| 1 | /* Parse a printf-style format string. |
| 2 | |
| 3 | Copyright (C) 1986-2015 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 "common-defs.h" |
| 21 | #include "format.h" |
| 22 | |
| 23 | struct format_piece * |
| 24 | parse_format_string (const char **arg) |
| 25 | { |
| 26 | const char *s; |
| 27 | char *f, *string; |
| 28 | const char *prev_start; |
| 29 | const char *percent_loc; |
| 30 | char *sub_start, *current_substring; |
| 31 | struct format_piece *pieces; |
| 32 | int next_frag; |
| 33 | int max_pieces; |
| 34 | enum argclass this_argclass; |
| 35 | |
| 36 | s = *arg; |
| 37 | |
| 38 | /* Parse the format-control string and copy it into the string STRING, |
| 39 | processing some kinds of escape sequence. */ |
| 40 | |
| 41 | f = string = (char *) alloca (strlen (s) + 1); |
| 42 | |
| 43 | while (*s != '"' && *s != '\0') |
| 44 | { |
| 45 | int c = *s++; |
| 46 | switch (c) |
| 47 | { |
| 48 | case '\0': |
| 49 | continue; |
| 50 | |
| 51 | case '\\': |
| 52 | switch (c = *s++) |
| 53 | { |
| 54 | case '\\': |
| 55 | *f++ = '\\'; |
| 56 | break; |
| 57 | case 'a': |
| 58 | *f++ = '\a'; |
| 59 | break; |
| 60 | case 'b': |
| 61 | *f++ = '\b'; |
| 62 | break; |
| 63 | case 'f': |
| 64 | *f++ = '\f'; |
| 65 | break; |
| 66 | case 'n': |
| 67 | *f++ = '\n'; |
| 68 | break; |
| 69 | case 'r': |
| 70 | *f++ = '\r'; |
| 71 | break; |
| 72 | case 't': |
| 73 | *f++ = '\t'; |
| 74 | break; |
| 75 | case 'v': |
| 76 | *f++ = '\v'; |
| 77 | break; |
| 78 | case '"': |
| 79 | *f++ = '"'; |
| 80 | break; |
| 81 | default: |
| 82 | /* ??? TODO: handle other escape sequences. */ |
| 83 | error (_("Unrecognized escape character \\%c in format string."), |
| 84 | c); |
| 85 | } |
| 86 | break; |
| 87 | |
| 88 | default: |
| 89 | *f++ = c; |
| 90 | } |
| 91 | } |
| 92 | |
| 93 | /* Terminate our escape-processed copy. */ |
| 94 | *f++ = '\0'; |
| 95 | |
| 96 | /* Whether the format string ended with double-quote or zero, we're |
| 97 | done with it; it's up to callers to complain about syntax. */ |
| 98 | *arg = s; |
| 99 | |
| 100 | /* Need extra space for the '\0's. Doubling the size is sufficient. */ |
| 101 | |
| 102 | current_substring = xmalloc (strlen (string) * 2 + 1000); |
| 103 | |
| 104 | max_pieces = strlen (string) + 2; |
| 105 | |
| 106 | pieces = XNEWVEC (struct format_piece, max_pieces); |
| 107 | |
| 108 | next_frag = 0; |
| 109 | |
| 110 | /* Now scan the string for %-specs and see what kinds of args they want. |
| 111 | argclass classifies the %-specs so we can give printf-type functions |
| 112 | something of the right size. */ |
| 113 | |
| 114 | f = string; |
| 115 | prev_start = string; |
| 116 | while (*f) |
| 117 | if (*f++ == '%') |
| 118 | { |
| 119 | int seen_hash = 0, seen_zero = 0, lcount = 0, seen_prec = 0; |
| 120 | int seen_space = 0, seen_plus = 0; |
| 121 | int seen_big_l = 0, seen_h = 0, seen_big_h = 0; |
| 122 | int seen_big_d = 0, seen_double_big_d = 0; |
| 123 | int bad = 0; |
| 124 | |
| 125 | /* Skip over "%%", it will become part of a literal piece. */ |
| 126 | if (*f == '%') |
| 127 | { |
| 128 | f++; |
| 129 | continue; |
| 130 | } |
| 131 | |
| 132 | sub_start = current_substring; |
| 133 | |
| 134 | strncpy (current_substring, prev_start, f - 1 - prev_start); |
| 135 | current_substring += f - 1 - prev_start; |
| 136 | *current_substring++ = '\0'; |
| 137 | |
| 138 | pieces[next_frag].string = sub_start; |
| 139 | pieces[next_frag].argclass = literal_piece; |
| 140 | next_frag++; |
| 141 | |
| 142 | percent_loc = f - 1; |
| 143 | |
| 144 | /* Check the validity of the format specifier, and work |
| 145 | out what argument it expects. We only accept C89 |
| 146 | format strings, with the exception of long long (which |
| 147 | we autoconf for). */ |
| 148 | |
| 149 | /* The first part of a format specifier is a set of flag |
| 150 | characters. */ |
| 151 | while (*f != '\0' && strchr ("0-+ #", *f)) |
| 152 | { |
| 153 | if (*f == '#') |
| 154 | seen_hash = 1; |
| 155 | else if (*f == '0') |
| 156 | seen_zero = 1; |
| 157 | else if (*f == ' ') |
| 158 | seen_space = 1; |
| 159 | else if (*f == '+') |
| 160 | seen_plus = 1; |
| 161 | f++; |
| 162 | } |
| 163 | |
| 164 | /* The next part of a format specifier is a width. */ |
| 165 | while (*f != '\0' && strchr ("0123456789", *f)) |
| 166 | f++; |
| 167 | |
| 168 | /* The next part of a format specifier is a precision. */ |
| 169 | if (*f == '.') |
| 170 | { |
| 171 | seen_prec = 1; |
| 172 | f++; |
| 173 | while (*f != '\0' && strchr ("0123456789", *f)) |
| 174 | f++; |
| 175 | } |
| 176 | |
| 177 | /* The next part of a format specifier is a length modifier. */ |
| 178 | if (*f == 'h') |
| 179 | { |
| 180 | seen_h = 1; |
| 181 | f++; |
| 182 | } |
| 183 | else if (*f == 'l') |
| 184 | { |
| 185 | f++; |
| 186 | lcount++; |
| 187 | if (*f == 'l') |
| 188 | { |
| 189 | f++; |
| 190 | lcount++; |
| 191 | } |
| 192 | } |
| 193 | else if (*f == 'L') |
| 194 | { |
| 195 | seen_big_l = 1; |
| 196 | f++; |
| 197 | } |
| 198 | /* Decimal32 modifier. */ |
| 199 | else if (*f == 'H') |
| 200 | { |
| 201 | seen_big_h = 1; |
| 202 | f++; |
| 203 | } |
| 204 | /* Decimal64 and Decimal128 modifiers. */ |
| 205 | else if (*f == 'D') |
| 206 | { |
| 207 | f++; |
| 208 | |
| 209 | /* Check for a Decimal128. */ |
| 210 | if (*f == 'D') |
| 211 | { |
| 212 | f++; |
| 213 | seen_double_big_d = 1; |
| 214 | } |
| 215 | else |
| 216 | seen_big_d = 1; |
| 217 | } |
| 218 | |
| 219 | switch (*f) |
| 220 | { |
| 221 | case 'u': |
| 222 | if (seen_hash) |
| 223 | bad = 1; |
| 224 | /* FALLTHROUGH */ |
| 225 | |
| 226 | case 'o': |
| 227 | case 'x': |
| 228 | case 'X': |
| 229 | if (seen_space || seen_plus) |
| 230 | bad = 1; |
| 231 | /* FALLTHROUGH */ |
| 232 | |
| 233 | case 'd': |
| 234 | case 'i': |
| 235 | if (lcount == 0) |
| 236 | this_argclass = int_arg; |
| 237 | else if (lcount == 1) |
| 238 | this_argclass = long_arg; |
| 239 | else |
| 240 | this_argclass = long_long_arg; |
| 241 | |
| 242 | if (seen_big_l) |
| 243 | bad = 1; |
| 244 | break; |
| 245 | |
| 246 | case 'c': |
| 247 | this_argclass = lcount == 0 ? int_arg : wide_char_arg; |
| 248 | if (lcount > 1 || seen_h || seen_big_l) |
| 249 | bad = 1; |
| 250 | if (seen_prec || seen_zero || seen_space || seen_plus) |
| 251 | bad = 1; |
| 252 | break; |
| 253 | |
| 254 | case 'p': |
| 255 | this_argclass = ptr_arg; |
| 256 | if (lcount || seen_h || seen_big_l) |
| 257 | bad = 1; |
| 258 | if (seen_prec) |
| 259 | bad = 1; |
| 260 | if (seen_hash || seen_zero || seen_space || seen_plus) |
| 261 | bad = 1; |
| 262 | break; |
| 263 | |
| 264 | case 's': |
| 265 | this_argclass = lcount == 0 ? string_arg : wide_string_arg; |
| 266 | if (lcount > 1 || seen_h || seen_big_l) |
| 267 | bad = 1; |
| 268 | if (seen_zero || seen_space || seen_plus) |
| 269 | bad = 1; |
| 270 | break; |
| 271 | |
| 272 | case 'e': |
| 273 | case 'f': |
| 274 | case 'g': |
| 275 | case 'E': |
| 276 | case 'G': |
| 277 | if (seen_big_h || seen_big_d || seen_double_big_d) |
| 278 | this_argclass = decfloat_arg; |
| 279 | else if (seen_big_l) |
| 280 | this_argclass = long_double_arg; |
| 281 | else |
| 282 | this_argclass = double_arg; |
| 283 | |
| 284 | if (lcount || seen_h) |
| 285 | bad = 1; |
| 286 | break; |
| 287 | |
| 288 | case '*': |
| 289 | error (_("`*' not supported for precision or width in printf")); |
| 290 | |
| 291 | case 'n': |
| 292 | error (_("Format specifier `n' not supported in printf")); |
| 293 | |
| 294 | case '\0': |
| 295 | error (_("Incomplete format specifier at end of format string")); |
| 296 | |
| 297 | default: |
| 298 | error (_("Unrecognized format specifier '%c' in printf"), *f); |
| 299 | } |
| 300 | |
| 301 | if (bad) |
| 302 | error (_("Inappropriate modifiers to " |
| 303 | "format specifier '%c' in printf"), |
| 304 | *f); |
| 305 | |
| 306 | f++; |
| 307 | |
| 308 | sub_start = current_substring; |
| 309 | |
| 310 | if (lcount > 1 && USE_PRINTF_I64) |
| 311 | { |
| 312 | /* Windows' printf does support long long, but not the usual way. |
| 313 | Convert %lld to %I64d. */ |
| 314 | int length_before_ll = f - percent_loc - 1 - lcount; |
| 315 | |
| 316 | strncpy (current_substring, percent_loc, length_before_ll); |
| 317 | strcpy (current_substring + length_before_ll, "I64"); |
| 318 | current_substring[length_before_ll + 3] = |
| 319 | percent_loc[length_before_ll + lcount]; |
| 320 | current_substring += length_before_ll + 4; |
| 321 | } |
| 322 | else if (this_argclass == wide_string_arg |
| 323 | || this_argclass == wide_char_arg) |
| 324 | { |
| 325 | /* Convert %ls or %lc to %s. */ |
| 326 | int length_before_ls = f - percent_loc - 2; |
| 327 | |
| 328 | strncpy (current_substring, percent_loc, length_before_ls); |
| 329 | strcpy (current_substring + length_before_ls, "s"); |
| 330 | current_substring += length_before_ls + 2; |
| 331 | } |
| 332 | else |
| 333 | { |
| 334 | strncpy (current_substring, percent_loc, f - percent_loc); |
| 335 | current_substring += f - percent_loc; |
| 336 | } |
| 337 | |
| 338 | *current_substring++ = '\0'; |
| 339 | |
| 340 | prev_start = f; |
| 341 | |
| 342 | pieces[next_frag].string = sub_start; |
| 343 | pieces[next_frag].argclass = this_argclass; |
| 344 | next_frag++; |
| 345 | } |
| 346 | |
| 347 | /* Record the remainder of the string. */ |
| 348 | |
| 349 | sub_start = current_substring; |
| 350 | |
| 351 | strncpy (current_substring, prev_start, f - prev_start); |
| 352 | current_substring += f - prev_start; |
| 353 | *current_substring++ = '\0'; |
| 354 | |
| 355 | pieces[next_frag].string = sub_start; |
| 356 | pieces[next_frag].argclass = literal_piece; |
| 357 | next_frag++; |
| 358 | |
| 359 | /* Record an end-of-array marker. */ |
| 360 | |
| 361 | pieces[next_frag].string = NULL; |
| 362 | pieces[next_frag].argclass = literal_piece; |
| 363 | |
| 364 | return pieces; |
| 365 | } |
| 366 | |
| 367 | void |
| 368 | free_format_pieces (struct format_piece *pieces) |
| 369 | { |
| 370 | if (!pieces) |
| 371 | return; |
| 372 | |
| 373 | /* We happen to know that all the string pieces are in the block |
| 374 | pointed to by the first string piece. */ |
| 375 | if (pieces[0].string) |
| 376 | xfree (pieces[0].string); |
| 377 | |
| 378 | xfree (pieces); |
| 379 | } |
| 380 | |
| 381 | void |
| 382 | free_format_pieces_cleanup (void *ptr) |
| 383 | { |
| 384 | void **location = ptr; |
| 385 | |
| 386 | if (location == NULL) |
| 387 | return; |
| 388 | |
| 389 | if (*location != NULL) |
| 390 | { |
| 391 | free_format_pieces (*location); |
| 392 | *location = NULL; |
| 393 | } |
| 394 | } |
| 395 | |