iconv/loop.c

   1 /* Conversion loop frame work.
   2    Copyright (C) 1998-2003, 2005, 2008, 2011 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4    Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998.
   5
   6    The GNU C Library is free software; you can redistribute it and/or
   7    modify it under the terms of the GNU Lesser General Public
   8    License as published by the Free Software Foundation; either
   9    version 2.1 of the License, or (at your option) any later version.
  10
  11    The GNU C Library is distributed in the hope that it will be useful,
  12    but WITHOUT ANY WARRANTY; without even the implied warranty of
  13    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14    Lesser General Public License for more details.
  15
  16    You should have received a copy of the GNU Lesser General Public
  17    License along with the GNU C Library; if not, write to the Free
  18    Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
  19    02111-1307 USA.  */
  20
  21 /* This file provides a frame for the reader loop in all conversion modules.
  22    The actual code must (of course) be provided in the actual module source
  23    code but certain actions can be written down generically, with some
  24    customization options which are these:
  25
  26      MIN_NEEDED_INPUT   minimal number of input bytes needed for the next
  27                         conversion.
  28      MIN_NEEDED_OUTPUT  minimal number of bytes produced by the next round
  29                         of conversion.
  30
  31      MAX_NEEDED_INPUT   you guess it, this is the maximal number of input
  32                         bytes needed.  It defaults to MIN_NEEDED_INPUT
  33      MAX_NEEDED_OUTPUT  likewise for output bytes.
  34
  35      LOOPFCT            name of the function created.  If not specified
  36                         the name is `loop' but this prevents the use
  37                         of multiple functions in the same file.
  38
  39      BODY               this is supposed to expand to the body of the loop.
  40                         The user must provide this.
  41
  42      EXTRA_LOOP_DECLS   extra arguments passed from converion loop call.
  43
  44      INIT_PARAMS        code to define and initialize variables from params.
  45      UPDATE_PARAMS      code to store result in params.
  46
  47      ONEBYTE_BODY       body of the specialized conversion function for a
  48                         single byte from the current character set to INTERNAL.
  49 */
  50
  51 #include <assert.h>
  52 #include <endian.h>
  53 #include <gconv.h>
  54 #include <stdint.h>
  55 #include <string.h>
  56 #include <wchar.h>
  57 #include <sys/param.h>          /* For MIN.  */
  58 #define __need_size_t
  59 #include <stddef.h>
  60
  61
  62 /* We have to provide support for machines which are not able to handled
  63    unaligned memory accesses.  Some of the character encodings have
  64    representations with a fixed width of 2 or 4 bytes.  But if we cannot
  65    access unaligned memory we still have to read byte-wise.  */
  66 #undef FCTNAME2
  67 #if defined _STRING_ARCH_unaligned || !defined DEFINE_UNALIGNED
  68 /* We can handle unaligned memory access.  */
  69 # define get16(addr) *((__const uint16_t *) (addr))
  70 # define get32(addr) *((__const uint32_t *) (addr))
  71
  72 /* We need no special support for writing values either.  */
  73 # define put16(addr, val) *((uint16_t *) (addr)) = (val)
  74 # define put32(addr, val) *((uint32_t *) (addr)) = (val)
  75
  76 # define FCTNAME2(name) name
  77 #else
  78 /* Distinguish between big endian and little endian.  */
  79 # if __BYTE_ORDER == __LITTLE_ENDIAN
  80 #  define get16(addr) \
  81      (((__const unsigned char *) (addr))[1] << 8                              \
  82       | ((__const unsigned char *) (addr))[0])
  83 #  define get32(addr) \
  84      (((((__const unsigned char *) (addr))[3] << 8                            \
  85         | ((__const unsigned char *) (addr))[2]) << 8                         \
  86        | ((__const unsigned char *) (addr))[1]) << 8                          \
  87       | ((__const unsigned char *) (addr))[0])
  88
  89 #  define put16(addr, val) \
  90      ({ uint16_t __val = (val);                                               \
  91         ((unsigned char *) (addr))[0] = __val;                                \
  92         ((unsigned char *) (addr))[1] = __val >> 8;                           \
  93         (void) 0; })
  94 #  define put32(addr, val) \
  95      ({ uint32_t __val = (val);                                               \
  96         ((unsigned char *) (addr))[0] = __val;                                \
  97         __val >>= 8;                                                          \
  98         ((unsigned char *) (addr))[1] = __val;                                \
  99         __val >>= 8;                                                          \
 100         ((unsigned char *) (addr))[2] = __val;                                \
 101         __val >>= 8;                                                          \
 102         ((unsigned char *) (addr))[3] = __val;                                \
 103         (void) 0; })
 104 # else
 105 #  define get16(addr) \
 106      (((__const unsigned char *) (addr))[0] << 8                              \
 107       | ((__const unsigned char *) (addr))[1])
 108 #  define get32(addr) \
 109      (((((__const unsigned char *) (addr))[0] << 8                            \
 110         | ((__const unsigned char *) (addr))[1]) << 8                         \
 111        | ((__const unsigned char *) (addr))[2]) << 8                          \
 112       | ((__const unsigned char *) (addr))[3])
 113
 114 #  define put16(addr, val) \
 115      ({ uint16_t __val = (val);                                               \
 116         ((unsigned char *) (addr))[1] = __val;                                \
 117         ((unsigned char *) (addr))[0] = __val >> 8;                           \
 118         (void) 0; })
 119 #  define put32(addr, val) \
 120      ({ uint32_t __val = (val);                                               \
 121         ((unsigned char *) (addr))[3] = __val;                                \
 122         __val >>= 8;                                                          \
 123         ((unsigned char *) (addr))[2] = __val;                                \
 124         __val >>= 8;                                                          \
 125         ((unsigned char *) (addr))[1] = __val;                                \
 126         __val >>= 8;                                                          \
 127         ((unsigned char *) (addr))[0] = __val;                                \
 128         (void) 0; })
 129 # endif
 130
 131 # define FCTNAME2(name) name##_unaligned
 132 #endif
 133 #define FCTNAME(name) FCTNAME2(name)
 134
 135
 136 /* We need at least one byte for the next round.  */
 137 #ifndef MIN_NEEDED_INPUT
 138 # error "MIN_NEEDED_INPUT definition missing"
 139 #elif MIN_NEEDED_INPUT < 1
 140 # error "MIN_NEEDED_INPUT must be >= 1"
 141 #endif
 142
 143 /* Let's see how many bytes we produce.  */
 144 #ifndef MAX_NEEDED_INPUT
 145 # define MAX_NEEDED_INPUT       MIN_NEEDED_INPUT
 146 #endif
 147
 148 /* We produce at least one byte in the next round.  */
 149 #ifndef MIN_NEEDED_OUTPUT
 150 # error "MIN_NEEDED_OUTPUT definition missing"
 151 #elif MIN_NEEDED_OUTPUT < 1
 152 # error "MIN_NEEDED_OUTPUT must be >= 1"
 153 #endif
 154
 155 /* Let's see how many bytes we produce.  */
 156 #ifndef MAX_NEEDED_OUTPUT
 157 # define MAX_NEEDED_OUTPUT      MIN_NEEDED_OUTPUT
 158 #endif
 159
 160 /* Default name for the function.  */
 161 #ifndef LOOPFCT
 162 # define LOOPFCT                loop
 163 #endif
 164
 165 /* Make sure we have a loop body.  */
 166 #ifndef BODY
 167 # error "Definition of BODY missing for function" LOOPFCT
 168 #endif
 169
 170
 171 /* If no arguments have to passed to the loop function define the macro
 172    as empty.  */
 173 #ifndef EXTRA_LOOP_DECLS
 174 # define EXTRA_LOOP_DECLS
 175 #endif
 176
 177 /* Allow using UPDATE_PARAMS in macros where #ifdef UPDATE_PARAMS test
 178    isn't possible.  */
 179 #ifndef UPDATE_PARAMS
 180 # define UPDATE_PARAMS do { } while (0)
 181 #endif
 182 #ifndef REINIT_PARAMS
 183 # define REINIT_PARAMS do { } while (0)
 184 #endif
 185
 186
 187 /* To make it easier for the writers of the modules, we define a macro
 188    to test whether we have to ignore errors.  */
 189 #define ignore_errors_p() \
 190   (irreversible != NULL && (flags & __GCONV_IGNORE_ERRORS))
 191
 192
 193 /* Error handling for the FROM_LOOP direction, with ignoring of errors.
 194    Note that we cannot use the do while (0) trick since `break' and
 195    `continue' must reach certain points.  */
 196 #define STANDARD_FROM_LOOP_ERR_HANDLER(Incr) \
 197   {                                                                           \
 198     result = __GCONV_ILLEGAL_INPUT;                                           \
 199                                                                               \
 200     if (! ignore_errors_p ())                                                 \
 201       break;                                                                  \
 202                                                                               \
 203     /* We ignore the invalid input byte sequence.  */                         \
 204     inptr += (Incr);                                                          \
 205     ++*irreversible;                                                          \
 206     /* But we keep result == __GCONV_ILLEGAL_INPUT, because of the constraint \
 207        that "iconv -c" must give the same exitcode as "iconv".  */            \
 208     continue;                                                                 \
 209   }
 210
 211 /* Error handling for the TO_LOOP direction, with use of transliteration/
 212    transcription functions and ignoring of errors.  Note that we cannot use
 213    the do while (0) trick since `break' and `continue' must reach certain
 214    points.  */
 215 #define STANDARD_TO_LOOP_ERR_HANDLER(Incr) \
 216   {                                                                           \
 217     struct __gconv_trans_data *trans;                                         \
 218                                                                               \
 219     result = __GCONV_ILLEGAL_INPUT;                                           \
 220                                                                               \
 221     if (irreversible == NULL)                                                 \
 222       /* This means we are in call from __gconv_transliterate.  In this       \
 223          case we are not doing any error recovery outself.  */                \
 224       break;                                                                  \
 225                                                                               \
 226     /* If needed, flush any conversion state, so that __gconv_transliterate   \
 227        starts with current shift state.  */                                   \
 228     UPDATE_PARAMS;                                                            \
 229                                                                               \
 230     /* First try the transliteration methods.  */                             \
 231     for (trans = step_data->__trans; trans != NULL; trans = trans->__next)    \
 232       {                                                                       \
 233         result = DL_CALL_FCT (trans->__trans_fct,                             \
 234                               (step, step_data, trans->__data, *inptrp,       \
 235                                &inptr, inend, &outptr, irreversible));        \
 236         if (result != __GCONV_ILLEGAL_INPUT)                                  \
 237           break;                                                              \
 238       }                                                                       \
 239                                                                               \
 240     REINIT_PARAMS;                                                            \
 241                                                                               \
 242     /* If any of them recognized the input continue with the loop.  */        \
 243     if (result != __GCONV_ILLEGAL_INPUT)                                      \
 244       {                                                                       \
 245         if (__builtin_expect (result == __GCONV_FULL_OUTPUT, 0))              \
 246           break;                                                              \
 247                                                                               \
 248         continue;                                                             \
 249       }                                                                       \
 250                                                                               \
 251     /* Next see whether we have to ignore the error.  If not, stop.  */       \
 252     if (! ignore_errors_p ())                                                 \
 253       break;                                                                  \
 254                                                                               \
 255     /* When we come here it means we ignore the character.  */                \
 256     ++*irreversible;                                                          \
 257     inptr += Incr;                                                            \
 258     /* But we keep result == __GCONV_ILLEGAL_INPUT, because of the constraint \
 259        that "iconv -c" must give the same exitcode as "iconv".  */            \
 260     continue;                                                                 \
 261   }
 262
 263
 264 /* Handling of Unicode 3.1 TAG characters.  Unicode recommends
 265    "If language codes are not relevant to the particular processing
 266     operation, then they should be ignored."  This macro is usually
 267    called right before  STANDARD_TO_LOOP_ERR_HANDLER (Incr).  */
 268 #define UNICODE_TAG_HANDLER(Character, Incr) \
 269   {                                                                           \
 270     /* TAG characters are those in the range U+E0000..U+E007F.  */            \
 271     if (((Character) >> 7) == (0xe0000 >> 7))                                 \
 272       {                                                                       \
 273         inptr += Incr;                                                        \
 274         continue;                                                             \
 275       }                                                                       \
 276   }
 277
 278
 279 /* The function returns the status, as defined in gconv.h.  */
 280 static inline int
 281 __attribute ((always_inline))
 282 FCTNAME (LOOPFCT) (struct __gconv_step *step,
 283                    struct __gconv_step_data *step_data,
 284                    const unsigned char **inptrp, const unsigned char *inend,
 285                    unsigned char **outptrp, const unsigned char *outend,
 286                    size_t *irreversible EXTRA_LOOP_DECLS)
 287 {
 288 #ifdef LOOP_NEED_STATE
 289   mbstate_t *state = step_data->__statep;
 290 #endif
 291 #ifdef LOOP_NEED_FLAGS
 292   int flags = step_data->__flags;
 293 #endif
 294 #ifdef LOOP_NEED_DATA
 295   void *data = step->__data;
 296 #endif
 297   int result = __GCONV_EMPTY_INPUT;
 298   const unsigned char *inptr = *inptrp;
 299   unsigned char *outptr = *outptrp;
 300
 301 #ifdef INIT_PARAMS
 302   INIT_PARAMS;
 303 #endif
 304
 305   while (inptr != inend)
 306     {
 307       /* `if' cases for MIN_NEEDED_OUTPUT ==/!= 1 is made to help the
 308          compiler generating better code.  They will be optimized away
 309          since MIN_NEEDED_OUTPUT is always a constant.  */
 310       if (MIN_NEEDED_INPUT > 1
 311           && __builtin_expect (inptr + MIN_NEEDED_INPUT > inend, 0))
 312         {
 313           /* We don't have enough input for another complete input
 314              character.  */
 315           result = __GCONV_INCOMPLETE_INPUT;
 316           break;
 317         }
 318       if ((MIN_NEEDED_OUTPUT != 1
 319            && __builtin_expect (outptr + MIN_NEEDED_OUTPUT > outend, 0))
 320           || (MIN_NEEDED_OUTPUT == 1
 321               && __builtin_expect (outptr >= outend, 0)))
 322         {
 323           /* Overflow in the output buffer.  */
 324           result = __GCONV_FULL_OUTPUT;
 325           break;
 326         }
 327
 328       /* Here comes the body the user provides.  It can stop with
 329          RESULT set to GCONV_INCOMPLETE_INPUT (if the size of the
 330          input characters vary in size), GCONV_ILLEGAL_INPUT, or
 331          GCONV_FULL_OUTPUT (if the output characters vary in size).  */
 332       BODY
 333     }
 334
 335   /* Update the pointers pointed to by the parameters.  */
 336   *inptrp = inptr;
 337   *outptrp = outptr;
 338   UPDATE_PARAMS;
 339
 340   return result;
 341 }
 342
 343
 344 /* Include the file a second time to define the function to handle
 345    unaligned access.  */
 346 #if !defined DEFINE_UNALIGNED && !defined _STRING_ARCH_unaligned \
 347     && MIN_NEEDED_INPUT != 1 && MAX_NEEDED_INPUT % MIN_NEEDED_INPUT == 0 \
 348     && MIN_NEEDED_OUTPUT != 1 && MAX_NEEDED_OUTPUT % MIN_NEEDED_OUTPUT == 0
 349 # undef get16
 350 # undef get32
 351 # undef put16
 352 # undef put32
 353 # undef unaligned
 354
 355 # define DEFINE_UNALIGNED
 356 # include "loop.c"
 357 # undef DEFINE_UNALIGNED
 358 #endif
 359
 360
 361 #if MAX_NEEDED_INPUT > 1
 362 # define SINGLE(fct) SINGLE2 (fct)
 363 # define SINGLE2(fct) fct##_single
 364 static inline int
 365 __attribute ((always_inline))
 366 SINGLE(LOOPFCT) (struct __gconv_step *step,
 367                  struct __gconv_step_data *step_data,
 368                  const unsigned char **inptrp, const unsigned char *inend,
 369                  unsigned char **outptrp, unsigned char *outend,
 370                  size_t *irreversible EXTRA_LOOP_DECLS)
 371 {
 372   mbstate_t *state = step_data->__statep;
 373 #ifdef LOOP_NEED_FLAGS
 374   int flags = step_data->__flags;
 375 #endif
 376 #ifdef LOOP_NEED_DATA
 377   void *data = step->__data;
 378 #endif
 379   int result = __GCONV_OK;
 380   unsigned char bytebuf[MAX_NEEDED_INPUT];
 381   const unsigned char *inptr = *inptrp;
 382   unsigned char *outptr = *outptrp;
 383   size_t inlen;
 384
 385 #ifdef INIT_PARAMS
 386   INIT_PARAMS;
 387 #endif
 388
 389 #ifdef UNPACK_BYTES
 390   UNPACK_BYTES
 391 #else
 392   /* Add the bytes from the state to the input buffer.  */
 393   for (inlen = 0; inlen < (size_t) (state->__count & 7); ++inlen)
 394     bytebuf[inlen] = state->__value.__wchb[inlen];
 395 #endif
 396
 397   /* Are there enough bytes in the input buffer?  */
 398   if (__builtin_expect (inptr + (MIN_NEEDED_INPUT - inlen) > inend, 0))
 399     {
 400       *inptrp = inend;
 401 #ifdef STORE_REST
 402       while (inptr < inend)
 403         bytebuf[inlen++] = *inptr++;
 404
 405       inptr = bytebuf;
 406       inptrp = &inptr;
 407       inend = &bytebuf[inlen];
 408
 409       STORE_REST
 410 #else
 411       /* We don't have enough input for another complete input
 412          character.  */
 413       while (inptr < inend)
 414         state->__value.__wchb[inlen++] = *inptr++;
 415 #endif
 416
 417       return __GCONV_INCOMPLETE_INPUT;
 418     }
 419
 420   /* Enough space in output buffer.  */
 421   if ((MIN_NEEDED_OUTPUT != 1 && outptr + MIN_NEEDED_OUTPUT > outend)
 422       || (MIN_NEEDED_OUTPUT == 1 && outptr >= outend))
 423     /* Overflow in the output buffer.  */
 424     return __GCONV_FULL_OUTPUT;
 425
 426   /*  Now add characters from the normal input buffer.  */
 427   do
 428     bytebuf[inlen++] = *inptr++;
 429   while (inlen < MAX_NEEDED_INPUT && inptr < inend);
 430
 431   inptr = bytebuf;
 432   inend = &bytebuf[inlen];
 433
 434   do
 435     {
 436       BODY
 437     }
 438   while (0);
 439
 440   /* Now we either have produced an output character and consumed all the
 441      bytes from the state and at least one more, or the character is still
 442      incomplete, or we have some other error (like illegal input character,
 443      no space in output buffer).  */
 444   if (__builtin_expect (inptr != bytebuf, 1))
 445     {
 446       /* We found a new character.  */
 447       assert (inptr - bytebuf > (state->__count & 7));
 448
 449       *inptrp += inptr - bytebuf - (state->__count & 7);
 450       *outptrp = outptr;
 451
 452       result = __GCONV_OK;
 453
 454       /* Clear the state buffer.  */
 455 #ifdef CLEAR_STATE
 456       CLEAR_STATE;
 457 #else
 458       state->__count &= ~7;
 459 #endif
 460     }
 461   else if (result == __GCONV_INCOMPLETE_INPUT)
 462     {
 463       /* This can only happen if we have less than MAX_NEEDED_INPUT bytes
 464          available.  */
 465       assert (inend != &bytebuf[MAX_NEEDED_INPUT]);
 466
 467       *inptrp += inend - bytebuf - (state->__count & 7);
 468 #ifdef STORE_REST
 469       inptrp = &inptr;
 470
 471       STORE_REST
 472 #else
 473       /* We don't have enough input for another complete input
 474          character.  */
 475       assert (inend - inptr > (state->__count & ~7));
 476       assert (inend - inptr <= 7);
 477       state->__count = (state->__count & ~7) | (inend - inptr);
 478       inlen = 0;
 479       while (inptr < inend)
 480         state->__value.__wchb[inlen++] = *inptr++;
 481 #endif
 482     }
 483
 484   return result;
 485 }
 486 # undef SINGLE
 487 # undef SINGLE2
 488 #endif
 489
 490
 491 #ifdef ONEBYTE_BODY
 492 /* Define the shortcut function for btowc.  */
 493 static wint_t
 494 gconv_btowc (struct __gconv_step *step, unsigned char c)
 495   ONEBYTE_BODY
 496 # define FROM_ONEBYTE gconv_btowc
 497 #endif
 498
 499
 500 /* We remove the macro definitions so that we can include this file again
 501    for the definition of another function.  */
 502 #undef MIN_NEEDED_INPUT
 503 #undef MAX_NEEDED_INPUT
 504 #undef MIN_NEEDED_OUTPUT
 505 #undef MAX_NEEDED_OUTPUT
 506 #undef LOOPFCT
 507 #undef BODY
 508 #undef LOOPFCT
 509 #undef EXTRA_LOOP_DECLS
 510 #undef INIT_PARAMS
 511 #undef UPDATE_PARAMS
 512 #undef REINIT_PARAMS
 513 #undef ONEBYTE_BODY
 514 #undef UNPACK_BYTES
 515 #undef CLEAR_STATE
 516 #undef LOOP_NEED_STATE
 517 #undef LOOP_NEED_FLAGS
 518 #undef LOOP_NEED_DATA
 519 #undef get16
 520 #undef get32
 521 #undef put16
 522 #undef put32
 523 #undef unaligned