From 6f1fb172db9781bbacb3ff4957b57b0bc2e49704 Mon Sep 17 00:00:00 2001
From: Jackson Gardner <jacksongardner@google.com>
Date: Fri, 18 Aug 2023 15:08:10 -0700
Subject: [PATCH] Allow optional codepoints to be expressed to the font subset
 generator. (#44864)

This is the engine side change to fix https://github.com/flutter/flutter/issues/132711. There will be a subsequent framework change to express the space character as an "optional" character.
---
 tools/font-subset/main.cc | 29 ++++++++++++++++++++++++-----
 tools/font-subset/test.py |  7 +++++++
 2 files changed, 31 insertions(+), 5 deletions(-)

diff --git a/tools/font-subset/main.cc b/tools/font-subset/main.cc
index fc9765b6161cf..346cb2f008eee 100644
--- a/tools/font-subset/main.cc
+++ b/tools/font-subset/main.cc
@@ -13,15 +13,22 @@
 
 #include "hb_wrappers.h"
 
-hb_codepoint_t ParseCodepoint(const std::string& arg) {
+hb_codepoint_t ParseCodepoint(std::string_view arg, bool& optional) {
+  constexpr std::string_view kOptionalPrefix = "optional:";
+  if (arg.substr(0, kOptionalPrefix.length()) == kOptionalPrefix) {
+    optional = true;
+    arg = arg.substr(kOptionalPrefix.length());
+  } else {
+    optional = false;
+  }
   uint64_t value = 0;
   // Check for \u123, u123, otherwise let strtoul work it out.
   if (arg[0] == 'u') {
-    value = strtoul(arg.c_str() + 1, nullptr, 16);
+    value = strtoul(arg.data() + 1, nullptr, 16);
   } else if (arg[0] == '\\' && arg[1] == 'u') {
-    value = strtoul(arg.c_str() + 2, nullptr, 16);
+    value = strtoul(arg.data() + 2, nullptr, 16);
   } else {
-    value = strtoul(arg.c_str(), nullptr, 0);
+    value = strtoul(arg.data(), nullptr, 0);
   }
   if (value == 0 || value > std::numeric_limits<hb_codepoint_t>::max()) {
     std::cerr << "The value '" << arg << "' (" << value
@@ -43,6 +50,10 @@ void Usage() {
                "and must be input as decimal numbers (123), hexadecimal "
                "numbers (0x7B), or unicode hexadecimal characters (\\u7B)."
             << std::endl;
+  std::cout << "Codepoints can be prefixed by the string \"optional:\" to "
+               "specify that the codepoint can be omitted if it isn't found "
+               "in the input font file."
+            << std::endl;
   std::cout << "Input terminates with a newline." << std::endl;
   std::cout
       << "This program will de-duplicate codepoints if the same codepoint is "
@@ -113,13 +124,21 @@ int main(int argc, char** argv) {
     hb_face_collect_unicodes(font_face.get(), actual_codepoints.get());
     std::string raw_codepoint;
     while (std::cin >> raw_codepoint) {
-      auto codepoint = ParseCodepoint(raw_codepoint);
+      bool optional = false;
+      auto codepoint =
+          ParseCodepoint(std::string_view(raw_codepoint), optional);
       if (!codepoint) {
         std::cerr << "Invalid codepoint for " << raw_codepoint << "; exiting."
                   << std::endl;
         return -1;
       }
+
       if (!hb_set_has(actual_codepoints.get(), codepoint)) {
+        if (optional) {
+          // Code point is optional, so omit it.
+          continue;
+        }
+
         std::cerr << "Codepoint " << raw_codepoint
                   << " not found in font, aborting." << std::endl;
         return -1;
diff --git a/tools/font-subset/test.py b/tools/font-subset/test.py
index 540b27c02744f..b8ad5051a6833 100755
--- a/tools/font-subset/test.py
+++ b/tools/font-subset/test.py
@@ -60,6 +60,13 @@
     (False, '1.ttf', MATERIAL_TTF, [r'57348'
                                    ]),  # False because different codepoint
     (True, '2.ttf', MATERIAL_TTF, [r'0xE003', r'0xE004']),
+    (True, '2.ttf', MATERIAL_TTF, [r'0xE003', r'optional:0xE004'
+                                  ]),  # Optional codepoint that is found
+    (True, '2.ttf', MATERIAL_TTF, [
+        r'0xE003',
+        r'0xE004',
+        r'optional:0x12',
+    ]),  # Optional codepoint that is not found
     (True, '2.ttf', MATERIAL_TTF, [
         r'0xE003',
         r'0xE004',