add text blank preprocess, test=asr (#3025)

runtime/engine/common/utils/blank_process.cc

@@ -0,0 +1,26 @@
+#include "utils/blank_process.h"
+
+namespace ppspeech {
+
+std::string BlankProcess(const std::string& str) {
+    std::string out = "";
+    int p = 0;
+    int end = str.size();
+    int q = -1;  //  last char of the output string
+    while (p != end) {
+        while (p != end && str[p] == ' ') {
+            p += 1;
+        }
+        if (p == end) 
+            return out;
+        if (q != -1 && isalpha(str[p]) && isalpha(str[q]) && str[p-1] == ' ')
+            // add a space when the last and current chars are in English and there have space(s) between them
+            out += ' ';
+        out += str[p];
+        q = p;
+        p += 1;
+    }
+    return out;
+}
+
+}  // namespace ppspeech

runtime/engine/common/utils/blank_process.h

@@ -0,0 +1,9 @@
+#include <string>
+#include <vector>
+#include <cctype>
+
+namespace ppspeech {
+
+std::string BlankProcess(const std::string& str);
+
+}  // namespace ppspeech

runtime/engine/common/utils/blank_process_test.cc

@@ -0,0 +1,26 @@
+#include "utils/blank_process.h"
+
+#include <gtest/gtest.h>
+#include <gmock/gmock.h>
+
+TEST(BlankProcess, BlankProcessTest) {
+    std::string test_str = "我 今天 去 了 超市 花了 120 元。";
+    std::string out_str = ppspeech::BlankProcess(test_str);
+    int ret = out_str.compare("我今天去了超市花了120元。");
+    EXPECT_EQ(ret, 0);
+
+    test_str = "how are you today";
+    out_str = ppspeech::BlankProcess(test_str);
+    ret = out_str.compare("how are you today");
+    EXPECT_EQ(ret, 0);
+
+    test_str = "我 的 paper 在 哪里？";
+    out_str = ppspeech::BlankProcess(test_str);
+    ret = out_str.compare("我的paper在哪里？");
+    EXPECT_EQ(ret, 0);
+
+    test_str = "我 今天     去 了 超市     花了   120 元。";
+    out_str = ppspeech::BlankProcess(test_str);
+    ret = out_str.compare("我今天去了超市花了120元。");
+    EXPECT_EQ(ret, 0);
+}