(chise_ft_get_value): Use escape forms for some special characters.

diff --git a/chise.c b/chise.c
index c24fd8e..3397991 100644 (file)
--- a/chise.c
+++ b/chise.c
@@ -170,49 +170,110 @@ int chise_ft_get_value (CHISE_Feature_Table *db,
   unsigned char key_buf[8];
 
   key_buf[0] = '?';
-  if (cid <= 0x7f)
+  if (cid == '\t')
+    {
+      key_buf[1] = '\\';
+      key_buf[2] = 't';
+      key_buf[3] = '\0';
+    }
+  else if (cid == '\n')
+    {
+      key_buf[1] = '\\';
+      key_buf[2] = 'n';
+      key_buf[3] = '\0';
+    }
+  else if (cid == '\r')
+    {
+      key_buf[1] = '\\';
+      key_buf[2] = 'r';
+      key_buf[3] = '\0';
+    }
+  else if (cid == 0x1C)
+    {
+      key_buf[1] = '\\';
+      key_buf[2] = '^';
+      key_buf[3] = '\\';
+      key_buf[4] = '\\';
+      key_buf[5] = '\0';
+    }
+  else if (cid <= 0x1F)
+    {
+      key_buf[1] = '\\';
+      key_buf[2] = '^';
+      key_buf[3] = '@' + cid;
+      key_buf[4] = '\0';
+    }
+  else if ( (cid == ' ') || (cid == '"') ||
+           (cid == '#') || (cid == '\'') ||
+           (cid == '(') || (cid == ')') ||
+           (cid == ',') || (cid == '.') ||
+           (cid == ';') || (cid == '?') ||
+           (cid == '[') || (cid == '\\') ||
+           (cid == ']') || (cid == '`') )
+    {
+      key_buf[1] = '\\';
+      key_buf[2] = cid;
+      key_buf[3] = '\0';
+    }
+  else if (cid <= 0x7E)
     {
       key_buf[1] = cid;
       key_buf[2] = '\0';
     }
-  else if (cid <= 0x7ff)
+  else if (cid == 0x7F)
+    {
+      key_buf[1] = '\\';
+      key_buf[2] = '^';
+      key_buf[3] = '?';
+      key_buf[4] = '\0';
+    }
+  else if (cid <= 0x9F)
+    {
+      key_buf[1] = '\\';
+      key_buf[2] = '^';
+      key_buf[3] = ((cid + '@') >> 6) | 0xC0;
+      key_buf[4] = ((cid + '@') & 0x3F) | 0x80;
+      key_buf[5] = '\0';
+
+    }
+  else if (cid <= 0x7FF)
     {
-      key_buf[1] = (cid >> 6) | 0xc0;
-      key_buf[2] = (cid & 0x3f) | 0x80;
+      key_buf[1] = (cid >> 6) | 0xC0;
+      key_buf[2] = (cid & 0x3F) | 0x80;
       key_buf[3] = '\0';
     }
-  else if (cid <= 0xffff)
+  else if (cid <= 0xFFFF)
     {
-      key_buf[1] = (cid >> 12) | 0xe0;
-      key_buf[2]= ((cid >>  6) & 0x3f) | 0x80;
-      key_buf[3]=  (cid        & 0x3f) | 0x80;
+      key_buf[1] = (cid >> 12) | 0xE0;
+      key_buf[2]= ((cid >>  6) & 0x3F) | 0x80;
+      key_buf[3]=  (cid        & 0x3F) | 0x80;
       key_buf[4] = '\0';
     }
-  else if (cid <= 0x1fffff)
+  else if (cid <= 0x1FFFFF)
     {
-      key_buf[1]=  (cid >> 18) | 0xf0;
-      key_buf[2]= ((cid >> 12) & 0x3f) | 0x80;
-      key_buf[3]= ((cid >>  6) & 0x3f) | 0x80;
-      key_buf[4]=  (cid        & 0x3f) | 0x80;
+      key_buf[1]=  (cid >> 18) | 0xF0;
+      key_buf[2]= ((cid >> 12) & 0x3F) | 0x80;
+      key_buf[3]= ((cid >>  6) & 0x3F) | 0x80;
+      key_buf[4]=  (cid        & 0x3F) | 0x80;
       key_buf[5] = '\0';
     }
-  else if (cid <= 0x3ffffff)
+  else if (cid <= 0x3FFFFFF)
     {
-      key_buf[1]=  (cid >> 24) | 0xf8;
-      key_buf[2]= ((cid >> 18) & 0x3f) | 0x80;
-      key_buf[3]= ((cid >> 12) & 0x3f) | 0x80;
-      key_buf[4]= ((cid >>  6) & 0x3f) | 0x80;
-      key_buf[5]=  (cid        & 0x3f) | 0x80;
+      key_buf[1]=  (cid >> 24) | 0xF8;
+      key_buf[2]= ((cid >> 18) & 0x3F) | 0x80;
+      key_buf[3]= ((cid >> 12) & 0x3F) | 0x80;
+      key_buf[4]= ((cid >>  6) & 0x3F) | 0x80;
+      key_buf[5]=  (cid        & 0x3F) | 0x80;
       key_buf[6] = '\0';
     }
   else
     {
-      key_buf[1]=  (cid >> 30) | 0xfc;
-      key_buf[2]= ((cid >> 24) & 0x3f) | 0x80;
-      key_buf[3]= ((cid >> 18) & 0x3f) | 0x80;
-      key_buf[4]= ((cid >> 12) & 0x3f) | 0x80;
-      key_buf[5]= ((cid >>  6) & 0x3f) | 0x80;
-      key_buf[6]=  (cid        & 0x3f) | 0x80;
+      key_buf[1]=  (cid >> 30) | 0xFC;
+      key_buf[2]= ((cid >> 24) & 0x3F) | 0x80;
+      key_buf[3]= ((cid >> 18) & 0x3F) | 0x80;
+      key_buf[4]= ((cid >> 12) & 0x3F) | 0x80;
+      key_buf[5]= ((cid >>  6) & 0x3F) | 0x80;
+      key_buf[6]=  (cid        & 0x3F) | 0x80;
       key_buf[7] = '\0';
     }
   return