| 28 | self.pos_len = pos_len |
| 29 | |
| 30 | def __call__(self, data): |
| 31 | text = data['label'] |
| 32 | if self.ch: |
| 33 | text, text_node_index, text_node_num = self.encodech(text) |
| 34 | if text is None: |
| 35 | return None |
| 36 | if len(text) > self.max_text_len: |
| 37 | return None |
| 38 | data['length'] = np.array(len(text)) |
| 39 | # text.insert(0, 0) |
| 40 | if self.pos_len: |
| 41 | text_pos_node = [i_ for i_ in range(len(text), -1, -1) |
| 42 | ] + [100] * (self.max_text_len - len(text)) |
| 43 | else: |
| 44 | text_pos_node = [1] * (len(text) + 1) + [0] * ( |
| 45 | self.max_text_len - len(text)) |
| 46 | |
| 47 | text.append(0) |
| 48 | text + [0] * (self.max_text_len - len(text)) |
| 49 | |
| 50 | text = text + [self.ignore_index |
| 51 | ] * (self.max_text_len + 1 - len(text)) |
| 52 | |
| 53 | data['label'] = np.array(text) |
| 54 | data['label_node'] = np.array(text_node_num + text_pos_node) |
| 55 | data['label_index'] = np.array(text_node_index) |
| 56 | # data['label_ctc'] = np.array(ctc_text) |
| 57 | return data |
| 58 | else: |
| 59 | text, text_char_node, ch_order = self.encode(text) |
| 60 | |
| 61 | if text is None: |
| 62 | return None |
| 63 | if len(text) > self.max_text_len: |
| 64 | return None |
| 65 | data['length'] = np.array(len(text)) |
| 66 | # text.insert(0, 0) |
| 67 | if self.pos_len: |
| 68 | text_pos_node = [i_ for i_ in range(len(text), -1, -1) |
| 69 | ] + [100] * (self.max_text_len - len(text)) |
| 70 | else: |
| 71 | text_pos_node = [1] * (len(text) + 1) + [0] * ( |
| 72 | self.max_text_len - len(text)) |
| 73 | |
| 74 | text.append(0) |
| 75 | |
| 76 | text = text + [self.ignore_index |
| 77 | ] * (self.max_text_len + 1 - len(text)) |
| 78 | data['label'] = np.array(text) |
| 79 | data['label_node'] = np.array(text_char_node + text_pos_node) |
| 80 | data['label_order'] = np.array(ch_order) |
| 81 | |
| 82 | return data |
| 83 | |
| 84 | def add_special_char(self, dict_character): |
| 85 | if self.use_sos: |