X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=mymail.c;h=125366a99ae21fa488227970dcef87519ee73427;hb=9dad01815a424140893ae31a10530c0208c21111;hp=877215cb1d0509e2822ecf1a7864b989b840adb5;hpb=72cd948c9f36ae5e79bf3317aa4e2cda0232c9ee;p=mymail.git diff --git a/mymail.c b/mymail.c index 877215c..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 @@ -45,8 +57,21 @@ struct parsable_field { 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'; +} + /********************************************************************/ /* malloc with error checking. */ @@ -76,9 +101,11 @@ void usage(FILE *out) { void search_in_db(const char *search_name, const char *search_regexp_string, FILE *db_file) { char raw_line[BUFFER_SIZE]; - char current_mail[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, @@ -90,42 +117,77 @@ void search_in_db(const char *search_name, const char *search_regexp_string, exit(EXIT_FAILURE); } + current_position_in_mail = 0; + already_written = 0; + while(fgets(raw_line, BUFFER_SIZE, db_file)) { name = raw_line; - value = raw_line; - while(*value && *value != ' ') value++; - *value = '\0'; value++; - while(*value && *value == ' ') value++; - - /* printf("LINE [%s] %s", name, value); */ + value = segment_next_field(raw_line); if(strcmp("mail", name) == 0) { - strcpy(current_mail, value); - /* printf("READING [%s]\n", current_mail); */ - } else { + 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) { - printf("%s", current_mail); + 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 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); + if(paranoid) { exit(EXIT_FAILURE); } + return; } in_header = 0; @@ -137,10 +199,10 @@ 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; @@ -148,26 +210,46 @@ 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 %s:%d\n", input_filename, position_in_file); + 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); @@ -178,8 +260,8 @@ 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] != '/'); } @@ -197,11 +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); @@ -216,10 +293,7 @@ void process_entry(const char *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); } } @@ -227,8 +301,7 @@ 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 }; @@ -260,6 +333,7 @@ int main(int argc, char **argv) { char c; int f; + paranoid = 0; action_index = 0; search_pattern = 0; @@ -284,8 +358,7 @@ int main(int argc, char **argv) { 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); @@ -352,20 +425,17 @@ int main(int argc, char **argv) { char *search_name; char *search_regexp_string; search_name = search_pattern; - search_regexp_string = search_pattern; - while(*search_regexp_string && *search_regexp_string != ' ') search_regexp_string++; - *search_regexp_string = '\0'; search_regexp_string++; - while(*search_regexp_string && *search_regexp_string == ' ') search_regexp_string++; + search_regexp_string = segment_next_field(search_pattern); if(!*search_regexp_string) { fprintf(stderr, "Syntax error in the search pattern.\n"); exit(EXIT_FAILURE); } - printf("Starting search in %s for field \"%s\" matching \"%s\".\n", - db_filename, - search_name, - search_regexp_string); + /* printf("Starting search in %s for field \"%s\" matching \"%s\".\n", */ + /* db_filename, */ + /* search_name, */ + /* search_regexp_string); */ db_file = fopen(db_filename, "r"); @@ -376,6 +446,7 @@ int main(int argc, char **argv) { strerror(errno)); exit(EXIT_FAILURE); } + search_in_db(search_name, search_regexp_string, db_file); fclose(db_file); @@ -383,5 +454,7 @@ int main(int argc, char **argv) { } } + free(db_filename); + exit(EXIT_SUCCESS); }