diff options
| author | Victor Chahuneau <vchahune@cs.cmu.edu> | 2012-09-06 16:44:53 +0100 |
|---|---|---|
| committer | Victor Chahuneau <vchahune@cs.cmu.edu> | 2012-09-06 16:44:53 +0100 |
| commit | c1b77250f656c4cff9f0e532d6b6644cb0dc993c (patch) | |
| tree | 93711d64a442124ca441e61d64b75b861c517120 /python/src/sa/data_array.pxi | |
| parent | fe9577d0ec51b7a5136d23885742780ca7f9ba8c (diff) | |
[cdec.sa] Fix API to make everyone happy
Diffstat (limited to 'python/src/sa/data_array.pxi')
| -rw-r--r-- | python/src/sa/data_array.pxi | 17 |
1 files changed, 9 insertions, 8 deletions
diff --git a/python/src/sa/data_array.pxi b/python/src/sa/data_array.pxi index 371708c5..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 public word2id + cdef public id2word cdef public IntList data cdef public IntList sent_id - cdef IntList sent_index + 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): @@ -44,17 +44,18 @@ cdef class DataArray: sent.append(self.id2word[self.data.arr[i]]) return sent - def get_sentence_position(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: |