summaryrefslogtreecommitdiff
path: root/klm/lm/vocab.cc
diff options
context:
space:
mode:
authorMichael Denkowski <michael.j.denkowski@gmail.com>2012-12-22 16:01:23 -0500
committerMichael Denkowski <michael.j.denkowski@gmail.com>2012-12-22 16:01:23 -0500
commit597d89c11db53e91bc011eab70fd613bbe6453e8 (patch)
tree83c87c07d1ff6d3ee4e3b1626f7eddd49c61095b /klm/lm/vocab.cc
parent65e958ff2678a41c22be7171456a63f002ef370b (diff)
parent201af2acd394415a05072fbd53d42584875aa4b4 (diff)
Merge branch 'master' of git://github.com/redpony/cdec
Diffstat (limited to 'klm/lm/vocab.cc')
-rw-r--r--klm/lm/vocab.cc7
1 files changed, 5 insertions, 2 deletions
diff --git a/klm/lm/vocab.cc b/klm/lm/vocab.cc
index 11c27518..fd7f96dc 100644
--- a/klm/lm/vocab.cc
+++ b/klm/lm/vocab.cc
@@ -116,7 +116,9 @@ WordIndex SortedVocabulary::Insert(const StringPiece &str) {
}
*end_ = hashed;
if (enumerate_) {
- strings_to_enumerate_[end_ - begin_].assign(str.data(), str.size());
+ void *copied = string_backing_.Allocate(str.size());
+ memcpy(copied, str.data(), str.size());
+ strings_to_enumerate_[end_ - begin_] = StringPiece(static_cast<const char*>(copied), str.size());
}
++end_;
// This is 1 + the offset where it was inserted to make room for unk.
@@ -126,7 +128,7 @@ WordIndex SortedVocabulary::Insert(const StringPiece &str) {
void SortedVocabulary::FinishedLoading(ProbBackoff *reorder_vocab) {
if (enumerate_) {
if (!strings_to_enumerate_.empty()) {
- util::PairedIterator<ProbBackoff*, std::string*> values(reorder_vocab + 1, &*strings_to_enumerate_.begin());
+ util::PairedIterator<ProbBackoff*, StringPiece*> values(reorder_vocab + 1, &*strings_to_enumerate_.begin());
util::JointSort(begin_, end_, values);
}
for (WordIndex i = 0; i < static_cast<WordIndex>(end_ - begin_); ++i) {
@@ -134,6 +136,7 @@ void SortedVocabulary::FinishedLoading(ProbBackoff *reorder_vocab) {
enumerate_->Add(i + 1, strings_to_enumerate_[i]);
}
strings_to_enumerate_.clear();
+ string_backing_.FreeAll();
} else {
util::JointSort(begin_, end_, reorder_vocab + 1);
}