X-Git-Url: https://www.fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=mymail.c;h=528730107f73b5d9def4a7bdf85c3ca15218a2ff;hb=5b468174fa4d6629faf3069e254a17c72bd18545;hp=dd84dbe4e1bcde6506dda4ea17710cc3b7b9c94b;hpb=875659a04836f530f81214fd16db6afbf6e16c1f;p=mymail.git diff --git a/mymail.c b/mymail.c index dd84dbe..5287301 100644 --- a/mymail.c +++ b/mymail.c @@ -25,7 +25,7 @@ directories containing mbox files, and create a db file containing for each mail a list of fields computed from the header, or (2) read such a db file and get all the mails matching regexp-defined - conditions on the fields. + conditions on the fields, to create a resulting mbox file. It is low-tech, simple, light and fast. @@ -46,13 +46,15 @@ #include #define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file" -#define VERSION "0.9.1" +#define VERSION "0.9.2" -#define MAX_NB_SEARCH_CONDITIONS 10 +#define MAX_NB_SEARCH_CONDITIONS 32 #define BUFFER_SIZE 65536 #define TOKEN_BUFFER_SIZE 1024 +#define 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$" + regex_t leading_from_line_regexp; /* Global variables! */ @@ -60,6 +62,7 @@ regex_t leading_from_line_regexp; int paranoid; int quiet; char *default_search_field; +int ignore_dot_files; /********************************************************************/ @@ -117,7 +120,7 @@ static struct parsable_field fields_to_parse[] = { { ID_FROM, REG_ICASE, - "^\\(from:\\|reply-to:\\|sender:\\|return-path:\\)", + "^\\(from\\|reply-to\\|sender\\|return-path\\): ", { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, @@ -162,11 +165,6 @@ char *parse_token(char *token_buffer, size_t token_buffer_size, return string; } -void remove_eof(char *c) { - while(*c && *c != '\n' && *c != '\r') c++; - *c = '\0'; -} - /********************************************************************/ /* malloc with error checking. */ @@ -205,9 +203,11 @@ void print_usage(FILE *out) { 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, " -m , --mbox-pattern \n"); + 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 \n"); + fprintf(out, " -d , --db-file-generate \n"); fprintf(out, " set the db filename for indexing\n"); fprintf(out, " -i, --index\n"); fprintf(out, " index mails\n"); @@ -233,12 +233,12 @@ time_t time_for_past_day(int day) { int ignore_entry(const char *name) { return - /* strcmp(name, ".") == 0 || */ - /* strcmp(name, "..") == 0 || */ - (name[0] == '.' && name[1] != '/'); + strcmp(name, ".") == 0 || + strcmp(name, "..") == 0 || + (ignore_dot_files && name[0] == '.' && name[1] != '/'); } -int is_a_leading_from_line(int last_mbox_line_was_empty, char *mbox_line) { +int is_a_leading_from_line(char *mbox_line) { return /* @@ -251,7 +251,6 @@ int is_a_leading_from_line(int last_mbox_line_was_empty, char *mbox_line) { */ - /* last_mbox_line_was_empty && */ strncmp(mbox_line, "From ", 5) == 0 && regexec(&leading_from_line_regexp, mbox_line, 0, 0, 0) == 0; } @@ -292,7 +291,9 @@ int mbox_line_match_search(struct search_condition *condition, (condition->field_id == ID_FROM && mbox_id == ID_LEADING_LINE) ) + && + regexec(&condition->regexp, mbox_value, 0, 0, 0) == 0; } } @@ -303,7 +304,6 @@ void update_body_hits(char *mail_filename, int position_in_mail, int *hits) { FILE *mail_file; int header, n; - int last_mbox_line_was_empty; char raw_mbox_line[BUFFER_SIZE]; int nb_body_hits; @@ -323,9 +323,10 @@ 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'); + /* last_mbox_line_was_empty = (raw_mbox_line[0] == '\n'); */ + /* if(last_mbox_line_was_empty) { header = 0; } */ - 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++) { @@ -340,30 +341,61 @@ void update_body_hits(char *mail_filename, int position_in_mail, } if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) || - (is_a_leading_from_line(last_mbox_line_was_empty, raw_mbox_line))) + (is_a_leading_from_line(raw_mbox_line))) + break; + } + } + + fclose(mail_file); +} + +void extract_mail(const char *mail_filename, unsigned long int position_in_mail, + FILE *output_file) { + char raw_mbox_line[BUFFER_SIZE]; + FILE *mail_file; + + mail_file = fopen(mail_filename, "r"); + + if(!mail_file) { + fprintf(stderr, + "mymail: Cannot open mbox '%s' for mail extraction.\n", + mail_filename); + exit(EXIT_FAILURE); + } + + fseek(mail_file, position_in_mail, SEEK_SET); + + if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) { + 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)) + ) break; + fprintf(output_file, "%s", raw_mbox_line); } } fclose(mail_file); } -void search_in_db(const char *db_filename, - int nb_search_conditions, - struct search_condition *search_conditions, - FILE *output_file) { +int 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; + int nb_extracted_mails; + + nb_extracted_mails = 0; if(!quiet) { printf("Searching in '%s' ... ", db_filename); @@ -380,6 +412,8 @@ void search_in_db(const char *db_filename, exit(EXIT_FAILURE); } + /* First, check the db file leading line integrity */ + 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, @@ -394,6 +428,8 @@ void search_in_db(const char *db_filename, exit(EXIT_FAILURE); } + /* Then parse the said db file */ + current_position_in_mail = 0; already_written = 0; @@ -424,8 +460,7 @@ void search_in_db(const char *db_filename, if(n == nb_search_conditions) { - /* all conditions but the body ones are fine, check the body - ones */ + /* Now check the body ones */ if(nb_body_conditions > 0) { update_body_hits(current_mail_filename, current_position_in_mail, @@ -444,33 +479,8 @@ void search_in_db(const char *db_filename, } 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); + nb_extracted_mails++; + extract_mail(current_mail_filename, current_position_in_mail, output_file); } } } @@ -478,9 +488,8 @@ void 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, TOKEN_BUFFER_SIZE, ' ', mbox_value); + mbox_value = parse_token(current_mail_filename, TOKEN_BUFFER_SIZE, '\n', mbox_value); current_position_in_mail = atol(position_in_file_string); - remove_eof(current_mail_filename); already_written = 0; } @@ -504,16 +513,19 @@ void search_in_db(const char *db_filename, printf("done.\n"); fflush(stdout); } + + return nb_extracted_mails; } -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) { +int 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]; + int nb_extracted_mails = 0; if(lstat(entry_name, &sb) != 0) { fprintf(stderr, @@ -529,9 +541,9 @@ void recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp, 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); + nb_extracted_mails += recursive_search_in_db(subname, db_filename_regexp, + nb_search_conditions, search_conditions, + output_file); } } closedir(dir); @@ -542,9 +554,12 @@ void recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp, 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); + nb_extracted_mails += + search_in_db(entry_name, nb_search_conditions, search_conditions, output_file); } } + + return nb_extracted_mails; } /*********************************************************************/ @@ -568,7 +583,7 @@ void index_mbox(const char *mbox_filename, char raw_mbox_line[BUFFER_SIZE], full_line[BUFFER_SIZE]; char *end_of_full_line; FILE *file; - int in_header, new_header, last_mbox_line_was_empty; + int in_header, new_header; unsigned long int position_in_file; file = fopen(mbox_filename, "r"); @@ -585,10 +600,9 @@ 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_mbox_line, BUFFER_SIZE, file)) { - if(is_a_leading_from_line(last_mbox_line_was_empty, raw_mbox_line)) { + if(is_a_leading_from_line(raw_mbox_line)) { if(in_header) { fprintf(stderr, "Got a ^\"From \" in the header in %s:%lu.\n", @@ -602,8 +616,6 @@ void index_mbox(const char *mbox_filename, 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); @@ -653,7 +665,7 @@ void index_mbox(const char *mbox_filename, } void recursive_index_mbox(FILE *db_file, - const char *entry_name, + const char *entry_name, regex_t *mbox_filename_regexp, int nb_fields_to_parse, struct parsable_field *fields_to_parse) { DIR *dir; struct dirent *dir_e; @@ -674,12 +686,17 @@ void recursive_index_mbox(FILE *db_file, 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_index_mbox(db_file, subname, nb_fields_to_parse, fields_to_parse); + recursive_index_mbox(db_file, subname, mbox_filename_regexp, + nb_fields_to_parse, fields_to_parse); } } closedir(dir); } else { - index_mbox(entry_name, nb_fields_to_parse, fields_to_parse, db_file); + const char *s = entry_name, *filename = s; + while(*s) { if(*s == '/') { filename = s+1; }; s++; } + if(!mbox_filename_regexp || regexec(mbox_filename_regexp, filename, 0, 0, 0) == 0) { + index_mbox(entry_name, nb_fields_to_parse, fields_to_parse, db_file); + } } } @@ -695,10 +712,11 @@ 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-file-generate", 1, 0, 'd' }, { "db-pattern", 1, 0, 'p' }, { "db-root", 1, 0, 'r' }, { "db-list", 1, 0, 'l' }, + { "mbox-pattern", 1, 0, 'm' }, { "search", 1, 0, 's' }, { "index", 0, 0, 'i' }, { "output", 1, 0, 'o' }, @@ -731,6 +749,13 @@ void init_condition(struct search_condition *condition, char *full_string) { 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; + } + else if(strcmp(search_field, "week") == 0) { condition->field_id = ID_INTERVAL; condition->interval_start = time(0) - 3600 * 24 * 7; @@ -743,10 +768,9 @@ void init_condition(struct search_condition *condition, char *full_string) { condition->interval_stop = 0; } - else if(strcmp(search_field, "24h") == 0 || - strcmp(search_field, "today") == 0) { + else if(strcmp(search_field, "year") == 0) { condition->field_id = ID_INTERVAL; - condition->interval_start = time(0) - 3600 * 24; + condition->interval_start = time(0) - 3600 * 24 * 365; condition->interval_stop = 0; } @@ -860,6 +884,7 @@ int main(int argc, char **argv) { char *db_filename_regexp_string; char *db_root_path; char *db_filename_list; + char *mbox_filename_regexp_string; char output_filename[PATH_MAX + 1]; int action_index; int error = 0, show_help = 0; @@ -867,12 +892,9 @@ int main(int argc, char **argv) { char c; 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)) { + if(regcomp(&leading_from_line_regexp, LEADING_FROM_LINE_REGEXP, 0)) { fprintf(stderr, "mymail: Cannot compile leading \"from\" line regexp. That is strange.\n"); exit(EXIT_FAILURE); @@ -884,8 +906,10 @@ int main(int argc, char **argv) { db_filename_regexp_string = 0; db_root_path = 0; db_filename_list = 0; + mbox_filename_regexp_string = 0; quiet = 0; default_search_field = 0; + ignore_dot_files = 1; setlocale(LC_ALL, ""); @@ -913,22 +937,46 @@ int main(int argc, char **argv) { break; case 'd': + if(db_filename) { + fprintf(stderr, "mymail: Can not set the db filename twice.\n"); + exit(EXIT_FAILURE); + } db_filename = strdup(optarg); break; case 'p': + if(db_filename_regexp_string) { + fprintf(stderr, "mymail: Can not set the db filename pattern twice.\n"); + exit(EXIT_FAILURE); + } db_filename_regexp_string = strdup(optarg); break; + case 'm': + if(mbox_filename_regexp_string) { + fprintf(stderr, "mymail: Can not set the mbox filename pattern twice.\n"); + exit(EXIT_FAILURE); + } + mbox_filename_regexp_string = strdup(optarg); + break; + case 'o': strncpy(output_filename, optarg, PATH_MAX); break; case 'r': + if(db_root_path) { + fprintf(stderr, "mymail: Can not set the db root path twice.\n"); + exit(EXIT_FAILURE); + } db_root_path = strdup(optarg); break; case 'l': + if(db_filename_list) { + fprintf(stderr, "mymail: Can not set the db filename list twice.\n"); + exit(EXIT_FAILURE); + } db_filename_list = strdup(optarg); break; @@ -951,6 +999,9 @@ int main(int argc, char **argv) { } } + /* Set all the values that may defined in the arguments, through + environment variables, or hard-coded */ + if(!db_filename) { char *default_db_filename = getenv("MYMAIL_DB_FILE"); @@ -987,21 +1038,16 @@ int main(int argc, char **argv) { } } - if(output_filename[0]) { - output_file = fopen(output_filename, "w"); + if(!mbox_filename_regexp_string) { + char *default_mbox_filename_regexp_string = getenv("MYMAIL_MBOX_PATTERN"); - if(!output_file) { - fprintf(stderr, - "mymail: Cannot open result file \"%s\" for writing: %s\n", - output_filename, - strerror(errno)); - exit(EXIT_FAILURE); + if(default_mbox_filename_regexp_string) { + mbox_filename_regexp_string = strdup(default_mbox_filename_regexp_string); } - } else { - output_file = stdout; - quiet = 1; } + /* Start the processing */ + if(error) { print_usage(stderr); exit(EXIT_FAILURE); @@ -1012,8 +1058,26 @@ int main(int argc, char **argv) { exit(EXIT_SUCCESS); } + /* mbox indexing */ + if(action_index) { FILE *db_file; + regex_t mbox_filename_regexp_static; + regex_t *mbox_filename_regexp; + + if(mbox_filename_regexp_string) { + if(regcomp(&mbox_filename_regexp_static, + mbox_filename_regexp_string, + 0)) { + fprintf(stderr, + "mymail: Syntax error in regexp \"%s\".\n", + mbox_filename_regexp_string); + exit(EXIT_FAILURE); + } + mbox_filename_regexp = &mbox_filename_regexp_static; + } else { + mbox_filename_regexp = 0; + } db_file = fopen(db_filename, "w"); @@ -1041,7 +1105,7 @@ int main(int argc, char **argv) { while(optind < argc) { recursive_index_mbox(db_file, - argv[optind], + argv[optind], mbox_filename_regexp, nb_fields_to_parse, fields_to_parse); optind++; } @@ -1049,13 +1113,37 @@ int main(int argc, char **argv) { fflush(db_file); fclose(db_file); + if(mbox_filename_regexp) { + regfree(mbox_filename_regexp); + } + for(f = 0; f < nb_fields_to_parse; f++) { regfree(&fields_to_parse[f].regexp); } } + /* Mail search */ + else { + FILE *output_file; + int nb_extracted_mails = 0; + + 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(nb_search_conditions > 0) { /* Recursive search if db_root_path is set */ @@ -1071,9 +1159,9 @@ int main(int argc, char **argv) { exit(EXIT_FAILURE); } - recursive_search_in_db(db_root_path, &db_filename_regexp, - nb_search_conditions, search_conditions, - output_file); + nb_extracted_mails += recursive_search_in_db(db_root_path, &db_filename_regexp, + nb_search_conditions, search_conditions, + output_file); regfree(&db_filename_regexp); } @@ -1090,7 +1178,8 @@ int main(int argc, char **argv) { 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); + nb_extracted_mails += + search_in_db(db_filename, nb_search_conditions, search_conditions, output_file); } } } @@ -1098,25 +1187,36 @@ int main(int argc, char **argv) { /* 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); + nb_extracted_mails += + search_in_db(argv[optind], nb_search_conditions, search_conditions, output_file); optind++; } } + + if(!quiet) { + if(nb_extracted_mails > 0) { + printf("Found %d matching mails.\n", nb_extracted_mails); + } else { + printf("No matching mail found.\n"); + } + } + + fflush(output_file); + + if(output_file != stdout) { + fclose(output_file); + } } 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); + free(mbox_filename_regexp_string); regfree(&leading_from_line_regexp);