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 "0.9.6"
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) {
257 (condition->field_id == mbox_id)
261 (condition->field_id == ID_PARTICIPANT && (mbox_id == ID_LEADING_LINE ||
262 mbox_id == ID_FROM ||
266 (condition->field_id == ID_FROM && mbox_id == ID_LEADING_LINE)
272 regexec(&condition->regexp, mbox_value, 0, 0, 0) == 0;
275 void update_body_hits(char *mail_filename, int position_in_mail,
276 int nb_search_conditions, struct search_condition *search_conditions,
277 int nb_body_conditions,
281 char raw_mbox_line[BUFFER_SIZE];
287 mail_file = safe_fopen(mail_filename, "r", "mbox for body scan");
289 fseek(mail_file, position_in_mail, SEEK_SET);
291 if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
292 while(nb_body_hits < nb_body_conditions) {
293 /* last_mbox_line_was_empty = (raw_mbox_line[0] == '\n'); */
294 /* if(last_mbox_line_was_empty) { header = 0; } */
296 if(raw_mbox_line[0] == '\n') { header = 0; }
299 for(n = 0; n < nb_search_conditions; n++) {
300 if(search_conditions[n].field_id == ID_BODY && !hits[n]) {
302 (regexec(&search_conditions[n].regexp, raw_mbox_line, 0, 0, 0) == 0);
310 if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
311 (is_a_leading_from_line(raw_mbox_line)))
319 void extract_mail(const char *mail_filename, unsigned long int position_in_mail,
321 char raw_mbox_line[BUFFER_SIZE];
324 mail_file = safe_fopen(mail_filename, "r", "mbox for mail extraction");
325 fseek(mail_file, position_in_mail, SEEK_SET);
327 if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
328 fprintf(output_file, "%s", raw_mbox_line);
330 if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
331 (is_a_leading_from_line(raw_mbox_line))
334 fprintf(output_file, "%s", raw_mbox_line);
341 int check_full_mail_match(char *current_mail_filename,
343 int nb_search_conditions,
344 struct search_condition *search_conditions,
345 int nb_body_conditions,
347 int current_position_in_mail) {
348 int n, nb_fulfilled_body_conditions;
350 for(n = 0; n < nb_search_conditions; n++) {
351 if(search_conditions[n].field_id == ID_TIME_INTERVAL) {
352 hits[n] = (mail_time >= search_conditions[n].interval_start &&
353 (search_conditions[n].interval_stop == 0 ||
354 mail_time <= search_conditions[n].interval_stop));
358 /* We first check all conditions but the body ones */
360 for(n = 0; n < nb_search_conditions &&
361 ((search_conditions[n].field_id == ID_BODY) ||
362 xor(hits[n], search_conditions[n].negation)); n++);
364 if(n == nb_search_conditions) {
366 /* Now check the body ones */
368 nb_fulfilled_body_conditions = 0;
370 if(nb_body_conditions > 0) {
371 update_body_hits(current_mail_filename, current_position_in_mail,
372 nb_search_conditions, search_conditions,
376 for(n = 0; n < nb_search_conditions; n++) {
377 if(search_conditions[n].field_id == ID_BODY &&
378 xor(hits[n], search_conditions[n].negation)) {
379 nb_fulfilled_body_conditions++;
383 return nb_body_conditions == nb_fulfilled_body_conditions;
389 /* We use the mail leading line time by default, and if we should and
390 can, we update with the Date: field */
392 void update_time(int mbox_id, const char *mbox_value, time_t *t) {
396 if(mbox_id == ID_LEADING_LINE) {
398 while(*c && *c != ' ') c++; while(*c && *c == ' ') c++;
399 strptime(c, "%a %b %e %k:%M:%S %Y", &tm);
402 if(!global_use_leading_time) {
403 if(mbox_id == ID_DATE) {
404 if(strptime(mbox_value, "%a, %d %b %Y %k:%M:%S", &tm) ||
405 strptime(mbox_value, "%d %b %Y %k:%M:%S", &tm)) {
413 int search_in_db(const char *db_filename,
414 int nb_search_conditions,
415 struct search_condition *search_conditions,
418 int hits[MAX_NB_SEARCH_CONDITIONS];
419 char raw_db_line[BUFFER_SIZE];
420 char current_mail_filename[PATH_MAX + 1];
421 unsigned long int current_position_in_mail;
422 char mbox_key[TOKEN_BUFFER_SIZE];
423 char position_in_file_string[TOKEN_BUFFER_SIZE];
424 const char *mbox_value;
426 int already_written, m, n;
427 int nb_body_conditions, need_time;
429 int nb_extracted_mails;
432 nb_extracted_mails = 0;
435 printf("Searching in '%s' ... ", db_filename);
439 db_file = safe_fopen(db_filename, "r", "index file for search");
441 /* First, check the db file leading line integrity */
443 if(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
444 if(strncmp(raw_db_line, MYMAIL_DB_MAGIC_TOKEN, strlen(MYMAIL_DB_MAGIC_TOKEN))) {
446 "mymail: Header line in '%s' does not match the mymail db format.\n",
452 "mymail: Cannot read the header line in '%s'.\n",
457 /* Then parse the said db file */
459 current_position_in_mail = 0;
462 for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
464 nb_body_conditions = 0;
468 for(n = 0; n < nb_search_conditions; n++) {
469 if(search_conditions[n].field_id == ID_BODY) {
470 nb_body_conditions++;
472 else if(search_conditions[n].field_id == ID_TIME_INTERVAL) {
477 strcpy(current_mail_filename, "");
479 while(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
480 mbox_value = parse_token(mbox_key, TOKEN_BUFFER_SIZE, ' ', raw_db_line);
482 if(strcmp("mail", mbox_key) == 0) {
483 if(current_mail_filename[0]) {
484 if(check_full_mail_match(current_mail_filename,
486 nb_search_conditions, search_conditions,
487 nb_body_conditions, hits, current_position_in_mail)) {
488 extract_mail(current_mail_filename, current_position_in_mail, output_file);
489 nb_extracted_mails++;
493 for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
494 mbox_value = parse_token(position_in_file_string, TOKEN_BUFFER_SIZE, ' ', mbox_value);
495 mbox_value = parse_token(current_mail_filename, PATH_MAX+1, '\n', mbox_value);
496 current_position_in_mail = atol(position_in_file_string);
502 for(m = 0; (m < MAX_ID) && mbox_id == -1; m++) {
503 if(strncmp(field_keys[m], mbox_key, strlen(mbox_key)) == 0) {
508 for(n = 0; n < nb_search_conditions; n++) {
509 hits[n] |= mbox_line_match_search(&search_conditions[n],
510 mbox_id, mbox_value);
514 update_time(mbox_id, mbox_value, &mail_time);
519 if(current_mail_filename[0]) {
520 if(check_full_mail_match(current_mail_filename,
522 nb_search_conditions, search_conditions,
523 nb_body_conditions, hits, current_position_in_mail)) {
524 extract_mail(current_mail_filename, current_position_in_mail, output_file);
525 nb_extracted_mails++;
536 return nb_extracted_mails;
539 int recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp,
540 int nb_search_conditions,
541 struct search_condition *search_conditions,
544 struct dirent *dir_e;
546 char subname[PATH_MAX + 1];
547 int nb_extracted_mails = 0;
549 if(lstat(entry_name, &sb) != 0) {
551 "mymail: Cannot stat \"%s\": %s\n",
557 /* printf("recursive_search_in_db %s\n", entry_name); */
559 dir = opendir(entry_name);
562 while((dir_e = readdir(dir))) {
563 if(!ignore_entry(dir_e->d_name)) {
564 snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
565 nb_extracted_mails += recursive_search_in_db(subname, db_filename_regexp,
566 nb_search_conditions, search_conditions,
574 const char *s = entry_name, *filename = entry_name;
575 while(*s) { if(*s == '/') { filename = s+1; } s++; }
577 if(regexec(db_filename_regexp, filename, 0, 0, 0) == 0) {
578 nb_extracted_mails +=
579 search_in_db(entry_name, nb_search_conditions, search_conditions, output_file);
583 return nb_extracted_mails;
586 /*********************************************************************/
588 void index_one_mbox_line(unsigned int nb_fields_to_parse,
589 struct parsable_field *fields_to_parse,
590 char *raw_mbox_line, FILE *db_file) {
593 for(f = 0; f < nb_fields_to_parse; f++) {
594 if(regexec(&fields_to_parse[f].regexp, raw_mbox_line, 1, &matches, 0) == 0) {
595 fprintf(db_file, "%s %s\n",
596 field_keys[fields_to_parse[f].id],
597 raw_mbox_line + matches.rm_eo);
602 void index_mbox(const char *mbox_filename,
603 int nb_fields_to_parse, struct parsable_field *fields_to_parse,
605 char raw_mbox_line[BUFFER_SIZE], full_line[BUFFER_SIZE];
606 char *end_of_full_line;
608 int in_header, new_header;
609 unsigned long int position_in_file;
611 file = safe_fopen(mbox_filename, "r", "mbox for indexing");
616 position_in_file = 0;
617 end_of_full_line = 0;
620 while(fgets(raw_mbox_line, BUFFER_SIZE, file)) {
621 if(is_a_leading_from_line(raw_mbox_line)) {
622 /* This starts a new mail */
625 "Got a ^\"From \" in the header in %s:%lu.\n",
626 mbox_filename, position_in_file);
627 fprintf(stderr, "%s", raw_mbox_line);
630 /* printf("LEADING_LINE %s", raw_mbox_line); */
634 } else if(raw_mbox_line[0] == '\n') {
637 /* We leave the header, index the current line */
639 /* printf("INDEX %s\n", full_line); */
640 index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
642 end_of_full_line = full_line;
643 *end_of_full_line = '\0';
649 fprintf(db_file, "mail %lu %s\n", position_in_file, mbox_filename);
653 if(raw_mbox_line[0] == ' ' || raw_mbox_line[0] == '\t') {
654 /* Continuation of a line */
655 char *start = raw_mbox_line;
656 while(*start == ' ' || *start == '\t') start++;
657 *(end_of_full_line++) = ' ';
658 strcpy(end_of_full_line, start);
659 while(*end_of_full_line && *end_of_full_line != '\n') {
662 *end_of_full_line = '\0';
666 /* Start a new header line, not a continuation */
669 /* printf("INDEX %s\n", full_line); */
670 index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
673 end_of_full_line = full_line;
674 strcpy(end_of_full_line, raw_mbox_line);
675 while(*end_of_full_line && *end_of_full_line != '\n') {
678 *end_of_full_line = '\0';
683 position_in_file += strlen(raw_mbox_line);
689 void recursive_index_mbox(FILE *db_file,
690 const char *entry_name, regex_t *mbox_filename_regexp,
691 int nb_fields_to_parse, struct parsable_field *fields_to_parse) {
693 struct dirent *dir_e;
695 char subname[PATH_MAX + 1];
697 if(lstat(entry_name, &sb) != 0) {
699 "mymail: Cannot stat \"%s\": %s\n",
705 dir = opendir(entry_name);
708 while((dir_e = readdir(dir))) {
709 if(!ignore_entry(dir_e->d_name)) {
710 snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
711 recursive_index_mbox(db_file, subname, mbox_filename_regexp,
712 nb_fields_to_parse, fields_to_parse);
717 const char *s = entry_name, *filename = s;
718 while(*s) { if(*s == '/') { filename = s+1; }; s++; }
719 if(!mbox_filename_regexp || regexec(mbox_filename_regexp, filename, 0, 0, 0) == 0) {
720 index_mbox(entry_name, nb_fields_to_parse, fields_to_parse, db_file);
725 /*********************************************************************/
727 /* For long options that have no equivalent short option, use a
728 non-character as a pseudo short option, starting with CHAR_MAX + 1. */
730 OPT_BASH_MODE = CHAR_MAX + 1
733 static struct option long_options[] = {
734 { "help", no_argument, 0, 'h' },
735 { "version", no_argument, 0, 'v' },
736 { "quiet", no_argument, 0, 'q' },
737 { "use-leading-time", no_argument, 0, 't' },
738 { "db-file-generate", 1, 0, 'd' },
739 { "db-pattern", 1, 0, 'p' },
740 { "db-root", 1, 0, 'r' },
741 { "db-list", 1, 0, 'l' },
742 { "mbox-pattern", 1, 0, 'm' },
743 { "search", 1, 0, 's' },
744 { "index", 0, 0, 'i' },
745 { "output", 1, 0, 'o' },
746 { "default-search", 1, 0, 'a' },
750 struct time_criterion {
753 int start_hour, end_hour;
757 /*********************************************************************/
759 static struct time_criterion time_criteria[] = {
761 { "8h", 0, 8, -1, -1 },
762 { "24h", 0, 24, -1, -1 },
763 { "48h", 0, 48, -1, -1 },
764 { "week", 0, 24 * 7, -1, -1 },
765 { "month", 0, 24 * 31, -1, -1 },
766 { "year", 0, 24 * 365, -1, -1 },
768 { "yesterday", 1, -1, -1, -1 },
769 { "today", 1, -1, -1, 0 },
771 { "monday", 1, -1, -1, 1 },
772 { "tuesday", 1, -1, -1, 2 },
773 { "wednesday", 1, -1, -1, 3 },
774 { "thursday", 1, -1, -1, 4 },
775 { "friday", 1, -1, -1, 5 },
776 { "saturday", 1, -1, -1, 6 },
777 { "sunday", 1, -1, -1, 7 },
781 /*********************************************************************/
783 time_t time_for_past_day(int day) {
790 delta_day = (7 + tm->tm_wday - day) % 7;
794 return t - (delta_day * 3600 * 24 + tm->tm_sec + 60 * tm->tm_min + 3600 * tm->tm_hour);
797 void init_condition(struct search_condition *condition, const char *full_string,
798 const char *default_search_field) {
799 char full_search_field[TOKEN_BUFFER_SIZE], *search_field;
803 string = parse_token(full_search_field, TOKEN_BUFFER_SIZE, ' ', full_string);
804 search_field = full_search_field;
806 if(search_field[0] == '!') {
808 condition->negation = 1;
810 condition->negation = 0;
813 condition->field_id = -1;
817 for(k = 0; k < sizeof(time_criteria) / sizeof(struct time_criterion); k++) {
818 if(strcmp(time_criteria[k].label, search_field) == 0) {
819 condition->field_id = ID_TIME_INTERVAL;
820 if(time_criteria[k].day_criterion) {
821 condition->interval_start = time_for_past_day(time_criteria[k].past_week_day);
822 condition->interval_stop = condition->interval_start + 3600 * 24;
824 condition->interval_start = time(0) - 3600 * time_criteria[k].start_hour;
825 if(time_criteria[k].end_hour >= 0) {
826 condition->interval_stop = time(0) - 3600 * time_criteria[k].end_hour;
828 condition->interval_stop = 0;
835 if(condition->field_id == -1) {
837 /* No time condition matched, look for the search fields */
839 for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) {
840 if(strncmp(field_keys[m], search_field, strlen(search_field)) == 0) {
841 condition->field_id = m;
845 /* None match, if there is a default search field, re-run the search with it */
847 if(condition->field_id == -1) {
848 if(default_search_field) {
849 for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) {
850 if(strncmp(field_keys[m],
851 default_search_field, strlen(default_search_field)) == 0) {
852 condition->field_id = m;
855 string = full_string;
856 if(string[0] == '!') { string++; }
860 if(condition->field_id == -1) {
862 "mymail: Syntax error in field key \"%s\".\n",
867 if(regcomp(&condition->regexp,
871 "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
873 field_keys[condition->field_id]);
879 void free_condition(struct search_condition *condition) {
880 if(condition->field_id != ID_TIME_INTERVAL) {
881 regfree(&condition->regexp);
885 /*********************************************************************/
886 /*********************************************************************/
887 /*********************************************************************/
889 int main(int argc, char **argv) {
890 char *db_filename = 0;
891 char *db_filename_regexp_string = 0;
892 char *db_root_path = 0;
893 char *db_filename_list = 0;
894 char *mbox_filename_regexp_string = 0;
895 char *default_search_field;
896 char output_filename[PATH_MAX + 1];
897 int action_index = 0;
898 int error = 0, show_help = 0;
899 const unsigned int nb_fields_to_parse =
900 sizeof(fields_to_parse) / sizeof(struct parsable_field);
903 unsigned int nb_search_conditions;
904 struct search_condition search_conditions[MAX_NB_SEARCH_CONDITIONS];
906 if(regcomp(&global_leading_from_line_regexp, LEADING_FROM_LINE_REGEXP_STRING, 0)) {
908 "mymail: Cannot compile leading \"from\" line regexp. That is strange.\n");
913 global_use_leading_time = 0;
914 default_search_field = 0;
915 strncpy(output_filename, "", PATH_MAX);
917 setlocale(LC_ALL, "");
919 nb_search_conditions = 0;
921 while ((c = getopt_long(argc, argv, "hvqip:s:d:r:l:o:a:m:",
922 long_options, NULL)) != -1) {
931 print_version(stdout);
939 global_use_leading_time = 1;
948 fprintf(stderr, "mymail: Can not set the db filename twice.\n");
951 db_filename = strdup(optarg);
955 if(db_filename_regexp_string) {
956 fprintf(stderr, "mymail: Can not set the db filename pattern twice.\n");
959 db_filename_regexp_string = strdup(optarg);
963 if(mbox_filename_regexp_string) {
964 fprintf(stderr, "mymail: Can not set the mbox filename pattern twice.\n");
967 mbox_filename_regexp_string = strdup(optarg);
971 strncpy(output_filename, optarg, PATH_MAX);
976 fprintf(stderr, "mymail: Can not set the db root path twice.\n");
979 db_root_path = strdup(optarg);
983 if(db_filename_list) {
984 fprintf(stderr, "mymail: Can not set the db filename list twice.\n");
987 db_filename_list = strdup(optarg);
991 if(nb_search_conditions == MAX_NB_SEARCH_CONDITIONS) {
992 fprintf(stderr, "mymail: Too many search patterns.\n");
995 init_condition(&search_conditions[nb_search_conditions], optarg, default_search_field);
996 nb_search_conditions++;
1000 default_search_field = optarg;
1009 /* Set all the values that may defined in the arguments, through
1010 environment variables, or hard-coded */
1012 db_filename = default_value(db_filename,
1016 db_filename_regexp_string = default_value(db_filename_regexp_string,
1020 db_root_path = default_value(db_root_path,
1024 db_filename_list = default_value(db_filename_list,
1028 mbox_filename_regexp_string = default_value(mbox_filename_regexp_string,
1029 "MYMAIL_MBOX_PATTERN",
1032 /* Start the processing */
1035 print_usage(stderr);
1040 print_usage(stdout);
1048 regex_t mbox_filename_regexp_static;
1049 regex_t *mbox_filename_regexp;
1051 if(mbox_filename_regexp_string) {
1052 if(regcomp(&mbox_filename_regexp_static,
1053 mbox_filename_regexp_string,
1056 "mymail: Syntax error in regexp \"%s\".\n",
1057 mbox_filename_regexp_string);
1060 mbox_filename_regexp = &mbox_filename_regexp_static;
1062 mbox_filename_regexp = 0;
1065 db_file = safe_fopen(db_filename, "w", "index file for indexing");
1067 for(f = 0; f < nb_fields_to_parse; f++) {
1068 if(regcomp(&fields_to_parse[f].regexp,
1069 fields_to_parse[f].regexp_string,
1070 fields_to_parse[f].cflags)) {
1072 "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
1073 fields_to_parse[f].regexp_string,
1074 field_keys[fields_to_parse[f].id]);
1079 fprintf(db_file, "%s version_%s raw\n", MYMAIL_DB_MAGIC_TOKEN, VERSION);
1081 while(optind < argc) {
1082 recursive_index_mbox(db_file,
1083 argv[optind], mbox_filename_regexp,
1084 nb_fields_to_parse, fields_to_parse);
1091 if(mbox_filename_regexp) {
1092 regfree(mbox_filename_regexp);
1095 for(f = 0; f < nb_fields_to_parse; f++) {
1096 regfree(&fields_to_parse[f].regexp);
1105 int nb_extracted_mails = 0;
1107 if(output_filename[0]) {
1108 output_file = safe_fopen(output_filename, "w", "result mbox");
1110 output_file = stdout;
1114 if(nb_search_conditions > 0) {
1116 /* Recursive search if db_root_path is set */
1119 regex_t db_filename_regexp;
1120 if(regcomp(&db_filename_regexp,
1121 db_filename_regexp_string,
1124 "mymail: Syntax error in regexp \"%s\".\n",
1125 db_filename_regexp_string);
1129 nb_extracted_mails += recursive_search_in_db(db_root_path, &db_filename_regexp,
1130 nb_search_conditions, search_conditions,
1133 regfree(&db_filename_regexp);
1136 /* Search in all db files listed in db_filename_list */
1138 if(db_filename_list) {
1139 char db_filename[PATH_MAX + 1];
1142 s = db_filename_list;
1145 s = parse_token(db_filename, PATH_MAX + 1, ';', s);
1147 if(db_filename[0]) {
1148 nb_extracted_mails +=
1149 search_in_db(db_filename, nb_search_conditions, search_conditions, output_file);
1154 /* Search in all db files listed in the command arguments */
1156 while(optind < argc) {
1157 nb_extracted_mails +=
1158 search_in_db(argv[optind], nb_search_conditions, search_conditions, output_file);
1164 if(nb_extracted_mails > 0) {
1165 printf("Found %d matching mails.\n", nb_extracted_mails);
1167 printf("No matching mail found.\n");
1171 fflush(output_file);
1173 if(output_file != stdout) {
1174 fclose(output_file);
1178 for(n = 0; n < nb_search_conditions; n++) {
1179 free_condition(&search_conditions[n]);
1183 free(db_filename_regexp_string);
1185 free(db_filename_list);
1186 free(mbox_filename_regexp_string);
1188 regfree(&global_leading_from_line_regexp);