chris's crappy lda

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@6 ec762483-ff6d-05da-a07a-a48fb63a330f

1 files changed, 49 insertions, 0 deletions

diff --git a/gi/clda/src/tdict.h b/gi/clda/src/tdict.h
new file mode 100644
index 00000000..97f145a1
--- /dev/null
+++ b/gi/clda/src/tdict.h
@@ -0,0 +1,49 @@
+#ifndef _TDICT_H_
+#define _TDICT_H_
+
+#include <string>
+#include <vector>
+#include "wordid.h"
+#include "dict.h"
+
+class Vocab;
+
+struct TD {
+
+  static Dict dict_;
+  static std::string empty;
+  static std::string space;
+
+  static std::string GetString(const std::vector<WordID>& str) {
+    std::string res;
+    for (std::vector<WordID>::const_iterator i = str.begin(); i != str.end(); ++i)
+      res += (i == str.begin() ? empty : space) + TD::Convert(*i);
+    return res;
+  }
+
+  static void ConvertSentence(const std::string& sent, std::vector<WordID>* ids) {
+    std::string s = sent;
+    int last = 0;
+    ids->clear();
+    for (int i=0; i < s.size(); ++i)
+      if (s[i] == 32 || s[i] == '\t') {
+        s[i]=0;
+        if (last != i) {
+          ids->push_back(Convert(&s[last]));
+        }
+        last = i + 1;
+      }
+    if (last != s.size())
+      ids->push_back(Convert(&s[last]));
+  }
+
+  static WordID Convert(const std::string& s) {
+    return dict_.Convert(s);
+  }
+
+  static const std::string& Convert(const WordID& w) {
+    return dict_.Convert(w);
+  }
+};
+
+#endif