#include <regex.h>
#define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file"
-#define VERSION "0.1"
+#define VERSION "0.9"
#define MAX_NB_SEARCH_PATTERNS 10
#define BUFFER_SIZE 65536
+char *db_filename;
+char *db_root_path;
+
+int paranoid;
+int action_index;
+
+/********************************************************************/
+
+enum {
+ ID_MAIL = 0,
+ ID_FROM,
+ ID_DEST,
+ ID_SUBJECT,
+ ID_PARTICIPANT,
+ MAX_ID
+};
+
+static char *field_names[] = {
+ "mail",
+ "from",
+ "dest",
+ "subject",
+ "part"
+};
+
+/********************************************************************/
+
+struct search_request {
+ int field_id;
+ int negation;
+ regex_t regexp;
+};
+
+/********************************************************************/
+
struct parsable_field {
- char *name;
+ 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_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 xor(int a, int b) {
+ return (a && !b) || (!a && b);
+}
char *segment_next_field(char *current) {
while(*current && *current != ' ') current++;
(name[0] == '.' && name[1] != '/');
}
+int mbox_line_match_search(struct search_request *request,
+ int mbox_id, char *mbox_value) {
+ return
+ (request->field_id == mbox_id ||
+ (request->field_id == ID_PARTICIPANT && (mbox_id == ID_FROM || mbox_id == ID_DEST)))
+ &&
+ regexec(&request->regexp, mbox_value, 0, 0, 0) == 0;
+}
+
void search_in_db(int nb_search_patterns,
- char **search_name, char **search_regexp_string,
+ struct search_request *search_requests,
FILE *db_file) {
int hits[MAX_NB_SEARCH_PATTERNS];
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 *name, *value;
- regex_t regexp[MAX_NB_SEARCH_PATTERNS];
- int already_written, n;
-
- for(n = 0; n < nb_search_patterns; n++) {
- if(regcomp(®exp[n],
- search_regexp_string[n],
- REG_ICASE)) {
- fprintf(stderr,
- "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
- search_regexp_string[n],
- search_name[n]);
- exit(EXIT_FAILURE);
- }
- }
+ char *mbox_name, *mbox_value;
+ int mbox_id;
+ int already_written, m, n;
+ int last_mbox_line_was_empty;
current_position_in_mail = 0;
already_written = 0;
for(n = 0; n < nb_search_patterns; n++) { hits[n] = 0; }
while(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
- name = raw_db_line;
- value = segment_next_field(raw_db_line);
+ mbox_name = raw_db_line;
+ mbox_value = segment_next_field(raw_db_line);
- if(strcmp("mail", name) == 0) {
+ if(strcmp("mail", mbox_name) == 0) {
char *position_in_file_string;
char *mail_filename;
- for(n = 0; n < nb_search_patterns && hits[n]; n++);
+ for(n = 0; n < nb_search_patterns && xor(hits[n], search_requests[n].negation); n++);
+
+ /* for(n = 0; n < nb_search_patterns && */
+ /* ((hits[n] && !search_requests[n].negation) || */
+ /* (!hits[n] && search_requests[n].negation)); n++); */
if(n == nb_search_patterns) {
FILE *mail_file;
+
mail_file = fopen(current_mail_filename, "r");
+
if(!mail_file) {
fprintf(stderr, "mymail: Cannot open mbox '%s'.\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;
printf("%s", raw_mbox_line);
- while(fgets(raw_mbox_line, BUFFER_SIZE, mail_file) &&
- strncmp(raw_mbox_line, "From ", 5)) {
+ 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');
printf("%s", raw_mbox_line);
}
}
+
fclose(mail_file);
}
for(n = 0; n < nb_search_patterns; n++) { hits[n] = 0; }
- position_in_file_string = value;
- mail_filename = segment_next_field(value);
+ position_in_file_string = mbox_value;
+ mail_filename = segment_next_field(mbox_value);
current_position_in_mail = atol(position_in_file_string);
strcpy(current_mail_filename, mail_filename);
}
else {
+ mbox_id = -1;
+ for(m = 0; (m < MAX_ID) && mbox_id == -1; m++) {
+ if(strncmp(field_names[m], mbox_name, strlen(mbox_name)) == 0) {
+ mbox_id = m;
+ }
+ }
for(n = 0; n < nb_search_patterns; n++) {
- hits[n] |=
- (strncmp(search_name[n], name, strlen(search_name[n])) == 0 &&
- regexec(®exp[n], value, 0, 0, 0) == 0);
+ hits[n] |= mbox_line_match_search(&search_requests[n],
+ mbox_id, mbox_value);
}
}
}
-
- for(n = 0; n < nb_search_patterns; n++) {
- regfree(®exp[n]);
- }
}
void recursive_search_in_db(const char *entry_name,
int nb_search_patterns,
- char **search_name, char **search_regexp_string) {
+ struct search_request *search_requests) {
DIR *dir;
struct dirent *dir_e;
struct stat sb;
snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
recursive_search_in_db(subname,
nb_search_patterns,
- search_name, search_regexp_string);
+ search_requests);
}
}
closedir(dir);
exit(EXIT_FAILURE);
}
- search_in_db(nb_search_patterns, search_name, search_regexp_string,
- db_file);
+ search_in_db(nb_search_patterns, search_requests, db_file);
fclose(db_file);
}
for(f = 0; f < nb_fields_to_parse; f++) {
if(regexec(&fields_to_parse[f].regexp, raw_mbox_line, 1, &matches, 0) == 0) {
fprintf(db_file, "%s %s\n",
- fields_to_parse[f].name,
+ field_names[fields_to_parse[f].id],
raw_mbox_line + matches.rm_eo);
}
}
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);
}
else {
- /* if(!((raw_mbox_line[0] >= 'a' && raw_mbox_line[0] <= 'z') || */
- /* (raw_mbox_line[0] >= 'A' && raw_mbox_line[0] <= 'Z'))) { */
- /* fprintf(stderr, */
- /* "Header line syntax error %s:%lu.\n", */
- /* mbox_filename, position_in_file); */
- /* fprintf(stderr, "%s", raw_mbox_line); */
- /* } */
+ /*
+ if(!((raw_mbox_line[0] >= 'a' && raw_mbox_line[0] <= 'z') ||
+ (raw_mbox_line[0] >= 'A' && raw_mbox_line[0] <= 'Z'))) {
+ fprintf(stderr,
+ "Header line syntax error %s:%lu.\n",
+ mbox_filename, position_in_file);
+ fprintf(stderr, "%s", raw_mbox_line);
+ }
+ */
if(full_line[0]) {
index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
{ 0, 0, 0, 0 }
};
-static struct parsable_field fields_to_parse[] = {
- {
- "from",
- "^\\([Ff][Rr][Oo][Mm]:\\|From\\) *",
- { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
- },
-
- {
- "dest",
- "^\\([Tt][Oo]\\|[Cc][Cc]\\|[Bb][Cc][Cc]\\): *",
- { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
- },
-
- {
- "subj",
- "^[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) {
char *search_pattern[MAX_NB_SEARCH_PATTERNS];
/* for(f = 0; f < argc; f++) { */
- /* printf("arg %d \"%s\"\n", f, argv[f]); */
+ /* printf("arg %d \"%s\"\n", f, argv[f]); */
/* } */
paranoid = 0;
}
}
- if(!db_root_path) {
- fprintf(stderr,
- "mymail: db root path is not set\n");
- exit(EXIT_FAILURE);
- }
-
-
if(error) {
print_usage(stderr);
exit(EXIT_FAILURE);
fprintf(stderr,
"mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
fields_to_parse[f].regexp_string,
- fields_to_parse[f].name);
+ field_names[fields_to_parse[f].id]);
exit(EXIT_FAILURE);
}
}
- 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(!db_root_path) {
+ fprintf(stderr,
+ "mymail: db root path is not set\n");
+ exit(EXIT_FAILURE);
+ }
+
if(nb_search_patterns > 0) {
- char *search_name[MAX_NB_SEARCH_PATTERNS];
- char *search_regexp_string[MAX_NB_SEARCH_PATTERNS];
- int n;
+ struct search_request search_requests[MAX_NB_SEARCH_PATTERNS];
+ char *search_regexp_string;
+ int m, n;
for(n = 0; n < nb_search_patterns; n++) {
- search_name[n] = search_pattern[n];
- search_regexp_string[n] = segment_next_field(search_pattern[n]);
- }
+ search_regexp_string = segment_next_field(search_pattern[n]);
- if(!*search_regexp_string) {
- fprintf(stderr,
- "Syntax error in the search pattern.\n");
- exit(EXIT_FAILURE);
+ if(search_pattern[n][0] == '!') {
+ search_pattern[n]++;
+ search_requests[n].negation = 1;
+ } else {
+ search_requests[n].negation = 0;
+ }
+
+ search_requests[n].field_id = -1;
+ for(m = 0; (m < MAX_ID) && search_requests[n].field_id == -1; m++) {
+ if(strncmp(field_names[m], search_pattern[n], strlen(search_pattern[n])) == 0) {
+ search_requests[n].field_id = m;
+ }
+ }
+
+ if(regcomp(&search_requests[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_requests[n].field_id]);
+ exit(EXIT_FAILURE);
+ }
}
recursive_search_in_db(db_root_path,
- nb_search_patterns, search_name, search_regexp_string);
+ nb_search_patterns, search_requests);
for(n = 0; n < nb_search_patterns; n++) {
free(search_pattern[n]);
projects / mymail.git / blobdiff