| 1 | /* Character set conversion support for GDB. |
| 2 | |
| 3 | Copyright (C) 2001-2017 Free Software Foundation, Inc. |
| 4 | |
| 5 | This file is part of GDB. |
| 6 | |
| 7 | This program is free software; you can redistribute it and/or modify |
| 8 | it under the terms of the GNU General Public License as published by |
| 9 | the Free Software Foundation; either version 3 of the License, or |
| 10 | (at your option) any later version. |
| 11 | |
| 12 | This program is distributed in the hope that it will be useful, |
| 13 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 15 | GNU General Public License for more details. |
| 16 | |
| 17 | You should have received a copy of the GNU General Public License |
| 18 | along with this program. If not, see <http://www.gnu.org/licenses/>. */ |
| 19 | |
| 20 | #include "defs.h" |
| 21 | #include "charset.h" |
| 22 | #include "gdbcmd.h" |
| 23 | #include "gdb_obstack.h" |
| 24 | #include "gdb_wait.h" |
| 25 | #include "charset-list.h" |
| 26 | #include "vec.h" |
| 27 | #include "environ.h" |
| 28 | #include "arch-utils.h" |
| 29 | #include "gdb_vecs.h" |
| 30 | #include <ctype.h> |
| 31 | |
| 32 | #ifdef USE_WIN32API |
| 33 | #include <windows.h> |
| 34 | #endif |
| 35 | \f |
| 36 | /* How GDB's character set support works |
| 37 | |
| 38 | GDB has three global settings: |
| 39 | |
| 40 | - The `current host character set' is the character set GDB should |
| 41 | use in talking to the user, and which (hopefully) the user's |
| 42 | terminal knows how to display properly. Most users should not |
| 43 | change this. |
| 44 | |
| 45 | - The `current target character set' is the character set the |
| 46 | program being debugged uses. |
| 47 | |
| 48 | - The `current target wide character set' is the wide character set |
| 49 | the program being debugged uses, that is, the encoding used for |
| 50 | wchar_t. |
| 51 | |
| 52 | There are commands to set each of these, and mechanisms for |
| 53 | choosing reasonable default values. GDB has a global list of |
| 54 | character sets that it can use as its host or target character |
| 55 | sets. |
| 56 | |
| 57 | The header file `charset.h' declares various functions that |
| 58 | different pieces of GDB need to perform tasks like: |
| 59 | |
| 60 | - printing target strings and characters to the user's terminal |
| 61 | (mostly target->host conversions), |
| 62 | |
| 63 | - building target-appropriate representations of strings and |
| 64 | characters the user enters in expressions (mostly host->target |
| 65 | conversions), |
| 66 | |
| 67 | and so on. |
| 68 | |
| 69 | To avoid excessive code duplication and maintenance efforts, |
| 70 | GDB simply requires a capable iconv function. Users on platforms |
| 71 | without a suitable iconv can use the GNU iconv library. */ |
| 72 | |
| 73 | \f |
| 74 | #ifdef PHONY_ICONV |
| 75 | |
| 76 | /* Provide a phony iconv that does as little as possible. Also, |
| 77 | arrange for there to be a single available character set. */ |
| 78 | |
| 79 | #undef GDB_DEFAULT_HOST_CHARSET |
| 80 | #ifdef USE_WIN32API |
| 81 | # define GDB_DEFAULT_HOST_CHARSET "CP1252" |
| 82 | #else |
| 83 | # define GDB_DEFAULT_HOST_CHARSET "ISO-8859-1" |
| 84 | #endif |
| 85 | #define GDB_DEFAULT_TARGET_CHARSET GDB_DEFAULT_HOST_CHARSET |
| 86 | #define GDB_DEFAULT_TARGET_WIDE_CHARSET "UTF-32" |
| 87 | #undef DEFAULT_CHARSET_NAMES |
| 88 | #define DEFAULT_CHARSET_NAMES GDB_DEFAULT_HOST_CHARSET , |
| 89 | |
| 90 | #undef iconv_t |
| 91 | #define iconv_t int |
| 92 | #undef iconv_open |
| 93 | #define iconv_open phony_iconv_open |
| 94 | #undef iconv |
| 95 | #define iconv phony_iconv |
| 96 | #undef iconv_close |
| 97 | #define iconv_close phony_iconv_close |
| 98 | |
| 99 | #undef ICONV_CONST |
| 100 | #define ICONV_CONST const |
| 101 | |
| 102 | /* We allow conversions from UTF-32, wchar_t, and the host charset. |
| 103 | We allow conversions to wchar_t and the host charset. |
| 104 | Return 1 if we are converting from UTF-32BE, 2 if from UTF32-LE, |
| 105 | 0 otherwise. This is used as a flag in calls to iconv. */ |
| 106 | |
| 107 | static iconv_t |
| 108 | phony_iconv_open (const char *to, const char *from) |
| 109 | { |
| 110 | if (strcmp (to, "wchar_t") && strcmp (to, GDB_DEFAULT_HOST_CHARSET)) |
| 111 | return -1; |
| 112 | |
| 113 | if (!strcmp (from, "UTF-32BE") || !strcmp (from, "UTF-32")) |
| 114 | return 1; |
| 115 | |
| 116 | if (!strcmp (from, "UTF-32LE")) |
| 117 | return 2; |
| 118 | |
| 119 | if (strcmp (from, "wchar_t") && strcmp (from, GDB_DEFAULT_HOST_CHARSET)) |
| 120 | return -1; |
| 121 | |
| 122 | return 0; |
| 123 | } |
| 124 | |
| 125 | static int |
| 126 | phony_iconv_close (iconv_t arg) |
| 127 | { |
| 128 | return 0; |
| 129 | } |
| 130 | |
| 131 | static size_t |
| 132 | phony_iconv (iconv_t utf_flag, const char **inbuf, size_t *inbytesleft, |
| 133 | char **outbuf, size_t *outbytesleft) |
| 134 | { |
| 135 | if (utf_flag) |
| 136 | { |
| 137 | enum bfd_endian endian |
| 138 | = utf_flag == 1 ? BFD_ENDIAN_BIG : BFD_ENDIAN_LITTLE; |
| 139 | while (*inbytesleft >= 4) |
| 140 | { |
| 141 | unsigned long c |
| 142 | = extract_unsigned_integer ((const gdb_byte *)*inbuf, 4, endian); |
| 143 | |
| 144 | if (c >= 256) |
| 145 | { |
| 146 | errno = EILSEQ; |
| 147 | return -1; |
| 148 | } |
| 149 | if (*outbytesleft < 1) |
| 150 | { |
| 151 | errno = E2BIG; |
| 152 | return -1; |
| 153 | } |
| 154 | **outbuf = c & 0xff; |
| 155 | ++*outbuf; |
| 156 | --*outbytesleft; |
| 157 | |
| 158 | *inbuf += 4; |
| 159 | *inbytesleft -= 4; |
| 160 | } |
| 161 | if (*inbytesleft) |
| 162 | { |
| 163 | /* Partial sequence on input. */ |
| 164 | errno = EINVAL; |
| 165 | return -1; |
| 166 | } |
| 167 | } |
| 168 | else |
| 169 | { |
| 170 | /* In all other cases we simply copy input bytes to the |
| 171 | output. */ |
| 172 | size_t amt = *inbytesleft; |
| 173 | |
| 174 | if (amt > *outbytesleft) |
| 175 | amt = *outbytesleft; |
| 176 | memcpy (*outbuf, *inbuf, amt); |
| 177 | *inbuf += amt; |
| 178 | *outbuf += amt; |
| 179 | *inbytesleft -= amt; |
| 180 | *outbytesleft -= amt; |
| 181 | if (*inbytesleft) |
| 182 | { |
| 183 | errno = E2BIG; |
| 184 | return -1; |
| 185 | } |
| 186 | } |
| 187 | |
| 188 | /* The number of non-reversible conversions -- but they were all |
| 189 | reversible. */ |
| 190 | return 0; |
| 191 | } |
| 192 | |
| 193 | #else /* PHONY_ICONV */ |
| 194 | |
| 195 | /* On systems that don't have EILSEQ, GNU iconv's iconv.h defines it |
| 196 | to ENOENT, while gnulib defines it to a different value. Always |
| 197 | map ENOENT to gnulib's EILSEQ, leaving callers agnostic. */ |
| 198 | |
| 199 | static size_t |
| 200 | gdb_iconv (iconv_t utf_flag, ICONV_CONST char **inbuf, size_t *inbytesleft, |
| 201 | char **outbuf, size_t *outbytesleft) |
| 202 | { |
| 203 | size_t ret; |
| 204 | |
| 205 | ret = iconv (utf_flag, inbuf, inbytesleft, outbuf, outbytesleft); |
| 206 | if (errno == ENOENT) |
| 207 | errno = EILSEQ; |
| 208 | return ret; |
| 209 | } |
| 210 | |
| 211 | #undef iconv |
| 212 | #define iconv gdb_iconv |
| 213 | |
| 214 | #endif /* PHONY_ICONV */ |
| 215 | |
| 216 | \f |
| 217 | /* The global lists of character sets and translations. */ |
| 218 | |
| 219 | |
| 220 | #ifndef GDB_DEFAULT_TARGET_CHARSET |
| 221 | #define GDB_DEFAULT_TARGET_CHARSET "ISO-8859-1" |
| 222 | #endif |
| 223 | |
| 224 | #ifndef GDB_DEFAULT_TARGET_WIDE_CHARSET |
| 225 | #define GDB_DEFAULT_TARGET_WIDE_CHARSET "UTF-32" |
| 226 | #endif |
| 227 | |
| 228 | static const char *auto_host_charset_name = GDB_DEFAULT_HOST_CHARSET; |
| 229 | static const char *host_charset_name = "auto"; |
| 230 | static void |
| 231 | show_host_charset_name (struct ui_file *file, int from_tty, |
| 232 | struct cmd_list_element *c, |
| 233 | const char *value) |
| 234 | { |
| 235 | if (!strcmp (value, "auto")) |
| 236 | fprintf_filtered (file, |
| 237 | _("The host character set is \"auto; currently %s\".\n"), |
| 238 | auto_host_charset_name); |
| 239 | else |
| 240 | fprintf_filtered (file, _("The host character set is \"%s\".\n"), value); |
| 241 | } |
| 242 | |
| 243 | static const char *target_charset_name = "auto"; |
| 244 | static void |
| 245 | show_target_charset_name (struct ui_file *file, int from_tty, |
| 246 | struct cmd_list_element *c, const char *value) |
| 247 | { |
| 248 | if (!strcmp (value, "auto")) |
| 249 | fprintf_filtered (file, |
| 250 | _("The target character set is \"auto; " |
| 251 | "currently %s\".\n"), |
| 252 | gdbarch_auto_charset (get_current_arch ())); |
| 253 | else |
| 254 | fprintf_filtered (file, _("The target character set is \"%s\".\n"), |
| 255 | value); |
| 256 | } |
| 257 | |
| 258 | static const char *target_wide_charset_name = "auto"; |
| 259 | static void |
| 260 | show_target_wide_charset_name (struct ui_file *file, |
| 261 | int from_tty, |
| 262 | struct cmd_list_element *c, |
| 263 | const char *value) |
| 264 | { |
| 265 | if (!strcmp (value, "auto")) |
| 266 | fprintf_filtered (file, |
| 267 | _("The target wide character set is \"auto; " |
| 268 | "currently %s\".\n"), |
| 269 | gdbarch_auto_wide_charset (get_current_arch ())); |
| 270 | else |
| 271 | fprintf_filtered (file, _("The target wide character set is \"%s\".\n"), |
| 272 | value); |
| 273 | } |
| 274 | |
| 275 | static const char *default_charset_names[] = |
| 276 | { |
| 277 | DEFAULT_CHARSET_NAMES |
| 278 | 0 |
| 279 | }; |
| 280 | |
| 281 | static const char **charset_enum; |
| 282 | |
| 283 | \f |
| 284 | /* If the target wide character set has big- or little-endian |
| 285 | variants, these are the corresponding names. */ |
| 286 | static const char *target_wide_charset_be_name; |
| 287 | static const char *target_wide_charset_le_name; |
| 288 | |
| 289 | /* The architecture for which the BE- and LE-names are valid. */ |
| 290 | static struct gdbarch *be_le_arch; |
| 291 | |
| 292 | /* A helper function which sets the target wide big- and little-endian |
| 293 | character set names, if possible. */ |
| 294 | |
| 295 | static void |
| 296 | set_be_le_names (struct gdbarch *gdbarch) |
| 297 | { |
| 298 | int i, len; |
| 299 | const char *target_wide; |
| 300 | |
| 301 | if (be_le_arch == gdbarch) |
| 302 | return; |
| 303 | be_le_arch = gdbarch; |
| 304 | |
| 305 | #ifdef PHONY_ICONV |
| 306 | /* Match the wide charset names recognized by phony_iconv_open. */ |
| 307 | target_wide_charset_le_name = "UTF-32LE"; |
| 308 | target_wide_charset_be_name = "UTF-32BE"; |
| 309 | #else |
| 310 | target_wide_charset_le_name = NULL; |
| 311 | target_wide_charset_be_name = NULL; |
| 312 | |
| 313 | target_wide = target_wide_charset_name; |
| 314 | if (!strcmp (target_wide, "auto")) |
| 315 | target_wide = gdbarch_auto_wide_charset (gdbarch); |
| 316 | |
| 317 | len = strlen (target_wide); |
| 318 | for (i = 0; charset_enum[i]; ++i) |
| 319 | { |
| 320 | if (strncmp (target_wide, charset_enum[i], len)) |
| 321 | continue; |
| 322 | if ((charset_enum[i][len] == 'B' |
| 323 | || charset_enum[i][len] == 'L') |
| 324 | && charset_enum[i][len + 1] == 'E' |
| 325 | && charset_enum[i][len + 2] == '\0') |
| 326 | { |
| 327 | if (charset_enum[i][len] == 'B') |
| 328 | target_wide_charset_be_name = charset_enum[i]; |
| 329 | else |
| 330 | target_wide_charset_le_name = charset_enum[i]; |
| 331 | } |
| 332 | } |
| 333 | # endif /* PHONY_ICONV */ |
| 334 | } |
| 335 | |
| 336 | /* 'Set charset', 'set host-charset', 'set target-charset', 'set |
| 337 | target-wide-charset', 'set charset' sfunc's. */ |
| 338 | |
| 339 | static void |
| 340 | validate (struct gdbarch *gdbarch) |
| 341 | { |
| 342 | iconv_t desc; |
| 343 | const char *host_cset = host_charset (); |
| 344 | const char *target_cset = target_charset (gdbarch); |
| 345 | const char *target_wide_cset = target_wide_charset_name; |
| 346 | |
| 347 | if (!strcmp (target_wide_cset, "auto")) |
| 348 | target_wide_cset = gdbarch_auto_wide_charset (gdbarch); |
| 349 | |
| 350 | desc = iconv_open (target_wide_cset, host_cset); |
| 351 | if (desc == (iconv_t) -1) |
| 352 | error (_("Cannot convert between character sets `%s' and `%s'"), |
| 353 | target_wide_cset, host_cset); |
| 354 | iconv_close (desc); |
| 355 | |
| 356 | desc = iconv_open (target_cset, host_cset); |
| 357 | if (desc == (iconv_t) -1) |
| 358 | error (_("Cannot convert between character sets `%s' and `%s'"), |
| 359 | target_cset, host_cset); |
| 360 | iconv_close (desc); |
| 361 | |
| 362 | /* Clear the cache. */ |
| 363 | be_le_arch = NULL; |
| 364 | } |
| 365 | |
| 366 | /* This is the sfunc for the 'set charset' command. */ |
| 367 | static void |
| 368 | set_charset_sfunc (const char *charset, int from_tty, |
| 369 | struct cmd_list_element *c) |
| 370 | { |
| 371 | /* CAREFUL: set the target charset here as well. */ |
| 372 | target_charset_name = host_charset_name; |
| 373 | validate (get_current_arch ()); |
| 374 | } |
| 375 | |
| 376 | /* 'set host-charset' command sfunc. We need a wrapper here because |
| 377 | the function needs to have a specific signature. */ |
| 378 | static void |
| 379 | set_host_charset_sfunc (const char *charset, int from_tty, |
| 380 | struct cmd_list_element *c) |
| 381 | { |
| 382 | validate (get_current_arch ()); |
| 383 | } |
| 384 | |
| 385 | /* Wrapper for the 'set target-charset' command. */ |
| 386 | static void |
| 387 | set_target_charset_sfunc (const char *charset, int from_tty, |
| 388 | struct cmd_list_element *c) |
| 389 | { |
| 390 | validate (get_current_arch ()); |
| 391 | } |
| 392 | |
| 393 | /* Wrapper for the 'set target-wide-charset' command. */ |
| 394 | static void |
| 395 | set_target_wide_charset_sfunc (const char *charset, int from_tty, |
| 396 | struct cmd_list_element *c) |
| 397 | { |
| 398 | validate (get_current_arch ()); |
| 399 | } |
| 400 | |
| 401 | /* sfunc for the 'show charset' command. */ |
| 402 | static void |
| 403 | show_charset (struct ui_file *file, int from_tty, |
| 404 | struct cmd_list_element *c, |
| 405 | const char *name) |
| 406 | { |
| 407 | show_host_charset_name (file, from_tty, c, host_charset_name); |
| 408 | show_target_charset_name (file, from_tty, c, target_charset_name); |
| 409 | show_target_wide_charset_name (file, from_tty, c, |
| 410 | target_wide_charset_name); |
| 411 | } |
| 412 | |
| 413 | \f |
| 414 | /* Accessor functions. */ |
| 415 | |
| 416 | const char * |
| 417 | host_charset (void) |
| 418 | { |
| 419 | if (!strcmp (host_charset_name, "auto")) |
| 420 | return auto_host_charset_name; |
| 421 | return host_charset_name; |
| 422 | } |
| 423 | |
| 424 | const char * |
| 425 | target_charset (struct gdbarch *gdbarch) |
| 426 | { |
| 427 | if (!strcmp (target_charset_name, "auto")) |
| 428 | return gdbarch_auto_charset (gdbarch); |
| 429 | return target_charset_name; |
| 430 | } |
| 431 | |
| 432 | const char * |
| 433 | target_wide_charset (struct gdbarch *gdbarch) |
| 434 | { |
| 435 | enum bfd_endian byte_order = gdbarch_byte_order (gdbarch); |
| 436 | |
| 437 | set_be_le_names (gdbarch); |
| 438 | if (byte_order == BFD_ENDIAN_BIG) |
| 439 | { |
| 440 | if (target_wide_charset_be_name) |
| 441 | return target_wide_charset_be_name; |
| 442 | } |
| 443 | else |
| 444 | { |
| 445 | if (target_wide_charset_le_name) |
| 446 | return target_wide_charset_le_name; |
| 447 | } |
| 448 | |
| 449 | if (!strcmp (target_wide_charset_name, "auto")) |
| 450 | return gdbarch_auto_wide_charset (gdbarch); |
| 451 | |
| 452 | return target_wide_charset_name; |
| 453 | } |
| 454 | |
| 455 | \f |
| 456 | /* Host character set management. For the time being, we assume that |
| 457 | the host character set is some superset of ASCII. */ |
| 458 | |
| 459 | char |
| 460 | host_letter_to_control_character (char c) |
| 461 | { |
| 462 | if (c == '?') |
| 463 | return 0177; |
| 464 | return c & 0237; |
| 465 | } |
| 466 | |
| 467 | /* Convert a host character, C, to its hex value. C must already have |
| 468 | been validated using isxdigit. */ |
| 469 | |
| 470 | int |
| 471 | host_hex_value (char c) |
| 472 | { |
| 473 | if (isdigit (c)) |
| 474 | return c - '0'; |
| 475 | if (c >= 'a' && c <= 'f') |
| 476 | return 10 + c - 'a'; |
| 477 | gdb_assert (c >= 'A' && c <= 'F'); |
| 478 | return 10 + c - 'A'; |
| 479 | } |
| 480 | |
| 481 | \f |
| 482 | /* Public character management functions. */ |
| 483 | |
| 484 | class iconv_wrapper |
| 485 | { |
| 486 | public: |
| 487 | |
| 488 | iconv_wrapper (const char *to, const char *from) |
| 489 | { |
| 490 | m_desc = iconv_open (to, from); |
| 491 | if (m_desc == (iconv_t) -1) |
| 492 | perror_with_name (_("Converting character sets")); |
| 493 | } |
| 494 | |
| 495 | ~iconv_wrapper () |
| 496 | { |
| 497 | iconv_close (m_desc); |
| 498 | } |
| 499 | |
| 500 | size_t convert (ICONV_CONST char **inp, size_t *inleft, char **outp, |
| 501 | size_t *outleft) |
| 502 | { |
| 503 | return iconv (m_desc, inp, inleft, outp, outleft); |
| 504 | } |
| 505 | |
| 506 | private: |
| 507 | |
| 508 | iconv_t m_desc; |
| 509 | }; |
| 510 | |
| 511 | void |
| 512 | convert_between_encodings (const char *from, const char *to, |
| 513 | const gdb_byte *bytes, unsigned int num_bytes, |
| 514 | int width, struct obstack *output, |
| 515 | enum transliterations translit) |
| 516 | { |
| 517 | size_t inleft; |
| 518 | ICONV_CONST char *inp; |
| 519 | unsigned int space_request; |
| 520 | |
| 521 | /* Often, the host and target charsets will be the same. */ |
| 522 | if (!strcmp (from, to)) |
| 523 | { |
| 524 | obstack_grow (output, bytes, num_bytes); |
| 525 | return; |
| 526 | } |
| 527 | |
| 528 | iconv_wrapper desc (to, from); |
| 529 | |
| 530 | inleft = num_bytes; |
| 531 | inp = (ICONV_CONST char *) bytes; |
| 532 | |
| 533 | space_request = num_bytes; |
| 534 | |
| 535 | while (inleft > 0) |
| 536 | { |
| 537 | char *outp; |
| 538 | size_t outleft, r; |
| 539 | int old_size; |
| 540 | |
| 541 | old_size = obstack_object_size (output); |
| 542 | obstack_blank (output, space_request); |
| 543 | |
| 544 | outp = (char *) obstack_base (output) + old_size; |
| 545 | outleft = space_request; |
| 546 | |
| 547 | r = desc.convert (&inp, &inleft, &outp, &outleft); |
| 548 | |
| 549 | /* Now make sure that the object on the obstack only includes |
| 550 | bytes we have converted. */ |
| 551 | obstack_blank_fast (output, -outleft); |
| 552 | |
| 553 | if (r == (size_t) -1) |
| 554 | { |
| 555 | switch (errno) |
| 556 | { |
| 557 | case EILSEQ: |
| 558 | { |
| 559 | int i; |
| 560 | |
| 561 | /* Invalid input sequence. */ |
| 562 | if (translit == translit_none) |
| 563 | error (_("Could not convert character " |
| 564 | "to `%s' character set"), to); |
| 565 | |
| 566 | /* We emit escape sequence for the bytes, skip them, |
| 567 | and try again. */ |
| 568 | for (i = 0; i < width; ++i) |
| 569 | { |
| 570 | char octal[5]; |
| 571 | |
| 572 | xsnprintf (octal, sizeof (octal), "\\%.3o", *inp & 0xff); |
| 573 | obstack_grow_str (output, octal); |
| 574 | |
| 575 | ++inp; |
| 576 | --inleft; |
| 577 | } |
| 578 | } |
| 579 | break; |
| 580 | |
| 581 | case E2BIG: |
| 582 | /* We ran out of space in the output buffer. Make it |
| 583 | bigger next time around. */ |
| 584 | space_request *= 2; |
| 585 | break; |
| 586 | |
| 587 | case EINVAL: |
| 588 | /* Incomplete input sequence. FIXME: ought to report this |
| 589 | to the caller somehow. */ |
| 590 | inleft = 0; |
| 591 | break; |
| 592 | |
| 593 | default: |
| 594 | perror_with_name (_("Internal error while " |
| 595 | "converting character sets")); |
| 596 | } |
| 597 | } |
| 598 | } |
| 599 | } |
| 600 | |
| 601 | \f |
| 602 | |
| 603 | /* Create a new iterator. */ |
| 604 | wchar_iterator::wchar_iterator (const gdb_byte *input, size_t bytes, |
| 605 | const char *charset, size_t width) |
| 606 | : m_input (input), |
| 607 | m_bytes (bytes), |
| 608 | m_width (width), |
| 609 | m_out (1) |
| 610 | { |
| 611 | m_desc = iconv_open (INTERMEDIATE_ENCODING, charset); |
| 612 | if (m_desc == (iconv_t) -1) |
| 613 | perror_with_name (_("Converting character sets")); |
| 614 | } |
| 615 | |
| 616 | wchar_iterator::~wchar_iterator () |
| 617 | { |
| 618 | if (m_desc != (iconv_t) -1) |
| 619 | iconv_close (m_desc); |
| 620 | } |
| 621 | |
| 622 | int |
| 623 | wchar_iterator::iterate (enum wchar_iterate_result *out_result, |
| 624 | gdb_wchar_t **out_chars, |
| 625 | const gdb_byte **ptr, |
| 626 | size_t *len) |
| 627 | { |
| 628 | size_t out_request; |
| 629 | |
| 630 | /* Try to convert some characters. At first we try to convert just |
| 631 | a single character. The reason for this is that iconv does not |
| 632 | necessarily update its outgoing arguments when it encounters an |
| 633 | invalid input sequence -- but we want to reliably report this to |
| 634 | our caller so it can emit an escape sequence. */ |
| 635 | out_request = 1; |
| 636 | while (m_bytes > 0) |
| 637 | { |
| 638 | ICONV_CONST char *inptr = (ICONV_CONST char *) m_input; |
| 639 | char *outptr = (char *) m_out.data (); |
| 640 | const gdb_byte *orig_inptr = m_input; |
| 641 | size_t orig_in = m_bytes; |
| 642 | size_t out_avail = out_request * sizeof (gdb_wchar_t); |
| 643 | size_t num; |
| 644 | size_t r = iconv (m_desc, &inptr, &m_bytes, &outptr, &out_avail); |
| 645 | |
| 646 | m_input = (gdb_byte *) inptr; |
| 647 | |
| 648 | if (r == (size_t) -1) |
| 649 | { |
| 650 | switch (errno) |
| 651 | { |
| 652 | case EILSEQ: |
| 653 | /* Invalid input sequence. We still might have |
| 654 | converted a character; if so, return it. */ |
| 655 | if (out_avail < out_request * sizeof (gdb_wchar_t)) |
| 656 | break; |
| 657 | |
| 658 | /* Otherwise skip the first invalid character, and let |
| 659 | the caller know about it. */ |
| 660 | *out_result = wchar_iterate_invalid; |
| 661 | *ptr = m_input; |
| 662 | *len = m_width; |
| 663 | m_input += m_width; |
| 664 | m_bytes -= m_width; |
| 665 | return 0; |
| 666 | |
| 667 | case E2BIG: |
| 668 | /* We ran out of space. We still might have converted a |
| 669 | character; if so, return it. Otherwise, grow the |
| 670 | buffer and try again. */ |
| 671 | if (out_avail < out_request * sizeof (gdb_wchar_t)) |
| 672 | break; |
| 673 | |
| 674 | ++out_request; |
| 675 | if (out_request > m_out.size ()) |
| 676 | m_out.resize (out_request); |
| 677 | continue; |
| 678 | |
| 679 | case EINVAL: |
| 680 | /* Incomplete input sequence. Let the caller know, and |
| 681 | arrange for future calls to see EOF. */ |
| 682 | *out_result = wchar_iterate_incomplete; |
| 683 | *ptr = m_input; |
| 684 | *len = m_bytes; |
| 685 | m_bytes = 0; |
| 686 | return 0; |
| 687 | |
| 688 | default: |
| 689 | perror_with_name (_("Internal error while " |
| 690 | "converting character sets")); |
| 691 | } |
| 692 | } |
| 693 | |
| 694 | /* We converted something. */ |
| 695 | num = out_request - out_avail / sizeof (gdb_wchar_t); |
| 696 | *out_result = wchar_iterate_ok; |
| 697 | *out_chars = m_out.data (); |
| 698 | *ptr = orig_inptr; |
| 699 | *len = orig_in - m_bytes; |
| 700 | return num; |
| 701 | } |
| 702 | |
| 703 | /* Really done. */ |
| 704 | *out_result = wchar_iterate_eof; |
| 705 | return -1; |
| 706 | } |
| 707 | |
| 708 | /* The charset.c module initialization function. */ |
| 709 | |
| 710 | static VEC (char_ptr) *charsets; |
| 711 | |
| 712 | #ifdef PHONY_ICONV |
| 713 | |
| 714 | static void |
| 715 | find_charset_names (void) |
| 716 | { |
| 717 | /* Cast is fine here, because CHARSETS is never released. Note that |
| 718 | the vec does not hold "const char *" pointers instead of "char *" |
| 719 | because the non-phony version stores heap-allocated strings in |
| 720 | it. */ |
| 721 | VEC_safe_push (char_ptr, charsets, (char *) GDB_DEFAULT_HOST_CHARSET); |
| 722 | VEC_safe_push (char_ptr, charsets, NULL); |
| 723 | } |
| 724 | |
| 725 | #else /* PHONY_ICONV */ |
| 726 | |
| 727 | /* Sometimes, libiconv redefines iconvlist as libiconvlist -- but |
| 728 | provides different symbols in the static and dynamic libraries. |
| 729 | So, configure may see libiconvlist but not iconvlist. But, calling |
| 730 | iconvlist is the right thing to do and will work. Hence we do a |
| 731 | check here but unconditionally call iconvlist below. */ |
| 732 | #if defined (HAVE_ICONVLIST) || defined (HAVE_LIBICONVLIST) |
| 733 | |
| 734 | /* A helper function that adds some character sets to the vector of |
| 735 | all character sets. This is a callback function for iconvlist. */ |
| 736 | |
| 737 | static int |
| 738 | add_one (unsigned int count, const char *const *names, void *data) |
| 739 | { |
| 740 | unsigned int i; |
| 741 | |
| 742 | for (i = 0; i < count; ++i) |
| 743 | VEC_safe_push (char_ptr, charsets, xstrdup (names[i])); |
| 744 | |
| 745 | return 0; |
| 746 | } |
| 747 | |
| 748 | static void |
| 749 | find_charset_names (void) |
| 750 | { |
| 751 | iconvlist (add_one, NULL); |
| 752 | VEC_safe_push (char_ptr, charsets, NULL); |
| 753 | } |
| 754 | |
| 755 | #else |
| 756 | |
| 757 | /* Return non-zero if LINE (output from iconv) should be ignored. |
| 758 | Older iconv programs (e.g. 2.2.2) include the human readable |
| 759 | introduction even when stdout is not a tty. Newer versions omit |
| 760 | the intro if stdout is not a tty. */ |
| 761 | |
| 762 | static int |
| 763 | ignore_line_p (const char *line) |
| 764 | { |
| 765 | /* This table is used to filter the output. If this text appears |
| 766 | anywhere in the line, it is ignored (strstr is used). */ |
| 767 | static const char * const ignore_lines[] = |
| 768 | { |
| 769 | "The following", |
| 770 | "not necessarily", |
| 771 | "the FROM and TO", |
| 772 | "listed with several", |
| 773 | NULL |
| 774 | }; |
| 775 | int i; |
| 776 | |
| 777 | for (i = 0; ignore_lines[i] != NULL; ++i) |
| 778 | { |
| 779 | if (strstr (line, ignore_lines[i]) != NULL) |
| 780 | return 1; |
| 781 | } |
| 782 | |
| 783 | return 0; |
| 784 | } |
| 785 | |
| 786 | static void |
| 787 | find_charset_names (void) |
| 788 | { |
| 789 | struct pex_obj *child; |
| 790 | const char *args[3]; |
| 791 | int err, status; |
| 792 | int fail = 1; |
| 793 | int flags; |
| 794 | gdb_environ iconv_env = gdb_environ::from_host_environ (); |
| 795 | char *iconv_program; |
| 796 | |
| 797 | /* Older iconvs, e.g. 2.2.2, don't omit the intro text if stdout is |
| 798 | not a tty. We need to recognize it and ignore it. This text is |
| 799 | subject to translation, so force LANGUAGE=C. */ |
| 800 | iconv_env.set ("LANGUAGE", "C"); |
| 801 | iconv_env.set ("LC_ALL", "C"); |
| 802 | |
| 803 | child = pex_init (PEX_USE_PIPES, "iconv", NULL); |
| 804 | |
| 805 | #ifdef ICONV_BIN |
| 806 | { |
| 807 | char *iconv_dir = relocate_gdb_directory (ICONV_BIN, |
| 808 | ICONV_BIN_RELOCATABLE); |
| 809 | iconv_program = concat (iconv_dir, SLASH_STRING, "iconv", NULL); |
| 810 | xfree (iconv_dir); |
| 811 | } |
| 812 | #else |
| 813 | iconv_program = xstrdup ("iconv"); |
| 814 | #endif |
| 815 | args[0] = iconv_program; |
| 816 | args[1] = "-l"; |
| 817 | args[2] = NULL; |
| 818 | flags = PEX_STDERR_TO_STDOUT; |
| 819 | #ifndef ICONV_BIN |
| 820 | flags |= PEX_SEARCH; |
| 821 | #endif |
| 822 | /* Note that we simply ignore errors here. */ |
| 823 | if (!pex_run_in_environment (child, flags, |
| 824 | args[0], const_cast<char **> (args), |
| 825 | iconv_env.envp (), |
| 826 | NULL, NULL, &err)) |
| 827 | { |
| 828 | FILE *in = pex_read_output (child, 0); |
| 829 | |
| 830 | /* POSIX says that iconv -l uses an unspecified format. We |
| 831 | parse the glibc and libiconv formats; feel free to add others |
| 832 | as needed. */ |
| 833 | |
| 834 | while (in != NULL && !feof (in)) |
| 835 | { |
| 836 | /* The size of buf is chosen arbitrarily. */ |
| 837 | char buf[1024]; |
| 838 | char *start, *r; |
| 839 | int len; |
| 840 | |
| 841 | r = fgets (buf, sizeof (buf), in); |
| 842 | if (!r) |
| 843 | break; |
| 844 | len = strlen (r); |
| 845 | if (len <= 3) |
| 846 | continue; |
| 847 | if (ignore_line_p (r)) |
| 848 | continue; |
| 849 | |
| 850 | /* Strip off the newline. */ |
| 851 | --len; |
| 852 | /* Strip off one or two '/'s. glibc will print lines like |
| 853 | "8859_7//", but also "10646-1:1993/UCS4/". */ |
| 854 | if (buf[len - 1] == '/') |
| 855 | --len; |
| 856 | if (buf[len - 1] == '/') |
| 857 | --len; |
| 858 | buf[len] = '\0'; |
| 859 | |
| 860 | /* libiconv will print multiple entries per line, separated |
| 861 | by spaces. Older iconvs will print multiple entries per |
| 862 | line, indented by two spaces, and separated by ", " |
| 863 | (i.e. the human readable form). */ |
| 864 | start = buf; |
| 865 | while (1) |
| 866 | { |
| 867 | int keep_going; |
| 868 | char *p; |
| 869 | |
| 870 | /* Skip leading blanks. */ |
| 871 | for (p = start; *p && *p == ' '; ++p) |
| 872 | ; |
| 873 | start = p; |
| 874 | /* Find the next space, comma, or end-of-line. */ |
| 875 | for ( ; *p && *p != ' ' && *p != ','; ++p) |
| 876 | ; |
| 877 | /* Ignore an empty result. */ |
| 878 | if (p == start) |
| 879 | break; |
| 880 | keep_going = *p; |
| 881 | *p = '\0'; |
| 882 | VEC_safe_push (char_ptr, charsets, xstrdup (start)); |
| 883 | if (!keep_going) |
| 884 | break; |
| 885 | /* Skip any extra spaces. */ |
| 886 | for (start = p + 1; *start && *start == ' '; ++start) |
| 887 | ; |
| 888 | } |
| 889 | } |
| 890 | |
| 891 | if (pex_get_status (child, 1, &status) |
| 892 | && WIFEXITED (status) && !WEXITSTATUS (status)) |
| 893 | fail = 0; |
| 894 | |
| 895 | } |
| 896 | |
| 897 | xfree (iconv_program); |
| 898 | pex_free (child); |
| 899 | |
| 900 | if (fail) |
| 901 | { |
| 902 | /* Some error occurred, so drop the vector. */ |
| 903 | free_char_ptr_vec (charsets); |
| 904 | charsets = NULL; |
| 905 | } |
| 906 | else |
| 907 | VEC_safe_push (char_ptr, charsets, NULL); |
| 908 | } |
| 909 | |
| 910 | #endif /* HAVE_ICONVLIST || HAVE_LIBICONVLIST */ |
| 911 | #endif /* PHONY_ICONV */ |
| 912 | |
| 913 | /* The "auto" target charset used by default_auto_charset. */ |
| 914 | static const char *auto_target_charset_name = GDB_DEFAULT_TARGET_CHARSET; |
| 915 | |
| 916 | const char * |
| 917 | default_auto_charset (void) |
| 918 | { |
| 919 | return auto_target_charset_name; |
| 920 | } |
| 921 | |
| 922 | const char * |
| 923 | default_auto_wide_charset (void) |
| 924 | { |
| 925 | return GDB_DEFAULT_TARGET_WIDE_CHARSET; |
| 926 | } |
| 927 | |
| 928 | |
| 929 | #ifdef USE_INTERMEDIATE_ENCODING_FUNCTION |
| 930 | /* Macro used for UTF or UCS endianness suffix. */ |
| 931 | #if WORDS_BIGENDIAN |
| 932 | #define ENDIAN_SUFFIX "BE" |
| 933 | #else |
| 934 | #define ENDIAN_SUFFIX "LE" |
| 935 | #endif |
| 936 | |
| 937 | /* The code below serves to generate a compile time error if |
| 938 | gdb_wchar_t type is not of size 2 nor 4, despite the fact that |
| 939 | macro __STDC_ISO_10646__ is defined. |
| 940 | This is better than a gdb_assert call, because GDB cannot handle |
| 941 | strings correctly if this size is different. */ |
| 942 | |
| 943 | extern char your_gdb_wchar_t_is_bogus[(sizeof (gdb_wchar_t) == 2 |
| 944 | || sizeof (gdb_wchar_t) == 4) |
| 945 | ? 1 : -1]; |
| 946 | |
| 947 | /* intermediate_encoding returns the charset used internally by |
| 948 | GDB to convert between target and host encodings. As the test above |
| 949 | compiled, sizeof (gdb_wchar_t) is either 2 or 4 bytes. |
| 950 | UTF-16/32 is tested first, UCS-2/4 is tested as a second option, |
| 951 | otherwise an error is generated. */ |
| 952 | |
| 953 | const char * |
| 954 | intermediate_encoding (void) |
| 955 | { |
| 956 | iconv_t desc; |
| 957 | static const char *stored_result = NULL; |
| 958 | char *result; |
| 959 | |
| 960 | if (stored_result) |
| 961 | return stored_result; |
| 962 | result = xstrprintf ("UTF-%d%s", (int) (sizeof (gdb_wchar_t) * 8), |
| 963 | ENDIAN_SUFFIX); |
| 964 | /* Check that the name is supported by iconv_open. */ |
| 965 | desc = iconv_open (result, host_charset ()); |
| 966 | if (desc != (iconv_t) -1) |
| 967 | { |
| 968 | iconv_close (desc); |
| 969 | stored_result = result; |
| 970 | return result; |
| 971 | } |
| 972 | /* Not valid, free the allocated memory. */ |
| 973 | xfree (result); |
| 974 | /* Second try, with UCS-2 type. */ |
| 975 | result = xstrprintf ("UCS-%d%s", (int) sizeof (gdb_wchar_t), |
| 976 | ENDIAN_SUFFIX); |
| 977 | /* Check that the name is supported by iconv_open. */ |
| 978 | desc = iconv_open (result, host_charset ()); |
| 979 | if (desc != (iconv_t) -1) |
| 980 | { |
| 981 | iconv_close (desc); |
| 982 | stored_result = result; |
| 983 | return result; |
| 984 | } |
| 985 | /* Not valid, free the allocated memory. */ |
| 986 | xfree (result); |
| 987 | /* No valid charset found, generate error here. */ |
| 988 | error (_("Unable to find a vaild charset for string conversions")); |
| 989 | } |
| 990 | |
| 991 | #endif /* USE_INTERMEDIATE_ENCODING_FUNCTION */ |
| 992 | |
| 993 | void |
| 994 | _initialize_charset (void) |
| 995 | { |
| 996 | /* The first element is always "auto". */ |
| 997 | VEC_safe_push (char_ptr, charsets, xstrdup ("auto")); |
| 998 | find_charset_names (); |
| 999 | |
| 1000 | if (VEC_length (char_ptr, charsets) > 1) |
| 1001 | charset_enum = (const char **) VEC_address (char_ptr, charsets); |
| 1002 | else |
| 1003 | charset_enum = default_charset_names; |
| 1004 | |
| 1005 | #ifndef PHONY_ICONV |
| 1006 | #ifdef HAVE_LANGINFO_CODESET |
| 1007 | /* The result of nl_langinfo may be overwritten later. This may |
| 1008 | leak a little memory, if the user later changes the host charset, |
| 1009 | but that doesn't matter much. */ |
| 1010 | auto_host_charset_name = xstrdup (nl_langinfo (CODESET)); |
| 1011 | /* Solaris will return `646' here -- but the Solaris iconv then does |
| 1012 | not accept this. Darwin (and maybe FreeBSD) may return "" here, |
| 1013 | which GNU libiconv doesn't like (infinite loop). */ |
| 1014 | if (!strcmp (auto_host_charset_name, "646") || !*auto_host_charset_name) |
| 1015 | auto_host_charset_name = "ASCII"; |
| 1016 | auto_target_charset_name = auto_host_charset_name; |
| 1017 | #elif defined (USE_WIN32API) |
| 1018 | { |
| 1019 | /* "CP" + x<=5 digits + paranoia. */ |
| 1020 | static char w32_host_default_charset[16]; |
| 1021 | |
| 1022 | snprintf (w32_host_default_charset, sizeof w32_host_default_charset, |
| 1023 | "CP%d", GetACP()); |
| 1024 | auto_host_charset_name = w32_host_default_charset; |
| 1025 | auto_target_charset_name = auto_host_charset_name; |
| 1026 | } |
| 1027 | #endif |
| 1028 | #endif |
| 1029 | |
| 1030 | add_setshow_enum_cmd ("charset", class_support, |
| 1031 | charset_enum, &host_charset_name, _("\ |
| 1032 | Set the host and target character sets."), _("\ |
| 1033 | Show the host and target character sets."), _("\ |
| 1034 | The `host character set' is the one used by the system GDB is running on.\n\ |
| 1035 | The `target character set' is the one used by the program being debugged.\n\ |
| 1036 | You may only use supersets of ASCII for your host character set; GDB does\n\ |
| 1037 | not support any others.\n\ |
| 1038 | To see a list of the character sets GDB supports, type `set charset <TAB>'."), |
| 1039 | /* Note that the sfunc below needs to set |
| 1040 | target_charset_name, because the 'set |
| 1041 | charset' command sets two variables. */ |
| 1042 | set_charset_sfunc, |
| 1043 | show_charset, |
| 1044 | &setlist, &showlist); |
| 1045 | |
| 1046 | add_setshow_enum_cmd ("host-charset", class_support, |
| 1047 | charset_enum, &host_charset_name, _("\ |
| 1048 | Set the host character set."), _("\ |
| 1049 | Show the host character set."), _("\ |
| 1050 | The `host character set' is the one used by the system GDB is running on.\n\ |
| 1051 | You may only use supersets of ASCII for your host character set; GDB does\n\ |
| 1052 | not support any others.\n\ |
| 1053 | To see a list of the character sets GDB supports, type `set host-charset <TAB>'."), |
| 1054 | set_host_charset_sfunc, |
| 1055 | show_host_charset_name, |
| 1056 | &setlist, &showlist); |
| 1057 | |
| 1058 | add_setshow_enum_cmd ("target-charset", class_support, |
| 1059 | charset_enum, &target_charset_name, _("\ |
| 1060 | Set the target character set."), _("\ |
| 1061 | Show the target character set."), _("\ |
| 1062 | The `target character set' is the one used by the program being debugged.\n\ |
| 1063 | GDB translates characters and strings between the host and target\n\ |
| 1064 | character sets as needed.\n\ |
| 1065 | To see a list of the character sets GDB supports, type `set target-charset'<TAB>"), |
| 1066 | set_target_charset_sfunc, |
| 1067 | show_target_charset_name, |
| 1068 | &setlist, &showlist); |
| 1069 | |
| 1070 | add_setshow_enum_cmd ("target-wide-charset", class_support, |
| 1071 | charset_enum, &target_wide_charset_name, |
| 1072 | _("\ |
| 1073 | Set the target wide character set."), _("\ |
| 1074 | Show the target wide character set."), _("\ |
| 1075 | The `target wide character set' is the one used by the program being debugged.\ |
| 1076 | \nIn particular it is the encoding used by `wchar_t'.\n\ |
| 1077 | GDB translates characters and strings between the host and target\n\ |
| 1078 | character sets as needed.\n\ |
| 1079 | To see a list of the character sets GDB supports, type\n\ |
| 1080 | `set target-wide-charset'<TAB>"), |
| 1081 | set_target_wide_charset_sfunc, |
| 1082 | show_target_wide_charset_name, |
| 1083 | &setlist, &showlist); |
| 1084 | } |