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 mymail is a simple mail indexer. It can:
26 (1) scan mbox files, and create a db file containing for each mail a
27 list of fields computed from its header.
29 (2) read such a db file, gets all the mails matching regexp-defined
30 conditions on the fields, and generates a resulting mbox file.
32 It is low-tech, simple, light and fast.
51 #define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file"
52 #define MYMAIL_VERSION "0.9.9"
54 #define MYMAIL_DB_FORMAT_VERSION 1
56 #define MAX_NB_SEARCH_CONDITIONS 32
58 #define BUFFER_SIZE 65536
59 #define TOKEN_BUFFER_SIZE 1024
61 #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\\) [ 0123][0-9] [0-9][0-9]:[0-9][0-9]:[0-9][0-9] [0-9][0-9][0-9][0-9]\n$"
63 /********************************************************************/
67 struct alias_node *next;
70 /* Global variables! */
73 int global_use_leading_time;
74 int global_nb_mails_max;
75 int global_discard_mail_from_the_future;
76 regex_t global_leading_from_line_regexp;
77 struct alias_node *global_alias_list;
78 time_t global_current_time;
80 /********************************************************************/
98 static char *field_keys[] = {
113 /********************************************************************/
115 struct search_condition {
117 regex_t db_value_regexp;
119 time_t time_start, time_stop;
122 /********************************************************************/
124 struct parsable_field {
131 static struct parsable_field fields_to_parse[] = {
136 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
142 "^\\(from\\|reply-to\\|sender\\|return-path\\): ",
143 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
149 "^\\(to\\|cc\\|bcc\\): ",
150 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
157 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
164 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
171 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
177 "^\\(in-reply-to\\|references\\): ",
178 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
183 /********************************************************************/
185 int xor(int a, int b) {
186 return (a && !b) || (!a && b);
189 const char *parse_token(char *token_buffer, size_t token_buffer_size,
190 char separator, const char *string) {
191 char *u = token_buffer;
193 while(*string == separator) { string++; }
195 while(u < token_buffer + token_buffer_size - 1 && *string && *string != separator) {
196 *(u++) = *(string++);
199 while(*string == separator) { string++; }
205 char *default_value(char *current_value,
206 const char *env_variable,
207 const char *hard_default_value) {
209 return current_value;
211 char *env_value = getenv(env_variable);
213 return strdup(env_value);
214 } else if(hard_default_value) {
215 return strdup(hard_default_value);
222 /********************************************************************/
224 void *safe_malloc(size_t n) {
228 "mymail: cannot allocate memory: %s\n", strerror(errno));
234 FILE *safe_fopen(const char *path, const char *mode, const char *comment) {
235 FILE *result = fopen(path, mode);
240 "mymail: Cannot open file '%s' (%s) with mode \"%s\": %s\n",
247 /*********************************************************************/
249 void print_version(FILE *out) {
250 fprintf(out, "mymail version %s (%s)\n", MYMAIL_VERSION, UNAME);
253 void print_usage(FILE *out) {
255 fprintf(out, "Written by Francois Fleuret <francois@fleuret.org>.\n");
257 fprintf(out, "Usage: mymail [options] [<mbox dir1> [<mbox dir2> ...]|<db file1> [<db file2> ...]]\n");
259 fprintf(out, " -h, --help\n");
260 fprintf(out, " show this help\n");
261 fprintf(out, " -v, --version\n");
262 fprintf(out, " print the version number\n");
263 fprintf(out, " -q, --quiet\n");
264 fprintf(out, " do not print information during search\n");
265 fprintf(out, " -t, --use-leading-time\n");
266 fprintf(out, " use the time stamp from the leading line of each mail and not the Date:\n");
267 fprintf(out, " field\n");
268 fprintf(out, " -f, --do-not-discard-mails-from-the-future\n");
269 fprintf(out, " do not ignore mails with a date more than 24h in the future\n");
270 fprintf(out, " -p <db filename pattern>, --db-pattern <db filename pattern>\n");
271 fprintf(out, " set the db filename pattern for recursive search\n");
272 fprintf(out, " -r <db root path>, --db-root <db root path>\n");
273 fprintf(out, " set the db root path for recursive search\n");
274 fprintf(out, " -l <db filename list>, --db-list <db filename list>\n");
275 fprintf(out, " set the semicolon-separated list of db files for search\n");
276 fprintf(out, " -m <mbox filename pattern>, --mbox-pattern <mbox filename pattern>\n");
277 fprintf(out, " set the mbox filename pattern for recursive search\n");
278 fprintf(out, " -s <search pattern>, --search <search pattern>\n");
279 fprintf(out, " search for matching mails in the db file\n");
280 fprintf(out, " -d <db filename>, --db-file-output <db filename>\n");
281 fprintf(out, " set the db filename for indexing\n");
282 fprintf(out, " -i, --index\n");
283 fprintf(out, " index mails\n");
284 fprintf(out, " -o <output filename>, --output <output filename>\n");
285 fprintf(out, " set the result file, use stdout if unset\n");
286 fprintf(out, " -n <max number of mails>, --nb-mails-max <max number of mails>\n");
287 fprintf(out, " set the maximum number of mails to extract\n");
288 fprintf(out, " -a <search field>, --default-search <search field>\n");
289 fprintf(out, " set the default search field\n");
293 /*********************************************************************/
295 int ignore_entry(const char *name) {
297 strcmp(name, ".") == 0 ||
298 strcmp(name, "..") == 0 ||
299 (name[0] == '.' && name[1] != '/');
302 int is_a_leading_from_line(char *mbox_line) {
304 strncmp(mbox_line, "From ", 5) == 0 &&
305 regexec(&global_leading_from_line_regexp, mbox_line, 0, 0, 0) == 0;
308 int db_line_match_search(struct search_condition *condition,
309 int db_key, const char *db_value) {
313 (condition->db_key == db_key)
317 (condition->db_key == ID_PARTICIPANT && (db_key == ID_LEADING_LINE ||
322 (condition->db_key == ID_FROM && db_key == ID_LEADING_LINE)
326 (condition->db_key == ID_THREAD_ID && (db_key == ID_MAIL_ID ||
327 db_key == ID_REFERENCE_ID))
332 regexec(&condition->db_value_regexp, db_value, 0, 0, 0) == 0;
335 void update_body_hits(char *mail_filename, int position_in_mail,
336 int nb_search_conditions, struct search_condition *search_conditions,
337 int nb_body_conditions,
341 char raw_mbox_line[BUFFER_SIZE];
347 mail_file = safe_fopen(mail_filename, "r", "mbox for body scan");
349 fseek(mail_file, position_in_mail, SEEK_SET);
351 if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
352 while(nb_body_hits < nb_body_conditions) {
353 if(raw_mbox_line[0] == '\n') { header = 0; }
356 for(n = 0; n < nb_search_conditions; n++) {
357 if(search_conditions[n].db_key == ID_BODY && !hits[n]) {
359 (regexec(&search_conditions[n].db_value_regexp, raw_mbox_line, 0, 0, 0) == 0);
367 if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
368 (is_a_leading_from_line(raw_mbox_line)))
376 void extract_mail(const char *mail_filename, unsigned long int position_in_mail,
378 char raw_mbox_line[BUFFER_SIZE];
381 /* printf("Extract\n"); */
383 mail_file = safe_fopen(mail_filename, "r", "mbox for mail extraction");
384 /* fchmod(fileno(mail_file), 0x660); */
385 fseek(mail_file, position_in_mail, SEEK_SET);
387 if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
388 fprintf(output_file, "%s", raw_mbox_line);
390 if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
391 is_a_leading_from_line(raw_mbox_line))
393 fprintf(output_file, "%s", raw_mbox_line);
400 int check_full_mail_match(char *current_mail_filename,
402 int nb_search_conditions,
403 struct search_condition *search_conditions,
404 int nb_body_conditions,
406 int current_position_in_mail) {
407 int n, nb_fulfilled_body_conditions;
409 for(n = 0; n < nb_search_conditions; n++) {
410 if(search_conditions[n].db_key == ID_TIME_INTERVAL) {
411 hits[n] = (mail_time >= search_conditions[n].time_start &&
412 (search_conditions[n].time_stop == 0 ||
413 mail_time <= search_conditions[n].time_stop));
417 /* We first check all conditions but the body ones */
419 for(n = 0; n < nb_search_conditions &&
420 ((search_conditions[n].db_key == ID_BODY) ||
421 xor(hits[n], search_conditions[n].negation)); n++);
423 if(n == nb_search_conditions) {
425 /* Now check the body ones */
427 nb_fulfilled_body_conditions = 0;
429 if(nb_body_conditions > 0) {
430 update_body_hits(current_mail_filename, current_position_in_mail,
431 nb_search_conditions, search_conditions,
435 for(n = 0; n < nb_search_conditions; n++) {
436 if(search_conditions[n].db_key == ID_BODY &&
437 xor(hits[n], search_conditions[n].negation)) {
438 nb_fulfilled_body_conditions++;
442 return nb_body_conditions == nb_fulfilled_body_conditions;
448 /* We use the mail leading line time by default, and if we should and
449 can, we update with the Date: field */
451 void update_time(int db_key, const char *db_value, time_t *t) {
455 memset(&tm, 0, sizeof(struct tm));
457 if(db_key == ID_LEADING_LINE) {
459 while(*c && *c != ' ') c++; while(*c && *c == ' ') c++;
460 /* printf("From %s", db_value); */
461 strptime(c, "%a %b %e %k:%M:%S %Y", &tm);
464 if(!global_use_leading_time) {
465 if(db_key == ID_DATE) {
466 if(strptime(db_value, "%a, %d %b %Y %k:%M:%S", &tm) ||
467 strptime(db_value, "%d %b %Y %k:%M:%S", &tm)) {
468 /* printf("Date: %s", db_value); */
476 int search_in_db(const char *db_filename,
477 int nb_extracted_mails,
478 int nb_search_conditions,
479 struct search_condition *search_conditions,
483 char raw_db_line[BUFFER_SIZE];
484 char current_mail_filename[PATH_MAX + 1];
485 char db_key_string[TOKEN_BUFFER_SIZE];
486 char position_in_file_string[TOKEN_BUFFER_SIZE];
487 unsigned long int current_position_in_mail;
488 const char *db_value;
490 int hits[MAX_NB_SEARCH_CONDITIONS];
491 int nb_body_conditions, need_time;
497 printf("Searching in '%s' ... ", db_filename);
501 db_file = safe_fopen(db_filename, "r", "index file for search");
503 /* First, check the db file leading line integrity */
505 if(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
506 if(strncmp(raw_db_line, MYMAIL_DB_MAGIC_TOKEN, strlen(MYMAIL_DB_MAGIC_TOKEN))) {
508 "mymail: Header line in '%s' does not match the mymail db format.\n",
514 "mymail: Cannot read the header line in '%s'.\n",
519 /* Then parse the said db file */
521 current_position_in_mail = 0;
523 for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
525 nb_body_conditions = 0;
526 need_time = global_discard_mail_from_the_future;
529 for(n = 0; n < nb_search_conditions; n++) {
530 if(search_conditions[n].db_key == ID_BODY) {
531 nb_body_conditions++;
533 else if(search_conditions[n].db_key == ID_TIME_INTERVAL) {
538 strcpy(current_mail_filename, "");
540 while(nb_extracted_mails < global_nb_mails_max &&
541 fgets(raw_db_line, BUFFER_SIZE, db_file)) {
544 char *s = raw_db_line;
545 while(*s && *s != '\n') { s++; }
548 db_value = parse_token(db_key_string, TOKEN_BUFFER_SIZE, ' ', raw_db_line);
550 if(strcmp("mail", db_key_string) == 0) {
551 if(current_mail_filename[0] &&
552 (!global_discard_mail_from_the_future || mail_time < global_current_time + 3600 * 24) &&
553 check_full_mail_match(current_mail_filename,
555 nb_search_conditions, search_conditions,
556 nb_body_conditions, hits, current_position_in_mail)) {
557 extract_mail(current_mail_filename, current_position_in_mail, output_file);
558 nb_extracted_mails++;
561 for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
562 db_value = parse_token(position_in_file_string, TOKEN_BUFFER_SIZE, ' ', db_value);
563 strncpy(current_mail_filename, db_value, PATH_MAX + 1);
564 current_position_in_mail = atol(position_in_file_string);
569 for(m = 0; (m < MAX_ID) && db_key == -1; m++) {
570 if(strncmp(field_keys[m], db_key_string, strlen(db_key_string)) == 0) {
575 for(n = 0; n < nb_search_conditions; n++) {
576 hits[n] |= db_line_match_search(&search_conditions[n],
581 update_time(db_key, db_value, &mail_time);
586 /* printf("global_discard_mail_from_the_future = %d\n",
587 global_discard_mail_from_the_future); */
589 if(nb_extracted_mails < global_nb_mails_max &&
590 current_mail_filename[0] &&
591 (!global_discard_mail_from_the_future || mail_time < global_current_time + 3600 * 24) &&
592 check_full_mail_match(current_mail_filename,
594 nb_search_conditions, search_conditions,
595 nb_body_conditions, hits, current_position_in_mail)) {
596 extract_mail(current_mail_filename, current_position_in_mail, output_file);
597 nb_extracted_mails++;
607 return nb_extracted_mails;
610 int recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp,
611 int nb_extracted_mails,
612 int nb_search_conditions,
613 struct search_condition *search_conditions,
616 struct dirent *dir_e;
618 char subname[PATH_MAX + 1];
620 if(lstat(entry_name, &sb) != 0) {
622 "mymail: Cannot stat \"%s\": %s\n",
628 /* printf("recursive_search_in_db %s\n", entry_name); */
630 dir = opendir(entry_name);
633 while((dir_e = readdir(dir)) &&
634 nb_extracted_mails < global_nb_mails_max) {
635 if(!ignore_entry(dir_e->d_name)) {
636 snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
637 nb_extracted_mails = recursive_search_in_db(subname, db_filename_regexp,
639 nb_search_conditions, search_conditions,
647 const char *s = entry_name, *filename = entry_name;
648 while(*s) { if(*s == '/') { filename = s+1; } s++; }
650 if(regexec(db_filename_regexp, filename, 0, 0, 0) == 0) {
652 search_in_db(entry_name,
654 nb_search_conditions, search_conditions, output_file);
658 return nb_extracted_mails;
661 /*********************************************************************/
663 void index_one_mbox_line(unsigned int nb_fields_to_parse,
664 struct parsable_field *fields_to_parse,
665 char *raw_mbox_line, FILE *db_file) {
668 for(f = 0; f < nb_fields_to_parse; f++) {
669 if(regexec(&fields_to_parse[f].regexp, raw_mbox_line, 1, &matches, 0) == 0) {
670 fprintf(db_file, "%s %s\n",
671 field_keys[fields_to_parse[f].id],
672 raw_mbox_line + matches.rm_eo);
677 void index_mbox(const char *mbox_filename,
678 int nb_fields_to_parse, struct parsable_field *fields_to_parse,
680 char raw_mbox_line[BUFFER_SIZE], full_line[BUFFER_SIZE];
681 char *end_of_full_line;
683 int in_header, new_header;
684 unsigned long int position_in_file;
686 file = safe_fopen(mbox_filename, "r", "mbox for indexing");
691 position_in_file = 0;
692 end_of_full_line = 0;
695 while(fgets(raw_mbox_line, BUFFER_SIZE, file)) {
696 if(is_a_leading_from_line(raw_mbox_line)) {
697 /* This starts a new mail */
700 "Got a ^\"From \" in the header in %s:%lu.\n",
701 mbox_filename, position_in_file);
702 fprintf(stderr, "%s", raw_mbox_line);
705 /* printf("LEADING_LINE %s", raw_mbox_line); */
709 } else if(raw_mbox_line[0] == '\n') {
712 /* We leave the header, index the current line */
714 /* printf("INDEX %s\n", full_line); */
715 index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
717 end_of_full_line = full_line;
718 *end_of_full_line = '\0';
724 fprintf(db_file, "mail %lu %s\n", position_in_file, mbox_filename);
728 if(raw_mbox_line[0] == ' ' || raw_mbox_line[0] == '\t') {
729 /* Continuation of a line */
730 char *start = raw_mbox_line;
731 while(*start == ' ' || *start == '\t') start++;
732 *(end_of_full_line++) = ' ';
733 strcpy(end_of_full_line, start);
734 while(*end_of_full_line && *end_of_full_line != '\n') {
737 *end_of_full_line = '\0';
741 /* Start a new header line, not a continuation */
744 /* printf("INDEX %s\n", full_line); */
745 index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
748 end_of_full_line = full_line;
749 strcpy(end_of_full_line, raw_mbox_line);
750 while(*end_of_full_line && *end_of_full_line != '\n') {
753 *end_of_full_line = '\0';
758 position_in_file += strlen(raw_mbox_line);
764 void recursive_index_mbox(FILE *db_file,
765 const char *entry_name, regex_t *mbox_filename_regexp,
766 int nb_fields_to_parse, struct parsable_field *fields_to_parse) {
768 struct dirent *dir_e;
770 char subname[PATH_MAX + 1];
772 if(lstat(entry_name, &sb) != 0) {
774 "mymail: Cannot stat \"%s\": %s\n",
780 dir = opendir(entry_name);
783 while((dir_e = readdir(dir))) {
784 if(!ignore_entry(dir_e->d_name)) {
785 snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
786 recursive_index_mbox(db_file, subname, mbox_filename_regexp,
787 nb_fields_to_parse, fields_to_parse);
792 const char *s = entry_name, *filename = s;
793 while(*s) { if(*s == '/') { filename = s+1; }; s++; }
794 if(!mbox_filename_regexp || regexec(mbox_filename_regexp, filename, 0, 0, 0) == 0) {
795 index_mbox(entry_name, nb_fields_to_parse, fields_to_parse, db_file);
800 /*********************************************************************/
802 /* For long options that have no equivalent short option, use a
803 non-character as a pseudo short option, starting with CHAR_MAX + 1. */
805 OPT_BASH_MODE = CHAR_MAX + 1
808 static struct option long_options[] = {
809 { "help", no_argument, 0, 'h' },
810 { "version", no_argument, 0, 'v' },
811 { "quiet", no_argument, 0, 'q' },
812 { "use-leading-time", no_argument, 0, 't' },
813 { "do-not-discard-mails-from-the-future", no_argument, 0, 'f' },
814 { "db-file-output", 1, 0, 'd' },
815 { "db-pattern", 1, 0, 'p' },
816 { "db-root", 1, 0, 'r' },
817 { "db-list", 1, 0, 'l' },
818 { "mbox-pattern", 1, 0, 'm' },
819 { "search", 1, 0, 's' },
820 { "index", 0, 0, 'i' },
821 { "output", 1, 0, 'o' },
822 { "default-search", 1, 0, 'a' },
823 { "nb-mails-max", 1, 0, 'n' },
827 struct time_criterion {
830 int start_hour, end_hour;
834 /*********************************************************************/
836 static struct time_criterion time_criteria[] = {
838 { "1h", 0, 1, -1, -1 },
839 { "8h", 0, 8, -1, -1 },
840 { "24h", 0, 24, -1, -1 },
841 { "48h", 0, 48, -1, -1 },
842 { "week", 0, 24 * 7, -1, -1 },
843 { "2weeks", 0, 24 * 14, -1, -1 },
844 { "month", 0, 24 * 31, -1, -1 },
845 { "trimester", 0, 24 * 92, -1, -1 },
846 { "year", 0, 24 * 365, -1, -1 },
848 { "yesterday", 1, -1, -1, -1 },
849 { "today", 1, -1, -1, 0 },
851 { "monday", 1, -1, -1, 1 },
852 { "tuesday", 1, -1, -1, 2 },
853 { "wednesday", 1, -1, -1, 3 },
854 { "thursday", 1, -1, -1, 4 },
855 { "friday", 1, -1, -1, 5 },
856 { "saturday", 1, -1, -1, 6 },
857 { "sunday", 1, -1, -1, 7 },
861 /*********************************************************************/
863 time_t time_for_past_day(int day) {
866 tm = localtime(&global_current_time);
868 delta_day = (7 + tm->tm_wday - day) % 7;
872 return global_current_time - (delta_day * 3600 * 24 + tm->tm_sec + 60 * tm->tm_min + 3600 * tm->tm_hour);
875 void init_condition(struct search_condition *condition, const char *full_string,
876 const char *default_search_field) {
877 char full_search_field[TOKEN_BUFFER_SIZE], *search_field;
880 struct alias_node *a;
882 for(a = global_alias_list; a; a = a->next) {
883 if(strcmp(full_string, a->alias) == 0) {
884 full_string = a->value;
889 string = parse_token(full_search_field, TOKEN_BUFFER_SIZE, ' ', full_string);
890 search_field = full_search_field;
892 if(search_field[0] == '!') {
894 condition->negation = 1;
896 condition->negation = 0;
899 condition->db_key = -1;
903 for(k = 0; k < sizeof(time_criteria) / sizeof(struct time_criterion); k++) {
904 if(strcmp(time_criteria[k].label, search_field) == 0) {
905 condition->db_key = ID_TIME_INTERVAL;
906 if(time_criteria[k].day_criterion) {
907 condition->time_start = time_for_past_day(time_criteria[k].past_week_day);
908 condition->time_stop = condition->time_start + 3600 * 24;
910 condition->time_start = global_current_time - 3600 * time_criteria[k].start_hour;
911 if(time_criteria[k].end_hour >= 0) {
912 condition->time_stop = global_current_time - 3600 * time_criteria[k].end_hour;
914 condition->time_stop = 0;
922 if(condition->db_key == -1) {
924 /* No time condition matched, look for the search fields */
926 for(m = 0; (m < MAX_ID) && condition->db_key == -1; m++) {
927 if(strncmp(field_keys[m], search_field, strlen(search_field)) == 0) {
928 condition->db_key = m;
932 /* None match, if there is a default search field, re-run the search with it */
934 if(condition->db_key == -1) {
935 if(default_search_field) {
936 for(m = 0; (m < MAX_ID) && condition->db_key == -1; m++) {
937 if(strncmp(field_keys[m],
938 default_search_field, strlen(default_search_field)) == 0) {
939 condition->db_key = m;
942 string = full_string;
943 if(string[0] == '!') { string++; }
947 if(condition->db_key == -1) {
949 "mymail: Syntax error in field key \"%s\".\n",
954 if(regcomp(&condition->db_value_regexp,
958 "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
960 field_keys[condition->db_key]);
966 void free_condition(struct search_condition *condition) {
967 if(condition->db_key != ID_TIME_INTERVAL) {
968 regfree(&condition->db_value_regexp);
972 const char *eat_space(const char *s) {
973 while(*s == ' ' || *s == '\t') { s++; }
977 void read_rc_file(const char *rc_filename) {
978 char raw_line[BUFFER_SIZE];
979 char command[TOKEN_BUFFER_SIZE], tmp_token[TOKEN_BUFFER_SIZE];
986 rc_file = fopen(rc_filename, "r");
990 while(fgets(raw_line, BUFFER_SIZE, rc_file)) {
992 while(*t) { if(*t == '\n') { *t = '\0'; }; t++; }
997 if(*s && *s != '#') {
998 s = parse_token(command, TOKEN_BUFFER_SIZE, ' ', s);
1000 if(strcmp(command, "alias") == 0) {
1001 struct alias_node *a = safe_malloc(sizeof(struct alias_node));
1002 a->next = global_alias_list;
1003 global_alias_list = a;
1006 s = parse_token(tmp_token, TOKEN_BUFFER_SIZE, '=', s);
1007 a->alias = strdup(tmp_token);
1010 a->value = strdup(s);
1012 fprintf(stderr, "%s:%d syntax error, missing alias value.\n",
1018 fprintf(stderr, "%s:%d syntax error, missing alias key.\n",
1024 fprintf(stderr, "%s:%d syntax error, unknown command '%s'.\n",
1038 /*********************************************************************/
1039 /*********************************************************************/
1040 /*********************************************************************/
1042 int main(int argc, char **argv) {
1043 char *db_filename = 0;
1044 char *db_filename_regexp_string = 0;
1045 char *db_root_path = 0;
1046 char *db_filename_list = 0;
1047 char *mbox_filename_regexp_string = 0;
1048 char *default_search_field;
1049 char output_filename[PATH_MAX + 1];
1050 char rc_filename[PATH_MAX + 1];
1051 int action_index = 0;
1052 int error = 0, show_help = 0;
1053 const unsigned int nb_fields_to_parse =
1054 sizeof(fields_to_parse) / sizeof(struct parsable_field);
1057 unsigned int nb_search_conditions;
1058 struct search_condition search_conditions[MAX_NB_SEARCH_CONDITIONS];
1059 struct alias_node *a, *b;
1061 /* Group and others have no access */
1062 umask(S_IRWXG | S_IRWXO);
1064 if(regcomp(&global_leading_from_line_regexp, LEADING_FROM_LINE_REGEXP_STRING, 0)) {
1066 "mymail: Cannot compile leading \"from\" line regexp. That is strange.\n");
1070 if(getenv("MYMAILRC")) {
1071 sprintf(rc_filename, "%s", getenv("MYMAILRC"));
1072 } else if(getenv("HOME")) {
1073 sprintf(rc_filename, "%s/.mymailrc", getenv("HOME"));
1075 rc_filename[0] = '\0';
1078 global_alias_list = 0;
1080 global_use_leading_time = 0;
1081 global_nb_mails_max = 250;
1082 global_discard_mail_from_the_future = 1;
1083 global_current_time = time(0);
1085 default_search_field = 0;
1086 strncpy(output_filename, "", PATH_MAX);
1088 if(rc_filename[0]) {
1089 read_rc_file(rc_filename);
1094 #warning Test code added on 2013 May 02 11:17:01
1095 struct alias_node *a;
1096 for(a = global_alias_list; a; a = a->next) {
1097 printf ("ALIAS [%s] [%s]\n", a->alias, a->value);
1102 setlocale(LC_ALL, "");
1104 nb_search_conditions = 0;
1106 while ((c = getopt_long(argc, argv, "hvqtfip:s:d:r:l:o:a:m:",
1107 long_options, NULL)) != -1) {
1116 print_version(stdout);
1124 global_use_leading_time = 1;
1128 global_discard_mail_from_the_future = 0;
1137 fprintf(stderr, "mymail: Can not set the db filename twice.\n");
1140 db_filename = strdup(optarg);
1144 if(db_filename_regexp_string) {
1145 fprintf(stderr, "mymail: Can not set the db filename pattern twice.\n");
1148 db_filename_regexp_string = strdup(optarg);
1152 if(mbox_filename_regexp_string) {
1153 fprintf(stderr, "mymail: Can not set the mbox filename pattern twice.\n");
1156 mbox_filename_regexp_string = strdup(optarg);
1160 strncpy(output_filename, optarg, PATH_MAX);
1165 fprintf(stderr, "mymail: Can not set the db root path twice.\n");
1168 db_root_path = strdup(optarg);
1172 if(db_filename_list) {
1173 fprintf(stderr, "mymail: Can not set the db filename list twice.\n");
1176 db_filename_list = strdup(optarg);
1180 if(nb_search_conditions == MAX_NB_SEARCH_CONDITIONS) {
1181 fprintf(stderr, "mymail: Too many search patterns.\n");
1184 init_condition(&search_conditions[nb_search_conditions], optarg, default_search_field);
1185 nb_search_conditions++;
1189 default_search_field = optarg;
1193 global_nb_mails_max = atoi(optarg);
1203 print_usage(stderr);
1208 print_usage(stdout);
1212 /* Set all the values that may defined in the arguments, through
1213 environment variables, or hard-coded */
1215 db_filename = default_value(db_filename,
1219 db_filename_regexp_string = default_value(db_filename_regexp_string,
1223 db_root_path = default_value(db_root_path,
1227 db_filename_list = default_value(db_filename_list,
1231 mbox_filename_regexp_string = default_value(mbox_filename_regexp_string,
1232 "MYMAIL_MBOX_PATTERN",
1239 regex_t mbox_filename_regexp_static;
1240 regex_t *mbox_filename_regexp;
1242 if(mbox_filename_regexp_string) {
1243 if(regcomp(&mbox_filename_regexp_static,
1244 mbox_filename_regexp_string,
1247 "mymail: Syntax error in regexp \"%s\".\n",
1248 mbox_filename_regexp_string);
1251 mbox_filename_regexp = &mbox_filename_regexp_static;
1253 mbox_filename_regexp = 0;
1256 db_file = safe_fopen(db_filename, "w", "index file for indexing");
1258 for(f = 0; f < nb_fields_to_parse; f++) {
1259 if(regcomp(&fields_to_parse[f].regexp,
1260 fields_to_parse[f].regexp_string,
1261 fields_to_parse[f].cflags)) {
1263 "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
1264 fields_to_parse[f].regexp_string,
1265 field_keys[fields_to_parse[f].id]);
1271 "%s version_%s format_%d raw\n",
1272 MYMAIL_DB_MAGIC_TOKEN,
1274 MYMAIL_DB_FORMAT_VERSION);
1276 while(optind < argc) {
1277 recursive_index_mbox(db_file,
1278 argv[optind], mbox_filename_regexp,
1279 nb_fields_to_parse, fields_to_parse);
1286 if(mbox_filename_regexp) {
1287 regfree(mbox_filename_regexp);
1290 for(f = 0; f < nb_fields_to_parse; f++) {
1291 regfree(&fields_to_parse[f].regexp);
1300 int nb_extracted_mails = 0;
1302 if(output_filename[0]) {
1303 output_file = safe_fopen(output_filename, "w", "result mbox");
1305 output_file = stdout;
1309 if(nb_search_conditions > 0) {
1311 /* Recursive search if db_root_path is set */
1314 regex_t db_filename_regexp;
1315 if(regcomp(&db_filename_regexp,
1316 db_filename_regexp_string,
1319 "mymail: Syntax error in regexp \"%s\".\n",
1320 db_filename_regexp_string);
1324 nb_extracted_mails = recursive_search_in_db(db_root_path, &db_filename_regexp,
1326 nb_search_conditions, search_conditions,
1329 regfree(&db_filename_regexp);
1332 /* Search in all db files listed in db_filename_list */
1334 if(db_filename_list) {
1335 char db_filename[PATH_MAX + 1];
1338 s = db_filename_list;
1341 s = parse_token(db_filename, PATH_MAX + 1, ';', s);
1343 if(db_filename[0]) {
1344 nb_extracted_mails =
1345 search_in_db(db_filename,
1347 nb_search_conditions, search_conditions, output_file);
1352 /* Search in all db files listed in the command arguments */
1354 while(optind < argc) {
1355 nb_extracted_mails =
1356 search_in_db(argv[optind],
1358 nb_search_conditions, search_conditions, output_file);
1364 if(nb_extracted_mails > 0) {
1365 printf("Found %d matching mails.\n", nb_extracted_mails);
1367 printf("No matching mail found.\n");
1371 fflush(output_file);
1373 if(output_file != stdout) {
1374 fclose(output_file);
1378 for(n = 0; n < nb_search_conditions; n++) {
1379 free_condition(&search_conditions[n]);
1382 a = global_alias_list;
1392 free(db_filename_regexp_string);
1394 free(db_filename_list);
1395 free(mbox_filename_regexp_string);
1397 regfree(&global_leading_from_line_regexp);