azul_layout/font.rs
1//! Font parsing, metrics extraction, and subsetting.
2//!
3//! This module provides the core font infrastructure for text layout and PDF generation:
4//! - `loading`: System font cache construction and font reload errors
5//! - `mock`: Mock font implementation for testing without real font files
6//! - `parsed`: Full font parsing via allsorts (outlines, metrics, shaping tables, subsetting)
7
8#![cfg(feature = "font_loading")]
9
10use azul_css::{AzString, U8Vec};
11use rust_fontconfig::{FcFontCache, OwnedFontSource};
12
13pub mod loading {
14 #![cfg(feature = "std")]
15 #![cfg(feature = "font_loading")]
16 #![cfg_attr(not(feature = "std"), no_std)]
17
18 use std::io::Error as IoError;
19
20 use azul_css::{AzString, StringVec, U8Vec};
21 use rust_fontconfig::FcFontCache;
22
23 #[cfg(not(miri))]
24 #[must_use] pub fn build_font_cache() -> FcFontCache {
25 FcFontCache::build()
26 }
27
28 #[cfg(miri)]
29 pub fn build_font_cache() -> FcFontCache {
30 FcFontCache::default()
31 }
32
33 #[derive(Debug)]
34 pub enum FontReloadError {
35 Io(IoError, AzString),
36 FontNotFound(AzString),
37 FontLoadingNotActive(AzString),
38 }
39
40 impl Clone for FontReloadError {
41 fn clone(&self) -> Self {
42 use self::FontReloadError::{Io, FontNotFound, FontLoadingNotActive};
43 match self {
44 Io(err, path) => Io(IoError::new(err.kind(), "Io Error"), path.clone()),
45 FontNotFound(id) => FontNotFound(id.clone()),
46 FontLoadingNotActive(id) => FontLoadingNotActive(id.clone()),
47 }
48 }
49 }
50
51 azul_core::impl_display!(FontReloadError, {
52 Io(err, path_buf) => format!("Could not load \"{}\" - IO error: {}", path_buf.as_str(), err),
53 FontNotFound(id) => format!("Could not locate system font: \"{:?}\" found", id),
54 FontLoadingNotActive(id) => format!("Could not load system font: \"{:?}\": crate was not compiled with --features=\"font_loading\"", id)
55 });
56}
57pub mod mock {
58 //! Mock font implementation for testing text layout.
59 //!
60 //! Provides a `MockFont` that simulates font behavior without requiring
61 //! actual font files, useful for unit testing text layout functionality.
62
63 use std::collections::BTreeMap;
64
65 use crate::text3::cache::LayoutFontMetrics;
66
67 /// A mock font implementation for testing text layout without real fonts.
68 ///
69 /// This allows testing text shaping, layout, and rendering code paths
70 /// without needing to load actual TrueType/OpenType font files.
71 #[derive(Debug, Clone)]
72 pub struct MockFont {
73 /// Font metrics (ascent, descent, etc.).
74 pub font_metrics: LayoutFontMetrics,
75 /// Width of the space character in font units.
76 pub space_width: Option<usize>,
77 /// Horizontal advance widths keyed by glyph ID.
78 pub glyph_advances: BTreeMap<u16, u16>,
79 /// Glyph bounding box sizes (width, height) keyed by glyph ID.
80 pub glyph_sizes: BTreeMap<u16, (i32, i32)>,
81 /// Unicode codepoint to glyph ID mapping.
82 pub glyph_indices: BTreeMap<u32, u16>,
83 }
84
85 impl MockFont {
86 /// Creates a new `MockFont` with the given font metrics.
87 #[must_use] pub const fn new(font_metrics: LayoutFontMetrics) -> Self {
88 Self {
89 font_metrics,
90 space_width: Some(10),
91 glyph_advances: BTreeMap::new(),
92 glyph_sizes: BTreeMap::new(),
93 glyph_indices: BTreeMap::new(),
94 }
95 }
96
97 /// Sets the space character width.
98 #[must_use] pub const fn with_space_width(mut self, width: usize) -> Self {
99 self.space_width = Some(width);
100 self
101 }
102
103 /// Adds a horizontal advance value for a glyph.
104 #[must_use] pub fn with_glyph_advance(mut self, glyph_index: u16, advance: u16) -> Self {
105 self.glyph_advances.insert(glyph_index, advance);
106 self
107 }
108
109 /// Adds a bounding box size for a glyph.
110 #[must_use] pub fn with_glyph_size(mut self, glyph_index: u16, size: (i32, i32)) -> Self {
111 self.glyph_sizes.insert(glyph_index, size);
112 self
113 }
114
115 /// Adds a Unicode codepoint to glyph ID mapping.
116 #[must_use] pub fn with_glyph_index(mut self, unicode: u32, index: u16) -> Self {
117 self.glyph_indices.insert(unicode, index);
118 self
119 }
120 }
121}
122
123pub mod parsed {
124 use core::fmt;
125 use std::{collections::BTreeMap, sync::Arc};
126
127 use allsorts::{
128 binary::read::ReadScope,
129 font_data::FontData,
130 layout::{GDEFTable, LayoutCache, LayoutCacheData, GPOS, GSUB},
131 outline::{OutlineBuilder, OutlineSink},
132 pathfinder_geometry::{line_segment::LineSegment2F, vector::Vector2F},
133 subset::{subset as allsorts_subset, whole_font, CmapTarget, SubsetProfile},
134 tables::{
135 cmap::owned::CmapSubtable as OwnedCmapSubtable,
136 glyf::{
137 Glyph, GlyfVisitorContext, LocaGlyf, Point,
138 VariableGlyfContext, VariableGlyfContextStore,
139 },
140 kern::owned::KernTable,
141 FontTableProvider, HheaTable, MaxpTable,
142 },
143 tag,
144 };
145 use azul_core::resources::{
146 GlyphOutline, GlyphOutlineOperation, OutlineCubicTo, OutlineLineTo, OutlineMoveTo,
147 OutlineQuadTo, OwnedGlyphBoundingBox,
148 };
149 use azul_css::props::basic::FontMetrics as CssFontMetrics;
150
151 // Mock font module for testing
152 pub use crate::font::mock::MockFont;
153 use crate::text3::cache::LayoutFontMetrics;
154
155 /// Cached GSUB table for glyph substitution operations.
156 pub type GsubCache = Arc<LayoutCacheData<GSUB>>;
157 /// Cached GPOS table for glyph positioning operations.
158 pub type GposCache = Arc<LayoutCacheData<GPOS>>;
159
160 /// Monotonic-clock nanos since process start. Used to timestamp
161 /// `ParsedFont.last_used` for LRU eviction. Cheap (single
162 /// `Instant::now`); resolution is plenty fine for "did this
163 /// face get touched in the last N seconds" decisions. Exposed
164 /// `pub(crate)` so `FontManager::evict_unused` reads from the
165 /// same clock as `last_used` writes.
166 #[allow(clippy::cast_possible_truncation)] // bounded graphics/coord/font/fixed-point/debug-marker cast
167 pub(crate) fn monotonic_now_nanos() -> u64 {
168 // Safe: `Instant::elapsed` against the same launch instant is
169 // monotonic and never overflows in any realistic process
170 // lifetime (>500 years).
171 use std::sync::OnceLock;
172 use std::time::Instant;
173 static LAUNCH: OnceLock<Instant> = OnceLock::new();
174 let start = LAUNCH.get_or_init(Instant::now);
175 start.elapsed().as_nanos() as u64
176 }
177
178 /// Glyph-outline decoder state. See the
179 /// [`ParsedFont::loca_glyf`] field docs for the full description.
180 #[derive(Clone)]
181 pub(crate) enum LocaGlyfState {
182 /// Ready to decode immediately, or known to have no outline
183 /// data. `None` covers both CFF fonts and fonts where the
184 /// loca+glyf parse failed.
185 ///
186 /// This variant *cannot* be evicted by
187 /// [`crate::text3::cache::FontManager::evict_unused`]: there
188 /// are no source bytes retained to re-decode from. The eager
189 /// `from_bytes` path (tests, `with_source_bytes` PDF callers)
190 /// produces this variant.
191 Loaded(Option<Arc<std::sync::Mutex<LocaGlyf>>>),
192 /// Font bytes retained for lazy `LocaGlyf` construction.
193 ///
194 /// `loaded` is `Mutex<Option<…>>` (not `OnceLock`) so an
195 /// idle eviction can clear it back to `None`; the next
196 /// `get_or_decode_glyph` will re-parse from `bytes`. Two-step
197 /// double-check pattern in `resolve_loca_glyf` keeps the
198 /// expensive `LocaGlyf::load` outside the critical section.
199 Deferred {
200 bytes: Arc<rust_fontconfig::FontBytes>,
201 font_index: usize,
202 loaded: Arc<std::sync::Mutex<Option<Arc<std::sync::Mutex<LocaGlyf>>>>>,
203 },
204 }
205
206 /// Adapter that collects allsorts outline commands into our `GlyphOutline` format.
207 ///
208 /// Implements `OutlineSink` so it can be passed to `GlyfVisitorContext::visit()`.
209 /// This handles composite glyph resolution, transforms, and variable font
210 /// deltas automatically via allsorts internals.
211 struct GlyphOutlineCollector {
212 contours: Vec<GlyphOutline>,
213 current_contour: Vec<GlyphOutlineOperation>,
214 }
215
216 impl GlyphOutlineCollector {
217 const fn new() -> Self {
218 Self {
219 contours: Vec::new(),
220 current_contour: Vec::new(),
221 }
222 }
223
224 fn into_outlines(mut self) -> Vec<GlyphOutline> {
225 if !self.current_contour.is_empty() {
226 self.contours.push(GlyphOutline {
227 operations: std::mem::take(&mut self.current_contour).into(),
228 });
229 }
230 self.contours
231 }
232 }
233
234 impl OutlineSink for GlyphOutlineCollector {
235 #[allow(clippy::cast_possible_truncation)] // bounded graphics/coord/font/fixed-point/debug-marker cast
236 fn move_to(&mut self, to: Vector2F) {
237 if !self.current_contour.is_empty() {
238 self.contours.push(GlyphOutline {
239 operations: std::mem::take(&mut self.current_contour).into(),
240 });
241 }
242 self.current_contour.push(GlyphOutlineOperation::MoveTo(OutlineMoveTo {
243 x: to.x() as i16,
244 y: to.y() as i16,
245 }));
246 }
247
248 #[allow(clippy::cast_possible_truncation)] // bounded graphics/coord/font/fixed-point/debug-marker cast
249 fn line_to(&mut self, to: Vector2F) {
250 self.current_contour.push(GlyphOutlineOperation::LineTo(OutlineLineTo {
251 x: to.x() as i16,
252 y: to.y() as i16,
253 }));
254 }
255
256 #[allow(clippy::cast_possible_truncation)] // bounded graphics/coord/font/fixed-point/debug-marker cast
257 fn quadratic_curve_to(&mut self, ctrl: Vector2F, to: Vector2F) {
258 self.current_contour.push(GlyphOutlineOperation::QuadraticCurveTo(
259 OutlineQuadTo {
260 ctrl_1_x: ctrl.x() as i16,
261 ctrl_1_y: ctrl.y() as i16,
262 end_x: to.x() as i16,
263 end_y: to.y() as i16,
264 },
265 ));
266 }
267
268 #[allow(clippy::cast_possible_truncation)] // bounded graphics/coord/font/fixed-point/debug-marker cast
269 fn cubic_curve_to(&mut self, ctrl: LineSegment2F, to: Vector2F) {
270 self.current_contour.push(GlyphOutlineOperation::CubicCurveTo(
271 OutlineCubicTo {
272 ctrl_1_x: ctrl.from_x() as i16,
273 ctrl_1_y: ctrl.from_y() as i16,
274 ctrl_2_x: ctrl.to_x() as i16,
275 ctrl_2_y: ctrl.to_y() as i16,
276 end_x: to.x() as i16,
277 end_y: to.y() as i16,
278 },
279 ));
280 }
281
282 fn close(&mut self) {
283 self.current_contour.push(GlyphOutlineOperation::ClosePath);
284 self.contours.push(GlyphOutline {
285 operations: std::mem::take(&mut self.current_contour).into(),
286 });
287 }
288 }
289
290 /// Parsed font data with all required tables for text layout and PDF generation.
291 ///
292 /// This struct holds the parsed representation of a TrueType/OpenType font,
293 /// including glyph outlines, metrics, and shaping tables. It's used for:
294 /// - Text layout (via GSUB/GPOS tables)
295 /// - Glyph rendering (via glyf/CFF outlines)
296 /// - PDF font embedding (via font metrics and subsetting)
297 pub struct ParsedFont {
298 /// Hash of the font bytes for caching and equality checks.
299 pub hash: u64,
300 /// Layout-specific font metrics (ascent, descent, line gap).
301 pub font_metrics: LayoutFontMetrics,
302 /// PDF-specific detailed font metrics from HEAD, HHEA, OS/2 tables.
303 pub pdf_font_metrics: PdfFontMetrics,
304 /// Total number of glyphs in the font (from maxp table).
305 pub num_glyphs: u16,
306 /// Horizontal header table (hhea) containing global horizontal metrics.
307 pub hhea_table: HheaTable,
308 /// Offset+length into `original_bytes` for hmtx table (lazy: no copy).
309 pub hmtx_range: (usize, usize),
310 /// Offset+length into `original_bytes` for vmtx table (lazy: no copy).
311 pub vmtx_range: (usize, usize),
312 /// Vertical header table (vhea), same format as hhea. None if font has no vertical metrics.
313 pub vhea_table: Option<HheaTable>,
314 /// Maximum profile table (maxp) containing glyph count and memory hints.
315 pub maxp_table: MaxpTable,
316 /// Raw GSUB table bytes, kept as a `Vec<u8>` (tens to low-hundreds
317 /// of KiB) so the parsed `GsubCache` can be built on first shape
318 /// call instead of up-front. Access via [`ParsedFont::gsub`] —
319 /// that getter populates `gsub_cache_lazy` via `OnceLock` and
320 /// returns a borrow.
321 pub(crate) gsub_bytes: Option<Vec<u8>>,
322 /// Lazy GSUB cache: populated on first [`ParsedFont::gsub`] call.
323 /// `None` means "font has no GSUB table" *after* init attempt;
324 /// the `OnceLock` wrapper distinguishes "not yet initialised"
325 /// from "initialised to None".
326 pub(crate) gsub_cache_lazy: std::sync::OnceLock<Option<GsubCache>>,
327 /// Raw GPOS table bytes. Same lazy-parse arrangement as
328 /// `gsub_bytes` — see [`ParsedFont::gpos`].
329 pub(crate) gpos_bytes: Option<Vec<u8>>,
330 /// Lazy GPOS cache, populated on first [`ParsedFont::gpos`] call.
331 pub(crate) gpos_cache_lazy: std::sync::OnceLock<Option<GposCache>>,
332 /// Glyph definition table (GDEF) for glyph classification.
333 pub opt_gdef_table: Option<Arc<GDEFTable>>,
334 /// Legacy kerning table (kern) for fonts without GPOS.
335 pub opt_kern_table: Option<Arc<KernTable>>,
336 /// Monotonic-clock nanos at the most recent
337 /// [`ParsedFont::get_or_decode_glyph`] / `gsub()` / `gpos()`
338 /// call. `0` means "never touched". Used by
339 /// [`crate::text3::cache::FontManager::evict_unused`] to
340 /// decide which `LocaGlyfState::Deferred` faces to release.
341 pub(crate) last_used: Arc<std::sync::atomic::AtomicU64>,
342 /// `true` if this font is a variable font (carries a `gvar`
343 /// table). Cached at parse time so [`decode_glyph_inner`]
344 /// can short-circuit the variable-context construction for
345 /// the common non-variable case. Variable-glyph delta
346 /// application requires the source bytes to be retained,
347 /// so it only fires on the `LocaGlyfState::Deferred` path.
348 pub(crate) is_variable_font: bool,
349 /// Lazy outline cache. Populated on first
350 /// [`ParsedFont::get_or_decode_glyph`] call per `gid`; entries
351 /// are wrapped in `Arc` so callers can hold them without
352 /// keeping the lock. The space glyph (and `.notdef` when
353 /// present) are pre-inserted by `from_bytes_internal` so the
354 /// shaper's cmap-miss path has something to render without
355 /// racing with a decode.
356 ///
357 /// Tests that previously walked the public `glyph_records_decoded`
358 /// `BTreeMap` field now call
359 /// [`ParsedFont::prime_glyph_cache`] (decodes every glyph into
360 /// this cache) followed by
361 /// [`ParsedFont::for_each_decoded_glyph`] /
362 /// [`ParsedFont::glyph_cache_snapshot`] to walk the result.
363 // [az-web-lift] queue RwLock spins in lock_contended in single-threaded lifted wasm
364 // (only the pure-Rust queue RwLock is lifted; Mutex is Leaf-stubbed). Reuse
365 // rust_fontconfig::StLock (no-atomic single-threaded bypass). One of the 3 RwLocks total.
366 pub(crate) glyph_cache: Arc<rust_fontconfig::StLock<BTreeMap<u16, Arc<OwnedGlyph>>>>,
367 /// Glyph outline decoder state.
368 ///
369 /// - `Loaded(Some(arc))`: `LocaGlyf` is already loaded (owning
370 /// its own `Box<[u8]>` copy of the loca+glyf tables) and
371 /// ready to decode glyphs. Produced by the eager `from_bytes`
372 /// constructor path (tests).
373 /// - `Loaded(None)`: the font has no usable loca+glyf (CFF, or
374 /// a parse failure). Glyph outlines won't decode; the hmtx
375 /// advance fallback fills in the blanks.
376 /// - `Deferred`: we retain an `Arc<[u8]>` to the full font file
377 /// and the `font_index`; the first `get_or_decode_glyph` call
378 /// parses a fresh `FontData` / `TableProvider` from those
379 /// bytes and loads `LocaGlyf`, storing the result in the
380 /// `OnceLock`. Fonts that get resolved into a chain but are
381 /// never actually rasterized pay zero decode cost — this is
382 /// the big win for pages like `excel.html` where 20+ fallback
383 /// faces load but only a handful are touched.
384 pub(crate) loca_glyf: LocaGlyfState,
385 /// Cached width of the space character in font units.
386 pub space_width: Option<usize>,
387 /// Character-to-glyph mapping (cmap subtable).
388 pub cmap_subtable: Option<OwnedCmapSubtable>,
389 /// Mock font data for testing (replaces real font behavior).
390 pub mock: Option<Box<MockFont>>,
391 /// Reverse mapping: `glyph_id` -> cluster text (handles ligatures like "fi").
392 pub reverse_glyph_cache: BTreeMap<u16, String>,
393 /// Original font bytes — only retained for callers that need to
394 /// reconstruct or subset the font (PDF export). Layout / shaping /
395 /// raster never read this, so `ParsedFont::from_bytes` leaves it
396 /// as `None` by default and callers opt in via
397 /// [`ParsedFont::with_source_bytes`]. Shared across faces of the
398 /// same `.ttc` via the `Arc<FontBytes>` that
399 /// [`rust_fontconfig::FcFontCache::get_font_bytes`] returns —
400 /// for disk fonts the backing is an mmap so untouched pages
401 /// don't count toward RSS.
402 pub original_bytes: Option<Arc<rust_fontconfig::FontBytes>>,
403 /// Font index within collection (0 for single-font files).
404 pub original_index: usize,
405 /// GID to CID mapping for CFF fonts (required for PDF embedding).
406 pub index_to_cid: BTreeMap<u16, u16>,
407 /// Font type (TrueType outlines or OpenType CFF).
408 pub font_type: FontType,
409 /// PostScript font name from the NAME table.
410 pub font_name: Option<String>,
411 /// TrueType bytecode hinting instance (mutable interpreter state).
412 /// Wrapped in Mutex because hinting mutates internal state.
413 /// None for CFF fonts or fonts without hinting data.
414 pub hint_instance: Option<std::sync::Mutex<allsorts::hinting::HintInstance>>,
415 }
416
417 impl Clone for ParsedFont {
418 fn clone(&self) -> Self {
419 Self {
420 hash: self.hash,
421 font_metrics: self.font_metrics,
422 pdf_font_metrics: self.pdf_font_metrics,
423 num_glyphs: self.num_glyphs,
424 hhea_table: self.hhea_table.clone(),
425 hmtx_range: self.hmtx_range,
426 vmtx_range: self.vmtx_range,
427 vhea_table: self.vhea_table.clone(),
428 maxp_table: self.maxp_table.clone(),
429 // OnceLock<T: Clone>: Clone preserves the init state, so
430 // a clone of a parsed cache skips re-parse on first
431 // access. The raw bytes we keep around for lazy init
432 // are cloned too.
433 gsub_bytes: self.gsub_bytes.clone(),
434 gsub_cache_lazy: self.gsub_cache_lazy.clone(),
435 gpos_bytes: self.gpos_bytes.clone(),
436 gpos_cache_lazy: self.gpos_cache_lazy.clone(),
437 opt_gdef_table: self.opt_gdef_table.clone(),
438 opt_kern_table: self.opt_kern_table.clone(),
439 // Share the lazy cache and loca_glyf across clones: cheap
440 // Arc bump, amortises glyph decode across clones of the
441 // same face.
442 last_used: Arc::clone(&self.last_used),
443 is_variable_font: self.is_variable_font,
444 glyph_cache: Arc::clone(&self.glyph_cache),
445 // `LocaGlyfState` is `Clone` — for `Loaded` this is an
446 // `Arc::clone`; for `Deferred` it's an `Arc::clone` of
447 // the bytes + the `OnceLock`, so a clone of a face
448 // that's already decoded glyphs carries the decode.
449 loca_glyf: self.loca_glyf.clone(),
450 space_width: self.space_width,
451 cmap_subtable: self.cmap_subtable.clone(),
452 mock: self.mock.clone(),
453 reverse_glyph_cache: self.reverse_glyph_cache.clone(),
454 // Arc clone — O(1), just bumps refcount; no byte copy.
455 original_bytes: self.original_bytes.clone(),
456 original_index: self.original_index,
457 index_to_cid: self.index_to_cid.clone(),
458 font_type: self.font_type.clone(),
459 font_name: self.font_name.clone(),
460 // HintInstance has mutable interpreter state and is not Clone.
461 // Clones are used for PDF/serialization where hinting isn't needed.
462 hint_instance: None,
463 }
464 }
465 }
466
467 /// Distinguishes TrueType fonts from OpenType CFF fonts.
468 ///
469 /// This affects how glyph outlines are extracted and how the font
470 /// is embedded in PDF documents.
471 #[derive(Debug, Clone, PartialEq, Eq)]
472 pub enum FontType {
473 /// TrueType font with quadratic Bézier outlines in glyf table.
474 TrueType,
475 /// OpenType font with cubic Bézier outlines in CFF table.
476 /// Contains the serialized CFF data for PDF embedding.
477 OpenTypeCFF(Vec<u8>),
478 }
479
480 /// PDF-specific font metrics from HEAD, HHEA, and OS/2 tables.
481 ///
482 /// These metrics are used for PDF font descriptors and accurate
483 /// text positioning in generated PDF documents.
484 #[derive(Debug, Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Hash)]
485 #[repr(C)]
486 pub struct PdfFontMetrics {
487 // -- HEAD table fields --
488 /// Font units per em-square (typically 1000 or 2048).
489 pub units_per_em: u16,
490 /// Font flags (italic, bold, fixed-pitch, etc.).
491 pub font_flags: u16,
492 /// Minimum x-coordinate across all glyphs.
493 pub x_min: i16,
494 /// Minimum y-coordinate across all glyphs.
495 pub y_min: i16,
496 /// Maximum x-coordinate across all glyphs.
497 pub x_max: i16,
498 /// Maximum y-coordinate across all glyphs.
499 pub y_max: i16,
500
501 // -- HHEA table fields --
502 /// Typographic ascender (distance above baseline).
503 pub ascender: i16,
504 /// Typographic descender (distance below baseline, usually negative).
505 pub descender: i16,
506 /// Recommended line gap between lines of text.
507 pub line_gap: i16,
508 /// Maximum horizontal advance width across all glyphs.
509 pub advance_width_max: u16,
510 /// Caret slope rise for italic angle calculation.
511 pub caret_slope_rise: i16,
512 /// Caret slope run for italic angle calculation.
513 pub caret_slope_run: i16,
514
515 // -- OS/2 table fields (0 if table not present) --
516 /// Average width of lowercase letters.
517 pub x_avg_char_width: i16,
518 /// Visual weight class (100-900, 400=normal, 700=bold).
519 pub us_weight_class: u16,
520 /// Visual width class (1-9, 5=normal).
521 pub us_width_class: u16,
522 /// Thickness of strikeout stroke in font units.
523 pub y_strikeout_size: i16,
524 /// Vertical position of strikeout stroke.
525 pub y_strikeout_position: i16,
526 }
527
528 impl Default for PdfFontMetrics {
529 fn default() -> Self {
530 Self::zero()
531 }
532 }
533
534 impl PdfFontMetrics {
535 /// Returns zeroed metrics with `units_per_em` set to 1000 (standard PostScript default)
536 /// to avoid division-by-zero in scaling calculations.
537 #[must_use] pub const fn zero() -> Self {
538 Self {
539 units_per_em: 1000,
540 font_flags: 0,
541 x_min: 0,
542 y_min: 0,
543 x_max: 0,
544 y_max: 0,
545 ascender: 0,
546 descender: 0,
547 line_gap: 0,
548 advance_width_max: 0,
549 caret_slope_rise: 0,
550 caret_slope_run: 0,
551 x_avg_char_width: 0,
552 us_weight_class: 0,
553 us_width_class: 0,
554 y_strikeout_size: 0,
555 y_strikeout_position: 0,
556 }
557 }
558 }
559
560 /// Result of font subsetting operation.
561 ///
562 /// Contains the subsetted font bytes and a mapping from original
563 /// glyph IDs to new glyph IDs in the subset.
564 #[derive(Debug, Clone)]
565 pub struct SubsetFont {
566 /// The subsetted font file bytes (smaller than original).
567 pub bytes: Vec<u8>,
568 /// Mapping: original glyph ID -> (new subset glyph ID, source character).
569 pub glyph_mapping: BTreeMap<u16, (u16, char)>,
570 }
571
572 impl SubsetFont {
573 /// Return the changed text so that when rendering with the subset font (instead of the
574 /// original) the renderer will end up at the same glyph IDs as if we used the original text
575 /// on the original font
576 #[must_use] pub fn subset_text(&self, text: &str) -> String {
577 text.chars()
578 .filter_map(|c| {
579 self.glyph_mapping.values().find_map(|(ngid, ch)| {
580 if *ch == c {
581 char::from_u32(u32::from(*ngid))
582 } else {
583 None
584 }
585 })
586 })
587 .collect()
588 }
589 }
590
591 /// Hash-based equality: two fonts are considered equal if their content hash matches.
592 /// This is a performance optimization — hash collisions are possible but vanishingly
593 /// unlikely (~1/2^64).
594 impl PartialEq for ParsedFont {
595 fn eq(&self, other: &Self) -> bool {
596 self.hash == other.hash
597 }
598 }
599
600 impl Eq for ParsedFont {}
601
602 const FONT_B64_START: &str = "data:font/ttf;base64,";
603
604 impl serde::Serialize for ParsedFont {
605 fn serialize<S: serde::Serializer>(&self, serializer: S) -> Result<S::Ok, S::Error> {
606 use base64::Engine;
607 let s = format!(
608 "{FONT_B64_START}{}",
609 base64::prelude::BASE64_STANDARD.encode(self.to_bytes(None).unwrap_or_default())
610 );
611 s.serialize(serializer)
612 }
613 }
614
615 impl<'de> serde::Deserialize<'de> for ParsedFont {
616 fn deserialize<D: serde::Deserializer<'de>>(
617 deserializer: D,
618 ) -> Result<Self, D::Error> {
619 use base64::Engine;
620 let s = String::deserialize(deserializer)?;
621 let b64 = s.strip_prefix(FONT_B64_START).and_then(|b| base64::prelude::BASE64_STANDARD.decode(b).ok());
622
623 let mut warnings = Vec::new();
624 Self::from_bytes(&b64.unwrap_or_default(), 0, &mut warnings).ok_or_else(|| {
625 serde::de::Error::custom(format!("Font deserialization error: {warnings:?}"))
626 })
627 }
628 }
629
630 impl fmt::Debug for ParsedFont {
631 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
632 f.debug_struct("ParsedFont")
633 .field("hash", &self.hash)
634 .field("font_metrics", &self.font_metrics)
635 .field("num_glyphs", &self.num_glyphs)
636 .field("hhea_table", &self.hhea_table)
637 .field(
638 "hmtx_range",
639 &format_args!("<{} bytes>", self.hmtx_range.1),
640 )
641 .field("maxp_table", &self.maxp_table)
642 .field(
643 "glyph_cache",
644 &format_args!(
645 "{} entries (lazy)",
646 self.glyph_cache.read().map(|m| m.len()).unwrap_or(0),
647 ),
648 )
649 .field("space_width", &self.space_width)
650 .field("cmap_subtable", &self.cmap_subtable)
651 .finish_non_exhaustive()
652 }
653 }
654
655 /// Warning or error message generated during font parsing.
656 #[derive(Debug, Clone, PartialEq, Eq)]
657 pub struct FontParseWarning {
658 /// Severity level of this warning.
659 pub severity: FontParseWarningSeverity,
660 /// Human-readable description of the issue.
661 pub message: String,
662 }
663
664 /// Severity level for font parsing warnings.
665 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
666 pub enum FontParseWarningSeverity {
667 /// Informational message (not an error).
668 Info,
669 /// Warning that may affect font rendering.
670 Warning,
671 /// Error that prevents proper font usage.
672 Error,
673 }
674
675 impl FontParseWarning {
676 /// Creates an info-level message.
677 #[must_use] pub const fn info(message: String) -> Self {
678 Self {
679 severity: FontParseWarningSeverity::Info,
680 message,
681 }
682 }
683
684 /// Creates a warning-level message.
685 #[must_use] pub const fn warning(message: String) -> Self {
686 Self {
687 severity: FontParseWarningSeverity::Warning,
688 message,
689 }
690 }
691
692 /// Creates an error-level message.
693 #[must_use] pub const fn error(message: String) -> Self {
694 Self {
695 severity: FontParseWarningSeverity::Error,
696 message,
697 }
698 }
699 }
700
701 // WEB-LIFT FIX (2026-06-02): a `FontTableProvider` that scans the sfnt table directory
702 // by hand from the raw font bytes. allsorts' `OffsetTableFontProvider` produces garbage
703 // on the remill/web backend: (1) `ReadArray::read_item`'s nested-tuple `TableRecord` read
704 // returns `table_tag = 0` for EVERY record (proven: tags[7]=0x0000 while the bytes there
705 // are 0x68656164 'head'); (2) even a hand-rolled scan added to the *allsorts crate* sees a
706 // bad `self.scope.data()` (the ReadScope fat-pointer mis-lifts through provider
707 // construction, or allsorts-crate code lifts differently). This provider lives in
708 // azul-layout — whose identical byte reads PROVABLY work (the `from_provider` probe read
709 // num_tables=15 from these same `font_bytes`) — and reads the slice directly. KEEP.
710 #[inline]
711 fn manual_be16(d: &[u8], o: usize) -> u32 {
712 (u32::from(d[o]) << 8) | u32::from(d[o + 1])
713 }
714 #[inline]
715 fn manual_be32(d: &[u8], o: usize) -> u32 {
716 (u32::from(d[o]) << 24)
717 | (u32::from(d[o + 1]) << 16)
718 | (u32::from(d[o + 2]) << 8)
719 | u32::from(d[o + 3])
720 }
721
722 struct ManualTableProvider<'a> {
723 data: &'a [u8],
724 dir: usize, // byte offset of the first table record (offset-table base + 12)
725 num: usize, // number of table records
726 }
727
728 impl<'a> ManualTableProvider<'a> {
729 fn new(data: &'a [u8], font_index: usize) -> Option<Self> {
730 if data.len() < 12 {
731 return None;
732 }
733 let base = if manual_be32(data, 0) == 0x7474_6366 {
734 // 'ttcf' (TrueType Collection): the font_index'th offset-table offset.
735 let num_fonts = manual_be32(data, 8) as usize;
736 if font_index >= num_fonts || 12 + font_index * 4 + 4 > data.len() {
737 return None;
738 }
739 manual_be32(data, 12 + font_index * 4) as usize
740 } else {
741 0 // single font: offset table at the start
742 };
743 if base + 12 > data.len() {
744 return None;
745 }
746 Some(ManualTableProvider {
747 data,
748 dir: base + 12,
749 num: manual_be16(data, base + 4) as usize,
750 })
751 }
752 }
753
754 impl FontTableProvider for ManualTableProvider<'_> {
755 fn table_data(
756 &self,
757 tag: u32,
758 ) -> Result<Option<std::borrow::Cow<'_, [u8]>>, allsorts::error::ParseError> {
759 let mut i = 0;
760 while i < self.num {
761 let r = self.dir + i * 16;
762 if r + 16 > self.data.len() {
763 break;
764 }
765 if manual_be32(self.data, r) == tag {
766 let off = manual_be32(self.data, r + 8) as usize;
767 let len = manual_be32(self.data, r + 12) as usize;
768 return Ok(off
769 .checked_add(len)
770 .filter(|&e| e <= self.data.len())
771 .map(|e| std::borrow::Cow::Borrowed(&self.data[off..e])));
772 }
773 i += 1;
774 }
775 Ok(None)
776 }
777
778 fn has_table(&self, tag: u32) -> bool {
779 self.table_data(tag).ok().flatten().is_some()
780 }
781
782 fn table_tags(&self) -> Option<Vec<u32>> {
783 // DIAG (REVERT): sentinel 0xFADE as tags[0] proves THIS provider ran; then
784 // self.num pushes let me see if the usize field survived; then the real reads
785 // show if self.data (slice field) survived the struct move through generics.
786 let mut tags = Vec::with_capacity(self.num + 1);
787 tags.push(0x0000_FADE);
788 let mut i = 0;
789 while i < self.num {
790 let r = self.dir + i * 16;
791 if r + 4 > self.data.len() {
792 break;
793 }
794 tags.push(manual_be32(self.data, r));
795 i += 1;
796 }
797 Some(tags)
798 }
799 }
800
801 impl allsorts::tables::SfntVersion for ManualTableProvider<'_> {
802 fn sfnt_version(&self) -> u32 {
803 let base = self.dir.saturating_sub(12);
804 if base + 4 <= self.data.len() {
805 manual_be32(self.data, base)
806 } else {
807 0
808 }
809 }
810 }
811
812 impl ParsedFont {
813 /// Parse a font from bytes using allsorts
814 ///
815 /// # Arguments
816 /// * `font_bytes` - The font file data
817 /// * `font_index` - Index of the font in a font collection (0 for single fonts)
818 /// * `warnings` - Optional vector to collect parsing warnings
819 ///
820 /// # Returns
821 /// `Some(ParsedFont)` if parsing succeeds, `None` otherwise
822 ///
823 /// Note: Outlines are decoded lazily by `get_or_decode_glyph`;
824 /// `LocaGlyf::load` runs eagerly here. Use `from_bytes_shared`
825 /// for the lazy-LocaGlyf production path.
826 pub fn from_bytes(
827 font_bytes: &[u8],
828 font_index: usize,
829 warnings: &mut Vec<FontParseWarning>,
830 ) -> Option<Self> {
831 // `from_bytes` keeps the eager-LocaGlyf behaviour for the
832 // small number of callers (mainly tests) that don't have
833 // an `Arc<[u8]>` to keep alive for the lazy path.
834 let mut font = Self::from_bytes_internal(font_bytes, font_index, warnings, false)?;
835 // Retain an owned copy of the source bytes so the face can later be
836 // subset/embedded (PDF export, save->parse roundtrips). Callers pass a
837 // borrowed slice that may not outlive us, so we own it here. Mirrors
838 // `from_bytes_shared`, which retains the caller's `Arc<FontBytes>`.
839 if font.original_bytes.is_none() {
840 font.original_bytes = Some(Arc::new(
841 rust_fontconfig::FontBytes::Owned(Arc::from(font_bytes.to_vec())),
842 ));
843 }
844 Some(font)
845 }
846
847 /// Shared implementation of `from_bytes` / `from_bytes_shared`.
848 ///
849 /// `defer_loca_glyf = true` skips the `LocaGlyf::load` call
850 /// here so the caller (`from_bytes_shared`) can install a
851 /// `LocaGlyfState::Deferred` slot that re-parses on first
852 /// glyph decode. Saves the load-then-drop cycle the previous
853 /// arrangement paid (`from_bytes_shared` used to call
854 /// `from_bytes` and immediately replace the loaded `LocaGlyf`
855 /// with a Deferred slot, throwing away ~hundreds of KiB of
856 /// loca+glyf bytes per face for fonts in the chain that get
857 /// loaded but never rasterized).
858 fn from_bytes_internal(
859 font_bytes: &[u8],
860 font_index: usize,
861 warnings: &mut Vec<FontParseWarning>,
862 defer_loca_glyf: bool,
863 ) -> Option<Self> {
864 use allsorts::{binary::read::ReadScope, font_data::FontData};
865 fn provider_err(font_index: usize, e: impl fmt::Display) -> FontParseWarning {
866 FontParseWarning::error(format!(
867 "Failed to get table provider for font index {font_index}: {e}"
868 ))
869 }
870
871 let scope = ReadScope::new(font_bytes);
872 let font_file = match scope.read::<FontData<'_>>() {
873 Ok(ff) => ff,
874 Err(e) => {
875 warnings.push(FontParseWarning::error(format!(
876 "Failed to read font data: {e}"
877 )));
878 return None;
879 }
880 };
881 // FIX (2026-06-02): route OpenType fonts through the CONCRETE provider
882 // (`OffsetTableFontProvider`) instead of `FontData::table_provider`'s
883 // `Box<dyn FontTableProvider>`. On the lifted/web backend the trait-object
884 // VTABLE dispatch (allsorts font_data.rs:45 `self.provider.table_data(tag)`)
885 // mis-lifts: the vtable's fn-pointers are untranslated native addresses, so the
886 // indirect-call dispatcher routes the dyn call to the WRONG `table_data` impl,
887 // which returns a `Cow::Owned` garbage buffer → `HeadTable::read` errors → font
888 // parse returns None → text measures height 0. A concrete provider makes every
889 // `table_data` a DIRECT (monomorphized) call, which lifts correctly. Woff/Woff2
890 // keep the dyn path (they're not used on the web backend's embedded TTF).
891 match font_file {
892 FontData::OpenType(otf) => {
893 // Prefer the hand-rolled provider (reads font_bytes directly) over
894 // allsorts' OffsetTableFontProvider, whose lifted table reads are garbage
895 // on the web backend. Fall back to allsorts only if the manual layout
896 // parse can't recognise the sfnt (e.g. an unusual TTC).
897 if let Some(mp) = ManualTableProvider::new(font_bytes, font_index) {
898 Self::from_provider(mp, font_bytes, font_index, warnings, defer_loca_glyf)
899 } else {
900 match otf.table_provider(font_index) {
901 Ok(p) => Self::from_provider(
902 p,
903 font_bytes,
904 font_index,
905 warnings,
906 defer_loca_glyf,
907 ),
908 Err(e) => {
909 warnings.push(provider_err(font_index, e));
910 None
911 }
912 }
913 }
914 }
915 other => match other.table_provider(font_index) {
916 Ok(p) => {
917 Self::from_provider(p, font_bytes, font_index, warnings, defer_loca_glyf)
918 }
919 Err(e) => {
920 warnings.push(provider_err(font_index, e));
921 None
922 }
923 },
924 }
925 }
926
927 /// Build a `ParsedFont` from a concrete [`FontTableProvider`]. Split out of
928 /// `from_bytes_internal` (2026-06-02) so OpenType fonts use the concrete
929 /// `OffsetTableFontProvider` (direct `table_data` calls that lift correctly on
930 /// the web backend) rather than `FontData::table_provider`'s `Box<dyn>`, whose
931 /// trait-object vtable dispatch mis-lifts (wrong impl → Owned garbage → parse fail).
932 #[allow(clippy::cast_possible_truncation)] // bounded graphics/coord/font/fixed-point/debug-marker cast
933 #[allow(clippy::too_many_lines)] // large but cohesive: single-purpose layout/render/parse routine (one branch per case)
934 fn from_provider<P: FontTableProvider>(
935 provider: P,
936 font_bytes: &[u8],
937 font_index: usize,
938 warnings: &mut Vec<FontParseWarning>,
939 defer_loca_glyf: bool,
940 ) -> Option<Self> {
941 use std::{
942 collections::hash_map::DefaultHasher,
943 hash::{Hash, Hasher},
944 };
945
946 use allsorts::{
947 binary::read::ReadScope,
948 tables::{
949 cmap::{owned::CmapSubtable as OwnedCmapSubtable, CmapSubtable},
950 FontTableProvider, HeadTable, HheaTable, MaxpTable,
951 },
952 tag,
953 };
954
955 // Extract font name from NAME table early (before provider is moved).
956 // WEB-LIFT FIX (2026-06-02): NameTable::string_for_id decodes the NAME strings via
957 // `encoding_rs` (Mac Roman / UTF-16 charset state machines), whose jump-tables
958 // are NOT devirt'd by the remill lift → MISSING_BLOCK trap (proven: trap in
959 // encoding_rs::Decoder::decode_to_utf8). font_name is OPTIONAL metadata (NOT used
960 // for layout/metrics/shaping — those are binary head/hhea/maxp/cmap/glyf), so skip
961 // the NAME-string decode on the web backend to avoid encoding_rs entirely.
962 #[cfg(feature = "web_lift")]
963 let font_name: Option<String> = None;
964 #[cfg(not(feature = "web_lift"))]
965 let font_name = provider.table_data(tag::NAME).ok().and_then(|name_data| {
966 ReadScope::new(&name_data?)
967 .read::<allsorts::tables::NameTable<'_>>()
968 .ok()
969 .and_then(|name_table| {
970 name_table.string_for_id(allsorts::tables::NameTable::POSTSCRIPT_NAME)
971 })
972 });
973
974 // DIAG (2026-06-02, REVERT): pinpoint the web font-parse-fails root — does HEAD
975 // fail because table_data can't find/return the table (directory mis-lift) or
976 // because HeadTable::read errors (table-read mis-lift)? Surfaced via warnings.
977 let head_table = match provider.table_data(tag::HEAD) {
978 Ok(Some(head_cow)) => {
979 // DIAG: is the HEAD table data CORRECT (magicNumber 0x5F0F3CF5 @ off 12 →
980 // HeadTable::read mis-lifts) or WRONG bytes (directory offset mis-lift)?
981 let bb = head_cow.as_ref();
982 let magic = if bb.len() >= 16 {
983 (u32::from(bb[12]) << 24) | (u32::from(bb[13]) << 16)
984 | (u32::from(bb[14]) << 8) | u32::from(bb[15])
985 } else { 0 };
986 if let Ok(h) = ReadScope::new(&head_cow).read::<HeadTable>() { h } else {
987 // DIAG: surface the sliced offset (how wrong) as hex — "HO" + 8 hex
988 // of (head_cow.ptr - font_bytes.ptr). garbage→offset-read mis-lift;
989 // 00000000→base; plausible-but-wrong→record mapping. "RF"=bytes-OK.
990 let m = if magic == 0x5F0F_3CF5 {
991 "RF000000".to_string()
992 } else {
993 let off = (head_cow.as_ref().as_ptr() as usize)
994 .wrapping_sub(font_bytes.as_ptr() as usize);
995 let mut msg = String::new();
996 // B=Borrowed(slice of font_bytes, ptr-arith/base mis-lift) vs
997 // O=Owned(decompressed/copied Vec — wrong path for plain TTF).
998 msg.push(if matches!(head_cow, std::borrow::Cow::Borrowed(_)) { 'B' } else { 'O' });
999 msg.push_str("HO");
1000 let mut sh: i32 = 28;
1001 while sh >= 0 {
1002 let d = ((off >> sh) & 0xf) as u8;
1003 msg.push((if d < 10 { b'0' + d } else { b'a' + d - 10 }) as char);
1004 sh -= 4;
1005 }
1006 msg
1007 };
1008 warnings.push(FontParseWarning::error(m));
1009 return None;
1010 }
1011 }
1012 Ok(None) => {
1013 // DIAG (REVERT): bytes+len+read_item-count+dir all proved OK (N0fr0fc0fg1)
1014 // yet find_table_record(HEAD)=None though 'head' is rec[7] on disk. So
1015 // either read_item's table_tag FIELD is garbage, or tag::HEAD mis-lifts, or
1016 // the u32 == mis-lifts. t7 = tags[7] (should be 0x68656164 'head' low16
1017 // =6164); H = tag::HEAD low16 (should be 6164); f = ANY tag==HEAD via an
1018 // indexed compare loop (NOT .iter().any). "T<4h t7>H<4h HEAD>f<0|1>".
1019 // T6164 H6164 f1 → values+compare OK (won't reach here — HEAD found)
1020 // T6164 H6164 f0 → the u32 == comparison mis-lifts
1021 // T!=6164 → read_item table_tag FIELD garbage (tuple read mis-lift)
1022 // H!=6164 → tag::HEAD const mis-lifts
1023 #[allow(clippy::cast_possible_truncation)] // bounded graphics/coord/font/fixed-point/debug-marker cast
1024 fn hx(m: &mut String, val: u32, nibbles: i32) {
1025 let mut sh = (nibbles - 1) * 4;
1026 while sh >= 0 {
1027 let d = ((val >> sh) & 0xf) as u8;
1028 m.push((if d < 10 { b'0' + d } else { b'a' + d - 10 }) as char);
1029 sh -= 4;
1030 }
1031 }
1032 // DECISIVE: tags[8] (head, file off 124) reads 0 but tags[2] (off 28) is OK.
1033 // Read the SAME offsets from from_provider's LOCAL font_bytes param (proven
1034 // correct at off 4/12). If local@124 = 0x6865 'he' but provider tags[8]=0 ⇒
1035 // STORED-SLICE issue (provider self.data fat-ptr mis-lifts) → read locally.
1036 // If local@124 = 0 ⇒ the font CONST is only PARTIALLY MIRRORED into the wasm
1037 // (deep data-mirror gap) → table data is simply absent. local@93596 (=0x16f9c,
1038 // head TABLE data start) further maps the mirror: 'he'/nonzero vs 0.
1039 let loc124 = if font_bytes.len() >= 126 {
1040 (u32::from(font_bytes[124]) << 8) | u32::from(font_bytes[125])
1041 } else {
1042 0xEEEE
1043 };
1044 let loc_head = if font_bytes.len() >= 93598 {
1045 (u32::from(font_bytes[93596]) << 8) | u32::from(font_bytes[93597])
1046 } else {
1047 0xEEEE
1048 };
1049 let mut m = String::from("L"); // local font_bytes[124..126] (head dir record):
1050 hx(&mut m, loc124 & 0xffff, 4); // 6865 'he' = mirrored; 0000 = not
1051 m.push('H'); // local font_bytes[93596..] (head TABLE data, deep):
1052 hx(&mut m, loc_head & 0xffff, 4);
1053 warnings.push(FontParseWarning::error(m));
1054 return None;
1055 }
1056 Err(_) => {
1057 warnings.push(FontParseWarning::error("HEAD_DATAERR".to_string()));
1058 return None;
1059 }
1060 };
1061
1062 let maxp_table = provider
1063 .table_data(tag::MAXP)
1064 .ok()
1065 .and_then(|maxp_data| ReadScope::new(&maxp_data?).read::<MaxpTable>().ok())
1066 .unwrap_or(MaxpTable {
1067 num_glyphs: 0,
1068 version1_sub_table: None,
1069 });
1070
1071 let num_glyphs = maxp_table.num_glyphs as usize;
1072
1073 // Compute byte offset+length into font_bytes for hmtx/vmtx
1074 // instead of copying the table data. The provider returns a
1075 // borrowed slice for OpenType fonts, so we can derive the
1076 // offset via pointer arithmetic.
1077 let hmtx_range = provider
1078 .table_data(tag::HMTX)
1079 .ok()
1080 .and_then(|cow_opt| {
1081 let cow = cow_opt?;
1082 match cow {
1083 std::borrow::Cow::Borrowed(slice) => {
1084 let base = font_bytes.as_ptr() as usize;
1085 let ptr = slice.as_ptr() as usize;
1086 let offset = ptr.checked_sub(base)?;
1087 if offset + slice.len() <= font_bytes.len() {
1088 Some((offset, slice.len()))
1089 } else {
1090 None
1091 }
1092 }
1093 std::borrow::Cow::Owned(_) => None,
1094 }
1095 })
1096 .unwrap_or((0, 0));
1097
1098 let vmtx_range = provider
1099 .table_data(tag::VMTX)
1100 .ok()
1101 .and_then(|s| {
1102 let slice = s?;
1103 let base = font_bytes.as_ptr() as usize;
1104 let ptr = slice.as_ptr() as usize;
1105 let offset = ptr.checked_sub(base)?;
1106 if offset + slice.len() <= font_bytes.len() {
1107 Some((offset, slice.len()))
1108 } else {
1109 None
1110 }
1111 })
1112 .unwrap_or((0, 0));
1113
1114 // Parse vhea table (same format as hhea, used for vertical metrics)
1115 let vhea_table = provider
1116 .table_data(tag::VHEA)
1117 .ok()
1118 .and_then(|vhea_data| ReadScope::new(&vhea_data?).read::<HheaTable>().ok());
1119
1120 // hhea is required per the OpenType spec; return None if missing
1121 let hhea_table = provider
1122 .table_data(tag::HHEA)
1123 .ok()
1124 .and_then(|hhea_data| ReadScope::new(&hhea_data?).read::<HheaTable>().ok())?;
1125
1126 // Build layout-specific font metrics
1127 let font_metrics = LayoutFontMetrics {
1128 units_per_em: if head_table.units_per_em == 0 {
1129 1000
1130 } else {
1131 head_table.units_per_em
1132 },
1133 ascent: f32::from(hhea_table.ascender),
1134 descent: f32::from(hhea_table.descender),
1135 line_gap: f32::from(hhea_table.line_gap),
1136 x_height: None, // will be populated from OS/2 table via from_font_metrics if available
1137 cap_height: None,
1138 };
1139
1140 // Build PDF-specific font metrics
1141 let pdf_font_metrics =
1142 Self::parse_pdf_font_metrics(font_bytes, font_index, &head_table, &hhea_table);
1143
1144 // Use allsorts LocaGlyf for on-demand outline extraction. We
1145 // *load* LocaGlyf eagerly (it owns ~tens of KiB of loca +
1146 // ~hundreds of KiB of glyf bytes) but we *don't* decode any
1147 // glyph outlines up front — that's the big RSS win. Glyphs
1148 // are decoded by `ParsedFont::get_or_decode_glyph` on first
1149 // access from the CPU/GPU rasterizer.
1150 //
1151 // When `defer_loca_glyf` is set (production lazy path via
1152 // `from_bytes_shared`), we skip `LocaGlyf::load` here too —
1153 // the caller will overwrite the slot with
1154 // `LocaGlyfState::Deferred` carrying the source bytes
1155 // `Arc<[u8]>`, and the load happens on the first
1156 // `get_or_decode_glyph` call. This avoids parsing
1157 // ~hundreds of KiB per face for fonts that get resolved
1158 // into a chain but never actually rasterized (typical
1159 // for fallback fonts in CSS chains).
1160 let has_glyf = provider.has_table(tag::GLYF) && provider.has_table(tag::LOCA);
1161 // Cache `has_gvar` before `provider` gets moved into
1162 // `allsorts::font::Font::new(provider)` further down —
1163 // it's the cheapest way to detect a variable font and
1164 // avoids the borrow-after-move that a later
1165 // `provider.has_table(tag::GVAR)` would incur.
1166 let has_gvar = provider.has_table(tag::GVAR);
1167 let loca_glyf_opt: Option<Arc<std::sync::Mutex<LocaGlyf>>> = if has_glyf
1168 && !defer_loca_glyf
1169 {
1170 match LocaGlyf::load(&provider) {
1171 Ok(lg) => Some(Arc::new(std::sync::Mutex::new(lg))),
1172 Err(e) => {
1173 warnings.push(FontParseWarning::warning(format!(
1174 "Failed to load LocaGlyf: {e} — falling back to hmtx-only"
1175 )));
1176 None
1177 }
1178 }
1179 } else {
1180 None
1181 };
1182
1183 // Lazy `glyph_cache` starts empty; the space-glyph stub
1184 // below pre-inserts gid 0 / space so the shaper's
1185 // cmap-miss fallback has something to render without
1186 // racing with a decode.
1187
1188 let mut font_data_impl = allsorts::font::Font::new(provider).ok()?;
1189
1190 // Create TrueType hinting instance from font tables.
1191 // [az-web-lift] Skip on the web build. The lifted layout never grid-fits glyphs to a
1192 // pixel raster (it measures + ships a display list to JS), so hinting is never used.
1193 // Building it (HintInstance::new) runs the allsorts bytecode Interpreter
1194 // (Interpreter::new + ::dispatch — a large un-devirt'd opcode jump table the remill
1195 // lift can't resolve, plus ~700 op_* fns of closure bloat). This is INDEPENDENT of the
1196 // lift's jump-table devirt: even with a perfect lift, web has no use for hinting, and
1197 // hinted advances are lower-quality output than the plain scaled advance. Native keeps
1198 // real hinting unchanged.
1199 #[cfg(feature = "web_lift")]
1200 let hint_instance: Option<std::sync::Mutex<allsorts::hinting::HintInstance>> = None;
1201 #[cfg(not(feature = "web_lift"))]
1202 let hint_instance = allsorts::hinting::HintInstance::new(
1203 &font_data_impl.font_table_provider
1204 ).ok().flatten().map(std::sync::Mutex::new);
1205
1206 // Stash raw GSUB/GPOS bytes for lazy parse. Typical fonts
1207 // have ~tens of KiB of GSUB + a few-to-tens of KiB of GPOS —
1208 // dwarfed by glyph outlines — so we keep the bytes around
1209 // and only spend `LayoutTable::read` + `new_layout_cache`
1210 // cycles when the shaper actually needs them (via
1211 // `ParsedFont::gsub` / `::gpos`). For an ASCII run where no
1212 // substitution / kerning is required, we skip both entirely.
1213 let gsub_bytes = font_data_impl
1214 .font_table_provider
1215 .table_data(tag::GSUB)
1216 .ok()
1217 .flatten()
1218 .map(std::borrow::Cow::into_owned);
1219 let gpos_bytes = font_data_impl
1220 .font_table_provider
1221 .table_data(tag::GPOS)
1222 .ok()
1223 .flatten()
1224 .map(std::borrow::Cow::into_owned);
1225 let opt_gdef_table = font_data_impl.gdef_table().ok().and_then(|o| o);
1226 let num_glyphs = font_data_impl.num_glyphs();
1227
1228 let opt_kern_table = font_data_impl
1229 .kern_table()
1230 .ok()
1231 .and_then(|s| s);
1232
1233 let cmap_data = font_data_impl.cmap_subtable_data();
1234 let cmap_subtable = ReadScope::new(cmap_data);
1235 let cmap_subtable = cmap_subtable
1236 .read::<CmapSubtable<'_>>()
1237 .ok()
1238 .and_then(|s| s.to_owned());
1239
1240 // Font identity hash — used by `PartialEq` for ParsedFont.
1241 //
1242 // Previously we did `font_bytes.hash(&mut hasher)` over
1243 // the full mmap. That touched every page of the file
1244 // (a 40 MiB `.ttc` walked byte-for-byte) so the "lazy
1245 // mmap" ended up *fully resident* the moment we built
1246 // a `ParsedFont`. Cold RSS jumped ~40 MiB from this
1247 // single line.
1248 //
1249 // The hash doesn't need to be cryptographic — it just
1250 // has to disambiguate two `ParsedFont`s. `(len, first
1251 // 4 KiB, last 4 KiB, font_index)` is plenty unique and
1252 // only faults in the two header / trailer pages, which
1253 // shaping is going to need anyway.
1254 let mut hasher = DefaultHasher::new();
1255 (font_bytes.len() as u64).hash(&mut hasher);
1256 let head_len = font_bytes.len().min(4096);
1257 font_bytes[..head_len].hash(&mut hasher);
1258 let tail_start = font_bytes.len().saturating_sub(4096);
1259 font_bytes[tail_start..].hash(&mut hasher);
1260 font_index.hash(&mut hasher);
1261 let hash = hasher.finish();
1262
1263 let mut font = Self {
1264 hash,
1265 font_metrics,
1266 pdf_font_metrics,
1267 num_glyphs,
1268 hhea_table,
1269 hmtx_range,
1270 vmtx_range,
1271 vhea_table,
1272 maxp_table,
1273 gsub_bytes,
1274 gsub_cache_lazy: std::sync::OnceLock::new(),
1275 gpos_bytes,
1276 gpos_cache_lazy: std::sync::OnceLock::new(),
1277 opt_gdef_table,
1278 opt_kern_table,
1279 cmap_subtable,
1280 last_used: Arc::new(std::sync::atomic::AtomicU64::new(0)),
1281 is_variable_font: has_gvar,
1282 glyph_cache: Arc::new(rust_fontconfig::StLock::new(BTreeMap::new())),
1283 // Eager path: `from_bytes` loaded LocaGlyf immediately
1284 // (or set None if the font has no loca+glyf). Lazy
1285 // callers use `from_bytes_shared` which replaces this
1286 // with `LocaGlyfState::Deferred` before returning.
1287 loca_glyf: LocaGlyfState::Loaded(loca_glyf_opt),
1288 space_width: None,
1289 mock: None,
1290 reverse_glyph_cache: BTreeMap::new(),
1291 // Don't retain the source bytes by default — layout and
1292 // raster don't need them. PDF subsetting / `to_bytes`
1293 // callers opt in via `with_source_bytes`.
1294 original_bytes: None,
1295 original_index: font_index,
1296 index_to_cid: BTreeMap::new(), // Will be filled for CFF fonts
1297 font_type: FontType::TrueType, // Default, will be updated if CFF
1298 font_name,
1299 hint_instance,
1300 };
1301
1302 // Calculate space width
1303 let space_width = font.get_space_width_internal();
1304
1305 // Pre-decode the space glyph straight into the lazy
1306 // `glyph_cache`. Space typically has no outline, so the
1307 // decoder's outline visitor returns nothing useful and
1308 // we'd spin re-decoding it every shape — short-circuit
1309 // here with a hand-rolled record carrying the hmtx
1310 // advance.
1311 let _ = (|| {
1312 let space_gid = font.lookup_glyph_index(' ' as u32)?;
1313 {
1314 // StLock::read() is infallible (Result<_, Infallible>);
1315 // kept in a tight block so the guard drops at scope end.
1316 let Ok(cache) = font.glyph_cache.read();
1317 if cache.contains_key(&space_gid) {
1318 return None;
1319 }
1320 }
1321 let space_width_val = space_width?;
1322 // Only pre-cache when we actually know a non-zero advance. During
1323 // `from_bytes_internal` the source bytes are not attached yet, so
1324 // `hmtx` is unreadable and `get_space_width_internal` reads back 0;
1325 // caching that would pin every space to a 0 advance for the life of
1326 // the face. Skip it and let the space decode lazily once bytes are
1327 // attached (mock fonts that carry a real space width still cache).
1328 if space_width_val == 0 {
1329 return None;
1330 }
1331 let space_record = OwnedGlyph {
1332 bounding_box: OwnedGlyphBoundingBox {
1333 max_x: 0,
1334 max_y: 0,
1335 min_x: 0,
1336 min_y: 0,
1337 },
1338 horz_advance: space_width_val as u16,
1339 outline: Vec::new(),
1340 phantom_points: None,
1341 raw_points: None,
1342 raw_on_curve: None,
1343 raw_contour_ends: None,
1344 instructions: None,
1345 };
1346 {
1347 // StLock::write() is infallible (Result<_, Infallible>).
1348 let Ok(mut cache) = font.glyph_cache.write();
1349 cache.insert(space_gid, Arc::new(space_record));
1350 }
1351 Some(())
1352 })();
1353
1354 font.space_width = space_width;
1355
1356 Some(font)
1357 }
1358
1359 /// Attach the source font bytes to this `ParsedFont`, enabling
1360 /// [`ParsedFont::to_bytes`] and [`ParsedFont::subset`] (both of
1361 /// which the layout / shaping path never calls).
1362 ///
1363 /// Takes an `Arc<FontBytes>` so the same file's bytes can be
1364 /// shared across every face of a `.ttc` at zero extra cost —
1365 /// pair with [`rust_fontconfig::FcFontCache::get_font_bytes`].
1366 /// For ad-hoc PDF callers that have raw heap bytes, wrap them
1367 /// via `Arc::new(FontBytes::Owned(Arc::from(vec)))`.
1368 #[must_use]
1369 pub fn with_source_bytes(mut self, bytes: Arc<rust_fontconfig::FontBytes>) -> Self {
1370 self.original_bytes = Some(bytes);
1371 self
1372 }
1373
1374 /// Lazy-friendly constructor — identical to
1375 /// [`ParsedFont::from_bytes`] except that `LocaGlyf` is
1376 /// **not** loaded during the call. Instead, the supplied
1377 /// `Arc<[u8]>` is retained and `LocaGlyf::load` runs the first
1378 /// time [`get_or_decode_glyph`] needs glyph outlines for this
1379 /// face.
1380 ///
1381 /// Fonts that get resolved into a CSS fallback chain but are
1382 /// never actually rasterized (common on desktop — e.g. every
1383 /// face of HelveticaNeue.ttc loads, but only one or two are
1384 /// shaped) then pay zero loca/glyf cost.
1385 ///
1386 /// Production callers (the reftest harness, `LayoutWindow`,
1387 /// `cpurender`) should prefer this constructor. Tests that
1388 /// inspect `glyph_records_decoded` directly and don't want
1389 /// a lazy path keep using `from_bytes`.
1390 pub fn from_bytes_shared(
1391 bytes: Arc<rust_fontconfig::FontBytes>,
1392 font_index: usize,
1393 warnings: &mut Vec<FontParseWarning>,
1394 ) -> Option<Self> {
1395 // Skip the eager LocaGlyf::load via `defer_loca_glyf=true`
1396 // — saves the load-then-drop cycle the prior arrangement
1397 // paid (when this called `from_bytes`, allocated
1398 // ~hundreds of KiB of loca+glyf bytes, then immediately
1399 // replaced the slot with `Deferred` and dropped them).
1400 // `bytes.as_ref()` derefs FontBytes → &[u8] (mmap or owned
1401 // — same code path).
1402 let mut font = Self::from_bytes_internal(bytes.as_ref(), font_index, warnings, true)?;
1403 font.original_bytes = Some(bytes.clone());
1404 font.loca_glyf = LocaGlyfState::Deferred {
1405 bytes,
1406 font_index,
1407 loaded: Arc::new(std::sync::Mutex::new(None)),
1408 };
1409 Some(font)
1410 }
1411
1412 /// Resolve the current face's `LocaGlyf`, loading it lazily
1413 /// on first call when `loca_glyf` is `Deferred`. Returns
1414 /// `None` when the font has no usable loca+glyf (CFF fonts
1415 /// or parse failures).
1416 fn resolve_loca_glyf(&self) -> Option<Arc<std::sync::Mutex<LocaGlyf>>> {
1417 use allsorts::{
1418 binary::read::ReadScope,
1419 font_data::FontData,
1420 tables::FontTableProvider,
1421 };
1422 match &self.loca_glyf {
1423 LocaGlyfState::Loaded(inner) => inner.clone(),
1424 LocaGlyfState::Deferred { bytes, font_index, loaded } => {
1425 // Fast path: cached LocaGlyf is present.
1426 if let Ok(guard) = loaded.lock() {
1427 if let Some(arc) = guard.as_ref() {
1428 return Some(Arc::clone(arc));
1429 }
1430 }
1431 let _p = crate::probe::Probe::span("resolve_loca_glyf");
1432
1433 // Slow path: parse provider + load LocaGlyf without
1434 // holding the slot's lock (allsorts can take a
1435 // millisecond or two on a fresh load). Re-check
1436 // after acquiring the write lock so a parallel
1437 // decoder doesn't double-load.
1438 let scope = ReadScope::new(bytes.as_slice());
1439 let font_data = scope.read::<FontData<'_>>().ok()?;
1440 let provider = font_data.table_provider(*font_index).ok()?;
1441 // Gate on table presence to match the `from_bytes`
1442 // has_glyf check; avoids a spurious warning on
1443 // CFF fonts that sneak into the Deferred path.
1444 if !provider.has_table(tag::GLYF) || !provider.has_table(tag::LOCA) {
1445 return None;
1446 }
1447 // Zero-copy: keep `glyf` as a view into the already-resident
1448 // (mmap'd) font bytes instead of copying the whole table
1449 // (~20 MB for a large font) onto the heap. `bytes` is the
1450 // exact buffer `provider` reads from, so load_shared can
1451 // anchor the glyf range inside it (falling back to an owned
1452 // copy if it ever can't). Audit §3.3a.
1453 let owner: Arc<dyn AsRef<[u8]> + Send + Sync> = bytes.clone();
1454 let new_arc = LocaGlyf::load_shared(&provider, owner)
1455 .ok()
1456 .map(|lg| Arc::new(std::sync::Mutex::new(lg)))?;
1457
1458 if let Ok(mut guard) = loaded.lock() {
1459 if let Some(existing) = guard.as_ref() {
1460 return Some(Arc::clone(existing));
1461 }
1462 *guard = Some(Arc::clone(&new_arc));
1463 }
1464 Some(new_arc)
1465 }
1466 }
1467 }
1468
1469 /// Source bytes for PDF subsetting / table extraction.
1470 ///
1471 /// Looks in two places:
1472 /// - `original_bytes` (set by [`ParsedFont::with_source_bytes`]
1473 /// for legacy PDF-first construction).
1474 /// - `LocaGlyfState::Deferred.bytes` (set by
1475 /// [`ParsedFont::from_bytes_shared`] — the production lazy
1476 /// path, which already retains an `Arc<[u8]>` for the lazy
1477 /// loca/glyf loader).
1478 ///
1479 /// Returns `None` only for `ParsedFont`s built via the eager
1480 /// `from_bytes` path without an explicit `with_source_bytes`
1481 /// call — i.e. unit tests that load a font and don't touch
1482 /// PDF.
1483 pub fn source_bytes_for_subset(&self) -> Option<Arc<rust_fontconfig::FontBytes>> {
1484 if let Some(bytes) = &self.original_bytes {
1485 return Some(Arc::clone(bytes));
1486 }
1487 if let LocaGlyfState::Deferred { bytes, .. } = &self.loca_glyf {
1488 return Some(Arc::clone(bytes));
1489 }
1490 None
1491 }
1492
1493 /// Read the monotonic-clock nanos timestamp of the most
1494 /// recent [`get_or_decode_glyph`] call on this face, or `0`
1495 /// if it's never been touched.
1496 pub fn last_used_nanos(&self) -> u64 {
1497 self.last_used.load(std::sync::atomic::Ordering::Relaxed)
1498 }
1499
1500 /// Drop the cached `LocaGlyf` for this face if it's
1501 /// `Deferred`-with-bytes-retained — so the next
1502 /// [`get_or_decode_glyph`] re-parses from `bytes`. No-op for
1503 /// `Loaded` faces (no source bytes to fall back to).
1504 ///
1505 /// Used by [`crate::text3::cache::FontManager::evict_unused`]
1506 /// and exposed publicly so embedders can free memory under
1507 /// pressure on fonts they no longer need to render.
1508 pub fn evict_loca_glyf(&self) -> bool {
1509 match &self.loca_glyf {
1510 LocaGlyfState::Deferred { loaded, .. } => {
1511 if let Ok(mut guard) = loaded.lock() {
1512 if guard.is_some() {
1513 *guard = None;
1514 return true;
1515 }
1516 }
1517 false
1518 }
1519 LocaGlyfState::Loaded(_) => false,
1520 }
1521 }
1522
1523 /// Fetch the parsed GSUB cache if this font has one, parsing
1524 /// it from the retained `gsub_bytes` on first access.
1525 ///
1526 /// Moved out of the eager `from_bytes` path because most text
1527 /// runs never trigger GSUB — plain ASCII without ligatures is
1528 /// handled entirely by the cmap + hmtx fast path. Building
1529 /// `LayoutCacheData<GSUB>` up front reserved ~0.5–2 MiB per
1530 /// face just to throw it away on pages that don't shape
1531 /// complex scripts.
1532 pub fn gsub(&self) -> Option<&GsubCache> {
1533 self.gsub_cache_lazy
1534 .get_or_init(|| {
1535 use allsorts::{
1536 binary::read::ReadScope,
1537 layout::{new_layout_cache, LayoutTable, GSUB},
1538 };
1539 let bytes = self.gsub_bytes.as_ref()?;
1540 ReadScope::new(bytes)
1541 .read::<LayoutTable<GSUB>>()
1542 .ok()
1543 .map(new_layout_cache)
1544 })
1545 .as_ref()
1546 }
1547
1548 /// Fetch the parsed GPOS cache if this font has one, parsing
1549 /// it from the retained `gpos_bytes` on first access. See
1550 /// [`ParsedFont::gsub`] for the motivation.
1551 pub fn gpos(&self) -> Option<&GposCache> {
1552 self.gpos_cache_lazy
1553 .get_or_init(|| {
1554 use allsorts::{
1555 binary::read::ReadScope,
1556 layout::{new_layout_cache, LayoutTable, GPOS},
1557 };
1558 let bytes = self.gpos_bytes.as_ref()?;
1559 ReadScope::new(bytes)
1560 .read::<LayoutTable<GPOS>>()
1561 .ok()
1562 .map(new_layout_cache)
1563 })
1564 .as_ref()
1565 }
1566
1567 /// Fetch an `OwnedGlyph` for `gid`, decoding it on first access.
1568 ///
1569 /// Cached in the `Arc<RwLock<…>>` `glyph_cache` so subsequent
1570 /// calls (including across clones of this `ParsedFont`) hit the
1571 /// cache. Returns `None` when `gid >= num_glyphs` or the font
1572 /// has no loca+glyf and no hmtx entry for the glyph. For CFF
1573 /// fonts the returned record has an empty outline and an advance
1574 /// pulled from hmtx — matching the pre-lazy behaviour.
1575 ///
1576 /// Called on the rasterizer hot path; performance budget is a
1577 /// few µs per unique glyph (first hit) and an Arc bump + `BTreeMap`
1578 /// lookup (cache hits). The write lock is held only across the
1579 /// decode, not across the caller's use of the returned Arc.
1580 pub fn get_or_decode_glyph(&self, gid: u16) -> Option<Arc<OwnedGlyph>> {
1581 use std::sync::Arc;
1582 if usize::from(gid) >= self.num_glyphs as usize {
1583 return None;
1584 }
1585 // Bump the LRU timestamp so `FontManager::evict_unused`
1586 // can tell this face is still in use. Cheap atomic store
1587 // (Relaxed — eviction reads the same atomic and tolerates
1588 // a slightly stale value, which only causes "evict, then
1589 // re-load on next access" — never an incorrect render).
1590 self.last_used
1591 .store(monotonic_now_nanos(), std::sync::atomic::Ordering::Relaxed);
1592
1593 // Fast path: cache hit.
1594 {
1595 // StLock::read() is infallible; tight block drops the read
1596 // guard before the write lock below (deadlock avoidance).
1597 let Ok(cache) = self.glyph_cache.read();
1598 if let Some(existing) = cache.get(&gid) {
1599 return Some(Arc::clone(existing));
1600 }
1601 }
1602
1603 // Miss: decode. We drop the read lock before taking the
1604 // write lock to avoid deadlock, and we re-check on the way
1605 // in because another thread may have decoded the same glyph
1606 // in between.
1607 let record = self.decode_glyph_inner(gid);
1608 let arc = Arc::new(record);
1609 {
1610 // StLock::write() is infallible (Result<_, Infallible>).
1611 let Ok(mut cache) = self.glyph_cache.write();
1612 cache
1613 .entry(gid)
1614 .or_insert_with(|| Arc::clone(&arc));
1615 // If another thread beat us to the insert, return theirs
1616 // so all callers observe the same Arc.
1617 if let Some(winner) = cache.get(&gid) {
1618 return Some(Arc::clone(winner));
1619 }
1620 }
1621 Some(arc)
1622 }
1623
1624 /// Eagerly decode every glyph into the lazy `glyph_cache`,
1625 /// restoring the pre-lazy "every glyph is materialised at
1626 /// construction time" behaviour. Used by tests that iterate
1627 /// or compare against reference tooling, and by embedders
1628 /// that want a walkable view without driving every shape
1629 /// through `get_or_decode_glyph`.
1630 ///
1631 /// After `prime_glyph_cache`, callers can use
1632 /// [`ParsedFont::for_each_decoded_glyph`] or
1633 /// [`ParsedFont::glyph_cache_snapshot`] to observe the
1634 /// populated cache.
1635 #[allow(clippy::cast_possible_truncation)] // bounded graphics/coord/font/fixed-point/debug-marker cast
1636 pub fn prime_glyph_cache(&mut self) {
1637 let n = self.num_glyphs as usize;
1638 for glyph_index in 0..n {
1639 let gid = glyph_index as u16;
1640 drop(self.get_or_decode_glyph(gid));
1641 }
1642 }
1643
1644 /// Walk every entry currently in the lazy `glyph_cache`,
1645 /// invoking `f(gid, &OwnedGlyph)` for each. Holds a read
1646 /// lock for the duration; do not call back into the font
1647 /// from `f`. The cache is populated on demand by
1648 /// [`ParsedFont::get_or_decode_glyph`] (and bulk-prefilled
1649 /// by [`ParsedFont::prime_glyph_cache`]).
1650 pub fn for_each_decoded_glyph<F: FnMut(u16, &OwnedGlyph)>(&self, mut f: F) {
1651 {
1652 // StLock::read() is infallible (Result<_, Infallible>).
1653 let Ok(cache) = self.glyph_cache.read();
1654 for (gid, glyph) in cache.iter() {
1655 f(*gid, glyph.as_ref());
1656 }
1657 }
1658 }
1659
1660 /// Snapshot of the currently-decoded glyphs as a
1661 /// `BTreeMap<u16, Arc<OwnedGlyph>>`. Cheap (clones the
1662 /// Arcs, not the records). Used by callers that want to
1663 /// hand the map off across an API boundary; for in-place
1664 /// iteration prefer [`ParsedFont::for_each_decoded_glyph`].
1665 pub fn glyph_cache_snapshot(&self) -> BTreeMap<u16, Arc<OwnedGlyph>> {
1666 self.glyph_cache
1667 .read()
1668 .map(|c| c.clone())
1669 .unwrap_or_default()
1670 }
1671
1672 /// Core decode routine: produces one `OwnedGlyph` for `gid` by
1673 /// locking `loca_glyf` and running allsorts' outline visitor +
1674 /// raw-simple-glyph extraction. Factored out so both
1675 /// [`get_or_decode_glyph`] and [`prime_glyph_cache`] share it.
1676 ///
1677 /// Always returns an `OwnedGlyph` — if anything in the decode
1678 /// chain fails, falls back to an empty-outline record with the
1679 /// `hmtx` advance. This mirrors the pre-lazy behaviour where
1680 /// every gid ended up in `glyph_records_decoded`.
1681 fn hmtx_bytes(&self) -> &[u8] {
1682 let (off, len) = self.hmtx_range;
1683 if len == 0 { return &[]; }
1684 self.original_bytes.as_ref()
1685 .map_or(&[], |b| &b.as_ref()[off..off+len])
1686 }
1687
1688 fn vmtx_bytes(&self) -> &[u8] {
1689 let (off, len) = self.vmtx_range;
1690 if len == 0 { return &[]; }
1691 self.original_bytes.as_ref()
1692 .map_or(&[], |b| &b.as_ref()[off..off+len])
1693 }
1694
1695 #[allow(clippy::cast_possible_wrap)] // bounded graphics/coord/font/fixed-point/debug-marker cast
1696 fn decode_glyph_inner(&self, gid: u16) -> OwnedGlyph {
1697 let _p = crate::probe::Probe::span("decode_glyph");
1698 // [az-web-lift] use get_horizontal_advance (reads hmtx directly on the web build)
1699 // instead of allsorts::glyph_info::advance, whose lifted ReadArray parse has an
1700 // un-devirt'd jump table → MISSING_BLOCK → OOB during measure.
1701 let horz_advance = self.get_horizontal_advance(gid);
1702
1703 let mut record = OwnedGlyph {
1704 horz_advance,
1705 bounding_box: OwnedGlyphBoundingBox {
1706 min_x: 0,
1707 min_y: 0,
1708 max_x: horz_advance as i16,
1709 max_y: 0,
1710 },
1711 outline: Vec::new(),
1712 phantom_points: None,
1713 raw_points: None,
1714 raw_on_curve: None,
1715 raw_contour_ends: None,
1716 instructions: None,
1717 };
1718
1719 // Resolve the `LocaGlyf` for this face. For `Loaded` that's
1720 // a cheap `Arc::clone`; for `Deferred` this is where the
1721 // actual `LocaGlyf::load` happens on first access, paid once
1722 // per face that ever decodes a glyph.
1723 let Some(loca_glyf_arc) = self.resolve_loca_glyf() else {
1724 // No usable loca+glyf → CFF / OpenType-PostScript font
1725 // (Noto Sans/Serif CJK and most .otf). Decode the glyph
1726 // from the `CFF ` table instead; the TrueType-only glyf
1727 // path below can't see these, which left every CFF glyph
1728 // blank on the cpurender/headless path (CJK rendered as
1729 // empty space with the hmtx advance still reserved).
1730 self.decode_cff_glyph_into(gid, &mut record);
1731 return record;
1732 };
1733 let Ok(mut loca_glyf) = loca_glyf_arc.lock() else {
1734 return record;
1735 };
1736
1737 // Visit the outline. If this is a variable font (gvar
1738 // table present) AND we still have source bytes (only
1739 // the `LocaGlyfState::Deferred` path retains them), we
1740 // re-derive a `VariableGlyfContext` here so default-
1741 // instance vs designed-instance differences land in
1742 // the decoded outline. The chained `if let` pattern
1743 // keeps `provider` and `store` in scope for the
1744 // visit, which the borrow checker requires (the
1745 // store's `Cow::Borrowed(&[u8])` tables tie its
1746 // lifetime to the provider).
1747 //
1748 // Eager-`from_bytes` faces (no retained bytes) and
1749 // non-variable fonts skip the var-context machinery
1750 // and decode the default instance — same behaviour as
1751 // before R4.
1752 // [az-web-lift] The lifted web layout NEVER rasterizes (it measures + positions, then
1753 // ships a display list to JS) — so glyph OUTLINES + TrueType hinting raw-points are
1754 // never needed in wasm. Decoding them (allsorts GlyfVisitorContext::visit +
1755 // GlyphOutlineCollector::into_outlines, whose GlyphOutlineOperation match is a 5-arm
1756 // jump table the remill lift doesn't devirtualize → MISSING_BLOCK → OOB) crashes the
1757 // measure pass. Skip BOTH decode passes on the web build; the record keeps its hmtx
1758 // advance/metrics (set above) which is all text measurement needs.
1759 if !cfg!(feature = "web_lift") {
1760 let mut outline_done = false;
1761 if self.is_variable_font {
1762 if let LocaGlyfState::Deferred { bytes, .. } = &self.loca_glyf {
1763 let scope = ReadScope::new(bytes);
1764 if let Ok(font_data) =
1765 scope.read::<FontData<'_>>()
1766 {
1767 if let Ok(provider) = font_data.table_provider(self.original_index) {
1768 if let Ok(store) = VariableGlyfContextStore::read(&provider) {
1769 if let Ok(var_ctx) = VariableGlyfContext::new(&store) {
1770 let mut visitor = GlyfVisitorContext::new(
1771 &mut loca_glyf,
1772 Some(var_ctx),
1773 );
1774 let mut collector = GlyphOutlineCollector::new();
1775 if visitor.visit(gid, None, &mut collector).is_ok() {
1776 record.outline = collector.into_outlines();
1777 let (min_x, min_y, max_x, max_y) =
1778 compute_outline_bbox(&record.outline);
1779 record.bounding_box = OwnedGlyphBoundingBox {
1780 min_x,
1781 min_y,
1782 max_x,
1783 max_y,
1784 };
1785 outline_done = true;
1786 }
1787 }
1788 }
1789 }
1790 }
1791 }
1792 }
1793 if !outline_done {
1794 let mut visitor =
1795 GlyfVisitorContext::new(&mut loca_glyf, None);
1796 let mut collector = GlyphOutlineCollector::new();
1797 if visitor.visit(gid, None, &mut collector).is_ok() {
1798 record.outline = collector.into_outlines();
1799 let (min_x, min_y, max_x, max_y) =
1800 compute_outline_bbox(&record.outline);
1801 record.bounding_box = OwnedGlyphBoundingBox {
1802 min_x,
1803 min_y,
1804 max_x,
1805 max_y,
1806 };
1807 }
1808 }
1809
1810 // Second pass: pull raw SimpleGlyph data for TrueType
1811 // bytecode hinting. LocaGlyf caches the `Arc<Glyph>`
1812 // internally so this lookup is cheap after the first call.
1813 if let Ok(glyph_arc) = loca_glyf.glyph(gid) {
1814 // `is_on_curve` moved onto the `SimpleGlyphFlagExt` trait in
1815 // allsorts 0.17 (SimpleGlyphFlags is now a BitFlags alias).
1816 use allsorts::tables::glyf::SimpleGlyphFlagExt;
1817 if let allsorts::tables::glyf::Glyph::Simple(sg) = glyph_arc.as_ref() {
1818 record.raw_points = Some(
1819 sg.coordinates.iter().map(|(_, pt)| (pt.0, pt.1)).collect(),
1820 );
1821 record.raw_on_curve = Some(
1822 sg.coordinates.iter().map(|(f, _)| f.is_on_curve()).collect(),
1823 );
1824 record.raw_contour_ends = Some(sg.end_pts_of_contours.clone());
1825 record.instructions = Some(sg.instructions.to_vec());
1826 }
1827 }
1828 } // [az-web-lift] end skip glyph outline/hinting decode on web
1829
1830 record
1831 }
1832
1833 /// Decode a single glyph outline from the `CFF ` (OpenType
1834 /// PostScript) table into `record`. Used for fonts with no `glyf`
1835 /// table — `decode_glyph_inner`'s TrueType path returns an empty
1836 /// outline for them, so without this every CFF glyph rasterised as
1837 /// blank on the CPU renderer. Notably this hit ALL CJK text: the
1838 /// installed Noto Sans/Serif CJK fonts are CID-keyed CFF. allsorts'
1839 /// `CFFOutlines` feeds the same `GlyphOutlineCollector` the glyf
1840 /// path uses and resolves CID-keyed local subrs internally.
1841 fn decode_cff_glyph_into(&self, gid: u16, record: &mut OwnedGlyph) {
1842 use allsorts::cff::{outline::CFFOutlines, CFF};
1843
1844 let Some(ref original) = self.original_bytes else {
1845 return;
1846 };
1847 let bytes: &[u8] = original.as_slice();
1848 let Ok(font_data) = ReadScope::new(bytes).read::<FontData<'_>>() else {
1849 return;
1850 };
1851 let Ok(provider) = font_data.table_provider(self.original_index) else {
1852 return;
1853 };
1854 let Ok(Some(cff_data)) = provider.table_data(tag::CFF) else {
1855 return;
1856 };
1857 let Ok(cff) = ReadScope::new(&cff_data).read::<CFF<'_>>() else {
1858 return;
1859 };
1860 let mut outlines = CFFOutlines { table: &cff };
1861 let mut collector = GlyphOutlineCollector::new();
1862 if outlines.visit(gid, None, &mut collector).is_ok() {
1863 record.outline = collector.into_outlines();
1864 let (min_x, min_y, max_x, max_y) = compute_outline_bbox(&record.outline);
1865 record.bounding_box = OwnedGlyphBoundingBox {
1866 min_x,
1867 min_y,
1868 max_x,
1869 max_y,
1870 };
1871 }
1872 }
1873
1874 /// Parse PDF-specific font metrics from HEAD, HHEA, and OS/2 tables
1875 fn parse_pdf_font_metrics(
1876 font_bytes: &[u8],
1877 font_index: usize,
1878 head_table: &allsorts::tables::HeadTable,
1879 hhea_table: &HheaTable,
1880 ) -> PdfFontMetrics {
1881 use allsorts::{
1882 binary::read::ReadScope,
1883 font_data::FontData,
1884 tables::{os2::Os2, FontTableProvider},
1885 tag,
1886 };
1887
1888 let scope = ReadScope::new(font_bytes);
1889 let font_file = scope.read::<FontData<'_>>().ok();
1890 let provider = font_file
1891 .as_ref()
1892 .and_then(|ff| ff.table_provider(font_index).ok());
1893
1894 let os2_table = provider
1895 .as_ref()
1896 .and_then(|p| p.table_data(tag::OS_2).ok())
1897 .and_then(|os2_data| {
1898 let data = os2_data?;
1899 let scope = ReadScope::new(&data);
1900 scope.read_dep::<Os2>(data.len()).ok()
1901 });
1902
1903 // Base metrics from HEAD and HHEA (always present)
1904 let base = PdfFontMetrics {
1905 units_per_em: head_table.units_per_em,
1906 font_flags: head_table.flags,
1907 x_min: head_table.x_min,
1908 y_min: head_table.y_min,
1909 x_max: head_table.x_max,
1910 y_max: head_table.y_max,
1911 ascender: hhea_table.ascender,
1912 descender: hhea_table.descender,
1913 line_gap: hhea_table.line_gap,
1914 advance_width_max: hhea_table.advance_width_max,
1915 caret_slope_rise: hhea_table.caret_slope_rise,
1916 caret_slope_run: hhea_table.caret_slope_run,
1917 ..PdfFontMetrics::zero()
1918 };
1919
1920 // Add OS/2 metrics if available
1921 os2_table
1922 .map_or(base, |os2| PdfFontMetrics {
1923 x_avg_char_width: os2.x_avg_char_width,
1924 us_weight_class: os2.us_weight_class,
1925 us_width_class: os2.us_width_class,
1926 y_strikeout_size: os2.y_strikeout_size,
1927 y_strikeout_position: os2.y_strikeout_position,
1928 ..base
1929 })
1930 }
1931
1932 /// Returns the width of the space character in font units.
1933 ///
1934 /// This is used internally for text layout calculations.
1935 /// Returns `None` if the font has no space glyph or its width cannot be determined.
1936 fn get_space_width_internal(&self) -> Option<usize> {
1937 if let Some(mock) = self.mock.as_ref() {
1938 return mock.space_width;
1939 }
1940 let glyph_index = self.lookup_glyph_index(' ' as u32)?;
1941
1942 // [az-web-lift] use get_horizontal_advance (direct hmtx on web) instead of
1943 // allsorts::glyph_info::advance (un-devirt'd jump table → OOB).
1944 Some(self.get_horizontal_advance(glyph_index) as usize)
1945 }
1946
1947 /// Look up the glyph index for a Unicode codepoint
1948 pub fn lookup_glyph_index(&self, codepoint: u32) -> Option<u16> {
1949 let cmap = self.cmap_subtable.as_ref()?;
1950 cmap.map_glyph(codepoint).ok().flatten()
1951 }
1952
1953 /// Get the horizontal advance width for a glyph in font units.
1954 ///
1955 /// Pulled straight from the `hmtx` table — no glyph-outline
1956 /// decode. Called once per shaped glyph per layout pass, so
1957 /// avoiding the lazy decode here is a meaningful win over
1958 /// routing through `get_or_decode_glyph`.
1959 pub fn get_horizontal_advance(&self, glyph_index: u16) -> u16 {
1960 if let Some(mock) = self.mock.as_ref() {
1961 return mock.glyph_advances.get(&glyph_index).copied().unwrap_or(0);
1962 }
1963 // [az-web-lift] Read the hmtx advance DIRECTLY (a plain longHorMetric table lookup)
1964 // instead of allsorts::glyph_info::advance, whose lifted binary `ReadArray` parse has
1965 // an un-devirt'd jump table → MISSING_BLOCK → OOB during text measure. Identical result
1966 // for non-variable fonts (the web fallback font is non-variable); native keeps the
1967 // allsorts path (variable-font deltas etc.).
1968 #[cfg(feature = "web_lift")]
1969 {
1970 let hmtx = self.hmtx_bytes();
1971 let num = usize::from(self.hhea_table.num_h_metrics);
1972 if num == 0 {
1973 return 0;
1974 }
1975 let idx = (glyph_index as usize).min(num - 1);
1976 let off = idx * 4;
1977 return if off + 2 <= hmtx.len() {
1978 ((hmtx[off] as u16) << 8) | (hmtx[off + 1] as u16)
1979 } else {
1980 0
1981 };
1982 }
1983 #[cfg(not(feature = "web_lift"))]
1984 {
1985 allsorts::glyph_info::advance(
1986 &self.maxp_table,
1987 &self.hhea_table,
1988 self.hmtx_bytes(),
1989 glyph_index,
1990 )
1991 .unwrap_or_default()
1992 }
1993 }
1994
1995 /// Get the hinted advance width in pixels for a glyph at the given ppem.
1996 ///
1997 /// For glyphs with outlines, runs TrueType bytecode hinting to get the
1998 /// grid-fitted advance from phantom points. For glyphs without outlines
1999 /// (e.g. space), rounds the scaled advance to the pixel grid, matching
2000 /// `FreeType`'s behavior.
2001 ///
2002 /// Returns `None` if hinting is not available or fails.
2003 #[allow(clippy::cast_precision_loss)] // bounded graphics/coord/font/fixed-point/debug-marker cast
2004 pub fn get_hinted_advance_px(&self, glyph_index: u16, ppem: u16) -> Option<f32> {
2005 // [az-web-lift] No pixel grid-fitting on the web (measure-only): return None so the
2006 // caller falls back to the plain scaled advance. Hard-cfg (not a runtime `if cfg!`)
2007 // so the whole hinting body — get_or_decode_glyph's outline path AND set_ppem →
2008 // allsorts Interpreter::dispatch (opcode jump table → OOB) — is removed from the lift
2009 // closure entirely. SEPARATE concern from the transpiler's jump-table devirt: web has
2010 // no use for hinted advances regardless of lift quality. Native is unchanged.
2011 #[cfg(feature = "web_lift")]
2012 {
2013 let _ = (glyph_index, ppem);
2014 None
2015 }
2016 #[cfg(not(feature = "web_lift"))]
2017 {
2018 use allsorts::hinting::f26dot6::{compute_scale, F26Dot6};
2019 let glyph = self.get_or_decode_glyph(glyph_index)?;
2020
2021 let upem = self.font_metrics.units_per_em;
2022 if upem == 0 || ppem == 0 {
2023 return None;
2024 }
2025
2026 // Check if we even have a hint instance
2027 let hint_mutex = self.hint_instance.as_ref()?;
2028
2029 let scale = compute_scale(ppem, upem);
2030 // Round the LIVE hmtx advance, not the decoded glyph's cached
2031 // `horz_advance`. The space glyph is eagerly pre-cached during
2032 // `from_bytes_internal` before `original_bytes` is attached, so its
2033 // cached advance can be a stale 0; and this function only ever rounds
2034 // the scaled hmtx advance to the pixel grid anyway (see below), never
2035 // the hinted phantom point. For every correctly-decoded glyph the two
2036 // are identical, so this is a no-op except for the stale-space case.
2037 let hmtx_advance = self.get_horizontal_advance(glyph_index);
2038 let adv_f26dot6 = F26Dot6::from_funits(i32::from(hmtx_advance), scale);
2039
2040 // For glyphs with outline data, run bytecode hinting
2041 if let (Some(raw_points), Some(raw_on_curve), Some(raw_contour_ends)) = (
2042 glyph.raw_points.as_ref(),
2043 glyph.raw_on_curve.as_ref(),
2044 glyph.raw_contour_ends.as_ref(),
2045 ) {
2046 let instructions = glyph.instructions.as_deref().unwrap_or(&[]);
2047 let mut hint = hint_mutex.lock().ok()?;
2048 hint.set_ppem(ppem, f64::from(ppem)).ok()?;
2049 drop(hint);
2050
2051 let points_f26dot6: Vec<(i32, i32)> = raw_points
2052 .iter()
2053 .map(|&(x, y)| {
2054 let sx = F26Dot6::from_funits(i32::from(x), scale);
2055 let sy = F26Dot6::from_funits(i32::from(y), scale);
2056 (sx.to_bits(), sy.to_bits())
2057 })
2058 .collect();
2059 }
2060
2061 // Use the scaled advance rounded to pixel grid, NOT the hinted
2062 // phantom point. Some glyph programs apply ClearType-specific SHPIX
2063 // adjustments to the advance phantom point that are wrong for
2064 // non-ClearType rendering. The rounded scaled advance matches
2065 // FreeType's DEFAULT mode advance output (and, for glyphs without an
2066 // outline such as space, FreeType's phantom-point pre-rounding).
2067 let rounded = (adv_f26dot6.to_bits() + 32) & !63;
2068 Some(rounded as f32 / 64.0)
2069 } // [az-web-lift] end #[cfg(not(web_lift))] hinting body
2070 }
2071
2072 /// Get the number of glyphs in this font
2073 pub const fn num_glyphs(&self) -> u16 {
2074 self.num_glyphs
2075 }
2076
2077 /// Check if this font has a glyph for the given codepoint
2078 pub fn has_glyph(&self, codepoint: u32) -> bool {
2079 self.lookup_glyph_index(codepoint).is_some()
2080 }
2081
2082 /// Get vertical metrics for a glyph (for vertical text layout).
2083 ///
2084 /// Uses vhea+vmtx tables (same binary format as hhea+hmtx).
2085 /// Returns None if font has no vertical metrics tables.
2086 #[allow(clippy::suboptimal_flops)] // mul_add not guaranteed faster/available without target +fma; keep explicit a*b+c
2087 pub fn get_vertical_metrics(
2088 &self,
2089 glyph_id: u16,
2090 ) -> Option<crate::text3::cache::VerticalMetrics> {
2091 let vhea = self.vhea_table.as_ref()?;
2092 if self.vmtx_range.1 == 0 {
2093 return None;
2094 }
2095 let vert_advance = f32::from(allsorts::glyph_info::advance(
2096 &self.maxp_table, vhea, self.vmtx_bytes(), glyph_id,
2097 ).ok()?);
2098
2099 let units_per_em = f32::from(self.font_metrics.units_per_em);
2100 let scale = if units_per_em > 0.0 { 1.0 / units_per_em } else { 0.001 };
2101
2102 // Vertical bearing: approximate from glyph bbox if available
2103 let (bearing_x, bearing_y) = self.get_or_decode_glyph(glyph_id)
2104 .map_or((0.0, 0.0), |g| {
2105 let bbox = &g.bounding_box;
2106 // tsb (top side bearing): origin_y - max_y
2107 // lsb for vertical: center the glyph horizontally
2108 let width = f32::from(bbox.max_x - bbox.min_x);
2109 (-(width / 2.0) * scale, (vert_advance * scale) - (f32::from(bbox.max_y) * scale))
2110 });
2111
2112 Some(crate::text3::cache::VerticalMetrics {
2113 advance: vert_advance * scale,
2114 bearing_x,
2115 bearing_y,
2116 origin_y: self.font_metrics.ascent * scale,
2117 })
2118 }
2119
2120 /// Get layout-specific font metrics
2121 #[allow(clippy::suboptimal_flops)] // mul_add not guaranteed faster/available without target +fma; keep explicit a*b+c
2122 pub fn get_font_metrics(&self) -> LayoutFontMetrics {
2123 // Ensure descent is positive (OpenType may have negative descent)
2124 let descent = if self.font_metrics.descent > 0.0 {
2125 self.font_metrics.descent
2126 } else {
2127 -self.font_metrics.descent
2128 };
2129
2130 LayoutFontMetrics {
2131 ascent: self.font_metrics.ascent,
2132 descent,
2133 line_gap: self.font_metrics.line_gap,
2134 units_per_em: self.font_metrics.units_per_em,
2135 x_height: self.font_metrics.x_height,
2136 cap_height: self.font_metrics.cap_height,
2137 }
2138 }
2139
2140 /// Convert the `ParsedFont` back to bytes using `allsorts::whole_font`
2141 /// This reconstructs the entire font from the parsed data
2142 ///
2143 /// Source bytes come from either the explicit
2144 /// [`ParsedFont::with_source_bytes`] handle (PDF-first
2145 /// construction) *or* the `LocaGlyfState::Deferred` slot
2146 /// installed by [`ParsedFont::from_bytes_shared`]. The
2147 /// production lazy path retains bytes for the lazy `LocaGlyf`
2148 /// loader, so PDF subsetting Just Works without an extra
2149 /// `with_source_bytes` call.
2150 ///
2151 /// # Arguments
2152 /// * `tags` - Optional list of specific table tags to include (None = all tables)
2153 /// # Errors
2154 ///
2155 /// Returns an error string if serializing the font fails.
2156 pub fn to_bytes(&self, tags: Option<&[u32]>) -> Result<Vec<u8>, String> {
2157 let source = self.source_bytes_for_subset().ok_or_else(|| {
2158 "ParsedFont::to_bytes requires source bytes; construct via \
2159 ParsedFont::from_bytes_shared (production lazy path) or \
2160 attach via ParsedFont::with_source_bytes"
2161 .to_string()
2162 })?;
2163 let scope = ReadScope::new(source.as_slice());
2164 let font_file = scope.read::<FontData<'_>>().map_err(|e| e.to_string())?;
2165 let provider = font_file
2166 .table_provider(self.original_index)
2167 .map_err(|e| e.to_string())?;
2168
2169 let tags_to_use = tags.unwrap_or(&[
2170 tag::CMAP,
2171 tag::HEAD,
2172 tag::HHEA,
2173 tag::HMTX,
2174 tag::MAXP,
2175 tag::NAME,
2176 tag::OS_2,
2177 tag::POST,
2178 tag::GLYF,
2179 tag::LOCA,
2180 ]);
2181
2182 whole_font(&provider, tags_to_use).map_err(|e| e.to_string())
2183 }
2184
2185 /// Create a subset font containing only the specified glyph IDs
2186 /// Returns the subset font bytes and a mapping from old to new glyph IDs
2187 ///
2188 /// # Arguments
2189 /// * `glyph_ids` - The glyph IDs to include in the subset (glyph 0/.notdef is always
2190 /// included)
2191 /// * `cmap_target` - Target cmap format (Unicode for web, `MacRoman` for compatibility)
2192 ///
2193 /// # Returns
2194 /// A tuple of (`subset_font_bytes`, `glyph_mapping`) where `glyph_mapping` maps
2195 /// `original_glyph_id` -> (`new_glyph_id`, `original_char`)
2196 #[allow(clippy::cast_possible_truncation)] // bounded graphics/coord/font/fixed-point/debug-marker cast
2197 /// # Errors
2198 ///
2199 /// Returns an error string if subsetting the font fails.
2200 pub fn subset(
2201 &self,
2202 glyph_ids: &[(u16, char)],
2203 cmap_target: CmapTarget,
2204 ) -> Result<(Vec<u8>, BTreeMap<u16, (u16, char)>), String> {
2205 let source = self.source_bytes_for_subset().ok_or_else(|| {
2206 "ParsedFont::subset requires source bytes; construct via \
2207 ParsedFont::from_bytes_shared (production lazy path) or \
2208 attach via ParsedFont::with_source_bytes"
2209 .to_string()
2210 })?;
2211 let scope = ReadScope::new(source.as_slice());
2212 let font_file = scope.read::<FontData<'_>>().map_err(|e| e.to_string())?;
2213 let provider = font_file
2214 .table_provider(self.original_index)
2215 .map_err(|e| e.to_string())?;
2216
2217 // Build glyph mapping: original_id -> (new_id, char)
2218 let glyph_mapping: BTreeMap<u16, (u16, char)> = glyph_ids
2219 .iter()
2220 .enumerate()
2221 .map(|(new_id, &(original_id, ch))| (original_id, (new_id as u16, ch)))
2222 .collect();
2223
2224 // Extract just the glyph IDs for subsetting
2225 let ids: Vec<u16> = glyph_ids.iter().map(|(id, _)| *id).collect();
2226
2227 // Use PDF profile for embedding fonts in PDFs
2228 let font_bytes = allsorts_subset(&provider, &ids, &SubsetProfile::Pdf, cmap_target)
2229 .map_err(|e| format!("Subset error: {e:?}"))?;
2230
2231 Ok((font_bytes, glyph_mapping))
2232 }
2233
2234 /// Get the width of a glyph in font units (internal, unscaled)
2235 pub fn get_glyph_width_internal(&self, glyph_index: u16) -> Option<usize> {
2236 allsorts::glyph_info::advance(
2237 &self.maxp_table,
2238 &self.hhea_table,
2239 self.hmtx_bytes(),
2240 glyph_index,
2241 )
2242 .ok()
2243 .map(|s| s as usize)
2244 }
2245
2246 /// Get the width of the space character (unscaled font units)
2247 #[inline]
2248 pub const fn get_space_width(&self) -> Option<usize> {
2249 self.space_width
2250 }
2251
2252 /// Add glyph-to-text mapping to reverse cache
2253 /// This should be called during text shaping when we know both the source text and
2254 /// resulting glyphs
2255 pub fn cache_glyph_mapping(&mut self, glyph_id: u16, cluster_text: &str) {
2256 self.reverse_glyph_cache
2257 .insert(glyph_id, cluster_text.to_string());
2258 }
2259
2260 /// Get the cluster text that produced a specific glyph ID
2261 /// Returns the original text that was shaped into this glyph (handles ligatures correctly)
2262 pub fn get_glyph_cluster_text(&self, glyph_id: u16) -> Option<&str> {
2263 self.reverse_glyph_cache.get(&glyph_id).map(String::as_str)
2264 }
2265
2266 /// Get the first character from the cluster text for a glyph ID
2267 /// This is useful for PDF `ToUnicode` `CMap` generation which requires single character
2268 /// mappings
2269 pub fn get_glyph_primary_char(&self, glyph_id: u16) -> Option<char> {
2270 self.reverse_glyph_cache
2271 .get(&glyph_id)
2272 .and_then(|text| text.chars().next())
2273 }
2274
2275 /// Clear the reverse glyph cache (useful for memory management)
2276 pub fn clear_glyph_cache(&mut self) {
2277 self.reverse_glyph_cache.clear();
2278 }
2279
2280 /// Get the bounding box size of a glyph (unscaled units) - for PDF
2281 /// Returns (width, height) in font units
2282 pub fn get_glyph_bbox_size(&self, glyph_index: u16) -> Option<(i32, i32)> {
2283 let g = self.get_or_decode_glyph(glyph_index)?;
2284 let glyph_width = i32::from(g.horz_advance);
2285 let glyph_height = i32::from(g.bounding_box.max_y) - i32::from(g.bounding_box.min_y);
2286 Some((glyph_width, glyph_height))
2287 }
2288 }
2289
2290 /// Compute the bounding box from collected glyph outlines.
2291 fn compute_outline_bbox(outlines: &[GlyphOutline]) -> (i16, i16, i16, i16) {
2292 let mut min_x = i16::MAX;
2293 let mut min_y = i16::MAX;
2294 let mut max_x = i16::MIN;
2295 let mut max_y = i16::MIN;
2296 let mut has_points = false;
2297
2298 for outline in outlines {
2299 for op in outline.operations.as_slice() {
2300 let points: &[(i16, i16)] = match op {
2301 GlyphOutlineOperation::MoveTo(m) => &[(m.x, m.y)],
2302 GlyphOutlineOperation::LineTo(l) => &[(l.x, l.y)],
2303 GlyphOutlineOperation::QuadraticCurveTo(q) => {
2304 // Check both control and end point for bbox
2305 min_x = min_x.min(q.ctrl_1_x).min(q.end_x);
2306 min_y = min_y.min(q.ctrl_1_y).min(q.end_y);
2307 max_x = max_x.max(q.ctrl_1_x).max(q.end_x);
2308 max_y = max_y.max(q.ctrl_1_y).max(q.end_y);
2309 has_points = true;
2310 continue;
2311 }
2312 GlyphOutlineOperation::CubicCurveTo(c) => {
2313 min_x = min_x.min(c.ctrl_1_x).min(c.ctrl_2_x).min(c.end_x);
2314 min_y = min_y.min(c.ctrl_1_y).min(c.ctrl_2_y).min(c.end_y);
2315 max_x = max_x.max(c.ctrl_1_x).max(c.ctrl_2_x).max(c.end_x);
2316 max_y = max_y.max(c.ctrl_1_y).max(c.ctrl_2_y).max(c.end_y);
2317 has_points = true;
2318 continue;
2319 }
2320 GlyphOutlineOperation::ClosePath => continue,
2321 };
2322 for &(x, y) in points {
2323 min_x = min_x.min(x);
2324 min_y = min_y.min(y);
2325 max_x = max_x.max(x);
2326 max_y = max_y.max(y);
2327 has_points = true;
2328 }
2329 }
2330 }
2331
2332 if has_points {
2333 (min_x, min_y, max_x, max_y)
2334 } else {
2335 (0, 0, 0, 0)
2336 }
2337 }
2338
2339 #[derive(Debug, Clone)]
2340 pub struct OwnedGlyph {
2341 pub bounding_box: OwnedGlyphBoundingBox,
2342 pub horz_advance: u16,
2343 pub outline: Vec<GlyphOutline>,
2344 pub phantom_points: Option<[Point; 4]>,
2345 /// Raw TrueType points in font units (for hinting). None for composite/CFF glyphs.
2346 pub raw_points: Option<Vec<(i16, i16)>>,
2347 /// On-curve flags for each raw point.
2348 pub raw_on_curve: Option<Vec<bool>>,
2349 /// Contour end-point indices (TrueType).
2350 pub raw_contour_ends: Option<Vec<u16>>,
2351 /// Per-glyph TrueType hinting instructions.
2352 pub instructions: Option<Vec<u8>>,
2353 }
2354
2355 // --- ParsedFontTrait Implementation for ParsedFont ---
2356
2357 impl crate::text3::cache::ShallowClone for ParsedFont {
2358 fn shallow_clone(&self) -> Self {
2359 self.clone() // ParsedFont::clone uses Arc internally, so it's shallow
2360 }
2361 }
2362
2363 impl crate::text3::cache::ParsedFontTrait for ParsedFont {
2364 fn shape_text(
2365 &self,
2366 text: &str,
2367 script: crate::font_traits::Script,
2368 language: crate::font_traits::Language,
2369 direction: crate::font_traits::BidiDirection,
2370 style: &crate::font_traits::StyleProperties,
2371 ) -> Result<Vec<crate::font_traits::Glyph>, crate::font_traits::LayoutError> {
2372 // Call the existing shape_text_for_parsed_font method (defined in default.rs)
2373 crate::text3::default::shape_text_for_parsed_font(
2374 self, text, script, language, direction, style,
2375 )
2376 }
2377
2378 fn get_hash(&self) -> u64 {
2379 self.hash
2380 }
2381
2382 fn get_glyph_size(
2383 &self,
2384 glyph_id: u16,
2385 font_size_px: f32,
2386 ) -> Option<azul_core::geom::LogicalSize> {
2387 self.get_or_decode_glyph(glyph_id).map(|record| {
2388 let units_per_em = f32::from(self.font_metrics.units_per_em);
2389 let scale_factor = if units_per_em > 0.0 {
2390 font_size_px / units_per_em
2391 } else {
2392 0.01
2393 };
2394 let bbox = &record.bounding_box;
2395 azul_core::geom::LogicalSize {
2396 width: f32::from(bbox.max_x - bbox.min_x) * scale_factor,
2397 height: f32::from(bbox.max_y - bbox.min_y) * scale_factor,
2398 }
2399 })
2400 }
2401
2402 fn get_hyphen_glyph_and_advance(&self, font_size: f32) -> Option<(u16, f32)> {
2403 let glyph_id = self.lookup_glyph_index('-' as u32)?;
2404 let advance_units = self.get_horizontal_advance(glyph_id);
2405 let scale_factor = if self.font_metrics.units_per_em > 0 {
2406 font_size / f32::from(self.font_metrics.units_per_em)
2407 } else {
2408 return None;
2409 };
2410 let scaled_advance = f32::from(advance_units) * scale_factor;
2411 Some((glyph_id, scaled_advance))
2412 }
2413
2414 fn get_kashida_glyph_and_advance(&self, font_size: f32) -> Option<(u16, f32)> {
2415 let glyph_id = self.lookup_glyph_index('\u{0640}' as u32)?;
2416 let advance_units = self.get_horizontal_advance(glyph_id);
2417 let scale_factor = if self.font_metrics.units_per_em > 0 {
2418 font_size / f32::from(self.font_metrics.units_per_em)
2419 } else {
2420 return None;
2421 };
2422 let scaled_advance = f32::from(advance_units) * scale_factor;
2423 Some((glyph_id, scaled_advance))
2424 }
2425
2426 fn has_glyph(&self, codepoint: u32) -> bool {
2427 self.lookup_glyph_index(codepoint).is_some()
2428 }
2429
2430 fn get_vertical_metrics(
2431 &self,
2432 glyph_id: u16,
2433 ) -> Option<crate::text3::cache::VerticalMetrics> {
2434 self.get_vertical_metrics(glyph_id)
2435 }
2436
2437 fn get_font_metrics(&self) -> LayoutFontMetrics {
2438 self.font_metrics
2439 }
2440
2441 fn num_glyphs(&self) -> u16 {
2442 self.num_glyphs
2443 }
2444
2445 fn get_space_width(&self) -> Option<usize> {
2446 self.space_width
2447 }
2448 }
2449
2450 /// Build an agg-rust `PathStorage` from an `OwnedGlyph` outline (in font units, Y-up → Y-down).
2451 ///
2452 /// Returns `None` if the glyph has no outline operations (e.g. space).
2453 /// The caller is responsible for applying scale and translation transforms.
2454 #[cfg(feature = "cpurender")]
2455 #[must_use] pub fn build_glyph_path(glyph: &OwnedGlyph) -> Option<agg_rust::path_storage::PathStorage> {
2456 use agg_rust::{basics::PATH_FLAGS_NONE, path_storage::PathStorage};
2457
2458 let mut path = PathStorage::new();
2459 let mut has_ops = false;
2460 for outline in &glyph.outline {
2461 for op in outline.operations.as_slice() {
2462 has_ops = true;
2463 match op {
2464 GlyphOutlineOperation::MoveTo(OutlineMoveTo { x, y }) => {
2465 path.move_to(f64::from(*x), -f64::from(*y));
2466 }
2467 GlyphOutlineOperation::LineTo(OutlineLineTo { x, y }) => {
2468 path.line_to(f64::from(*x), -f64::from(*y));
2469 }
2470 GlyphOutlineOperation::QuadraticCurveTo(OutlineQuadTo {
2471 ctrl_1_x, ctrl_1_y, end_x, end_y,
2472 }) => {
2473 path.curve3(
2474 f64::from(*ctrl_1_x), -f64::from(*ctrl_1_y),
2475 f64::from(*end_x), -f64::from(*end_y),
2476 );
2477 }
2478 GlyphOutlineOperation::CubicCurveTo(OutlineCubicTo {
2479 ctrl_1_x, ctrl_1_y, ctrl_2_x, ctrl_2_y, end_x, end_y,
2480 }) => {
2481 path.curve4(
2482 f64::from(*ctrl_1_x), -f64::from(*ctrl_1_y),
2483 f64::from(*ctrl_2_x), -f64::from(*ctrl_2_y),
2484 f64::from(*end_x), -f64::from(*end_y),
2485 );
2486 }
2487 GlyphOutlineOperation::ClosePath => {
2488 path.close_polygon(PATH_FLAGS_NONE);
2489 }
2490 }
2491 }
2492 }
2493 if !has_ops {
2494 return None;
2495 }
2496 Some(path)
2497 }
2498}