diff options
author | Paul Baltescu <pauldb89@gmail.com> | 2013-02-01 16:11:10 +0000 |
---|---|---|
committer | Paul Baltescu <pauldb89@gmail.com> | 2013-02-01 16:11:10 +0000 |
commit | 252fb164c208ec8f3005f8a652eb3b48c0644e3d (patch) | |
tree | 7199cb668e77ef89c7bcccb37d70554e3b52c2a5 /extractor/phrase_builder.cc | |
parent | 4ab84a0be28fdb6c0c421fe5ba5e09cfa298f2d1 (diff) |
Second working commit.
Diffstat (limited to 'extractor/phrase_builder.cc')
-rw-r--r-- | extractor/phrase_builder.cc | 24 |
1 files changed, 24 insertions, 0 deletions
diff --git a/extractor/phrase_builder.cc b/extractor/phrase_builder.cc index 7f3447e5..c4e0c2ed 100644 --- a/extractor/phrase_builder.cc +++ b/extractor/phrase_builder.cc @@ -19,3 +19,27 @@ Phrase PhraseBuilder::Build(const vector<int>& symbols) { } return phrase; } + +Phrase PhraseBuilder::Extend(const Phrase& phrase, bool start_x, bool end_x) { + vector<int> symbols = phrase.Get(); + int num_nonterminals = 0; + if (start_x) { + num_nonterminals = 1; + symbols.insert(symbols.begin(), + vocabulary->GetNonterminalIndex(num_nonterminals)); + } + + for (size_t i = start_x; i < symbols.size(); ++i) { + if (vocabulary->IsTerminal(symbols[i])) { + ++num_nonterminals; + symbols[i] = vocabulary->GetNonterminalIndex(num_nonterminals); + } + } + + if (end_x) { + ++num_nonterminals; + symbols.push_back(vocabulary->GetNonterminalIndex(num_nonterminals)); + } + + return Build(symbols); +} |