From 064b591a1bc1521af1e8f805f90e1dabc8fd6ac5 Mon Sep 17 00:00:00 2001
From: Wez Furlong <wez@wezfurlong.org>
Date: Thu, 8 Apr 2021 09:12:59 -0700
Subject: [PATCH] fonts: remove ttf_parser, compute coverage from freetype

Replaces the last usage of ttf-parser with calling into freetype.

This removes a source of inconsistency, as ttf-parser doesn't support
all of the things that freetype does.

Notably, this prevents a weird error from blowing up codepoint coverage
calculations on a system where I have helvetica.bdf in my font dir for
long-forgotten reasons.
---
 Cargo.lock                 |  9 +--------
 wezterm-font/Cargo.toml    |  1 -
 wezterm-font/src/db.rs     | 30 +++++++-----------------------
 wezterm-font/src/ftwrap.rs | 32 ++++++++++++++++++++++++++++++++
 4 files changed, 40 insertions(+), 32 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index 7afe670c7..3d3282b5d 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2542,7 +2542,7 @@ version = "0.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9f923fb806c46266c02ab4a5b239735c144bdeda724a50ed058e5226f594cde3"
 dependencies = [
- "ttf-parser 0.6.2",
+ "ttf-parser",
 ]
 
 [[package]]
@@ -3919,12 +3919,6 @@ version = "0.6.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3e5d7cd7ab3e47dda6e56542f4bbf3824c15234958c6e1bd6aaa347e93499fdc"
 
-[[package]]
-name = "ttf-parser"
-version = "0.12.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "85e00391c1f3d171490a3f8bd79999b0002ae38d3da0d6a3a306c754b053d71b"
-
 [[package]]
 name = "typenum"
 version = "1.13.0"
@@ -4432,7 +4426,6 @@ dependencies = [
  "termwiz",
  "thiserror",
  "tinyvec",
- "ttf-parser 0.12.0",
  "unicode-general-category 0.3.0",
  "unicode-segmentation",
  "walkdir",
diff --git a/wezterm-font/Cargo.toml b/wezterm-font/Cargo.toml
index 62513f56f..a162b2186 100644
--- a/wezterm-font/Cargo.toml
+++ b/wezterm-font/Cargo.toml
@@ -22,7 +22,6 @@ rangeset = { path = "../rangeset" }
 termwiz = { path = "../termwiz" }
 thiserror = "1.0"
 tinyvec = "1.1" # Note: constrained by the allsorts crate
-ttf-parser = "0.12"
 unicode-segmentation = "1.7"
 unicode-general-category = "0.3"
 walkdir = "2"
diff --git a/wezterm-font/src/db.rs b/wezterm-font/src/db.rs
index 17a6b67fb..ccad259e3 100644
--- a/wezterm-font/src/db.rs
+++ b/wezterm-font/src/db.rs
@@ -1,11 +1,11 @@
 //! A font-database to keep track of fonts that we've located
 
+use crate::ftwrap::Library;
 use crate::parser::{load_built_in_fonts, parse_and_collect_font_info, FontMatch, ParsedFont};
 use crate::FontDataHandle;
-use anyhow::{anyhow, Context};
+use anyhow::Context;
 use config::{Config, FontAttributes};
 use rangeset::RangeSet;
-use std::borrow::Cow;
 use std::cmp::Ordering;
 use std::collections::{HashMap, HashSet};
 use std::sync::{Arc, Mutex};
@@ -20,28 +20,12 @@ impl Entry {
     /// Parses out the underlying TTF data and produces a RangeSet holding
     /// the set of codepoints for which the font has coverage.
     fn compute_coverage(&self) -> anyhow::Result<RangeSet<u32>> {
-        use ttf_parser::Face;
-        let (data, index) = match &self.handle {
-            FontDataHandle::Memory { data, index, .. } => (data.clone(), *index),
-            FontDataHandle::OnDisk { path, index, .. } => {
-                let data = std::fs::read(path)
-                    .with_context(|| anyhow!("reading font data from {}", path.display()))?;
-                (Cow::Owned(data), *index)
-            }
-        };
+        let lib = Library::new()?;
+        let face = lib
+            .face_from_locator(&self.handle)
+            .with_context(|| format!("freetype parsing {:?}", self.handle))?;
 
-        let face = Face::from_slice(&data, index)
-            .with_context(|| format!("ttf_parser parsing {:?}", self.handle))?;
-        let mut coverage = RangeSet::new();
-
-        for table in face.character_mapping_subtables() {
-            if table.is_unicode() {
-                table.codepoints(|cp| coverage.add(cp));
-                break;
-            }
-        }
-
-        Ok(coverage)
+        Ok(face.compute_coverage())
     }
 
     /// Computes the intersection of the wanted set of codepoints with
diff --git a/wezterm-font/src/ftwrap.rs b/wezterm-font/src/ftwrap.rs
index 74c5c9a7f..0b513b07d 100644
--- a/wezterm-font/src/ftwrap.rs
+++ b/wezterm-font/src/ftwrap.rs
@@ -5,6 +5,7 @@ use crate::parser::ParsedFont;
 use anyhow::{anyhow, Context};
 use config::{configuration, FreeTypeLoadTarget};
 pub use freetype::*;
+use rangeset::RangeSet;
 use std::borrow::Cow;
 use std::convert::TryInto;
 use std::ffi::CStr;
@@ -228,6 +229,37 @@ impl Face {
         unsafe { ((*self.face).style_flags & FT_STYLE_FLAG_ITALIC as FT_Long) != 0 }
     }
 
+    pub fn compute_coverage(&self) -> RangeSet<u32> {
+        let mut coverage = RangeSet::new();
+
+        for encoding in &[
+            FT_Encoding::FT_ENCODING_UNICODE,
+            FT_Encoding::FT_ENCODING_MS_SYMBOL,
+        ] {
+            if unsafe { FT_Select_Charmap(self.face, *encoding) } != 0 {
+                continue;
+            }
+
+            let mut glyph = 0;
+            let mut ucs4 = unsafe { FT_Get_First_Char(self.face, &mut glyph) };
+            while glyph != 0 {
+                coverage.add(ucs4);
+                ucs4 = unsafe { FT_Get_Next_Char(self.face, ucs4, &mut glyph) };
+            }
+
+            if *encoding == FT_Encoding::FT_ENCODING_MS_SYMBOL {
+                // Fontconfig duplicates F000..F0FF to 0000..00FF
+                for ucs4 in 0xf00..0xf100 {
+                    if coverage.contains(ucs4) {
+                        coverage.add(ucs4 - 0xf000);
+                    }
+                }
+            }
+        }
+
+        coverage
+    }
+
     /// This is a wrapper around set_char_size and select_size
     /// that accounts for some weirdness with eg: color emoji
     pub fn set_font_size(&mut self, point_size: f64, dpi: u32) -> anyhow::Result<(f64, f64)> {