| 76 | } |
| 77 | |
| 78 | std::optional<size_t> GDMRModelObject::addDoc(PyObject* words, PyObject* numericMetadata, const std::string& metadata, PyObject* multiMetadata, bool ignoreEmptyWords) |
| 79 | { |
| 80 | if (isPrepared) throw py::RuntimeError{ "cannot add_doc() after train()" }; |
| 81 | auto* inst = getInst<tomoto::IGDMRModel>(); |
| 82 | if (PyUnicode_Check(words)) |
| 83 | { |
| 84 | if (PyErr_WarnEx(PyExc_RuntimeWarning, "`words` should be an iterable of str.", 1)) throw py::ExcPropagation{}; |
| 85 | } |
| 86 | |
| 87 | tomoto::RawDoc raw = buildRawDoc(words); |
| 88 | raw.misc["metadata"] = metadata; |
| 89 | if (multiMetadata) |
| 90 | { |
| 91 | vector<string> multiMetadataValue; |
| 92 | if (!py::toCpp<vector<string>>(multiMetadata, multiMetadataValue)) |
| 93 | { |
| 94 | throw py::ValueError{ |
| 95 | "`multi_metadata` must be an instance of `List[str]` (but given " + py::repr(multiMetadata) + ")" |
| 96 | }; |
| 97 | } |
| 98 | raw.misc["multi_metadata"] = move(multiMetadataValue); |
| 99 | } |
| 100 | vector<tomoto::Float> nmd; |
| 101 | if (!py::toCpp(numericMetadata, nmd)) |
| 102 | { |
| 103 | throw py::ValueError{ |
| 104 | "`numeric_metadata` must be an iterable of float (but given " + py::repr(numericMetadata) + ")" |
| 105 | }; |
| 106 | } |
| 107 | for (auto x : nmd) |
| 108 | { |
| 109 | if (!isfinite(x)) throw py::ValueError{ "`numeric_metadata` has non-finite value (" + py::reprFromCpp(nmd) + ")." }; |
| 110 | } |
| 111 | raw.misc["numeric_metadata"] = move(nmd); |
| 112 | try |
| 113 | { |
| 114 | auto ret = inst->addDoc(raw); |
| 115 | return py::buildPyValue(ret); |
| 116 | } |
| 117 | catch (const tomoto::exc::EmptyWordArgument&) |
| 118 | { |
| 119 | if (ignoreEmptyWords) |
| 120 | { |
| 121 | return std::nullopt; |
| 122 | } |
| 123 | else |
| 124 | { |
| 125 | throw; |
| 126 | } |
| 127 | } |
| 128 | } |
| 129 | |
| 130 | py::UniqueCObj<DocumentObject> GDMRModelObject::makeDoc(PyObject* words, PyObject* numericMetadata, const std::string& metadata, PyObject* multiMetadata) |
| 131 | { |
nothing calls this directly
no test coverage detected