| 6 | using namespace std; |
| 7 | |
| 8 | PTModelObject::PTModelObject(size_t tw, size_t minCnt, size_t minDf, size_t rmTop, |
| 9 | size_t k, std::optional<size_t> p, PyObject* alpha, float eta, |
| 10 | PyObject* seed, PyObject* corpus, PyObject* transform) |
| 11 | { |
| 12 | tomoto::PTArgs margs; |
| 13 | margs.k = k; |
| 14 | if (alpha) |
| 15 | { |
| 16 | margs.alpha = broadcastObj<tomoto::Float>(alpha, margs.k, |
| 17 | [&]() { return "`alpha` must be an instance of `float` or `List[float]` with length `k` (given " + py::repr(alpha) + ")"; } |
| 18 | ); |
| 19 | } |
| 20 | margs.eta = eta; |
| 21 | margs.p = p.value_or(0); |
| 22 | if (seed && seed != Py_None && !py::toCpp<size_t>(seed, margs.seed)) |
| 23 | { |
| 24 | throw py::ValueError{ "`seed` must be an integer or None." }; |
| 25 | } |
| 26 | |
| 27 | if (margs.p == 0) margs.p = margs.k * 10; |
| 28 | |
| 29 | inst = tomoto::IPTModel::create((tomoto::TermWeight)tw, margs); |
| 30 | if (!inst) throw py::ValueError{ "unknown `tw` value" }; |
| 31 | isPrepared = false; |
| 32 | seedGiven = !!seed; |
| 33 | minWordCnt = minCnt; |
| 34 | minWordDf = minDf; |
| 35 | removeTopWord = rmTop; |
| 36 | |
| 37 | insertCorpus(corpus, transform); |
| 38 | } |
| 39 | |
| 40 | size_t DocumentObject::getPseudoDocId() const |
| 41 | { |