/* method: "get_text_raw(s)\nGet the unevaluated Text from Text objects." */
static utf8_string Shape_get_text_raw(const Object& self){
auto text = dynamic_cast<const ObjText*>(&self);
if (!text){
throw ValueError(space_sep(self.GetType(), "does not support text."));
}
return text->GetRawString();
}
/* method: "get_text_lines()->(s,...)\n
Returns the evaluated text from a Text-object split into lines. Takes
the bounding rectangle in consideration, so that the lines are split in
the same way as they appear in Faint." */
static text_lines_t Shape_get_text_lines(const Object& self){
auto text = dynamic_cast<const ObjText*>(&self);
if (!text){
throw TypeError(space_sep(self.GetType(), "does not support text."));
}
NullExpressionContext ctx;
return split_evaluated(ctx, *text);
}
static void faintwindow_init(faintWindowObject&){
// Prevent instantiation from Python, since the AppContext can't be
// provided from there.
throw TypeError(space_sep("FaintWindow can not be instantiated.",
"Use the 'window'-object instead."));
}
utf8_string Name() const override{
return space_sep("Delete", m_object->GetType(), "Point");
}
bool ParallelCorpus::ReadPartialAlignmentFile(const string& filename) {
std::ifstream in(filename.c_str());
if (!in.good()) {
return false;
}
typedef boost::tokenizer<boost::char_separator<char> > tokenizer;
boost::char_separator<char> tab_sep("\t", "", boost::keep_empty_tokens);
boost::char_separator<char> space_sep(" ");
string line;
set<pair<int, int> > alignment;
PartialAlignment partial_alignment;
int doc_index = 0;
InitPartialAlignment(doc_pairs_.at(doc_index), &partial_alignment);
while (getline(in, line)) {
vector<std::string> tokens;
tokenizer line_tokenizer(line, tab_sep);
for (tokenizer::iterator it = line_tokenizer.begin();
it != line_tokenizer.end(); ++it) {
string token = *it;
tokens.push_back(token);
}
if (tokens.size() == 3) {
// (source_index)\t(0 or more positive target indices)\t(0 or more
// negative target indices)
int source = atoi(tokens[0].c_str());
vector<int> pos_targets, neg_targets;
tokenizer index_tokenizer(tokens[1], space_sep);
for (tokenizer::iterator it = index_tokenizer.begin();
it != index_tokenizer.end(); ++it) {
pos_targets.push_back(atoi(it->c_str()));
}
index_tokenizer.assign(tokens[2], space_sep);
for (tokenizer::iterator it = index_tokenizer.begin();
it != index_tokenizer.end(); ++it) {
neg_targets.push_back(atoi(it->c_str()));
}
for (int t = 0; t < neg_targets.size(); ++t) {
int target = neg_targets.at(t);
partial_alignment[source][target] = false;
}
// If we encounter at least one positive alignment for the source
// sentence, we consider all target sentences as false. Otherwise, we only
// consider the seen target sentences as false.
// TODO
/*
if (pos_targets.size() > 0) {
for (int t = 0; t < doc_pairs_.at(doc_index).second.size(); ++t) {
partial_alignment[source][t] = false;
}
} */
for (int t = 0; t < pos_targets.size(); ++t) {
int target = pos_targets.at(t);
partial_alignment[source][target] = true;
alignment.insert(std::make_pair(source, target));
}
} else {
// An empty line indicates a document boundary
alignments_.push_back(alignment);
alignment.clear();
partial_alignments_.push_back(partial_alignment);
doc_index++;
if (doc_index < doc_pairs_.size()) {
InitPartialAlignment(doc_pairs_.at(doc_index), &partial_alignment);
}
}
}
in.close();
if (alignments_.size() != doc_pairs_.size()) {
return false;
}
return true;
}
static void as_init(AppContext&){
// Prevent instantiation from Python, since the expected context
// can't be provided from there.
throw TypeError(space_sep(ACTIVE_SETTINGS_NAME, "can not be instantiated."));
}
