//Assign true labels to all clusters, and find members of each cluster.
void Clustering::finalize_labeling() {
assert(!done_labeling);
std::vector<int> valid_labels;
std::vector<bool> flag(N_,false);
for (int i=0; i<N_; ++i) {
cluster_alias_[i] = get_true_label(i);
if (!flag[cluster_alias_[i]]) {
flag[cluster_alias_[i]] = true;
valid_labels.push_back(cluster_alias_[i]);
}
}
const int num_valid_labels = valid_labels.size();
//std::cout << "# of valid labels " << num_valid_labels << std::endl;
cluster_members.resize(num_valid_labels);
cluster_labels_.resize(N_);
std::vector<int> label_map(N_);
for (int label=0; label<num_valid_labels; ++label) {
label_map[valid_labels[label]] = label;
}
for (int i=0; i<N_; ++i) {
const int new_label = cluster_labels_[i] = label_map[cluster_alias_[i]];
assert(new_label>=0);
assert(new_label<cluster_members.size());
cluster_members[new_label].push_back(i);
}
//std::cout << "len of cluster members " << cluster_members.size() << std::endl;
done_labeling = true;
}
idn_result_t
idn_res_decodename(idn_resconf_t ctx, idn_action_t actions, const char *from,
char *to, size_t tolen) {
idn_converter_t local_converter = NULL;
idn_converter_t idn_converter = NULL;
idn_delimitermap_t delimiter_mapper;
idn_result_t r;
labellist_t labels = NULL, l;
unsigned long *buffer = NULL;
unsigned long *saved_name = NULL;
size_t buffer_length;
int idn_is_ace;
assert(ctx != NULL && from != NULL && to != NULL);
TRACE(("idn_res_decodename(actions=%s, from=\"%s\", tolen=%d)\n",
idn__res_actionstostring(actions),
idn__debug_xstring(from, 50), (int)tolen));
if (actions & ~DECODE_MASK) {
WARNING(("idn_res_decodename: invalid actions 0x%x\n",
actions));
r = idn_invalid_action;
goto ret;
}
if (!initialized)
idn_res_initialize();
if (!enabled || actions == 0) {
r = copy_verbatim(from, to, tolen);
goto ret;
} else if (tolen <= 0) {
r = idn_buffer_overflow;
goto ret;
}
if (actions & IDN_DECODE_QUERY) {
#ifndef WITHOUT_ICONV
actions |= (IDN_DELIMMAP | IDN_MAP | IDN_NORMALIZE | \
IDN_PROHCHECK | IDN_BIDICHECK | IDN_IDNCONV | \
IDN_RTCHECK | IDN_LOCALCONV);
#else
actions |= (IDN_DELIMMAP | IDN_MAP | IDN_NORMALIZE | \
IDN_PROHCHECK | IDN_BIDICHECK | IDN_IDNCONV | \
IDN_RTCHECK);
#endif
}
/*
* Convert `from' to UCS4.
*/
local_converter = idn_resconf_getlocalconverter(ctx);
#ifndef WITHOUT_ICONV
if (local_converter == NULL) {
r = idn_invalid_name;
goto ret;
}
#endif
idn_converter = idn_resconf_getidnconverter(ctx);
if (idn_converter != NULL &&
idn_converter_isasciicompatible(idn_converter))
idn_is_ace = 1;
else
idn_is_ace = 0;
buffer_length = tolen * 2;
TRACE(("res idndecode(name=\"%s\")\n", idn__debug_xstring(from, 50)));
for (;;) {
void *new_buffer;
new_buffer = realloc(buffer, sizeof(*buffer) * buffer_length);
if (new_buffer == NULL) {
r = idn_nomemory;
goto ret;
}
buffer = (unsigned long *)new_buffer;
if ((actions & IDN_IDNCONV) &&
idn_converter != NULL && !idn_is_ace) {
r = idn_converter_convtoucs4(idn_converter, from,
buffer, buffer_length);
} else {
r = idn_ucs4_utf8toucs4(from, buffer, buffer_length);
}
if (r == idn_success)
break;
else if (r != idn_buffer_overflow)
goto ret;
buffer_length *= 2;
}
if (*buffer == '\0') {
if (tolen <= 0) {
r = idn_buffer_overflow;
goto ret;
}
*to = '\0';
r = idn_success;
goto ret;
}
/*
* Delimiter map.
*/
if (actions & IDN_DELIMMAP) {
TRACE(("res delimitermap(name=\"%s\")\n",
idn__debug_ucs4xstring(buffer, 50)));
delimiter_mapper = idn_resconf_getdelimitermap(ctx);
if (delimiter_mapper != NULL) {
r = idn_delimitermap_map(delimiter_mapper, buffer,
buffer, buffer_length);
idn_delimitermap_destroy(delimiter_mapper);
if (r != idn_success)
goto ret;
}
TRACE(("res delimitermap(): success (name=\"%s\")\n",
idn__debug_ucs4xstring(buffer, 50)));
}
/*
* Split the name into a list of labels.
*/
r = labellist_create(buffer, &labels);
if (r != idn_success)
goto ret;
/*
* Perform conversions and tests.
*/
for (l = labellist_tail(labels); l != NULL;
l = labellist_previous(l)) {
free(saved_name);
saved_name = NULL;
if (!idn__util_ucs4isasciirange(labellist_getname(l))) {
if (actions & IDN_MAP) {
r = label_map(ctx, l);
if (r != idn_success)
goto ret;
}
if (actions & IDN_NORMALIZE) {
r = label_normalize(ctx, l);
if (r != idn_success)
goto ret;
}
if (actions & IDN_PROHCHECK) {
r = label_prohcheck(ctx, l);
if (r == idn_prohibited) {
labellist_undo(l);
continue;
} else if (r != idn_success) {
goto ret;
}
}
if (actions & IDN_UNASCHECK) {
r = label_unascheck(ctx, l);
if (r == idn_prohibited) {
labellist_undo(l);
continue;
} else if (r != idn_success) {
goto ret;
}
}
if (actions & IDN_BIDICHECK) {
r = label_bidicheck(ctx, l);
if (r == idn_prohibited) {
labellist_undo(l);
continue;
} else if (r != idn_success) {
goto ret;
}
}
}
if ((actions & IDN_IDNCONV) && idn_is_ace) {
saved_name = idn_ucs4_strdup(labellist_getname(l));
if (saved_name == NULL) {
r = idn_nomemory;
goto ret;
}
r = label_idndecode(ctx, l);
if (r == idn_invalid_encoding) {
labellist_undo(l);
continue;
} else if (r != idn_success) {
goto ret;
}
}
if ((actions & IDN_RTCHECK) && saved_name != NULL) {
r = label_rtcheck(ctx, actions, l, saved_name);
if (r == idn_invalid_encoding) {
labellist_undo(l);
continue;
} else if (r != idn_success) {
goto ret;
}
}
#ifndef WITHOUT_ICONV
if (actions & IDN_LOCALCONV) {
r = label_localdecodecheck(ctx, l);
if (r != idn_success)
goto ret;
}
#endif
}
/*
* Concat a list of labels to a name.
*/
for (;;) {
void *new_buffer;
new_buffer = realloc(buffer, sizeof(*buffer) * buffer_length);
if (new_buffer == NULL) {
r = idn_nomemory;
goto ret;
}
buffer = (unsigned long *)new_buffer;
r = labellist_getnamelist(labels, buffer, buffer_length);
if (r == idn_success)
break;
else if (r != idn_buffer_overflow)
goto ret;
buffer_length *= 2;
}
if (actions & IDN_LOCALCONV) {
r = idn_converter_convfromucs4(local_converter, buffer, to,
tolen);
} else {
r = idn_ucs4_ucs4toutf8(buffer, to, tolen);
}
ret:
if (r == idn_success) {
TRACE(("idn_res_decodename(): success (to=\"%s\")\n",
idn__debug_xstring(to, 50)));
} else {
TRACE(("idn_res_decodename(): %s\n", idn_result_tostring(r)));
}
free(saved_name);
free(buffer);
if (local_converter != NULL)
idn_converter_destroy(local_converter);
if (idn_converter != NULL)
idn_converter_destroy(idn_converter);
if (labels != NULL)
labellist_destroy(labels);
return (r);
}
static idn_result_t
label_rtcheck(idn_resconf_t ctx, idn_action_t actions, labellist_t label,
const unsigned long *original_name) {
labellist_t rt_label = NULL;
const unsigned long *rt_name;
const unsigned long *cur_name;
idn_result_t r;
cur_name = labellist_getname(label);
TRACE(("res rtcheck(label=\"%s\", org_label=\"%s\")\n",
idn__debug_ucs4xstring(cur_name, 50),
idn__debug_ucs4xstring(original_name, 50)));
r = labellist_create(cur_name, &rt_label);
if (r != idn_success)
goto ret;
if (rt_label == NULL) {
if (*original_name == '\0')
r = idn_success;
else
r = idn_invalid_encoding;
goto ret;
}
if (!idn__util_ucs4isasciirange(labellist_getname(rt_label))) {
r = label_map(ctx, rt_label);
if (r != idn_success)
goto ret;
r = label_normalize(ctx, rt_label);
if (r != idn_success)
goto ret;
r = label_prohcheck(ctx, rt_label);
if (r != idn_success)
goto ret;
if (actions & IDN_UNASCHECK) {
r = label_unascheck(ctx, rt_label);
if (r != idn_success)
goto ret;
}
r = label_bidicheck(ctx, rt_label);
if (r != idn_success)
goto ret;
}
if (actions & IDN_ASCCHECK) {
r = label_asccheck(ctx, rt_label);
if (r != idn_success)
goto ret;
}
if (!idn__util_ucs4isasciirange(labellist_getname(rt_label))) {
r = label_idnencode_ace(ctx, rt_label);
if (r != idn_success)
goto ret;
}
r = label_lencheck_ace(ctx, rt_label);
if (r != idn_success)
goto ret;
rt_name = labellist_getname(rt_label);
if (idn_ucs4_strcasecmp(rt_name, original_name) != 0) {
TRACE(("res rtcheck(): round trip failed, org =\"%s\", rt=\"%s\"\n",
idn__debug_ucs4xstring(original_name, 50),
idn__debug_ucs4xstring(rt_name, 50)));
r = idn_invalid_encoding;
goto ret;
}
r = idn_success;
ret:
if (r != idn_nomemory && r != idn_success)
r = idn_invalid_encoding;
TRACE(("res rtcheck(): %s\n", idn_result_tostring(r)));
if (rt_label != NULL)
labellist_destroy(rt_label);
return (r);
}
