From ff95e19bacf5e4cd08c5cfba35b23a227ac34274 Mon Sep 17 00:00:00 2001 From: Francois Fleuret Date: Thu, 31 Jan 2013 09:01:17 +0100 Subject: [PATCH] Now search recursively in multiple db files + lots of little updates. --- mymail.c | 194 ++++++++++++++++++++++++++++++++++++++++++------------- 1 file changed, 148 insertions(+), 46 deletions(-) diff --git a/mymail.c b/mymail.c index 2617d02..c7d41d7 100644 --- a/mymail.c +++ b/mymail.c @@ -44,6 +44,7 @@ #include #include +#define MYMAIL_DB_MAGIC_TOKEN "mymail_index_file" #define VERSION "0.1" #define BUFFER_SIZE 65536 @@ -55,6 +56,9 @@ struct parsable_field { }; char *db_filename; +char *db_root_path; + +int multi_db_files; char *search_pattern; int paranoid; @@ -112,10 +116,18 @@ void print_usage(FILE *out) { /*********************************************************************/ +int ignore_entry(const char *name) { + return + /* strcmp(name, ".") == 0 || */ + /* strcmp(name, "..") == 0 || */ + (name[0] == '.' && name[1] != '/'); +} + 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]; + char current_mail_filename[PATH_MAX + 1]; unsigned long int current_position_in_mail; char *name, *value; regex_t regexp; @@ -172,6 +184,68 @@ void search_in_db(const char *search_name, const char *search_regexp_string, regfree(®exp); } +void recursive_search_in_db(const char *entry_name, + const char *search_name, const char *search_regexp_string) { + DIR *dir; + struct dirent *dir_e; + struct stat sb; + char raw_line[BUFFER_SIZE]; + char subname[PATH_MAX + 1]; + + if(lstat(entry_name, &sb) != 0) { + fprintf(stderr, + "mymail: Can not stat \"%s\": %s\n", + entry_name, + strerror(errno)); + exit(EXIT_FAILURE); + } + + dir = opendir(entry_name); + + if(dir) { + while((dir_e = readdir(dir))) { + if(!ignore_entry(dir_e->d_name)) { + snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name); + recursive_search_in_db(subname, search_name, search_regexp_string); + } + } + closedir(dir); + } else { + const char *s = entry_name, *filename = entry_name; + while(*s) { if(*s == '/') { filename = s+1; } s++; } + + if(strcmp(filename, db_filename) == 0) { + FILE *db_file = fopen(entry_name, "r"); + + if(!db_file) { + fprintf(stderr, + "mymail: Can not open \"%s\" for reading: %s\n", + db_filename, + strerror(errno)); + exit(EXIT_FAILURE); + } + + if(fgets(raw_line, BUFFER_SIZE, db_file)) { + if(strncmp(raw_line, MYMAIL_DB_MAGIC_TOKEN, strlen(MYMAIL_DB_MAGIC_TOKEN))) { + fprintf(stderr, + "mymail: Header line in '%s' does not match the mymail db format.\n", + entry_name); + exit(EXIT_FAILURE); + } + } else { + fprintf(stderr, + "mymail: Can not read the header line in '%s'.\n", + entry_name); + exit(EXIT_FAILURE); + } + + search_in_db(search_name, search_regexp_string, db_file); + + fclose(db_file); + } + } +} + /*********************************************************************/ void index_one_mbox_line(int nb_fields_to_parse, struct parsable_field *fields_to_parse, @@ -187,7 +261,7 @@ void index_one_mbox_line(int nb_fields_to_parse, struct parsable_field *fields_t } } -void index_mbox(const char *input_filename, +void index_mbox(const char *mbox_filename, int nb_fields_to_parse, struct parsable_field *fields_to_parse, FILE *db_file) { char raw_line[BUFFER_SIZE], full_line[BUFFER_SIZE]; @@ -196,10 +270,10 @@ void index_mbox(const char *input_filename, int in_header, new_header; unsigned long int position_in_file; - file = fopen(input_filename, "r"); + file = fopen(mbox_filename, "r"); if(!file) { - fprintf(stderr, "mymail: Can not open '%s'.\n", input_filename); + fprintf(stderr, "mymail: Can not open '%s'.\n", mbox_filename); if(paranoid) { exit(EXIT_FAILURE); } return; } @@ -216,7 +290,7 @@ void index_mbox(const char *input_filename, if(in_header) { fprintf(stderr, "Got a ^\"From \" in the header in %s:%lu.\n", - input_filename, position_in_file); + mbox_filename, position_in_file); fprintf(stderr, "%s", raw_line); if(paranoid) { exit(EXIT_FAILURE); } } @@ -226,11 +300,9 @@ void index_mbox(const char *input_filename, if(in_header) { in_header = 0; } } - /* printf("PARSE %d %s", in_header, raw_line); */ - if(in_header) { if(new_header) { - fprintf(db_file, "mail %lu %s\n", position_in_file, input_filename); + fprintf(db_file, "mail %lu %s\n", position_in_file, mbox_filename); new_header = 0; } @@ -250,7 +322,7 @@ void index_mbox(const char *input_filename, /* (raw_line[0] >= 'A' && raw_line[0] <= 'Z'))) { */ /* fprintf(stderr, */ /* "Header line syntax error %s:%lu.\n", */ - /* input_filename, position_in_file); */ + /* mbox_filename, position_in_file); */ /* fprintf(stderr, "%s", raw_line); */ /* } */ @@ -274,42 +346,34 @@ void index_mbox(const char *input_filename, fclose(file); } -int ignore_entry(const char *name) { - return - /* 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, - FILE *db_file) { +void recursive_index_mbox(FILE *db_file, + const char *entry_name, + int nb_fields_to_parse, struct parsable_field *fields_to_parse) { DIR *dir; struct dirent *dir_e; struct stat sb; char subname[PATH_MAX + 1]; - if(lstat(dir_name, &sb) != 0) { + if(lstat(entry_name, &sb) != 0) { fprintf(stderr, "mymail: Can not stat \"%s\": %s\n", - dir_name, + entry_name, strerror(errno)); exit(EXIT_FAILURE); } - dir = opendir(dir_name); + dir = opendir(entry_name); if(dir) { - printf("Processing directory '%s'.\n", 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, db_file); + snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name); + recursive_index_mbox(db_file, subname, nb_fields_to_parse, fields_to_parse); } } closedir(dir); } else { - index_mbox(dir_name, nb_fields_to_parse, fields_to_parse, db_file); + index_mbox(entry_name, nb_fields_to_parse, fields_to_parse, db_file); } } @@ -325,8 +389,10 @@ static struct option long_options[] = { { "help", no_argument, 0, 'h' }, { "version", no_argument, 0, 'v' }, { "db-file", 1, 0, 'd' }, + { "db-root", 1, 0, 'p' }, { "search-pattern", 1, 0, 's' }, { "index", 0, 0, 'i' }, + { "multi-db-files", 0, 0, 'm' }, { 0, 0, 0, 0 } }; @@ -342,6 +408,13 @@ static struct parsable_field fields_to_parse[] = { "^\\([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) { @@ -353,10 +426,13 @@ int main(int argc, char **argv) { paranoid = 0; action_index = 0; search_pattern = 0; + db_filename = 0; + db_root_path = 0; + multi_db_files = 0; setlocale(LC_ALL, ""); - while ((c = getopt_long(argc, argv, "hvip:s:", + while ((c = getopt_long(argc, argv, "hvimp:s:d:p:", long_options, NULL)) != -1) { switch(c) { @@ -373,9 +449,16 @@ int main(int argc, char **argv) { action_index = 1; break; - case 'p': + case 'm': + multi_db_files = 1; + break; + + case 'd': db_filename = strdup(optarg); - /* printf("db_filename=\"%s\"\n", db_filename); */ + break; + + case 'p': + db_root_path = strdup(optarg); break; case 's': @@ -394,10 +477,32 @@ int main(int argc, char **argv) { if(!db_filename) { char *default_db_filename = getenv("MYMAIL_DB_FILE"); - if(!default_db_filename) { default_db_filename = "/tmp/mymail.db"; } + + if(!default_db_filename) { + if(multi_db_files) { + default_db_filename = "mymail.db"; + } else { + default_db_filename = "/tmp/mymail.db"; + } + } + db_filename = strdup(default_db_filename); } + if(!db_root_path) { + char *default_db_root_path = getenv("MYMAIL_DB_ROOT"); + + if(!default_db_root_path) { + if(multi_db_files) { + default_db_root_path = "mymail.db"; + } else { + default_db_root_path = "/tmp/mymail.db"; + } + } + + db_root_path = strdup(default_db_root_path); + } + if(error) { print_usage(stderr); exit(EXIT_FAILURE); @@ -409,7 +514,10 @@ int main(int argc, char **argv) { } if(action_index) { - FILE *db_file = fopen(db_filename, "w"); + FILE *db_file; + + db_file = fopen(db_filename, "w"); + if(!db_file) { fprintf(stderr, "mymail: Can not open \"%s\" for writing: %s\n", @@ -430,9 +538,12 @@ int main(int argc, char **argv) { } } + fprintf(db_file, "%s version_%s raw version\n", MYMAIL_DB_MAGIC_TOKEN, VERSION); + while(optind < argc) { - process_entry(argv[optind], - nb_fields_to_parse, fields_to_parse, db_file); + recursive_index_mbox(db_file, + argv[optind], + nb_fields_to_parse, fields_to_parse); optind++; } @@ -444,36 +555,27 @@ int main(int argc, char **argv) { } 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); } - 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); + recursive_search_in_db(db_root_path, search_name, search_regexp_string); - fclose(db_file); free(search_pattern); } } free(db_filename); + free(db_root_path); exit(EXIT_SUCCESS); } -- 2.20.1