|
1
|
|
|
#!/usr/bin/env python |
|
2
|
|
|
# -*- coding: utf-8 -*- |
|
3
|
|
|
from __future__ import unicode_literals |
|
4
|
|
|
from pkg_resources import resource_filename |
|
5
|
|
|
from .processors import * |
|
6
|
|
|
from .sentiment import SentimentAnalysisAPI |
|
7
|
|
|
from .odin import OdinAPI |
|
8
|
|
|
import os |
|
9
|
|
|
import shlex |
|
10
|
|
|
import os |
|
11
|
|
|
import subprocess as sp |
|
12
|
|
|
import requests |
|
13
|
|
|
import time |
|
14
|
|
|
import sys |
|
15
|
|
|
import logging |
|
16
|
|
|
|
|
17
|
|
|
|
|
18
|
|
|
class ProcessorsAPI(object): |
|
19
|
|
|
|
|
20
|
|
|
PROC_VAR = 'PROCESSORS_SERVER' |
|
21
|
|
|
|
|
22
|
|
|
def __init__(self, port, hostname="127.0.0.1", timeout=120, jvm_mem="-Xmx3G", jar_path=None, keep_alive=False, log_file=None): |
|
23
|
|
|
|
|
24
|
|
|
self.hostname = hostname |
|
25
|
|
|
self.port = port |
|
26
|
|
|
self.make_address(hostname, port) |
|
27
|
|
|
self._start_command = "java {} -cp {} NLPServer {}" # mem, jar path, port |
|
28
|
|
|
self.timeout = timeout |
|
29
|
|
|
self.jvm_mem = jvm_mem |
|
30
|
|
|
# whether or not to stop the server when the object is destroyed |
|
31
|
|
|
self.keep_alive = keep_alive |
|
32
|
|
|
# how long to wait between requests |
|
33
|
|
|
self.wait_time = 2 |
|
34
|
|
|
# processors |
|
35
|
|
|
self.default = Processor(self.address) |
|
36
|
|
|
self.fastnlp = FastNLPProcessor(self.address) |
|
37
|
|
|
self.bionlp = BioNLPProcessor(self.address) |
|
38
|
|
|
# sentiment |
|
39
|
|
|
self.sentiment = SentimentAnalysisAPI(self.address) |
|
40
|
|
|
# odin |
|
41
|
|
|
self.odin = OdinAPI(self.address) |
|
42
|
|
|
# use the os module's devnull for compatibility with python 2.7 |
|
43
|
|
|
#self.DEVNULL = open(os.devnull, 'wb') |
|
44
|
|
|
self.logger = logging.getLogger(__name__) |
|
45
|
|
|
self.log_file = self.prepare_log_file(log_file) |
|
46
|
|
|
|
|
47
|
|
|
# resolve jar path |
|
48
|
|
|
self.resolve_jar_path(jar_path) |
|
49
|
|
|
# attempt to establish connection with server |
|
50
|
|
|
self.establish_connection() |
|
51
|
|
|
|
|
52
|
|
|
def prepare_log_file(self, lf): |
|
53
|
|
|
""" |
|
54
|
|
|
Configure logger and return file path for logging |
|
55
|
|
|
""" |
|
56
|
|
|
# log_file |
|
57
|
|
|
log_file = os.path.expanduser(os.path.join("~", ".py-processors.log")) if not lf else os.path.expanduser(lf) |
|
58
|
|
|
# configure logger |
|
59
|
|
|
self.logger.setLevel(logging.DEBUG) |
|
60
|
|
|
# create console handler and set level to info |
|
61
|
|
|
handler = logging.StreamHandler() |
|
62
|
|
|
handler.setLevel(logging.INFO) |
|
63
|
|
|
formatter = logging.Formatter("%(levelname)s - %(message)s") |
|
64
|
|
|
handler.setFormatter(formatter) |
|
65
|
|
|
self.logger.addHandler(handler) |
|
66
|
|
|
# create debug file handler and set level to debug |
|
67
|
|
|
handler = logging.FileHandler(log_file, "w") |
|
68
|
|
|
handler.setLevel(logging.DEBUG) |
|
69
|
|
|
formatter = logging.Formatter("%(levelname)s - %(message)s") |
|
70
|
|
|
handler.setFormatter(formatter) |
|
71
|
|
|
self.logger.addHandler(handler) |
|
72
|
|
|
return log_file |
|
73
|
|
|
|
|
74
|
|
|
def annotate(self, text): |
|
75
|
|
|
""" |
|
76
|
|
|
Uses default processor (CoreNLP) to annotate text. Included for backwards compatibility. |
|
77
|
|
|
""" |
|
78
|
|
|
return self.default.annotate(text) |
|
79
|
|
|
|
|
80
|
|
|
def annotate_from_sentences(self, sentences): |
|
81
|
|
|
""" |
|
82
|
|
|
Uses default processor (CoreNLP) to annotate a list of segmented sentences. |
|
83
|
|
|
""" |
|
84
|
|
|
return self.default.annotate_from_sentences(sentences) |
|
85
|
|
|
|
|
86
|
|
|
def establish_connection(self): |
|
87
|
|
|
""" |
|
88
|
|
|
Attempt to connect to a server (assumes server is running) |
|
89
|
|
|
""" |
|
90
|
|
|
if self.annotate("Blah"): |
|
91
|
|
|
print("Connection with server established!") |
|
92
|
|
|
else: |
|
93
|
|
|
try: |
|
94
|
|
|
# Attempt to start the server |
|
95
|
|
|
self._start_server() |
|
96
|
|
|
except Exception as e: |
|
97
|
|
|
if not os.path.exists(self.jar_path): |
|
98
|
|
|
print("\nprocessors-server.jar not found at {}.".format(self.jar_path)) |
|
99
|
|
|
print("Unable to start server. Please start the server manually with .start_server(\"path/to/processors-server.jar\")") |
|
100
|
|
|
print("\n{}".format(e)) |
|
101
|
|
|
|
|
102
|
|
|
def resolve_jar_path(self, jar_path): |
|
103
|
|
|
""" |
|
104
|
|
|
Attempts to preferentially set value of self.jar_path |
|
105
|
|
|
""" |
|
106
|
|
|
# Preference 1: if a .jar is given, check to see if the path is valid |
|
107
|
|
|
if jar_path: |
|
108
|
|
|
print("Using provided path") |
|
109
|
|
|
jp = os.path.expanduser(jar_path) |
|
110
|
|
|
# check if path is valid |
|
111
|
|
|
if os.path.exists(jp): |
|
112
|
|
|
self.jar_path = jp |
|
113
|
|
|
else: |
|
114
|
|
|
# Preference 2: if a PROCESSORS_SERVER environment variable is defined, check its validity |
|
115
|
|
|
if ProcessorsAPI.PROC_VAR in os.environ: |
|
116
|
|
|
print("Using path given via $PROCESSORS_SERVER") |
|
117
|
|
|
jp = os.path.expanduser(os.environ[ProcessorsAPI.PROC_VAR]) |
|
118
|
|
|
# check if path is valid |
|
119
|
|
|
if os.path.exists(jp): |
|
120
|
|
|
self.jar_path = jp |
|
121
|
|
|
else: |
|
122
|
|
|
print("WARNING: {0} path is invalid. \nPlease verify this entry in your environment:\n\texport {0}=/path/to/processors-server.jar".format(ProcessorsAPI.PROC_VAR)) |
|
123
|
|
|
# Preference 3: attempt to use the processors-sever.jar downloaded when this package was installed |
|
124
|
|
|
else: |
|
125
|
|
|
print("Using default") |
|
126
|
|
|
self.jar_path = resource_filename(__name__, "processors-server.jar") |
|
127
|
|
|
|
|
128
|
|
|
def start_server(self, jar_path=None, timeout=120): |
|
129
|
|
|
""" |
|
130
|
|
|
Starts processors-sever.jar |
|
131
|
|
|
""" |
|
132
|
|
|
self.timeout = int(float(timeout)/2) |
|
133
|
|
|
if jar_path: |
|
134
|
|
|
self.jar_path = jar_path |
|
135
|
|
|
self._start_server() |
|
136
|
|
|
|
|
137
|
|
|
def stop_server(self, port=None): |
|
138
|
|
|
""" |
|
139
|
|
|
Sends a poison pill to the server and waits for shutdown response |
|
140
|
|
|
""" |
|
141
|
|
|
port = port or self.port |
|
142
|
|
|
address = "http://{}:{}".format(self.hostname, port) |
|
143
|
|
|
shutdown_address = "{}/shutdown".format(address) |
|
144
|
|
|
# attempt shutdown |
|
145
|
|
|
try: |
|
146
|
|
|
response = requests.post(shutdown_address) |
|
147
|
|
|
if response: |
|
148
|
|
|
print(response.content.decode("utf-8")) |
|
149
|
|
|
return True |
|
150
|
|
|
# will fail if the server is already down |
|
151
|
|
|
except Exception as e: |
|
152
|
|
|
pass |
|
153
|
|
|
return False |
|
154
|
|
|
|
|
155
|
|
|
def _start_server(self, port=None): |
|
156
|
|
|
""" |
|
157
|
|
|
"Private" method called by start_server() |
|
158
|
|
|
""" |
|
159
|
|
|
if port: |
|
160
|
|
|
self.port = port |
|
161
|
|
|
# build the command |
|
162
|
|
|
cmd = self._start_command.format(self.jvm_mem, self.jar_path, self.port) |
|
163
|
|
|
self._process = sp.Popen(shlex.split(cmd), |
|
164
|
|
|
shell=False, |
|
165
|
|
|
stderr=open(self.log_file, 'wb'), |
|
166
|
|
|
stdout=open(self.log_file, 'wb'), |
|
167
|
|
|
universal_newlines=True) |
|
168
|
|
|
|
|
169
|
|
|
self.logger.info("Starting processors-server ({}) ...".format(cmd)) |
|
170
|
|
|
print("\nWaiting for server...") |
|
171
|
|
|
|
|
172
|
|
|
progressbar_length = int(self.timeout/self.wait_time) |
|
173
|
|
|
for i in range(progressbar_length): |
|
174
|
|
|
try: |
|
175
|
|
|
success = self.annotate("blah") |
|
176
|
|
|
if success: |
|
177
|
|
|
print("\n\nConnection with processors-server established ({})".format(self.address)) |
|
178
|
|
|
return True |
|
179
|
|
|
sys.stdout.write("\r[{:{}}]".format('='*i, progressbar_length)) |
|
180
|
|
|
time.sleep(self.wait_time) |
|
181
|
|
|
except Exception as e: |
|
182
|
|
|
raise(e) |
|
183
|
|
|
|
|
184
|
|
|
# if the server still hasn't started, raise an Exception |
|
185
|
|
|
raise Exception("Couldn't connect to processors-server. Is the port in use?") |
|
186
|
|
|
|
|
187
|
|
|
def make_address(self, hostname, port): |
|
188
|
|
|
# update hostname |
|
189
|
|
|
self.hostname = hostname |
|
190
|
|
|
# update port |
|
191
|
|
|
self.port = port |
|
192
|
|
|
# update address |
|
193
|
|
|
self.address = "http://{}:{}".format(self.hostname, self.port) |
|
194
|
|
|
|
|
195
|
|
|
def _get_path(self, p): |
|
196
|
|
|
""" |
|
197
|
|
|
Expand a user-specified path. Supports "~" shortcut. |
|
198
|
|
|
""" |
|
199
|
|
|
return os.path.abspath(os.path.normpath(os.path.expanduser(p))) |
|
200
|
|
|
|
|
201
|
|
|
def __del__(self): |
|
202
|
|
|
""" |
|
203
|
|
|
Stop server unless otherwise specified |
|
204
|
|
|
""" |
|
205
|
|
|
if not self.keep_alive: |
|
206
|
|
|
try: |
|
207
|
|
|
self.stop_server() |
|
208
|
|
|
# close our file object |
|
209
|
|
|
#self.DEVNULL.close() |
|
210
|
|
|
print("Successfully shut down processors-server!") |
|
211
|
|
|
except Exception as e: |
|
212
|
|
|
self.logger.debug(e) |
|
213
|
|
|
print("Couldn't kill processors-server. Was server started externally?") |
|
214
|
|
|
|
myedibleenso /
py-processors
