From: Francois Fleuret Date: Sun, 3 Feb 2013 18:06:18 +0000 (+0100) Subject: Added the body condition. X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?p=mymail.git;a=commitdiff_plain;h=b1fd04286d1c0f0764b5cfb20769bf9d29dde5d1 Added the body condition. --- diff --git a/mymail.1 b/mymail.1 index 877a9f7..7d6c963 100644 --- a/mymail.1 +++ b/mymail.1 @@ -82,6 +82,11 @@ or fields From:, Reply-To:, To:, Cc:, or Bcc: matches the regexp. \fBs \fR (subject) selects mails whose field Subject: matches the regexp. .TP \fBd \fR (date) selects mails whose field Date: matches the regexp. +.TP +\fBb \fR (body) selects mails whose body matches the +regexp. This needs to read fully the original mboxes and can take a +long time. Other conditions have to be fulfilled before the body is +read though. .SH "EXAMPLES" @@ -108,7 +113,8 @@ mymail --db-pattern '\\.db$' --db-root /tmp/mymail --search 'p bob.something' -- .SH "BUGS" -None known, probably many. +The search in the mail bodies does not decode mimencoding mails, hence +will not catch patterns in encoded text. .SH "AUTHOR" diff --git a/mymail.c b/mymail.c index 21ad5ea..48f25fa 100644 --- a/mymail.c +++ b/mymail.c @@ -47,7 +47,7 @@ #define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file" #define VERSION "0.9" -#define MAX_NB_SEARCH_REQUESTS 10 +#define MAX_NB_SEARCH_CONDITIONS 10 #define BUFFER_SIZE 65536 @@ -68,6 +68,7 @@ enum { ID_SUBJECT, ID_DATE, ID_PARTICIPANT, + ID_BODY, MAX_ID }; @@ -77,12 +78,13 @@ static char *field_names[] = { "to", "subject", "date", - "part" + "part", + "body" }; /********************************************************************/ -struct search_request { +struct search_condition { int field_id; int negation; regex_t regexp; @@ -194,19 +196,19 @@ int ignore_entry(const char *name) { (name[0] == '.' && name[1] != '/'); } -int mbox_line_match_search(struct search_request *request, +int mbox_line_match_search(struct search_condition *condition, int mbox_id, char *mbox_value) { return - (request->field_id == mbox_id || - (request->field_id == ID_PARTICIPANT && (mbox_id == ID_FROM || mbox_id == ID_TO))) + (condition->field_id == mbox_id || + (condition->field_id == ID_PARTICIPANT && (mbox_id == ID_FROM || mbox_id == ID_TO))) && - regexec(&request->regexp, mbox_value, 0, 0, 0) == 0; + regexec(&condition->regexp, mbox_value, 0, 0, 0) == 0; } -void search_in_db(int nb_search_requests, - struct search_request *search_requests, +void search_in_db(int nb_search_conditions, + struct search_condition *search_conditions, FILE *db_file) { - int hits[MAX_NB_SEARCH_REQUESTS]; + 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]; @@ -215,11 +217,19 @@ void search_in_db(int nb_search_requests, int mbox_id; int already_written, m, n; int last_mbox_line_was_empty; + int nb_body_conditions, nb_fulfilled_body_conditions; current_position_in_mail = 0; already_written = 0; - for(n = 0; n < nb_search_requests; n++) { hits[n] = 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++; + } + } while(fgets(raw_db_line, BUFFER_SIZE, db_file)) { mbox_name = raw_db_line; @@ -229,39 +239,88 @@ void search_in_db(int nb_search_requests, char *position_in_file_string; char *mail_filename; - for(n = 0; n < nb_search_requests && xor(hits[n], search_requests[n].negation); n++); + /* We first check all conditions but the body ones */ - /* for(n = 0; n < nb_search_requests && */ - /* ((hits[n] && !search_requests[n].negation) || */ - /* (!hits[n] && search_requests[n].negation)); n++); */ + 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_requests) { - FILE *mail_file; + if(n == nb_search_conditions) { - mail_file = fopen(current_mail_filename, "r"); + /* all conditions but the body ones are fine, check the body + ones */ - if(!mail_file) { - fprintf(stderr, "mymail: Cannot open mbox '%s'.\n", current_mail_filename); - exit(EXIT_FAILURE); + nb_fulfilled_body_conditions = 0; + + if(nb_body_conditions > 0) { + FILE *mail_file; + int header; + + header = 1; + mail_file = fopen(current_mail_filename, "r"); + + if(!mail_file) { + fprintf(stderr, "mymail: Cannot open mbox '%s'.\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)) { + while(nb_fulfilled_body_conditions < 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]) { + fprintf(stderr, "** %d %s", header, raw_mbox_line); + nb_fulfilled_body_conditions++; + } + } + } + } + + if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) || + (last_mbox_line_was_empty && strncmp(raw_mbox_line, "From ", 5) == 0)) break; + } + } + + fclose(mail_file); } - fseek(mail_file, current_position_in_mail, SEEK_SET); + 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'.\n", current_mail_filename); + exit(EXIT_FAILURE); + } - if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) { - last_mbox_line_was_empty = 1; - printf("%s", raw_mbox_line); - while(1) { - if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) || - (last_mbox_line_was_empty && strncmp(raw_mbox_line, "From ", 5) == 0)) break; - last_mbox_line_was_empty = (raw_mbox_line[0] == '\n'); + fseek(mail_file, current_position_in_mail, SEEK_SET); + + if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) { + last_mbox_line_was_empty = 1; printf("%s", raw_mbox_line); + while(1) { + if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) || + (last_mbox_line_was_empty && strncmp(raw_mbox_line, "From ", 5) == 0)) break; + last_mbox_line_was_empty = (raw_mbox_line[0] == '\n'); + printf("%s", raw_mbox_line); + } } - } - fclose(mail_file); + fclose(mail_file); + } } - for(n = 0; n < nb_search_requests; n++) { hits[n] = 0; } + for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; } position_in_file_string = mbox_value; mail_filename = segment_next_field(mbox_value); @@ -279,8 +338,8 @@ void search_in_db(int nb_search_requests, mbox_id = m; } } - for(n = 0; n < nb_search_requests; n++) { - hits[n] |= mbox_line_match_search(&search_requests[n], + for(n = 0; n < nb_search_conditions; n++) { + hits[n] |= mbox_line_match_search(&search_conditions[n], mbox_id, mbox_value); } } @@ -288,7 +347,7 @@ void search_in_db(int nb_search_requests, } void recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp, - int nb_search_requests, struct search_request *search_requests) { + int nb_search_conditions, struct search_condition *search_conditions) { DIR *dir; struct dirent *dir_e; struct stat sb; @@ -310,7 +369,7 @@ void recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp, 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_requests, search_requests); + nb_search_conditions, search_conditions); } } closedir(dir); @@ -345,7 +404,7 @@ void recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp, exit(EXIT_FAILURE); } - search_in_db(nb_search_requests, search_requests, db_file); + search_in_db(nb_search_conditions, search_conditions, db_file); fclose(db_file); } @@ -515,8 +574,8 @@ int main(int argc, char **argv) { const int nb_fields_to_parse = sizeof(fields_to_parse) / sizeof(struct parsable_field); char c; int f; - int nb_search_requests; - char *search_request_strings[MAX_NB_SEARCH_REQUESTS]; + int nb_search_conditions; + char *search_condition_strings[MAX_NB_SEARCH_CONDITIONS]; /* for(f = 0; f < argc; f++) { */ /* printf("arg %d \"%s\"\n", f, argv[f]); */ @@ -530,7 +589,7 @@ int main(int argc, char **argv) { setlocale(LC_ALL, ""); - nb_search_requests = 0; + nb_search_conditions = 0; while ((c = getopt_long(argc, argv, "hvip:s:d:r:l:", long_options, NULL)) != -1) { @@ -566,11 +625,11 @@ int main(int argc, char **argv) { break; case 's': - if(nb_search_requests == MAX_NB_SEARCH_REQUESTS) { + if(nb_search_conditions == MAX_NB_SEARCH_CONDITIONS) { fprintf(stderr, "mymail: Too many search patterns.\n"); exit(EXIT_FAILURE); } - search_request_strings[nb_search_requests++] = strdup(optarg); + search_condition_strings[nb_search_conditions++] = strdup(optarg); break; default: @@ -668,43 +727,43 @@ int main(int argc, char **argv) { else { - if(nb_search_requests > 0) { - struct search_request search_requests[MAX_NB_SEARCH_REQUESTS]; + if(nb_search_conditions > 0) { + struct search_condition search_conditions[MAX_NB_SEARCH_CONDITIONS]; char *search_field, *search_regexp_string; int m, n; - for(n = 0; n < nb_search_requests; n++) { - search_field = search_request_strings[n]; - search_regexp_string = segment_next_field(search_request_strings[n]); + for(n = 0; n < nb_search_conditions; n++) { + search_field = search_condition_strings[n]; + search_regexp_string = segment_next_field(search_condition_strings[n]); if(search_field[0] == '!') { search_field++; - search_requests[n].negation = 1; + search_conditions[n].negation = 1; } else { - search_requests[n].negation = 0; + search_conditions[n].negation = 0; } - search_requests[n].field_id = -1; - for(m = 0; (m < MAX_ID) && search_requests[n].field_id == -1; m++) { + search_conditions[n].field_id = -1; + for(m = 0; (m < MAX_ID) && search_conditions[n].field_id == -1; m++) { if(strncmp(field_names[m], search_field, strlen(search_field)) == 0) { - search_requests[n].field_id = m; + search_conditions[n].field_id = m; } } - if(search_requests[n].field_id == -1) { + if(search_conditions[n].field_id == -1) { fprintf(stderr, "mymail: Syntax error in field name \"%s\".\n", search_field); exit(EXIT_FAILURE); } - if(regcomp(&search_requests[n].regexp, + if(regcomp(&search_conditions[n].regexp, search_regexp_string, REG_ICASE)) { fprintf(stderr, "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n", search_regexp_string, - field_names[search_requests[n].field_id]); + field_names[search_conditions[n].field_id]); exit(EXIT_FAILURE); } } @@ -723,7 +782,7 @@ int main(int argc, char **argv) { } recursive_search_in_db(db_root_path, &db_filename_regexp, - nb_search_requests, search_requests); + nb_search_conditions, search_conditions); regfree(&db_filename_regexp); } @@ -754,7 +813,7 @@ int main(int argc, char **argv) { exit(EXIT_FAILURE); } - search_in_db(nb_search_requests, search_requests, db_file); + search_in_db(nb_search_conditions, search_conditions, db_file); fclose(db_file); } @@ -774,15 +833,15 @@ int main(int argc, char **argv) { exit(EXIT_FAILURE); } - search_in_db(nb_search_requests, search_requests, db_file); + search_in_db(nb_search_conditions, search_conditions, db_file); fclose(db_file); optind++; } - for(n = 0; n < nb_search_requests; n++) { - regfree(&search_requests[n].regexp); - free(search_request_strings[n]); + for(n = 0; n < nb_search_conditions; n++) { + regfree(&search_conditions[n].regexp); + free(search_condition_strings[n]); } } }