D-Bus 1.6.12
|
00001 /* -*- mode: C; c-file-style: "gnu"; indent-tabs-mode: nil; -*- */ 00002 /* dbus-shell.c Shell command line utility functions. 00003 * 00004 * Copyright (C) 2002, 2003 Red Hat, Inc. 00005 * Copyright (C) 2003 CodeFactory AB 00006 * 00007 * Licensed under the Academic Free License version 2.1 00008 * 00009 * This program is free software; you can redistribute it and/or modify 00010 * it under the terms of the GNU General Public License as published by 00011 * the Free Software Foundation; either version 2 of the License, or 00012 * (at your option) any later version. 00013 * 00014 * This program is distributed in the hope that it will be useful, 00015 * but WITHOUT ANY WARRANTY; without even the implied warranty of 00016 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 00017 * GNU General Public License for more details. 00018 * 00019 * You should have received a copy of the GNU General Public License 00020 * along with this program; if not, write to the Free Software 00021 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 00022 * 00023 */ 00024 00025 #include <config.h> 00026 #include <string.h> 00027 #include "dbus-internals.h" 00028 #include "dbus-list.h" 00029 #include "dbus-memory.h" 00030 #include "dbus-protocol.h" 00031 #include "dbus-shell.h" 00032 #include "dbus-string.h" 00033 00034 /* Single quotes preserve the literal string exactly. escape 00035 * sequences are not allowed; not even \' - if you want a ' 00036 * in the quoted text, you have to do something like 'foo'\''bar' 00037 * 00038 * Double quotes allow $ ` " \ and newline to be escaped with backslash. 00039 * Otherwise double quotes preserve things literally. 00040 */ 00041 00042 static dbus_bool_t 00043 unquote_string_inplace (char* str, char** end) 00044 { 00045 char* dest; 00046 char* s; 00047 char quote_char; 00048 00049 dest = s = str; 00050 00051 quote_char = *s; 00052 00053 if (!(*s == '"' || *s == '\'')) 00054 { 00055 *end = str; 00056 return FALSE; 00057 } 00058 00059 /* Skip the initial quote mark */ 00060 ++s; 00061 00062 if (quote_char == '"') 00063 { 00064 while (*s) 00065 { 00066 _dbus_assert(s > dest); /* loop invariant */ 00067 00068 switch (*s) 00069 { 00070 case '"': 00071 /* End of the string, return now */ 00072 *dest = '\0'; 00073 ++s; 00074 *end = s; 00075 return TRUE; 00076 00077 case '\\': 00078 /* Possible escaped quote or \ */ 00079 ++s; 00080 switch (*s) 00081 { 00082 case '"': 00083 case '\\': 00084 case '`': 00085 case '$': 00086 case '\n': 00087 *dest = *s; 00088 ++s; 00089 ++dest; 00090 break; 00091 00092 default: 00093 /* not an escaped char */ 00094 *dest = '\\'; 00095 ++dest; 00096 /* ++s already done. */ 00097 break; 00098 } 00099 break; 00100 00101 default: 00102 *dest = *s; 00103 ++dest; 00104 ++s; 00105 break; 00106 } 00107 00108 _dbus_assert(s > dest); /* loop invariant */ 00109 } 00110 } 00111 else 00112 { 00113 while (*s) 00114 { 00115 _dbus_assert(s > dest); /* loop invariant */ 00116 00117 if (*s == '\'') 00118 { 00119 /* End of the string, return now */ 00120 *dest = '\0'; 00121 ++s; 00122 *end = s; 00123 return TRUE; 00124 } 00125 else 00126 { 00127 *dest = *s; 00128 ++dest; 00129 ++s; 00130 } 00131 00132 _dbus_assert(s > dest); /* loop invariant */ 00133 } 00134 } 00135 00136 /* If we reach here this means the close quote was never encountered */ 00137 00138 *dest = '\0'; 00139 00140 *end = s; 00141 return FALSE; 00142 } 00143 00168 char* 00169 _dbus_shell_unquote (const char *quoted_string) 00170 { 00171 char *unquoted; 00172 char *end; 00173 char *start; 00174 char *ret; 00175 DBusString retval; 00176 00177 unquoted = _dbus_strdup (quoted_string); 00178 if (unquoted == NULL) 00179 return NULL; 00180 00181 start = unquoted; 00182 end = unquoted; 00183 if (!_dbus_string_init (&retval)) 00184 { 00185 dbus_free (unquoted); 00186 return NULL; 00187 } 00188 00189 /* The loop allows cases such as 00190 * "foo"blah blah'bar'woo foo"baz"la la la\'\''foo' 00191 */ 00192 while (*start) 00193 { 00194 /* Append all non-quoted chars, honoring backslash escape 00195 */ 00196 00197 while (*start && !(*start == '"' || *start == '\'')) 00198 { 00199 if (*start == '\\') 00200 { 00201 /* all characters can get escaped by backslash, 00202 * except newline, which is removed if it follows 00203 * a backslash outside of quotes 00204 */ 00205 00206 ++start; 00207 if (*start) 00208 { 00209 if (*start != '\n') 00210 { 00211 if (!_dbus_string_append_byte (&retval, *start)) 00212 goto error; 00213 } 00214 ++start; 00215 } 00216 } 00217 else 00218 { 00219 if (!_dbus_string_append_byte (&retval, *start)) 00220 goto error; 00221 ++start; 00222 } 00223 } 00224 00225 if (*start) 00226 { 00227 if (!unquote_string_inplace (start, &end)) 00228 goto error; 00229 else 00230 { 00231 if (!_dbus_string_append (&retval, start)) 00232 goto error; 00233 start = end; 00234 } 00235 } 00236 } 00237 00238 ret = _dbus_strdup (_dbus_string_get_data (&retval)); 00239 if (!ret) 00240 goto error; 00241 00242 dbus_free (unquoted); 00243 _dbus_string_free (&retval); 00244 00245 return ret; 00246 00247 error: 00248 dbus_free (unquoted); 00249 _dbus_string_free (&retval); 00250 return NULL; 00251 } 00252 00253 /* _dbus_shell_parse_argv() does a semi-arbitrary weird subset of the way 00254 * the shell parses a command line. We don't do variable expansion, 00255 * don't understand that operators are tokens, don't do tilde expansion, 00256 * don't do command substitution, no arithmetic expansion, IFS gets ignored, 00257 * don't do filename globs, don't remove redirection stuff, etc. 00258 * 00259 * READ THE UNIX98 SPEC on "Shell Command Language" before changing 00260 * the behavior of this code. 00261 * 00262 * Steps to parsing the argv string: 00263 * 00264 * - tokenize the string (but since we ignore operators, 00265 * our tokenization may diverge from what the shell would do) 00266 * note that tokenization ignores the internals of a quoted 00267 * word and it always splits on spaces, not on IFS even 00268 * if we used IFS. We also ignore "end of input indicator" 00269 * (I guess this is control-D?) 00270 * 00271 * Tokenization steps, from UNIX98 with operator stuff removed, 00272 * are: 00273 * 00274 * 1) "If the current character is backslash, single-quote or 00275 * double-quote (\, ' or ") and it is not quoted, it will affect 00276 * quoting for subsequent characters up to the end of the quoted 00277 * text. The rules for quoting are as described in Quoting 00278 * . During token recognition no substitutions will be actually 00279 * performed, and the result token will contain exactly the 00280 * characters that appear in the input (except for newline 00281 * character joining), unmodified, including any embedded or 00282 * enclosing quotes or substitution operators, between the quote 00283 * mark and the end of the quoted text. The token will not be 00284 * delimited by the end of the quoted field." 00285 * 00286 * 2) "If the current character is an unquoted newline character, 00287 * the current token will be delimited." 00288 * 00289 * 3) "If the current character is an unquoted blank character, any 00290 * token containing the previous character is delimited and the 00291 * current character will be discarded." 00292 * 00293 * 4) "If the previous character was part of a word, the current 00294 * character will be appended to that word." 00295 * 00296 * 5) "If the current character is a "#", it and all subsequent 00297 * characters up to, but excluding, the next newline character 00298 * will be discarded as a comment. The newline character that 00299 * ends the line is not considered part of the comment. The 00300 * "#" starts a comment only when it is at the beginning of a 00301 * token. Since the search for the end-of-comment does not 00302 * consider an escaped newline character specially, a comment 00303 * cannot be continued to the next line." 00304 * 00305 * 6) "The current character will be used as the start of a new word." 00306 * 00307 * 00308 * - for each token (word), perform portions of word expansion, namely 00309 * field splitting (using default whitespace IFS) and quote 00310 * removal. Field splitting may increase the number of words. 00311 * Quote removal does not increase the number of words. 00312 * 00313 * "If the complete expansion appropriate for a word results in an 00314 * empty field, that empty field will be deleted from the list of 00315 * fields that form the completely expanded command, unless the 00316 * original word contained single-quote or double-quote characters." 00317 * - UNIX98 spec 00318 * 00319 * 00320 */ 00321 00322 static dbus_bool_t 00323 delimit_token (DBusString *token, 00324 DBusList **retval, 00325 DBusError *error) 00326 { 00327 char *str; 00328 00329 str = _dbus_strdup (_dbus_string_get_data (token)); 00330 if (!str) 00331 { 00332 _DBUS_SET_OOM (error); 00333 return FALSE; 00334 } 00335 00336 if (!_dbus_list_append (retval, str)) 00337 { 00338 dbus_free (str); 00339 _DBUS_SET_OOM (error); 00340 return FALSE; 00341 } 00342 00343 return TRUE; 00344 } 00345 00346 static DBusList* 00347 tokenize_command_line (const char *command_line, DBusError *error) 00348 { 00349 char current_quote; 00350 const char *p; 00351 DBusString current_token; 00352 DBusList *retval = NULL; 00353 dbus_bool_t quoted;; 00354 00355 current_quote = '\0'; 00356 quoted = FALSE; 00357 p = command_line; 00358 00359 if (!_dbus_string_init (¤t_token)) 00360 { 00361 _DBUS_SET_OOM (error); 00362 return NULL; 00363 } 00364 00365 while (*p) 00366 { 00367 if (current_quote == '\\') 00368 { 00369 if (*p == '\n') 00370 { 00371 /* we append nothing; backslash-newline become nothing */ 00372 } 00373 else 00374 { 00375 if (!_dbus_string_append_byte (¤t_token, '\\') || 00376 !_dbus_string_append_byte (¤t_token, *p)) 00377 { 00378 _DBUS_SET_OOM (error); 00379 goto error; 00380 } 00381 } 00382 00383 current_quote = '\0'; 00384 } 00385 else if (current_quote == '#') 00386 { 00387 /* Discard up to and including next newline */ 00388 while (*p && *p != '\n') 00389 ++p; 00390 00391 current_quote = '\0'; 00392 00393 if (*p == '\0') 00394 break; 00395 } 00396 else if (current_quote) 00397 { 00398 if (*p == current_quote && 00399 /* check that it isn't an escaped double quote */ 00400 !(current_quote == '"' && quoted)) 00401 { 00402 /* close the quote */ 00403 current_quote = '\0'; 00404 } 00405 00406 /* Everything inside quotes, and the close quote, 00407 * gets appended literally. 00408 */ 00409 00410 if (!_dbus_string_append_byte (¤t_token, *p)) 00411 { 00412 _DBUS_SET_OOM (error); 00413 goto error; 00414 } 00415 } 00416 else 00417 { 00418 switch (*p) 00419 { 00420 case '\n': 00421 if (!delimit_token (¤t_token, &retval, error)) 00422 goto error; 00423 00424 _dbus_string_free (¤t_token); 00425 00426 if (!_dbus_string_init (¤t_token)) 00427 { 00428 _DBUS_SET_OOM (error); 00429 goto init_error; 00430 } 00431 00432 break; 00433 00434 case ' ': 00435 case '\t': 00436 /* If the current token contains the previous char, delimit 00437 * the current token. A nonzero length 00438 * token should always contain the previous char. 00439 */ 00440 if (_dbus_string_get_length (¤t_token) > 0) 00441 { 00442 if (!delimit_token (¤t_token, &retval, error)) 00443 goto error; 00444 00445 _dbus_string_free (¤t_token); 00446 00447 if (!_dbus_string_init (¤t_token)) 00448 { 00449 _DBUS_SET_OOM (error); 00450 goto init_error; 00451 } 00452 00453 } 00454 00455 /* discard all unquoted blanks (don't add them to a token) */ 00456 break; 00457 00458 00459 /* single/double quotes are appended to the token, 00460 * escapes are maybe appended next time through the loop, 00461 * comment chars are never appended. 00462 */ 00463 00464 case '\'': 00465 case '"': 00466 if (!_dbus_string_append_byte (¤t_token, *p)) 00467 { 00468 _DBUS_SET_OOM (error); 00469 goto error; 00470 } 00471 00472 /* FALL THRU */ 00473 00474 case '#': 00475 case '\\': 00476 current_quote = *p; 00477 break; 00478 00479 default: 00480 /* Combines rules 4) and 6) - if we have a token, append to it, 00481 * otherwise create a new token. 00482 */ 00483 if (!_dbus_string_append_byte (¤t_token, *p)) 00484 { 00485 _DBUS_SET_OOM (error); 00486 goto error; 00487 } 00488 break; 00489 } 00490 } 00491 00492 /* We need to count consecutive backslashes mod 2, 00493 * to detect escaped doublequotes. 00494 */ 00495 if (*p != '\\') 00496 quoted = FALSE; 00497 else 00498 quoted = !quoted; 00499 00500 ++p; 00501 } 00502 00503 if (!delimit_token (¤t_token, &retval, error)) 00504 goto error; 00505 00506 if (current_quote) 00507 { 00508 dbus_set_error_const (error, DBUS_ERROR_INVALID_ARGS, "Unclosed quotes in command line"); 00509 goto error; 00510 } 00511 00512 if (retval == NULL) 00513 { 00514 dbus_set_error_const (error, DBUS_ERROR_INVALID_ARGS, "No tokens found in command line"); 00515 goto error; 00516 } 00517 00518 _dbus_string_free (¤t_token); 00519 00520 return retval; 00521 00522 error: 00523 _dbus_string_free (¤t_token); 00524 00525 init_error: 00526 if (retval) 00527 { 00528 _dbus_list_foreach (&retval, (DBusForeachFunction) dbus_free, NULL); 00529 _dbus_list_clear (&retval); 00530 } 00531 00532 return NULL; 00533 } 00534 00552 dbus_bool_t 00553 _dbus_shell_parse_argv (const char *command_line, 00554 int *argcp, 00555 char ***argvp, 00556 DBusError *error) 00557 { 00558 /* Code based on poptParseArgvString() from libpopt */ 00559 int argc = 0; 00560 char **argv = NULL; 00561 DBusList *tokens = NULL; 00562 int i; 00563 DBusList *tmp_list; 00564 00565 if (!command_line) 00566 { 00567 _dbus_verbose ("Command line is NULL\n"); 00568 return FALSE; 00569 } 00570 00571 tokens = tokenize_command_line (command_line, error); 00572 if (tokens == NULL) 00573 { 00574 _dbus_verbose ("No tokens for command line '%s'\n", command_line); 00575 return FALSE; 00576 } 00577 00578 /* Because we can't have introduced any new blank space into the 00579 * tokens (we didn't do any new expansions), we don't need to 00580 * perform field splitting. If we were going to honor IFS or do any 00581 * expansions, we would have to do field splitting on each word 00582 * here. Also, if we were going to do any expansion we would need to 00583 * remove any zero-length words that didn't contain quotes 00584 * originally; but since there's no expansion we know all words have 00585 * nonzero length, unless they contain quotes. 00586 * 00587 * So, we simply remove quotes, and don't do any field splitting or 00588 * empty word removal, since we know there was no way to introduce 00589 * such things. 00590 */ 00591 00592 argc = _dbus_list_get_length (&tokens); 00593 argv = dbus_new (char *, argc + 1); 00594 if (!argv) 00595 { 00596 _DBUS_SET_OOM (error); 00597 goto error; 00598 } 00599 00600 i = 0; 00601 tmp_list = tokens; 00602 while (tmp_list) 00603 { 00604 argv[i] = _dbus_shell_unquote (tmp_list->data); 00605 00606 if (!argv[i]) 00607 { 00608 int j; 00609 for (j = 0; j < i; j++) 00610 dbus_free(argv[j]); 00611 00612 dbus_free (argv); 00613 _DBUS_SET_OOM (error); 00614 goto error; 00615 } 00616 00617 tmp_list = _dbus_list_get_next_link (&tokens, tmp_list); 00618 ++i; 00619 } 00620 argv[argc] = NULL; 00621 00622 _dbus_list_foreach (&tokens, (DBusForeachFunction) dbus_free, NULL); 00623 _dbus_list_clear (&tokens); 00624 00625 if (argcp) 00626 *argcp = argc; 00627 00628 if (argvp) 00629 *argvp = argv; 00630 else 00631 dbus_free_string_array (argv); 00632 00633 return TRUE; 00634 00635 error: 00636 _dbus_list_foreach (&tokens, (DBusForeachFunction) dbus_free, NULL); 00637 _dbus_list_clear (&tokens); 00638 00639 return FALSE; 00640 00641 }