Fix LRC same-line collapsing sometimes deleting parts of a line

Fixes #354

src/lyric_auto_edit.cpp

@@ -27,7 +27,7 @@ std::optional<LyricData> auto_edit::RunAutoEdit(AutoEditType type, const LyricDa
 
 std::optional<LyricData> auto_edit::ReplaceHtmlEscapedChars(const LyricData& lyrics)
 {
-    LyricDataUnstructured unstructured = parsers::lrc::serialise(lyrics);
+    std::string text = from_tstring(parsers::lrc::expand_text(lyrics, false));
     std::pair<std::string_view, char> replacements[] =
     {
         {"&amp;", '&'},
@@ -41,12 +41,12 @@ std::optional<LyricData> auto_edit::ReplaceHtmlEscapedChars(const LyricData& lyr
     for(auto [escaped, replacement] : replacements)
     {
         size_t current_index = 0;
-        while(current_index < unstructured.text.length())
+        while(current_index < text.length())
         {
-            size_t next_index = unstructured.text.find(escaped, current_index);
+            size_t next_index = text.find(escaped, current_index);
             if(next_index == std::string::npos) break;
 
-            unstructured.text.replace(next_index, escaped.length(), 1, replacement);
+            text.replace(next_index, escaped.length(), 1, replacement);
             current_index = next_index + 1;
             replace_count++;
         }
@@ -55,7 +55,7 @@ std::optional<LyricData> auto_edit::ReplaceHtmlEscapedChars(const LyricData& lyr
 
     if(replace_count > 0)
     {
-        return {parsers::lrc::parse(unstructured)};
+        return {parsers::lrc::parse(lyrics, text)};
     }
     else
     {
@@ -253,32 +253,31 @@ std::optional<LyricData> auto_edit::FixMalformedTimestamps(const LyricData& lyri
         }
     };
 
-    LyricDataUnstructured new_lyrics = parsers::lrc::serialise(lyrics);
-
+    std::string text = from_tstring(parsers::lrc::expand_text(lyrics, false));
     int change_count = 0;
-    size_t current_index = new_lyrics.text.find('[');
+    size_t current_index = text.find('[');
     while(current_index != std::string::npos)
     {
         bool changed = false;
 
-        size_t end_index = new_lyrics.text.find(']', current_index);
+        size_t end_index = text.find(']', current_index);
         if(end_index == std::string::npos)
         {
             break;
         }
-        std::string_view tag {new_lyrics.text.c_str() + current_index, end_index - current_index + 1};
+        std::string_view tag {text.c_str() + current_index, end_index - current_index + 1};
         changed |= fix_decimal_separator(tag);
 
         if(changed)
         {
             change_count++;
         }
-        current_index = new_lyrics.text.find('[', current_index+1);
+        current_index = text.find('[', current_index+1);
     }
 
     if(change_count > 0)
     {
-        return {parsers::lrc::parse(new_lyrics)};
+        return {parsers::lrc::parse(lyrics, text)};
     }
     else
     {

src/lyric_io.cpp

@@ -75,42 +75,7 @@ bool io::save_lyrics(metadb_handle_ptr track, const metadb_v2_rec_t& track_info,
         return false;
     }
 
-    std::string text;
-    if(lyrics.IsTimestamped() && preferences::saving::merge_equivalent_lrc_lines())
-    {
-        LyricData merged_lyrics = lyrics;
-        const auto lexicographic_sort = [](const auto& lhs, const auto& rhs){ return lhs.text < rhs.text; };
-        std::stable_sort(merged_lyrics.lines.begin(), merged_lyrics.lines.end(), lexicographic_sort);
-        std::vector<LyricDataLine>::iterator equal_begin = merged_lyrics.lines.begin();
-
-        while(equal_begin != merged_lyrics.lines.end())
-        {
-            std::vector<LyricDataLine>::iterator equal_end = equal_begin + 1;
-            while((equal_end != merged_lyrics.lines.end()) && (equal_begin->text == equal_end->text) && (equal_end->timestamp != DBL_MAX))
-            {
-                equal_end++;
-            }
-
-            // NOTE: We don't need to move equal_begin back one because we don't add
-            //       the first timestamp to the string. That'll happen as part of the
-            //       normal printing below.
-            for(auto iter=equal_end-1; iter!=equal_begin; iter--)
-            {
-                equal_begin->text = to_tstring(parsers::lrc::print_timestamp(iter->timestamp)) + equal_begin->text;
-            }
-            equal_begin = merged_lyrics.lines.erase(equal_begin+1, equal_end);
-        }
-
-        const auto timestamp_sort = [](const auto& lhs, const auto& rhs){ return lhs.timestamp < rhs.timestamp; };
-        std::stable_sort(merged_lyrics.lines.begin(), merged_lyrics.lines.end(), timestamp_sort);
-
-        text = from_tstring(parsers::lrc::expand_text(merged_lyrics));
-    }
-    else
-    {
-        text = from_tstring(parsers::lrc::expand_text(lyrics));
-    }
-
+    const std::string text = from_tstring(parsers::lrc::expand_text(lyrics, preferences::saving::merge_equivalent_lrc_lines()));
     try
     {
         std::string output_path = source->save(track, track_info, lyrics.IsTimestamped(), text, allow_overwrite, abort);

src/main.cpp

@@ -53,6 +53,7 @@ void OpenLyricsVersion::get_about_message(pfc::string_base & out)
     "- Fix album art backgrounds disappearing when playing the same track twice\n"
     "- Fix the 'encoding' tag in LRC files being considered a lyric line\n"
     "- Fix rounding errors causing timestamps to randomly change when saving\n"
+    "- Fix LRC same-line collapsing sometimes deleting parts of a line\n"
     "- Many, many minor non-functional internal code improvements\n"
     "\n";
     out += "Version 1.9 (2024-06-12):\n"

src/parsers.h

@@ -18,9 +18,9 @@ namespace lrc
     bool try_parse_timestamp(std::string_view tag, double& out_timestamp);
 
     LyricData parse(const LyricDataUnstructured& input);
-    LyricDataUnstructured serialise(const LyricData& input);
+    LyricData parse(const LyricDataCommon& metadata, std::string text_utf8);
 
-    std::tstring expand_text(const LyricData& data);
+    std::tstring expand_text(const LyricData& data, bool merge_equivalent_lrc_lines);
 } // namespace lrc
 
 } // namespace parsers