const int usage_train()
{
print("Usage: salad train [options]\n"
"\n"
"I/O options:\n"
" -i, --input <file> The input filename.\n"
" -f, --input-format <fmt> Sets the format of input. This option might be \n"
" one of " IOMODES ".\n"
#ifdef USE_REGEX_FILTER
" --input-filter <regex> The regular expression for filtering input lines\n"
" or filenames respectively.\n"
#endif
" --batch-size <num> Set the size of batches that are read and \n"
" processed in one go (Default: %"ZU").\n"
#ifdef USE_NETWORK
" -p, --pcap-filter <str> Filter expression for the PCAP library in case\n"
" network data is processed (Default: %s).\n"
" --client-only Only consider the client-side of the network\n"
" communication.\n"
" --server-only Only consider the server-side of the network\n"
" communication.\n"
#endif
" -u, --update-model In case the specified output file exists and\n"
" contains a valid model this flag indicates\n"
" that that model should be update rather than\n"
" recreated from scratch.\n"
" -o, --output <file> The output filename.\n"
#ifdef USE_ARCHIVES
// If there is no libarchive support we can only make use of text-based configurations.
" -F, --output-format <fmt> Sets the format of output. This option might be \n"
" one of " SALAD_OUTPUTFMTS ".\n"
#endif
"\n"
"Feature options:\n"
" -n, --ngram-len <num> Set length of n-grams (Default: %"ZU").\n"
" -d, --ngram-delim <delim> Set delimiters for the use of word/ token n-grams.\n"
" If omitted or empty byte n-grams are used.\n"
" --binary Indicates to use bit n-grams rather than byte\n"
" or token n-grams and consequently, disables the\n"
" --ngram-delim option.\n"
" -s, --filter-size <num> Set the size of the bloom filter as bits of\n"
" the index (Default: %u).\n"
" --hash-set <hashes> Set the hash set to be used: 'simple', 'simple2'\n"
" or 'murmur' (Default: '%s').\n"
"\n"
"Generic options:\n"
" -e, --echo-params Echo used parameters and settings.\n"
" -q, --quiet Suppress all output but warning and errors.\n"
" -h, --help Print this help screen.\n",
/* --batch-size */ (SIZE_T) DEFAULT_CONFIG.batch_size,
#ifdef USE_NETWORK
/* --pcap-filter */ DEFAULT_CONFIG.pcap_filter,
#endif
/* --ngram-len */ (SIZE_T) DEFAULT_CONFIG.ngram_length,
/* --filter-size */ DEFAULT_CONFIG.filter_size,
/* --hash-set */ hashset_to_string(DEFAULT_CONFIG.hash_set));
return EXIT_SUCCESS;
}
const int usage_inspect()
{
print("Usage: salad inspect [options]\n"
"\n"
"I/O options:\n"
" -i, --input <file> The input filename.\n"
" -f, --input-format <fmt> Sets the format of input. This option might be \n"
" one of " IOMODES ".\n"
#ifdef USE_REGEX_FILTER
" --input-filter <regex> The regular expression for filtering input lines\n"
" or filenames respectively.\n"
#endif
" --batch-size <num> Set the size of batches that are read and \n"
" processed in one go (Default: %"ZU").\n"
#ifdef USE_NETWORK
" -p, --pcap-filter <str> Filter expression for the PCAP library in case\n"
" network data is processed (Default: %s).\n"
" --client-only Only consider the client-side of the network\n"
" communication.\n"
" --server-only Only consider the server-side of the network\n"
" communication.\n"
#endif
" -b, --bloom <file> The bloom filter to be used.\n"
" -o, --output <file> The output filename.\n"
"\n"
"Feature options:\n"
" -n, --ngram-len <num> Set length of n-grams (Default: %"ZU").\n"
" -d, --ngram-delim <delim> Set delimiters for the use of word/ token n-grams.\n"
" If omitted or empty byte n-grams are used.\n"
" --binary Indicates to use bit n-grams rather than byte\n"
" or token n-grams and consequently, disables the\n"
" --ngram-delim option.\n"
" -s, --filter-size <num> Set the size of the bloom filter as bits of\n"
" the index (Default: %u).\n"
" --hash-set <hashes> Set the hash set to be used: 'simple', 'simple2'\n"
" or 'murmur' (Default: '%s').\n"
"\n"
"Generic options:\n"
" -e, --echo-params Echo used parameters and settings.\n"
" -h, --help Print this help screen.\n",
/* --batch-size */ (SIZE_T) DEFAULT_CONFIG.batch_size,
#ifdef USE_NETWORK
/* --pcap-filter */ DEFAULT_CONFIG.pcap_filter,
#endif
/* --ngram-len */ (SIZE_T) DEFAULT_CONFIG.ngram_length,
/* --filter-size */ DEFAULT_CONFIG.filter_size,
/* --hash-set */ hashset_to_string(DEFAULT_CONFIG.hash_set));
return EXIT_SUCCESS;
}
const saladstate_t parse_traininglike_options_ex(int argc, char* argv[], config_t* const config,
const char *shortopts, const struct option *longopts)
{
assert(argv != NULL);
assert(config != NULL);
int option, bs = FALSE, fo = FALSE;
while ((option = getopt_long(argc, argv, shortopts, longopts, NULL)) != -1)
{
switch (option)
{
case 'i':
config->input = optarg;
break;
case 'f':
config->input_type = as_iomode(optarg);
break;
case OPTION_INPUTFILTER:
config->input_filter = optarg;
break;
case OPTION_BATCHSIZE:
{
int batch_size = atoi(optarg); // TODO: strtol
if (batch_size <= 0)
{
warn("Illegal batch size specified.\n");
// This is not true in case of network data as input. Therefore,
// we simply suppress this output at this point.
// warn("Defaulting to: %u\n", (unsigned int) config->batch_size);
}
else
{
bs = TRUE;
config->batch_size = batch_size;
}
break;
}
#ifdef USE_NETWORK
case 'p':
config->pcap_filter = optarg;
break;
#endif
case 'b':
config->bloom = optarg;
break;
case 'u':
config->update_model = TRUE;
break;
case 'o':
config->output = optarg;
break;
case 'n':
{
fo = TRUE;
int ngramLength = atoi(optarg); // TODO: strtol
if (ngramLength <= 0)
{
warn("Illegal n-gram length specified.");
warn("Defaulting to: %u\n", (unsigned int) config->ngramLength);
}
else config->ngramLength = ngramLength;
break;
}
case 'd':
fo = TRUE;
config->delimiter = optarg;
break;
case 's':
{
fo = TRUE;
int filter_size = atoi(optarg); // TODO: strtol
if (filter_size <= 0)
{
warn("Illegal filter size specified.");
warn("Defaulting to: %u\n", (unsigned int) config->filter_size);
}
else config->filter_size = filter_size;
break;
}
case OPTION_HASHSET:
{
fo = TRUE;
hashset_t hashset = to_hashset(optarg);
if (hashset == HASHES_UNDEFINED)
{
warn("Illegal hash set specified.");
warn("Defaulting to: %s\n", hashset_to_string(config->hash_set));
}
else config->hash_set = hashset;
break;
}
case 'e':
config->echo_params = TRUE;
break;
case '?':
case 'h':
return SALAD_HELP_TRAIN;
default:
// In order to catch program argument that correspond to
// features that were excluded at compile time.
fprintf(stderr, "invalid option -- '%c'\n", option);
return SALAD_HELP_TRAIN;
}
}
config->transfer_spec = !fo;
if (check_input(config, TRUE, bs) == EXIT_FAILURE) return SALAD_EXIT;
if (check_output(config) == EXIT_FAILURE) return SALAD_EXIT;
if (config->echo_params)
{
if (config->update_model && config->transfer_spec) {
// cf. salad_train_stub
} else {
echo_options(config);
}
}
return SALAD_RUN;
}
const saladstate_t parse_traininglike_options_ex(int argc, char* argv[], config_t* const config,
const char *shortopts, const struct option *longopts)
{
assert(argv != NULL);
assert(config != NULL);
char* end; // For parsing numbers with strto*
int conly = FALSE, sonly = FALSE;
int option, bs = FALSE, fo = FALSE;
while ((option = getopt_long(argc, argv, shortopts, longopts, NULL)) != -1)
{
switch (option)
{
case 'i':
config->input = optarg;
break;
case 'f':
config->input_type = as_inputmode(optarg);
break;
case OPTION_INPUTFILTER:
config->input_filter = optarg;
break;
case OPTION_BATCHSIZE:
{
const long long int batch_size = strtoll(optarg, &end, 10);
if (batch_size <= 0)
{
warn("Illegal batch size specified.\n");
// This is not true in case of network data as input. Therefore,
// we simply suppress this output at this point.
// warn("Defaulting to: %u\n", (unsigned int) config->batch_size);
}
else
{
bs = TRUE;
config->batch_size = (size_t) MIN(SIZE_MAX, (unsigned long) MAX(0, batch_size));
}
break;
}
#ifdef USE_NETWORK
case 'p':
config->pcap_filter = optarg;
break;
case OPTION_NETCLIENT:
conly = TRUE;
break;
case OPTION_NETSERVER:
sonly = TRUE;
break;
#endif
case 'b':
config->bloom = optarg;
break;
case 'u':
config->update_model = TRUE;
break;
case 'o':
config->output = optarg;
break;
case 'F':
config->output_type = as_outputmode(optarg);
break;
case 'n':
{
fo = TRUE;
const long long int ngram_length = strtoll(optarg, &end, 10);
if (ngram_length <= 0)
{
warn("Illegal n-gram length specified.");
warn("Defaulting to: %"ZU"\n", (SIZE_T) config->ngram_length);
}
else config->ngram_length = (size_t) MIN(SIZE_MAX, (unsigned long) ngram_length);
break;
}
case 'd':
fo = TRUE;
config->delimiter = optarg;
break;
case OPTION_BINARY:
config->binary_ngrams = TRUE;
break;
case 's':
{
fo = TRUE;
const long long int filter_size = strtoll(optarg, &end, 10);
if (filter_size <= 0)
{
warn("Illegal filter size specified.");
warn("Defaulting to: %u\n", (unsigned int) config->filter_size);
}
else config->filter_size = (unsigned int) MIN(UINT_MAX, (unsigned long) MAX(0, filter_size));
break;
}
case OPTION_HASHSET:
{
fo = TRUE;
hashset_t hashset = to_hashset(optarg);
if (hashset == HASHES_UNDEFINED)
{
warn("Illegal hash set specified.");
warn("Defaulting to: %s\n", hashset_to_string(config->hash_set));
}
else config->hash_set = hashset;
break;
}
case 'e':
config->echo_params = TRUE;
break;
case 'q':
log_level = WARNING;
break;
case '?':
case 'h':
log_level = STATUS;
return SALAD_HELP_TRAIN;
default:
// In order to catch program argument that correspond to
// features that were excluded at compile time.
fprintf(stderr, "invalid option -- '%c'\n", option);
return SALAD_HELP_TRAIN;
}
}
config->transfer_spec = !fo;
if (config->binary_ngrams && config->ngram_length > MASK_BITSIZE)
{
error("When using binary n-grams currently only a maximal");
error("length of %u bits is supported.", MASK_BITSIZE);
return SALAD_EXIT;
}
if (check_netparams(config, conly, sonly) == EXIT_FAILURE) return SALAD_HELP_TRAIN;
if (check_input(config, TRUE, bs) == EXIT_FAILURE) return SALAD_EXIT;
if (check_output(config) == EXIT_FAILURE) return SALAD_EXIT;
if (config->echo_params)
{
if (config->update_model && config->transfer_spec) {
// cf. salad_train_stub
} else {
echo_options(config);
}
}
return SALAD_RUN;
}
