Removed comment.

[jmdict-cli.git] / kana2romaji.cpp

diff --git a/kana2romaji.cpp b/kana2romaji.cpp
index b02562f1ebf70b34bd277fcd03b137e90f88a44e..17c9aea99ba06c3d68ef3167b57356968fe5a75e 100644 (file)
--- a/kana2romaji.cpp
+++ b/kana2romaji.cpp
@@ -27,7 +27,7 @@ using namespace std;
 
 namespace  {
 void utfchar(const string& from, string::size_type pos, string& to) {
-    string::value_type first = from[pos];
+    unsigned first = from[pos];
     if ((first & 0x80) == 0)
         to = from[pos];
     else {
@@ -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 +308,46 @@ 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["　"] = " ";
+}
+
+void remove_quote_1(
+  string::size_type const pos,
+  string &rom)
+{
+  // if we encounter something like
+  // "ki" + '\1' 
+  // remove the previous character of \1 and the \1
+  // if \1 is followed by an 'y' remove that also
+  rom.erase(
+    pos - 1,
+    (pos + 1 < rom.size()
+     && rom[pos + 1] == 'y'
+    )
+      ? 3
+      : 2);
 }
 
 void kana2romaji(const string& kana, string& rom) {
@@ -350,7 +367,6 @@ void kana2romaji(const string& kana, string& rom) {
     }
     for (string::size_type pos = 0; pos < rom.size(); ++pos)
         if (rom[pos] == '\1') {
-            string::size_type count = 1;
 
             if (pos > 2) {
                string const pred = rom.substr(pos - 3, 3);
@@ -358,8 +374,8 @@ void kana2romaji(const string& kana, string& rom) {
                   pred == "shi" ||
                   pred == "dzi"
                ) {
-                 count = (pos + 1 < rom.size() && rom[pos + 1] == 'y') ? 3 : 2;
-                 rom.erase(pos - 1, count);
+                 remove_quote_1(pos, rom);
+                 pos -= 2;
                  continue;
                }
             }
@@ -372,26 +388,47 @@ void kana2romaji(const string& kana, string& rom) {
                   pred == "ri" ||
                   pred == "gi" ||
                   pred == "ji" ||
+                  pred == "hi" ||
                   pred == "bi" ||
                   pred == "pi"
                 )
                {
-                  count = (pos + 1 < rom.size() && rom[pos + 1] == 'y' && pred[0] == 'j') ? 3 : 2;
-                  rom.erase(pos - 1, count);
-                  continue;
+                 // shorten "ji\1y" to "j"
+                 // otherwise remove "\1" and the preceding character
+                 // but not the y
+                 rom.erase(
+                   pos - 1,
+                   (pos + 1 < rom.size()
+                    && rom[pos + 1] == 'y'
+                    && pred[0] == 'j')
+                      ? 3
+                      : 2);
+                  pos -= 2;
+               }
+               else if(
+                  pred == "fu" ||
+                  pred == "de" ||
+                  pred == "te" ||
+                  pred == "vu")
+               {
+                  remove_quote_1(pos, rom);
+                  pos -= 2;
                }
-                
-                if(pred == "fu" ||
-                   pred == "de" ||
-                   pred == "te" ||
-                   pred == "vu")
-                {
-                  count = (pos + 1 < rom.size() && rom[pos + 1] == 'y') ? 3 : 2;
-                  rom.erase(pos - 1, count);
-                  continue;
-                }
+            }
+            else
+            {
+              rom.erase(pos);
+              --pos;
             }
         }
-        else if (rom[pos] == '\2' && pos + 1 < rom.size())
+        else if (rom[pos] == '\2')
+        {
+          if(pos + 1 < rom.size())
             rom[pos] = rom[pos + 1];
+          else
+          {
+            rom.erase(pos);
+            --pos;
+          }
+        }
 }