|
1
|
|
|
#!/usr/bin/env python |
|
2
|
|
|
# -*- coding: utf-8 -*- |
|
3
|
|
|
|
|
4
|
|
|
# use data structures |
|
5
|
|
|
from __future__ import unicode_literals |
|
6
|
|
|
from .ds import Document, Sentence, Dependencies |
|
7
|
|
|
from .utils import post_json |
|
8
|
|
|
import json |
|
9
|
|
|
|
|
10
|
|
|
|
|
11
|
|
|
class Processor(object): |
|
12
|
|
|
|
|
13
|
|
|
def __init__(self, address): |
|
14
|
|
|
self.service = "{}/annotate".format(address) |
|
15
|
|
|
|
|
16
|
|
|
def _message_to_json_dict(self, msg): |
|
17
|
|
|
return post_json(self.service, msg.to_JSON()) |
|
18
|
|
|
|
|
19
|
|
|
def _annotate_message(self, msg): |
|
20
|
|
|
annotated_text = post_json(self.service, msg.to_JSON()) |
|
21
|
|
|
return Document.load_from_JSON(annotated_text) |
|
22
|
|
|
|
|
23
|
|
|
def annotate(self, text): |
|
24
|
|
|
try: |
|
25
|
|
|
# load json and build Sentences and Document |
|
26
|
|
|
msg = Message(text) |
|
27
|
|
|
return self._annotate_message(msg) |
|
28
|
|
|
|
|
29
|
|
|
except Exception as e: |
|
30
|
|
|
#print(e) |
|
31
|
|
|
return None |
|
32
|
|
|
|
|
33
|
|
|
def annotate_from_sentences(self, sentences): |
|
34
|
|
|
""" |
|
35
|
|
|
Annotate text that has already been segmented into sentences. |
|
36
|
|
|
""" |
|
37
|
|
|
try: |
|
38
|
|
|
# load json from str interable and build Sentences and Document |
|
39
|
|
|
msg = SentencesMessage(sentences) |
|
40
|
|
|
return self._annotate_message(msg) |
|
41
|
|
|
|
|
42
|
|
|
except Exception as e: |
|
43
|
|
|
#print(e) |
|
44
|
|
|
return None |
|
45
|
|
|
|
|
46
|
|
|
class FastNLPProcessor(Processor): |
|
47
|
|
|
|
|
48
|
|
|
def __init__(self, address): |
|
49
|
|
|
self.service = "{}/fastnlp/annotate".format(address) |
|
50
|
|
|
|
|
51
|
|
|
def annotate(self, text): |
|
52
|
|
|
return super(FastNLPProcessor, self).annotate(text) |
|
53
|
|
|
|
|
54
|
|
|
|
|
55
|
|
|
class BioNLPProcessor(Processor): |
|
56
|
|
|
|
|
57
|
|
|
def __init__(self, address): |
|
58
|
|
|
self.service = "{}/bionlp/annotate".format(address) |
|
59
|
|
|
|
|
60
|
|
|
def annotate(self, text): |
|
61
|
|
|
return super(BioNLPProcessor, self).annotate(text) |
|
62
|
|
|
|
|
63
|
|
|
|
|
64
|
|
|
class Message(object): |
|
65
|
|
|
def __init__(self, text): |
|
66
|
|
|
self.text = text |
|
67
|
|
|
|
|
68
|
|
|
def to_JSON_dict(self): |
|
69
|
|
|
jdict = dict() |
|
70
|
|
|
jdict["text"] = self.text |
|
71
|
|
|
return jdict |
|
72
|
|
|
|
|
73
|
|
|
def to_JSON(self): |
|
74
|
|
|
return json.dumps(self.to_JSON_dict(), sort_keys=True, indent=4) |
|
75
|
|
|
|
|
76
|
|
|
|
|
77
|
|
|
class SentencesMessage(object): |
|
78
|
|
|
def __init__(self, sentences): |
|
79
|
|
|
self.sentences = sentences |
|
80
|
|
|
|
|
81
|
|
|
def to_JSON_dict(self): |
|
82
|
|
|
jdict = dict() |
|
83
|
|
|
jdict["sentences"] = self.sentences |
|
84
|
|
|
return jdict |
|
85
|
|
|
|
|
86
|
|
|
def to_JSON(self): |
|
87
|
|
|
return json.dumps(self.to_JSON_dict(), sort_keys=True, indent=4) |
|
88
|
|
|
|
myedibleenso /
py-processors
