Porting to Python 3

stanford_corenlp_pywrapper/__init__.py

@@ -1 +1 @@
-from sockwrap import *
+from .sockwrap import *

stanford_corenlp_pywrapper/sockwrap.py

@@ -74,7 +74,7 @@ def command(mode=None, configfile=None, configdict=None, comm_mode=None,
     else: assert False, "need comm_mode to be SOCKET or PIPE but got " + repr(comm_mode)
 
 
-    cmd = """exec {java_command} {java_options} -cp '{classpath}' 
+    cmd = """exec {java_command} {java_options} -cp '{classpath}'
     corenlp.SocketServer {comm_info} {more_config}"""
     return cmd.format(**d).replace("\n", " ")
 
@@ -85,7 +85,7 @@ class SubprocessCrashed(Exception):
 
 class CoreNLP:
 
-    def __init__(self, mode=None, 
+    def __init__(self, mode=None,
             configfile=None, configdict=None,
             corenlp_jars=(
                 "/home/sw/corenlp/stanford-corenlp-full-2015-04-20/*",
@@ -168,7 +168,7 @@ def start_server(self):
         if self.comm_mode=='PIPE':
             if not os.path.exists(self.outpipe):
                 os.mkfifo(self.outpipe)
-        
+
         cmd = command(**self.__dict__)
         LOG.info("Starting java subprocess, and waiting for signal it's ready, with command: %s" % cmd)
         self.proc = subprocess.Popen(cmd, shell=True, stdin=subprocess.PIPE)
@@ -178,7 +178,7 @@ def start_server(self):
             sock = self.get_socket(num_retries=100, retry_interval=STARTUP_BUSY_WAIT_INTERVAL_SEC)
             sock.close()
         elif self.comm_mode=='PIPE':
-            self.outpipe_fp = open(self.outpipe, 'r')
+            self.outpipe_fp = open(self.outpipe, 'rb')
 
         while True:
             # This loop is for if you have timeouts for the socket connection
@@ -191,7 +191,7 @@ def start_server(self):
                 assert ret == "PONG", "Bad return data on startup ping: " + ret
                 LOG.info("Successful ping. The server has started.")
                 break
-            except socket.error, e:
+            except socket.error as e:
                 LOG.info("Waiting for startup: ping got exception: %s %s" % (type(e), e))
                 LOG.info("pausing before retry")
                 time.sleep(STARTUP_BUSY_WAIT_INTERVAL_SEC)
@@ -251,7 +251,7 @@ def send_command_and_parse_result(self, cmd, timeout, raw=False):
                 LOG.warning("Bad JSON length %d, starts with: %s" % (len(data), repr(data[:1000])))
                 return None
             return decoded
-        except socket.timeout, e:
+        except socket.timeout as e:
             LOG.info("Socket timeout happened, returning None: %s %s" % (type(e), e))
             return None
             # This is tricky. maybe the process is running smoothly but just
@@ -266,14 +266,13 @@ def send_command_and_get_string_result(self, cmd, timeout):
             sock.sendall(cmd + "\n")
             size_info_str = sock.recv(8)
         elif self.comm_mode == 'PIPE':
-            self.proc.stdin.write(cmd + "\n")
+            self.proc.stdin.write((cmd+"\n").encode('utf-8'))
             self.proc.stdin.flush()
             size_info_str = self.outpipe_fp.read(8)
 
         # java "long" is 8 bytes, which python struct calls "long long".
         # java default byte ordering is big-endian.
-        size_info = struct.unpack('>Q', size_info_str)[0]
-        # print "size expected", size_info
+        size_info = struct.unpack(('>Q').encode('utf-8'), size_info_str)[0]
 
         chunks = []
         curlen = lambda: sum(len(x) for x in chunks)
@@ -283,7 +282,7 @@ def send_command_and_get_string_result(self, cmd, timeout):
                 data = sock.recv(remaining_size)
             elif self.comm_mode == 'PIPE':
                 data = self.outpipe_fp.read(remaining_size)
-            chunks.append(data)
+            chunks.append(data.decode("utf-8"))
             if curlen() >= size_info: break
             if len(chunks) > 1000:
                 LOG.warning("Incomplete value from server")
@@ -329,7 +328,7 @@ def test_paths():
 def assert_no_java(msg=""):
     ps_output = os.popen("ps wux").readlines()
     javalines = [x for x in ps_output if re.search(r'\bbin/java\b', x)]
-    print ''.join(javalines)
+    print (''.join(javalines))
     assert len(javalines) == 0, msg
 
 # def test_doctimeout():
@@ -345,8 +344,8 @@ def assert_no_java(msg=""):
     import sys
     if sys.argv[1]=='modes':
         for mode,d in MODES_items:
-            print "  * `%s`: %s" % (mode, d['description'])
+            print ("  * `%s`: %s" % (mode, d['description']))
     if sys.argv[1]=='modes_json':
         # import json as stdjson
         # print stdjson.dumps(MODES, indent=4)
-        print '"%s"' % json.dumps(MODES).replace('"', r'\"')
+        print ('"%s"' % json.dumps(MODES).replace('"', r'\"'))