summaryrefslogtreecommitdiff
path: root/klm/util/tokenize_piece.hh
diff options
context:
space:
mode:
Diffstat (limited to 'klm/util/tokenize_piece.hh')
-rw-r--r--klm/util/tokenize_piece.hh12
1 files changed, 12 insertions, 0 deletions
diff --git a/klm/util/tokenize_piece.hh b/klm/util/tokenize_piece.hh
index c7e1c863..4a7f5460 100644
--- a/klm/util/tokenize_piece.hh
+++ b/klm/util/tokenize_piece.hh
@@ -54,6 +54,18 @@ class AnyCharacter {
StringPiece chars_;
};
+class AnyCharacterLast {
+ public:
+ explicit AnyCharacterLast(const StringPiece &chars) : chars_(chars) {}
+
+ StringPiece Find(const StringPiece &in) const {
+ return StringPiece(std::find_end(in.data(), in.data() + in.size(), chars_.data(), chars_.data() + chars_.size()), 1);
+ }
+
+ private:
+ StringPiece chars_;
+};
+
template <class Find, bool SkipEmpty = false> class TokenIter : public boost::iterator_facade<TokenIter<Find, SkipEmpty>, const StringPiece, boost::forward_traversal_tag> {
public:
TokenIter() {}