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 strptime(mbox_value, "%a, %d %b %Y %k:%M:%S", &tm);
275 return (t >= condition->interval_start &&
276 (condition->interval_stop == 0 ||
277 t <= condition->interval_stop));
286 (condition->field_id == mbox_id)
290 (condition->field_id == ID_PARTICIPANT && (mbox_id == ID_LEADING_LINE ||
291 mbox_id == ID_FROM ||
295 (condition->field_id == ID_FROM && mbox_id == ID_LEADING_LINE)
301 regexec(&condition->regexp, mbox_value, 0, 0, 0) == 0;
305 void update_body_hits(char *mail_filename, int position_in_mail,
306 int nb_search_conditions, struct search_condition *search_conditions,
307 int nb_body_conditions,
311 char raw_mbox_line[BUFFER_SIZE];
317 mail_file = safe_fopen(mail_filename, "r", "mbox for body scan");
319 fseek(mail_file, position_in_mail, SEEK_SET);
321 if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
322 while(nb_body_hits < nb_body_conditions) {
323 /* last_mbox_line_was_empty = (raw_mbox_line[0] == '\n'); */
324 /* if(last_mbox_line_was_empty) { header = 0; } */
326 if(raw_mbox_line[0] == '\n') { header = 0; }
329 for(n = 0; n < nb_search_conditions; n++) {
330 if(search_conditions[n].field_id == ID_BODY && !hits[n]) {
332 (regexec(&search_conditions[n].regexp, raw_mbox_line, 0, 0, 0) == 0);
340 if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
341 (is_a_leading_from_line(raw_mbox_line)))
349 void extract_mail(const char *mail_filename, unsigned long int position_in_mail,
351 char raw_mbox_line[BUFFER_SIZE];
354 mail_file = safe_fopen(mail_filename, "r", "mbox for mail extraction");
355 fseek(mail_file, position_in_mail, SEEK_SET);
357 if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
358 fprintf(output_file, "%s", raw_mbox_line);
360 if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
361 (is_a_leading_from_line(raw_mbox_line))
364 fprintf(output_file, "%s", raw_mbox_line);
371 int search_in_db(const char *db_filename,
372 int nb_search_conditions,
373 struct search_condition *search_conditions,
376 int hits[MAX_NB_SEARCH_CONDITIONS];
377 char raw_db_line[BUFFER_SIZE];
378 char current_mail_filename[PATH_MAX + 1];
379 unsigned long int current_position_in_mail;
380 char mbox_key[TOKEN_BUFFER_SIZE];
381 const char *mbox_value;
383 int already_written, m, n;
384 int nb_body_conditions, nb_fulfilled_body_conditions;
386 int nb_extracted_mails;
388 nb_extracted_mails = 0;
391 printf("Searching in '%s' ... ", db_filename);
395 db_file = safe_fopen(db_filename, "r", "index file for search");
397 /* First, check the db file leading line integrity */
399 if(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
400 if(strncmp(raw_db_line, MYMAIL_DB_MAGIC_TOKEN, strlen(MYMAIL_DB_MAGIC_TOKEN))) {
402 "mymail: Header line in '%s' does not match the mymail db format.\n",
408 "mymail: Cannot read the header line in '%s'.\n",
413 /* Then parse the said db file */
415 current_position_in_mail = 0;
418 for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
420 nb_body_conditions = 0;
421 for(n = 0; n < nb_search_conditions; n++) {
422 if(search_conditions[n].field_id == ID_BODY) {
423 nb_body_conditions++;
427 strcpy(current_mail_filename, "");
429 while(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
430 mbox_value = parse_token(mbox_key, TOKEN_BUFFER_SIZE, ' ', raw_db_line);
432 if(strcmp("mail", mbox_key) == 0) {
433 char position_in_file_string[TOKEN_BUFFER_SIZE];
435 if(current_mail_filename[0]) {
437 /* We first check all conditions but the body ones */
439 for(n = 0; n < nb_search_conditions &&
440 ((search_conditions[n].field_id == ID_BODY) ||
441 xor(hits[n], search_conditions[n].negation)); n++);
443 if(n == nb_search_conditions) {
445 /* Now check the body ones */
447 nb_fulfilled_body_conditions = 0;
449 if(nb_body_conditions > 0) {
450 update_body_hits(current_mail_filename, current_position_in_mail,
451 nb_search_conditions, search_conditions,
455 for(n = 0; n < nb_search_conditions; n++) {
456 if(search_conditions[n].field_id == ID_BODY &&
457 xor(hits[n], search_conditions[n].negation)) {
458 nb_fulfilled_body_conditions++;
463 if(nb_body_conditions == nb_fulfilled_body_conditions) {
464 nb_extracted_mails++;
465 extract_mail(current_mail_filename, current_position_in_mail, output_file);
470 for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
472 mbox_value = parse_token(position_in_file_string, TOKEN_BUFFER_SIZE, ' ', mbox_value);
473 mbox_value = parse_token(current_mail_filename, TOKEN_BUFFER_SIZE, '\n', mbox_value);
474 current_position_in_mail = atol(position_in_file_string);
480 for(m = 0; (m < MAX_ID) && mbox_id == -1; m++) {
481 if(strncmp(field_keys[m], mbox_key, strlen(mbox_key)) == 0) {
485 for(n = 0; n < nb_search_conditions; n++) {
486 hits[n] |= mbox_line_match_search(&search_conditions[n],
487 mbox_id, mbox_value);
499 return nb_extracted_mails;
502 int recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp,
503 int nb_search_conditions,
504 struct search_condition *search_conditions,
507 struct dirent *dir_e;
509 char subname[PATH_MAX + 1];
510 int nb_extracted_mails = 0;
512 if(lstat(entry_name, &sb) != 0) {
514 "mymail: Cannot stat \"%s\": %s\n",
520 /* printf("recursive_search_in_db %s\n", entry_name); */
522 dir = opendir(entry_name);
525 while((dir_e = readdir(dir))) {
526 if(!ignore_entry(dir_e->d_name)) {
527 snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
528 nb_extracted_mails += recursive_search_in_db(subname, db_filename_regexp,
529 nb_search_conditions, search_conditions,
537 const char *s = entry_name, *filename = entry_name;
538 while(*s) { if(*s == '/') { filename = s+1; } s++; }
540 if(regexec(db_filename_regexp, filename, 0, 0, 0) == 0) {
541 nb_extracted_mails +=
542 search_in_db(entry_name, nb_search_conditions, search_conditions, output_file);
546 return nb_extracted_mails;
549 /*********************************************************************/
551 void index_one_mbox_line(unsigned int nb_fields_to_parse,
552 struct parsable_field *fields_to_parse,
553 char *raw_mbox_line, FILE *db_file) {
556 for(f = 0; f < nb_fields_to_parse; f++) {
557 if(regexec(&fields_to_parse[f].regexp, raw_mbox_line, 1, &matches, 0) == 0) {
558 fprintf(db_file, "%s %s\n",
559 field_keys[fields_to_parse[f].id],
560 raw_mbox_line + matches.rm_eo);
565 void index_mbox(const char *mbox_filename,
566 int nb_fields_to_parse, struct parsable_field *fields_to_parse,
568 char raw_mbox_line[BUFFER_SIZE], full_line[BUFFER_SIZE];
569 char *end_of_full_line;
571 int in_header, new_header;
572 unsigned long int position_in_file;
574 file = safe_fopen(mbox_filename, "r", "mbox for indexing");
579 position_in_file = 0;
580 end_of_full_line = 0;
583 while(fgets(raw_mbox_line, BUFFER_SIZE, file)) {
584 if(is_a_leading_from_line(raw_mbox_line)) {
585 /* This starts a new mail */
588 "Got a ^\"From \" in the header in %s:%lu.\n",
589 mbox_filename, position_in_file);
590 fprintf(stderr, "%s", raw_mbox_line);
593 /* printf("LEADING_LINE %s", raw_mbox_line); */
597 } else if(raw_mbox_line[0] == '\n') {
600 /* We leave the header, index the current line */
602 /* printf("INDEX %s\n", full_line); */
603 index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
605 end_of_full_line = full_line;
606 *end_of_full_line = '\0';
612 fprintf(db_file, "mail %lu %s\n", position_in_file, mbox_filename);
616 if(raw_mbox_line[0] == ' ' || raw_mbox_line[0] == '\t') {
617 /* Continuation of a line */
618 char *start = raw_mbox_line;
619 while(*start == ' ' || *start == '\t') start++;
620 *(end_of_full_line++) = ' ';
621 strcpy(end_of_full_line, start);
622 while(*end_of_full_line && *end_of_full_line != '\n') {
625 *end_of_full_line = '\0';
629 /* Start a new header line, not a continuation */
632 /* printf("INDEX %s\n", full_line); */
633 index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
636 end_of_full_line = full_line;
637 strcpy(end_of_full_line, raw_mbox_line);
638 while(*end_of_full_line && *end_of_full_line != '\n') {
641 *end_of_full_line = '\0';
646 position_in_file += strlen(raw_mbox_line);
652 void recursive_index_mbox(FILE *db_file,
653 const char *entry_name, regex_t *mbox_filename_regexp,
654 int nb_fields_to_parse, struct parsable_field *fields_to_parse) {
656 struct dirent *dir_e;
658 char subname[PATH_MAX + 1];
660 if(lstat(entry_name, &sb) != 0) {
662 "mymail: Cannot stat \"%s\": %s\n",
668 dir = opendir(entry_name);
671 while((dir_e = readdir(dir))) {
672 if(!ignore_entry(dir_e->d_name)) {
673 snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
674 recursive_index_mbox(db_file, subname, mbox_filename_regexp,
675 nb_fields_to_parse, fields_to_parse);
680 const char *s = entry_name, *filename = s;
681 while(*s) { if(*s == '/') { filename = s+1; }; s++; }
682 if(!mbox_filename_regexp || regexec(mbox_filename_regexp, filename, 0, 0, 0) == 0) {
683 index_mbox(entry_name, nb_fields_to_parse, fields_to_parse, db_file);
688 /*********************************************************************/
690 /* For long options that have no equivalent short option, use a
691 non-character as a pseudo short option, starting with CHAR_MAX + 1. */
693 OPT_BASH_MODE = CHAR_MAX + 1
696 static struct option long_options[] = {
697 { "help", no_argument, 0, 'h' },
698 { "version", no_argument, 0, 'v' },
699 { "quiet", no_argument, 0, 'q' },
700 { "use-leading-time", no_argument, 0, 't' },
701 { "db-file-generate", 1, 0, 'd' },
702 { "db-pattern", 1, 0, 'p' },
703 { "db-root", 1, 0, 'r' },
704 { "db-list", 1, 0, 'l' },
705 { "mbox-pattern", 1, 0, 'm' },
706 { "search", 1, 0, 's' },
707 { "index", 0, 0, 'i' },
708 { "output", 1, 0, 'o' },
709 { "default-search", 1, 0, 'a' },
713 struct time_criterion {
715 int start_hour, end_hour;
719 /*********************************************************************/
721 static struct time_criterion time_criteria[] = {
724 { "today", 24, -1, -1 },
725 { "24h", 24, -1, -1 },
726 { "48h", 48, -1, -1 },
727 { "week", 24 * 7, -1, -1 },
728 { "month", 24 * 31, -1, -1 },
729 { "year", 24 * 365, -1, -1 },
731 { "yesterday", 48, 24, -1 },
733 { "monday", -1, -1, 1 },
734 { "tuesday", -1, -1, 2 },
735 { "wednesday", -1, -1, 3 },
736 { "thursday", -1, -1, 4 },
737 { "friday", -1, -1, 5 },
738 { "saturday", -1, -1, 6 },
739 { "sunday", -1, -1, 7 },
743 /*********************************************************************/
745 time_t time_for_past_day(int day) {
751 delta_day = (7 + tm->tm_wday - day) % 7;
752 if(delta_day == 0) { delta_day = 7; }
753 return t - (delta_day * 3600 * 24 + tm->tm_sec + 60 * tm->tm_min + 3600 * tm->tm_hour);
756 void init_condition(struct search_condition *condition, const char *full_string,
757 const char *default_search_field) {
758 char full_search_field[TOKEN_BUFFER_SIZE], *search_field;
762 string = parse_token(full_search_field, TOKEN_BUFFER_SIZE, ' ', full_string);
763 search_field = full_search_field;
765 if(search_field[0] == '!') {
767 condition->negation = 1;
769 condition->negation = 0;
772 condition->field_id = -1;
776 for(k = 0; k < sizeof(time_criteria) / sizeof(struct time_criterion); k++) {
777 if(strcmp(time_criteria[k].label, search_field) == 0) {
778 condition->field_id = ID_INTERVAL;
779 if(time_criteria[k].past_week_day < 0) {
780 condition->interval_start = time(0) - 3600 * time_criteria[k].start_hour;
781 if(time_criteria[k].end_hour >= 0) {
782 condition->interval_stop = time(0) - 3600 * time_criteria[k].end_hour;
784 condition->interval_stop = 0;
787 condition->interval_start = time_for_past_day(time_criteria[k].past_week_day);
788 condition->interval_stop = condition->interval_start + 3600 * 24;
794 if(condition->field_id == -1) {
796 /* No time condition matched, look for the search fields */
798 for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) {
799 if(strncmp(field_keys[m], search_field, strlen(search_field)) == 0) {
800 condition->field_id = m;
804 /* None match, if there is a default search field, re-run the search with it */
806 if(condition->field_id == -1) {
807 if(default_search_field) {
808 for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) {
809 if(strncmp(field_keys[m],
810 default_search_field, strlen(default_search_field)) == 0) {
811 condition->field_id = m;
814 string = full_string;
815 if(string[0] == '!') { string++; }
819 if(condition->field_id == -1) {
821 "mymail: Syntax error in field key \"%s\".\n",
826 if(regcomp(&condition->regexp,
830 "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
832 field_keys[condition->field_id]);
838 void free_condition(struct search_condition *condition) {
839 if(condition->field_id != ID_INTERVAL) {
840 regfree(&condition->regexp);
844 /*********************************************************************/
845 /*********************************************************************/
846 /*********************************************************************/
848 int main(int argc, char **argv) {
849 char *db_filename = 0;
850 char *db_filename_regexp_string = 0;
851 char *db_root_path = 0;
852 char *db_filename_list = 0;
853 char *mbox_filename_regexp_string = 0;
854 char *default_search_field;
855 char output_filename[PATH_MAX + 1];
856 int action_index = 0;
857 int error = 0, show_help = 0;
858 const unsigned int nb_fields_to_parse =
859 sizeof(fields_to_parse) / sizeof(struct parsable_field);
862 unsigned int nb_search_conditions;
863 struct search_condition search_conditions[MAX_NB_SEARCH_CONDITIONS];
865 if(regcomp(&global_leading_from_line_regexp, LEADING_FROM_LINE_REGEXP_STRING, 0)) {
867 "mymail: Cannot compile leading \"from\" line regexp. That is strange.\n");
872 global_use_leading_time = 0;
873 default_search_field = 0;
874 strncpy(output_filename, "", PATH_MAX);
876 setlocale(LC_ALL, "");
878 nb_search_conditions = 0;
880 while ((c = getopt_long(argc, argv, "hvqip:s:d:r:l:o:a:m:",
881 long_options, NULL)) != -1) {
890 print_version(stdout);
898 global_use_leading_time = 1;
907 fprintf(stderr, "mymail: Can not set the db filename twice.\n");
910 db_filename = strdup(optarg);
914 if(db_filename_regexp_string) {
915 fprintf(stderr, "mymail: Can not set the db filename pattern twice.\n");
918 db_filename_regexp_string = strdup(optarg);
922 if(mbox_filename_regexp_string) {
923 fprintf(stderr, "mymail: Can not set the mbox filename pattern twice.\n");
926 mbox_filename_regexp_string = strdup(optarg);
930 strncpy(output_filename, optarg, PATH_MAX);
935 fprintf(stderr, "mymail: Can not set the db root path twice.\n");
938 db_root_path = strdup(optarg);
942 if(db_filename_list) {
943 fprintf(stderr, "mymail: Can not set the db filename list twice.\n");
946 db_filename_list = strdup(optarg);
950 if(nb_search_conditions == MAX_NB_SEARCH_CONDITIONS) {
951 fprintf(stderr, "mymail: Too many search patterns.\n");
954 init_condition(&search_conditions[nb_search_conditions], optarg, default_search_field);
955 nb_search_conditions++;
959 default_search_field = optarg;
968 /* Set all the values that may defined in the arguments, through
969 environment variables, or hard-coded */
971 db_filename = default_value(db_filename,
975 db_filename_regexp_string = default_value(db_filename_regexp_string,
979 db_root_path = default_value(db_root_path,
983 db_filename_list = default_value(db_filename_list,
987 mbox_filename_regexp_string = default_value(mbox_filename_regexp_string,
988 "MYMAIL_MBOX_PATTERN",
991 /* Start the processing */
1007 regex_t mbox_filename_regexp_static;
1008 regex_t *mbox_filename_regexp;
1010 if(mbox_filename_regexp_string) {
1011 if(regcomp(&mbox_filename_regexp_static,
1012 mbox_filename_regexp_string,
1015 "mymail: Syntax error in regexp \"%s\".\n",
1016 mbox_filename_regexp_string);
1019 mbox_filename_regexp = &mbox_filename_regexp_static;
1021 mbox_filename_regexp = 0;
1024 db_file = safe_fopen(db_filename, "w", "index file for indexing");
1026 for(f = 0; f < nb_fields_to_parse; f++) {
1027 if(regcomp(&fields_to_parse[f].regexp,
1028 fields_to_parse[f].regexp_string,
1029 fields_to_parse[f].cflags)) {
1031 "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
1032 fields_to_parse[f].regexp_string,
1033 field_keys[fields_to_parse[f].id]);
1038 fprintf(db_file, "%s version_%s raw\n", MYMAIL_DB_MAGIC_TOKEN, VERSION);
1040 while(optind < argc) {
1041 recursive_index_mbox(db_file,
1042 argv[optind], mbox_filename_regexp,
1043 nb_fields_to_parse, fields_to_parse);
1050 if(mbox_filename_regexp) {
1051 regfree(mbox_filename_regexp);
1054 for(f = 0; f < nb_fields_to_parse; f++) {
1055 regfree(&fields_to_parse[f].regexp);
1064 int nb_extracted_mails = 0;
1066 if(output_filename[0]) {
1067 output_file = safe_fopen(output_filename, "w", "result mbox");
1069 output_file = stdout;
1073 if(nb_search_conditions > 0) {
1075 /* Recursive search if db_root_path is set */
1078 regex_t db_filename_regexp;
1079 if(regcomp(&db_filename_regexp,
1080 db_filename_regexp_string,
1083 "mymail: Syntax error in regexp \"%s\".\n",
1084 db_filename_regexp_string);
1088 nb_extracted_mails += recursive_search_in_db(db_root_path, &db_filename_regexp,
1089 nb_search_conditions, search_conditions,
1092 regfree(&db_filename_regexp);
1095 /* Search in all db files listed in db_filename_list */
1097 if(db_filename_list) {
1098 char db_filename[PATH_MAX + 1];
1101 s = db_filename_list;
1104 s = parse_token(db_filename, PATH_MAX + 1, ';', s);
1106 if(db_filename[0]) {
1107 nb_extracted_mails +=
1108 search_in_db(db_filename, nb_search_conditions, search_conditions, output_file);
1113 /* Search in all db files listed in the command arguments */
1115 while(optind < argc) {
1116 nb_extracted_mails +=
1117 search_in_db(argv[optind], nb_search_conditions, search_conditions, output_file);
1123 if(nb_extracted_mails > 0) {
1124 printf("Found %d matching mails.\n", nb_extracted_mails);
1126 printf("No matching mail found.\n");
1130 fflush(output_file);
1132 if(output_file != stdout) {
1133 fclose(output_file);
1137 for(n = 0; n < nb_search_conditions; n++) {
1138 free_condition(&search_conditions[n]);
1142 free(db_filename_regexp_string);
1144 free(db_filename_list);
1145 free(mbox_filename_regexp_string);
1147 regfree(&global_leading_from_line_regexp);