Reading *alloc_reading(const Reading& o) {
Reading *r = pool_get(pool_readings);
if (r == 0) {
r = new Reading(o);
}
else {
r->mapped = o.mapped;
r->deleted = o.deleted;
r->noprint = o.noprint;
r->matched_target = false;
r->matched_tests = false;
r->baseform = o.baseform;
r->hash = o.hash;
r->hash_plain = o.hash_plain;
r->number = o.number + 100;
r->tags_bloom = o.tags_bloom;
r->tags_plain_bloom = o.tags_plain_bloom;
r->tags_textual_bloom = o.tags_textual_bloom;
r->mapping = o.mapping;
r->parent = o.parent;
r->next = o.next;
r->hit_by = o.hit_by;
r->tags_list = o.tags_list;
r->tags = o.tags;
r->tags_plain = o.tags_plain;
r->tags_textual = o.tags_textual;
r->tags_numerical = o.tags_numerical;
if (r->next) {
r->next = alloc_reading(*r->next);
}
}
return r;
}
Reading *Reading::allocateReading(Cohort *p) {
return alloc_reading(p);
}
Reading *Reading::allocateReading(const Reading& r) {
return alloc_reading(r);
}
std::vector<Cohort*> MweSplitApplicator::splitMwe(Cohort* cohort) {
constexpr UChar rtrimblank[] = { ' ', '\n', '\r', '\t', 0 };
constexpr UChar textprefix[] = { ':', 0 };
std::vector<Cohort*> cos;
size_t n_wftags = 0;
size_t n_goodreadings = 0;
for (auto rter1 : cohort->readings) {
if (maybeWfTag(rter1) != NULL) {
++n_wftags;
}
++n_goodreadings;
}
if (n_wftags < n_goodreadings) {
if (n_wftags > 0) {
u_fprintf(ux_stderr, "WARNING: Line %u: Some but not all main-readings of %S had wordform-tags (not completely mwe-disambiguated?), not splitting.\n", cohort->line_number, cohort->wordform->tag.c_str());
// We also don't split if wordform-tags were only on sub-readings, but should we warn on such faulty input?
}
cos.push_back(cohort);
return cos;
}
for (auto r : cohort->readings) {
size_t pos = std::numeric_limits<size_t>::max();
Reading* prev = NULL; // prev == NULL || prev->next == rNew (or a ->next of rNew)
for (auto sub = r; sub; sub = sub->next) {
const Tag* wfTag = maybeWfTag(sub);
if (wfTag == NULL) {
prev = prev->next;
}
else {
++pos;
Cohort* c;
while (cos.size() < pos + 1) {
c = alloc_cohort(cohort->parent);
c->global_number = gWindow->cohort_counter++;
cohort->parent->appendCohort(c);
cos.push_back(c);
}
c = cos[pos];
const size_t wfEnd = wfTag->tag.size() - 3; // index before the final '>"'
const size_t i = 1 + wfTag->tag.find_last_not_of(rtrimblank, wfEnd);
const UString& wf = wfTag->tag.substr(0, i) + wfTag->tag.substr(wfEnd + 1);
if (c->wordform != 0 && wf != c->wordform->tag) {
u_fprintf(ux_stderr, "WARNING: Line %u: Ambiguous wordform-tags for same cohort, '%S' vs '%S', not splitting.\n", numLines, wf.c_str(), c->wordform->tag.c_str());
cos.clear();
cos.push_back(cohort);
return cos;
}
c->wordform = addTag(wf);
if (i < wfEnd + 1) {
c->text = textprefix + wfTag->tag.substr(i, wfEnd + 1 - i);
}
Reading* rNew = alloc_reading(*sub);
for (size_t i = 0; i < rNew->tags_list.size(); ++i) {
auto& tter = rNew->tags_list[i];
if (tter == wfTag->hash || tter == rNew->parent->wordform->hash) {
rNew->tags_list.erase(rNew->tags_list.begin() + i);
rNew->tags.erase(tter);
}
}
cos[pos]->appendReading(rNew);
rNew->parent = cos[pos];
if (prev != NULL) {
free_reading(prev->next);
prev->next = 0;
}
prev = rNew;
}
}
}
if (cos.size() == 0) {
u_fprintf(ux_stderr, "WARNING: Line %u: Tried splitting %S, but got no new cohorts; shouldn't happen.", numLines, cohort->wordform->tag.c_str());
cos.push_back(cohort);
}
// The last word forms are the top readings:
cos[0]->text = cohort->text;
std::reverse(cos.begin(), cos.end());
return cos;
}
