]> git.lyx.org Git - lyx.git/blobdiff - src/TexRow.cpp
Fix text direction issue for InsetInfo in RTL context
[lyx.git] / src / TexRow.cpp
index 24e0f36937d45ff86e4032d6c92ba5f46972a2cb..460a141600194adccca48685bab7af85cac7ef2d 100644 (file)
@@ -242,30 +242,63 @@ void TexRow::append(TexRow other)
 pair<TexRow::TextEntry, TexRow::TextEntry>
 TexRow::getEntriesFromRow(int const row) const
 {
+       // FIXME: Take math entries into account, take table cells into account and
+       //        get rid of the ad hoc special text entry for each row.
+       //
+       // FIXME: A yellow note alone on its paragraph makes the reverse-search on
+       //        the subsequent line inaccurate. Get rid of text entries that
+       //        correspond to no output by delaying their addition, at the level
+       //        of otexrowstream, until a character is actually output.
+       //
        LYXERR(Debug::LATEX, "getEntriesFromRow: row " << row << " requested");
+
        // check bounds for row - 1, our target index
        if (row <= 0)
                return {text_none, text_none};
        size_t const i = static_cast<size_t>(row - 1);
        if (i >= rowlist_.size())
                return {text_none, text_none};
+
        // find the start entry
-       size_t j = i;
-       while (j > 0 && isNone(rowlist_[j].getTextEntry()))
-               --j;
-       TextEntry start = rowlist_[j].getTextEntry();
+       TextEntry const start = [&]() {
+               for (size_t j = i; j > 0; --j) {
+                       if (!isNone(rowlist_[j].getTextEntry()))
+                               return rowlist_[j].getTextEntry();
+                       // Check the absence of begin_document at row j. The begin_document row
+                       // entry is used to prevent mixing of body and preamble.
+                       for (RowEntry entry : rowlist_[j])
+                               if (entry.type == begin_document)
+                                       return text_none;
+               }
+               return text_none;
+       } ();
+
        // find the end entry
-       j = i + 1;
-       while (j < rowlist_.size() && isNone(rowlist_[j].getTextEntry()))
-               ++j;
-       TextEntry end =
-               (j < rowlist_.size()) ? rowlist_[j].getTextEntry()
-                                     : TextEntry{start.id, -1}; // last position
+       TextEntry end = [&]() {
+               if (isNone(start))
+                       return text_none;
+               // select up to the last position of the starting paragraph as a
+               // fallback
+               TextEntry last_pos = {start.id, -1};
+               // find the next occurence of paragraph start.id
+               for (size_t j = i + 1; j < rowlist_.size(); ++j) {
+                       for (RowEntry entry : rowlist_[j]) {
+                               if (entry.type == begin_document)
+                                       // what happens in the preamble remains in the preamble
+                                       return last_pos;
+                               if (entry.type == text_entry && entry.text.id == start.id)
+                                       return entry.text;
+                       }
+               }
+               return last_pos;
+       } ();
+
        // The following occurs for a displayed math inset for instance (for good
        // reasons involving subtleties of the algorithm in getRowFromDocIterator).
        // We want this inset selected.
        if (start.id == end.id && start.pos == end.pos)
                ++end.pos;
+
        return {start, end};
 }
 
@@ -494,6 +527,7 @@ TexRow::RowListIterator TexRow::end() const
 
 pair<int,int> TexRow::rowFromDocIterator(DocIterator const & dit) const
 {
+       // Do not change anything in this algorithm if unsure.
        bool beg_found = false;
        bool end_is_next = true;
        int end_offset = 1;