From 5b468174fa4d6629faf3069e254a17c72bd18545 Mon Sep 17 00:00:00 2001 From: Francois Fleuret Date: Sun, 10 Feb 2013 12:49:04 +0100 Subject: [PATCH] Added the --mbox-pattern option. --- mymail.1 | 5 ++- mymail.c | 119 ++++++++++++++++++++++++++++++++++++++++++++----------- 2 files changed, 100 insertions(+), 24 deletions(-) diff --git a/mymail.1 b/mymail.1 index e48633b..95b83f0 100644 --- a/mymail.1 +++ b/mymail.1 @@ -40,11 +40,14 @@ set the db root path for recursive search \fB-l \fR, \fB--db-list \fR set the semicolon-separated list of db files for search .TP +\fB-m \fR, \fB--mbox-pattern \fR +set the mbox filename pattern for recursive search +.TP \fB-s \fR, \fB--search \fR search for matching mails in the db file. Multiple search requests can be combined, and only mails fulfilling all of them will be extracted. .TP -\fB-d \fR, \fB--db-file \fR +\fB-d \fR, \fB--db-file-generate \fR set the db filename for indexing .TP \fB-i\fR, \fB--index\fR diff --git a/mymail.c b/mymail.c index f9b0148..5287301 100644 --- a/mymail.c +++ b/mymail.c @@ -203,9 +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"); @@ -663,7 +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; @@ -684,12 +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); + } } } @@ -705,10 +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' }, @@ -876,6 +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; @@ -883,7 +892,6 @@ 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, LEADING_FROM_LINE_REGEXP, 0)) { @@ -898,6 +906,7 @@ 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; @@ -928,22 +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; @@ -966,6 +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"); @@ -1002,21 +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); @@ -1027,8 +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"); @@ -1056,7 +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++; } @@ -1064,14 +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 */ @@ -1129,21 +1201,22 @@ int main(int argc, char **argv) { } } + 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); -- 2.39.5