HACK: If guessing English input, try Romaji input as well

[jmdict-cli.git] / kana2romaji.cpp

diff --git a/kana2romaji.cpp b/kana2romaji.cpp
index d0da9f396d8693efa0237abfcab3bd3d66f42c43..8508f764c1794ad46e27605236bac4f83427635d 100644 (file)
--- a/kana2romaji.cpp
+++ b/kana2romaji.cpp
@@ -218,7 +218,7 @@ void initRomaji() {
     romaji["プ"] = "pu";
     romaji["ペ"] = "pe";
     romaji["ポ"] = "po";
-    romaji["ã\83¼"] = "";
+    romaji["ã\83»"] = " ";
     
     // -- double width letters ------
     romaji["Ａ"] = "A";
@@ -292,9 +292,7 @@ void initRomaji() {
     romaji["＄"] = "$";
     romaji["％"] = "%";
     romaji["＆"] = "&";
-    romaji["＇"] = "'"; // TODO:
-    romaji["（"] = "(";
-    romaji["）"] = ")";
+    romaji["＇"] = "'";
     romaji["＊"] = "*";
     romaji["＋"] = "+";
     romaji["，"] = ",";
@@ -310,27 +308,31 @@ void initRomaji() {
     romaji["？"] = "?";
     romaji["＠"] = "@";
 
+    romaji["〔"] = "(";
+    romaji["〕"] = ")";
+    romaji["（"] = "(";
+    romaji["）"] = ")";
     romaji["［"] = "[";
-    romaji["＼"] = "\\";
     romaji["］"] = "]";
+    romaji["【"] = "[";
+    romaji["】"] = "]";
+    romaji["｛"] = "{";
+    romaji["｝"] = "}";
+    romaji["＼"] = "\\";
     romaji["＾"] = "^";
     romaji["＿"] = "_";
     romaji["｀"] = "`";
-
-    romaji["｛"] = "{";
     romaji["｜"] = "|";
-    romaji["｝"] = "}";
     romaji["～"] = "~";
-
-
-    // don't know where those belong to
+    romaji["ー"] = "";
+    romaji["。"] = ".";
+    romaji["、"] = ",";
     romaji["〜"] = "~";
-    romaji["、"] = ","; // TODO:
     romaji["−"] = "-";
-    
-    romaji["　"] = " ";
     romaji["―"] = "-";
-    romaji["・"] = "-"; // FIXME
+    romaji["　"] = " ";
+
+    romaji["ー"] = "\3";
 }
 
 void remove_quote_1(
@@ -359,7 +361,7 @@ void kana2romaji(const string& kana, string& rom) {
         if (trans == romaji.end()) {
             rom += ch;
             if (ch.size() > 1)
-                cout << "Don't know how to translate '" << ch << "' in '" << kana << "' to romaji." << endl;
+                cout << "Don't know how to translate '" << ch << "' in '" << kana << "' to romaji.\n";
         }
         else
             rom += trans->second;
@@ -372,13 +374,16 @@ void kana2romaji(const string& kana, string& rom) {
                string const pred = rom.substr(pos - 3, 3);
                if(pred == "chi" ||
                   pred == "shi" ||
-                  pred == "dzi"
+                  pred == "dzi" ||
+                  pred == "tsu" ||
+                  pred == "shi"
                ) {
                  remove_quote_1(pos, rom);
                  pos -= 2;
+                 continue;
                }
             }
-            else if (pos > 1) {
+            if (pos > 1) {
              
                string const pred = rom.substr(pos - 2, 2);
                if(pred == "ki" ||
@@ -403,6 +408,7 @@ void kana2romaji(const string& kana, string& rom) {
                       ? 3
                       : 2);
                   pos -= 2;
+                 continue;
                }
                else if(
                   pred == "fu" ||
@@ -412,22 +418,93 @@ void kana2romaji(const string& kana, string& rom) {
                {
                   remove_quote_1(pos, rom);
                   pos -= 2;
+                 continue;
+               }
+               else if(
+                 pred == "su" ||
+                 pred == "zu" ||
+                 pred == "te" ||
+                 pred == "de" ||
+                 pred == "ku" ||
+                 pred == "gu" ||
+                 pred == "mu"
+               )
+               {
+                 rom[pos - 1] = 'w'; 
+                 rom.erase(pos, 1);
+                 --pos;
+                 continue;
+               }
+               else if(
+                 pred == "to" ||
+                 pred == "do"
+               )
+               {
+                 rom[pos - 1] = 'h';
+                 rom.erase(pos, 1);
+                 --pos;
+                 continue;
                }
             }
+
+            if (pos > 0) {
+              char const pred = rom[pos - 1];
+
+              switch(pred)
+              {
+              case 'a':
+              case 'i':
+              case 'u':
+              case 'e':
+              case 'o':
+                rom.erase(pos, 1);
+                --pos;
+                break;
+              default:
+                cout << "Encountered a special character in " << kana << " but don't know what to do with it.\n";
+              }
+            }
             else
             {
-              rom.erase(pos);
+              rom.erase(pos, 1);
               --pos;
             }
         }
         else if (rom[pos] == '\2')
         {
-          if(pos + 1 < rom.size())
+         // two tsu may follow each other, so just remove them
+          if(pos + 1 < rom.size() && rom[pos + 1] != '\2')
             rom[pos] = rom[pos + 1];
           else
           {
-            rom.erase(pos);
+            rom.erase(pos, 1);
             --pos;
           }
         }
+        else if (rom[pos] == '\3')
+        {
+          if(pos == 0)
+          {
+            if(rom.size() == 1)
+              rom = "chouon";
+            else
+            {
+              cout << "ー is the first letter of " << kana << ". Don't know how to translate this.\n";
+              rom.erase(pos, 1);
+              --pos;
+            }
+          }
+          else
+            rom[pos] = rom[pos-1];
+        }
+
+  for (string::size_type pos = 0; pos < rom.size(); ++pos)
+    switch(rom[pos])
+    {
+    case '\1':
+    case '\2':
+    case '\3':
+      cout << "Failed to translate " << kana << '\n';
+      return;
+    }
 }