From f6767c857f85d90ad7017ba416b159bcb2d1907f Mon Sep 17 00:00:00 2001
From: Arthur Tazhitdinov <lisnake@gmail.com>
Date: Fri, 26 Dec 2025 06:03:38 +0500
Subject: [PATCH] Rename trimTrailingPunctuation to trimSurroundingPunctuation
 and update logic to remove surrounding punctuation; add explicit hyphen
 handling in breakOffsets function.

---
 .../Epub/hyphenation/HyphenationCommon.cpp    |  5 ++-
 lib/Epub/Epub/hyphenation/HyphenationCommon.h |  2 +-
 lib/Epub/Epub/hyphenation/Hyphenator.cpp      | 39 ++++++++++++++++++-
 3 files changed, 43 insertions(+), 3 deletions(-)

diff --git a/lib/Epub/Epub/hyphenation/HyphenationCommon.cpp b/lib/Epub/Epub/hyphenation/HyphenationCommon.cpp
index fb8f26b1..c7eeb691 100644
--- a/lib/Epub/Epub/hyphenation/HyphenationCommon.cpp
+++ b/lib/Epub/Epub/hyphenation/HyphenationCommon.cpp
@@ -92,7 +92,10 @@ bool isPunctuation(const uint32_t cp) {
   }
 }
 
-void trimTrailingPunctuation(std::vector<CodepointInfo>& cps) {
+void trimSurroundingPunctuation(std::vector<CodepointInfo>& cps) {
+  while (!cps.empty() && isPunctuation(cps.front().value)) {
+    cps.erase(cps.begin());
+  }
   while (!cps.empty() && isPunctuation(cps.back().value)) {
     cps.pop_back();
   }
diff --git a/lib/Epub/Epub/hyphenation/HyphenationCommon.h b/lib/Epub/Epub/hyphenation/HyphenationCommon.h
index 7e6f4eee..9a6b69aa 100644
--- a/lib/Epub/Epub/hyphenation/HyphenationCommon.h
+++ b/lib/Epub/Epub/hyphenation/HyphenationCommon.h
@@ -28,6 +28,6 @@ bool isCyrillicConsonant(uint32_t cp);
 bool isAlphabetic(uint32_t cp);
 bool isVowel(uint32_t cp);
 bool isPunctuation(uint32_t cp);
-void trimTrailingPunctuation(std::vector<CodepointInfo>& cps);
+void trimSurroundingPunctuation(std::vector<CodepointInfo>& cps);
 
 Script detectScript(const std::vector<CodepointInfo>& cps);
diff --git a/lib/Epub/Epub/hyphenation/Hyphenator.cpp b/lib/Epub/Epub/hyphenation/Hyphenator.cpp
index cecdc38a..2ac3bf9d 100644
--- a/lib/Epub/Epub/hyphenation/Hyphenator.cpp
+++ b/lib/Epub/Epub/hyphenation/Hyphenator.cpp
@@ -48,6 +48,32 @@ std::vector<CodepointInfo> collectCodepoints(const std::string& word) {
   return cps;
 }
 
+bool isExplicitHyphen(const uint32_t cp) { return cp == '-' || cp == 0x2010; }
+
+std::vector<size_t> collectExplicitHyphenIndexes(const std::vector<CodepointInfo>& cps) {
+  std::vector<size_t> indexes;
+  for (size_t i = 0; i < cps.size(); ++i) {
+    if (!isExplicitHyphen(cps[i].value)) {
+      continue;
+    }
+    if (i == 0 || i + 1 >= cps.size()) {
+      continue;
+    }
+    if (!isAlphabetic(cps[i - 1].value) || !isAlphabetic(cps[i + 1].value)) {
+      continue;
+    }
+    const size_t breakIndex = i + 1;
+    if (breakIndex >= cps.size()) {
+      continue;
+    }
+    if (breakIndex == 0) {
+      continue;
+    }
+    indexes.push_back(breakIndex);
+  }
+  return indexes;
+}
+
 // Rejects words containing punctuation or digits unless forced.
 bool hasOnlyAlphabetic(const std::vector<CodepointInfo>& cps) {
   if (cps.empty()) {
@@ -93,11 +119,22 @@ std::vector<size_t> Hyphenator::breakOffsets(const std::string& word, const bool
   }
 
   auto cps = collectCodepoints(word);
-  trimTrailingPunctuation(cps);
+  trimSurroundingPunctuation(cps);
   if (cps.size() < MIN_PREFIX_CP + MIN_SUFFIX_CP) {
     return {};
   }
 
+  if (auto explicitIndexes = collectExplicitHyphenIndexes(cps); !explicitIndexes.empty()) {
+    std::sort(explicitIndexes.begin(), explicitIndexes.end());
+    explicitIndexes.erase(std::unique(explicitIndexes.begin(), explicitIndexes.end()), explicitIndexes.end());
+    std::vector<size_t> byteOffsets;
+    byteOffsets.reserve(explicitIndexes.size());
+    for (const size_t idx : explicitIndexes) {
+      byteOffsets.push_back(byteOffsetForIndex(cps, idx));
+    }
+    return byteOffsets;
+  }
+
   std::vector<size_t> indexes = hasOnlyAlphabetic(cps) ? collectBreakIndexes(cps) : std::vector<size_t>();
   if (includeFallback) {
     for (size_t idx = MIN_PREFIX_CP; idx + MIN_SUFFIX_CP <= cps.size(); ++idx) {