#include <regex.h>
#define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file"
-#define VERSION "0.1"
+#define VERSION "0.9.1"
-#define MAX_NB_SEARCH_PATTERNS 10
+#define MAX_NB_SEARCH_CONDITIONS 10
#define BUFFER_SIZE 65536
+char *db_filename;
+char *db_filename_regexp_string;
+char *db_root_path;
+char *db_filename_list;
+char output_filename[PATH_MAX + 1];
+
+int paranoid;
+int action_index;
+int quiet;
+
+/********************************************************************/
+
enum {
- ID_MAIL,
+ ID_MAIL = 0,
ID_FROM,
- ID_DEST,
+ ID_TO,
ID_SUBJECT,
+ ID_DATE,
ID_PARTICIPANT,
+ ID_BODY,
MAX_ID
};
static char *field_names[] = {
"mail",
"from",
- "dest",
- "subj",
- "part"
+ "to",
+ "subject",
+ "date",
+ "part",
+ "body"
};
+/********************************************************************/
+
+struct search_condition {
+ 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_TO,
+ "^\\([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 }
+ },
+
+ {
+ ID_DATE,
+ "^[Dd][Aa][Tt][Ee]: ",
+ { 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++;
print_version(out);
fprintf(out, "Written by Francois Fleuret <francois@fleuret.org>.\n");
fprintf(out, "\n");
- fprintf(out, "Usage: mymail [options] [<mbox dir1> [<mbox dir2> ...]]\n");
+ fprintf(out, "Usage: mymail [options] [<mbox dir1> [<mbox dir2> ...]|<db file1> [<db file2> ...]]\n");
fprintf(out, "\n");
fprintf(out, " -h, --help\n");
fprintf(out, " show this help\n");
fprintf(out, " -v, --version\n");
fprintf(out, " print the version number\n");
- fprintf(out, " -i, --index\n");
- fprintf(out, " index mails\n");
+ fprintf(out, " -q, --quiet\n");
+ fprintf(out, " do not print information during search\n");
+ fprintf(out, " -p <db filename pattern>, --db-pattern <db filename pattern>\n");
+ fprintf(out, " set the db filename pattern for recursive search\n");
+ fprintf(out, " -r <db root path>, --db-root <db root path>\n");
+ 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, " -s <search pattern>, --search <search pattern>\n");
- fprintf(out, " search for matching mails in the data-base file\n");
+ fprintf(out, " search for matching mails in the db file\n");
fprintf(out, " -d <db filename>, --db-file <db filename>\n");
- fprintf(out, " set the data-base file\n");
- fprintf(out, " -r <db root path>, --db-root <db root path>\n");
- fprintf(out, " set the data-base root path for recursive search\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");
}
/*********************************************************************/
(name[0] == '.' && name[1] != '/');
}
-int mbox_line_match_search(int search_id, regex_t *search_regexp,
+int mbox_line_match_search(struct search_condition *condition,
int mbox_id, char *mbox_value) {
return
- (search_id == mbox_id ||
- (search_id == ID_PARTICIPANT && (mbox_id == ID_FROM || mbox_id == ID_DEST)))
+ (condition->field_id == mbox_id ||
+ (condition->field_id == ID_PARTICIPANT && (mbox_id == ID_FROM || mbox_id == ID_TO)))
&&
- regexec(search_regexp, mbox_value, 0, 0, 0) == 0;
+ regexec(&condition->regexp, mbox_value, 0, 0, 0) == 0;
}
-void search_in_db(int nb_search_patterns,
- int *search_ids, char **search_regexp_strings,
- FILE *db_file) {
- int hits[MAX_NB_SEARCH_PATTERNS];
+void search_in_db(FILE *db_file,
+ 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, *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;
+ int nb_body_conditions, nb_fulfilled_body_conditions;
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_conditions; n++) { hits[n] = 0; }
+
+ nb_body_conditions = 0;
+ for(n = 0; n < nb_search_conditions; n++) {
+ if(search_conditions[n].field_id == ID_BODY) {
+ nb_body_conditions++;
+ }
+ }
+
+ strcpy(current_mail_filename, "");
while(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
mbox_name = raw_db_line;
char *position_in_file_string;
char *mail_filename;
- for(n = 0; n < nb_search_patterns && hits[n]; n++);
+ if(current_mail_filename[0]) {
- if(n == nb_search_patterns) {
- FILE *mail_file;
+ /* We first check all conditions but the body ones */
- mail_file = fopen(current_mail_filename, "r");
+ for(n = 0; n < nb_search_conditions &&
+ ((search_conditions[n].field_id == ID_BODY) ||
+ xor(hits[n], search_conditions[n].negation)); n++);
- if(!mail_file) {
- fprintf(stderr, "mymail: Cannot open mbox '%s'.\n", current_mail_filename);
- exit(EXIT_FAILURE);
- }
+ if(n == nb_search_conditions) {
+
+ /* all conditions but the body ones are fine, check the body
+ ones */
+
+ nb_fulfilled_body_conditions = 0;
+
+ if(nb_body_conditions > 0) {
+ FILE *mail_file;
+ int header;
- fseek(mail_file, current_position_in_mail, SEEK_SET);
+ header = 1;
+ mail_file = fopen(current_mail_filename, "r");
- if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
- printf("%s", raw_mbox_line);
- while(fgets(raw_mbox_line, BUFFER_SIZE, mail_file) &&
- strncmp(raw_mbox_line, "From ", 5)) {
- printf("%s", raw_mbox_line);
+ if(!mail_file) {
+ fprintf(stderr,
+ "mymail: Cannot open mbox '%s' for body scan.\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)) {
+ while(nb_fulfilled_body_conditions < nb_body_conditions) {
+ last_mbox_line_was_empty = (raw_mbox_line[0] == '\n');
+
+ if(last_mbox_line_was_empty) { header = 0; }
+
+ if(!header) {
+ for(n = 0; n < nb_search_conditions; n++) {
+ if(search_conditions[n].field_id == ID_BODY && !hits[n]) {
+ hits[n] =
+ (regexec(&search_conditions[n].regexp, raw_mbox_line, 0, 0, 0) == 0);
+ if(hits[n]) {
+ nb_fulfilled_body_conditions++;
+ }
+ }
+ }
+ }
+
+ if(!fgets(raw_mbox_line, BUFFER_SIZE, mail_file) ||
+ (last_mbox_line_was_empty && strncmp(raw_mbox_line, "From ", 5) == 0))
+ break;
+ }
+ }
+
+ fclose(mail_file);
}
- }
- fclose(mail_file);
+ 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 = 1;
+ fprintf(output_file, "%s", raw_mbox_line);
+ 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');
+ fprintf(output_file, "%s", raw_mbox_line);
+ }
+ }
+
+ fclose(mail_file);
+ }
+ }
}
- for(n = 0; n < nb_search_patterns; n++) { hits[n] = 0; }
+ for(n = 0; n < nb_search_conditions; n++) { hits[n] = 0; }
position_in_file_string = mbox_value;
mail_filename = segment_next_field(mbox_value);
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_conditions; n++) {
+ hits[n] |= mbox_line_match_search(&search_conditions[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) {
+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) {
DIR *dir;
struct dirent *dir_e;
struct stat sb;
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,
- nb_search_patterns,
- search_ids, search_regexp_strings);
+ recursive_search_in_db(subname, db_filename_regexp,
+ nb_search_conditions, search_conditions,
+ output_file);
}
}
closedir(dir);
- } else {
+ }
+
+ else {
const char *s = entry_name, *filename = entry_name;
while(*s) { if(*s == '/') { filename = s+1; } s++; }
- if(strcmp(filename, db_filename) == 0) {
+ if(regexec(db_filename_regexp, filename, 0, 0, 0) == 0) {
FILE *db_file = fopen(entry_name, "r");
+ if(!quiet) {
+ printf("Searching in '%s' ... ", entry_name);
+ }
+
if(!db_file) {
fprintf(stderr,
"mymail: Cannot open \"%s\" for reading: %s\n",
exit(EXIT_FAILURE);
}
- search_in_db(nb_search_patterns, search_ids, search_regexp_strings,
- db_file);
+ search_in_db(db_file, nb_search_conditions, search_conditions, output_file);
fclose(db_file);
+
+ if(!quiet) {
+ printf("done.\n");
+ }
}
}
}
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");
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",
}
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);
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-pattern", 1, 0, 'p' },
{ "db-root", 1, 0, 'r' },
+ { "db-list", 1, 0, 'l' },
{ "search", 1, 0, 's' },
{ "index", 0, 0, 'i' },
+ { "output", 1, 0, 'o' },
{ 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) {
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];
-
- /* for(f = 0; f < argc; f++) { */
- /* printf("arg %d \"%s\"\n", f, argv[f]); */
- /* } */
+ int nb_search_conditions;
+ char *search_condition_strings[MAX_NB_SEARCH_CONDITIONS];
+ FILE *output_file;
paranoid = 0;
action_index = 0;
db_filename = 0;
db_root_path = 0;
+ db_filename_list = 0;
+ quiet = 0;
setlocale(LC_ALL, "");
- nb_search_patterns = 0;
+ nb_search_conditions = 0;
- while ((c = getopt_long(argc, argv, "hvip:s:d:r:",
+ while ((c = getopt_long(argc, argv, "hvqip:s:d:r:l:o:",
long_options, NULL)) != -1) {
switch(c) {
print_version(stdout);
break;
+ case 'q':
+ quiet = 1;
+ break;
+
case 'i':
action_index = 1;
break;
db_filename = strdup(optarg);
break;
+ case 'p':
+ db_filename_regexp_string = strdup(optarg);
+ break;
+
+ case 'o':
+ strncpy(output_filename, optarg, PATH_MAX);
+ break;
+
case 'r':
db_root_path = strdup(optarg);
break;
+ case 'l':
+ db_filename_list = strdup(optarg);
+ break;
+
case 's':
- if(nb_search_patterns == MAX_NB_SEARCH_PATTERNS) {
+ if(nb_search_conditions == MAX_NB_SEARCH_CONDITIONS) {
fprintf(stderr, "mymail: Too many search patterns.\n");
exit(EXIT_FAILURE);
}
- search_pattern[nb_search_patterns++] = strdup(optarg);
+ search_condition_strings[nb_search_conditions++] = strdup(optarg);
break;
default:
db_filename = strdup(default_db_filename);
}
+ if(!db_filename_regexp_string) {
+ char *default_db_filename_regexp_string = getenv("MYMAIL_DB_PATTERN");
+
+ if(!default_db_filename_regexp_string) {
+ default_db_filename_regexp_string = "^mymail.db$";
+ }
+
+ db_filename_regexp_string = strdup(default_db_filename_regexp_string);
+ }
+
if(!db_root_path) {
char *default_db_root_path = getenv("MYMAIL_DB_ROOT");
}
}
- if(!db_root_path) {
- fprintf(stderr,
- "mymail: db root path is not set\n");
- exit(EXIT_FAILURE);
+ if(!db_filename_list) {
+ char *default_db_filename_list = getenv("MYMAIL_DB_LIST");
+
+ if(default_db_filename_list) {
+ db_filename_list = strdup(default_db_filename_list);
+ }
}
+ 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(error) {
print_usage(stderr);
}
}
- 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,
else {
- if(nb_search_patterns > 0) {
- int search_ids[MAX_NB_SEARCH_PATTERNS];
- char *search_regexp_strings[MAX_NB_SEARCH_PATTERNS];
+ if(nb_search_conditions > 0) {
+ struct search_condition search_conditions[MAX_NB_SEARCH_CONDITIONS];
+ 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_conditions; n++) {
+ search_field = search_condition_strings[n];
+ search_regexp_string = segment_next_field(search_condition_strings[n]);
+
+ if(search_field[0] == '!') {
+ search_field++;
+ search_conditions[n].negation = 1;
+ } else {
+ search_conditions[n].negation = 0;
+ }
+
+ search_conditions[n].field_id = -1;
+ for(m = 0; (m < MAX_ID) && search_conditions[n].field_id == -1; m++) {
+ if(strncmp(field_names[m], search_field, strlen(search_field)) == 0) {
+ search_conditions[n].field_id = m;
}
}
+
+ if(search_conditions[n].field_id == -1) {
+ fprintf(stderr,
+ "mymail: Syntax error in field name \"%s\".\n",
+ search_field);
+ exit(EXIT_FAILURE);
+ }
+
+ if(regcomp(&search_conditions[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_conditions[n].field_id]);
+ exit(EXIT_FAILURE);
+ }
}
- if(!*search_regexp_strings) {
- fprintf(stderr,
- "Syntax error in the search pattern.\n");
- exit(EXIT_FAILURE);
+ /* Recursive search if db_root_path is set */
+
+ if(db_root_path) {
+ regex_t db_filename_regexp;
+ if(regcomp(&db_filename_regexp,
+ db_filename_regexp_string,
+ 0)) {
+ fprintf(stderr,
+ "mymail: Syntax error in regexp \"%s\".\n",
+ db_filename_regexp_string);
+ exit(EXIT_FAILURE);
+ }
+
+ recursive_search_in_db(db_root_path, &db_filename_regexp,
+ nb_search_conditions, search_conditions,
+ output_file);
+
+ regfree(&db_filename_regexp);
}
- recursive_search_in_db(db_root_path,
- nb_search_patterns, search_ids, search_regexp_strings);
+ /* Search in all db files listed in db_filename_list */
+
+ if(db_filename_list) {
+ char db_filename[PATH_MAX + 1];
+ char *s, *t;
+ FILE *db_file;
- for(n = 0; n < nb_search_patterns; n++) {
- free(search_pattern[n]);
+ s = db_filename_list;
+
+ while(*s) {
+ t = db_filename;
+ while(*s == ';') { s++; }
+ while(*s && *s != ';') { *t++ = *s++; }
+ *t++ = '\0';
+
+ if(db_filename[0]) {
+ db_file = fopen(db_filename, "r");
+
+ if(!db_file) {
+ fprintf(stderr,
+ "mymail: Cannot open \"%s\" for reading: %s\n",
+ argv[optind],
+ strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
+ search_in_db(db_file, nb_search_conditions, search_conditions, output_file);
+
+ fclose(db_file);
+ }
+ }
+ }
+
+ /* Search in all db files listed in the command arguments */
+
+ while(optind < argc) {
+ FILE *db_file = fopen(argv[optind], "r");
+
+ if(!db_file) {
+ fprintf(stderr,
+ "mymail: Cannot open \"%s\" for reading: %s\n",
+ argv[optind],
+ strerror(errno));
+ exit(EXIT_FAILURE);
+ }
+
+ search_in_db(db_file, nb_search_conditions, search_conditions, output_file);
+
+ fclose(db_file);
+ optind++;
+ }
+
+ for(n = 0; n < nb_search_conditions; n++) {
+ regfree(&search_conditions[n].regexp);
+ free(search_condition_strings[n]);
}
}
}
+ if(output_file != stdout) {
+ fclose(output_file);
+ }
+
free(db_filename);
+ free(db_filename_regexp_string);
free(db_root_path);
+ free(db_filename_list);
exit(EXIT_SUCCESS);
}