*
*/
+/*
+
+ 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.
+
+ It is low-tech, simple, light and fast.
+
+*/
+
#define _GNU_SOURCE
#include <stdio.h>
#define VERSION "0.1"
-#define BUFFER_SIZE 16384
+#define BUFFER_SIZE 65536
struct parsable_field {
char *name;
char *db_filename;
char *search_pattern;
+int paranoid;
int action_index;
char *segment_next_field(char *current) {
/*********************************************************************/
-void usage(FILE *out) {
+void print_version(FILE *out) {
fprintf(out, "mymail version %s (%s)\n", VERSION, UNAME);
+}
+
+void print_usage(FILE *out) {
+ print_version(out);
fprintf(out, "Written by Francois Fleuret <francois@fleuret.org>.\n");
fprintf(out, "\n");
fprintf(out, "Usage: mymail [options] [<filename1> [<filename2> ...]]\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, " -d <db filename>, --db-file <db filename>\n");
+ fprintf(out, " set the data-base file\n");
+ fprintf(out, " -s <search pattern>, --search <search pattern>\n");
+ fprintf(out, " search for matching mails in the data-base file\n");
}
/*********************************************************************/
FILE *db_file) {
char raw_line[BUFFER_SIZE];
char current_mail_filename[BUFFER_SIZE];
- unsigned int current_position_in_mail;
+ unsigned long int current_position_in_mail;
char *name, *value;
regex_t regexp;
int already_written;
name = raw_line;
value = segment_next_field(raw_line);
- /* printf("LINE [%s] %s", name, value); */
-
if(strcmp("mail", name) == 0) {
char *position_in_file_string = value;
char *mail_filename = segment_next_field(value);
- current_position_in_mail = atoi(position_in_file_string);
+ current_position_in_mail = atol(position_in_file_string);
strcpy(current_mail_filename, mail_filename);
remove_eof(current_mail_filename);
- /* printf("READING [%s]\n", current_mail_filename); */
already_written = 0;
- } else if(!already_written) {
+ }
+
+ else if(!already_written) {
if(strcmp(search_name, name) == 0 && regexec(®exp, value, 0, 0, 0) == 0) {
FILE *mail_file;
- /* printf("%s:%u\n", current_mail_filename, current_position_in_mail); */
mail_file = fopen(current_mail_filename, "r");
if(!mail_file) {
- fprintf(stderr, "mymail: Can not open `%s'.\n", current_mail_filename);
+ fprintf(stderr, "mymail: Can not open '%s'.\n", current_mail_filename);
exit(EXIT_FAILURE);
}
fseek(mail_file, current_position_in_mail, SEEK_SET);
}
}
}
-}
+ regfree(®exp);
+}
/*********************************************************************/
-void read_file(const char *input_filename,
- int nb_fields_to_parse, struct parsable_field *fields_to_parse,
- FILE *db_file) {
- char raw_line[BUFFER_SIZE];
+void index_one_mbox_line(int nb_fields_to_parse, struct parsable_field *fields_to_parse,
+ char *raw_line, FILE *db_file) {
+ regmatch_t matches;
+ int f;
+ for(f = 0; f < nb_fields_to_parse; f++) {
+ if(regexec(&fields_to_parse[f].regexp, raw_line, 1, &matches, 0) == 0) {
+ fprintf(db_file, "%s %s\n",
+ fields_to_parse[f].name,
+ raw_line + matches.rm_eo);
+ }
+ }
+}
+
+void index_mbox(const char *input_filename,
+ int nb_fields_to_parse, struct parsable_field *fields_to_parse,
+ FILE *db_file) {
+ char raw_line[BUFFER_SIZE], full_line[BUFFER_SIZE];
+ char *end_of_full_line;
FILE *file;
int in_header, new_header;
- unsigned int position_in_file;
+ unsigned long int position_in_file;
file = fopen(input_filename, "r");
if(!file) {
- fprintf(stderr, "mymail: Can not open `%s'.\n", input_filename);
- exit(EXIT_FAILURE);
+ fprintf(stderr, "mymail: Can not open '%s'.\n", input_filename);
+ if(paranoid) { exit(EXIT_FAILURE); }
+ return;
}
in_header = 0;
new_header = 0;
position_in_file = 0;
+ end_of_full_line = 0;
+ full_line[0] = '\0';
while(fgets(raw_line, BUFFER_SIZE, file)) {
if(strncmp(raw_line, "From ", 5) == 0) {
if(in_header) {
fprintf(stderr,
- "Got a 'From ' in the header in %s:%u.\n",
+ "Got a ^\"From \" in the header in %s:%lu.\n",
input_filename, position_in_file);
fprintf(stderr, "%s", raw_line);
- exit(EXIT_FAILURE);
+ if(paranoid) { exit(EXIT_FAILURE); }
}
in_header = 1;
new_header = 1;
if(in_header) { in_header = 0; }
}
- /* if(in_header) { */
- /* printf("LINE.H %s", raw_line); */
- /* } else { */
- /* printf("LINE.B %s", raw_line); */
- /* } */
+ /* printf("PARSE %d %s", in_header, raw_line); */
if(in_header) {
- int f;
- regmatch_t matches;
if(new_header) {
- fprintf(db_file, "mail %u %s\n", position_in_file, input_filename);
+ fprintf(db_file, "mail %lu %s\n", position_in_file, input_filename);
new_header = 0;
}
- for(f = 0; f < nb_fields_to_parse; f++) {
- if(regexec(&fields_to_parse[f].regexp, raw_line, 1, &matches, 0) == 0) {
- fprintf(db_file, "%s %s",
- fields_to_parse[f].name,
- raw_line + matches.rm_eo);
+
+ if(raw_line[0] == ' ' || raw_line[0] == '\t') {
+ char *start = raw_line;
+ while(*start == ' ' || *start == '\t') start++;
+ *(end_of_full_line++) = ' ';
+ strcpy(end_of_full_line, start);
+ while(*end_of_full_line && *end_of_full_line != '\n') {
+ end_of_full_line++;
+ }
+ *end_of_full_line = '\0';
+ }
+
+ else {
+ /* if(!((raw_line[0] >= 'a' && raw_line[0] <= 'z') || */
+ /* (raw_line[0] >= 'A' && raw_line[0] <= 'Z'))) { */
+ /* fprintf(stderr, */
+ /* "Header line syntax error %s:%lu.\n", */
+ /* input_filename, position_in_file); */
+ /* fprintf(stderr, "%s", raw_line); */
+ /* } */
+
+ if(full_line[0]) {
+ index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
+ }
+
+ end_of_full_line = full_line;
+ strcpy(end_of_full_line, raw_line);
+ while(*end_of_full_line && *end_of_full_line != '\n') {
+ end_of_full_line++;
}
+ *end_of_full_line = '\0';
}
+
}
position_in_file += strlen(raw_line);
int ignore_entry(const char *name) {
return
- strcmp(name, ".") == 0 ||
- strcmp(name, "..") == 0 ||
+ /* strcmp(name, ".") == 0 || */
+ /* strcmp(name, "..") == 0 || */
(name[0] == '.' && name[1] != '/');
}
dir_name,
strerror(errno));
exit(EXIT_FAILURE);
- } else {
- }
-
- if(S_ISLNK(sb.st_mode)) {
- return;
}
dir = opendir(dir_name);
}
closedir(dir);
} else {
- if(S_ISREG(sb.st_mode)) {
- /* printf("Processing regular file '%s'.\n", dir_name); */
- read_file(dir_name, nb_fields_to_parse, fields_to_parse, db_file);
- }
+ index_mbox(dir_name, nb_fields_to_parse, fields_to_parse, db_file);
}
}
/* For long options that have no equivalent short option, use a
non-character as a pseudo short option, starting with CHAR_MAX + 1. */
-enum
-{
+enum {
OPT_BASH_MODE = CHAR_MAX + 1
};
static struct option long_options[] = {
{ "help", no_argument, 0, 'h' },
- { "db-prefix", 1, 0, 'p' },
+ { "version", no_argument, 0, 'v' },
+ { "db-file", 1, 0, 'd' },
{ "search-pattern", 1, 0, 's' },
{ "index", 0, 0, 'i' },
{ 0, 0, 0, 0 }
char c;
int f;
+ paranoid = 0;
action_index = 0;
search_pattern = 0;
setlocale(LC_ALL, "");
- while ((c = getopt_long(argc, argv, "hip:s:",
+ while ((c = getopt_long(argc, argv, "hvip:s:",
long_options, NULL)) != -1) {
switch(c) {
show_help = 1;
break;
+ case 'v':
+ print_version(stdout);
+ break;
+
case 'i':
action_index = 1;
break;
case 'p':
db_filename = strdup(optarg);
+ /* printf("db_filename=\"%s\"\n", db_filename); */
break;
case 's':
if(search_pattern) {
- fprintf(stderr,
- "mymail: Search pattern already defined.\n");
+ fprintf(stderr, "mymail: Search pattern already defined.\n");
exit(EXIT_FAILURE);
}
search_pattern = strdup(optarg);
}
if(!db_filename) {
- db_filename = strdup("/tmp/mymail");
+ char *default_db_filename = getenv("MYMAIL_DB_FILE");
+ if(!default_db_filename) { default_db_filename = "/tmp/mymail.db"; }
+ db_filename = strdup(default_db_filename);
}
if(error) {
- usage(stderr);
+ print_usage(stderr);
exit(EXIT_FAILURE);
}
if(show_help) {
- usage(stdout);
+ print_usage(stdout);
exit(EXIT_SUCCESS);
}
exit(EXIT_FAILURE);
}
- /* printf("Starting search in %s for field \"%s\" matching \"%s\".\n", */
- /* db_filename, */
- /* search_name, */
- /* search_regexp_string); */
-
db_file = fopen(db_filename, "r");
if(!db_file) {
}
}
+ free(db_filename);
+
exit(EXIT_SUCCESS);
}