+ fprintf(out, " -h, --help\n");
+ fprintf(out, " show this help\n");
+ fprintf(out, " -v, --version\n");
+ fprintf(out, " print the version number\n");
+ fprintf(out, " -q, --quiet\n");
+ fprintf(out, " do not print information during search\n");
+ fprintf(out, " -p <db filename pattern>, --db-pattern <db filename pattern>\n");
+ fprintf(out, " set the db filename pattern for recursive search\n");
+ fprintf(out, " -r <db root path>, --db-root <db root path>\n");
+ fprintf(out, " set the db root path for recursive search\n");
+ fprintf(out, " -l <db filename list>, --db-list <db filename list>\n");
+ fprintf(out, " set the semicolon-separated list of db files for search\n");
+ fprintf(out, " -s <search pattern>, --search <search pattern>\n");
+ fprintf(out, " search for matching mails in the db file\n");
+ fprintf(out, " -d <db filename>, --db-file <db filename>\n");
+ fprintf(out, " set the db filename for indexing\n");
+ fprintf(out, " -i, --index\n");
+ fprintf(out, " index mails\n");
+ fprintf(out, " -o <output filename>, --output <output filename>\n");
+ fprintf(out, " set the result file, use stdout if unset\n");
+ fprintf(out, " -a <search field>, --default-search <search field>\n");
+ fprintf(out, " set the default search field\n");
+}
+
+/*********************************************************************/
+
+time_t time_for_past_day(int day) {
+ time_t t;
+ struct tm *tm;
+ int delta_day;
+ t = time(0);
+ tm = localtime(&t);
+ delta_day = (7 + tm->tm_wday - day) % 7 + 1;
+ return t - delta_day * 3600 * 24 + tm->tm_sec + 60 * tm->tm_min + 3600 * tm->tm_hour;
+}
+
+/*********************************************************************/
+
+int ignore_entry(const char *name) {
+ return
+ /* strcmp(name, ".") == 0 || */
+ /* strcmp(name, "..") == 0 || */
+ (name[0] == '.' && name[1] != '/');
+}
+
+int is_a_leading_from_line(char *s) {
+ return strncmp(s, "From ", 5) == 0 &&
+ regexec(&leading_from_line_regexp, s, 0, 0, 0) == 0;
+}
+
+int mbox_line_match_search(struct search_condition *condition,
+ int mbox_id, char *mbox_value) {
+
+ if(condition->field_id == ID_INTERVAL) {
+ if(mbox_id == ID_LEADING_LINE) {
+ char *c;
+ time_t t;
+ struct tm tm;
+
+ c = mbox_value;
+ while(*c && *c != ' ') c++; while(*c && *c == ' ') c++;
+ strptime(c, "%a %b %e %k:%M:%S %Y", &tm);
+ t = mktime(&tm);
+
+ return (t >= condition->interval_start &&
+ (condition->interval_stop == 0 ||
+ t <= condition->interval_stop));
+ } else {
+ return 0;
+ }
+ } else {
+ return
+ (
+
+ (condition->field_id == mbox_id)
+
+ ||
+
+ (condition->field_id == ID_PARTICIPANT && (mbox_id == ID_LEADING_LINE ||
+ mbox_id == ID_FROM ||
+ mbox_id == ID_TO))
+ ||
+
+ (condition->field_id == ID_FROM && mbox_id == ID_LEADING_LINE)
+
+ )
+ &&
+ regexec(&condition->regexp, mbox_value, 0, 0, 0) == 0;
+ }
+}
+
+void update_body_hits(char *mail_filename, int position_in_mail,
+ int nb_search_conditions, struct search_condition *search_conditions,
+ int nb_body_conditions,
+ int *hits) {
+ FILE *mail_file;
+ int header, n;
+ int last_mbox_line_was_empty;
+ char raw_mbox_line[BUFFER_SIZE];
+ int nb_body_hits;
+
+ nb_body_hits = 0;
+
+ header = 1;
+ mail_file = fopen(mail_filename, "r");
+
+ if(!mail_file) {
+ fprintf(stderr,
+ "mymail: Cannot open mbox '%s' for body scan.\n",
+ mail_filename);
+ exit(EXIT_FAILURE);
+ }
+
+ fseek(mail_file, position_in_mail, SEEK_SET);
+
+ if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
+ while(nb_body_hits < nb_body_conditions) {
+ last_mbox_line_was_empty = (raw_mbox_line[0] == '\n');
+
+ if(last_mbox_line_was_empty) { header = 0; }
+
+ if(!header) {
+ for(n = 0; n < nb_search_conditions; n++) {
+ if(search_conditions[n].field_id == ID_BODY && !hits[n]) {
+ hits[n] =
+ (regexec(&search_conditions[n].regexp, raw_mbox_line, 0, 0, 0) == 0);
+ if(hits[n]) {
+ nb_body_hits++;
+ }
+ }
+ }
+ }
+
+ if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
+ (last_mbox_line_was_empty &&
+ is_a_leading_from_line(raw_mbox_line)))
+ break;
+ }
+ }
+
+ fclose(mail_file);
+}
+
+void search_in_db(const char *db_filename,
+ int nb_search_conditions,
+ struct search_condition *search_conditions,
+ FILE *output_file) {
+
+ int hits[MAX_NB_SEARCH_CONDITIONS];
+ char raw_db_line[BUFFER_SIZE];
+ char raw_mbox_line[BUFFER_SIZE];
+ char current_mail_filename[PATH_MAX + 1];
+ unsigned long int current_position_in_mail;
+ char mbox_name[TOKEN_BUFFER_SIZE], *mbox_value;
+ int mbox_id;
+ int already_written, m, n;
+ int last_mbox_line_was_empty;
+ int nb_body_conditions, nb_fulfilled_body_conditions;
+ FILE *db_file;
+
+ if(!quiet) {
+ printf("Searching in '%s' ... ", db_filename);
+ fflush(stdout);
+ }
+
+ db_file = fopen(db_filename, "r");
+
+ if(!db_file) {
+ fprintf(stderr,
+ "mymail: Cannot open \"%s\" for reading: %s\n",
+ db_filename,
+ strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
+ if(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
+ if(strncmp(raw_db_line, MYMAIL_DB_MAGIC_TOKEN, strlen(MYMAIL_DB_MAGIC_TOKEN))) {
+ fprintf(stderr,
+ "mymail: Header line in '%s' does not match the mymail db format.\n",
+ db_filename);
+ exit(EXIT_FAILURE);
+ }
+ } else {
+ fprintf(stderr,
+ "mymail: Cannot read the header line in '%s'.\n",
+ db_filename);
+ exit(EXIT_FAILURE);
+ }
+
+ current_position_in_mail = 0;
+ already_written = 0;
+
+ for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
+
+ nb_body_conditions = 0;
+ for(n = 0; n < nb_search_conditions; n++) {
+ if(search_conditions[n].field_id == ID_BODY) {
+ nb_body_conditions++;
+ }
+ }
+
+ strcpy(current_mail_filename, "");
+
+ while(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
+ mbox_value = parse_token(mbox_name, TOKEN_BUFFER_SIZE, ' ', raw_db_line);
+
+ if(strcmp("mail", mbox_name) == 0) {
+ char position_in_file_string[TOKEN_BUFFER_SIZE];
+
+ if(current_mail_filename[0]) {
+
+ /* We first check all conditions but the body ones */
+
+ for(n = 0; n < nb_search_conditions &&
+ ((search_conditions[n].field_id == ID_BODY) ||
+ xor(hits[n], search_conditions[n].negation)); n++);
+
+ if(n == nb_search_conditions) {
+
+ /* all conditions but the body ones are fine, check the body
+ ones */
+
+ if(nb_body_conditions > 0) {
+ update_body_hits(current_mail_filename, current_position_in_mail,
+ nb_search_conditions, search_conditions,
+ nb_body_conditions,
+ hits);
+ }
+
+ nb_fulfilled_body_conditions = 0;
+
+ for(n = 0; n < nb_search_conditions; n++) {
+ if(search_conditions[n].field_id == ID_BODY &&
+ xor(hits[n], search_conditions[n].negation)) {
+ nb_fulfilled_body_conditions++;
+ }
+ }
+
+ if(nb_body_conditions == nb_fulfilled_body_conditions) {
+ FILE *mail_file;
+
+ mail_file = fopen(current_mail_filename, "r");
+
+ if(!mail_file) {
+ fprintf(stderr,
+ "mymail: Cannot open mbox '%s' for mail extraction.\n",
+ current_mail_filename);
+ exit(EXIT_FAILURE);
+ }
+
+ fseek(mail_file, current_position_in_mail, SEEK_SET);
+
+ if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
+ last_mbox_line_was_empty = 0;
+ fprintf(output_file, "%s", raw_mbox_line);
+ while(1) {
+ if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
+ (last_mbox_line_was_empty &&
+ is_a_leading_from_line(raw_mbox_line))
+ )
+ break;
+ last_mbox_line_was_empty = (raw_mbox_line[0] == '\n');
+ fprintf(output_file, "%s", raw_mbox_line);
+ }
+ }
+
+ fclose(mail_file);
+ }
+ }
+ }
+
+ for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
+
+ mbox_value = parse_token(position_in_file_string, TOKEN_BUFFER_SIZE, ' ', mbox_value);
+ mbox_value = parse_token(current_mail_filename, TOKEN_BUFFER_SIZE, ' ', mbox_value);
+ current_position_in_mail = atol(position_in_file_string);
+ remove_eof(current_mail_filename);
+ already_written = 0;
+ }
+
+ else {
+ mbox_id = -1;
+ for(m = 0; (m < MAX_ID) && mbox_id == -1; m++) {
+ if(strncmp(field_names[m], mbox_name, strlen(mbox_name)) == 0) {
+ mbox_id = m;
+ }
+ }
+ for(n = 0; n < nb_search_conditions; n++) {
+ hits[n] |= mbox_line_match_search(&search_conditions[n],
+ mbox_id, mbox_value);
+ }
+ }
+ }
+
+ fclose(db_file);
+
+ if(!quiet) {
+ printf("done.\n");
+ fflush(stdout);
+ }
+}
+
+void recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp,
+ int nb_search_conditions,
+ struct search_condition *search_conditions,
+ FILE *output_file) {
+ DIR *dir;
+ struct dirent *dir_e;
+ struct stat sb;
+ char subname[PATH_MAX + 1];
+
+ if(lstat(entry_name, &sb) != 0) {
+ fprintf(stderr,
+ "mymail: Cannot stat \"%s\": %s\n",
+ entry_name,
+ strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
+ dir = opendir(entry_name);
+
+ if(dir) {
+ while((dir_e = readdir(dir))) {
+ if(!ignore_entry(dir_e->d_name)) {
+ snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
+ recursive_search_in_db(subname, db_filename_regexp,
+ nb_search_conditions, search_conditions,
+ output_file);
+ }
+ }
+ closedir(dir);
+ }
+
+ else {
+ const char *s = entry_name, *filename = entry_name;
+ while(*s) { if(*s == '/') { filename = s+1; } s++; }
+
+ if(regexec(db_filename_regexp, filename, 0, 0, 0) == 0) {
+ search_in_db(entry_name, nb_search_conditions, search_conditions, output_file);
+ }
+ }
+}
+
+/*********************************************************************/
+
+void index_one_mbox_line(int nb_fields_to_parse, struct parsable_field *fields_to_parse,
+ char *raw_mbox_line, FILE *db_file) {
+ regmatch_t matches;
+ int f;
+ for(f = 0; f < nb_fields_to_parse; f++) {
+ if(regexec(&fields_to_parse[f].regexp, raw_mbox_line, 1, &matches, 0) == 0) {
+ fprintf(db_file, "%s %s\n",
+ field_names[fields_to_parse[f].id],
+ raw_mbox_line + matches.rm_eo);
+ }
+ }