/*
- This command is a dumb mail indexer. It can either (1) scan
- 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, to create a resulting mbox file.
+ mymail is a simple mail indexer. It can:
+
+ (1) scan mbox files, and create a db file containing for each mail a
+ list of fields computed from its header.
+
+ (2) read such a db file, gets all the mails matching regexp-defined
+ conditions on the fields, and generates a resulting mbox file.
It is low-tech, simple, light and fast.
#include <time.h>
#define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file"
-#define VERSION "0.9.6"
+#define MYMAIL_VERSION "0.9.9"
+
+#define MYMAIL_DB_FORMAT_VERSION 1
#define MAX_NB_SEARCH_CONDITIONS 32
#define BUFFER_SIZE 65536
#define TOKEN_BUFFER_SIZE 1024
-#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$"
+#define LEADING_FROM_LINE_REGEXP_STRING "^From .*\\(Mon\\|Tue\\|Wed\\|Thu\\|Fri\\|Sat\\|Sun\\) \\(Jan\\|Feb\\|Mar\\|Apr\\|May\\|Jun\\|Jul\\|Aug\\|Sep\\|Oct\\|Nov\\|Dec\\) [ 0123][0-9] [0-9][0-9]:[0-9][0-9]:[0-9][0-9] [0-9][0-9][0-9][0-9]\n$"
+
+/********************************************************************/
+
+struct alias_node {
+ char *alias, *value;
+ struct alias_node *next;
+};
/* Global variables! */
int global_quiet;
int global_use_leading_time;
-
+int global_nb_mails_max;
regex_t global_leading_from_line_regexp;
+struct alias_node *global_alias_list;
/********************************************************************/
ID_PARTICIPANT,
ID_BODY,
ID_TIME_INTERVAL,
+ ID_MAIL_ID,
+ ID_REFERENCE_ID,
MAX_ID
};
"date",
"part",
"body",
- "interval"
+ "interval",
+ "mailid",
+ "references"
};
/********************************************************************/
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
},
+ {
+ ID_MAIL_ID,
+ REG_ICASE,
+ "^message-id: ",
+ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
+ },
+
+ {
+ ID_REFERENCE_ID,
+ REG_ICASE,
+ "^\\(in-reply-to\\|references\\): ",
+ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
+ },
+
};
/********************************************************************/
const char *parse_token(char *token_buffer, size_t token_buffer_size,
char separator, const char *string) {
char *u = token_buffer;
+
while(*string == separator) { string++; }
- while(u < token_buffer + token_buffer_size - 1 && *string &&
- *string != separator) {
+
+ while(u < token_buffer + token_buffer_size - 1 && *string && *string != separator) {
*(u++) = *(string++);
}
+
while(*string == separator) { string++; }
+
*u = '\0';
return string;
}
}
}
+/********************************************************************/
+
+void *safe_malloc(size_t n) {
+ void *p = malloc(n);
+ if(!p && n != 0) {
+ fprintf(stderr,
+ "mymail: cannot allocate memory: %s\n", strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+ return p;
+}
+
FILE *safe_fopen(const char *path, const char *mode, const char *comment) {
FILE *result = fopen(path, mode);
if(result) {
return result;
} else {
fprintf(stderr,
- "mymail: Cannot open file '%s' (%s) with mode \"%s\".\n",
- path, comment, mode);
+ "mymail: Cannot open file '%s' (%s) with mode \"%s\": %s\n",
+ path, comment, mode,
+ strerror(errno));
exit(EXIT_FAILURE);
}
}
/*********************************************************************/
void print_version(FILE *out) {
- fprintf(out, "mymail version %s (%s)\n", VERSION, UNAME);
+ fprintf(out, "mymail version %s (%s)\n", MYMAIL_VERSION, UNAME);
}
void print_usage(FILE *out) {
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-generate <db filename>\n");
+ fprintf(out, " -d <db filename>, --db-file-output <db filename>\n");
fprintf(out, " set the db filename for indexing\n");
fprintf(out, " -i, --index\n");
fprintf(out, " index mails\n");
fprintf(out, " -o <output filename>, --output <output filename>\n");
fprintf(out, " set the result file, use stdout if unset\n");
+ fprintf(out, " -n <max number of mails>, --nb-mails-max <max number of mails>\n");
+ fprintf(out, " set the maximum number of mails to extract\n");
fprintf(out, " -a <search field>, --default-search <search field>\n");
fprintf(out, " set the default search field\n");
}
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'); */
- /* if(last_mbox_line_was_empty) { header = 0; } */
-
if(raw_mbox_line[0] == '\n') { header = 0; }
if(!header) {
char raw_mbox_line[BUFFER_SIZE];
FILE *mail_file;
+ /* printf("Extract\n"); */
+
mail_file = safe_fopen(mail_filename, "r", "mbox for mail extraction");
fseek(mail_file, position_in_mail, SEEK_SET);
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))
- )
+ is_a_leading_from_line(raw_mbox_line))
break;
fprintf(output_file, "%s", raw_mbox_line);
}
const char *c;
struct tm tm;
+ memset(&tm, 0, sizeof(struct tm));
+
if(db_key == ID_LEADING_LINE) {
c = db_value;
while(*c && *c != ' ') c++; while(*c && *c == ' ') c++;
+ /* printf("From %s", db_value); */
strptime(c, "%a %b %e %k:%M:%S %Y", &tm);
*t = mktime(&tm);
} else {
if(db_key == ID_DATE) {
if(strptime(db_value, "%a, %d %b %Y %k:%M:%S", &tm) ||
strptime(db_value, "%d %b %Y %k:%M:%S", &tm)) {
+ /* printf("Date: %s", db_value); */
*t = mktime(&tm);
}
}
}
int search_in_db(const char *db_filename,
+ int nb_extracted_mails,
int nb_search_conditions,
struct search_condition *search_conditions,
FILE *output_file) {
int db_key;
int hits[MAX_NB_SEARCH_CONDITIONS];
int nb_body_conditions, need_time;
- int nb_extracted_mails;
time_t mail_time;
int m, n;
- nb_extracted_mails = 0;
-
if(!global_quiet) {
printf("Searching in '%s' ... ", db_filename);
fflush(stdout);
strcpy(current_mail_filename, "");
- while(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
+ while(nb_extracted_mails < global_nb_mails_max &&
+ fgets(raw_db_line, BUFFER_SIZE, db_file)) {
+
+ /* Removes the CR */
+ char *s = raw_db_line;
+ while(*s && *s != '\n') { s++; }
+ *s = '\0';
+
db_value = parse_token(db_key_string, TOKEN_BUFFER_SIZE, ' ', raw_db_line);
if(strcmp("mail", db_key_string) == 0) {
for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
db_value = parse_token(position_in_file_string, TOKEN_BUFFER_SIZE, ' ', db_value);
- db_value = parse_token(current_mail_filename, PATH_MAX+1, '\n', db_value);
+ strncpy(current_mail_filename, db_value, PATH_MAX + 1);
current_position_in_mail = atol(position_in_file_string);
}
}
}
- if(current_mail_filename[0]) {
- if(check_full_mail_match(current_mail_filename,
- mail_time,
- nb_search_conditions, search_conditions,
- nb_body_conditions, hits, current_position_in_mail)) {
- extract_mail(current_mail_filename, current_position_in_mail, output_file);
- nb_extracted_mails++;
- }
+ if(nb_extracted_mails < global_nb_mails_max &&
+ current_mail_filename[0] &&
+ check_full_mail_match(current_mail_filename,
+ mail_time,
+ nb_search_conditions, search_conditions,
+ nb_body_conditions, hits, current_position_in_mail)) {
+ extract_mail(current_mail_filename, current_position_in_mail, output_file);
+ nb_extracted_mails++;
}
fclose(db_file);
}
int recursive_search_in_db(const char *entry_name, regex_t *db_filename_regexp,
+ int nb_extracted_mails,
int nb_search_conditions,
struct search_condition *search_conditions,
FILE *output_file) {
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,
dir = opendir(entry_name);
if(dir) {
- while((dir_e = readdir(dir))) {
+ while((dir_e = readdir(dir)) &&
+ nb_extracted_mails < global_nb_mails_max) {
if(!ignore_entry(dir_e->d_name)) {
snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
- nb_extracted_mails += 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_extracted_mails,
+ 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) {
- nb_extracted_mails +=
- search_in_db(entry_name, nb_search_conditions, search_conditions, output_file);
+ nb_extracted_mails =
+ search_in_db(entry_name,
+ nb_extracted_mails,
+ nb_search_conditions, search_conditions, output_file);
}
}
{ "version", no_argument, 0, 'v' },
{ "quiet", no_argument, 0, 'q' },
{ "use-leading-time", no_argument, 0, 't' },
- { "db-file-generate", 1, 0, 'd' },
+ { "db-file-output", 1, 0, 'd' },
{ "db-pattern", 1, 0, 'p' },
{ "db-root", 1, 0, 'r' },
{ "db-list", 1, 0, 'l' },
{ "index", 0, 0, 'i' },
{ "output", 1, 0, 'o' },
{ "default-search", 1, 0, 'a' },
+ { "nb-mails-max", 1, 0, 'n' },
{ 0, 0, 0, 0 }
};
{ "48h", 0, 48, -1, -1 },
{ "week", 0, 24 * 7, -1, -1 },
{ "month", 0, 24 * 31, -1, -1 },
+ { "trimester", 0, 24 * 92, -1, -1 },
{ "year", 0, 24 * 365, -1, -1 },
{ "yesterday", 1, -1, -1, -1 },
char full_search_field[TOKEN_BUFFER_SIZE], *search_field;
unsigned int k, m;
const char *string;
+ struct alias_node *a;
+
+ for(a = global_alias_list; a; a = a->next) {
+ if(strcmp(full_string, a->alias) == 0) {
+ full_string = a->value;
+ break;
+ }
+ }
string = parse_token(full_search_field, TOKEN_BUFFER_SIZE, ' ', full_string);
search_field = full_search_field;
condition->time_stop = 0;
}
}
+
break;
}
}
}
}
+const char *eat_space(const char *s) {
+ while(*s == ' ' || *s == '\t') { s++; }
+ return s;
+}
+
+void read_rc_file(const char *rc_filename) {
+ char raw_line[BUFFER_SIZE];
+ char command[TOKEN_BUFFER_SIZE], tmp_token[TOKEN_BUFFER_SIZE];
+
+ FILE *rc_file;
+ int line_number;
+ const char *s;
+ char *t;
+
+ rc_file = fopen(rc_filename, "r");
+
+ if(rc_file) {
+ line_number = 1;
+ while(fgets(raw_line, BUFFER_SIZE, rc_file)) {
+ t = raw_line;
+ while(*t) { if(*t == '\n') { *t = '\0'; }; t++; }
+
+ s = raw_line;
+ s = eat_space(s);
+
+ if(*s && *s != '#') {
+ s = parse_token(command, TOKEN_BUFFER_SIZE, ' ', s);
+
+ if(strcmp(command, "alias") == 0) {
+ struct alias_node *a = safe_malloc(sizeof(struct alias_node));
+ a->next = global_alias_list;
+ global_alias_list = a;
+ if(s) {
+ s = eat_space(s);
+ s = parse_token(tmp_token, TOKEN_BUFFER_SIZE, '=', s);
+ a->alias = strdup(tmp_token);
+ if(s) {
+ s = eat_space(s);
+ a->value = strdup(s);
+ } else {
+ fprintf(stderr, "%s:%d syntax error, missing alias value.\n",
+ rc_filename,
+ line_number);
+ exit(EXIT_FAILURE);
+ }
+ } else {
+ fprintf(stderr, "%s:%d syntax error, missing alias key.\n",
+ rc_filename,
+ line_number);
+ exit(EXIT_FAILURE);
+ }
+ } else {
+ fprintf(stderr, "%s:%d syntax error, unknown command '%s'.\n",
+ rc_filename,
+ line_number,
+ command);
+ exit(EXIT_FAILURE);
+ }
+ }
+
+ line_number++;
+ }
+ fclose(rc_file);
+ }
+}
+
/*********************************************************************/
/*********************************************************************/
/*********************************************************************/
char *mbox_filename_regexp_string = 0;
char *default_search_field;
char output_filename[PATH_MAX + 1];
+ char rc_filename[PATH_MAX + 1];
int action_index = 0;
int error = 0, show_help = 0;
const unsigned int nb_fields_to_parse =
unsigned int f, n;
unsigned int nb_search_conditions;
struct search_condition search_conditions[MAX_NB_SEARCH_CONDITIONS];
+ struct alias_node *a, *b;
if(regcomp(&global_leading_from_line_regexp, LEADING_FROM_LINE_REGEXP_STRING, 0)) {
fprintf(stderr,
exit(EXIT_FAILURE);
}
+ if(getenv("MYMAILRC")) {
+ sprintf(rc_filename, "%s", getenv("MYMAILRC"));
+ } else if(getenv("HOME")) {
+ sprintf(rc_filename, "%s/.mymailrc", getenv("HOME"));
+ } else {
+ rc_filename[0] = '\0';
+ }
+
+ global_alias_list = 0;
global_quiet = 0;
global_use_leading_time = 0;
+ global_nb_mails_max = 250;
+
default_search_field = 0;
strncpy(output_filename, "", PATH_MAX);
+ if(rc_filename[0]) {
+ read_rc_file(rc_filename);
+ }
+
+ /*
+ {
+ #warning Test code added on 2013 May 02 11:17:01
+ struct alias_node *a;
+ for(a = global_alias_list; a; a = a->next) {
+ printf ("ALIAS [%s] [%s]\n", a->alias, a->value);
+ }
+ }
+ */
+
setlocale(LC_ALL, "");
nb_search_conditions = 0;
default_search_field = optarg;
break;
+ case 'n':
+ global_nb_mails_max = atoi(optarg);
+ break;
+
default:
error = 1;
break;
}
}
+ if(error) {
+ print_usage(stderr);
+ exit(EXIT_FAILURE);
+ }
+
+ if(show_help) {
+ print_usage(stdout);
+ exit(EXIT_SUCCESS);
+ }
+
/* Set all the values that may defined in the arguments, through
environment variables, or hard-coded */
"MYMAIL_MBOX_PATTERN",
0);
- /* Start the processing */
-
- if(error) {
- print_usage(stderr);
- exit(EXIT_FAILURE);
- }
-
- if(show_help) {
- print_usage(stdout);
- exit(EXIT_SUCCESS);
- }
-
/* mbox indexing */
if(action_index) {
}
}
- fprintf(db_file, "%s version_%s raw\n", MYMAIL_DB_MAGIC_TOKEN, VERSION);
+ fprintf(db_file,
+ "%s version_%s format_%d raw\n",
+ MYMAIL_DB_MAGIC_TOKEN,
+ MYMAIL_VERSION,
+ MYMAIL_DB_FORMAT_VERSION);
while(optind < argc) {
recursive_index_mbox(db_file,
exit(EXIT_FAILURE);
}
- nb_extracted_mails += 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_extracted_mails,
+ nb_search_conditions, search_conditions,
+ output_file);
regfree(&db_filename_regexp);
}
s = parse_token(db_filename, PATH_MAX + 1, ';', s);
if(db_filename[0]) {
- nb_extracted_mails +=
- search_in_db(db_filename, nb_search_conditions, search_conditions, output_file);
+ nb_extracted_mails =
+ search_in_db(db_filename,
+ nb_extracted_mails,
+ nb_search_conditions, search_conditions, output_file);
}
}
}
/* Search in all db files listed in the command arguments */
while(optind < argc) {
- nb_extracted_mails +=
- search_in_db(argv[optind], nb_search_conditions, search_conditions, output_file);
+ nb_extracted_mails =
+ search_in_db(argv[optind],
+ nb_extracted_mails,
+ nb_search_conditions, search_conditions, output_file);
optind++;
}
}
free_condition(&search_conditions[n]);
}
+ a = global_alias_list;
+ while(a) {
+ b = a->next;
+ free(a->alias);
+ free(a->value);
+ free(a);
+ a = b;
+ }
+
free(db_filename);
free(db_filename_regexp_string);
free(db_root_path);