(BODY to IBM937): Correct several mistakes in buffer and table handling.
[kopensolaris-gnu/glibc.git] / iconvdata / ibm937.c
index 98e9683..4fee56a 100644 (file)
@@ -1,5 +1,5 @@
 /* Conversion to and from IBM937.
-   Copyright (C) 2000 Free Software Foundation, Inc.
+   Copyright (C) 2000, 2001 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
    Contributed by Masahide Washizawa <washi@yamato.ibm.co.jp>, 2000.
 
 #define CHARSET_NAME   "IBM937//"
 #define FROM_LOOP      from_ibm937
 #define TO_LOOP                to_ibm937
+#define MIN_NEEDED_FROM        1
+#define MAX_NEEDED_FROM        2
+#define MIN_NEEDED_TO  4
+#define MAX_NEEDED_TO  4
+#define PREPARE_LOOP \
+  int save_curcs;                                                            \
+  int *curcsp = &data->__statep->__count;
+#define EXTRA_LOOP_ARGS                , curcsp
 
 /* Definitions of initialization and destructor function.  */
 #define DEFINE_INIT    1
 #define DEFINE_FINI    1
 
-#define MIN_NEEDED_FROM        1
-#define MIN_NEEDED_TO  4
+
+/* Since this is a stateful encoding we have to provide code which resets
+   the output state to the initial state.  This has to be done during the
+   flushing.  */
+#define EMIT_SHIFT_TO_INIT \
+  if ((data->__statep->__count & ~7) != sb)                                  \
+    {                                                                        \
+      if (FROM_DIRECTION)                                                    \
+       data->__statep->__count &= 7;                                         \
+      else                                                                   \
+       {                                                                     \
+         unsigned char *outbuf = data->__outbuf;                             \
+                                                                             \
+         /* We are not in the initial state.  To switch back we have         \
+            to emit `SI'.  */                                                \
+         if (__builtin_expect (outbuf >= data->__outbufend, 0))              \
+           /* We don't have enough room in the output buffer.  */            \
+           status = __GCONV_FULL_OUTPUT;                                     \
+         else                                                                \
+           {                                                                 \
+             /* Write out the shift sequence.  */                            \
+             *outbuf++ = SI;                                                 \
+             data->__outbuf = outbuf;                                        \
+             data->__statep->__count &= 7;                                   \
+           }                                                                 \
+       }                                                                     \
+    }
+
+
+/* Since we might have to reset input pointer we must be able to save
+   and retore the state.  */
+#define SAVE_RESET_STATE(Save) \
+  if (Save)                                                                  \
+    save_curcs = *curcsp;                                                    \
+  else                                                                       \
+    *curcsp = save_curcs
+
 
 /* Current codeset type.  */
 enum
 {
-  init = 0,
-  sb,
-  db
+  sb = 0,
+  db = 64
 };
 
 /* First, define the conversion function from IBM-937 to UCS4.  */
 #define MIN_NEEDED_INPUT       MIN_NEEDED_FROM
+#define MAX_NEEDED_INPUT       MAX_NEEDED_FROM
 #define MIN_NEEDED_OUTPUT      MIN_NEEDED_TO
-#define INIT_PARAMS            int curcs = init;
 #define LOOPFCT                FROM_LOOP
 #define BODY \
   {                                                                          \
@@ -59,12 +101,6 @@ enum
                                                                              \
     if (__builtin_expect (ch, 0) == SO)                                              \
       {                                                                              \
-       if (__builtin_expect (inptr + 1 >= inend, 0))                         \
-         {                                                                   \
-           result = __GCONV_INCOMPLETE_INPUT;                                \
-           break;                                                            \
-         }                                                                   \
-                                                                             \
        /* Shift OUT, change to DBCS converter.  */                           \
        if (curcs == db)                                                      \
          {                                                                   \
@@ -73,16 +109,10 @@ enum
          }                                                                   \
        curcs = db;                                                           \
        ++inptr;                                                              \
-       ch = *inptr;                                                          \
+       continue;                                                             \
       }                                                                              \
     else if (__builtin_expect (ch, 0) == SI)                                 \
       {                                                                              \
-       if (__builtin_expect (inptr + 1 >= inend, 0))                         \
-         {                                                                   \
-           result = __GCONV_INCOMPLETE_INPUT;                                \
-           break;                                                            \
-         }                                                                   \
-                                                                             \
        /* Shift IN, change to SBCS converter.  */                            \
        if (curcs == sb)                                                      \
          {                                                                   \
@@ -91,10 +121,10 @@ enum
          }                                                                   \
        curcs = sb;                                                           \
        ++inptr;                                                              \
-       ch = *inptr;                                                          \
+       continue;                                                             \
       }                                                                              \
                                                                              \
-    if (curcs == sb || curcs == init)                                        \
+    if (curcs == sb)                                                         \
       {                                                                              \
        /* Use the UCS4 table for single byte.  */                            \
        ch = __ibm937sb_to_ucs4[ch];                                          \
@@ -107,19 +137,20 @@ enum
                break;                                                        \
              }                                                               \
            ++*irreversible;                                                  \
-           ++inptr;                                                          \
-           continue;                                                         \
          }                                                                   \
        else                                                                  \
          {                                                                   \
            put32 (outptr, ch);                                               \
            outptr += 4;                                                      \
-           inptr++;                                                          \
          }                                                                   \
+       ++inptr;                                                              \
       }                                                                              \
-    else if (curcs == db)                                                    \
+    else                                                                     \
       {                                                                              \
        /* Use the IBM937 table for double byte.  */                          \
+                                                                             \
+       assert (curcs == db);                                                 \
+                                                                             \
        ch = ibm937db_to_ucs4(inptr[0], inptr[1]);                            \
        if (__builtin_expect (ch, L'\1') == L'\0' && *inptr != '\0')          \
          {                                                                   \
@@ -130,24 +161,25 @@ enum
                break;                                                        \
              }                                                               \
            ++*irreversible;                                                  \
-           inptr += 2;                                                       \
-           continue;                                                         \
          }                                                                   \
        else                                                                  \
          {                                                                   \
            put32 (outptr, ch);                                               \
            outptr += 4;                                                      \
-           inptr += 2;                                                       \
          }                                                                   \
+       inptr += 2;                                                           \
       }                                                                              \
   }
 #define LOOP_NEED_FLAGS
+#define EXTRA_LOOP_DECLS       , int *curcsp
+#define INIT_PARAMS            int curcs = *curcsp & ~7
+#define UPDATE_PARAMS          *curcsp = curcs
 #include <iconv/loop.c>
 
 /* Next, define the other direction.  */
 #define MIN_NEEDED_INPUT       MIN_NEEDED_TO
 #define MIN_NEEDED_OUTPUT      MIN_NEEDED_FROM
-#define INIT_PARAMS            int curcs = init;
+#define MAX_NEEDED_OUTPUT      MAX_NEEDED_FROM
 #define LOOPFCT                        TO_LOOP
 #define BODY \
   {                                                                          \
@@ -156,13 +188,15 @@ enum
                                                                              \
     /* Use the UCS4 table for single byte.  */                               \
     cp = __ucs4_to_ibm937sb[ch];                                             \
-    if (__builtin_expect (ch >= sizeof (__ucs4_to_ibm937sb)                  \
-                         / sizeof (__ucs4_to_ibm937sb[0]), 0)                \
+    if (__builtin_expect (ch >= (sizeof (__ucs4_to_ibm937sb)                 \
+                                / sizeof (__ucs4_to_ibm937sb[0])), 0)        \
        || (__builtin_expect (cp[0], '\1') == '\0' && ch != 0))               \
       {                                                                              \
        /* Use the UCS4 table for double byte.  */                            \
        cp = __ucs4_to_ibm937db[ch];                                          \
-       if (__builtin_expect (cp[0], '\1') == '\0' && ch != 0)                \
+       if (__builtin_expect (ch >= (sizeof (__ucs4_to_ibm937db)              \
+                                    / sizeof (__ucs4_to_ibm937db[0])), 0)    \
+           || __builtin_expect (cp[0], '\1') == '\0')                        \
          {                                                                   \
            /* This is an illegal character.  */                              \
            if (! ignore_errors_p ())                                         \
@@ -174,16 +208,16 @@ enum
          }                                                                   \
        else                                                                  \
          {                                                                   \
-           if (curcs == init || curcs == sb)                                 \
+           if (curcs == sb)                                                  \
              {                                                               \
                *outptr++ = SO;                                               \
-               if (__builtin_expect (outptr == outend, 0))                   \
-                 {                                                           \
-                   result = __GCONV_FULL_OUTPUT;                             \
-                   break;                                                    \
-                 }                                                           \
                curcs = db;                                                   \
              }                                                               \
+           if (__builtin_expect (outptr + 1 >= outend, 0))                   \
+             {                                                               \
+               result = __GCONV_FULL_OUTPUT;                                 \
+               break;                                                        \
+             }                                                               \
            *outptr++ = cp[0];                                                \
            *outptr++ = cp[1];                                                \
          }                                                                   \
@@ -193,13 +227,13 @@ enum
        if (curcs == db)                                                      \
          {                                                                   \
            *outptr++ = SI;                                                   \
+           curcs = sb;                                                       \
            if (__builtin_expect (outptr == outend, 0))                       \
              {                                                               \
                result = __GCONV_FULL_OUTPUT;                                 \
                break;                                                        \
              }                                                               \
          }                                                                   \
-       curcs = sb;                                                           \
        *outptr++ = cp[0];                                                    \
       }                                                                              \
                                                                              \
@@ -207,6 +241,9 @@ enum
     inptr += 4;                                                                      \
   }
 #define LOOP_NEED_FLAGS
+#define EXTRA_LOOP_DECLS       , int *curcsp
+#define INIT_PARAMS            int curcs = *curcsp & ~7
+#define UPDATE_PARAMS          *curcsp = curcs
 #include <iconv/loop.c>
 
 /* Now define the toplevel functions.  */