From 80686d4e567bae579ea39e009826a2de92cd4ace Mon Sep 17 00:00:00 2001
From: redpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Wed, 11 Aug 2010 02:37:10 +0000
Subject: major refactor, break bad circular deps

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@509 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 utils/tdict.h | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
 create mode 100644 utils/tdict.h

(limited to 'utils/tdict.h')

diff --git a/utils/tdict.h b/utils/tdict.h
new file mode 100644
index 00000000..a7b3ee1c
--- /dev/null
+++ b/utils/tdict.h
@@ -0,0 +1,50 @@
+#ifndef _TDICT_H_
+#define _TDICT_H_
+
+#include <string>
+#include <vector>
+#include "wordid.h"
+#include <assert.h>
+
+class Vocab;
+
+struct TD {
+  /* // disabled for now
+  static const int reserved_begin=10; // allow room for SRI special tokens e.g. unk ss se pause.  tokens until this get "<FILLERi>"
+  static const int n_reserved=10; // 0...n_reserved-1 get token '<RESERVEDi>'
+  static inline WordID reserved(int i) {
+    assert(i>=0 && i<n_reserved);
+    return (WordID)(reserved_begin+i);
+  }
+  static inline WordID begin() {
+    return reserved(n_reserved);
+  }
+  */
+  static const WordID max_wordid=0x7fffffff;
+  static const WordID none=(WordID)-1; // Vocab_None
+  static char const* const ss_str;  //="<s>";
+  static char const* const se_str;  //="</s>";
+  static char const* const unk_str; //="<unk>";
+  static WordID ss,se,unk; // x=Convert(x_str)
+  static WordID end(); // next id to be assigned; [begin,end) give the non-reserved tokens seen so far
+  static Vocab dict_;
+  static void ConvertSentence(std::string const& sent, std::vector<WordID>* ids);
+  static void GetWordIDs(const std::vector<std::string>& strings, std::vector<WordID>* ids);
+  static std::string GetString(const std::vector<WordID>& str);
+  static std::string GetString(WordID const* i,WordID const* e);
+  static int AppendString(const WordID& w, int pos, int bufsize, char* buffer);
+  static unsigned int NumWords();
+  static WordID Convert(const std::string& s);
+  static WordID Convert(char const* s);
+  static const char* Convert(WordID w);
+};
+
+struct ToTD {
+  typedef WordID result_type;
+  result_type operator()(std::string const& t) const {
+    return TD::Convert(t);
+  }
+};
+
+
+#endif
-- 
cgit v1.2.3