static int
apply_must_form_a_cycle(Postprocessor *pp,Sublinkage *sublinkage,pp_rule *rule)
{
/* Returns TRUE if the linkage is connected when ignoring the links
whose names are in the given list of link names.
Actually, what it does is this: it returns FALSE if the connectivity
of the subgraph reachable from word 0 changes as a result of deleting
these links. */
List_o_links *lol;
int w;
for (w=0; w<pp->pp_data.length; w++) {
for (lol = pp->pp_data.word_links[w]; lol != NULL; lol = lol->next) {
if (w > lol->word) continue; /* only consider each edge once */
if (!pp_linkset_match(rule->link_set, sublinkage->link[lol->link]->name)) continue;
memset(pp->visited, 0, pp->pp_data.length*(sizeof pp->visited[0]));
reachable_without_dfs(pp, sublinkage, w, lol->word, w);
if (!pp->visited[lol->word]) return FALSE;
}
}
for (lol = pp->pp_data.links_to_ignore; lol != NULL; lol = lol->next) {
w = sublinkage->link[lol->link]->l;
/* (w, lol->word) are the left and right ends of the edge we're considering */
if (!pp_linkset_match(rule->link_set, sublinkage->link[lol->link]->name)) continue;
memset(pp->visited, 0, pp->pp_data.length*(sizeof pp->visited[0]));
reachable_without_dfs(pp, sublinkage, w, lol->word, w);
if (!pp->visited[lol->word]) return FALSE;
}
return TRUE;
}
/**
* Returns true if the linkage is connected when ignoring the links
* whose names are in the given list of link names.
* Actually, what it does is this: it returns false if the connectivity
* of the subgraph reachable from word 0 changes as a result of deleting
* these links.
*/
static bool
apply_must_form_a_cycle(PP_data *pp_data, Linkage sublinkage, pp_rule *rule)
{
List_o_links *lol;
size_t w;
for (w = 0; w < pp_data->num_words; w++)
{
for (lol = pp_data->word_links[w]; lol != NULL; lol = lol->next)
{
if (w > lol->word) continue; /* only consider each edge once */
if (!pp_linkset_match(rule->link_set, sublinkage->link_array[lol->link].link_name)) continue;
clear_visited(pp_data);
reachable_without_dfs(pp_data, sublinkage, w, lol->word, w);
if (!pp_data->visited[lol->word]) return false;
}
}
for (lol = pp_data->links_to_ignore; lol != NULL; lol = lol->next)
{
w = sublinkage->link_array[lol->link].lw;
/* (w, lol->word) are the left and right ends of the edge we're considering */
if (!pp_linkset_match(rule->link_set, sublinkage->link_array[lol->link].link_name)) continue;
clear_visited(pp_data);
reachable_without_dfs(pp_data, sublinkage, w, lol->word, w);
assert(lol->word < pp_data->num_words, "Bad word index");
if (!pp_data->visited[lol->word]) return false;
}
return true;
}
static void d_depth_first_search(Postprocessor *pp, Linkage sublinkage,
size_t w, size_t root, size_t right, size_t start_link)
{
List_o_links * lol;
PP_data *pp_data = &pp->pp_data;
assert(w < pp_data->num_words, "Bad word index");
pp_data->visited[w] = true;
for (lol = pp_data->word_links[w]; lol != NULL; lol = lol->next)
{
if ((lol->word < w) && (lol->link != start_link) && (w != root))
{
add_link_to_domain(pp_data, lol->link);
}
}
for (lol = pp_data->word_links[w]; lol != NULL; lol = lol->next)
{
assert(lol->word < pp_data->num_words, "Bad word index");
if (!pp_data->visited[lol->word] && !(w == root && lol->word >= right) &&
!(w == root && lol->word < root) &&
!(lol->word < root && lol->word < w &&
pp_linkset_match(pp->knowledge->restricted_links,
sublinkage->link_array[lol->link].link_name)))
{
d_depth_first_search(pp,sublinkage,lol->word,root,right,start_link);
}
}
}
static void connectivity_dfs(Postprocessor *pp, Sublinkage *sublinkage,
int w, pp_linkset *ls)
{
List_o_links *lol;
pp->visited[w] = TRUE;
for (lol = pp->pp_data.word_links[w]; lol != NULL; lol = lol->next) {
if (!pp->visited[lol->word] &&
!pp_linkset_match(ls, sublinkage->link[lol->link]->name))
connectivity_dfs(pp, sublinkage, lol->word, ls);
}
}
static void connectivity_dfs(Postprocessor *pp, Linkage sublinkage,
int w, pp_linkset *ls)
{
List_o_links *lol;
assert(w < pp_data->num_words, "Bad word index");
pp_data->visited[w] = true;
for (lol = pp_data->word_links[w]; lol != NULL; lol = lol->next)
{
if (!pp_data->visited[lol->word] &&
!pp_linkset_match(ls, sublinkage->link[lol->link]->name))
connectivity_dfs(pp, sublinkage, lol->word, ls);
}
}
/**
* fill in the pp->pp_data.word_links array with a list of words
* neighboring each word (actually a list of links). This is an
* undirected graph.
*/
static void build_graph(Postprocessor *pp, Linkage sublinkage)
{
size_t link;
List_o_links * lol;
PP_data *pp_data = &pp->pp_data;
/* Get more size, if needed */
if (pp_data->wowlen <= pp_data->num_words)
{
size_t newsz;
pp_data->wowlen += pp_data->num_words;
newsz = pp_data->wowlen * sizeof(List_o_links *);
pp_data->word_links = (List_o_links **) realloc(
pp_data->word_links, newsz);
}
memset(pp_data->word_links, 0, pp_data->wowlen * sizeof(List_o_links *));
for (link = 0; link < sublinkage->num_links; link++)
{
assert (sublinkage->link_array[link].lw != SIZE_MAX);
if (NULL == sublinkage->link_array[link].link_name) continue;
if (pp_linkset_match(pp->knowledge->ignore_these_links,
sublinkage->link_array[link].link_name))
{
lol = (List_o_links *) malloc(sizeof(List_o_links));
lol->next = pp_data->links_to_ignore;
pp_data->links_to_ignore = lol;
lol->link = link;
lol->word = sublinkage->link_array[link].rw;
continue;
}
lol = (List_o_links *) malloc(sizeof(List_o_links));
lol->next = pp_data->word_links[sublinkage->link_array[link].lw];
pp_data->word_links[sublinkage->link_array[link].lw] = lol;
lol->link = link;
lol->word = sublinkage->link_array[link].rw;
lol = (List_o_links *) malloc(sizeof(List_o_links));
lol->next = pp_data->word_links[sublinkage->link_array[link].rw];
pp_data->word_links[sublinkage->link_array[link].rw] = lol;
lol->link = link;
lol->word = sublinkage->link_array[link].lw;
}
}
static void bad_depth_first_search(Postprocessor *pp, Sublinkage *sublinkage,
int w, int root, int start_link)
{
List_o_links * lol;
pp->visited[w] = TRUE;
for (lol = pp->pp_data.word_links[w]; lol != NULL; lol = lol->next) {
if ((lol->word < w) && (lol->link != start_link) && (w != root)) {
add_link_to_domain(pp, lol->link);
}
}
for (lol = pp->pp_data.word_links[w]; lol != NULL; lol = lol->next) {
if ((!pp->visited[lol->word]) && !(w == root && lol->word < w) &&
!(lol->word < root && lol->word < w &&
pp_linkset_match(pp->knowledge->restricted_links,
sublinkage->link[lol->link]->name)))
bad_depth_first_search(pp, sublinkage, lol->word, root, start_link);
}
}
static void build_graph(Postprocessor *pp, Sublinkage *sublinkage)
{
/* fill in the pp->pp_data.word_links array with a list of words neighboring each
word (actually a list of links). The dir fields are not set, since this
(after fat-link-extraction) is an undirected graph. */
int i, link;
List_o_links * lol;
for (i=0; i<pp->pp_data.length; i++)
pp->pp_data.word_links[i] = NULL;
for (link=0; link<sublinkage->num_links; link++)
{
if (sublinkage->link[link]->l == -1) continue;
if (pp_linkset_match(pp->knowledge->ignore_these_links, sublinkage->link[link]->name)) {
lol = (List_o_links *) xalloc(sizeof(List_o_links));
lol->next = pp->pp_data.links_to_ignore;
pp->pp_data.links_to_ignore = lol;
lol->link = link;
lol->word = sublinkage->link[link]->r;
continue;
}
lol = (List_o_links *) xalloc(sizeof(List_o_links));
lol->next = pp->pp_data.word_links[sublinkage->link[link]->l];
pp->pp_data.word_links[sublinkage->link[link]->l] = lol;
lol->link = link;
lol->word = sublinkage->link[link]->r;
lol = (List_o_links *) xalloc(sizeof(List_o_links));
lol->next = pp->pp_data.word_links[sublinkage->link[link]->r];
pp->pp_data.word_links[sublinkage->link[link]->r] = lol;
lol->link = link;
lol->word = sublinkage->link[link]->l;
}
}
static void build_domains(Postprocessor *pp, Linkage sublinkage)
{
size_t link, i, d;
const char *s;
PP_data *pp_data = &pp->pp_data;
pp_data->N_domains = 0;
for (link = 0; link<sublinkage->num_links; link++)
{
assert (sublinkage->link_array[link].lw != SIZE_MAX);
if (NULL == sublinkage->link_array[link].link_name) continue;
s = sublinkage->link_array[link].link_name;
if (pp_linkset_match(pp->knowledge->ignore_these_links, s)) continue;
if (pp_linkset_match(pp->knowledge->domain_starter_links, s))
{
setup_domain_array(pp, s, link);
if (pp_linkset_match(pp->knowledge->domain_contains_links, s))
add_link_to_domain(pp_data, link);
clear_visited(pp_data);
depth_first_search(pp, sublinkage, sublinkage->link_array[link].rw,
sublinkage->link_array[link].lw, link);
}
else
if (pp_linkset_match(pp->knowledge->urfl_domain_starter_links, s))
{
setup_domain_array(pp, s, link);
/* always add the starter link to its urfl domain */
add_link_to_domain(pp_data, link);
clear_visited(pp_data);
bad_depth_first_search(pp, sublinkage,sublinkage->link_array[link].rw,
sublinkage->link_array[link].lw, link);
}
else
if (pp_linkset_match(pp->knowledge->urfl_only_domain_starter_links, s))
{
setup_domain_array(pp, s, link);
/* do not add the starter link to its urfl_only domain */
clear_visited(pp_data);
d_depth_first_search(pp, sublinkage, sublinkage->link_array[link].lw,
sublinkage->link_array[link].lw,
sublinkage->link_array[link].rw, link);
}
else
if (pp_linkset_match(pp->knowledge->left_domain_starter_links, s))
{
setup_domain_array(pp, s, link);
/* do not add the starter link to a left domain */
clear_visited(pp_data);
left_depth_first_search(pp, sublinkage, sublinkage->link_array[link].lw,
sublinkage->link_array[link].rw, link);
}
}
/* sort the domains by size */
qsort((void *) pp_data->domain_array,
pp_data->N_domains,
sizeof(Domain),
(int (*)(const void *, const void *)) domain_compare);
/* sanity check: all links in all domains have a legal domain name */
for (d = 0; d < pp_data->N_domains; d++)
{
i = find_domain_name(pp, pp_data->domain_array[d].string);
if (i == SIZE_MAX)
prt_error("Error: post_process(): Need an entry for %s in LINK_TYPE_TABLE",
pp_data->domain_array[d].string);
pp_data->domain_array[d].type = i;
}
}
static void build_domains(Postprocessor *pp, Sublinkage *sublinkage)
{
int link, i, d;
char *s;
pp->pp_data.N_domains = 0;
for (link = 0; link<sublinkage->num_links; link++) {
if (sublinkage->link[link]->l == -1) continue;
s = sublinkage->link[link]->name;
if (pp_linkset_match(pp->knowledge->ignore_these_links, s)) continue;
if (pp_linkset_match(pp->knowledge->domain_starter_links, s))
{
setup_domain_array(pp, pp->pp_data.N_domains, s, link);
if (pp_linkset_match(pp->knowledge->domain_contains_links, s))
add_link_to_domain(pp, link);
depth_first_search(pp,sublinkage,sublinkage->link[link]->r,
sublinkage->link[link]->l, link);
pp->pp_data.N_domains++;
assert(pp->pp_data.N_domains<PP_MAX_DOMAINS, "raise value of PP_MAX_DOMAINS");
}
else {
if (pp_linkset_match(pp->knowledge->urfl_domain_starter_links,s))
{
setup_domain_array(pp, pp->pp_data.N_domains, s, link);
/* always add the starter link to its urfl domain */
add_link_to_domain(pp, link);
bad_depth_first_search(pp,sublinkage,sublinkage->link[link]->r,
sublinkage->link[link]->l,link);
pp->pp_data.N_domains++;
assert(pp->pp_data.N_domains<PP_MAX_DOMAINS,"raise PP_MAX_DOMAINS value");
}
else
if (pp_linkset_match(pp->knowledge->urfl_only_domain_starter_links,s))
{
setup_domain_array(pp, pp->pp_data.N_domains, s, link);
/* do not add the starter link to its urfl_only domain */
d_depth_first_search(pp,sublinkage, sublinkage->link[link]->l,
sublinkage->link[link]->l,
sublinkage->link[link]->r,link);
pp->pp_data.N_domains++;
assert(pp->pp_data.N_domains<PP_MAX_DOMAINS,"raise PP_MAX_DOMAINS value");
}
else
if (pp_linkset_match(pp->knowledge->left_domain_starter_links,s))
{
setup_domain_array(pp, pp->pp_data.N_domains, s, link);
/* do not add the starter link to a left domain */
left_depth_first_search(pp,sublinkage, sublinkage->link[link]->l,
sublinkage->link[link]->r,link);
pp->pp_data.N_domains++;
assert(pp->pp_data.N_domains<PP_MAX_DOMAINS,"raise PP_MAX_DOMAINS value");
}
}
}
/* sort the domains by size */
qsort((void *) pp->pp_data.domain_array,
pp->pp_data.N_domains,
sizeof(Domain),
(int (*)(const void *, const void *)) domain_compare);
/* sanity check: all links in all domains have a legal domain name */
for (d=0; d<pp->pp_data.N_domains; d++) {
i = find_domain_name(pp, pp->pp_data.domain_array[d].string);
if (i==-1)
error("\tpost_process: Need an entry for %s in LINK_TYPE_TABLE",
pp->pp_data.domain_array[d].string);
pp->pp_data.domain_array[d].type = i;
}
}