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.
#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$"
+#define LEADING_FROM_LINE_REGEXP_STRING "^From [^ ]* \\(Mon\\|Tue\\|Wed\\|Thu\\|Fri\\|Sat\\|Sun\\) \\(Jan\\|Feb\\|Mar\\|Apr\\|May\\|Jun\\|Jul\\|Aug\\|Sep\\|Oct\\|Nov\\|Dec\\) [ 123][0-9] [0-9][0-9]:[0-9][0-9]:[0-9][0-9] [0-9][0-9][0-9][0-9]\n$"
regex_t leading_from_line_regexp;
return string;
}
+char *default_value(char *current_value,
+ const char *env_variable,
+ const char *hard_default_value) {
+ if(current_value) {
+ return current_value;
+ } else {
+ char *env_value = getenv(env_variable);
+ if(env_value) {
+ return strdup(env_value);
+ } else if(hard_default_value) {
+ return strdup(hard_default_value);
+ } else {
+ return 0;
+ }
+ }
+}
+
/********************************************************************/
/* malloc with error checking. */
fprintf(out, " set the db root path for recursive search\n");
fprintf(out, " -l <db filename list>, --db-list <db filename list>\n");
fprintf(out, " set the semicolon-separated list of db files for search\n");
+ fprintf(out, " -m <mbox filename pattern>, --mbox-pattern <mbox filename pattern>\n");
+ fprintf(out, " set the mbox filename pattern for recursive search\n");
fprintf(out, " -s <search pattern>, --search <search pattern>\n");
fprintf(out, " search for matching mails in the db file\n");
- fprintf(out, " -d <db filename>, --db-file <db filename>\n");
+ fprintf(out, " -d <db filename>, --db-file-generate <db filename>\n");
fprintf(out, " set the db filename for indexing\n");
fprintf(out, " -i, --index\n");
fprintf(out, " index mails\n");
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];
int already_written, m, n;
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);
}
if(nb_body_conditions == nb_fulfilled_body_conditions) {
+ nb_extracted_mails++;
extract_mail(current_mail_filename, current_position_in_mail, output_file);
}
}
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,
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);
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;
}
/*********************************************************************/
}
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;
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);
+ }
}
}
{ "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' },
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;
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)) {
+ if(regcomp(&leading_from_line_regexp, LEADING_FROM_LINE_REGEXP_STRING, 0)) {
fprintf(stderr,
"mymail: Cannot compile leading \"from\" line regexp. That is strange.\n");
exit(EXIT_FAILURE);
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;
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;
}
}
- if(!db_filename) {
- char *default_db_filename = getenv("MYMAIL_DB_FILE");
-
- if(!default_db_filename) {
- default_db_filename = "mymail.db";
- }
-
- db_filename = strdup(default_db_filename);
- }
-
- if(!db_filename_regexp_string) {
- char *default_db_filename_regexp_string = getenv("MYMAIL_DB_PATTERN");
+ /* Set all the values that may defined in the arguments, through
+ environment variables, or hard-coded */
- if(!default_db_filename_regexp_string) {
- default_db_filename_regexp_string = "^mymail.db$";
- }
+ db_filename = default_value(db_filename,
+ "MYMAIL_DB_FILE",
+ "mymail.db");
- db_filename_regexp_string = strdup(default_db_filename_regexp_string);
- }
+ db_filename_regexp_string = default_value(db_filename_regexp_string,
+ "MYMAIL_DB_FILE",
+ "^mymail.db$");
- if(!db_root_path) {
- char *default_db_root_path = getenv("MYMAIL_DB_ROOT");
+ db_root_path = default_value(db_root_path,
+ "MYMAIL_DB_ROOT",
+ 0);
- if(default_db_root_path) {
- db_root_path = strdup(default_db_root_path);
- }
- }
-
- if(!db_filename_list) {
- char *default_db_filename_list = getenv("MYMAIL_DB_LIST");
+ db_filename_list = default_value(db_filename_list,
+ "MYMAIL_DB_LIST",
+ 0);
- if(default_db_filename_list) {
- db_filename_list = strdup(default_db_filename_list);
- }
- }
+ mbox_filename_regexp_string = default_value(mbox_filename_regexp_string,
+ "MYMAIL_MBOX_PATTERN",
+ "mbox$");
- 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;
- }
+ /* Start the processing */
if(error) {
print_usage(stderr);
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");
while(optind < argc) {
recursive_index_mbox(db_file,
- argv[optind],
+ argv[optind], mbox_filename_regexp,
nb_fields_to_parse, fields_to_parse);
optind++;
}
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 */
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);
}
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);
}
}
}
/* 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);
projects / mymail.git / blobdiff