2 Bacula® - The Network Backup Solution
4 Copyright (C) 2000-2007 Free Software Foundation Europe e.V.
6 The main author of Bacula is Kern Sibbald, with contributions from
7 many others, a complete list can be found in the file AUTHORS.
8 This program is Free Software; you can redistribute it and/or
9 modify it under the terms of version two of the GNU General Public
10 License as published by the Free Software Foundation and included
13 This program is distributed in the hope that it will be useful, but
14 WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, write to the Free Software
20 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
23 Bacula® is a registered trademark of John Walker.
24 The licensor of Bacula is the Free Software Foundation Europe
25 (FSFE), Fiduciary Program, Sumatrastrasse 25, 8006 Zürich,
26 Switzerland, email:ftf@fsfeurope.org.
29 * Lexical scanner for Bacula configuration file
40 extern int debug_level;
42 /* Debug level for this source file */
43 static const int dbglvl = 5000;
46 * Scan to "logical" end of line. I.e. end of line,
47 * or semicolon, but stop on T_EOB (same as end of
48 * line except it is not eaten).
50 void scan_to_eol(LEX *lc)
53 Dmsg0(dbglvl, "start scan to eof\n");
54 while ((token = lex_get_token(lc, T_ALL)) != T_EOL) {
63 * Get next token, but skip EOL
65 int scan_to_next_not_eol(LEX * lc)
69 token = lex_get_token(lc, T_ALL);
70 } while (token == T_EOL);
75 * Format a scanner error message
77 static void s_err(const char *file, int line, LEX *lc, const char *msg, ...)
83 va_start(arg_ptr, msg);
84 bvsnprintf(buf, sizeof(buf), msg, arg_ptr);
87 if (lc->err_type == 0) { /* M_ERROR_TERM by default */
88 lc->err_type = M_ERROR_TERM;
91 if (lc->line_no > lc->begin_line_no) {
92 bsnprintf(more, sizeof(more),
93 _("Problem probably begins at line %d.\n"), lc->begin_line_no);
97 if (lc->line_no > 0) {
98 e_msg(file, line, lc->err_type, 0, _("Config error: %s\n"
99 " : line %d, col %d of file %s\n%s\n%s"),
100 buf, lc->line_no, lc->col_no, lc->fname, lc->line, more);
102 e_msg(file, line, lc->err_type, 0, _("Config error: %s\n"), buf);
106 void lex_set_default_error_handler(LEX *lf)
108 lf->scan_error = s_err;
112 * Set err_type used in error_handler
113 * return the old value
115 int lex_set_error_handler_error_type(LEX *lf, int err_type)
117 int old = lf->err_type;
118 lf->err_type = err_type;
123 * Free the current file, and retrieve the contents
124 * of the previous packet if any.
126 LEX *lex_close_file(LEX *lf)
131 Emsg0(M_ABORT, 0, _("Close of NULL file\n"));
133 Dmsg1(dbglvl, "Close lex file: %s\n", lf->fname);
137 close_bpipe(lf->bpipe);
142 Dmsg1(dbglvl, "Close cfg file %s\n", lf->fname);
145 of->options = lf->options; /* preserve options */
146 memcpy(lf, of, sizeof(LEX));
147 Dmsg1(dbglvl, "Restart scan of cfg file %s\n", of->fname);
157 * Open a new configuration file. We push the
158 * state of the current file (lf) so that we
159 * can do includes. This is a bit of a hammer.
160 * Instead of passing back the pointer to the
161 * new packet, I simply replace the contents
162 * of the caller's packet with the new packet,
163 * and link the contents of the old packet into
167 LEX *lex_open_file(LEX *lf, const char *filename, LEX_ERROR_HANDLER *scan_error)
173 char *fname = bstrdup(filename);
176 if (fname[0] == '|') {
177 if ((bpipe = open_bpipe(fname+1, 0, "rb")) == NULL) {
182 } else if ((fd = fopen(fname, "rb")) == NULL) {
186 Dmsg1(400, "Open config file: %s\n", fname);
187 nf = (LEX *)malloc(sizeof(LEX));
189 memcpy(nf, lf, sizeof(LEX));
190 memset(lf, 0, sizeof(LEX));
191 lf->next = nf; /* if have lf, push it behind new one */
192 lf->options = nf->options; /* preserve user options */
194 * preserve err_type to prevent bacula exiting on 'reload'
195 * if config is invalid. Fixes bug #877
197 lf->err_type = nf->err_type;
199 lf = nf; /* start new packet */
200 memset(lf, 0, sizeof(LEX));
201 lex_set_error_handler_error_type(lf, M_ERROR_TERM);
204 lf->scan_error = scan_error;
206 lex_set_default_error_handler(lf);
211 lf->state = lex_none;
213 Dmsg1(dbglvl, "Return lex=%x\n", lf);
218 * Get the next character from the input.
219 * Returns the character or
220 * L_EOF if end of file
221 * L_EOL if end of line
223 int lex_get_char(LEX *lf)
225 if (lf->ch == L_EOF) {
226 Emsg0(M_ABORT, 0, _("get_char: called after EOF."
227 " You may have a open double quote without the closing double quote.\n"));
229 if (lf->ch == L_EOL) {
230 if (bfgets(lf->line, MAXSTRING, lf->fd) == NULL) {
239 Dmsg2(1000, "fget line=%d %s", lf->line_no, lf->line);
241 lf->ch = (uint8_t)lf->line[lf->col_no];
247 Dmsg2(dbglvl, "lex_get_char: %c %d\n", lf->ch, lf->ch);
251 void lex_unget_char(LEX *lf)
253 if (lf->ch == L_EOL) {
254 lf->ch = 0; /* End of line, force read of next one */
256 lf->col_no--; /* Backup to re-read char */
263 * Add a character to the current string
265 static void add_str(LEX *lf, int ch)
267 if (lf->str_len >= MAXSTRING-3) {
268 Emsg3(M_ERROR_TERM, 0, _(
269 _("Config token too long, file: %s, line %d, begins at line %d\n")),
270 lf->fname, lf->line_no, lf->begin_line_no);
272 lf->str[lf->str_len++] = ch;
273 lf->str[lf->str_len] = 0;
279 static void begin_str(LEX *lf, int ch)
286 lf->begin_line_no = lf->line_no; /* save start string line no */
290 static const char *lex_state_to_str(int state)
293 case lex_none: return _("none");
294 case lex_comment: return _("comment");
295 case lex_number: return _("number");
296 case lex_ip_addr: return _("ip_addr");
297 case lex_identifier: return _("identifier");
298 case lex_string: return _("string");
299 case lex_quoted_string: return _("quoted_string");
300 case lex_include: return _("include");
301 case lex_include_quoted_string: return _("include_quoted_string");
302 case lex_utf8_bom: return _("UTF-8 Byte Order Mark");
303 case lex_utf16_le_bom: return _("UTF-16le Byte Order Mark");
304 default: return "??????";
310 * Convert a lex token to a string
311 * used for debug/error printing.
313 const char *lex_tok_to_str(int token)
316 case L_EOF: return "L_EOF";
317 case L_EOL: return "L_EOL";
318 case T_NONE: return "T_NONE";
319 case T_NUMBER: return "T_NUMBER";
320 case T_IPADDR: return "T_IPADDR";
321 case T_IDENTIFIER: return "T_IDENTIFIER";
322 case T_UNQUOTED_STRING: return "T_UNQUOTED_STRING";
323 case T_QUOTED_STRING: return "T_QUOTED_STRING";
324 case T_BOB: return "T_BOB";
325 case T_EOB: return "T_EOB";
326 case T_EQUALS: return "T_EQUALS";
327 case T_ERROR: return "T_ERROR";
328 case T_EOF: return "T_EOF";
329 case T_COMMA: return "T_COMMA";
330 case T_EOL: return "T_EOL";
331 case T_UTF8_BOM: return "T_UTF8_BOM";
332 case T_UTF16_BOM: return "T_UTF16_BOM";
333 default: return "??????";
337 static uint32_t scan_pint(LEX *lf, char *str)
340 if (!is_a_number(str)) {
341 scan_err1(lf, _("expected a positive integer number, got: %s"), str);
345 val = str_to_int64(str);
346 if (errno != 0 || val < 0) {
347 scan_err1(lf, _("expected a positive integer number, got: %s"), str);
351 return (uint32_t)val;
356 * Get the next token from the input
360 lex_get_token(LEX *lf, int expect)
364 bool esc_next = false;
365 /* Unicode files, especially on Win32, may begin with a "Byte Order Mark"
366 to indicate which transmission format the file is in. The codepoint for
367 this mark is U+FEFF and is represented as the octets EF-BB-BF in UTF-8
368 and as FF-FE in UTF-16le(little endian) and FE-FF in UTF-16(big endian).
369 We use a distinct state for UTF-8 and UTF-16le, and use bom_bytes_seen
370 to tell which byte we are expecting. */
371 int bom_bytes_seen = 0;
373 Dmsg0(dbglvl, "enter lex_get_token\n");
374 while (token == T_NONE) {
375 ch = lex_get_char(lf);
378 Dmsg2(dbglvl, "Lex state lex_none ch=%d,%x\n", ch, ch);
382 if (lf->options & LOPT_NO_IDENT || lf->options & LOPT_STRING) {
383 lf->state = lex_string;
385 lf->state = lex_identifier;
391 if (lf->options & LOPT_STRING) {
392 lf->state = lex_string;
394 lf->state = lex_number;
399 Dmsg0(dbglvl, "Enter lex_none switch\n");
403 Dmsg0(dbglvl, "got L_EOF set token=T_EOF\n");
406 lf->state = lex_comment;
417 lf->state = lex_quoted_string;
429 if (expect != T_SKIP_EOL) {
430 token = T_EOL; /* treat ; like EOL */
434 Dmsg0(dbglvl, "got L_EOL set token=T_EOL\n");
435 if (expect != T_SKIP_EOL) {
440 lf->state = lex_include;
443 case 0xEF: /* probably a UTF-8 BOM */
444 case 0xFF: /* probably a UTF-16le BOM */
445 case 0xFE: /* probably a UTF-16be BOM (error)*/
446 if (lf->line_no != 1 || lf->col_no != 1)
448 lf->state = lex_string;
453 lf->state = lex_utf8_bom;
454 } else if (ch == 0xFF) {
455 lf->state = lex_utf16_le_bom;
457 scan_err0(lf, _("This config file appears to be in an "
458 "unsupported Unicode format (UTF-16be). Please resave as UTF-8\n"));
464 lf->state = lex_string;
470 Dmsg1(dbglvl, "Lex state lex_comment ch=%x\n", ch);
472 lf->state = lex_none;
473 if (expect != T_SKIP_EOL) {
476 } else if (ch == L_EOF) {
481 Dmsg2(dbglvl, "Lex state lex_number ch=%x %c\n", ch, ch);
486 /* Might want to allow trailing specifications here */
492 /* A valid number can be terminated by the following */
493 if (B_ISSPACE(ch) || ch == L_EOL || ch == ',' || ch == ';') {
495 lf->state = lex_none;
497 lf->state = lex_string;
506 Dmsg1(dbglvl, "Lex state lex_ip_addr ch=%x\n", ch);
509 Dmsg1(dbglvl, "Lex state lex_string ch=%x\n", ch);
514 if (ch == '\n' || ch == L_EOL || ch == '=' || ch == '}' || ch == '{' ||
515 ch == '\r' || ch == ';' || ch == ',' || ch == '#' || (B_ISSPACE(ch)) ) {
517 token = T_UNQUOTED_STRING;
518 lf->state = lex_none;
524 Dmsg2(dbglvl, "Lex state lex_identifier ch=%x %c\n", ch, ch);
528 } else if (B_ISSPACE(ch)) {
530 } else if (ch == '\n' || ch == L_EOL || ch == '=' || ch == '}' || ch == '{' ||
531 ch == '\r' || ch == ';' || ch == ',' || ch == '"' || ch == '#') {
533 token = T_IDENTIFIER;
534 lf->state = lex_none;
536 } else if (ch == L_EOF) {
538 lf->state = lex_none;
542 /* Some non-alpha character => string */
543 lf->state = lex_string;
546 case lex_quoted_string:
547 Dmsg2(dbglvl, "Lex state lex_quoted_string ch=%x %c\n", ch, ch);
566 token = T_QUOTED_STRING;
567 lf->state = lex_none;
572 case lex_include_quoted_string:
587 /* Keep the original LEX so we can print an error if the included file can't be opened. */
589 /* Skip the double quote when restarting parsing */
592 lf->state = lex_none;
593 lf = lex_open_file(lf, lf->str, lf->scan_error);
596 scan_err2(lfori, _("Cannot open included config file %s: %s\n"),
597 lfori->str, be.bstrerror());
604 case lex_include: /* scanning a filename */
610 lf->state = lex_include_quoted_string;
615 if (B_ISSPACE(ch) || ch == '\n' || ch == L_EOL || ch == '}' || ch == '{' ||
616 ch == ';' || ch == ',' || ch == '"' || ch == '#') {
617 /* Keep the original LEX so we can print an error if the included file can't be opened. */
620 lf->state = lex_none;
621 lf = lex_open_file(lf, lf->str, lf->scan_error);
624 scan_err2(lfori, _("Cannot open included config file %s: %s\n"),
625 lfori->str, be.bstrerror());
633 /* we only end up in this state if we have read an 0xEF
634 as the first byte of the file, indicating we are probably
635 reading a UTF-8 file */
636 if (ch == 0xBB && bom_bytes_seen == 1) {
638 } else if (ch == 0xBF && bom_bytes_seen == 2) {
640 lf->state = lex_none;
645 case lex_utf16_le_bom:
646 /* we only end up in this state if we have read an 0xFF
647 as the first byte of the file -- indicating that we are
648 probably dealing with an Intel based (little endian) UTF-16 file*/
651 lf->state = lex_none;
657 Dmsg4(dbglvl, "ch=%d state=%s token=%s %c\n", ch, lex_state_to_str(lf->state),
658 lex_tok_to_str(token), ch);
660 Dmsg2(dbglvl, "lex returning: line %d token: %s\n", lf->line_no, lex_tok_to_str(token));
664 * Here is where we check to see if the user has set certain
665 * expectations (e.g. 32 bit integer). If so, we do type checking
666 * and possible additional scanning (e.g. for range).
670 lf->pint32_val = scan_pint(lf, lf->str);
671 lf->pint32_val2 = lf->pint32_val;
676 if (token == T_NUMBER) {
677 lf->pint32_val = scan_pint(lf, lf->str);
678 lf->pint32_val2 = lf->pint32_val;
681 char *p = strchr(lf->str, '-');
683 scan_err2(lf, _("expected an integer or a range, got %s: %s"),
684 lex_tok_to_str(token), lf->str);
688 *p++ = 0; /* terminate first half of range */
689 lf->pint32_val = scan_pint(lf, lf->str);
690 lf->pint32_val2 = scan_pint(lf, p);
691 token = T_PINT32_RANGE;
696 if (token != T_NUMBER || !is_a_number(lf->str)) {
697 scan_err2(lf, _("expected an integer number, got %s: %s"),
698 lex_tok_to_str(token), lf->str);
703 lf->int32_val = (int32_t)str_to_int64(lf->str);
705 scan_err2(lf, _("expected an integer number, got %s: %s"),
706 lex_tok_to_str(token), lf->str);
714 Dmsg2(dbglvl, "int64=:%s: %f\n", lf->str, strtod(lf->str, NULL));
715 if (token != T_NUMBER || !is_a_number(lf->str)) {
716 scan_err2(lf, _("expected an integer number, got %s: %s"),
717 lex_tok_to_str(token), lf->str);
722 lf->int64_val = str_to_int64(lf->str);
724 scan_err2(lf, _("expected an integer number, got %s: %s"),
725 lex_tok_to_str(token), lf->str);
733 if (token != T_IDENTIFIER && token != T_UNQUOTED_STRING && token != T_QUOTED_STRING) {
734 scan_err2(lf, _("expected a name, got %s: %s"),
735 lex_tok_to_str(token), lf->str);
737 } else if (lf->str_len > MAX_RES_NAME_LENGTH) {
738 scan_err3(lf, _("name %s length %d too long, max is %d\n"), lf->str,
739 lf->str_len, MAX_RES_NAME_LENGTH);
745 if (token != T_IDENTIFIER && token != T_UNQUOTED_STRING && token != T_QUOTED_STRING) {
746 scan_err2(lf, _("expected a string, got %s: %s"),
747 lex_tok_to_str(token), lf->str);
756 break; /* no expectation given */
758 lf->token = token; /* set possible new token */