From e3082e64b3ab8a8aa0777d63be69eb8b6d50a654 Mon Sep 17 00:00:00 2001
From: Sam Atman
Date: Tue, 8 Jul 2025 12:12:20 -0400
Subject: Add Words.zig example to README

---
 src/Words.zig      | 17 +++++++++++++++++
 src/code_point.zig |  3 +++
 2 files changed, 20 insertions(+)

(limited to 'src')

diff --git a/src/Words.zig b/src/Words.zig
index af82562..617c34d 100644
--- a/src/Words.zig
+++ b/src/Words.zig
@@ -674,6 +674,23 @@ test "ext_pict" {
     try testing.expect(ext_pict.isMatch("\u{2701}"));
 }
 
+test "Words" {
+    const wb = try Words.init(testing.allocator);
+    defer wb.deinit(testing.allocator);
+    const word_str = "Metonym   Μετωνύμιο メトニム";
+    var w_iter = wb.iterator(word_str);
+    try testing.expectEqualStrings("Metonym", w_iter.next().?.bytes(word_str));
+    // Spaces are "words" too!
+    try testing.expectEqualStrings("   ", w_iter.next().?.bytes(word_str));
+    const in_greek = w_iter.next().?;
+    for (in_greek.offset..in_greek.offset + in_greek.len) |i| {
+        const at_index = wb.wordAtIndex(word_str, i).bytes(word_str);
+        try testing.expectEqualStrings("Μετωνύμιο", at_index);
+    }
+    _ = w_iter.next();
+    try testing.expectEqualStrings("メトニム", w_iter.next().?.bytes(word_str));
+}
+
 test wordAtIndex {
     const wb = try Words.init(testing.allocator);
     defer wb.deinit(testing.allocator);
diff --git a/src/code_point.zig b/src/code_point.zig
index 16648af..7a638af 100644
--- a/src/code_point.zig
+++ b/src/code_point.zig
@@ -121,6 +121,9 @@ pub fn decodeAtCursor(bytes: []const u8, cursor: *uoffset) ?CodePoint {
     }
     if (st == RUNE_REJECT or cursor.* == bytes.len) {
         @branchHint(.cold);
+        // This, and the branch below, detect truncation, the
+        // only invalid state handled differently by the Maximal
+        // Subparts algorithm.
         if (state_dfa[@intCast(u8dfa[byte])] == RUNE_REJECT) {
             cursor.* -= 2; // +1
             return .{
-- 
cgit v1.2.3