From 74b9f7cb88cd21cfac3c2e50c8efb802485df0c5 Mon Sep 17 00:00:00 2001
From: Determinant <ted.sybil@gmail.com>
Date: Fri, 29 May 2015 23:06:58 +0800
Subject: init

---
 tnet_io/KaldiLib/Tokenizer.cc | 53 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 53 insertions(+)
 create mode 100644 tnet_io/KaldiLib/Tokenizer.cc

(limited to 'tnet_io/KaldiLib/Tokenizer.cc')

diff --git a/tnet_io/KaldiLib/Tokenizer.cc b/tnet_io/KaldiLib/Tokenizer.cc
new file mode 100644
index 0000000..0c49050
--- /dev/null
+++ b/tnet_io/KaldiLib/Tokenizer.cc
@@ -0,0 +1,53 @@
+#include "Tokenizer.h"
+#include "string.h"
+
+namespace TNet
+{
+  //****************************************************************************
+  //****************************************************************************
+  void
+  Tokenizer::
+  AddString(const char* pString)
+  {
+    // copy into string struct, which is more convenient
+    std::string       aux_string(pString);
+    std::string       aux_record;
+    std::string::size_type    cur_pos = 0;
+    std::string::size_type    old_pos = 0;
+    std::string::size_type    search_start = 0;
+
+    // make sure we have enough space
+    aux_record.reserve(aux_string.length());
+
+    // find all of separators and make a list of tokens
+    while(old_pos < std::string::npos) {
+      // find the next separator
+      cur_pos = aux_string.find_first_of(mSeparator, search_start);
+
+      // if backslash is in front of separator, ignore this separator
+      if (cur_pos != 0 && cur_pos != std::string::npos && 
+          pString[cur_pos - 1] == '\\') {
+        search_start = cur_pos + 1;
+        continue;
+      }
+
+      // we don't want to have empty records
+      if (!(cur_pos == old_pos && mSkipEmpty)) {
+        // extract token
+        aux_record.insert(0, pString+old_pos, cur_pos==std::string::npos ? strlen(pString+old_pos) : cur_pos - old_pos);
+        // insert to list
+        this->push_back(aux_record);
+
+        // we don't need the contents of the token
+        aux_record.erase();
+      }
+
+      // update old position so that it points behind the separator
+      old_pos = cur_pos < std::string::npos ? cur_pos + 1 : cur_pos;
+      search_start = old_pos;
+    }
+  }
+
+
+} // namespace TNet
+
-- 
cgit v1.2.3-70-g09d2