Merge pull request #7 from shineware/#6_add_heap_size_param

JVM Heap size 반영 및 테스트 코드 추가

Author: gamamoe

python/MANIFEST.in

@@ -3,3 +3,4 @@ include README.md
 include requirements.txt
 include PyKomoran/libs/**
 include PyKomoran/models_*/**
+include PyKomoran/tests/test_data/**

python/PyKomoran/core.py

@@ -22,13 +22,13 @@ class Komoran:
     Komoran Wrapper class
     """
 
-    def __init__(self, model_path="./models_full"):
+    def __init__(self, model_path="./models_full", max_heap=1024):
         self._base_path = os.path.dirname(os.path.realpath(__file__))
         self._model_path = os.path.abspath(os.path.join(self._base_path, model_path))
 
         assert os.path.exists(self._model_path)
 
-        jvm.init_jvm()
+        jvm.init_jvm(max_heap)
         self._komoran = jvm.get_jvm().kr.co.shineware.nlp.pykomoran.KomoranEntryPoint()
 
         self._komoran.init(self._model_path)

python/PyKomoran/jvm.py

@@ -10,7 +10,7 @@
 jvm_gateway = None
 
 
-def init_jvm(jar_path="./libs", max_heap=1024):
+def init_jvm(max_heap, jar_path="./libs"):
     base_path = os.path.dirname(os.path.realpath(__file__))
     jar_path = os.path.abspath(os.path.join(base_path, jar_path))
 

python/PyKomoran/tests/_jvm_test.py python/PyKomoran/tests/ajvm_test.pypython/PyKomoran/tests/_jvm_test.py renamed to python/PyKomoran/tests/ajvm_test.py

@@ -1,16 +1,17 @@
-""" This test filename has _ prefix because this test should run before running other tests """
+""" This test filename has 'a' prefix because this test should run before running other tests """
 import nose
+
 from PyKomoran.jvm import *
 
+global jvm_gateway
 if jvm_gateway is not None:
     jvm_gateway.shutdown()
 test_jvm1 = None
-test_jvm2 = None
 
 
 def test_to_before_init_Jvm():
     """
-    before jvm_init(), jvm_gateway should be None
+    JVM Test: before jvm_init(), jvm_gateway should be None
     :return:
     """
     global jvm_gateway
@@ -20,7 +21,7 @@ def test_to_before_init_Jvm():
 
 def test_to_get_Jvm_before_init():
     """
-    before jvm_init(), get_jvm() should be None
+    JVM Test: before jvm_init(), get_jvm() should be None
     :return:
     """
     global test_jvm1
@@ -32,25 +33,25 @@ def test_to_get_Jvm_before_init():
 
 def test_to_init_Jvm():
     """
-    when call jvm_init() first time, jvm object should be returned
+    JVM Test: when call jvm_init() first time, jvm object should be returned
     :return:
     """
     global test_jvm1
 
-    test_jvm1 = init_jvm()
+    test_jvm1 = init_jvm(1024)
 
     assert test_jvm1 is not None
 
 
 def test_to_duplicate_init_Jvm():
     """
-    when call jvm_init() more than once, None should be returned
+    JVM Test: when call jvm_init() more than once, None should be returned
     :return:
     """
     global test_jvm1
 
-    test_jvm2 = init_jvm()
-    test_jvm3 = init_jvm()
+    test_jvm2 = init_jvm(1024)
+    test_jvm3 = init_jvm(1024)
 
     assert test_jvm1 is not None
     assert test_jvm2 is None
@@ -59,7 +60,7 @@ def test_to_duplicate_init_Jvm():
 
 def test_to_get_Jvm_after_init():
     """
-    when call get_jvm() after init_jvm(), jvm object should be returned
+    JVM Test: when call get_jvm() after init_jvm(), jvm object should be returned
     :return:
     """
     test_jvm2 = get_jvm()
@@ -69,7 +70,7 @@ def test_to_get_Jvm_after_init():
 
 def test_to_duplicate_get_Jvm_after_init():
     """
-    when call get_jvm() after init_jvm() more than once, returned values should be same
+    JVM Test: when call get_jvm() after init_jvm() more than once, returned values should be same
     :return:
     """
     global test_jvm1

python/PyKomoran/tests/core_test.py

@@ -1,4 +1,6 @@
+import os
 import nose
+
 from PyKomoran.core import *
 from PyKomoran.type import *
 
@@ -8,7 +10,7 @@
 
 def test_to_init_Komoran():
     """
-    init Komoran with default model (models_full)
+    Core Test: init Komoran with default model (models_full)
     :return:
     """
     global komoran
@@ -19,19 +21,9 @@ def test_to_init_Komoran():
     assert komoran._komoran.isInitialized()
 
 
-def test_to_set_user_dic():
-    # TODO: implement test_to_set_user_dic() test code
-    pass
-
-
-def test_to_set_fw_dic():
-    # TODO: implement test_to_set_fw_dic() test code
-    pass
-
-
 def test_to_analyze_get_nouns():
     """
-    analyze test string with get_nouns() and check result is as expected
+    Core Test: analyze with get_nouns()
     :return:
     """
     global komoran
@@ -46,7 +38,7 @@ def test_to_analyze_get_nouns():
 
 def test_to_analyze_get_morphes_by_tags():
     """
-     analyze test string with get_morphes_by_tags() and check result is as expected
+     Core Test: analyze with get_morphes_by_tags()
     :return:
     """
     global komoran
@@ -61,7 +53,7 @@ def test_to_analyze_get_morphes_by_tags():
 
 def test_to_analyze_get_morphes_by_invalid_tags():
     """
-    analyze test string with get_morphes_by_tags(tag_list=['INVALID','POS']) and invalid tag_list and check result is as expected
+    Core Test: analyze with get_morphes_by_tags(tag_list=['INVALID','POS']) & invalid tag_list
     :return:
     """
     global komoran
@@ -76,7 +68,7 @@ def test_to_analyze_get_morphes_by_invalid_tags():
 
 def test_to_analyze_get_morphes_by_no_given_tags():
     """
-    analyze test string with get_morphes_by_tags(tag_list=[]) and check result is as expected
+    Core Test: analyze with get_morphes_by_tags(tag_list=[])
     :return:
     """
     global komoran
@@ -91,7 +83,7 @@ def test_to_analyze_get_morphes_by_no_given_tags():
 
 def test_to_analyze_get_plain_text():
     """
-    analyze test string with get_plain_text() and check result is as expected
+    Core Test: analyze with get_plain_text()
     :return:
     """
     global komoran
@@ -107,7 +99,7 @@ def test_to_analyze_get_plain_text():
 
 def test_to_analyze_get_token_list_with_flatten():
     """
-    analyze test string with get_token_list(flatten=False,use_pos_name=False) and check result is as expected
+    Core Test: analyze with get_token_list(flatten=False,use_pos_name=False)
     :return:
     """
     global komoran
@@ -145,7 +137,7 @@ def test_to_analyze_get_token_list_with_flatten():
 
 def test_to_analyze_get_token_list_with_flatten_and_use_pos_name():
     """
-    analyze test string with get_token_list(flatten=True,use_pos_name=True) and check result is as expected
+    Core Test: analyze with get_token_list(flatten=True,use_pos_name=True)
     :return:
     """
     global komoran
@@ -183,7 +175,7 @@ def test_to_analyze_get_token_list_with_flatten_and_use_pos_name():
 
 def test_to_analyze_get_token_list_without_flatten():
     """
-    analyze test string with get_token_list(flatten=False,use_pos_name=False) and check result is as expected
+    Core Test: analyze with get_token_list(flatten=False,use_pos_name=False)
     :return:
     """
     global komoran
@@ -222,7 +214,7 @@ def test_to_analyze_get_token_list_without_flatten():
 
 def test_to_analyze_get_token_list_without_flatten_and_use_pos_name():
     """
-    analyze test string with get_token_list(flatten=False,use_pos_name=True) and check result is as expected
+    Core Test: analyze with get_token_list(flatten=False,use_pos_name=True)
     :return:
     """
     global komoran
@@ -261,7 +253,7 @@ def test_to_analyze_get_token_list_without_flatten_and_use_pos_name():
 
 def test_to_analyze_get_list():
     """
-    analyze test string with get_list() and check result is as expected
+    Core Test: analyze with get_list()
     :return:
     """
     global komoran
@@ -291,5 +283,110 @@ def test_to_analyze_get_list():
     # @formatter:on
 
 
+def test_to_set_user_dic():
+    """
+    Core Test: test with set_user_dic()
+    :return:
+    """
+    global komoran
+
+    if komoran is None:
+        komoran = Komoran(model_path='./models_full')
+
+    tokens = komoran.get_token_list("테스트 단어")
+
+    # @formatter:off
+    assert isinstance(tokens, list)
+    assert len(tokens) == 2
+    assert isinstance(tokens[0], Token)
+    assert tokens[0] == Token({
+                            'morph': '테스트',
+                            'pos': 'NNP',
+                            'beginIndex': 0,
+                            'endIndex': 3
+                        })
+    assert tokens[1] == Token({
+                            'morph': '단어',
+                            'pos': 'NNG',
+                            'beginIndex': 4,
+                            'endIndex': 6
+                        })
+    # @formatter:on
+
+    base_path = os.path.dirname(os.path.realpath(__file__))
+    komoran.set_user_dic(os.path.join(base_path, "./test_data/dic.user"))
+
+    tokens = komoran.get_token_list("테스트 단어")
+
+    # @formatter:off
+    assert isinstance(tokens, list)
+    assert len(tokens) == 1
+    assert isinstance(tokens[0], Token)
+    assert tokens[0] == Token({
+                            'morph': '테스트 단어',
+                            'pos': 'NNP',
+                            'beginIndex': 0,
+                            'endIndex': 6
+                        })
+    # @formatter:on
+
+    pass
+
+
+def test_to_set_fw_dic():
+    # TODO: implement test_to_set_fw_dic() test code
+    """
+    Core Test: test with set_fw_dic()
+    :return:
+    """
+    global komoran
+
+    if komoran is None:
+        komoran = Komoran(model_path='./models_full')
+
+    tokens = komoran.get_token_list("테스트")
+
+    # @formatter:off
+    assert isinstance(tokens, list)
+    assert len(tokens) == 1
+    assert isinstance(tokens[0], Token)
+    assert tokens[0] == Token({
+                            'morph': '테스트',
+                            'pos': 'NNP',
+                            'beginIndex': 0,
+                            'endIndex': 3
+                        })
+    # @formatter:on
+
+    base_path = os.path.dirname(os.path.realpath(__file__))
+    komoran.set_fw_dic(os.path.join(base_path, "./test_data/fwd.user"))
+
+    tokens = komoran.get_token_list("테스트")
+
+    # @formatter:off
+    assert isinstance(tokens, list)
+    assert len(tokens) == 3
+    assert isinstance(tokens[0], Token)
+    assert tokens[0] == Token({
+                            'morph': '테',
+                            'pos': 'NNG',
+                            'beginIndex': 0,
+                            'endIndex': 3           # TODO: Check and fix KOMORAN
+                        })
+    assert tokens[1] == Token({
+                            'morph': '스',
+                            'pos': 'NNG',
+                            'beginIndex': 0,        # TODO: Check and fix KOMORAN
+                            'endIndex': 3           # TODO: Check and fix KOMORAN
+                        })
+    assert tokens[2] == Token({
+                            'morph': '트',
+                            'pos': 'NNG',
+                            'beginIndex': 0,        # TODO: Check and fix KOMORAN
+                            'endIndex': 3           # TODO: Check and fix KOMORAN
+                        })
+    # @formatter:on
+
+
 if __name__ == '__main__':
     nose.runmodule()

python/PyKomoran/tests/test_data/dic.user

@@ -0,0 +1 @@
+테스트 단어

python/PyKomoran/tests/test_data/fwd.user

@@ -0,0 +1 @@
+테스트	테/NNG 스/NNG 트/NNG

python/PyKomoran/tests/type_test.py

@@ -4,7 +4,7 @@
 
 def test_to_init_Token():
     """
-    init Token using given Dict, and validate given values with Token methods
+    Type Test: init Token using given Dict, and validate given values with Token methods
     :return:
     """
     # @formatter:off
@@ -33,7 +33,7 @@ def test_to_init_Token():
 
 def test_to_init_Token_using_Pos_name():
     """
-    init Token using given Dict with use_pos_name parameter, and validate given values with Token methods
+    Type Test: init Token using given Dict with use_pos_name parameter, and validate given values with Token methods
     :return:
     """
     # @formatter:off
@@ -62,7 +62,7 @@ def test_to_init_Token_using_Pos_name():
 
 def test_to_init_Pair():
     """
-    init Pair using given Dict, and validate given values with Pair methods
+    Type Test: init Pair using given Dict, and validate given values with Pair methods
     :return:
     """
     # @formatter:off
@@ -85,7 +85,7 @@ def test_to_init_Pair():
 
 def test_to_pos_table():
     """
-    check pos_table is initialized well
+    Type Test: check pos_table is initialized well
     :return:
     """
     global pos_table