class pylucene_test(unittest.TestCase):
def setUp(self):
if FLAG_TO_CONTROL_JVM == 0:
self.pylucene = PyLucene(startJVM=True)
set_FLAG_TO_CONTROL_JVM()
else:
self.pylucene = PyLucene(startJVM=False)
def tearDown(self):
shutil.rmtree(self.pylucene.STORE_DIR)
# testing functions
def test_pylucene_01(self):
"""
test in search in general_info field with a big set of docs
"""
for i in range(0, 200):
finalDate = "%d/%d/1900" % (i, i)
doc_dict = {
"doc_id": i,
"general_info": "title subtitle",
"subject": "subject",
"source": "source",
"initial_date": "1800",
"final_date": finalDate,
"content": "content1",
}
self.pylucene.index_doc(doc_dict)
docs = self.pylucene.search_docs(value="title", field="general_info")
self.assertEquals(200, len(docs))
for i in range(len(docs)):
self.assertEquals(docs[i].get("general_info"), "title subtitle")
self.assertEquals(docs[i].get("subject"), "subject")
self.assertEquals(docs[i].get("source"), "source")
self.assertEquals(docs[i].get("initial_date"), "1800")
# no guaranteed order
# self.assertEquals(docs[i].get("final_date"), "%d/%d/1900" % (i,i))
self.assertEquals(docs[i].get("content"), None)
def setUp(self):
if FLAG_TO_CONTROL_JVM == 0:
self.pylucene = PyLucene(startJVM=True)
set_FLAG_TO_CONTROL_JVM()
else:
self.pylucene = PyLucene(startJVM=False)
class pylucene_test(unittest.TestCase):
def setUp(self):
if FLAG_TO_CONTROL_JVM == 0:
self.pylucene = PyLucene(startJVM=True)
set_FLAG_TO_CONTROL_JVM()
else:
self.pylucene = PyLucene(startJVM=False)
def tearDown(self):
shutil.rmtree(self.pylucene.STORE_DIR)
# testing functions
# def test_pylucene_01(self):
# """
# test in search in general_info field
# """
#
# for i in range(0, 3):
# doc_dict = {
# "doc_id" : str(i),
# "general_info" : "title subtitle",
# "subject" : "subject",
# "source" : "source",
# "initial_date" : "1800",
# "final_date" : "%d/%d/1900" % (i, i),
# "content" : "content1"
# }
#
# self.pylucene.index_doc(doc_dict)
#
# docs = self.pylucene.search_docs(value="title", field="general_info")
#
# self.assertEquals(3, len(docs))
#
#
#
# def test_pylucene_02(self):
# """
# test match with field tokenized and in general search field
# """
#
# for i in range(0, 3):
# doc_dict = {
# "doc_id" : str(i),
# "general_info" : "title subtitle",
# "subject" : "subject",
# "source" : "source",
# "initial_date" : "1800",
# "final_date" : "%d/%d/1900" % (i, i),
# "content" : "content1"
# }
#
# self.pylucene.index_doc(doc_dict)
#
# docs = self.pylucene.search_docs("title")
#
# self.assertEquals(3, len(docs))
#
#
# def test_pylucene_03(self):
# """
# test perfect match with a specific field
# """
#
# for i in range(0, 3):
# doc_dict = {
# "doc_id" : str(i),
# "general_info" : "title subtitle",
# "subject" : "subject%d" % (i),
# "source" : "source",
# "initial_date" : "1800",
# "final_date" : "%d/%d/1900" % (i, i),
# "content" : "content1"
# }
#
# self.pylucene.index_doc(doc_dict)
#
# docs = self.pylucene.search_docs(value="subject1", field="subject")
#
# self.assertEquals(1, len(docs))
#
#
# def test_pylucene_04(self):
# """
# test search for date fields (year/month/day)
# """
#
# for i in range(0, 3):
# doc_dict = {
# "doc_id" : str(i),
# "general_info" : "title subtitle",
# "subject" : "subject%d" % (i),
# "source" : "source",
# "initial_date" : "1800",
# "final_date" : "%d/%d/1900" % (i, i),
# "content" : "content1"
# }
#
# self.pylucene.index_doc(doc_dict)
#
# docs = self.pylucene.search_docs(value="1900", field="final_date")
#
# self.assertEquals(3, len(docs))
#
#
# def test_pylucene_05(self):
# """
# test search for number
# """
#
# for i in range(0, 3):
# doc_dict = {
# "doc_id" : str(i),
# "general_info" : "title subtitle",
# "subject" : "subject%d" % (i),
# "source" : "source",
# "initial_date" : "1800",
# "final_date" : "%d/%d/1900" % (i, i),
# "content" : "content1"
# }
#
# self.pylucene.index_doc(doc_dict)
#
# docs = self.pylucene.search_docs(value="19", field="final_date")
#
# self.assertEquals(0, len(docs))
def test_pylucene_06(self):
"""
test search for specific value in final_date field
"""
for i in range(0, 3):
doc_dict = {
"doc_id" : str(i),
"general_info" : "title subtitle",
"subject" : "subject%d" % (i),
"source" : "source",
"initial_date" : "1800",
"final_date" : "%d/%d/1900" % (i, i),
"content" : "content1",
}
self.pylucene.index_doc(doc_dict)
docs = self.pylucene.search_docs(value="2/2/1900", field="final_date")
self.assertEquals(3, len(docs))
self.assertEquals(str(2), docs[0].get("doc_id"))
def test_pylucene_07(self):
"""
test search for specific value in final_date field with different separators
"""
for i in range(0, 3):
doc_dict = {
"doc_id" : str(i),
"general_info" : "title subtitle",
"subject" : "subject%d" % (i),
"source" : "source",
"initial_date" : "1800",
"final_date" : "%d/%d/1900" % (i, i),
"content" : "content1",
}
self.pylucene.index_doc(doc_dict)
docs = self.pylucene.search_docs(value="2-2-1900", field="final_date")
self.assertEquals(3, len(docs))
self.assertEquals(str(2), docs[0].get("doc_id"))