Require minimum length at right hand side of apostrophe too

This commit is contained in:
jpirnay
2026-03-05 22:08:14 +01:00
parent fd84ea92a5
commit b2812c7b24
2 changed files with 13 additions and 3 deletions
+11 -2
View File
@@ -98,6 +98,8 @@ void appendSegmentPatternBreaks(const std::vector<CodepointInfo>& cps, const Lan
void appendApostropheContractionBreaks(const std::vector<CodepointInfo>& cps,
std::vector<Hyphenator::BreakInfo>& outBreaks) {
constexpr size_t kMinLeftSegmentLen = 3;
constexpr size_t kMinRightSegmentLen = 2;
size_t segmentStart = 0;
for (size_t i = 0; i < cps.size(); ++i) {
@@ -111,8 +113,15 @@ void appendApostropheContractionBreaks(const std::vector<CodepointInfo>& cps,
}
}
// Avoid stranding very short clitics like "l'" or "d'" at line end.
if (leftPrefixLen >= 3) {
size_t rightSuffixLen = 0;
for (size_t j = i + 1; j < cps.size() && !isSegmentSeparator(cps[j].value); ++j) {
if (isAlphabetic(cps[j].value)) {
++rightSuffixLen;
}
}
// Avoid stranding short clitics like "l'"/"d'" or tiny suffixes like "'t".
if (leftPrefixLen >= kMinLeftSegmentLen && rightSuffixLen >= kMinRightSegmentLen) {
outBreaks.push_back({cps[i + 1].byteOffset, false});
}
}
+2 -1
View File
@@ -26,7 +26,8 @@ class Hyphenator {
// 2. Apostrophe contractions between letters (e.g. all'improvviso).
// Liang patterns are run per alphabetic segment around apostrophes.
// A direct break at the apostrophe boundary is allowed only when the left
// segment has at least 3 letters to avoid stranding short clitics (e.g. l', d').
// segment has at least 3 letters and the right segment has at least 2 letters,
// avoiding short clitics (e.g. l', d') and short contraction tails (e.g. can't).
// 3. Language-specific Liang patterns (e.g. German de_patterns).
// Example: "Quadratkilometer" -> Qua|drat|ki|lo|me|ter.
// 4. Fallback every-N-chars splitting (only when includeFallback is true AND no