diff options
| author | Paul Baltescu <pauldb89@gmail.com> | 2013-05-23 20:42:26 +0100 |
|---|---|---|
| committer | Paul Baltescu <pauldb89@gmail.com> | 2013-05-23 20:42:26 +0100 |
| commit | 4ab38fecf0f49c5132a5d5ea748d03e0acfa36c2 (patch) | |
| tree | 8e250d99f4a381eae865a15b6eb43b0f77b68479 /decoder/scfg_translator.cc | |
| parent | 35d2c095ba6a912272f2fcc8322ece4213ada82b (diff) | |
| parent | 1d96c9c9c521edfca1d10e4d8c5064f79fda3ec5 (diff) | |
Merge branch 'master' of github.com:pauldb89/cdec
Diffstat (limited to 'decoder/scfg_translator.cc')
| -rw-r--r-- | decoder/scfg_translator.cc | 7 |
1 files changed, 6 insertions, 1 deletions
diff --git a/decoder/scfg_translator.cc b/decoder/scfg_translator.cc index 3b43b586..6f0b003b 100644 --- a/decoder/scfg_translator.cc +++ b/decoder/scfg_translator.cc @@ -12,6 +12,7 @@ #include "grammar.h" #include "bottom_up_parser.h" #include "sentence_metadata.h" +#include "stringlib.h" #include "tdict.h" #include "viterbi.h" #include "verbose.h" @@ -68,7 +69,11 @@ PassThroughGrammar::PassThroughGrammar(const Lattice& input, const string& cat, const int j = alts[k].dist2next + i; const string& src = TD::Convert(alts[k].label); if (ss.count(alts[k].label) == 0) { - TRulePtr pt(new TRule("[" + cat + "] ||| " + src + " ||| " + src + " ||| PassThrough=1")); + int length = static_cast<int>(log(UTF8StringLen(src)) / log(1.6)) + 1; + if (length > 6) length = 6; + string len_feat = "PassThrough_0=1"; + len_feat[12] += length; + TRulePtr pt(new TRule("[" + cat + "] ||| " + src + " ||| " + src + " ||| PassThrough=1 " + len_feat)); pt->a_.push_back(AlignmentPoint(0,0)); AddRule(pt); RefineRule(pt, ctf_level); |