From 04a084f6c805c4e7c049809d809c026d98be3936 Mon Sep 17 00:00:00 2001
From: Arthur Tazhitdinov <lisnake@gmail.com>
Date: Wed, 14 Jan 2026 19:17:29 +0500
Subject: [PATCH] refactor: Simplify break information handling in
 buildExplicitBreakInfos and enhance clarity in liangBreakIndexes

---
 lib/Epub/Epub/hyphenation/Hyphenator.cpp      | 22 -------------------
 .../Epub/hyphenation/LiangHyphenation.cpp     | 15 +++++++++----
 2 files changed, 11 insertions(+), 26 deletions(-)

diff --git a/lib/Epub/Epub/hyphenation/Hyphenator.cpp b/lib/Epub/Epub/hyphenation/Hyphenator.cpp
index 3436a90b..5ad2bb85 100644
--- a/lib/Epub/Epub/hyphenation/Hyphenator.cpp
+++ b/lib/Epub/Epub/hyphenation/Hyphenator.cpp
@@ -48,28 +48,6 @@ std::vector<Hyphenator::BreakInfo> buildExplicitBreakInfos(const std::vector<Cod
     breaks.push_back({cps[i + 1].byteOffset, isSoftHyphen(cp)});
   }
 
-  if (breaks.empty()) {
-    return breaks;
-  }
-
-  // Sort by byte offset so we can deduplicate sequential markers in-place.
-  std::sort(breaks.begin(), breaks.end(), [](const Hyphenator::BreakInfo& lhs, const Hyphenator::BreakInfo& rhs) {
-    return lhs.byteOffset < rhs.byteOffset;
-  });
-
-  // Deduplicate in-place: merge entries at same offset while retaining "needs hyphen" flag.
-  size_t writePos = 0;
-  for (size_t readPos = 1; readPos < breaks.size(); ++readPos) {
-    if (breaks[readPos].byteOffset == breaks[writePos].byteOffset) {
-      // Merge: explicit hyphen wins over soft hyphen at same offset.
-      breaks[writePos].requiresInsertedHyphen =
-          breaks[writePos].requiresInsertedHyphen || breaks[readPos].requiresInsertedHyphen;
-    } else {
-      breaks[++writePos] = breaks[readPos];
-    }
-  }
-  breaks.resize(writePos + 1);
-
   return breaks;
 }
 
diff --git a/lib/Epub/Epub/hyphenation/LiangHyphenation.cpp b/lib/Epub/Epub/hyphenation/LiangHyphenation.cpp
index 54f82662..e88c953a 100644
--- a/lib/Epub/Epub/hyphenation/LiangHyphenation.cpp
+++ b/lib/Epub/Epub/hyphenation/LiangHyphenation.cpp
@@ -354,36 +354,43 @@ std::vector<size_t> liangBreakIndexes(const std::vector<CodepointInfo>& cps,
     return {};
   }
 
+  // Liang scores: one entry per augmented char (leading/trailing dots included).
   std::vector<uint8_t> scores(augmented.charCount(), 0);
 
+  // Walk every starting character position and stream bytes through the trie.
   for (size_t charStart = 0; charStart < augmented.charByteOffsets.size(); ++charStart) {
-    size_t byteStart = augmented.charByteOffsets[charStart];
+    const size_t byteStart = augmented.charByteOffsets[charStart];
     AutomatonState state = root;
+
     for (size_t cursor = byteStart; cursor < augmented.bytes.size(); ++cursor) {
       AutomatonState next;
       if (!transition(automaton, state, augmented.bytes[cursor], next)) {
-        break;
+        break;  // No more matches for this prefix.
       }
       state = next;
 
       if (state.levels && state.levelsLen > 0) {
         size_t offset = 0;
+        // Each packed byte stores the byte-distance delta and the Liang level digit.
         for (size_t i = 0; i < state.levelsLen; ++i) {
           const uint8_t packed = state.levels[i];
           const size_t dist = static_cast<size_t>(packed / 10);
           const uint8_t level = static_cast<uint8_t>(packed % 10);
+
           offset += dist;
           const size_t splitByte = byteStart + offset;
           if (splitByte >= augmented.byteToCharIndex.size()) {
             continue;
           }
+
           const int32_t boundary = augmented.byteToCharIndex[splitByte];
           if (boundary < 0) {
-            continue;
+            continue;  // Mid-codepoint byte, wait for the next one.
           }
           if (boundary < 2 || boundary + 2 > static_cast<int32_t>(augmented.charCount())) {
-            continue;
+            continue;  // Skip splits that land in the leading/trailing sentinels.
           }
+
           const size_t idx = static_cast<size_t>(boundary);
           if (idx >= scores.size()) {
             continue;