diff options
Diffstat (limited to 'python/src/sa/data_array.pxi')
| -rw-r--r-- | python/src/sa/data_array.pxi | 25 |
1 files changed, 13 insertions, 12 deletions
diff --git a/python/src/sa/data_array.pxi b/python/src/sa/data_array.pxi index 9f62dc0a..2a8ea921 100644 --- a/python/src/sa/data_array.pxi +++ b/python/src/sa/data_array.pxi @@ -7,11 +7,11 @@ from libc.stdlib cimport malloc, realloc, free from libc.string cimport memset, strcpy cdef class DataArray: - cdef word2id - cdef id2word - cdef IntList data - cdef IntList sent_id - cdef IntList sent_index + cdef public word2id + cdef public id2word + cdef public IntList data + cdef public IntList sent_id + cdef public IntList sent_index cdef bint use_sent_id def __cinit__(self, from_binary=None, from_text=None, side=None, bint use_sent_id=False): @@ -32,10 +32,10 @@ cdef class DataArray: def __len__(self): return len(self.data) - def getSentId(self, i): + def get_sentence_id(self, i): return self.sent_id.arr[i] - def getSent(self, i): + def get_sentence(self, i): cdef int j, start, stop sent = [] start = self.sent_index.arr[i] @@ -44,17 +44,18 @@ cdef class DataArray: sent.append(self.id2word[self.data.arr[i]]) return sent - def getSentPos(self, loc): - return loc - self.sent_index.arr[self.sent_id.arr[loc]] - def get_id(self, word): if not word in self.word2id: self.word2id[word] = len(self.id2word) self.id2word.append(word) return self.word2id[word] - def get_word(self, id): - return self.id2word[id] + def __getitem__(self, loc): + return self.id2word[self.data.arr[loc]] + + def get_sentence_bounds(self, loc): + cdef int sid = self.sent_id.arr[loc] + return (self.sent_index.arr[sid], self.sent_index.arr[sid+1]) def write_text(self, char* filename): with open(filename, "w") as f: |