From 703a824c1bb7fc41535c9515c5a2209d04899d19 Mon Sep 17 00:00:00 2001
From: Jose Colon Rodriguez
Date: Wed, 14 Feb 2024 12:02:02 -0400
Subject: Code reorg; Added UCD

---
 src/Grapheme.zig | 40 +++++++++++++++++++++++++++++-----------
 1 file changed, 29 insertions(+), 11 deletions(-)

(limited to 'src/Grapheme.zig')

diff --git a/src/Grapheme.zig b/src/Grapheme.zig
index d739159..9f09255 100644
--- a/src/Grapheme.zig
+++ b/src/Grapheme.zig
@@ -308,29 +308,36 @@ test "Segmentation ZWJ and ZWSP emoji sequences" {
 }
 
 // Grapheme break state.
-fn hasXpic(state: *const u3) bool {
+// Extended Pictographic (emoji)
+inline fn hasXpic(state: *const u3) bool {
     return state.* & 1 == 1;
 }
-
-fn setXpic(state: *u3) void {
+inline fn setXpic(state: *u3) void {
     state.* |= 1;
 }
-
-fn unsetXpic(state: *u3) void {
+inline fn unsetXpic(state: *u3) void {
     state.* ^= 1;
 }
-
-fn hasRegional(state: *const u3) bool {
+// Regional Indicatior (flags)
+inline fn hasRegional(state: *const u3) bool {
     return state.* & 2 == 2;
 }
-
-fn setRegional(state: *u3) void {
+inline fn setRegional(state: *u3) void {
     state.* |= 2;
 }
-
-fn unsetRegional(state: *u3) void {
+inline fn unsetRegional(state: *u3) void {
     state.* ^= 2;
 }
+// Indic Conjunct
+inline fn hasIndic(state: *const u3) bool {
+    return state.* & 2 == 4;
+}
+inline fn setIndic(state: *u3) void {
+    state.* |= 4;
+}
+inline fn unsetIndic(state: *u3) void {
+    state.* ^= 4;
+}
 
 /// `graphemeBreak` returns true only if a grapheme break point is required
 /// between `cp1` and `cp2`. `state` should start out as 0. If calling
@@ -345,6 +352,8 @@ pub fn graphemeBreak(
 ) bool {
     // GB11: Emoji Extend* ZWJ x Emoji
     if (!hasXpic(state) and emoji.isExtendedPictographic(cp1)) setXpic(state);
+    // GB9c: Indic Conjunct Break
+    // if (!hasIndic(state) and indic.isConsonant(cp1)) setIndic(state);
 
     // GB3: CR x LF
     if (cp1 == '\r' and cp2 == '\n') return false;
@@ -400,6 +409,15 @@ pub fn graphemeBreak(
         return false;
     }
 
+    // GB9c: Indic Conjunct Break
+    // if (hasIndic(state) and
+    //     indic.isLinker(cp1) and
+    //     indic.isConsonant(cp2))
+    // {
+    //     unsetIndic(state);
+    //     return false;
+    // }
+
     return true;
 }
 
-- 
cgit v1.2.3