X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=mymail.c;h=27b48f4fe8a5f2f481d46932e4ae4b266eba78be;hb=b18f24afc58a95f309f33e06ba7b2e67db06bea8;hp=0593f621d3045b0e3a7b392c15cccbd1c9a0fc08;hpb=f364f9c86b041f3541f314443d53bc9c5cb77b7c;p=mymail.git diff --git a/mymail.c b/mymail.c index 0593f62..27b48f4 100644 --- a/mymail.c +++ b/mymail.c @@ -45,14 +45,22 @@ #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_root_path; + +int paranoid; +int action_index; + +/********************************************************************/ + enum { - ID_MAIL, + ID_MAIL = 0, ID_FROM, ID_DEST, ID_SUBJECT, @@ -64,21 +72,52 @@ static char *field_names[] = { "mail", "from", "dest", - "subj", + "subject", "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_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 xor(int a, int b) { + return (a && !b) || (!a && b); +} char *segment_next_field(char *current) { while(*current && *current != ' ') current++; @@ -141,44 +180,32 @@ int ignore_entry(const char *name) { (name[0] == '.' && name[1] != '/'); } -int mbox_line_match_search(int search_id, regex_t *search_regexp, +int mbox_line_match_search(struct search_request *request, int mbox_id, char *mbox_value) { return - (search_id == mbox_id || - (search_id == ID_PARTICIPANT && (mbox_id == ID_FROM || mbox_id == ID_DEST))) + (request->field_id == mbox_id || + (request->field_id == ID_PARTICIPANT && (mbox_id == ID_FROM || mbox_id == ID_DEST))) && - regexec(search_regexp, mbox_value, 0, 0, 0) == 0; + regexec(&request->regexp, mbox_value, 0, 0, 0) == 0; } -void search_in_db(int nb_search_patterns, - int *search_ids, char **search_regexp_strings, +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]; unsigned long int current_position_in_mail; char *mbox_name, *mbox_value; int mbox_id; - regex_t search_regexps[MAX_NB_SEARCH_PATTERNS]; int already_written, m, n; - - for(n = 0; n < nb_search_patterns; n++) { - if(regcomp(&search_regexps[n], - search_regexp_strings[n], - REG_ICASE)) { - fprintf(stderr, - "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n", - search_regexp_strings[n], - field_names[search_ids[n]]); - exit(EXIT_FAILURE); - } - } + 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; @@ -188,9 +215,13 @@ 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"); @@ -203,9 +234,12 @@ 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); } } @@ -213,7 +247,7 @@ 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); @@ -231,21 +265,17 @@ void search_in_db(int nb_search_patterns, mbox_id = m; } } - for(n = 0; n < nb_search_patterns; n++) { - hits[n] |= mbox_line_match_search(search_ids[n], &search_regexps[n], + for(n = 0; n < nb_search_requests; n++) { + hits[n] |= mbox_line_match_search(&search_requests[n], mbox_id, mbox_value); } } } - - for(n = 0; n < nb_search_patterns; n++) { - regfree(&search_regexps[n]); - } } void recursive_search_in_db(const char *entry_name, - int nb_search_patterns, - int *search_ids, char **search_regexp_strings) { + int nb_search_requests, + struct search_request *search_requests) { DIR *dir; struct dirent *dir_e; struct stat sb; @@ -267,8 +297,8 @@ void recursive_search_in_db(const char *entry_name, 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_ids, search_regexp_strings); + nb_search_requests, + search_requests); } } closedir(dir); @@ -301,8 +331,7 @@ void recursive_search_in_db(const char *entry_name, exit(EXIT_FAILURE); } - search_in_db(nb_search_patterns, search_ids, search_regexp_strings, - db_file); + search_in_db(nb_search_requests, search_requests, db_file); fclose(db_file); } @@ -330,7 +359,7 @@ 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"); @@ -347,9 +376,10 @@ 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", @@ -359,10 +389,12 @@ 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); @@ -460,27 +492,6 @@ static struct option long_options[] = { { 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) { @@ -488,11 +499,11 @@ 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]); */ + /* printf("arg %d \"%s\"\n", f, argv[f]); */ /* } */ paranoid = 0; @@ -502,7 +513,7 @@ int main(int argc, char **argv) { setlocale(LC_ALL, ""); - nb_search_patterns = 0; + nb_search_requests = 0; while ((c = getopt_long(argc, argv, "hvip:s:d:r:", long_options, NULL)) != -1) { @@ -530,11 +541,11 @@ int main(int argc, char **argv) { 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: @@ -561,13 +572,6 @@ int main(int argc, char **argv) { } } - if(!db_root_path) { - fprintf(stderr, - "mymail: db root path is not set\n"); - exit(EXIT_FAILURE); - } - - if(error) { print_usage(stderr); exit(EXIT_FAILURE); @@ -603,7 +607,7 @@ 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, @@ -621,32 +625,59 @@ int main(int argc, char **argv) { else { - if(nb_search_patterns > 0) { - int search_ids[MAX_NB_SEARCH_PATTERNS]; - char *search_regexp_strings[MAX_NB_SEARCH_PATTERNS]; + if(!db_root_path) { + fprintf(stderr, + "mymail: db root path is not set\n"); + exit(EXIT_FAILURE); + } + + 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_strings[n] = segment_next_field(search_pattern[n]); - search_ids[n] = -1; - for(m = 0; (m < MAX_ID) && search_ids[n] == -1; m++) { - if(strncmp(field_names[m], search_pattern[n], strlen(search_pattern[n])) == 0) { - search_ids[n] = m; + 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_field, strlen(search_field)) == 0) { + search_requests[n].field_id = m; } } - } - if(!*search_regexp_strings) { - fprintf(stderr, - "Syntax error in the search pattern.\n"); - exit(EXIT_FAILURE); + 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)) { + fprintf(stderr, + "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n", + search_regexp_string, + field_names[search_requests[n].field_id]); + exit(EXIT_FAILURE); + } } recursive_search_in_db(db_root_path, - nb_search_patterns, search_ids, search_regexp_strings); + nb_search_requests, search_requests); - 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]); } } }