X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=mymail.c;h=125366a99ae21fa488227970dcef87519ee73427;hb=9dad01815a424140893ae31a10530c0208c21111;hp=dc7c58c2d4429bb7b31aeded25a99ddaf8bffa15;hpb=8dea757944e71980e7b76ede4553eb1fb457c1ec;p=mymail.git diff --git a/mymail.c b/mymail.c index dc7c58c..125366a 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 @@ -40,10 +52,25 @@ struct parsable_field { char *name; char *regexp_string; regex_t regexp; - FILE *db_file; }; -char *db_filename_prefix; +char *db_filename; +char *search_pattern; + +int paranoid; +int action_index; + +char *segment_next_field(char *current) { + while(*current && *current != ' ') current++; + *current = '\0'; current++; + while(*current && *current == ' ') current++; + return current; +} + +void remove_eof(char *c) { + while(*c && *c != '\n' && *c != '\r') c++; + *c = '\0'; +} /********************************************************************/ @@ -69,21 +96,102 @@ void usage(FILE *out) { fprintf(out, "\n"); } -void read_file(const char *input_filename, - int nb_fields_to_parse, struct parsable_field *fields_to_parse) { +/*********************************************************************/ + +void search_in_db(const char *search_name, const char *search_regexp_string, + FILE *db_file) { char raw_line[BUFFER_SIZE]; + char current_mail_filename[BUFFER_SIZE]; + unsigned long int current_position_in_mail; + char *name, *value; + regex_t regexp; + int already_written; + + if(regcomp(®exp, + search_regexp_string, + REG_ICASE)) { + fprintf(stderr, + "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n", + search_regexp_string, + search_name); + exit(EXIT_FAILURE); + } + + current_position_in_mail = 0; + already_written = 0; + + while(fgets(raw_line, BUFFER_SIZE, db_file)) { + name = raw_line; + value = segment_next_field(raw_line); + + if(strcmp("mail", name) == 0) { + char *position_in_file_string = value; + char *mail_filename = segment_next_field(value); + current_position_in_mail = atol(position_in_file_string); + strcpy(current_mail_filename, mail_filename); + remove_eof(current_mail_filename); + already_written = 0; + } + + else if(!already_written) { + 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); + exit(EXIT_FAILURE); + } + fseek(mail_file, current_position_in_mail, SEEK_SET); + if(fgets(raw_line, BUFFER_SIZE, mail_file)) { + printf("%s", raw_line); + while(fgets(raw_line, BUFFER_SIZE, mail_file) && + strncmp(raw_line, "From ", 5)) { + printf("%s", raw_line); + } + } + fclose(mail_file); + already_written = 1; + } + } + } + + regfree(®exp); +} + +/*********************************************************************/ + +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; - unsigned int position_in_file; + int in_header, new_header; + 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); + if(paranoid) { exit(EXIT_FAILURE); } + return; } in_header = 0; + new_header = 0; position_in_file = 0; @@ -91,32 +199,57 @@ void read_file(const char *input_filename, 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; } else if(strncmp(raw_line, "\n", 1) == 0) { 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; - for(f = 0; f < nb_fields_to_parse; f++) { - if(regexec(&fields_to_parse[f].regexp, raw_line, 1, &matches, 0) == 0) { - fprintf(fields_to_parse[f].db_file, "%s:%d %s", - input_filename, position_in_file, - raw_line + matches.rm_eo); + if(new_header) { + fprintf(db_file, "mail %lu %s\n", position_in_file, input_filename); + new_header = 0; + } + + 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); @@ -127,13 +260,14 @@ void read_file(const char *input_filename, int ignore_entry(const char *name) { return - strcmp(name, ".") == 0 || - strcmp(name, "..") == 0 || + /* strcmp(name, ".") == 0 || */ + /* strcmp(name, "..") == 0 || */ (name[0] == '.' && name[1] != '/'); } void process_entry(const char *dir_name, - int nb_fields_to_parse, struct parsable_field *fields_to_parse) { + int nb_fields_to_parse, struct parsable_field *fields_to_parse, + FILE *db_file) { DIR *dir; struct dirent *dir_e; struct stat sb; @@ -145,11 +279,6 @@ void process_entry(const char *dir_name, dir_name, strerror(errno)); exit(EXIT_FAILURE); - } else { - } - - if(S_ISLNK(sb.st_mode)) { - return; } dir = opendir(dir_name); @@ -159,15 +288,12 @@ void process_entry(const char *dir_name, while((dir_e = readdir(dir))) { if(!ignore_entry(dir_e->d_name)) { snprintf(subname, PATH_MAX, "%s/%s", dir_name, dir_e->d_name); - process_entry(subname, nb_fields_to_parse, fields_to_parse); + process_entry(subname, nb_fields_to_parse, fields_to_parse, db_file); } } 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); - } + index_mbox(dir_name, nb_fields_to_parse, fields_to_parse, db_file); } } @@ -175,28 +301,29 @@ 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' }, + { "search-pattern", 1, 0, 's' }, + { "index", 0, 0, 'i' }, { 0, 0, 0, 0 } }; static struct parsable_field fields_to_parse[] = { { "from", - "^[Ff][Rr][Oo][Mm]: *", - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, 0 + "^\\([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 }, 0 + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, }; @@ -206,9 +333,13 @@ int main(int argc, char **argv) { char c; int f; + paranoid = 0; + action_index = 0; + search_pattern = 0; + setlocale(LC_ALL, ""); - while ((c = getopt_long(argc, argv, "hp:", + while ((c = getopt_long(argc, argv, "hip:s:", long_options, NULL)) != -1) { switch(c) { @@ -217,8 +348,20 @@ int main(int argc, char **argv) { show_help = 1; break; + case 'i': + action_index = 1; + break; + case 'p': - db_filename_prefix = strdup(optarg); + db_filename = strdup(optarg); + break; + + case 's': + if(search_pattern) { + fprintf(stderr, "mymail: Search pattern already defined.\n"); + exit(EXIT_FAILURE); + } + search_pattern = strdup(optarg); break; default: @@ -227,8 +370,8 @@ int main(int argc, char **argv) { } } - if(!db_filename_prefix) { - db_filename_prefix = strdup("/tmp/mymail_"); + if(!db_filename) { + db_filename = strdup("/tmp/mymail"); } if(error) { @@ -241,11 +384,9 @@ int main(int argc, char **argv) { exit(EXIT_SUCCESS); } - for(f = 0; f < nb_fields_to_parse; f++) { - char db_filename[BUFFER_SIZE]; - sprintf(db_filename, "%s%s", db_filename_prefix, fields_to_parse[f].name); - fields_to_parse[f].db_file = fopen(db_filename, "w"); - if(!fields_to_parse[f].db_file) { + if(action_index) { + FILE *db_file = fopen(db_filename, "w"); + if(!db_file) { fprintf(stderr, "mymail: Can not open \"%s\" for writing: %s\n", db_filename, @@ -253,29 +394,67 @@ int main(int argc, char **argv) { exit(EXIT_FAILURE); } - printf("Initialized %s.\n", db_filename); + for(f = 0; f < nb_fields_to_parse; f++) { + if(regcomp(&fields_to_parse[f].regexp, + fields_to_parse[f].regexp_string, + REG_ICASE)) { + fprintf(stderr, + "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n", + fields_to_parse[f].regexp_string, + fields_to_parse[f].name); + exit(EXIT_FAILURE); + } + } + + while(optind < argc) { + process_entry(argv[optind], + nb_fields_to_parse, fields_to_parse, db_file); + optind++; + } - if(regcomp(&fields_to_parse[f].regexp, - fields_to_parse[f].regexp_string, - REG_ICASE)) { - fprintf(stderr, - "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n", - fields_to_parse[f].regexp_string, - fields_to_parse[f].name); - exit(EXIT_FAILURE); + fclose(db_file); + + for(f = 0; f < nb_fields_to_parse; f++) { + regfree(&fields_to_parse[f].regexp); } } - while(optind < argc) { - process_entry(argv[optind], - nb_fields_to_parse, fields_to_parse); - optind++; - } + else { + if(search_pattern) { + FILE *db_file; + char *search_name; + char *search_regexp_string; + search_name = search_pattern; + search_regexp_string = segment_next_field(search_pattern); + if(!*search_regexp_string) { + fprintf(stderr, + "Syntax error in the search pattern.\n"); + exit(EXIT_FAILURE); + } - for(f = 0; f < nb_fields_to_parse; f++) { - fclose(fields_to_parse[f].db_file); - regfree(&fields_to_parse[f].regexp); + /* 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) { + fprintf(stderr, + "mymail: Can not open \"%s\" for reading: %s\n", + db_filename, + strerror(errno)); + exit(EXIT_FAILURE); + } + + search_in_db(search_name, search_regexp_string, db_file); + + fclose(db_file); + free(search_pattern); + } } + free(db_filename); + exit(EXIT_SUCCESS); }