X-Git-Url: https://www.fleuret.org/cgi-bin/gitweb/gitweb.cgi?p=mymail.git;a=blobdiff_plain;f=mymail.c;h=88c01f509a5da89808282ef097696831815c96c9;hp=c07e8a8cc63707b7ad63624c84ab50d6a5170852;hb=00434484dd0b89eb91cb6ea9be03eeaee1f97751;hpb=ecf45088cdb6b6a3b02416063fe2aa591d997bdd diff --git a/mymail.c b/mymail.c index c07e8a8..88c01f5 100644 --- a/mymail.c +++ b/mymail.c @@ -46,14 +46,16 @@ #include #define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file" -#define VERSION "0.9.6" +#define MYMAIL_VERSION "0.9.8" + +#define MYMAIL_DB_FORMAT_VERSION 1 #define MAX_NB_SEARCH_CONDITIONS 32 #define BUFFER_SIZE 65536 #define TOKEN_BUFFER_SIZE 1024 -#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$" +#define LEADING_FROM_LINE_REGEXP_STRING "^From .*\\(Mon\\|Tue\\|Wed\\|Thu\\|Fri\\|Sat\\|Sun\\) \\(Jan\\|Feb\\|Mar\\|Apr\\|May\\|Jun\\|Jul\\|Aug\\|Sep\\|Oct\\|Nov\\|Dec\\) [ 0123][0-9] [0-9][0-9]:[0-9][0-9]:[0-9][0-9] [0-9][0-9][0-9][0-9]\n$" /* Global variables! */ @@ -92,10 +94,10 @@ static char *field_keys[] = { /********************************************************************/ struct search_condition { - int field_id; + int db_key; + regex_t db_value_regexp; int negation; - regex_t regexp; - time_t interval_start, interval_stop; + time_t time_start, time_stop; }; /********************************************************************/ @@ -154,12 +156,15 @@ int xor(int a, int b) { const char *parse_token(char *token_buffer, size_t token_buffer_size, char separator, const char *string) { char *u = token_buffer; + while(*string == separator) { string++; } - while(u < token_buffer + token_buffer_size - 1 && *string && - *string != separator) { + + while(u < token_buffer + token_buffer_size - 1 && *string && *string != separator) { *(u++) = *(string++); } + while(*string == separator) { string++; } + *u = '\0'; return string; } @@ -196,7 +201,7 @@ FILE *safe_fopen(const char *path, const char *mode, const char *comment) { /*********************************************************************/ void print_version(FILE *out) { - fprintf(out, "mymail version %s (%s)\n", VERSION, UNAME); + fprintf(out, "mymail version %s (%s)\n", MYMAIL_VERSION, UNAME); } void print_usage(FILE *out) { @@ -224,7 +229,7 @@ void print_usage(FILE *out) { fprintf(out, " set the mbox filename pattern for recursive search\n"); fprintf(out, " -s , --search \n"); fprintf(out, " search for matching mails in the db file\n"); - fprintf(out, " -d , --db-file-generate \n"); + fprintf(out, " -d , --db-file-output \n"); fprintf(out, " set the db filename for indexing\n"); fprintf(out, " -i, --index\n"); fprintf(out, " index mails\n"); @@ -249,27 +254,27 @@ int is_a_leading_from_line(char *mbox_line) { regexec(&global_leading_from_line_regexp, mbox_line, 0, 0, 0) == 0; } -int mbox_line_match_search(struct search_condition *condition, - int mbox_id, const char *mbox_value) { +int db_line_match_search(struct search_condition *condition, + int db_key, const char *db_value) { return ( - (condition->field_id == mbox_id) + (condition->db_key == db_key) || - (condition->field_id == ID_PARTICIPANT && (mbox_id == ID_LEADING_LINE || - mbox_id == ID_FROM || - mbox_id == ID_TO)) + (condition->db_key == ID_PARTICIPANT && (db_key == ID_LEADING_LINE || + db_key == ID_FROM || + db_key == ID_TO)) || - (condition->field_id == ID_FROM && mbox_id == ID_LEADING_LINE) + (condition->db_key == ID_FROM && db_key == ID_LEADING_LINE) ) && - regexec(&condition->regexp, mbox_value, 0, 0, 0) == 0; + regexec(&condition->db_value_regexp, db_value, 0, 0, 0) == 0; } void update_body_hits(char *mail_filename, int position_in_mail, @@ -290,16 +295,13 @@ void update_body_hits(char *mail_filename, int position_in_mail, 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(raw_mbox_line[0] == '\n') { header = 0; } if(!header) { for(n = 0; n < nb_search_conditions; n++) { - if(search_conditions[n].field_id == ID_BODY && !hits[n]) { + if(search_conditions[n].db_key == ID_BODY && !hits[n]) { hits[n] = - (regexec(&search_conditions[n].regexp, raw_mbox_line, 0, 0, 0) == 0); + (regexec(&search_conditions[n].db_value_regexp, raw_mbox_line, 0, 0, 0) == 0); if(hits[n]) { nb_body_hits++; } @@ -317,10 +319,12 @@ void update_body_hits(char *mail_filename, int position_in_mail, } void extract_mail(const char *mail_filename, unsigned long int position_in_mail, - FILE *output_file) { + FILE *output_file) { char raw_mbox_line[BUFFER_SIZE]; FILE *mail_file; + /* printf("Extract\n"); */ + mail_file = safe_fopen(mail_filename, "r", "mbox for mail extraction"); fseek(mail_file, position_in_mail, SEEK_SET); @@ -328,8 +332,7 @@ void extract_mail(const char *mail_filename, unsigned long int position_in_mail, fprintf(output_file, "%s", raw_mbox_line); while(1) { if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) || - (is_a_leading_from_line(raw_mbox_line)) - ) + is_a_leading_from_line(raw_mbox_line)) break; fprintf(output_file, "%s", raw_mbox_line); } @@ -348,17 +351,17 @@ int check_full_mail_match(char *current_mail_filename, int n, nb_fulfilled_body_conditions; for(n = 0; n < nb_search_conditions; n++) { - if(search_conditions[n].field_id == ID_TIME_INTERVAL) { - hits[n] = (mail_time >= search_conditions[n].interval_start && - (search_conditions[n].interval_stop == 0 || - mail_time <= search_conditions[n].interval_stop)); + if(search_conditions[n].db_key == ID_TIME_INTERVAL) { + hits[n] = (mail_time >= search_conditions[n].time_start && + (search_conditions[n].time_stop == 0 || + mail_time <= search_conditions[n].time_stop)); } } /* We first check all conditions but the body ones */ for(n = 0; n < nb_search_conditions && - ((search_conditions[n].field_id == ID_BODY) || + ((search_conditions[n].db_key == ID_BODY) || xor(hits[n], search_conditions[n].negation)); n++); if(n == nb_search_conditions) { @@ -374,7 +377,7 @@ int check_full_mail_match(char *current_mail_filename, hits); for(n = 0; n < nb_search_conditions; n++) { - if(search_conditions[n].field_id == ID_BODY && + if(search_conditions[n].db_key == ID_BODY && xor(hits[n], search_conditions[n].negation)) { nb_fulfilled_body_conditions++; } @@ -389,20 +392,22 @@ int check_full_mail_match(char *current_mail_filename, /* We use the mail leading line time by default, and if we should and can, we update with the Date: field */ -void update_time(int mbox_id, const char *mbox_value, time_t *t) { +void update_time(int db_key, const char *db_value, time_t *t) { const char *c; struct tm tm; - if(mbox_id == ID_LEADING_LINE) { - c = mbox_value; + if(db_key == ID_LEADING_LINE) { + c = db_value; while(*c && *c != ' ') c++; while(*c && *c == ' ') c++; + /* printf("From %s", db_value); */ strptime(c, "%a %b %e %k:%M:%S %Y", &tm); *t = mktime(&tm); } else { if(!global_use_leading_time) { - if(mbox_id == ID_DATE) { - if(strptime(mbox_value, "%a, %d %b %Y %k:%M:%S", &tm) || - strptime(mbox_value, "%d %b %Y %k:%M:%S", &tm)) { + if(db_key == ID_DATE) { + if(strptime(db_value, "%a, %d %b %Y %k:%M:%S", &tm) || + strptime(db_value, "%d %b %Y %k:%M:%S", &tm)) { + /* printf("Date: %s", db_value); */ *t = mktime(&tm); } } @@ -418,11 +423,11 @@ int search_in_db(const char *db_filename, FILE *db_file; char raw_db_line[BUFFER_SIZE]; char current_mail_filename[PATH_MAX + 1]; - char mbox_key[TOKEN_BUFFER_SIZE]; + char db_key_string[TOKEN_BUFFER_SIZE]; char position_in_file_string[TOKEN_BUFFER_SIZE]; unsigned long int current_position_in_mail; - const char *mbox_value; - int mbox_id; + const char *db_value; + int db_key; int hits[MAX_NB_SEARCH_CONDITIONS]; int nb_body_conditions, need_time; int nb_extracted_mails; @@ -466,10 +471,10 @@ int search_in_db(const char *db_filename, mail_time = 0; for(n = 0; n < nb_search_conditions; n++) { - if(search_conditions[n].field_id == ID_BODY) { + if(search_conditions[n].db_key == ID_BODY) { nb_body_conditions++; } - else if(search_conditions[n].field_id == ID_TIME_INTERVAL) { + else if(search_conditions[n].db_key == ID_TIME_INTERVAL) { need_time = 1; } } @@ -477,9 +482,9 @@ int search_in_db(const char *db_filename, strcpy(current_mail_filename, ""); while(fgets(raw_db_line, BUFFER_SIZE, db_file)) { - mbox_value = parse_token(mbox_key, TOKEN_BUFFER_SIZE, ' ', raw_db_line); + db_value = parse_token(db_key_string, TOKEN_BUFFER_SIZE, ' ', raw_db_line); - if(strcmp("mail", mbox_key) == 0) { + if(strcmp("mail", db_key_string) == 0) { if(current_mail_filename[0]) { if(check_full_mail_match(current_mail_filename, mail_time, @@ -491,26 +496,26 @@ int search_in_db(const char *db_filename, } 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, PATH_MAX+1, '\n', mbox_value); + db_value = parse_token(position_in_file_string, TOKEN_BUFFER_SIZE, ' ', db_value); + db_value = parse_token(current_mail_filename, PATH_MAX+1, '\n', db_value); current_position_in_mail = atol(position_in_file_string); } else { - mbox_id = -1; - for(m = 0; (m < MAX_ID) && mbox_id == -1; m++) { - if(strncmp(field_keys[m], mbox_key, strlen(mbox_key)) == 0) { - mbox_id = m; + db_key = -1; + for(m = 0; (m < MAX_ID) && db_key == -1; m++) { + if(strncmp(field_keys[m], db_key_string, strlen(db_key_string)) == 0) { + db_key = m; } } for(n = 0; n < nb_search_conditions; n++) { - hits[n] |= mbox_line_match_search(&search_conditions[n], - mbox_id, mbox_value); + hits[n] |= db_line_match_search(&search_conditions[n], + db_key, db_value); } if(need_time) { - update_time(mbox_id, mbox_value, &mail_time); + update_time(db_key, db_value, &mail_time); } } } @@ -734,7 +739,7 @@ static struct option long_options[] = { { "version", no_argument, 0, 'v' }, { "quiet", no_argument, 0, 'q' }, { "use-leading-time", no_argument, 0, 't' }, - { "db-file-generate", 1, 0, 'd' }, + { "db-file-output", 1, 0, 'd' }, { "db-pattern", 1, 0, 'p' }, { "db-root", 1, 0, 'r' }, { "db-list", 1, 0, 'l' }, @@ -762,6 +767,7 @@ static struct time_criterion time_criteria[] = { { "48h", 0, 48, -1, -1 }, { "week", 0, 24 * 7, -1, -1 }, { "month", 0, 24 * 31, -1, -1 }, + { "trimester", 0, 24 * 92, -1, -1 }, { "year", 0, 24 * 365, -1, -1 }, { "yesterday", 1, -1, -1, -1 }, @@ -809,46 +815,47 @@ void init_condition(struct search_condition *condition, const char *full_string, condition->negation = 0; } - condition->field_id = -1; + condition->db_key = -1; - /* Recently */ + /* Time condition */ for(k = 0; k < sizeof(time_criteria) / sizeof(struct time_criterion); k++) { if(strcmp(time_criteria[k].label, search_field) == 0) { - condition->field_id = ID_TIME_INTERVAL; + condition->db_key = ID_TIME_INTERVAL; if(time_criteria[k].day_criterion) { - condition->interval_start = time_for_past_day(time_criteria[k].past_week_day); - condition->interval_stop = condition->interval_start + 3600 * 24; + condition->time_start = time_for_past_day(time_criteria[k].past_week_day); + condition->time_stop = condition->time_start + 3600 * 24; } else { - condition->interval_start = time(0) - 3600 * time_criteria[k].start_hour; + condition->time_start = time(0) - 3600 * time_criteria[k].start_hour; if(time_criteria[k].end_hour >= 0) { - condition->interval_stop = time(0) - 3600 * time_criteria[k].end_hour; + condition->time_stop = time(0) - 3600 * time_criteria[k].end_hour; } else { - condition->interval_stop = 0; + condition->time_stop = 0; } } + break; } } - if(condition->field_id == -1) { + if(condition->db_key == -1) { /* No time condition matched, look for the search fields */ - for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) { + for(m = 0; (m < MAX_ID) && condition->db_key == -1; m++) { if(strncmp(field_keys[m], search_field, strlen(search_field)) == 0) { - condition->field_id = m; + condition->db_key = m; } } /* None match, if there is a default search field, re-run the search with it */ - if(condition->field_id == -1) { + if(condition->db_key == -1) { if(default_search_field) { - for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) { + for(m = 0; (m < MAX_ID) && condition->db_key == -1; m++) { if(strncmp(field_keys[m], default_search_field, strlen(default_search_field)) == 0) { - condition->field_id = m; + condition->db_key = m; } } string = full_string; @@ -856,28 +863,28 @@ void init_condition(struct search_condition *condition, const char *full_string, } } - if(condition->field_id == -1) { + if(condition->db_key == -1) { fprintf(stderr, "mymail: Syntax error in field key \"%s\".\n", search_field); exit(EXIT_FAILURE); } - if(regcomp(&condition->regexp, + if(regcomp(&condition->db_value_regexp, string, REG_ICASE)) { fprintf(stderr, "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n", string, - field_keys[condition->field_id]); + field_keys[condition->db_key]); exit(EXIT_FAILURE); } } } void free_condition(struct search_condition *condition) { - if(condition->field_id != ID_TIME_INTERVAL) { - regfree(&condition->regexp); + if(condition->db_key != ID_TIME_INTERVAL) { + regfree(&condition->db_value_regexp); } } @@ -1075,7 +1082,7 @@ int main(int argc, char **argv) { } } - fprintf(db_file, "%s version_%s raw\n", MYMAIL_DB_MAGIC_TOKEN, VERSION); + fprintf(db_file, "%s version_%s format_%d raw\n", MYMAIL_DB_MAGIC_TOKEN, MYMAIL_VERSION, MYMAIL_DB_FORMAT_VERSION); while(optind < argc) { recursive_index_mbox(db_file,