Renamed to PyCobalt

.dockerignore

@@ -1,3 +1,9 @@
 examples/
 .gitignore
 README.md
+__pycache__/
+.idea/
+tests/
+run.py
+.git/
+build/

Dockerfile

@@ -2,21 +2,16 @@ FROM python:3.6.0
 MAINTAINER Open Semantics Group <osemantics@gmail.com>
 
 # update old version
-RUN pip list --local --outdated --format=freeze | cut -d= -f 1 | xargs -n1 pip install -U
+RUN pip list --local --outdated --format=freeze | cut -d= -f 1 | xargs -n1 pip install -U && \
+    pip install 'uwsgi==2.0.14'
 
+# add application
+ADD . /app/
 WORKDIR /app
-ADD requirements.txt /app/
-
 # Project requirements
-RUN pip install -r requirements.txt
-# .. plus some NLTK corpora
-RUN python -m nltk.downloader names punkt
-
-# Deployment requirements
-RUN pip install 'uwsgi==2.0.14'
-
-# finally add application
-ADD pycobalt/ service.py uwsgi.ini /app/
+RUN pip install -r requirements.txt \
+    && python -m nltk.downloader names punkt
+RUN python setup.py install
 
 # Go!
-CMD [ "uwsgi", "--ini", "uwsgi.ini" ]
+CMD [ "uwsgi", "--ini", "uwsgi.ini" ]

VERSION

@@ -1 +1 @@
-2.0.0-beta.1
+1.1.0-beta.3

docker-compose.yml

@@ -0,0 +1,21 @@
+version: '3'
+
+services:
+
+  coreference:
+    depends_on:
+      - corenlp
+    build:
+      context: .
+      dockerfile: Dockerfile
+    environment:
+      - PYCOBALT_CORENLP=http://corenlp:9000
+    ports:
+      - "5128:5128"
+    links:
+      - corenlp
+
+  corenlp:
+    image: "corenlp:370"
+    expose:
+      - "9000"

pycobalt/resolving.py

@@ -266,7 +266,7 @@ def __first_name_mention(name_references: List[NameReference],
             entity_tokens = []
             for token in sentence.tokens:
                 if len(token.word) > 0 and token.word[0].isupper():
-                    entity_tokens.append(token.word)
+                    entity_tokens.append(token)
                 else:
                     return NameReference(sentence_index=0,
                                          tokens=entity_tokens,

requirements.txt

@@ -1,4 +1,5 @@
-joblib==0.11
-pycorenlp==0.3.0
-Flask==0.12
-nose
+nltk
+joblib
+pycorenlp
+Flask
+nose

service.py

@@ -1,6 +1,7 @@
 # -*- coding: utf-8 -*-
 
 import datetime
+import sys
 
 from flask import Flask, jsonify, make_response, request
 

setup.py

@@ -0,0 +1,13 @@
+#!/usr/bin/env python
+
+from distutils.core import setup
+
+with open('VERSION', 'r') as i:
+    version = i.read().strip()
+
+setup(name='PyCobalt',
+      version=version,
+      description='Coreference Resolution',
+      author='Bernhard Bermeitinger',
+      packages=['pycobalt'],
+     )

tests/test_WikiArticle.py

@@ -1,5 +1,6 @@
 import logging
 import unittest
+import nose
 
 import os
 
@@ -8,11 +9,11 @@
 logging.getLogger('requests').setLevel(logging.WARNING)
 
 
-class TestResolverForBarackObama(unittest.TestCase):
+class TestResolverWithActualData(unittest.TestCase):
     def setUp(self):
         self.maxDiff = None
 
-    def test_resolver_for_first_paragraphs(self):
+    def test_resolver_for_first_paragraphs_BarackObama(self):
         with open(os.path.join('examples', 'short_input.txt')) as i:
             sentences = [l for l in i]
 
@@ -34,7 +35,7 @@ def test_resolver_for_first_paragraphs(self):
         for s, e in zip(substituted, expected):
             self.assertEquals(s, e)
 
-    def test_resolver_for_full_article(self):
+    def test_resolver_for_article_BarackObama(self):
         with open(os.path.join('examples', 'Barack_Obama.input.txt')) as i:
             sentences = [l for l in i]
 
@@ -56,6 +57,19 @@ def test_resolver_for_full_article(self):
         for s, e in zip(substituted, expected):
             self.assertEquals(s, e)
 
+    def test_resolver_for_full_article_Paris(self):
+        with open(os.path.join('examples', 'Paris.input.txt')) as i:
+            text = i.read()
+
+        self.assertIsNotNone(text)
+
+        sentences, substitutions = resolve(text, "")
+
+        self.assertIsNotNone(sentences)
+        self.assertEquals(len(sentences), 657)
+        self.assertIsNotNone(substitutions)
+        self.assertGreater(len(substitutions), 100)
+
 
 if __name__ == '__main__':
-    unittest.main()
+    nose.main()