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.
#include <time.h>
#define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file"
-#define VERSION "0.9.1"
+#define VERSION "0.9.2"
-#define MAX_NB_SEARCH_CONDITIONS 10
+#define MAX_NB_SEARCH_CONDITIONS 32
#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$"
+
regex_t leading_from_line_regexp;
-char *db_filename;
-char *db_filename_regexp_string;
-char *db_root_path;
-char *db_filename_list;
-char output_filename[PATH_MAX + 1];
+/* Global variables! */
int paranoid;
-int action_index;
int quiet;
-
-time_t being_today;
+char *default_search_field;
+int ignore_dot_files;
/********************************************************************/
struct parsable_field {
int id;
+ int cflags;
char *regexp_string;
regex_t regexp;
};
static struct parsable_field fields_to_parse[] = {
{
ID_LEADING_LINE,
+ 0,
"^From ",
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
},
{
ID_FROM,
- "^\\([Ff][Rr][Oo][Mm]:\\|[Rr][Ee][Pp][Ll][Yy]-[Tt][Oo]:\\|[Ss][Ee][Nn][Dd][Ee][Rr]:\\)",
+ REG_ICASE,
+ "^\\(from\\|reply-to\\|sender\\|return-path\\): ",
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
},
{
ID_TO,
- "^\\([Tt][Oo]\\|[Cc][Cc]\\|[Bb][Cc][Cc]\\): ",
+ REG_ICASE,
+ "^\\(to\\|cc\\|bcc\\): ",
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
},
{
ID_SUBJECT,
- "^[Ss][Uu][Bb][Jj][Ee][Cc][Tt]: ",
+ REG_ICASE,
+ "^subject: ",
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
},
{
ID_DATE,
- "^[Dd][Aa][Tt][Ee]: ",
+ REG_ICASE,
+ "^date: ",
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
},
return string;
}
-void remove_eof(char *c) {
- while(*c && *c != '\n' && *c != '\r') c++;
- *c = '\0';
-}
-
/********************************************************************/
/* malloc with error checking. */
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, " -a <search field>, --default-search <search field>\n");
+ fprintf(out, " set the default search field\n");
+}
+
+/*********************************************************************/
+
+time_t time_for_past_day(int day) {
+ time_t t;
+ struct tm *tm;
+ int delta_day;
+ t = time(0);
+ tm = localtime(&t);
+ delta_day = (7 + tm->tm_wday - day) % 7 + 1;
+ return t - delta_day * 3600 * 24 + tm->tm_sec + 60 * tm->tm_min + 3600 * tm->tm_hour;
}
/*********************************************************************/
int ignore_entry(const char *name) {
return
- /* strcmp(name, ".") == 0 || */
- /* strcmp(name, "..") == 0 || */
- (name[0] == '.' && name[1] != '/');
+ strcmp(name, ".") == 0 ||
+ strcmp(name, "..") == 0 ||
+ (ignore_dot_files && name[0] == '.' && name[1] != '/');
}
-int is_a_leading_from_line(char *s) {
- return strncmp(s, "From ", 5) == 0 &&
- regexec(&leading_from_line_regexp, s, 0, 0, 0) == 0;
+int is_a_leading_from_line(char *mbox_line) {
+ return
+
+ /*
+
+ The mbox man page in qmail documentation states:
+
+ > The reader should not attempt to take advantage of the fact
+ > that every From_ line (past the beginning of the file) is
+ > preceded by a blank line.
+
+ */
+
+ strncmp(mbox_line, "From ", 5) == 0 &&
+ regexec(&leading_from_line_regexp, mbox_line, 0, 0, 0) == 0;
}
int mbox_line_match_search(struct search_condition *condition,
(condition->field_id == ID_FROM && mbox_id == ID_LEADING_LINE)
)
+
&&
+
regexec(&condition->regexp, mbox_value, 0, 0, 0) == 0;
}
}
int *hits) {
FILE *mail_file;
int header, n;
- int last_mbox_line_was_empty;
char raw_mbox_line[BUFFER_SIZE];
int nb_body_hits;
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');
+ /* last_mbox_line_was_empty = (raw_mbox_line[0] == '\n'); */
+ /* if(last_mbox_line_was_empty) { header = 0; } */
- if(last_mbox_line_was_empty) { header = 0; }
+ if(raw_mbox_line[0] == '\n') { header = 0; }
if(!header) {
for(n = 0; n < nb_search_conditions; n++) {
}
if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
- (last_mbox_line_was_empty &&
- is_a_leading_from_line(raw_mbox_line)))
+ (is_a_leading_from_line(raw_mbox_line)))
break;
}
}
fclose(mail_file);
}
-void search_in_db(const char *db_filename,
- int nb_search_conditions,
- struct search_condition *search_conditions,
- FILE *output_file) {
+void extract_mail(const char *mail_filename, unsigned long int position_in_mail,
+ FILE *output_file) {
+ char raw_mbox_line[BUFFER_SIZE];
+ FILE *mail_file;
+
+ mail_file = fopen(mail_filename, "r");
+
+ if(!mail_file) {
+ fprintf(stderr,
+ "mymail: Cannot open mbox '%s' for mail extraction.\n",
+ mail_filename);
+ exit(EXIT_FAILURE);
+ }
+
+ fseek(mail_file, position_in_mail, SEEK_SET);
+
+ if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
+ 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))
+ )
+ break;
+ fprintf(output_file, "%s", raw_mbox_line);
+ }
+ }
+
+ fclose(mail_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];
- char raw_mbox_line[BUFFER_SIZE];
char current_mail_filename[PATH_MAX + 1];
unsigned long int current_position_in_mail;
char mbox_name[TOKEN_BUFFER_SIZE], *mbox_value;
int mbox_id;
int already_written, m, n;
- int last_mbox_line_was_empty;
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);
exit(EXIT_FAILURE);
}
+ /* First, check the db file leading line integrity */
+
if(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
if(strncmp(raw_db_line, MYMAIL_DB_MAGIC_TOKEN, strlen(MYMAIL_DB_MAGIC_TOKEN))) {
fprintf(stderr,
exit(EXIT_FAILURE);
}
+ /* Then parse the said db file */
+
current_position_in_mail = 0;
already_written = 0;
if(n == nb_search_conditions) {
- /* all conditions but the body ones are fine, check the body
- ones */
+ /* Now check the body ones */
if(nb_body_conditions > 0) {
update_body_hits(current_mail_filename, current_position_in_mail,
}
if(nb_body_conditions == nb_fulfilled_body_conditions) {
- FILE *mail_file;
-
- mail_file = fopen(current_mail_filename, "r");
-
- if(!mail_file) {
- fprintf(stderr,
- "mymail: Cannot open mbox '%s' for mail extraction.\n",
- current_mail_filename);
- exit(EXIT_FAILURE);
- }
-
- fseek(mail_file, current_position_in_mail, SEEK_SET);
-
- if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
- last_mbox_line_was_empty = 0;
- fprintf(output_file, "%s", raw_mbox_line);
- while(1) {
- if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
- (last_mbox_line_was_empty &&
- is_a_leading_from_line(raw_mbox_line))
- )
- break;
- last_mbox_line_was_empty = (raw_mbox_line[0] == '\n');
- fprintf(output_file, "%s", raw_mbox_line);
- }
- }
-
- fclose(mail_file);
+ nb_extracted_mails++;
+ extract_mail(current_mail_filename, current_position_in_mail, output_file);
}
}
}
for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
mbox_value = parse_token(position_in_file_string, TOKEN_BUFFER_SIZE, ' ', mbox_value);
- mbox_value = parse_token(current_mail_filename, TOKEN_BUFFER_SIZE, ' ', mbox_value);
+ mbox_value = parse_token(current_mail_filename, TOKEN_BUFFER_SIZE, '\n', mbox_value);
current_position_in_mail = atol(position_in_file_string);
- remove_eof(current_mail_filename);
already_written = 0;
}
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;
}
/*********************************************************************/
char raw_mbox_line[BUFFER_SIZE], full_line[BUFFER_SIZE];
char *end_of_full_line;
FILE *file;
- int in_header, new_header, last_mbox_line_was_empty;
+ int in_header, new_header;
unsigned long int position_in_file;
file = fopen(mbox_filename, "r");
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(last_mbox_line_was_empty &&
- is_a_leading_from_line(raw_mbox_line)) {
+ if(is_a_leading_from_line(raw_mbox_line)) {
if(in_header) {
fprintf(stderr,
"Got a ^\"From \" in the header in %s:%lu.\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);
{ "search", 1, 0, 's' },
{ "index", 0, 0, 'i' },
{ "output", 1, 0, 'o' },
+ { "default-search", 1, 0, 'a' },
{ 0, 0, 0, 0 }
};
/*********************************************************************/
-void init_condition(struct search_condition *condition, char *string) {
+void init_condition(struct search_condition *condition, char *full_string) {
char full_search_field[TOKEN_BUFFER_SIZE], *search_field;
int m;
+ char *string;
- string = parse_token(full_search_field, TOKEN_BUFFER_SIZE, ' ', string);
+ string = parse_token(full_search_field, TOKEN_BUFFER_SIZE, ' ', full_string);
search_field = full_search_field;
if(search_field[0] == '!') {
condition->negation = 0;
}
- if(strcmp(search_field, "6h") == 0) {
+ /* Recently */
+
+ if(strcmp(search_field, "8h") == 0) {
condition->field_id = ID_INTERVAL;
- condition->interval_start = time(0) - 3600 * 6;
+ condition->interval_start = time(0) - 3600 * 8;
condition->interval_stop = 0;
}
condition->interval_stop = 0;
}
+ else if(strcmp(search_field, "week") == 0) {
+ condition->field_id = ID_INTERVAL;
+ condition->interval_start = time(0) - 3600 * 24 * 7;
+ condition->interval_stop = 0;
+ }
+
+ else if(strcmp(search_field, "month") == 0) {
+ condition->field_id = ID_INTERVAL;
+ condition->interval_start = time(0) - 3600 * 24 * 31;
+ condition->interval_stop = 0;
+ }
+
+ else if(strcmp(search_field, "year") == 0) {
+ condition->field_id = ID_INTERVAL;
+ condition->interval_start = time(0) - 3600 * 24 * 365;
+ condition->interval_stop = 0;
+ }
+
+ /* Yesterday */
+
else if(strcmp(search_field, "yesterday") == 0) {
condition->field_id = ID_INTERVAL;
condition->interval_start = time(0) - 2 * 3600 * 24;
- condition->interval_stop = time(0) - 3600 * 24;
+ condition->interval_stop = condition->interval_start + 3600 * 24;
+ }
+
+ /* Week days */
+
+ else if(strcmp(search_field, "monday") == 0) {
+ condition->field_id = ID_INTERVAL;
+ condition->interval_start = time_for_past_day(1);
+ condition->interval_stop = condition->interval_start + 3600 * 24;
+ }
+
+ else if(strcmp(search_field, "tuesday") == 0) {
+ condition->field_id = ID_INTERVAL;
+ condition->interval_start = time_for_past_day(2);
+ condition->interval_stop = condition->interval_start + 3600 * 24;
+ }
+
+ else if(strcmp(search_field, "wednesday") == 0) {
+ condition->field_id = ID_INTERVAL;
+ condition->interval_start = time_for_past_day(3);
+ condition->interval_stop = condition->interval_start + 3600 * 24;
+ }
+
+ else if(strcmp(search_field, "thursday") == 0) {
+ condition->field_id = ID_INTERVAL;
+ condition->interval_start = time_for_past_day(4);
+ condition->interval_stop = condition->interval_start + 3600 * 24;
+ }
+
+ else if(strcmp(search_field, "friday") == 0) {
+ condition->field_id = ID_INTERVAL;
+ condition->interval_start = time_for_past_day(5);
+ condition->interval_stop = condition->interval_start + 3600 * 24;
+ }
+
+ else if(strcmp(search_field, "saturday") == 0) {
+ condition->field_id = ID_INTERVAL;
+ condition->interval_start = time_for_past_day(6);
+ condition->interval_stop = condition->interval_start + 3600 * 24;
+ }
+
+ else if(strcmp(search_field, "sunday") == 0) {
+ condition->field_id = ID_INTERVAL;
+ condition->interval_start = time_for_past_day(7);
+ condition->interval_stop = condition->interval_start + 3600 * 24;
}
else {
+
+ /* header-related conditions */
+
condition->field_id = -1;
for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) {
}
}
+ if(condition->field_id == -1) {
+ if(default_search_field) {
+ for(m = 0; (m < MAX_ID) && condition->field_id == -1; m++) {
+ if(strncmp(field_names[m],
+ default_search_field, strlen(default_search_field)) == 0) {
+ condition->field_id = m;
+ }
+ }
+ string = full_string;
+ }
+ }
+
if(condition->field_id == -1) {
fprintf(stderr,
"mymail: Syntax error in field name \"%s\".\n",
/*********************************************************************/
int main(int argc, char **argv) {
+ char *db_filename;
+ char *db_filename_regexp_string;
+ char *db_root_path;
+ char *db_filename_list;
+ char output_filename[PATH_MAX + 1];
+ int action_index;
int error = 0, show_help = 0;
const int nb_fields_to_parse = sizeof(fields_to_parse) / sizeof(struct parsable_field);
char c;
FILE *output_file;
struct search_condition search_conditions[MAX_NB_SEARCH_CONDITIONS];
- if(regcomp(&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$",
- 0)) {
+ if(regcomp(&leading_from_line_regexp, LEADING_FROM_LINE_REGEXP, 0)) {
fprintf(stderr,
"mymail: Cannot compile leading \"from\" line regexp. That is strange.\n");
exit(EXIT_FAILURE);
paranoid = 0;
action_index = 0;
db_filename = 0;
+ db_filename_regexp_string = 0;
db_root_path = 0;
db_filename_list = 0;
quiet = 0;
+ default_search_field = 0;
+ ignore_dot_files = 1;
setlocale(LC_ALL, "");
nb_search_conditions = 0;
- while ((c = getopt_long(argc, argv, "hvqip:s:d:r:l:o:",
+ while ((c = getopt_long(argc, argv, "hvqip:s:d:r:l:o:a:",
long_options, NULL)) != -1) {
switch(c) {
nb_search_conditions++;
break;
+ case 'a':
+ default_search_field = optarg;
+ break;
+
default:
error = 1;
break;
for(f = 0; f < nb_fields_to_parse; f++) {
if(regcomp(&fields_to_parse[f].regexp,
fields_to_parse[f].regexp_string,
- REG_ICASE)) {
+ fields_to_parse[f].cflags)) {
fprintf(stderr,
"mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
fields_to_parse[f].regexp_string,
}
else {
+ int nb_extracted_mails = 0;
if(nb_search_conditions > 0) {
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");
+ }
+ }
+
}
for(n = 0; n < nb_search_conditions; n++) {