X-Git-Url: https://www.fleuret.org/cgi-bin/gitweb/gitweb.cgi?p=mymail.git;a=blobdiff_plain;f=mymail.c;h=dd84dbe4e1bcde6506dda4ea17710cc3b7b9c94b;hp=c7d41d77d120d980ebede5558439ed06f5e6859e;hb=875659a04836f530f81214fd16db6afbf6e16c1f;hpb=ff95e19bacf5e4cd08c5cfba35b23a227ac34274 diff --git a/mymail.c b/mymail.c index c7d41d7..dd84dbe 100644 --- a/mymail.c +++ b/mymail.c @@ -43,32 +43,123 @@ #include #include #include +#include #define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file" -#define VERSION "0.1" +#define VERSION "0.9.1" + +#define MAX_NB_SEARCH_CONDITIONS 10 #define BUFFER_SIZE 65536 +#define TOKEN_BUFFER_SIZE 1024 + +regex_t leading_from_line_regexp; + +/* Global variables! */ + +int paranoid; +int quiet; +char *default_search_field; + +/********************************************************************/ + +enum { + ID_MAIL = 0, + ID_LEADING_LINE, + ID_FROM, + ID_TO, + ID_SUBJECT, + ID_DATE, + ID_PARTICIPANT, + ID_BODY, + ID_INTERVAL, + MAX_ID +}; + +static char *field_names[] = { + "mail", + "lead", + "from", + "to", + "subject", + "date", + "part", + "body", + "interval" +}; + +/********************************************************************/ + +struct search_condition { + int field_id; + int negation; + regex_t regexp; + time_t interval_start, interval_stop; +}; + +/********************************************************************/ struct parsable_field { - char *name; + int id; + int cflags; char *regexp_string; regex_t regexp; }; -char *db_filename; -char *db_root_path; +static struct parsable_field fields_to_parse[] = { + { + ID_LEADING_LINE, + 0, + "^From ", + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } + }, -int multi_db_files; -char *search_pattern; + { + ID_FROM, + REG_ICASE, + "^\\(from:\\|reply-to:\\|sender:\\|return-path:\\)", + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } + }, -int paranoid; -int action_index; + { + ID_TO, + REG_ICASE, + "^\\(to\\|cc\\|bcc\\): ", + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } + }, + + { + ID_SUBJECT, + REG_ICASE, + "^subject: ", + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } + }, -char *segment_next_field(char *current) { - while(*current && *current != ' ') current++; - *current = '\0'; current++; - while(*current && *current == ' ') current++; - return current; + { + ID_DATE, + REG_ICASE, + "^date: ", + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } + }, + +}; + +/********************************************************************/ + +int xor(int a, int b) { + return (a && !b) || (!a && b); +} + +char *parse_token(char *token_buffer, size_t token_buffer_size, + char separator, char *string) { + char *u = token_buffer; + while(u < token_buffer + token_buffer_size - 1 && *string && + *string != separator) { + *(u++) = *(string++); + } + while(*string == separator) string++; + *u = '\0'; + return string; } void remove_eof(char *c) { @@ -84,7 +175,7 @@ void *safe_malloc(size_t n) { void *p = malloc(n); if(!p && n != 0) { fprintf(stderr, - "mymail: can not allocate memory: %s\n", strerror(errno)); + "mymail: cannot allocate memory: %s\n", strerror(errno)); exit(EXIT_FAILURE); } return p; @@ -100,18 +191,42 @@ void print_usage(FILE *out) { print_version(out); fprintf(out, "Written by Francois Fleuret .\n"); fprintf(out, "\n"); - fprintf(out, "Usage: mymail [options] [ [ ...]]\n"); + fprintf(out, "Usage: mymail [options] [ [ ...]| [ ...]]\n"); fprintf(out, "\n"); 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-pattern \n"); + fprintf(out, " set the db filename pattern for recursive search\n"); + fprintf(out, " -r , --db-root \n"); + fprintf(out, " set the db root path for recursive search\n"); + fprintf(out, " -l , --db-list \n"); + fprintf(out, " set the semicolon-separated list of db files for search\n"); + fprintf(out, " -s , --search \n"); + fprintf(out, " search for matching mails in the db file\n"); + fprintf(out, " -d , --db-file \n"); + fprintf(out, " set the db filename for indexing\n"); fprintf(out, " -i, --index\n"); fprintf(out, " index mails\n"); - fprintf(out, " -d , --db-file \n"); - fprintf(out, " set the data-base file\n"); - fprintf(out, " -s , --search \n"); - fprintf(out, " search for matching mails in the data-base file\n"); + fprintf(out, " -o , --output \n"); + fprintf(out, " set the result file, use stdout if unset\n"); + fprintf(out, " -a , --default-search \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; } /*********************************************************************/ @@ -123,78 +238,286 @@ int ignore_entry(const char *name) { (name[0] == '.' && name[1] != '/'); } -void search_in_db(const char *search_name, const char *search_regexp_string, - FILE *db_file) { +int is_a_leading_from_line(int last_mbox_line_was_empty, char *mbox_line) { + return + + /* + + The mbox man page in qmail documentation states: + + > The reader should not attempt to take advantage of the fact + > that every From_ line (past the beginning of the file) is + > preceded by a blank line. + + */ + + /* last_mbox_line_was_empty && */ + strncmp(mbox_line, "From ", 5) == 0 && + regexec(&leading_from_line_regexp, mbox_line, 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) || + (is_a_leading_from_line(last_mbox_line_was_empty, raw_mbox_line))) + break; + } + } - char raw_line[BUFFER_SIZE]; + 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 *name, *value; - regex_t regexp; - int already_written; + 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); + } - if(regcomp(®exp, - search_regexp_string, - REG_ICASE)) { + db_file = fopen(db_filename, "r"); + + if(!db_file) { fprintf(stderr, - "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n", - search_regexp_string, - search_name); + "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; - while(fgets(raw_line, BUFFER_SIZE, db_file)) { - name = raw_line; - value = segment_next_field(raw_line); + for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; } - if(strcmp("mail", name) == 0) { - char *position_in_file_string = value; - char *mail_filename = segment_next_field(value); + 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) || + (is_a_leading_from_line(last_mbox_line_was_empty, 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); - strcpy(current_mail_filename, mail_filename); remove_eof(current_mail_filename); already_written = 0; } - else if(!already_written) { - if(strcmp(search_name, name) == 0 && regexec(®exp, value, 0, 0, 0) == 0) { - FILE *mail_file; - mail_file = fopen(current_mail_filename, "r"); - if(!mail_file) { - fprintf(stderr, "mymail: Can not open '%s'.\n", current_mail_filename); - exit(EXIT_FAILURE); - } - fseek(mail_file, current_position_in_mail, SEEK_SET); - if(fgets(raw_line, BUFFER_SIZE, mail_file)) { - printf("%s", raw_line); - while(fgets(raw_line, BUFFER_SIZE, mail_file) && - strncmp(raw_line, "From ", 5)) { - printf("%s", raw_line); - } + 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; } - fclose(mail_file); - already_written = 1; + } + for(n = 0; n < nb_search_conditions; n++) { + hits[n] |= mbox_line_match_search(&search_conditions[n], + mbox_id, mbox_value); } } } - regfree(®exp); + fclose(db_file); + + if(!quiet) { + printf("done.\n"); + fflush(stdout); + } } -void recursive_search_in_db(const char *entry_name, - const char *search_name, const char *search_regexp_string) { +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 raw_line[BUFFER_SIZE]; char subname[PATH_MAX + 1]; if(lstat(entry_name, &sb) != 0) { fprintf(stderr, - "mymail: Can not stat \"%s\": %s\n", + "mymail: Cannot stat \"%s\": %s\n", entry_name, strerror(errno)); exit(EXIT_FAILURE); @@ -206,42 +529,20 @@ void recursive_search_in_db(const char *entry_name, 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, search_name, search_regexp_string); + recursive_search_in_db(subname, db_filename_regexp, + nb_search_conditions, search_conditions, + output_file); } } closedir(dir); - } else { + } + + else { const char *s = entry_name, *filename = entry_name; while(*s) { if(*s == '/') { filename = s+1; } s++; } - if(strcmp(filename, db_filename) == 0) { - FILE *db_file = fopen(entry_name, "r"); - - if(!db_file) { - fprintf(stderr, - "mymail: Can not open \"%s\" for reading: %s\n", - db_filename, - strerror(errno)); - exit(EXIT_FAILURE); - } - - if(fgets(raw_line, BUFFER_SIZE, db_file)) { - if(strncmp(raw_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", - entry_name); - exit(EXIT_FAILURE); - } - } else { - fprintf(stderr, - "mymail: Can not read the header line in '%s'.\n", - entry_name); - exit(EXIT_FAILURE); - } - - search_in_db(search_name, search_regexp_string, db_file); - - fclose(db_file); + if(regexec(db_filename_regexp, filename, 0, 0, 0) == 0) { + search_in_db(entry_name, nb_search_conditions, search_conditions, output_file); } } } @@ -249,14 +550,14 @@ void recursive_search_in_db(const char *entry_name, /*********************************************************************/ void index_one_mbox_line(int nb_fields_to_parse, struct parsable_field *fields_to_parse, - char *raw_line, FILE *db_file) { + 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_line, 1, &matches, 0) == 0) { + if(regexec(&fields_to_parse[f].regexp, raw_mbox_line, 1, &matches, 0) == 0) { fprintf(db_file, "%s %s\n", - fields_to_parse[f].name, - raw_line + matches.rm_eo); + field_names[fields_to_parse[f].id], + raw_mbox_line + matches.rm_eo); } } } @@ -264,16 +565,16 @@ void index_one_mbox_line(int nb_fields_to_parse, struct parsable_field *fields_t void index_mbox(const char *mbox_filename, int nb_fields_to_parse, struct parsable_field *fields_to_parse, FILE *db_file) { - char raw_line[BUFFER_SIZE], full_line[BUFFER_SIZE]; + char raw_mbox_line[BUFFER_SIZE], full_line[BUFFER_SIZE]; char *end_of_full_line; FILE *file; - int in_header, new_header; + int in_header, new_header, last_mbox_line_was_empty; unsigned long int position_in_file; file = fopen(mbox_filename, "r"); if(!file) { - fprintf(stderr, "mymail: Can not open '%s'.\n", mbox_filename); + fprintf(stderr, "mymail: Cannot open '%s'.\n", mbox_filename); if(paranoid) { exit(EXIT_FAILURE); } return; } @@ -284,30 +585,33 @@ void index_mbox(const char *mbox_filename, position_in_file = 0; end_of_full_line = 0; full_line[0] = '\0'; + last_mbox_line_was_empty = 1; - while(fgets(raw_line, BUFFER_SIZE, file)) { - if(strncmp(raw_line, "From ", 5) == 0) { + while(fgets(raw_mbox_line, BUFFER_SIZE, file)) { + if(is_a_leading_from_line(last_mbox_line_was_empty, raw_mbox_line)) { if(in_header) { fprintf(stderr, "Got a ^\"From \" in the header in %s:%lu.\n", mbox_filename, position_in_file); - fprintf(stderr, "%s", raw_line); + fprintf(stderr, "%s", raw_mbox_line); if(paranoid) { exit(EXIT_FAILURE); } } in_header = 1; new_header = 1; - } else if(strncmp(raw_line, "\n", 1) == 0) { + } else if(raw_mbox_line[0] == '\n') { if(in_header) { in_header = 0; } } + last_mbox_line_was_empty = (raw_mbox_line[0] == '\n'); + if(in_header) { if(new_header) { fprintf(db_file, "mail %lu %s\n", position_in_file, mbox_filename); new_header = 0; } - if(raw_line[0] == ' ' || raw_line[0] == '\t') { - char *start = raw_line; + if(raw_mbox_line[0] == ' ' || raw_mbox_line[0] == '\t') { + char *start = raw_mbox_line; while(*start == ' ' || *start == '\t') start++; *(end_of_full_line++) = ' '; strcpy(end_of_full_line, start); @@ -318,20 +622,22 @@ void index_mbox(const char *mbox_filename, } else { - /* if(!((raw_line[0] >= 'a' && raw_line[0] <= 'z') || */ - /* (raw_line[0] >= 'A' && raw_line[0] <= 'Z'))) { */ - /* fprintf(stderr, */ - /* "Header line syntax error %s:%lu.\n", */ - /* mbox_filename, position_in_file); */ - /* fprintf(stderr, "%s", raw_line); */ - /* } */ + /* + if(!((raw_mbox_line[0] >= 'a' && raw_mbox_line[0] <= 'z') || + (raw_mbox_line[0] >= 'A' && raw_mbox_line[0] <= 'Z'))) { + fprintf(stderr, + "Header line syntax error %s:%lu.\n", + mbox_filename, position_in_file); + fprintf(stderr, "%s", raw_mbox_line); + } + */ if(full_line[0]) { index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file); } end_of_full_line = full_line; - strcpy(end_of_full_line, raw_line); + strcpy(end_of_full_line, raw_mbox_line); while(*end_of_full_line && *end_of_full_line != '\n') { end_of_full_line++; } @@ -340,7 +646,7 @@ void index_mbox(const char *mbox_filename, } - position_in_file += strlen(raw_line); + position_in_file += strlen(raw_mbox_line); } fclose(file); @@ -356,7 +662,7 @@ void recursive_index_mbox(FILE *db_file, if(lstat(entry_name, &sb) != 0) { fprintf(stderr, - "mymail: Can not stat \"%s\": %s\n", + "mymail: Cannot stat \"%s\": %s\n", entry_name, strerror(errno)); exit(EXIT_FAILURE); @@ -388,51 +694,204 @@ enum { static struct option long_options[] = { { "help", no_argument, 0, 'h' }, { "version", no_argument, 0, 'v' }, + { "quiet", no_argument, 0, 'q' }, { "db-file", 1, 0, 'd' }, - { "db-root", 1, 0, 'p' }, - { "search-pattern", 1, 0, 's' }, + { "db-pattern", 1, 0, 'p' }, + { "db-root", 1, 0, 'r' }, + { "db-list", 1, 0, 'l' }, + { "search", 1, 0, 's' }, { "index", 0, 0, 'i' }, - { "multi-db-files", 0, 0, 'm' }, + { "output", 1, 0, 'o' }, + { "default-search", 1, 0, 'a' }, { 0, 0, 0, 0 } }; -static struct parsable_field fields_to_parse[] = { - { - "from", - "^\\([Ff][Rr][Oo][Mm]:\\|From\\) *", - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } - }, +/*********************************************************************/ - { - "dest", - "^\\([Tt][Oo]\\|[Cc][Cc]\\|[Bb][Cc][Cc]\\): *", - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } - }, +void init_condition(struct search_condition *condition, char *full_string) { + char full_search_field[TOKEN_BUFFER_SIZE], *search_field; + int m; + char *string; - { - "subj", - "^[Ss][Uu][Bb][Jj][Ee][Cc][Tt]: *", - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } - }, + string = parse_token(full_search_field, TOKEN_BUFFER_SIZE, ' ', full_string); + search_field = full_search_field; -}; + if(search_field[0] == '!') { + search_field++; + condition->negation = 1; + } else { + condition->negation = 0; + } + + /* Recently */ + + if(strcmp(search_field, "8h") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time(0) - 3600 * 8; + condition->interval_stop = 0; + } + + else if(strcmp(search_field, "week") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time(0) - 3600 * 24 * 7; + condition->interval_stop = 0; + } + + else if(strcmp(search_field, "month") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time(0) - 3600 * 24 * 31; + condition->interval_stop = 0; + } + + else if(strcmp(search_field, "24h") == 0 || + strcmp(search_field, "today") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time(0) - 3600 * 24; + condition->interval_stop = 0; + } + + /* Yesterday */ + + else if(strcmp(search_field, "yesterday") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time(0) - 2 * 3600 * 24; + condition->interval_stop = condition->interval_start + 3600 * 24; + } + + /* Week days */ + + else if(strcmp(search_field, "monday") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time_for_past_day(1); + condition->interval_stop = condition->interval_start + 3600 * 24; + } + + else if(strcmp(search_field, "tuesday") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time_for_past_day(2); + condition->interval_stop = condition->interval_start + 3600 * 24; + } + + else if(strcmp(search_field, "wednesday") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time_for_past_day(3); + condition->interval_stop = condition->interval_start + 3600 * 24; + } + + else if(strcmp(search_field, "thursday") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time_for_past_day(4); + condition->interval_stop = condition->interval_start + 3600 * 24; + } + + else if(strcmp(search_field, "friday") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time_for_past_day(5); + condition->interval_stop = condition->interval_start + 3600 * 24; + } + + else if(strcmp(search_field, "saturday") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time_for_past_day(6); + condition->interval_stop = condition->interval_start + 3600 * 24; + } + + else if(strcmp(search_field, "sunday") == 0) { + condition->field_id = ID_INTERVAL; + condition->interval_start = time_for_past_day(7); + condition->interval_stop = condition->interval_start + 3600 * 24; + } + + else { + + /* header-related conditions */ + + condition->field_id = -1; + + for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) { + if(strncmp(field_names[m], search_field, strlen(search_field)) == 0) { + condition->field_id = m; + } + } + + if(condition->field_id == -1) { + if(default_search_field) { + for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) { + if(strncmp(field_names[m], + default_search_field, strlen(default_search_field)) == 0) { + condition->field_id = m; + } + } + string = full_string; + } + } + + if(condition->field_id == -1) { + fprintf(stderr, + "mymail: Syntax error in field name \"%s\".\n", + search_field); + exit(EXIT_FAILURE); + } + + if(regcomp(&condition->regexp, + string, + REG_ICASE)) { + fprintf(stderr, + "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n", + string, + field_names[condition->field_id]); + exit(EXIT_FAILURE); + } + } +} + +void free_condition(struct search_condition *condition) { + if(condition->field_id != ID_INTERVAL) { + regfree(&condition->regexp); + } +} + +/*********************************************************************/ +/*********************************************************************/ +/*********************************************************************/ int main(int argc, char **argv) { + char *db_filename; + char *db_filename_regexp_string; + char *db_root_path; + char *db_filename_list; + char output_filename[PATH_MAX + 1]; + int action_index; int error = 0, show_help = 0; const int nb_fields_to_parse = sizeof(fields_to_parse) / sizeof(struct parsable_field); char c; - int f; + int f, n; + int nb_search_conditions; + FILE *output_file; + struct search_condition search_conditions[MAX_NB_SEARCH_CONDITIONS]; + + if(regcomp(&leading_from_line_regexp, + "^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$", + 0)) { + fprintf(stderr, + "mymail: Cannot compile leading \"from\" line regexp. That is strange.\n"); + exit(EXIT_FAILURE); + } paranoid = 0; action_index = 0; - search_pattern = 0; db_filename = 0; + db_filename_regexp_string = 0; db_root_path = 0; - multi_db_files = 0; + db_filename_list = 0; + quiet = 0; + default_search_field = 0; setlocale(LC_ALL, ""); - while ((c = getopt_long(argc, argv, "hvimp:s:d:p:", + nb_search_conditions = 0; + + while ((c = getopt_long(argc, argv, "hvqip:s:d:r:l:o:a:", long_options, NULL)) != -1) { switch(c) { @@ -445,12 +904,12 @@ int main(int argc, char **argv) { print_version(stdout); break; - case 'i': - action_index = 1; + case 'q': + quiet = 1; break; - case 'm': - multi_db_files = 1; + case 'i': + action_index = 1; break; case 'd': @@ -458,15 +917,32 @@ int main(int argc, char **argv) { break; case 'p': + db_filename_regexp_string = strdup(optarg); + break; + + case 'o': + strncpy(output_filename, optarg, PATH_MAX); + break; + + case 'r': db_root_path = strdup(optarg); break; + case 'l': + db_filename_list = strdup(optarg); + break; + case 's': - if(search_pattern) { - fprintf(stderr, "mymail: Search pattern already defined.\n"); + if(nb_search_conditions == MAX_NB_SEARCH_CONDITIONS) { + fprintf(stderr, "mymail: Too many search patterns.\n"); exit(EXIT_FAILURE); } - search_pattern = strdup(optarg); + init_condition(&search_conditions[nb_search_conditions], optarg); + nb_search_conditions++; + break; + + case 'a': + default_search_field = optarg; break; default: @@ -479,28 +955,51 @@ int main(int argc, char **argv) { char *default_db_filename = getenv("MYMAIL_DB_FILE"); if(!default_db_filename) { - if(multi_db_files) { - default_db_filename = "mymail.db"; - } else { - default_db_filename = "/tmp/mymail.db"; - } + default_db_filename = "mymail.db"; } db_filename = strdup(default_db_filename); } + if(!db_filename_regexp_string) { + char *default_db_filename_regexp_string = getenv("MYMAIL_DB_PATTERN"); + + if(!default_db_filename_regexp_string) { + default_db_filename_regexp_string = "^mymail.db$"; + } + + db_filename_regexp_string = strdup(default_db_filename_regexp_string); + } + if(!db_root_path) { char *default_db_root_path = getenv("MYMAIL_DB_ROOT"); - if(!default_db_root_path) { - if(multi_db_files) { - default_db_root_path = "mymail.db"; - } else { - default_db_root_path = "/tmp/mymail.db"; - } + if(default_db_root_path) { + db_root_path = strdup(default_db_root_path); } + } + + if(!db_filename_list) { + char *default_db_filename_list = getenv("MYMAIL_DB_LIST"); - db_root_path = strdup(default_db_root_path); + if(default_db_filename_list) { + db_filename_list = strdup(default_db_filename_list); + } + } + + if(output_filename[0]) { + output_file = fopen(output_filename, "w"); + + if(!output_file) { + fprintf(stderr, + "mymail: Cannot open result file \"%s\" for writing: %s\n", + output_filename, + strerror(errno)); + exit(EXIT_FAILURE); + } + } else { + output_file = stdout; + quiet = 1; } if(error) { @@ -520,7 +1019,7 @@ int main(int argc, char **argv) { if(!db_file) { fprintf(stderr, - "mymail: Can not open \"%s\" for writing: %s\n", + "mymail: Cannot open \"%s\" for writing: %s\n", db_filename, strerror(errno)); exit(EXIT_FAILURE); @@ -529,16 +1028,16 @@ int main(int argc, char **argv) { for(f = 0; f < nb_fields_to_parse; f++) { if(regcomp(&fields_to_parse[f].regexp, fields_to_parse[f].regexp_string, - REG_ICASE)) { + fields_to_parse[f].cflags)) { fprintf(stderr, "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n", fields_to_parse[f].regexp_string, - fields_to_parse[f].name); + field_names[fields_to_parse[f].id]); exit(EXIT_FAILURE); } } - fprintf(db_file, "%s version_%s raw version\n", MYMAIL_DB_MAGIC_TOKEN, VERSION); + fprintf(db_file, "%s version_%s raw\n", MYMAIL_DB_MAGIC_TOKEN, VERSION); while(optind < argc) { recursive_index_mbox(db_file, @@ -547,6 +1046,7 @@ int main(int argc, char **argv) { optind++; } + fflush(db_file); fclose(db_file); for(f = 0; f < nb_fields_to_parse; f++) { @@ -556,26 +1056,69 @@ int main(int argc, char **argv) { else { - if(search_pattern) { - char *search_name; - char *search_regexp_string; - search_name = search_pattern; - search_regexp_string = segment_next_field(search_pattern); + if(nb_search_conditions > 0) { - if(!*search_regexp_string) { - fprintf(stderr, - "Syntax error in the search pattern.\n"); - exit(EXIT_FAILURE); + /* Recursive search if db_root_path is set */ + + if(db_root_path) { + regex_t db_filename_regexp; + if(regcomp(&db_filename_regexp, + db_filename_regexp_string, + 0)) { + fprintf(stderr, + "mymail: Syntax error in regexp \"%s\".\n", + db_filename_regexp_string); + exit(EXIT_FAILURE); + } + + recursive_search_in_db(db_root_path, &db_filename_regexp, + nb_search_conditions, search_conditions, + output_file); + + regfree(&db_filename_regexp); } - recursive_search_in_db(db_root_path, search_name, search_regexp_string); + /* Search in all db files listed in db_filename_list */ + + if(db_filename_list) { + char db_filename[PATH_MAX + 1]; + char *s; + + s = db_filename_list; - free(search_pattern); + while(*s) { + s = parse_token(db_filename, PATH_MAX + 1, ';', s); + + if(db_filename[0]) { + search_in_db(db_filename, nb_search_conditions, search_conditions, output_file); + } + } + } + + /* Search in all db files listed in the command arguments */ + + while(optind < argc) { + search_in_db(argv[optind], nb_search_conditions, search_conditions, output_file); + optind++; + } } } + for(n = 0; n < nb_search_conditions; n++) { + free_condition(&search_conditions[n]); + } + + if(output_file != stdout) { + fflush(output_file); + fclose(output_file); + } + free(db_filename); + free(db_filename_regexp_string); free(db_root_path); + free(db_filename_list); + + regfree(&leading_from_line_regexp); exit(EXIT_SUCCESS); }