X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=mymail.c;h=001d14da9e4df232b82fa467e32faad0d77ff1e6;hb=b0bc71a636e5c43262dc7a1cb99ce4cdb92d2286;hp=99a76d895b7fc38d03e21a99f732590980564107;hpb=d763d790abe7d93b4572c38b261a4f480a8aa961;p=mymail.git diff --git a/mymail.c b/mymail.c index 99a76d8..001d14d 100644 --- a/mymail.c +++ b/mymail.c @@ -19,6 +19,18 @@ * */ +/* + + 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 @@ -34,7 +46,7 @@ #define VERSION "0.1" -#define BUFFER_SIZE 16384 +#define BUFFER_SIZE 65536 struct parsable_field { char *name; @@ -45,6 +57,7 @@ struct parsable_field { char *db_filename; char *search_pattern; +int paranoid; int action_index; char *segment_next_field(char *current) { @@ -121,11 +134,11 @@ void search_in_db(const char *search_name, const char *search_regexp_string, } else if(!already_written) { - if(strcmp(search_name, name) == 0 && regexec(®exp, value, 0, 0, 0) == 0) { + if(strcmp(search_name, name) == 0 && regexec(®exp, value, 0, 0, 0) == 0) { FILE *mail_file; 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); @@ -145,13 +158,26 @@ void search_in_db(const char *search_name, const char *search_regexp_string, 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 long int position_in_file; @@ -159,23 +185,25 @@ void read_file(const char *input_filename, 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; 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:%lu.\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; @@ -183,26 +211,46 @@ void read_file(const char *input_filename, 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 %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); @@ -234,10 +282,6 @@ void process_entry(const char *dir_name, exit(EXIT_FAILURE); } - /* if(S_ISLNK(sb.st_mode)) { */ - /* return; */ - /* } */ - dir = opendir(dir_name); if(dir) { @@ -250,9 +294,7 @@ void process_entry(const char *dir_name, } closedir(dir); } else { - /* if(S_ISREG(sb.st_mode)) { */ - 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); } } @@ -260,14 +302,13 @@ void process_entry(const char *dir_name, /* 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' }, + { "db-file", 1, 0, 'p' }, { "search-pattern", 1, 0, 's' }, { "index", 0, 0, 'i' }, { 0, 0, 0, 0 } @@ -293,6 +334,7 @@ int main(int argc, char **argv) { char c; int f; + paranoid = 0; action_index = 0; search_pattern = 0; @@ -313,12 +355,12 @@ int main(int argc, char **argv) { 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); @@ -331,7 +373,9 @@ int main(int argc, char **argv) { } 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) { @@ -393,9 +437,9 @@ int main(int argc, char **argv) { } /* printf("Starting search in %s for field \"%s\" matching \"%s\".\n", */ - /* db_filename, */ - /* search_name, */ - /* search_regexp_string); */ + /* db_filename, */ + /* search_name, */ + /* search_regexp_string); */ db_file = fopen(db_filename, "r");