3 * Copyright (c) 2013 Francois Fleuret
4 * Written by Francois Fleuret <francois@fleuret.org>
6 * This file is part of mymail.
8 * mymail is free software: you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License version 3 as
10 * published by the Free Software Foundation.
12 * mymail is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with mymail. If not, see <http://www.gnu.org/licenses/>.
24 This command is a dumb mail indexer. It can either (1) scan
25 directories containing mbox files, and create a db file containing
26 for each mail a list of fields computed from the header, or (2)
27 read such a db file and get all the mails matching regexp-defined
28 conditions on the fields, to create a resulting mbox file.
30 It is low-tech, simple, light and fast.
48 #define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file"
49 #define VERSION "0.9.5"
51 #define MAX_NB_SEARCH_CONDITIONS 32
53 #define BUFFER_SIZE 65536
54 #define TOKEN_BUFFER_SIZE 1024
56 #define LEADING_FROM_LINE_REGEXP_STRING "^From [^ ]* \\(Mon\\|Tue\\|Wed\\|Thu\\|Fri\\|Sat\\|Sun\\) \\(Jan\\|Feb\\|Mar\\|Apr\\|May\\|Jun\\|Jul\\|Aug\\|Sep\\|Oct\\|Nov\\|Dec\\) [ 123][0-9] [0-9][0-9]:[0-9][0-9]:[0-9][0-9] [0-9][0-9][0-9][0-9]\n$"
58 /* Global variables! */
61 int global_use_leading_time;
63 regex_t global_leading_from_line_regexp;
65 /********************************************************************/
80 static char *field_keys[] = {
92 /********************************************************************/
94 struct search_condition {
98 time_t interval_start, interval_stop;
101 /********************************************************************/
103 struct parsable_field {
110 static struct parsable_field fields_to_parse[] = {
115 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
121 "^\\(from\\|reply-to\\|sender\\|return-path\\): ",
122 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
128 "^\\(to\\|cc\\|bcc\\): ",
129 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
136 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
143 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
148 /********************************************************************/
150 int xor(int a, int b) {
151 return (a && !b) || (!a && b);
154 const char *parse_token(char *token_buffer, size_t token_buffer_size,
155 char separator, const char *string) {
156 char *u = token_buffer;
157 while(*string == separator) { string++; }
158 while(u < token_buffer + token_buffer_size - 1 && *string &&
159 *string != separator) {
160 *(u++) = *(string++);
162 while(*string == separator) { string++; }
167 char *default_value(char *current_value,
168 const char *env_variable,
169 const char *hard_default_value) {
171 return current_value;
173 char *env_value = getenv(env_variable);
175 return strdup(env_value);
176 } else if(hard_default_value) {
177 return strdup(hard_default_value);
184 FILE *safe_fopen(const char *path, const char *mode, const char *comment) {
185 FILE *result = fopen(path, mode);
190 "mymail: Cannot open file '%s' (%s) with mode \"%s\".\n",
191 path, comment, mode);
196 /*********************************************************************/
198 void print_version(FILE *out) {
199 fprintf(out, "mymail version %s (%s)\n", VERSION, UNAME);
202 void print_usage(FILE *out) {
204 fprintf(out, "Written by Francois Fleuret <francois@fleuret.org>.\n");
206 fprintf(out, "Usage: mymail [options] [<mbox dir1> [<mbox dir2> ...]|<db file1> [<db file2> ...]]\n");
208 fprintf(out, " -h, --help\n");
209 fprintf(out, " show this help\n");
210 fprintf(out, " -v, --version\n");
211 fprintf(out, " print the version number\n");
212 fprintf(out, " -q, --quiet\n");
213 fprintf(out, " do not print information during search\n");
214 fprintf(out, " -t, --use-leading-time\n");
215 fprintf(out, " use the time stamp from the leading line of each mail and not the Date:\n");
216 fprintf(out, " field\n");
217 fprintf(out, " -p <db filename pattern>, --db-pattern <db filename pattern>\n");
218 fprintf(out, " set the db filename pattern for recursive search\n");
219 fprintf(out, " -r <db root path>, --db-root <db root path>\n");
220 fprintf(out, " set the db root path for recursive search\n");
221 fprintf(out, " -l <db filename list>, --db-list <db filename list>\n");
222 fprintf(out, " set the semicolon-separated list of db files for search\n");
223 fprintf(out, " -m <mbox filename pattern>, --mbox-pattern <mbox filename pattern>\n");
224 fprintf(out, " set the mbox filename pattern for recursive search\n");
225 fprintf(out, " -s <search pattern>, --search <search pattern>\n");
226 fprintf(out, " search for matching mails in the db file\n");
227 fprintf(out, " -d <db filename>, --db-file-generate <db filename>\n");
228 fprintf(out, " set the db filename for indexing\n");
229 fprintf(out, " -i, --index\n");
230 fprintf(out, " index mails\n");
231 fprintf(out, " -o <output filename>, --output <output filename>\n");
232 fprintf(out, " set the result file, use stdout if unset\n");
233 fprintf(out, " -a <search field>, --default-search <search field>\n");
234 fprintf(out, " set the default search field\n");
237 /*********************************************************************/
239 int ignore_entry(const char *name) {
241 strcmp(name, ".") == 0 ||
242 strcmp(name, "..") == 0 ||
243 (name[0] == '.' && name[1] != '/');
246 int is_a_leading_from_line(char *mbox_line) {
248 strncmp(mbox_line, "From ", 5) == 0 &&
249 regexec(&global_leading_from_line_regexp, mbox_line, 0, 0, 0) == 0;
252 int mbox_line_match_search(struct search_condition *condition,
253 int mbox_id, const char *mbox_value) {
255 if(condition->field_id == ID_INTERVAL) {
259 if(global_use_leading_time) {
260 if(mbox_id == ID_LEADING_LINE) {
262 while(*c && *c != ' ') c++; while(*c && *c == ' ') c++;
263 strptime(c, "%a %b %e %k:%M:%S %Y", &tm);
265 return (t >= condition->interval_start &&
266 (condition->interval_stop == 0 ||
267 t <= condition->interval_stop));
272 if(mbox_id == ID_DATE) {
273 if(!strptime(mbox_value, "%a, %d %b %Y %k:%M:%S", &tm)) {
274 strptime(mbox_value, "%d %b %Y %k:%M:%S", &tm);
277 return (t >= condition->interval_start &&
278 (condition->interval_stop == 0 ||
279 t <= condition->interval_stop));
288 (condition->field_id == mbox_id)
292 (condition->field_id == ID_PARTICIPANT && (mbox_id == ID_LEADING_LINE ||
293 mbox_id == ID_FROM ||
297 (condition->field_id == ID_FROM && mbox_id == ID_LEADING_LINE)
303 regexec(&condition->regexp, mbox_value, 0, 0, 0) == 0;
307 void update_body_hits(char *mail_filename, int position_in_mail,
308 int nb_search_conditions, struct search_condition *search_conditions,
309 int nb_body_conditions,
313 char raw_mbox_line[BUFFER_SIZE];
319 mail_file = safe_fopen(mail_filename, "r", "mbox for body scan");
321 fseek(mail_file, position_in_mail, SEEK_SET);
323 if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
324 while(nb_body_hits < nb_body_conditions) {
325 /* last_mbox_line_was_empty = (raw_mbox_line[0] == '\n'); */
326 /* if(last_mbox_line_was_empty) { header = 0; } */
328 if(raw_mbox_line[0] == '\n') { header = 0; }
331 for(n = 0; n < nb_search_conditions; n++) {
332 if(search_conditions[n].field_id == ID_BODY && !hits[n]) {
334 (regexec(&search_conditions[n].regexp, raw_mbox_line, 0, 0, 0) == 0);
342 if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
343 (is_a_leading_from_line(raw_mbox_line)))
351 void extract_mail(const char *mail_filename, unsigned long int position_in_mail,
353 char raw_mbox_line[BUFFER_SIZE];
356 mail_file = safe_fopen(mail_filename, "r", "mbox for mail extraction");
357 fseek(mail_file, position_in_mail, SEEK_SET);
359 if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
360 fprintf(output_file, "%s", raw_mbox_line);
362 if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
363 (is_a_leading_from_line(raw_mbox_line))
366 fprintf(output_file, "%s", raw_mbox_line);
373 int check_full_mail_match(char *current_mail_filename,
374 int nb_search_conditions,
375 struct search_condition *search_conditions,
376 int nb_body_conditions,
378 int current_position_in_mail) {
379 int n, nb_fulfilled_body_conditions;
381 /* We first check all conditions but the body ones */
383 for(n = 0; n < nb_search_conditions &&
384 ((search_conditions[n].field_id == ID_BODY) ||
385 xor(hits[n], search_conditions[n].negation)); n++);
387 if(n == nb_search_conditions) {
389 /* Now check the body ones */
391 nb_fulfilled_body_conditions = 0;
393 if(nb_body_conditions > 0) {
394 update_body_hits(current_mail_filename, current_position_in_mail,
395 nb_search_conditions, search_conditions,
399 for(n = 0; n < nb_search_conditions; n++) {
400 if(search_conditions[n].field_id == ID_BODY &&
401 xor(hits[n], search_conditions[n].negation)) {
402 nb_fulfilled_body_conditions++;
406 return nb_body_conditions == nb_fulfilled_body_conditions;
412 int search_in_db(const char *db_filename,
413 int nb_search_conditions,
414 struct search_condition *search_conditions,
417 int hits[MAX_NB_SEARCH_CONDITIONS];
418 char raw_db_line[BUFFER_SIZE];
419 char current_mail_filename[PATH_MAX + 1];
420 unsigned long int current_position_in_mail;
421 char mbox_key[TOKEN_BUFFER_SIZE];
422 const char *mbox_value;
424 int already_written, m, n;
425 int nb_body_conditions;
427 int nb_extracted_mails;
429 nb_extracted_mails = 0;
432 printf("Searching in '%s' ... ", db_filename);
436 db_file = safe_fopen(db_filename, "r", "index file for search");
438 /* First, check the db file leading line integrity */
440 if(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
441 if(strncmp(raw_db_line, MYMAIL_DB_MAGIC_TOKEN, strlen(MYMAIL_DB_MAGIC_TOKEN))) {
443 "mymail: Header line in '%s' does not match the mymail db format.\n",
449 "mymail: Cannot read the header line in '%s'.\n",
454 /* Then parse the said db file */
456 current_position_in_mail = 0;
459 for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
461 nb_body_conditions = 0;
462 for(n = 0; n < nb_search_conditions; n++) {
463 if(search_conditions[n].field_id == ID_BODY) {
464 nb_body_conditions++;
468 strcpy(current_mail_filename, "");
470 while(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
471 mbox_value = parse_token(mbox_key, TOKEN_BUFFER_SIZE, ' ', raw_db_line);
473 if(strcmp("mail", mbox_key) == 0) {
474 char position_in_file_string[TOKEN_BUFFER_SIZE];
475 if(current_mail_filename[0] &&
476 check_full_mail_match(current_mail_filename,
477 nb_search_conditions, search_conditions,
478 nb_body_conditions, hits, current_position_in_mail)) {
479 nb_extracted_mails++;
480 extract_mail(current_mail_filename, current_position_in_mail, output_file);
482 for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
483 mbox_value = parse_token(position_in_file_string, TOKEN_BUFFER_SIZE, ' ', mbox_value);
484 mbox_value = parse_token(current_mail_filename, TOKEN_BUFFER_SIZE, '\n', mbox_value);
485 current_position_in_mail = atol(position_in_file_string);
491 for(m = 0; (m < MAX_ID) && mbox_id == -1; m++) {
492 if(strncmp(field_keys[m], mbox_key, strlen(mbox_key)) == 0) {
496 for(n = 0; n < nb_search_conditions; n++) {
497 hits[n] |= mbox_line_match_search(&search_conditions[n],
498 mbox_id, mbox_value);
503 if(current_mail_filename[0] &&
504 check_full_mail_match(current_mail_filename,
505 nb_search_conditions, search_conditions,
506 nb_body_conditions, hits, current_position_in_mail)) {
507 nb_extracted_mails++;
508 extract_mail(current_mail_filename, current_position_in_mail, output_file);
518 return nb_extracted_mails;
521 int recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp,
522 int nb_search_conditions,
523 struct search_condition *search_conditions,
526 struct dirent *dir_e;
528 char subname[PATH_MAX + 1];
529 int nb_extracted_mails = 0;
531 if(lstat(entry_name, &sb) != 0) {
533 "mymail: Cannot stat \"%s\": %s\n",
539 /* printf("recursive_search_in_db %s\n", entry_name); */
541 dir = opendir(entry_name);
544 while((dir_e = readdir(dir))) {
545 if(!ignore_entry(dir_e->d_name)) {
546 snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
547 nb_extracted_mails += recursive_search_in_db(subname, db_filename_regexp,
548 nb_search_conditions, search_conditions,
556 const char *s = entry_name, *filename = entry_name;
557 while(*s) { if(*s == '/') { filename = s+1; } s++; }
559 if(regexec(db_filename_regexp, filename, 0, 0, 0) == 0) {
560 nb_extracted_mails +=
561 search_in_db(entry_name, nb_search_conditions, search_conditions, output_file);
565 return nb_extracted_mails;
568 /*********************************************************************/
570 void index_one_mbox_line(unsigned int nb_fields_to_parse,
571 struct parsable_field *fields_to_parse,
572 char *raw_mbox_line, FILE *db_file) {
575 for(f = 0; f < nb_fields_to_parse; f++) {
576 if(regexec(&fields_to_parse[f].regexp, raw_mbox_line, 1, &matches, 0) == 0) {
577 fprintf(db_file, "%s %s\n",
578 field_keys[fields_to_parse[f].id],
579 raw_mbox_line + matches.rm_eo);
584 void index_mbox(const char *mbox_filename,
585 int nb_fields_to_parse, struct parsable_field *fields_to_parse,
587 char raw_mbox_line[BUFFER_SIZE], full_line[BUFFER_SIZE];
588 char *end_of_full_line;
590 int in_header, new_header;
591 unsigned long int position_in_file;
593 file = safe_fopen(mbox_filename, "r", "mbox for indexing");
598 position_in_file = 0;
599 end_of_full_line = 0;
602 while(fgets(raw_mbox_line, BUFFER_SIZE, file)) {
603 if(is_a_leading_from_line(raw_mbox_line)) {
604 /* This starts a new mail */
607 "Got a ^\"From \" in the header in %s:%lu.\n",
608 mbox_filename, position_in_file);
609 fprintf(stderr, "%s", raw_mbox_line);
612 /* printf("LEADING_LINE %s", raw_mbox_line); */
616 } else if(raw_mbox_line[0] == '\n') {
619 /* We leave the header, index the current line */
621 /* printf("INDEX %s\n", full_line); */
622 index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
624 end_of_full_line = full_line;
625 *end_of_full_line = '\0';
631 fprintf(db_file, "mail %lu %s\n", position_in_file, mbox_filename);
635 if(raw_mbox_line[0] == ' ' || raw_mbox_line[0] == '\t') {
636 /* Continuation of a line */
637 char *start = raw_mbox_line;
638 while(*start == ' ' || *start == '\t') start++;
639 *(end_of_full_line++) = ' ';
640 strcpy(end_of_full_line, start);
641 while(*end_of_full_line && *end_of_full_line != '\n') {
644 *end_of_full_line = '\0';
648 /* Start a new header line, not a continuation */
651 /* printf("INDEX %s\n", full_line); */
652 index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
655 end_of_full_line = full_line;
656 strcpy(end_of_full_line, raw_mbox_line);
657 while(*end_of_full_line && *end_of_full_line != '\n') {
660 *end_of_full_line = '\0';
665 position_in_file += strlen(raw_mbox_line);
671 void recursive_index_mbox(FILE *db_file,
672 const char *entry_name, regex_t *mbox_filename_regexp,
673 int nb_fields_to_parse, struct parsable_field *fields_to_parse) {
675 struct dirent *dir_e;
677 char subname[PATH_MAX + 1];
679 if(lstat(entry_name, &sb) != 0) {
681 "mymail: Cannot stat \"%s\": %s\n",
687 dir = opendir(entry_name);
690 while((dir_e = readdir(dir))) {
691 if(!ignore_entry(dir_e->d_name)) {
692 snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
693 recursive_index_mbox(db_file, subname, mbox_filename_regexp,
694 nb_fields_to_parse, fields_to_parse);
699 const char *s = entry_name, *filename = s;
700 while(*s) { if(*s == '/') { filename = s+1; }; s++; }
701 if(!mbox_filename_regexp || regexec(mbox_filename_regexp, filename, 0, 0, 0) == 0) {
702 index_mbox(entry_name, nb_fields_to_parse, fields_to_parse, db_file);
707 /*********************************************************************/
709 /* For long options that have no equivalent short option, use a
710 non-character as a pseudo short option, starting with CHAR_MAX + 1. */
712 OPT_BASH_MODE = CHAR_MAX + 1
715 static struct option long_options[] = {
716 { "help", no_argument, 0, 'h' },
717 { "version", no_argument, 0, 'v' },
718 { "quiet", no_argument, 0, 'q' },
719 { "use-leading-time", no_argument, 0, 't' },
720 { "db-file-generate", 1, 0, 'd' },
721 { "db-pattern", 1, 0, 'p' },
722 { "db-root", 1, 0, 'r' },
723 { "db-list", 1, 0, 'l' },
724 { "mbox-pattern", 1, 0, 'm' },
725 { "search", 1, 0, 's' },
726 { "index", 0, 0, 'i' },
727 { "output", 1, 0, 'o' },
728 { "default-search", 1, 0, 'a' },
732 struct time_criterion {
734 int start_hour, end_hour;
738 /*********************************************************************/
740 static struct time_criterion time_criteria[] = {
743 { "24h", 24, -1, -1 },
744 { "48h", 48, -1, -1 },
745 { "week", 24 * 7, -1, -1 },
746 { "month", 24 * 31, -1, -1 },
747 { "year", 24 * 365, -1, -1 },
749 { "yesterday", 48, 24, -1 },
751 { "today", -1, -1, 0 },
752 { "monday", -1, -1, 1 },
753 { "tuesday", -1, -1, 2 },
754 { "wednesday", -1, -1, 3 },
755 { "thursday", -1, -1, 4 },
756 { "friday", -1, -1, 5 },
757 { "saturday", -1, -1, 6 },
758 { "sunday", -1, -1, 7 },
762 /*********************************************************************/
764 time_t time_for_past_day(int day) {
771 delta_day = (7 + tm->tm_wday - day) % 7;
775 return t - (delta_day * 3600 * 24 + tm->tm_sec + 60 * tm->tm_min + 3600 * tm->tm_hour);
778 void init_condition(struct search_condition *condition, const char *full_string,
779 const char *default_search_field) {
780 char full_search_field[TOKEN_BUFFER_SIZE], *search_field;
784 string = parse_token(full_search_field, TOKEN_BUFFER_SIZE, ' ', full_string);
785 search_field = full_search_field;
787 if(search_field[0] == '!') {
789 condition->negation = 1;
791 condition->negation = 0;
794 condition->field_id = -1;
798 for(k = 0; k < sizeof(time_criteria) / sizeof(struct time_criterion); k++) {
799 if(strcmp(time_criteria[k].label, search_field) == 0) {
800 condition->field_id = ID_INTERVAL;
801 if(time_criteria[k].past_week_day < 0) {
802 condition->interval_start = time(0) - 3600 * time_criteria[k].start_hour;
803 if(time_criteria[k].end_hour >= 0) {
804 condition->interval_stop = time(0) - 3600 * time_criteria[k].end_hour;
806 condition->interval_stop = 0;
809 condition->interval_start = time_for_past_day(time_criteria[k].past_week_day);
810 condition->interval_stop = condition->interval_start + 3600 * 24;
816 if(condition->field_id == -1) {
818 /* No time condition matched, look for the search fields */
820 for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) {
821 if(strncmp(field_keys[m], search_field, strlen(search_field)) == 0) {
822 condition->field_id = m;
826 /* None match, if there is a default search field, re-run the search with it */
828 if(condition->field_id == -1) {
829 if(default_search_field) {
830 for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) {
831 if(strncmp(field_keys[m],
832 default_search_field, strlen(default_search_field)) == 0) {
833 condition->field_id = m;
836 string = full_string;
837 if(string[0] == '!') { string++; }
841 if(condition->field_id == -1) {
843 "mymail: Syntax error in field key \"%s\".\n",
848 if(regcomp(&condition->regexp,
852 "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
854 field_keys[condition->field_id]);
860 void free_condition(struct search_condition *condition) {
861 if(condition->field_id != ID_INTERVAL) {
862 regfree(&condition->regexp);
866 /*********************************************************************/
867 /*********************************************************************/
868 /*********************************************************************/
870 int main(int argc, char **argv) {
871 char *db_filename = 0;
872 char *db_filename_regexp_string = 0;
873 char *db_root_path = 0;
874 char *db_filename_list = 0;
875 char *mbox_filename_regexp_string = 0;
876 char *default_search_field;
877 char output_filename[PATH_MAX + 1];
878 int action_index = 0;
879 int error = 0, show_help = 0;
880 const unsigned int nb_fields_to_parse =
881 sizeof(fields_to_parse) / sizeof(struct parsable_field);
884 unsigned int nb_search_conditions;
885 struct search_condition search_conditions[MAX_NB_SEARCH_CONDITIONS];
887 if(regcomp(&global_leading_from_line_regexp, LEADING_FROM_LINE_REGEXP_STRING, 0)) {
889 "mymail: Cannot compile leading \"from\" line regexp. That is strange.\n");
894 global_use_leading_time = 0;
895 default_search_field = 0;
896 strncpy(output_filename, "", PATH_MAX);
898 setlocale(LC_ALL, "");
900 nb_search_conditions = 0;
902 while ((c = getopt_long(argc, argv, "hvqip:s:d:r:l:o:a:m:",
903 long_options, NULL)) != -1) {
912 print_version(stdout);
920 global_use_leading_time = 1;
929 fprintf(stderr, "mymail: Can not set the db filename twice.\n");
932 db_filename = strdup(optarg);
936 if(db_filename_regexp_string) {
937 fprintf(stderr, "mymail: Can not set the db filename pattern twice.\n");
940 db_filename_regexp_string = strdup(optarg);
944 if(mbox_filename_regexp_string) {
945 fprintf(stderr, "mymail: Can not set the mbox filename pattern twice.\n");
948 mbox_filename_regexp_string = strdup(optarg);
952 strncpy(output_filename, optarg, PATH_MAX);
957 fprintf(stderr, "mymail: Can not set the db root path twice.\n");
960 db_root_path = strdup(optarg);
964 if(db_filename_list) {
965 fprintf(stderr, "mymail: Can not set the db filename list twice.\n");
968 db_filename_list = strdup(optarg);
972 if(nb_search_conditions == MAX_NB_SEARCH_CONDITIONS) {
973 fprintf(stderr, "mymail: Too many search patterns.\n");
976 init_condition(&search_conditions[nb_search_conditions], optarg, default_search_field);
977 nb_search_conditions++;
981 default_search_field = optarg;
990 /* Set all the values that may defined in the arguments, through
991 environment variables, or hard-coded */
993 db_filename = default_value(db_filename,
997 db_filename_regexp_string = default_value(db_filename_regexp_string,
1001 db_root_path = default_value(db_root_path,
1005 db_filename_list = default_value(db_filename_list,
1009 mbox_filename_regexp_string = default_value(mbox_filename_regexp_string,
1010 "MYMAIL_MBOX_PATTERN",
1013 /* Start the processing */
1016 print_usage(stderr);
1021 print_usage(stdout);
1029 regex_t mbox_filename_regexp_static;
1030 regex_t *mbox_filename_regexp;
1032 if(mbox_filename_regexp_string) {
1033 if(regcomp(&mbox_filename_regexp_static,
1034 mbox_filename_regexp_string,
1037 "mymail: Syntax error in regexp \"%s\".\n",
1038 mbox_filename_regexp_string);
1041 mbox_filename_regexp = &mbox_filename_regexp_static;
1043 mbox_filename_regexp = 0;
1046 db_file = safe_fopen(db_filename, "w", "index file for indexing");
1048 for(f = 0; f < nb_fields_to_parse; f++) {
1049 if(regcomp(&fields_to_parse[f].regexp,
1050 fields_to_parse[f].regexp_string,
1051 fields_to_parse[f].cflags)) {
1053 "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
1054 fields_to_parse[f].regexp_string,
1055 field_keys[fields_to_parse[f].id]);
1060 fprintf(db_file, "%s version_%s raw\n", MYMAIL_DB_MAGIC_TOKEN, VERSION);
1062 while(optind < argc) {
1063 recursive_index_mbox(db_file,
1064 argv[optind], mbox_filename_regexp,
1065 nb_fields_to_parse, fields_to_parse);
1072 if(mbox_filename_regexp) {
1073 regfree(mbox_filename_regexp);
1076 for(f = 0; f < nb_fields_to_parse; f++) {
1077 regfree(&fields_to_parse[f].regexp);
1086 int nb_extracted_mails = 0;
1088 if(output_filename[0]) {
1089 output_file = safe_fopen(output_filename, "w", "result mbox");
1091 output_file = stdout;
1095 if(nb_search_conditions > 0) {
1097 /* Recursive search if db_root_path is set */
1100 regex_t db_filename_regexp;
1101 if(regcomp(&db_filename_regexp,
1102 db_filename_regexp_string,
1105 "mymail: Syntax error in regexp \"%s\".\n",
1106 db_filename_regexp_string);
1110 nb_extracted_mails += recursive_search_in_db(db_root_path, &db_filename_regexp,
1111 nb_search_conditions, search_conditions,
1114 regfree(&db_filename_regexp);
1117 /* Search in all db files listed in db_filename_list */
1119 if(db_filename_list) {
1120 char db_filename[PATH_MAX + 1];
1123 s = db_filename_list;
1126 s = parse_token(db_filename, PATH_MAX + 1, ';', s);
1128 if(db_filename[0]) {
1129 nb_extracted_mails +=
1130 search_in_db(db_filename, nb_search_conditions, search_conditions, output_file);
1135 /* Search in all db files listed in the command arguments */
1137 while(optind < argc) {
1138 nb_extracted_mails +=
1139 search_in_db(argv[optind], nb_search_conditions, search_conditions, output_file);
1145 if(nb_extracted_mails > 0) {
1146 printf("Found %d matching mails.\n", nb_extracted_mails);
1148 printf("No matching mail found.\n");
1152 fflush(output_file);
1154 if(output_file != stdout) {
1155 fclose(output_file);
1159 for(n = 0; n < nb_search_conditions; n++) {
1160 free_condition(&search_conditions[n]);
1164 free(db_filename_regexp_string);
1166 free(db_filename_list);
1167 free(mbox_filename_regexp_string);
1169 regfree(&global_leading_from_line_regexp);