bogoconfig.c

一个C语言写的快速贝叶斯垃圾邮件过滤工具

 ** The functions will exit if there's an error, for example if
 ** there are leftover command line arguments.
 */

static void process_arglist(int argc, char **argv, priority_t precedence, int pass)
{
    ex_t exitcode;

    if (pass != PASS_1_CLI) {
	optind = opterr = 1;
#ifdef __EMX__
	optind = 0;
#endif
	/* don't use #ifdef here: */
#if HAVE_DECL_OPTRESET
	optreset = 1;
#endif
    }

    while (1) 
    {
	int option;
	int option_index = 0;
	int this_option_optind = optind ? optind : 1;
	const char *name;

	option = getopt_long(argc, argv, OPTIONS,
			     longopts_bogofilter, &option_index);

	if (option == -1)
	    break;

	name = (option_index == 0) ? argv[this_option_optind] : longopts_bogofilter[option_index].name;

#ifdef EXCESSIVE_DEBUG
	if (getenv("BOGOFILTER_DEBUG_OPTIONS")) {
	    fprintf(stderr, "config: option=%c (%d), optind=%d, opterr=%d, optarg=%s\n", 
		    isprint((unsigned char)option) ? option : '_', option, 
		    optind, opterr, optarg ? optarg : "(null)");
	}
#endif

	process_arg(option, name, optarg, precedence, pass);
    }

    if (pass == PASS_1_CLI) {
	if (run_type == RUN_UNKNOWN)
	    run_type = RUN_NORMAL;
    }

    if (pass == PASS_3_CLI) {
	exitcode = validate_args();
	if (exitcode) 
	    exit(exitcode);

	if (bulk_mode == B_NORMAL && optind < argc) {
	    fprintf(stderr, "Extra arguments given, first: %s. Aborting.\n", argv[optind]);
	    exit(EX_ERROR);
	}

	if (inv_terse_mode) {
	    verbose = max(1, verbose); 	/* force printing */
	    set_terse_mode_format(inv_terse_mode);
	}
    }

    return;
}

void process_arg(int option, const char *name, const char *val, priority_t precedence, arg_pass_t pass)
{
    switch (option)
    {
    case '?':
	if (pass == PASS_1_CLI)
	    fprintf(stderr, "Unknown option '%s'.\n", name);
	break;

    case 'b':
	bulk_mode = B_STDIN;
	fpin = NULL;	/* Ensure that input file isn't stdin */
	break;

    case 'B':
	bulk_mode = B_CMDLINE;
	break;

    case 'c':
    case O_CONFIG_FILE:
	if (pass == PASS_1_CLI) {
	    if (!read_config_file(val, false, !quiet, PR_CFG_USER, longopts_bogofilter)) {
		fprintf(stderr, "Cannot open %s: %s\n", val, strerror(errno));
		exit(EX_ERROR);
	    }
	}

	/*@fallthrough@*/
	/* fall through to suppress reading config files */

    case 'C':
	if (pass == PASS_1_CLI)
	    suppress_config_file = true;
	break;

    case O_USER_CONFIG_FILE:
	user_config_file = get_string(name, val);
	break;

    case 'D':
	dbgout = stdout;
	break;

    case 'e':
    case O_HAM_TRUE:
	nonspam_exits_zero = true;
	break;

    case 'h':
	help(stdout);
	exit(EX_OK);

    case 'I':
	if (pass == PASS_1_CLI)
	    bogoreader_name(val);
	break;

    case 'L':
	logtag = xstrdup(val);
	/*@fallthrough@*/

    case 'l':
	logflag = true;
	break;

    case 'M':
	mbox_mode = true;
	break;

    case 'n':
	run_type = check_run_type(REG_GOOD, REG_SPAM | UNREG_GOOD);
	break;

    case 'N':
	run_type = check_run_type(UNREG_GOOD, REG_GOOD | UNREG_SPAM);
	break;

    case 'O':
	if (pass == PASS_1_CLI)
	    xstrlcpy(outfname, val, sizeof(outfname));
	break;

    case 'p':
	passthrough = true;
	break;

    case 'q':
	quiet = true;
	break;

    case 'Q':
	if (pass == PASS_1_CLI)
	    query += 1;
	break;

    case 'R':
	Rtable = true;
	break;

    case 's':
	run_type = check_run_type(REG_SPAM, REG_GOOD | UNREG_SPAM);
	break;

    case 'S':
	run_type = check_run_type(UNREG_SPAM, REG_SPAM | UNREG_GOOD);
	break;

    case 'u':
	run_type |= RUN_UPDATE;
	break;
	
    case 'U':
	unsure_stats = (val == NULL) ? true : get_bool(name, val);
	break;

    case 'v':
	if (pass == PASS_1_CLI)
	    verbose++;
	break;

    case 'x':
	set_debug_mask( val );
	break;

    case 'X':
	set_bogotest( val );
	break;

    case 'y':		/* date as YYYYMMDD */
    {
	const char *str = val ? val : optarg;
	YYYYMMDD date = string_to_date(str);
	if (date != 0 && date < 19990000) {
	    fprintf(stderr, "Date format for '-y' option is YYYYMMDD\n");
	    exit(EX_ERROR);
	}
	set_date( date );
	break;
    }
	
    case ':':
	fprintf(stderr, "Option %s requires an argument.\n", name);
	exit(EX_ERROR);

    case '-':
	if (pass == PASS_3_CLI)
	    process_config_option(val, true, precedence, longopts_bogofilter);
	break;

    case 'd':
	if (pass != PASS_1_CLI)
	    set_wordlist_dir(val, precedence);
	break;

    case O_NS_ESF:
	if (pass != PASS_1_CLI)
	    get_double(name, val, &ns_esf);
	break;

    case O_SP_ESF:
	if (pass != PASS_1_CLI)
	    get_double(name, val, &sp_esf);
	break;

    case 'H':
	header_line_markup = (val == NULL) ? false : get_bool(name, val);
	break;

    case 'k':
	db_cachesize=atoi(val);
	break;

    case 'm':
	if (pass != PASS_1_CLI) {
	    comma_parse(option, val, &min_dev, &robs, &robx);
	    if (DEBUG_CONFIG(1))
		fprintf(dbgout, "md %6.4f, rs %6.4f, rx %6.4f\n", min_dev, robs, robx);
	}
	break;

    case O_MIN_DEV:
	if (pass != PASS_1_CLI)
	    get_double(name, val, &min_dev);
	break;

    case O_ROBS:
	if (pass != PASS_1_CLI)
	    get_double(name, val, &robs);
	break;

    case O_ROBX:
	if (pass != PASS_1_CLI)
	    get_double(name, val, &robx);
	break;

    case 'o':
	if (pass != PASS_1_CLI) {
	    comma_parse(option, val, &spam_cutoff, &ham_cutoff, NULL);
	    if (DEBUG_CONFIG(1))
		fprintf(dbgout, "sc %6.4f, hc %6.4f\n", spam_cutoff, ham_cutoff);
	}
	break;

    case O_SPAM_CUTOFF:
	if (pass != PASS_1_CLI)
	    get_double(name, val, &spam_cutoff);
	break;

    case O_HAM_CUTOFF:
	if (pass != PASS_1_CLI)
	    get_double(name, val, &ham_cutoff);
	break;

    case 't':
	terse = true;
	break;

    case 'T':			/* invariant terse mode */
	terse = true;
	if (pass == PASS_1_CLI)
	    inv_terse_mode += 1;
	break;

    case 'V':
	print_version();
	exit(EX_OK);

    case O_BLOCK_ON_SUBNETS:		block_on_subnets = get_bool(name, val);			break;
    case O_CHARSET_DEFAULT:		charset_default = get_string(name, val);		break;
    case O_HEADER_FORMAT:		header_format = get_string(name, val);			break;
    case O_LOG_HEADER_FORMAT:		log_header_format = get_string(name, val);		break;
    case O_LOG_UPDATE_FORMAT:		log_update_format = get_string(name, val);		break;
    case O_MAX_TOKEN_LEN:		max_token_len=atoi(val);				break;
    case O_MIN_TOKEN_LEN:		min_token_len=atoi(val);				break;
    case O_MAX_MULTI_TOKEN_LEN:		max_multi_token_len=atoi(val);				break;
    case O_MULTI_TOKEN_COUNT:		multi_token_count=atoi(val);				break;
    case O_REPLACE_NONASCII_CHARACTERS:	replace_nonascii_characters = get_bool(name, val);	break;
    case O_SPAMICITY_FORMATS:		set_spamicity_formats(val);				break;
    case O_SPAMICITY_TAGS:		set_spamicity_tags(val);				break;
    case O_SPAM_HEADER_NAME:		spam_header_name = get_string(name, val);		break;
    case O_SPAM_SUBJECT_TAG:		spam_subject_tag = get_string(name, val);		break;
    case O_STATS_IN_HEADER:		stats_in_header = get_bool(name, val);			break;
    case O_TERSE:			terse = get_bool(name, val);				break;
    case O_TERSE_FORMAT:		terse_format = get_string(name, val);			break;
    case O_THRESH_UPDATE:		get_double(name, val, &thresh_update);			break;
    case O_TIMESTAMP:			timestamp_tokens = get_bool(name, val);			break;
    case O_UNSURE_SUBJECT_TAG:		unsure_subject_tag = get_string(name, val);		break;
    case O_UNICODE:			encoding = get_bool(name, val) ? E_UNICODE : E_RAW;	break;
    case O_WORDLIST:			configure_wordlist(val);				break;

    case O_DB_TRANSACTION:		eTransaction = get_txn(name, val);			break;

    default:
#ifndef	DISABLE_TRANSACTIONS
	if (!dsm_options_bogofilter(option, name, val))
#endif
	{
	    fprintf(stderr, "Invalid option '%s'.\n", name);
	    exit(EX_ERROR);
	}
    }
}

#define	Q1	if (query >= 1)
#define	Q2	if (query >= 2)

#define YN(b) (b ? "Yes" : "No")
#define NB(b) ((b != NULL && *b != '\0') ? b : "''")

rc_t query_config(void)
{
    Q1 fprintf(stdout, "# %s version %s\n", progname, version);
    Q1 fprintf(stdout, "\n");
    Q1 fprintf(stdout, "%-11s = %0.6f  # (%8.2e)\n", "robx", robx, robx);
    Q1 fprintf(stdout, "%-11s = %0.6f  # (%8.2e)\n", "robs", robs, robs);
    Q1 fprintf(stdout, "%-11s = %0.6f  # (%8.2e)\n", "min_dev", min_dev, min_dev);
    Q1 fprintf(stdout, "%-11s = %0.6f  # (%8.2e)\n", "ham_cutoff", ham_cutoff, ham_cutoff);
    Q1 fprintf(stdout, "%-11s = %0.6f  # (%8.2e)\n", "spam_cutoff", spam_cutoff, spam_cutoff);
    Q1 fprintf(stdout, "%-11s = %0.6f  # (%8.2e)\n", "ns_esf", ns_esf, ns_esf);
    Q1 fprintf(stdout, "%-11s = %0.6f  # (%8.2e)\n", "sp_esf", sp_esf, sp_esf);
    Q1 fprintf(stdout, "\n");
    Q1 fprintf(stdout, "%-17s = %s\n",    "block-on-subnets",    YN(block_on_subnets));
    Q1 fprintf(stdout, "%-17s = %s\n",    "encoding",		 (encoding != E_UNICODE) ? "raw" : "utf-8");
    Q1 fprintf(stdout, "%-17s = %s\n",    "charset-default",     charset_default);
    Q1 fprintf(stdout, "%-17s = %s\n",    "replace-nonascii-characters", YN(replace_nonascii_characters));
    Q2 fprintf(stdout, "%-17s = %s\n",    "no-header-tags",      YN(header_line_markup));
    Q1 fprintf(stdout, "%-17s = %s\n",    "stats-in-header",     YN(stats_in_header));
    Q2 fprintf(stdout, "%-17s = %s\n",    "report-unsure",       YN(unsure_stats));
    Q1 fprintf(stdout, "%-17s = %0.6f\n", "thresh-update",       thresh_update);
    Q1 fprintf(stdout, "%-17s = %s\n",    "timestamp",           YN(timestamp_tokens));
    Q2 fprintf(stdout, "%-17s = %ld\n",   "timestamp-date",      (long int)today);
    Q1 fprintf(stdout, "\n");
    Q1 fprintf(stdout, "%-17s = %s\n", "spam-header-name",    spam_header_name);
    Q1 fprintf(stdout, "%-17s = %s\n", "spam-subject-tag",    NB(spam_subject_tag));
    Q1 fprintf(stdout, "%-17s = %s\n", "unsure-subject-tag",  NB(unsure_subject_tag));
    Q2 fprintf(stdout, "%-18s = %s\n", "syslog-tag",          NB(logtag));
    Q1 fprintf(stdout, "%-17s = %s\n", "header-format",       header_format);
    Q1 fprintf(stdout, "%-17s = %s\n", "terse-format",        terse_format);
    Q1 fprintf(stdout, "%-17s = %s\n", "log-header-format",   log_header_format);
    Q1 fprintf(stdout, "%-17s = %s\n", "log-update-format",   log_update_format);
    Q1 display_tag_array("spamicity-tags   ", spamicity_tags);
    Q1 display_tag_array("spamicity-formats", spamicity_formats);

    Q2 fprintf(stdout, "\n");

    Q2 fprintf(stdout, "%-18s = %s\n", "no-config-file",   YN(suppress_config_file));
    Q2 fprintf(stdout, "%-18s = %s\n", "config-file",      NB(config_file_name));
    Q2 fprintf(stdout, "%-18s = %s\n", "user-config-file", NB(user_config_file));
    Q2 fprintf(stdout, "\n");

    Q2 fprintf(stdout, "%-18s = %s\n", "bogofilter-dir",        bogohome);
    Q2 display_wordlists(word_lists, "%-18s   ");
    Q2 fprintf(stdout, "\n");

#ifndef	DISABLE_TRANSACTIONS
    Q2 fprintf(stdout, "%-18s = %lu\n", "db-cachesize",         (unsigned long)db_cachesize);

#ifdef	ENABLE_TRANSACTIONS
#ifdef	HAVE_DECL_DB_CREATE
    Q2 fprintf(stdout, "%-18s = %s\n", "db-log-autoremove",     YN(db_log_autoremove));
#ifdef	FUTURE_DB_OPTIONS
    Q2 fprintf(stdout, "%-18s = %s\n", "db-log-txn-durable",    YN(db_txn_durable));
#endif
#endif
#endif
#endif

    return RC_OK;
}

static void display_tag_array(const char *label, FIELD *array)
{
    int i;
    int count = (ham_cutoff < EPS) ? 2 : 3;

    fprintf(stdout, "%s =", label);
    for (i = 0; i < count; i += 1)
	fprintf(stdout, "%s %s", (i == 0) ? "" : ",", array[i]);
    fprintf(stdout, "\n");
}