static pp_linkset *read_link_set(pp_knowledge *k,
const char *label, String_set *ss)
{
/* read link set, marked by label in knowledge file, into a set of links
whose handle is returned. Return NULL if link set not defined in file,
in which case the set is taken to be empty. */
int n_strings,i;
pp_linkset *ls;
if (!pp_lexer_set_label(k->lt, label))
{
if (verbosity_level(+D_PPK))
prt_error("Warning: File %s: Link set %s not defined: assuming empty\n",
k->path, label);
n_strings = 0;
}
else
{
n_strings = pp_lexer_count_tokens_of_label(k->lt);
if (-1 == n_strings) return &LINK_SET_ERROR;
}
ls = pp_linkset_open(n_strings);
for (i=0; i<n_strings; i++)
pp_linkset_add(ls,
string_set_add(pp_lexer_get_next_token_of_label(k->lt),ss));
return ls;
}
void prune_irrelevant_rules(Postprocessor *pp)
{
/* call this (a) after having called post_process_scan_linkage() on all
generated linkages, but (b) before calling post_process() on any
particular linkage. Here we mark all rules which we know (from having
accumulated a set of link names appearing in *any* linkage) won't
ever be needed. */
pp_rule *rule;
int coIDX, cnIDX, rcoIDX=0, rcnIDX=0;
/* If we didn't scan any linkages, there's no pruning to be done. */
if (pp_linkset_population(pp->set_of_links_of_sentence)==0) return;
for (coIDX=0;;coIDX++)
{
rule = &(pp->knowledge->contains_one_rules[coIDX]);
if (rule->msg==NULL) break;
if (pp_linkset_match_bw(pp->set_of_links_of_sentence, rule->selector))
{
/* mark rule as being relevant to this sentence */
pp->relevant_contains_one_rules[rcoIDX++] = coIDX;
pp_linkset_add(pp->set_of_links_in_an_active_rule, rule->selector);
}
}
pp->relevant_contains_one_rules[rcoIDX] = -1; /* end sentinel */
for (cnIDX=0;;cnIDX++)
{
rule = &(pp->knowledge->contains_none_rules[cnIDX]);
if (rule->msg==NULL) break;
if (pp_linkset_match_bw(pp->set_of_links_of_sentence, rule->selector))
{
pp->relevant_contains_none_rules[rcnIDX++] = cnIDX;
pp_linkset_add(pp->set_of_links_in_an_active_rule, rule->selector);
}
}
pp->relevant_contains_none_rules[rcnIDX] = -1;
if (verbosity>1) {
printf("Saw %i unique link names in all linkages.\n",
pp_linkset_population(pp->set_of_links_of_sentence));
printf("Using %i 'contains one' rules and %i 'contains none' rules\n",
rcoIDX, rcnIDX);
}
}
static void read_contains_rules(pp_knowledge *k, const char *label,
pp_rule **rules, int *nRules)
{
/* Reading the 'contains_one_rules' and reading the
'contains_none_rules' into their respective arrays */
int n_commas, n_tokens, i, r;
const char *p;
const char **tokens;
if (!pp_lexer_set_label(k->lt, label)) {
*nRules = 0;
if (verbosity>0) printf("PP warning: Not using any %s rules\n", label);
}
else {
n_commas = pp_lexer_count_commas_of_label(k->lt);
*nRules = (n_commas + 1)/3;
}
*rules = (pp_rule*) xalloc ((1+*nRules)*sizeof(pp_rule));
for (r=0; r<*nRules; r++)
{
/* first read link */
tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
if (n_tokens>1)
{
prt_error("Fatal Error: post_process: Invalid syntax in %s (rule %i)",label,r+1);
exit(1);
}
(*rules)[r].selector = string_set_add(tokens[0], k->string_set);
/* read link set */
tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
(*rules)[r].link_set = pp_linkset_open(n_tokens);
(*rules)[r].link_set_size = n_tokens;
(*rules)[r].link_array = (const char **) xalloc((1+n_tokens)*sizeof(const char*));
for (i=0; i<n_tokens; i++)
{
p = string_set_add(tokens[i], k->string_set);
pp_linkset_add((*rules)[r].link_set, p);
(*rules)[r].link_array[i] = p;
}
(*rules)[r].link_array[i]=0; /* NULL-terminator */
/* read error message */
tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
if (n_tokens>1)
{
prt_error("Fatal Error: post_process: Invalid syntax in %s (rule %i)",label,r+1);
exit(1);
}
(*rules)[r].msg = string_set_add(tokens[0], k->string_set);
}
/* sentinel entry */
(*rules)[*nRules].msg = 0;
}
/**
* Call this (a) after having called post_process_scan_linkage() on all
* generated linkages, but (b) before calling post_process() on any
* particular linkage. Here we mark all rules which we know (from having
* accumulated a set of link names appearing in *any* linkage) that won't
* ever be needed.
*/
static void prune_irrelevant_rules(Postprocessor *pp)
{
pp_rule *rule;
int coIDX, cnIDX, rcoIDX = 0, rcnIDX = 0;
/* If we didn't scan any linkages, there's no pruning to be done. */
if (pp_linkset_population(pp->set_of_links_of_sentence) == 0) return;
for (coIDX = 0; ; coIDX++)
{
rule = &(pp->knowledge->contains_one_rules[coIDX]);
if (rule->msg == NULL) break;
if (pp_linkset_match_bw(pp->set_of_links_of_sentence, rule->selector))
{
/* Mark rule as being relevant to this sentence */
pp->relevant_contains_one_rules[rcoIDX++] = coIDX;
pp_linkset_add(pp->set_of_links_in_an_active_rule, rule->selector);
}
}
pp->relevant_contains_one_rules[rcoIDX] = -1; /* end sentinel */
for (cnIDX = 0; ; cnIDX++)
{
rule = &(pp->knowledge->contains_none_rules[cnIDX]);
if (rule->msg == NULL) break;
if (pp_linkset_match_bw(pp->set_of_links_of_sentence, rule->selector))
{
pp->relevant_contains_none_rules[rcnIDX++] = cnIDX;
pp_linkset_add(pp->set_of_links_in_an_active_rule, rule->selector);
}
}
pp->relevant_contains_none_rules[rcnIDX] = -1;
if (debug_level(5))
{
printf("PP: Saw %zd unique link names in all linkages.\n",
pp_linkset_population(pp->set_of_links_of_sentence));
printf("PP: Using %i 'contains one' rules and %i 'contains none' rules\n",
rcoIDX, rcnIDX);
}
}
/**
* During a first pass (prior to actual post-processing of the linkages
* of a sentence), call this once for every generated linkage. Here we
* simply maintain a set of "seen" link names for rule pruning, later on.
*/
void post_process_scan_linkage(Postprocessor *pp, Linkage linkage)
{
size_t i;
if (pp == NULL) return;
for (i = 0; i < linkage->num_links; i++)
{
assert(linkage->link_array[i].lw != SIZE_MAX);
pp_linkset_add(pp->set_of_links_of_sentence,
linkage->link_array[i].link_name);
}
}
static void initialize_set_of_links_starting_bounded_domain(pp_knowledge *k)
{
int i,j,d,domain_of_rule;
k->set_of_links_starting_bounded_domain =
pp_linkset_open(PP_MAX_UNIQUE_LINK_NAMES);
for (i=0; k->bounded_rules[i].msg!=0; i++)
{
domain_of_rule = k->bounded_rules[i].domain;
for (j=0; (d=(k->starting_link_lookup_table[j].domain))!=-1; j++)
if (d==domain_of_rule)
pp_linkset_add(k->set_of_links_starting_bounded_domain,
k->starting_link_lookup_table[j].starting_link);
}
}
static bool read_form_a_cycle_rules(pp_knowledge *k, const char *label)
{
size_t n_commas, n_tokens;
size_t r, i;
pp_linkset *lsHandle;
const char **tokens;
if (!pp_lexer_set_label(k->lt, label)) {
k->n_form_a_cycle_rules = 0;
if (verbosity_level(+D_PPK))
prt_error("Warning: File %s: Not using any 'form a cycle' rules\n",
k->path);
}
else {
n_commas = pp_lexer_count_commas_of_label(k->lt);
k->n_form_a_cycle_rules = (n_commas + 1)/2;
}
k->form_a_cycle_rules=
(pp_rule*) malloc ((1+k->n_form_a_cycle_rules)*sizeof(pp_rule));
for (r=0; r<k->n_form_a_cycle_rules; r++)
{
/* read link set */
tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
if (n_tokens <= 0)
{
prt_error("Error: File %s: Syntax error\n", k->path);
return false;
}
lsHandle = pp_linkset_open(n_tokens);
for (i=0; i<n_tokens; i++)
pp_linkset_add(lsHandle,string_set_add(tokens[i], k->string_set));
k->form_a_cycle_rules[r].link_set = lsHandle;
/* read error message */
tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
if (n_tokens > 1)
{
prt_error("Error: File %s: Invalid syntax (rule %zu of %s)\n",
k->path, r+1,label);
return false;
}
k->form_a_cycle_rules[r].msg = string_set_add(tokens[0], k->string_set);
k->form_a_cycle_rules[r].use_count = 0;
}
/* sentinel entry */
k->form_a_cycle_rules[k->n_form_a_cycle_rules].msg = 0;
k->form_a_cycle_rules[k->n_form_a_cycle_rules].use_count = 0;
return true;
}
void post_process_scan_linkage(Postprocessor *pp, Parse_Options opts,
Sentence sent, Sublinkage *sublinkage)
{
/* During a first pass (prior to actual post-processing of the linkages
of a sentence), call this once for every generated linkage. Here we
simply maintain a set of "seen" link names for rule pruning later on */
char *p;
int i;
if (pp==NULL) return;
if (sent->length < opts->twopass_length) return;
for (i=0; i<sublinkage->num_links; i++)
{
if (sublinkage->link[i]->l == -1) continue;
p=string_set_add(sublinkage->link[i]->name,pp->sentence_link_name_set);
pp_linkset_add(pp->set_of_links_of_sentence, p);
}
}
static void read_form_a_cycle_rules(pp_knowledge *k, const char *label)
{
int n_commas, n_tokens, r, i;
pp_linkset *lsHandle;
const char **tokens;
if (!pp_lexer_set_label(k->lt, label)) {
k->n_form_a_cycle_rules = 0;
if (verbosity>0)
printf("PP warning: Not using any 'form a cycle' rules\n");
}
else {
n_commas = pp_lexer_count_commas_of_label(k->lt);
k->n_form_a_cycle_rules = (n_commas + 1)/2;
}
k->form_a_cycle_rules=
(pp_rule*) xalloc ((1+k->n_form_a_cycle_rules)*sizeof(pp_rule));
for (r=0; r<k->n_form_a_cycle_rules; r++)
{
/* read link set */
tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
if (n_tokens <= 0)
{
prt_error("Fatal Error: syntax error in knowledge file");
exit(1);
}
lsHandle = pp_linkset_open(n_tokens);
for (i=0; i<n_tokens; i++)
pp_linkset_add(lsHandle,string_set_add(tokens[i], k->string_set));
k->form_a_cycle_rules[r].link_set=lsHandle;
/* read error message */
tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
if (n_tokens > 1)
{
prt_error("Fatal Error: post_process: Invalid syntax (rule %i of %s)",r+1,label);
exit(1);
}
k->form_a_cycle_rules[r].msg=string_set_add(tokens[0],k->string_set);
}
/* sentinel entry */
k->form_a_cycle_rules[k->n_form_a_cycle_rules].msg = 0;
}
static pp_linkset *read_link_set(pp_knowledge *k,
const char *label, String_set *ss)
{
/* read link set, marked by label in knowledge file, into a set of links
whose handle is returned. Return NULL if link set not defined in file,
in which case the set is taken to be empty. */
int n_strings,i;
pp_linkset *ls;
if (!pp_lexer_set_label(k->lt, label)) {
if (verbosity>0)
printf("PP warning: Link set %s not defined: assuming empty.\n",label);
n_strings = 0;
}
else n_strings = pp_lexer_count_tokens_of_label(k->lt);
ls = pp_linkset_open(n_strings);
for (i=0; i<n_strings; i++)
pp_linkset_add(ls,
string_set_add(pp_lexer_get_next_token_of_label(k->lt),ss));
return ls;
}
static bool read_contains_rules(pp_knowledge *k, const char *label,
pp_rule **rules, size_t *nRules)
{
/* Reading the 'contains_one_rules' and reading the
'contains_none_rules' into their respective arrays */
size_t n_tokens, i, r;
int n_commas;
const char *p;
const char **tokens;
if (!pp_lexer_set_label(k->lt, label)) {
*nRules = 0;
if (verbosity_level(+D_PPK))
prt_error("Warning: File %s: Not using any %s rules\n", k->path, label);
}
else {
n_commas = pp_lexer_count_commas_of_label(k->lt);
if (-1 == n_commas) return false;
*nRules = (n_commas + 1)/3;
}
*rules = (pp_rule*) malloc ((1+*nRules)*sizeof(pp_rule));
for (r=0; r<*nRules; r++)
{
/* first read link */
tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
if (n_tokens > 1)
{
prt_error("Error: File %s: Invalid syntax in %s (rule %zu)\n",
k->path, label, r+1);
return false;
}
(*rules)[r].selector = string_set_add(tokens[0], k->string_set);
/* read link set */
tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
(*rules)[r].link_set = pp_linkset_open(n_tokens);
(*rules)[r].link_set_size = n_tokens;
(*rules)[r].link_array = (const char **) malloc((1+n_tokens)*sizeof(const char*));
for (i=0; i<n_tokens; i++)
{
p = string_set_add(tokens[i], k->string_set);
pp_linkset_add((*rules)[r].link_set, p);
(*rules)[r].link_array[i] = p;
}
(*rules)[r].link_array[i]=0; /* NULL-terminator */
/* read error message */
tokens = pp_lexer_get_next_group_of_tokens_of_label(k->lt, &n_tokens);
if (n_tokens > 1)
{
prt_error("Error: File %s: Invalid syntax in %s (rule %zu)\n",
k->path, label, r+1);
return false;
}
(*rules)[r].msg = string_set_add(tokens[0], k->string_set);
(*rules)[r].use_count = 0;
}
/* sentinel entry */
(*rules)[*nRules].msg = 0;
(*rules)[*nRules].use_count = 0;
return true;
}