X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=mymail.c;h=21ad5eabff678aaad71b8bb24b164803369f06ea;hb=622b0a6df1bcb8e072facf472e4568609ab2a321;hp=35a0344a971234dfb8f4a8591ff434e95e94c1c7;hpb=4494ac307c7d814e9ee3a1e6e1ead3e1cb35f781;p=mymail.git diff --git a/mymail.c b/mymail.c index 35a0344..21ad5ea 100644 --- a/mymail.c +++ b/mymail.c @@ -45,17 +45,28 @@ #include #define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file" -#define VERSION "0.1" +#define VERSION "0.9" -#define MAX_NB_SEARCH_PATTERNS 10 +#define MAX_NB_SEARCH_REQUESTS 10 #define BUFFER_SIZE 65536 +char *db_filename; +char *db_filename_regexp_string; +char *db_root_path; +char *db_filename_list; + +int paranoid; +int action_index; + +/********************************************************************/ + enum { - ID_MAIL, + ID_MAIL = 0, ID_FROM, - ID_DEST, + ID_TO, ID_SUBJECT, + ID_DATE, ID_PARTICIPANT, MAX_ID }; @@ -63,28 +74,60 @@ enum { static char *field_names[] = { "mail", "from", - "dest", + "to", "subject", + "date", "part" }; +/********************************************************************/ + struct search_request { int field_id; int negation; regex_t regexp; }; +/********************************************************************/ + struct parsable_field { int id; char *regexp_string; regex_t regexp; }; -char *db_filename; -char *db_root_path; +static struct parsable_field fields_to_parse[] = { + { + ID_FROM, + "^\\(From \\|[Ff][Rr][Oo][Mm]:\\|[R][r][E][e][P][p][L][l][Y][y]-[T][t][O][o]:\\)", + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } + }, -int paranoid; -int action_index; + { + ID_TO, + "^\\([Tt][Oo]\\|[Cc][Cc]\\|[Bb][Cc][Cc]\\): ", + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } + }, + + { + ID_SUBJECT, + "^[Ss][Uu][Bb][Jj][Ee][Cc][Tt]: ", + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } + }, + + { + ID_DATE, + "^[Dd][Aa][Tt][Ee]: ", + { 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 *segment_next_field(char *current) { while(*current && *current != ' ') current++; @@ -122,20 +165,24 @@ 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, " -i, --index\n"); - fprintf(out, " index mails\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 data-base file\n"); + fprintf(out, " search for matching mails in the db file\n"); fprintf(out, " -d , --db-file \n"); - fprintf(out, " set the data-base file\n"); - fprintf(out, " -r , --db-root \n"); - fprintf(out, " set the data-base root path for recursive search\n"); + fprintf(out, " set the db filename for indexing\n"); + fprintf(out, " -i, --index\n"); + fprintf(out, " index mails\n"); } /*********************************************************************/ @@ -151,15 +198,15 @@ int mbox_line_match_search(struct search_request *request, int mbox_id, char *mbox_value) { return (request->field_id == mbox_id || - (request->field_id == ID_PARTICIPANT && (mbox_id == ID_FROM || mbox_id == ID_DEST))) + (request->field_id == ID_PARTICIPANT && (mbox_id == ID_FROM || mbox_id == ID_TO))) && regexec(&request->regexp, mbox_value, 0, 0, 0) == 0; } -void search_in_db(int nb_search_patterns, +void search_in_db(int nb_search_requests, struct search_request *search_requests, FILE *db_file) { - int hits[MAX_NB_SEARCH_PATTERNS]; + int hits[MAX_NB_SEARCH_REQUESTS]; char raw_db_line[BUFFER_SIZE]; char raw_mbox_line[BUFFER_SIZE]; char current_mail_filename[PATH_MAX + 1]; @@ -167,11 +214,12 @@ void search_in_db(int nb_search_patterns, char *mbox_name, *mbox_value; int mbox_id; int already_written, m, n; + int last_mbox_line_was_empty; current_position_in_mail = 0; already_written = 0; - for(n = 0; n < nb_search_patterns; n++) { hits[n] = 0; } + for(n = 0; n < nb_search_requests; n++) { hits[n] = 0; } while(fgets(raw_db_line, BUFFER_SIZE, db_file)) { mbox_name = raw_db_line; @@ -181,9 +229,13 @@ void search_in_db(int nb_search_patterns, char *position_in_file_string; char *mail_filename; - for(n = 0; n < nb_search_patterns && hits[n]; n++); + for(n = 0; n < nb_search_requests && xor(hits[n], search_requests[n].negation); n++); + + /* for(n = 0; n < nb_search_requests && */ + /* ((hits[n] && !search_requests[n].negation) || */ + /* (!hits[n] && search_requests[n].negation)); n++); */ - if(n == nb_search_patterns) { + if(n == nb_search_requests) { FILE *mail_file; mail_file = fopen(current_mail_filename, "r"); @@ -196,9 +248,12 @@ void search_in_db(int nb_search_patterns, 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(fgets(raw_mbox_line, BUFFER_SIZE, mail_file) && - strncmp(raw_mbox_line, "From ", 5)) { + 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); } } @@ -206,7 +261,7 @@ void search_in_db(int nb_search_patterns, fclose(mail_file); } - for(n = 0; n < nb_search_patterns; n++) { hits[n] = 0; } + for(n = 0; n < nb_search_requests; n++) { hits[n] = 0; } position_in_file_string = mbox_value; mail_filename = segment_next_field(mbox_value); @@ -224,7 +279,7 @@ void search_in_db(int nb_search_patterns, mbox_id = m; } } - for(n = 0; n < nb_search_patterns; n++) { + for(n = 0; n < nb_search_requests; n++) { hits[n] |= mbox_line_match_search(&search_requests[n], mbox_id, mbox_value); } @@ -232,9 +287,8 @@ void search_in_db(int nb_search_patterns, } } -void recursive_search_in_db(const char *entry_name, - int nb_search_patterns, - struct search_request *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) { DIR *dir; struct dirent *dir_e; struct stat sb; @@ -255,17 +309,18 @@ 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, - nb_search_patterns, - search_requests); + recursive_search_in_db(subname, db_filename_regexp, + nb_search_requests, search_requests); } } 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) { + if(regexec(db_filename_regexp, filename, 0, 0, 0) == 0) { FILE *db_file = fopen(entry_name, "r"); if(!db_file) { @@ -290,7 +345,7 @@ void recursive_search_in_db(const char *entry_name, exit(EXIT_FAILURE); } - search_in_db(nb_search_patterns, search_requests, db_file); + search_in_db(nb_search_requests, search_requests, db_file); fclose(db_file); } @@ -318,7 +373,7 @@ 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; + int in_header, new_header, last_mbox_line_was_empty; unsigned long int position_in_file; file = fopen(mbox_filename, "r"); @@ -335,9 +390,10 @@ 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(strncmp(raw_mbox_line, "From ", 5) == 0) { + if(last_mbox_line_was_empty && strncmp(raw_mbox_line, "From ", 5) == 0) { if(in_header) { fprintf(stderr, "Got a ^\"From \" in the header in %s:%lu.\n", @@ -347,10 +403,12 @@ void index_mbox(const char *mbox_filename, } in_header = 1; new_header = 1; - } else if(strncmp(raw_mbox_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); @@ -442,33 +500,14 @@ static struct option long_options[] = { { "help", no_argument, 0, 'h' }, { "version", no_argument, 0, 'v' }, { "db-file", 1, 0, 'd' }, + { "db-pattern", 1, 0, 'p' }, { "db-root", 1, 0, 'r' }, + { "db-list", 1, 0, 'l' }, { "search", 1, 0, 's' }, { "index", 0, 0, 'i' }, { 0, 0, 0, 0 } }; -static struct parsable_field fields_to_parse[] = { - { - ID_FROM, - "^\\([Ff][Rr][Oo][Mm]:\\|From\\) *", - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } - }, - - { - ID_DEST, - "^\\([Tt][Oo]\\|[Cc][Cc]\\|[Bb][Cc][Cc]\\): *", - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } - }, - - { - ID_SUBJECT, - "^[Ss][Uu][Bb][Jj][Ee][Cc][Tt]: *", - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } - }, - -}; - /*********************************************************************/ int main(int argc, char **argv) { @@ -476,8 +515,8 @@ 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_patterns; - char *search_pattern[MAX_NB_SEARCH_PATTERNS]; + int nb_search_requests; + char *search_request_strings[MAX_NB_SEARCH_REQUESTS]; /* for(f = 0; f < argc; f++) { */ /* printf("arg %d \"%s\"\n", f, argv[f]); */ @@ -487,12 +526,13 @@ int main(int argc, char **argv) { action_index = 0; db_filename = 0; db_root_path = 0; + db_filename_list = 0; setlocale(LC_ALL, ""); - nb_search_patterns = 0; + nb_search_requests = 0; - while ((c = getopt_long(argc, argv, "hvip:s:d:r:", + while ((c = getopt_long(argc, argv, "hvip:s:d:r:l:", long_options, NULL)) != -1) { switch(c) { @@ -513,16 +553,24 @@ int main(int argc, char **argv) { db_filename = strdup(optarg); break; + case 'p': + db_filename_regexp_string = strdup(optarg); + break; + case 'r': db_root_path = strdup(optarg); break; + case 'l': + db_filename_list = strdup(optarg); + break; + case 's': - if(nb_search_patterns == MAX_NB_SEARCH_PATTERNS) { + if(nb_search_requests == MAX_NB_SEARCH_REQUESTS) { fprintf(stderr, "mymail: Too many search patterns.\n"); exit(EXIT_FAILURE); } - search_pattern[nb_search_patterns++] = strdup(optarg); + search_request_strings[nb_search_requests++] = strdup(optarg); break; default: @@ -541,6 +589,16 @@ int main(int argc, char **argv) { 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"); @@ -549,12 +607,13 @@ int main(int argc, char **argv) { } } - if(!db_root_path) { - fprintf(stderr, - "mymail: db root path is not set\n"); - exit(EXIT_FAILURE); - } + if(!db_filename_list) { + char *default_db_filename_list = getenv("MYMAIL_DB_LIST"); + if(default_db_filename_list) { + db_filename_list = strdup(default_db_filename_list); + } + } if(error) { print_usage(stderr); @@ -591,7 +650,7 @@ int main(int argc, char **argv) { } } - 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, @@ -609,21 +668,36 @@ int main(int argc, char **argv) { else { - if(nb_search_patterns > 0) { - struct search_request search_requests[MAX_NB_SEARCH_PATTERNS]; - char *search_regexp_string; + if(nb_search_requests > 0) { + struct search_request search_requests[MAX_NB_SEARCH_REQUESTS]; + char *search_field, *search_regexp_string; int m, n; - for(n = 0; n < nb_search_patterns; n++) { - search_regexp_string = segment_next_field(search_pattern[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]); + + if(search_field[0] == '!') { + search_field++; + search_requests[n].negation = 1; + } else { + search_requests[n].negation = 0; + } search_requests[n].field_id = -1; for(m = 0; (m < MAX_ID) && search_requests[n].field_id == -1; m++) { - if(strncmp(field_names[m], search_pattern[n], strlen(search_pattern[n])) == 0) { + if(strncmp(field_names[m], search_field, strlen(search_field)) == 0) { search_requests[n].field_id = m; } } + if(search_requests[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, search_regexp_string, REG_ICASE)) { @@ -635,17 +709,88 @@ int main(int argc, char **argv) { } } - recursive_search_in_db(db_root_path, - nb_search_patterns, search_requests); + /* 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_requests, search_requests); + + regfree(&db_filename_regexp); + } + + /* Search in all db files listed in db_filename_list */ + + if(db_filename_list) { + char db_filename[PATH_MAX + 1]; + char *s, *t; + FILE *db_file; + + s = db_filename_list; + + while(*s) { + t = db_filename; + while(*s == ';') { s++; } + while(*s && *s != ';') { *t++ = *s++; } + *t++ = '\0'; + + if(db_filename[0]) { + db_file = fopen(db_filename, "r"); + + if(!db_file) { + fprintf(stderr, + "mymail: Cannot open \"%s\" for reading: %s\n", + argv[optind], + strerror(errno)); + exit(EXIT_FAILURE); + } + + search_in_db(nb_search_requests, search_requests, db_file); + + fclose(db_file); + } + } + } + + /* Search in all db files listed in the command arguments */ + + while(optind < argc) { + FILE *db_file = fopen(argv[optind], "r"); + + if(!db_file) { + fprintf(stderr, + "mymail: Cannot open \"%s\" for reading: %s\n", + argv[optind], + strerror(errno)); + exit(EXIT_FAILURE); + } + + search_in_db(nb_search_requests, search_requests, db_file); + + fclose(db_file); + optind++; + } - for(n = 0; n < nb_search_patterns; n++) { - free(search_pattern[n]); + for(n = 0; n < nb_search_requests; n++) { + regfree(&search_requests[n].regexp); + free(search_request_strings[n]); } } } free(db_filename); + free(db_filename_regexp_string); free(db_root_path); + free(db_filename_list); exit(EXIT_SUCCESS); }