/**
* Create and initialize a Parse_Options object
*/
Parse_Options parse_options_create(void)
{
Parse_Options po;
init_memusage();
po = (Parse_Options) malloc(sizeof(struct Parse_Options_s));
/* Here's where the values are initialized */
/* The parse_options_set_(verbosity|debug|test) functions set also the
* corresponding global variables. So these globals are initialized
* here too. */
verbosity = po->verbosity = 1;
debug = po->debug = (char *)"";
test = po->test = (char *)"";
/* A cost of 2.7 allows the usual cost-2 connectors, plus the
* assorted fractional costs, without going to cost 3.0, which
* is used only during panic-parsing.
* XXX In the long run, this should be fetched from the dictionary
* (and should probably not be a parse option).
*/
po->disjunct_cost = 2.7;
po->min_null_count = 0;
po->max_null_count = 0;
po->islands_ok = false;
po->use_sat_solver = false;
po->use_viterbi = false;
po->linkage_limit = 100;
#if defined HAVE_HUNSPELL || defined HAVE_ASPELL
po->use_spell_guess = 7;
#else
po->use_spell_guess = 0;
#endif /* defined HAVE_HUNSPELL || defined HAVE_ASPELL */
#ifdef XXX_USE_CORPUS
/* Use the corpus cost model, if available.
* It really does a better job at parse ranking.
* Err .. sometimes ...
*/
po->cost_model.compare_fn = &CORP_compare_parse;
po->cost_model.type = CORPUS;
#else /* USE_CORPUS */
po->cost_model.compare_fn = &VDAL_compare_parse;
po->cost_model.type = VDAL;
#endif /* USE_CORPUS */
po->short_length = 16;
po->all_short = false;
po->perform_pp_prune = true;
po->twopass_length = 30;
po->repeatable_rand = true;
po->resources = resources_create();
po->use_cluster_disjuncts = false;
po->display_morphology = false;
return po;
}
/**
* Create and initialize a Parse_Options object
*/
Parse_Options parse_options_create(void)
{
Parse_Options po;
init_memusage();
po = (Parse_Options) xalloc(sizeof(struct Parse_Options_s));
/* Here's where the values are initialized */
po->verbosity = 1;
po->debug = (char *)"";
po->test = (char *)"";
po->linkage_limit = 100;
/* A cost of 2.7 allows the usual cost-2 connectors, plus the
* assorted fractional costs, without going to cost 3.0, which
* is used only during panic-parsing.
*/
po->disjunct_cost = 2.7; /* 3.0 is needed for Russian dicts */
po->min_null_count = 0;
po->max_null_count = 0;
po->islands_ok = false;
po->use_spell_guess = 7;
po->use_sat_solver = false;
po->use_viterbi = false;
#ifdef XXX_USE_CORPUS
/* Use the corpus cost model, if available.
* It really does a better job at parse ranking.
* Err .. sometimes ...
*/
po->cost_model.compare_fn = &CORP_compare_parse;
po->cost_model.type = CORPUS;
#else /* USE_CORPUS */
po->cost_model.compare_fn = &VDAL_compare_parse;
po->cost_model.type = VDAL;
#endif /* USE_CORPUS */
po->short_length = 16;
po->all_short = false;
po->twopass_length = 30;
po->repeatable_rand = true;
po->resources = resources_create();
po->use_cluster_disjuncts = false;
po->display_morphology = false;
return po;
}
/**
* Use "string" as the input dictionary. All of the other parts,
* including post-processing, affix table, etc, are NULL.
* This routine is intended for unit-testing ONLY.
*/
Dictionary dictionary_create_from_utf8(const char * input)
{
Dictionary dictionary = NULL;
char * lang;
init_memusage();
lang = get_default_locale();
if (lang && *lang) {
dictionary = dictionary_six_str(lang, input, "string",
NULL, NULL, NULL, NULL);
free(lang);
} else {
/* Default to en when locales are broken (e.g. WIN32) */
dictionary = dictionary_six_str("en", input, "string",
NULL, NULL, NULL, NULL);
}
return dictionary;
}
Dictionary dictionary_create_from_file(const char * lang)
{
Dictionary dictionary;
init_memusage();
if (lang && *lang)
{
char * dict_name;
char * pp_name;
char * cons_name;
char * affix_name;
char * regex_name;
dict_name = join_path(lang, "4.0.dict");
pp_name = join_path(lang, "4.0.knowledge");
cons_name = join_path(lang, "4.0.constituent-knowledge");
affix_name = join_path(lang, "4.0.affix");
regex_name = join_path(lang, "4.0.regex");
dictionary = dictionary_six(lang, dict_name, pp_name, cons_name,
affix_name, regex_name);
free(regex_name);
free(affix_name);
free(cons_name);
free(pp_name);
free(dict_name);
}
else
{
prt_error("Error: No language specified!");
dictionary = NULL;
}
return dictionary;
}