup a positioned glyph's texture location
//! if let Ok(Some((uv_rect, screen_rect))) = cache.rect_for(0, &glyph) {
//! // Generate vertex data, etc
//! }
//! # Ok(())
//! # }
//! ```
use crate::{point, vector, GlyphId, Point, PositionedGlyph, Rect, Vector};
use linked_hash_map::LinkedHashMap;
use rustc_hash::{FxHashMap, FxHasher};
use std::collections::{HashMap, HashSet};
use std::error;
use std::fmt;
use std::hash::BuildHasherDefault;
type FxBuildHasher = BuildHasherDefault<FxHasher>;
/// Texture coordinates (floating point) of the quad for a glyph in the cache,
/// as well as the pixel-space (integer) coordinates that this region should be
/// drawn at.
pub type TextureCoords = (Rect<f32>, Rect<i32>);
type FontId = usize;
/// Indicates where a glyph texture is stored in the cache
/// (row position, glyph index in row)
type TextureRowGlyphIndex = (u32, u32);
/// Texture lookup key that uses scale & offset as integers attained
/// by dividing by the relevant tolerance.
#[derive(Copy, Clone, Debug, PartialEq, Eq, Hash)]
struct LossyGlyphInfo {
font_id: FontId,
glyph_id: GlyphId,
/// x & y scales divided by `scale_tolerance` & rounded
scale_over_tolerance: (u32, u32),
/// Normalised subpixel positions divided by `position_tolerance` & rounded
///
/// `u16` is enough as subpixel position `[-0.5, 0.5]` converted to `[0, 1]`
/// divided by the min `position_tolerance` (`0.001`) is small.
offset_over_tolerance: (u16, u16),
}
#[derive(Debug, Clone, PartialEq, Eq)]
struct ByteArray2d {
inner_array: Vec<u8>,
row: usize,
col: usize,
}
impl ByteArray2d {
#[inline]
pub fn zeros(row: usize, col: usize) -> Self {
ByteArray2d {
inner_array: vec![0; row * col],
row,
col,
}
}
#[inline]
fn as_slice(&self) -> &[u8] {
self.inner_array.as_slice()
}
#[inline]
fn get_vec_index(&self, row: usize, col: usize) -> usize {
debug_assert!(
row < self.row,
"row out of range: row={}, given={}",
self.row,
row
);
debug_assert!(
col < self.col,
"column out of range: col={}, given={}",
self.col,
col
);
row * self.col + col
}
}
impl std::ops::Index<(usize, usize)> for ByteArray2d {
type Output = u8;
#[inline]
fn index(&self, (row, col): (usize, usize)) -> &u8 {
&self.inner_array[self.get_vec_index(row, col)]
}
}
impl std::ops::IndexMut<(usize, usize)> for ByteArray2d {
#[inline]
fn index_mut(&mut self, (row, col): (usize, usize)) -> &mut u8 {
let vec_index = self.get_vec_index(row, col);
&mut self.inner_array[vec_index]
}
}
/// Row of pixel data
struct Row {
/// Row pixel height
height: u32,
/// Pixel width current in use by glyphs
width: u32,
glyphs: Vec<GlyphTexInfo>,
}
struct GlyphTexInfo {
glyph_info: LossyGlyphInfo,
/// Actual (lossless) normalised subpixel offset of rasterized glyph
offset: Vector<f32>,
tex_coords: Rect<u32>,
}
trait PaddingAware {
fn unpadded(self) -> Self;
}
impl PaddingAware for Rect<u32> {
/// A padded texture has 1 extra pixel on all sides
fn unpadded(mut self) -> Self {
self.min.x += 1;
self.min.y += 1;
self.max.x -= 1;
self.max.y -= 1;
self
}
}
/// An implementation of a dynamic GPU glyph cache. See the module documentation
/// for more information.
pub struct Cache<'font> {
scale_tolerance: f32,
position_tolerance: f32,
width: u32,
height: u32,
rows: LinkedHashMap<u32, Row, FxBuildHasher>,
/// Mapping of row gaps bottom -> top
space_start_for_end: FxHashMap<u32, u32>,
/// Mapping of row gaps top -> bottom
space_end_for_start: FxHashMap<u32, u32>,
queue: Vec<(FontId, PositionedGlyph<'font>)>,
all_glyphs: FxHashMap<LossyGlyphInfo, TextureRowGlyphIndex>,
pad_glyphs: bool,
align_4x4: bool,
multithread: bool,
}
/// Builder & rebuilder for `Cache`.
///
/// # Example
///
/// ```
/// use rusttype::gpu_cache::Cache;
///
/// // Create a cache with all default values set explicitly
/// // equivalent to `Cache::builder().build()`
/// let default_cache = Cache::builder()
/// .dimensions(256, 256)
/// .scale_tolerance(0.1)
/// .position_tolerance(0.1)
/// .pad_glyphs(true)
/// .align_4x4(false)
/// .multithread(true)
/// .build();
///
/// // Create a cache with all default values, except with a dimension of 1024x1024
/// let bigger_cache = Cache::builder().dimensions(1024, 1024).build();
/// ```
#[derive(Debug, Clone)]
pub struct CacheBuilder {
dimensions: (u32, u32),
scale_tolerance: f32,
position_tolerance: f32,
pad_glyphs: bool,
align_4x4: bool,
multithread: bool,
}
impl Default for CacheBuilder {
fn default() -> Self {
Self {
dimensions: (256, 256),
scale_tolerance: 0.1,
position_tolerance: 0.1,
pad_glyphs: true,
align_4x4: false,
multithread: true,
}
}
}
impl CacheBuilder {
/// `width` & `height` dimensions of the 2D texture that will hold the
/// cache contents on the GPU.
///
/// This must match the dimensions of the actual texture used, otherwise
/// `cache_queued` will try to cache into coordinates outside the bounds of
/// the texture.
///
/// # Example (set to default value)
///
/// ```
/// # use rusttype::gpu_cache::Cache;
/// let cache = Cache::builder().dimensions(256, 256).build();
/// ```
pub fn dimensions(mut self, width: u32, height: u32) -> Self {
self.dimensions = (width, height);
self
}
/// Specifies the tolerances (maximum allowed difference) for judging
/// whether an existing glyph in the cache is close enough to the
/// requested glyph in scale to be used in its place. Due to floating
/// point inaccuracies a min value of `0.001` is enforced.
///
/// Both `scale_tolerance` and `position_tolerance` are measured in pixels.
///
/// Tolerances produce even steps for scale and subpixel position. Only a
/// single glyph texture will be used within a single step. For example,
/// `scale_tolerance = 0.1` will have a step `9.95-10.05` so similar glyphs
/// with scale `9.98` & `10.04` will match.
///
/// A typical application will produce results with no perceptible
/// inaccuracies with `scale_tolerance` and `position_tolerance` set to
/// 0.1. Depending on the target DPI higher tolerance may be acceptable.
///
/// # Example (set to default value)
///
/// ```
/// # use rusttype::gpu_cache::Cache;
/// let cache = Cache::builder().scale_tolerance(0.1).build();
/// ```
pub fn scale_tolerance<V: Into<f32>>(mut self, scale_tolerance: V) -> Self {
self.scale_tolerance = scale_tolerance.into();
self
}
/// Specifies the tolerances (maximum allowed difference) for judging
/// whether an existing glyph in the cache is close enough to the requested
/// glyph in subpixel offset to be used in its place. Due to floating
/// point inaccuracies a min value of `0.001` is enforced.
///
/// Both `scale_tolerance` and `position_tolerance` are measured in pixels.
///
/// Tolerances produce even steps for scale and subpixel position. Only a
/// single glyph texture will be used within a single step. For example,
/// `scale_tolerance = 0.1` will have a step `9.95-10.05` so similar glyphs
/// with scale `9.98` & `10.04` will match.
///
/// Note that since `position_tolerance` is a tolerance of subpixel
/// offsets, setting it to 1.0 or higher is effectively a "don't care"
/// option.
///
/// A typical application will produce results with no perceptible
/// inaccuracies with `scale_tolerance` and `position_tolerance` set to
/// 0.1. Depending on the target DPI higher tolerance may be acceptable.
///
/// # Example (set to default value)
///
/// ```
/// # use rusttype::gpu_cache::Cache;
/// let cache = Cache::builder().position_tolerance(0.1).build();
/// ```
pub fn position_tolerance<V: Into<f32>>(mut self, position_tolerance: V) -> Self {
self.position_tolerance = position_tolerance.into();
self
}
/// Pack glyphs in texture with a padding of a single zero alpha pixel to
/// avoid bleeding from interpolated shader texture lookups near edges.
///
/// If glyphs are never transformed this may be set to `false` to slightly
/// improve the glyph packing.
///
/// # Example (set to default value)
///
/// ```
/// # use rusttype::gpu_cache::Cache;
/// let cache = Cache::builder().pad_glyphs(true).build();
/// ```
pub fn pad_glyphs(mut self, pad_glyphs: bool) -> Self {
self.pad_glyphs = pad_glyphs;
self
}
/// Align glyphs in texture to 4x4 texel boundaries.
///
/// If your backend requires texture updates to be aligned to 4x4 texel
/// boundaries (e.g. WebGL), this should be set to `true`.
///
/// # Example (set to default value)
///
/// ```
/// # use rusttype::gpu_cache::Cache;
/// let cache = Cache::builder().align_4x4(false).build();
/// ```
pub fn align_4x4(mut self, align_4x4: bool) -> Self {
self.align_4x4 = align_4x4;
self
}
/// When multiple CPU cores are available spread rasterization work across
/// all cores.
///
/// Significantly reduces worst case latency in multicore environments.
///
/// # Platform-specific behaviour
///
/// This option has no effect on wasm32.
///
/// # Example (set to default value)
///
/// ```
/// # use rusttype::gpu_cache::Cache;
/// let cache = Cache::builder().multithread(true).build();
/// ```
pub fn multithread(mut self, multithread: bool) -> Self {
self.multithread = multithread;
self
}
fn validated(self) -> Self {
assert!(self.scale_tolerance >= 0.0);
assert!(self.position_tolerance >= 0.0);
let scale_tolerance = self.scale_tolerance.max(0.001);
let position_tolerance = self.position_tolerance.max(0.001);
#[cfg(not(target_arch = "wasm32"))]
let multithread = self.multithread && num_cpus::get() > 1;
Self {
scale_tolerance,
position_tolerance,
#[cfg(not(target_arch = "wasm32"))]
multithread,
..self
}
}
/// Constructs a new cache. Note that this is just the CPU side of the
/// cache. The GPU texture is managed by the user.
///
/// # Panics
///
/// `scale_tolerance` or `position_tolerance` are less than or equal to
/// zero.
///
/// # Example
///
/// ```
/// # use rusttype::gpu_cache::Cache;
/// let cache = Cache::builder().build();
/// ```
pub fn build<'a>(self) -> Cache<'a> {
let CacheBuilder {
dimensions: (width, height),
scale_tolerance,
position_tolerance,
pad_glyphs,
align_4x4,
multithread,
} = self.validated();
Cache {
scale_tolerance,
position_tolerance,
width,
height,
rows: LinkedHashMap::default(),
space_start_for_end: {
let mut m = HashMap::default();
m.insert(height, 0);
m
},
space_end_for_start: {
let mut m = HashMap::default();
m.insert(0, height);
m
},
queue: Vec::new(),
all_glyphs: HashMap::default(),
pad_glyphs,
align_4x4,
multithread,
}
}
/// Rebuilds a `Cache` with new attributes. All cached glyphs are cleared,
/// however the glyph queue is retained unmodified.
///
/// # Panics
///
/// `scale_tolerance` or `position_tolerance` are less than or equal to
/// zero.
///
/// # Example
///
/// ```
/// # use rusttype::gpu_cache::Cache;
/// # let mut cache = Cache::builder().build();
/// // Rebuild the cache with different dimensions
/// cache.to_builder().dimensions(768, 768).rebuild(&mut cache);
/// ```
pub fn rebuild(self, cache: &mut Cache) {
let CacheBuilder {
dimensions: (width, height),
scale_tolerance,
position_tolerance,
pad_glyphs,
align_4x4,
multithread,
} = self.validated();
cache.width = width;
cache.height = height;
cache.scale_tolerance = scale_tolerance;
cache.position_tolerance = position_tolerance;
cache.pad_glyphs = pad_glyphs;
cache.align_4x4 = align_4x4;
cache.multithread = multithread;
cache.clear();
}
}
/// Returned from `Cache::rect_for`.
#[derive(Copy, Clone, Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
pub enum CacheReadErr {
/// Indicates that the requested glyph is not present in the cache
GlyphNotCached,
}
impl fmt::Display for CacheReadErr {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match *self {
CacheReadErr::GlyphNotCached => "Glyph not cached",
}
.fmt(f)
}
}
impl error::Error for CacheReadErr {}
/// Returned from `Cache::cache_queued`.
#[derive(Copy, Clone, Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
pub enum CacheWriteErr {
/// At least one of the queued glyphs is too big to fit into the cache, even
/// if all other glyphs are removed.
GlyphTooLarge,
/// Not all of the requested glyphs can fit into the cache, even if the
/// cache is completely cleared before the attempt.
NoRoomForWholeQueue,
}
impl fmt::Display for CacheWriteErr {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match *self {
CacheWriteErr::GlyphTooLarge => "Glyph too large",
CacheWriteErr::NoRoomForWholeQueue => "No room for whole queue",
}
.fmt(f)
}
}
impl error::Error for CacheWriteErr {}
/// Successful method of caching of the queue.
#[derive(Copy, Clone, Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
pub enum CachedBy {
/// Added any additional glyphs into the texture without affecting
/// the position of any already cached glyphs in the latest queue.
///
/// Glyphs not in the latest queue may have been removed.
Adding,
/// Fit the glyph queue by re-ordering all glyph texture positions.
/// Previous texture positions are no longer valid.
Reordering,
}
fn normalised_offset_from_position(position: Point<f32>) -> Vector<f32> {
let mut offset = vector(position.x.fract(), position.y.fract());
if offset.x > 0.5 {
offset.x -= 1.0;
} else if offset.x < -0.5 {
offset.x += 1.0;
}
if offset.y > 0.5 {
offset.y -= 1.0;
} else if offset.y < -0.5 {
offset.y += 1.0;
}
offset
}
impl<'font> Cache<'font> {
/// Returns a default `CacheBuilder`.
#[inline]
pub fn builder() -> CacheBuilder {
CacheBuilder::default()
}
/// Returns the current scale tolerance for the cache.
pub fn scale_tolerance(&self) -> f32 {
self.scale_tolerance
}
/// Returns the current subpixel position tolerance for the cache.
pub fn position_tolerance(&self) -> f32 {
self.position_tolerance
}
/// Returns the cache texture dimensions assumed by the cache. For proper
/// operation this should match the dimensions of the used GPU texture.
pub fn dimensions(&self) -> (u32, u32) {
(self.width, self.height)
}
/// Queue a glyph for caching by the next call to `cache_queued`. `font_id`
/// is used to disambiguate glyphs from different fonts. The user should
/// ensure that `font_id` is unique to the font the glyph is from.
pub fn queue_glyph(&mut self, font_id: usize, glyph: PositionedGlyph<'font>) {
if glyph.pixel_bounding_box().is_some() {
self.queue.push((font_id, glyph));
}
}
/// Clears the cache. Does not affect the glyph queue.
pub fn clear(&mut self) {
self.rows.clear();
self.space_end_for_start.clear();
self.space_end_for_start.insert(0, self.height);
self.space_start_for_end.clear();
self.space_start_for_end.insert(self.height, 0);
self.all_glyphs.clear();
}
/// Clears the glyph queue.
pub fn clear_queue(&mut self) {
self.queue.clear();
}
/// Returns a `CacheBuilder` with this cache's attributes.
pub fn to_builder(&self) -> CacheBuilder {
CacheBuilder {
dimensions: (self.width, self.height),
position_tolerance: self.position_tolerance,
scale_tolerance: self.scale_tolerance,
pad_glyphs: self.pad_glyphs,
align_4x4: self.align_4x4,
multithread: self.multithread,
}
}
/// Returns glyph info with accuracy according to the set tolerances.
fn lossy_info_for(&self, font_id: FontId, glyph: &PositionedGlyph<'font>) -> LossyGlyphInfo {
let scale = glyph.scale();
let offset = normalised_offset_from_position(glyph.position());
LossyGlyphInfo {
font_id,
glyph_id: glyph.id(),
scale_over_tolerance: (
(scale.x / self.scale_tolerance + 0.5) as u32,
(scale.y / self.scale_tolerance + 0.5) as u32,
),
// convert [-0.5, 0.5] -> [0, 1] then divide
offset_over_tolerance: (
((offset.x + 0.5) / self.position_tolerance + 0.5) as u16,
((offset.y + 0.5) / self.position_tolerance + 0.5) as u16,
),
}
}
/// Caches the queued glyphs. If this is unsuccessful, the queue is
/// untouched. Any glyphs cached by previous calls to this function may be
/// removed from the cache to make room for the newly queued glyphs. Thus if
/// you want to ensure that a glyph is in the cache, the most recently
/// cached queue must have contained that glyph.
///
/// `uploader` is the user-provided function that should perform the texture
/// uploads to the GPU. The information provided is the rectangular region
/// to insert the pixel data into, and the pixel data itself. This data is
/// provided in horizontal scanline format (row major), with stride equal to
/// the rectangle width.
///
/// If successful returns a `CachedBy` that can indicate the validity of
/// previously cached glyph textures.
pub fn cache_queued<F: FnMut(Rect<u32>, &[u8])>(
&mut self,
mut uploader: F,
) -> Result<CachedBy, CacheWriteErr> {
let mut queue_success = true;
let from_empty = self.all_glyphs.is_empty();
{
let (mut in_use_rows, mut uncached_glyphs) = {
let mut in_use_rows =
HashSet::with_capacity_and_hasher(self.rows.len(), FxBuildHasher::default());
let mut uncached_glyphs = Vec::with_capacity(self.queue.len());
// divide glyphs into texture rows where a matching glyph texture
// already exists & glyphs where new textures must be cached
for (font_id, ref glyph) in &self.queue {
let glyph_info = self.lossy_info_for(*font_id, glyph);
if let Some((row, ..)) = self.all_glyphs.get(&glyph_info) {
in_use_rows.insert(*row);
} else {
uncached_glyphs.push((glyph, glyph_info));
}
}
(in_use_rows, uncached_glyphs)
};
for row in &in_use_rows {
self.rows.get_refresh(row);
}
// tallest first gives better packing
// can use 'sort_unstable' as order of equal elements is unimportant
uncached_glyphs
.sort_unstable_by_key(|(glyph, ..)| -glyph.pixel_bounding_box().unwrap().height());
self.all_glyphs.reserve(uncached_glyphs.len());
let mut draw_and_upload = Vec::with_capacity(uncached_glyphs.len());
'per_glyph: for (glyph, glyph_info) in uncached_glyphs {
// glyph may match a texture cached by a previous iteration
if self.all_glyphs.contains_key(&glyph_info) {
continue;
}
// Not cached, so add it:
let (unaligned_width, unaligned_height) = {
let bb = glyph.pixel_bounding_box().unwrap();
if self.pad_glyphs {
(bb.width() as u32 + 2, bb.height() as u32 + 2)
} else {
(bb.width() as u32, bb.height() as u32)
}
};
let (aligned_width, aligned_height) = if self.align_4x4 {
// align to the next 4x4 texel boundary
((unaligned_width + 3) & !3, (unaligned_height + 3) & !3)
} else {
(unaligned_width, unaligned_height)
};
if aligned_width >= self.width || aligned_height >= self.height {
return Result::Err(CacheWriteErr::GlyphTooLarge);
}
// find row to put the glyph in, most used rows first
let mut row_top = None;
for (top, row) in self.rows.iter().rev() {
if row.height >= aligned_height && self.width - row.width >= aligned_width {
// found a spot on an existing row
row_top = Some(*top);
break;
}
}
if row_top.is_none() {
let mut gap = None;
// See if there is space for a new row
for (start, end) in &self.space_end_for_start {
if end - start >= aligned_height {
gap = Some((*start, *end));
break;
}
}
if gap.is_none() {
// Remove old rows until room is available
while !self.rows.is_empty() {
// check that the oldest row isn't also in use
if !in_use_rows.contains(self.rows.front().unwrap().0) {
// Remove row
let (top, row) = self.rows.pop_front().unwrap();
for g in row.glyphs {
self.all_glyphs.remove(&g.glyph_info);
}
let (mut new_start, mut new_end) = (top, top + row.height);
// Update the free space maps
// Combine with neighbouring free space if possible
if let Some(end) = self.space_end_for_start.remove(&new_end) {
new_end = end;
}
if let Some(start) = self.space_start_for_end.remove(&new_start) {
new_start = start;
}
self.space_start_for_end.insert(new_end, new_start);
self.space_end_for_start.insert(new_start, new_end);
if new_end - new_start >= aligned_height {
// The newly formed gap is big enough
gap = Some((new_start, new_end));
break;
}
}
// all rows left are in use
// try a clean insert of all needed glyphs
// if that doesn't work, fail
else if from_empty {
// already trying a clean insert, don't do it again
return Err(CacheWriteErr::NoRoomForWholeQueue);
} else {
// signal that a retry is needed
queue_success = false;
break 'per_glyph;
}
}
}
let (gap_start, gap_end) = gap.unwrap();
// fill space for new row
let new_space_start = gap_start + aligned_height;
self.space_end_for_start.remove(&gap_start);
if new_space_start == gap_end {
self.space_start_for_end.remove(&gap_end);
} else {
self.space_end_for_start.insert(new_space_start, gap_end);
self.space_start_for_end.insert(gap_end, new_space_start);
}
// add the row
self.rows.insert(
gap_start,
Row {
width: 0,
height: aligned_height,
glyphs: Vec::new(),
},
);
row_top = Some(gap_start);
}
let row_top = row_top.unwrap();
// calculate the target rect
let row = self.rows.get_refresh(&row_top).unwrap();
let aligned_tex_coords = Rect {
min: point(row.width, row_top),
max: point(row.width + aligned_width, row_top + aligned_height),
};
let unaligned_tex_coords = Rect {
min: point(row.width, row_top),
max: point(row.width + unaligned_width, row_top + unaligned_height),
};
draw_and_upload.push((aligned_tex_coords, glyph));
// add the glyph to the row
row.glyphs.push(GlyphTexInfo {
glyph_info,
offset: normalised_offset_from_position(glyph.position()),
tex_coords: unaligned_tex_coords,
});
row.width += aligned_width;
in_use_rows.insert(row_top);
self.all_glyphs
.insert(glyph_info, (row_top, row.glyphs.len() as u32 - 1));
}
if queue_success {
#[cfg(not(target_arch = "wasm32"))]
{
let glyph_count = draw_and_upload.len();
if self.multithread && glyph_count > 1 {
// multithread rasterization
use crossbeam_deque::Steal;
use std::{
mem,
sync::mpsc::{self, TryRecvError},
};
let rasterize_queue = crossbeam_deque::Injector::new();
let (to_main, from_stealers) = mpsc::channel();
let pad_glyphs = self.pad_glyphs;
for el in draw_and_upload {
rasterize_queue.push(el);
}
crossbeam_utils::thread::scope(|scope| {
for _ in 0..num_cpus::get().min(glyph_count).saturating_sub(1) {
let rasterize_queue = &rasterize_queue;
let to_main = to_main.clone();
scope.spawn(move |_| loop {
match rasterize_queue.steal() {
Steal::Success((tex_coords, glyph)) => {
let pixels = draw_glyph(tex_coords, glyph, pad_glyphs);
to_main.send((tex_coords, pixels)).unwrap();
}
Steal::Empty => break,
Steal::Retry => {}
}
});
}
mem::drop(to_main);
let mut workers_finished = false;
loop {
match rasterize_queue.steal() {
Steal::Success((tex_coords, glyph)) => {
let pixels = draw_glyph(tex_coords, glyph, pad_glyphs);
uploader(tex_coords, pixels.as_slice());
}
Steal::Empty if workers_finished => break,
Steal::Empty | Steal::Retry => {}
}
while !workers_finished {
match from_stealers.try_recv() {
Ok((tex_coords, pixels)) => {
uploader(tex_coords, pixels.as_slice())
}
Err(TryRecvError::Disconnected) => workers_finished = true,
Err(TryRecvError::Empty) => break,
}
}
}
})
.unwrap();
} else {
// single thread rasterization
for (tex_coords, glyph) in draw_and_upload {
let pixels = draw_glyph(tex_coords, glyph, self.pad_glyphs);
uploader(tex_coords, pixels.as_slice());
}
}
}
#[cfg(target_arch = "wasm32")]
{
for (tex_coords, glyph) in draw_and_upload {
let pixels = draw_glyph(tex_coords, glyph, self.pad_glyphs);
uploader(tex_coords, pixels.as_slice());
}
}
}
}
if queue_success {
self.queue.clear();
Ok(CachedBy::Adding)
} else {
// clear the cache then try again with optimal packing
self.clear();
self.cache_queued(uploader).map(|_| CachedBy::Reordering)
}
}
/// Retrieves the (floating point) texture coordinates of the quad for a
/// glyph in the cache, as well as the pixel-space (integer) coordinates
/// that this region should be drawn at. These pixel-space coordinates
/// assume an origin at the top left of the quad. In the majority of cases
/// these pixel-space coordinates should be identical to the bounding box of
/// the input glyph. They only differ if the cache has returned a substitute
/// glyph that is deemed close enough to the requested glyph as specified by
/// the cache tolerance parameters.
///
/// A sucessful result is `Some` if the glyph is not an empty glyph (no
/// shape, and thus no rect to return).
///
/// Ensure that `font_id` matches the `font_id` that was passed to
/// `queue_glyph` with this `glyph`.
pub fn rect_for(
&self,
font_id: usize,
glyph: &PositionedGlyph,
) -> Result<Option<TextureCoords>, CacheReadErr> {
if glyph.pixel_bounding_box().is_none() {
return Ok(None);
}
let (row, index) = self
.all_glyphs
.get(&self.lossy_info_for(font_id, glyph))
.ok_or(CacheReadErr::GlyphNotCached)?;
let (tex_width, tex_height) = (self.width as f32, self.height as f32);
let GlyphTexInfo {
tex_coords: mut tex_rect,
offset: tex_offset,
..
} = self.rows[&row].glyphs[*index as usize];
if self.pad_glyphs {
tex_rect = tex_rect.unpadded();
}
let uv_rect = Rect {
min: point(
tex_rect.min.x as f32 / tex_width,
tex_rect.min.y as f32 / tex_height,
),
max: point(
tex_rect.max.x as f32 / tex_width,
tex_rect.max.y as f32 / tex_height,
),
};
let local_bb = glyph
.unpositioned()
.clone()
.positioned(point(0.0, 0.0) + tex_offset)
.pixel_bounding_box()
.unwrap();
let min_from_origin =
point(local_bb.min.x as f32, local_bb.min.y as f32) - (point(0.0, 0.0) + tex_offset);
let ideal_min = min_from_origin + glyph.position();
let min = point(ideal_min.x.round() as i32, ideal_min.y.round() as i32);
let bb_offset = min - local_bb.min;
let bb = Rect {
min,
max: local_bb.max + bb_offset,
};
Ok(Some((uv_rect, bb)))
}
}
#[inline]
fn draw_glyph(tex_coords: Rect<u32>, glyph: &PositionedGlyph<'_>, pad_glyphs: bool) -> ByteArray2d {
let mut pixels = ByteArray2d::zeros(tex_coords.height() as usize, tex_coords.width() as usize);
if pad_glyphs {
glyph.draw(|x, y, v| {
let v = (v * 255.0).round().max(0.0).min(255.0) as u8;
// `+ 1` accounts for top/left glyph padding
pixels[(y as usize + 1, x as usize + 1)] = v;
});
} else {
glyph.draw(|x, y, v| {
let v = (v * 255.0).round().max(0.0).min(255.0) as u8;
pixels[(y as usize, x as usize)] = v;
});
}
pixels
}
#[cfg(test)]
mod test {
use super::*;
use crate::{Font, FontCollection, Scale};
use approx::*;
#[test]
fn cache_test() {
let font_data = include_bytes!("../dev/fonts/wqy-microhei/WenQuanYiMicroHei.ttf");
let font = FontCollection::from_bytes(font_data as &[u8])
.unwrap()
.into_font()
.unwrap();
let mut cache = Cache::builder()
.dimensions(32, 32)
.scale_tolerance(0.1)
.position_tolerance(0.1)
.pad_glyphs(false)
.build();
let strings = [
("Hello World!", 15.0),
("Hello World!", 14.0),
("Hello World!", 10.0),
("Hello World!", 15.0),
("Hello World!", 14.0),
("Hello World!", 10.0),
];
for &(string, scale) in &strings {
println!("Caching {:?}", (string, scale));
for glyph in font.layout(string, Scale::uniform(scale), point(0.0, 0.0)) {
cache.queue_glyph(0, glyph);
}
cache.cache_queued(|_, _| {}).unwrap();
}
}
#[test]
fn need_to_check_whole_cache() {
let font_data = include_bytes!("../dev/fonts/wqy-microhei/WenQuanYiMicroHei.ttf");
let font = Font::from_bytes(font_data as &[u8]).unwrap();
let glyph = font.glyph('l');
let small = glyph.clone().scaled(Scale::uniform(10.0));
let large = glyph.clone().scaled(Scale::uniform(10.05));
let small_left = small.clone().positioned(point(0.0, 0.0));
let large_left = large.clone().positioned(point(0.0, 0.0));
let large_right = large.clone().positioned(point(-0.2, 0.0));
let mut cache = Cache::builder()
.dimensions(32, 32)
.scale_tolerance(0.1)
.position_tolerance(0.1)
.pad_glyphs(false)
.build();
cache.queue_glyph(0, small_left.clone());
// Next line is noop since it's within the scale tolerance of small_left:
cache.queue_glyph(0, large_left.clone());
cache.queue_glyph(0, large_right.clone());
cache.cache_queued(|_, _| {}).unwrap();
cache.rect_for(0, &small_left).unwrap();
cache.rect_for(0, &large_left).unwrap();
cache.rect_for(0, &large_right).unwrap();
}
#[test]
fn lossy_info() {
let font_data = include_bytes!("../dev/fonts/wqy-microhei/WenQuanYiMicroHei.ttf");
let font = Font::from_bytes(font_data as &[u8]).unwrap();
let glyph = font.glyph('l');
let small = glyph.clone().scaled(Scale::uniform(9.91));
let near = glyph.clone().scaled(Scale::uniform(10.09));
let far = glyph.clone().scaled(Scale::uniform(10.11));
let really_far = glyph.clone().scaled(Scale::uniform(12.0));
let small_pos = small.clone().positioned(point(0.0, 0.0));
let match_1 = near.clone().positioned(point(-10.0, -0.1));
let match_2 = near.clone().positioned(point(5.1, 0.24));
let match_3 = small.clone().positioned(point(-100.2, 50.1));
let miss_1 = far.clone().positioned(point(0.0, 0.0));
let miss_2 = really_far.clone().positioned(point(0.0, 0.0));
let miss_3 = small.clone().positioned(point(0.3, 0.0));
let cache = Cache::builder()
.scale_tolerance(0.2)
.position_tolerance(0.5)
.build();
let small_info = cache.lossy_info_for(0, &small_pos);
assert_eq!(small_info, cache.lossy_info_for(0, &match_1));
assert_eq!(small_info, cache.lossy_info_for(0, &match_2));
assert_eq!(small_info, cache.lossy_info_for(0, &match_3));
assert_ne!(small_info, cache.lossy_info_for(0, &miss_1));
assert_ne!(small_info, cache.lossy_info_for(0, &miss_2));
assert_ne!(small_info, cache.lossy_info_for(0, &miss_3));
}
#[test]
fn cache_to_builder() {
let cache = CacheBuilder {
dimensions: (32, 64),
scale_tolerance: 0.2,
position_tolerance: 0.3,
pad_glyphs: false,
align_4x4: false,
multithread: false,
}
.build();
let to_builder: CacheBuilder = cache.to_builder();
assert_eq!(to_builder.dimensions, (32, 64));
assert_relative_eq!(to_builder.scale_tolerance, 0.2);
assert_relative_eq!(to_builder.position_tolerance, 0.3);
assert_eq!(to_builder.pad_glyphs, false);
assert_eq!(to_builder.align_4x4, false);
assert_eq!(to_builder.multithread, false);
}
#[test]
fn builder_rebuild() {
let mut cache = Cache::builder()
.dimensions(32, 64)
.scale_tolerance(0.2)
.position_tolerance(0.3)
.pad_glyphs(false)
.align_4x4(true)
.multithread(true)
.build();
let font = Font::from_bytes(include_bytes!(
"../dev/fonts/wqy-microhei/WenQuanYiMicroHei.ttf"
) as &[u8])
.unwrap();
cache.queue_glyph(
0,
font.glyph('l')
.scaled(Scale::uniform(25.0))
.positioned(point(0.0, 0.0)),
);
cache.cache_queued(|_, _| {}).unwrap();
cache.queue_glyph(
0,
font.glyph('a')
.scaled(Scale::uniform(25.0))
.positioned(point(0.0, 0.0)),
);
Cache::builder()
.dimensions(64, 128)
.scale_tolerance(0.05)
.position_tolerance(0.15)
.pad_glyphs(true)
.align_4x4(false)
.multithread(false)
.rebuild(&mut cache);
assert_eq!(cache.width, 64);
assert_eq!(cache.height, 128);
assert_relative_eq!(cache.scale_tolerance, 0.05);
assert_relative_eq!(cache.position_tolerance, 0.15);
assert_eq!(cache.pad_glyphs, true);
assert_eq!(cache.align_4x4, false);
assert_eq!(cache.multithread, false);
assert!(
cache.all_glyphs.is_empty(),
"cache should have been cleared"
);
assert_eq!(cache.queue.len(), 1, "cache should have an unchanged queue");
}
/// Provide to caller that the cache was re-ordered to fit the latest queue
#[test]
fn return_cache_by_reordering() {
let font_data = include_bytes!("../dev/fonts/wqy-microhei/WenQuanYiMicroHei.ttf");
let font = FontCollection::from_bytes(font_data as &[u8])
.unwrap()
.into_font()
.unwrap();
let mut cache = Cache::builder()
.dimensions(36, 27)
.scale_tolerance(0.1)
.position_tolerance(0.1)
.build();
for glyph in font.layout("ABCDEFG", Scale::uniform(16.0), point(0.0, 0.0)) {
cache.queue_glyph(0, glyph);
}
assert_eq!(cache.cache_queued(|_, _| {}), Ok(CachedBy::Adding));
for glyph in font.layout("DEFGHIJK", Scale::uniform(16.0), point(0.0, 0.0)) {
cache.queue_glyph(0, glyph);
}
assert_eq!(cache.cache_queued(|_, _| {}), Ok(CachedBy::Reordering));
}
#[test]
fn align_4x4() {
// First, test align_4x4 disabled, to confirm non-4x4 alignment
align_4x4_helper(false, 5, 19);
// Now, test with align_4x4 enabled, to confirm 4x4 alignment
align_4x4_helper(true, 8, 20);
}
fn align_4x4_helper(align_4x4: bool, expected_width: u32, expected_height: u32) {
let mut cache = Cache::builder()
.dimensions(64, 64)
.align_4x4(align_4x4)
.build();
let font = Font::from_bytes(include_bytes!(
"../dev/fonts/wqy-microhei/WenQuanYiMicroHei.ttf"
) as &[u8])
.unwrap();
let glyph = font
.glyph('l')
.scaled(Scale::uniform(25.0))
.positioned(point(0.0, 0.0));
cache.queue_glyph(0, glyph.clone());
cache
.cache_queued(|rect, _| {
assert_eq!(rect.width(), expected_width);
assert_eq!(rect.height(), expected_height);
})
.unwrap();
let (uv_rect, _screen_rect) = cache.rect_for(0, &glyph).unwrap().unwrap();
assert_eq!(
uv_rect,
crate::Rect {
min: crate::point(0.015_625, 0.015_625),
max: crate::point(0.0625, 0.28125),
}
);
}
}