From 9975609bf3b39f0e1cd121995ac49aea30a6c48f Mon Sep 17 00:00:00 2001
From: Zufu Liu <unknown>
Date: Thu, 21 Oct 2021 22:15:57 +1100
Subject: Feature [feature-requests:#1417] Use backward iteration to find space
 / control character and text / punctuation boundaries in SafeSegment as will
 be simpler and faster in almost all cases. Simplify BreakFinder::Next calling
 SafeSegment.

---
 src/PositionCache.cxx | 21 ++++++++++-----------
 1 file changed, 10 insertions(+), 11 deletions(-)

(limited to 'src/PositionCache.cxx')

diff --git a/src/PositionCache.cxx b/src/PositionCache.cxx
index 6370edb33..c9f4e8793 100644
--- a/src/PositionCache.cxx
+++ b/src/PositionCache.cxx
@@ -755,21 +755,20 @@ TextSegment BreakFinder::Next() {
 		}
 		subBreak = prev;
 	}
+
 	// Splitting up a long run from prev to nextBreak in lots of approximately lengthEachSubdivision.
-	// For very long runs add extra breaks after spaces or if no spaces before low punctuation.
 	const int startSegment = subBreak;
-	if ((nextBreak - subBreak) <= lengthEachSubdivision) {
-		subBreak = -1;
-		return TextSegment(startSegment, nextBreak - startSegment);
+	const int remaining = nextBreak - startSegment;
+	int lengthSegment = remaining;
+	if (lengthSegment > lengthEachSubdivision) {
+		lengthSegment = static_cast<int>(pdoc->SafeSegment(std::string_view(&ll->chars[startSegment], lengthEachSubdivision)));
+	}
+	if (lengthSegment < remaining) {
+		subBreak += lengthSegment;
 	} else {
-		subBreak += pdoc->SafeSegment(&ll->chars[subBreak], lengthEachSubdivision);
-		if (subBreak >= nextBreak) {
-			subBreak = -1;
-			return TextSegment(startSegment, nextBreak - startSegment);
-		} else {
-			return TextSegment(startSegment, subBreak - startSegment);
-		}
+		subBreak = -1;
 	}
+	return TextSegment(startSegment, lengthSegment);
 }
 
 bool BreakFinder::More() const noexcept {
-- 
cgit v1.2.3