def test_get_window_two_result(self):
result = self.x.multiple_search_lim('тестов', 0, 1)
res1 = result['test_window_three.txt'][0]
res2 = result['test_window_three.txt'][1]
t = [Position(18, 24, 0), Position(18, 24, 1)]
ideal = {'test_window_three.txt': t}
self.assertEqual(result, ideal)
def test_get_window_two_result(self):
result = self.x.find_supplemented_window_lim('тестов', 1, 0, 1,
[(0, 2)])
res1 = result['test_window_three.txt'][0]
res2 = result['test_window_three.txt'][1]
result1 = self.x.find_supplemented_window_lim_v2(
'тестов', 1, 0, 1, [(0, 2)])
res11 = result1['test_window_three.txt'][0]
res21 = result1['test_window_three.txt'][1]
win1 = TokenWindow(self.strr3.replace('\n', ''), [Position(18, 24, 0)],
0, 25)
win2 = TokenWindow(self.strr4, [Position(18, 24, 1)], 0, 25)
ideal = {'test_window_three.txt': [win1, win2]}
self.assertEqual(res1.allString, win1.allString)
self.assertEqual(res1, win1)
self.assertEqual(res2.allString, win2.allString)
self.assertEqual(res2, win2)
self.assertEqual(result, ideal)
self.assertEqual(res11.allString, win1.allString)
self.assertEqual(res11, win1)
self.assertEqual(res21.allString, win2.allString)
self.assertEqual(res21, win2)
self.assertEqual(result1, ideal)
def test_get_window_two_result3(self):
result = self.x.multiple_search_lim_gen('тестов', 0, 1)
t = self.x.merge_and_sort_lists(
[[Position(18, 24, 0), Position(18, 24, 1)]])
ideal = {'test_window_three.txt': t}
self.assertEqual(list(result['test_window_three.txt']),
list(ideal['test_window_three.txt']))
def test_get_window_two_word(self):
result = self.x.find_supplemented_window_lim_v3(
'Мы работы', 2, 0, 1, [(0, 2)])
res = result['test_window_two.txt'][0]
win = TokenWindow(
self.strr2,
[Position(0, 2, 0), Position(32, 38, 0)], 0, 49)
ideal = {'test_window_two.txt': [win]}
self.assertEqual(res.allString, win.allString)
self.assertEqual(res, win)
self.assertEqual(result, ideal)
def test_two_identical_words(self):
test = open("text.txt", 'w' )
test.write("my my")
test.close()
self.indexer.index("text.txt")
words1 = dict(shelve.open("database"))
words2 = {
"my":{"text.txt": [Position(0, 2),
Position(3, 5)]
}}
self.assertEqual(words1, words2)
def test_indexing(self):
"""test if the program is working correctly
when indexing a single file
"""
f = open('test.txt', 'w')
f.write('this is a test test')
f.close()
self.i.indexing('test.txt')
db_dict = dict(shelve.open('dbase'))
dictionary = {
'this': {'test.txt': [Position(0, 4)]},
'is': {'test.txt': [Position(5, 7)]},
'a': {'test.txt': [Position(8, 9)]},
'test': {'test.txt': [Position(10, 14), Position(15, 19)]}
}
self.assertEqual(db_dict, dictionary)
def test_two_files(self):
test = open("text.txt", 'w' )
test.write("test")
test.close()
test = open("text1.txt", 'w' )
test.write("my my")
test.close()
self.indexer.indexing("text.txt")
self.indexer = Indexer('database')
self.indexer.indexing("text1.txt")
words1 = dict(shelve.open("database"))
words2 = {
"my":{"text.txt": [Position(0, 2),
Position(3, 5)]},
"test":{"text.txt": [Position(0, 4)]
}}
self.assertEqual(words1, words2)
def test_get_window_begin(self):
result = self.x.find_window('sun', 1)
res = result['test_window_one.txt'][0]
win = TokenWindow(self.strr, [Position(0, 3, 0)], 0, 10)
ideal = {'test_window_one.txt': [win]}
self.assertEqual(res.allString, win.allString)
self.assertEqual(res, win)
self.assertEqual(result, ideal)
def test_one_word(self):
file = open('test.txt', 'w')
file.write('indexator')
file.close()
self.indexator.indextie('test.txt')
data_dict = dict(shelve.open('database'))
dictionary = {'indexator': {'test.txt': [Position(0, 9)]}}
self.assertEqual(data_dict, dictionary)
def test_many_files(self):
file_one = open('testfile_1.txt', 'w')
file_one.write(' Ф 12 !!! @ # Alina is a student)))')
file_one.close()
file_two = open('testfile_2.txt', 'w')
file_two.write('Alina')
file_two.close()
self.indexator.get_index('testfile_1.txt')
self.indexator.get_index('testfile_2.txt')
del self.indexator
base_dict = dict(shelve.open('database'))
cool_result = {
'Ф': {
'testfile_1.txt': [Position(1, 2)]
},
'12': {
'testfile_1.txt': [Position(3, 5)]
},
'Alina': {
'testfile_1.txt': [Position(14, 19)],
'testfile_2.txt': [Position(0, 5)]
},
'is': {
'testfile_1.txt': [Position(20, 22)]
},
'a': {
'testfile_1.txt': [Position(23, 24)]
},
'student': {
'testfile_1.txt': [Position(25, 32)]
}
}
self.assertEqual(base_dict, cool_result)
os.remove('testfile_1.txt')
os.remove('testfile_2.txt')
def test_get_window_simple2(self):
result = self.x.find_supplemented_window_lim_v3(
'нашу', 2, 0, 1, [(0, 1)])
res = result['test_window_two.txt'][0]
win = TokenWindow(self.strr2, [Position(13, 17, 0)], 0, 49)
ideal = {'test_window_two.txt': [win]}
self.assertEqual(res.allString, win.allString)
self.assertEqual(res, win)
self.assertEqual(result, ideal)
def test_get_window_simple_two_line(self):
result = self.x.find_supplemented_window_lim_v3(
'Вторая', 1, 0, 1, [(0, 1)])
res = result['test_window_three.txt'][0]
win = TokenWindow(self.strr4, [Position(0, 6, 1)], 0, 25)
ideal = {'test_window_three.txt': [win]}
self.assertEqual(res.allString, win.allString)
self.assertEqual(res, win)
self.assertEqual(result, ideal)
def test_get_window_end(self):
result = self.x.find_supplemented_window_lim_v3(
'good', 1, 0, 1, [(0, 1)])
res = result['test_window_one.txt'][0]
win = TokenWindow(self.strr, [Position(46, 50, 0)], 0, 50)
ideal = {'test_window_one.txt': [win]}
self.assertEqual(res.allString, win.allString)
self.assertEqual(res, win)
self.assertEqual(result, ideal)
def test_many_words(self):
file = open('test.txt', 'w')
file.write('testing my indexator')
file.close()
self.indexator.indextie('test.txt')
data_dict = dict(shelve.open('database'))
dictionary = {
'testing': {
'test.txt': [Position(0, 7)]
},
'my': {
'test.txt': [Position(8, 10)]
},
'indexator': {
'test.txt': [Position(11, 20)]
}
}
self.assertEqual(data_dict, dictionary)
def test_get_window_simple(self):
result = self.x.find_window('tree', 2)
res = result['test_window_one.txt'][0]
win = TokenWindow(self.strr, [Position(11, 15, 0)], 0, 27)
ideal = {'test_window_one.txt': [win]}
self.assertEqual(res.allString, win.allString)
self.assertEqual(res, win)
self.assertEqual(result, ideal)
def test_get_window_two_result3(self):
result = self.x.find_supplemented_window_lim_v3(
'тестов', 1, 0, 1, [(1, 2)])
res1 = result['test_window_three.txt'][0]
win1 = TokenWindow(self.strr4.replace('\n', ''), [Position(18, 24, 1)],
0, 25)
ideal = {'test_window_three.txt': [win1]}
self.assertEqual(res1.allString, win1.allString)
self.assertEqual(res1, win1)
self.assertEqual(result, ideal)
def test_one_word(self):
test = open("text.txt", 'w' )
test.write("programming")
test.close()
self.indexer.indexing("text.txt")
words1 = dict(shelve.open("database"))
words2 = {
"programming":{"text.txt": [Position(0, 10)]
}}
self.assertEqual(words1, words2)
def test_two_files(self):
file1 = open('test1.txt', 'w')
file1.write('file number one')
file1.close()
self.indexator.indextie('test1.txt')
test2 = open('test2.txt', 'w')
test2.write('file number two')
test2.close()
self.indexator.indextie('test2.txt')
data_dict = dict(shelve.open('database'))
dictionary = {
'file': {
'test1.txt': [Position(0, 4)],
'test2.txt': [Position(0, 4)]
},
'number': {
'test1.txt': [Position(5, 11)],
'test2.txt': [Position(5, 11)]
},
'one': {
'test1.txt': [Position(12, 15)]
},
'two': {
'test2.txt': [Position(12, 15)]
}
}
self.assertEqual(data_dict, dictionary)
def test_database_one_token(self):
test_file = open('testfile.txt', 'w')
test_file.write('Alina')
test_file.close()
self.indexator.get_index('testfile.txt')
del self.indexator
base_dict = dict(shelve.open('database'))
cool_result = {
'Alina': {
'testfile.txt': [Position(0, 5)]
}
} # perfect dictionary
self.assertEqual(base_dict, cool_result)
os.remove('testfile.txt')
def test_database_continued(self):
file = open('test.txt', 'w')
file.write('Baikal')
file.close()
self.x.prescribe_index('test.txt')
file = open('test2.txt', 'w')
file.write('Baikal is a lake')
file.close()
self.x.prescribe_index('test2.txt')
files = os.listdir(path=".")
flag = False
for f in files:
if f.startswith('database.'):
flag = True
elif f.startswith('database'):
if f == 'database':
flag = True
self.assertEqual(flag, True)
db_dict = dict(self.x.db)
standart_result = {
'Baikal': {
'test2.txt': [Position(0, 6, 0)],
'test.txt': [Position(0, 6, 0)]
},
'is': {
'test2.txt': [Position(7, 9, 0)]
},
'a': {
'test2.txt': [Position(10, 11, 0)]
},
'lake': {
'test2.txt': [Position(12, 16, 0)]
}
}
self.assertEqual(db_dict, standart_result)
os.remove('test2.txt')
os.remove('test.txt')
def test_indexing_multiple_files(self):
"""test if the program is working correctly
when indexing multiple files
"""
p = open('test.txt', 'w')
p.write('is this a test?')
p.close()
self.i.indexing('test.txt')
s = open('ts.txt', 'w')
s.write('yes, it is a test')
s.close()
self.i.indexing('ts.txt')
db_dict = dict(shelve.open('dbase'))
dictionary = {
'is': {'test.txt': [Position(0, 2)], 'ts.txt': [Position(8, 10)]},
'this': {'test.txt': [Position(3, 7)]},
'a': {'test.txt': [Position(8, 9)], 'ts.txt': [Position(11, 12)]},
'test': {'test.txt': [Position(10, 14)], 'ts.txt': [Position(13, 17)]},
'yes': {'ts.txt': [Position(0, 3)]},
'it': {'ts.txt': [Position(5, 7)]}
}
self.assertEqual(db_dict, dictionary)
def test_get_window_simple(self):
result = self.x.find_supplemented_window_lim('tree', 2, 0, 1, [(0, 1)])
res = result['test_window_one.txt'][0]
result1 = self.x.find_supplemented_window_lim_v2(
'tree', 2, 0, 1, [(0, 1)])
res1 = result1['test_window_one.txt'][0]
win = TokenWindow(self.strr, [Position(11, 15, 0)], 0, 50)
ideal = {'test_window_one.txt': [win]}
self.assertEqual(res.allString, win.allString)
self.assertEqual(res, win)
self.assertEqual(result, ideal)
self.assertEqual(res1.allString, win.allString)
self.assertEqual(res1, win)
self.assertEqual(result1, ideal)
def test_database_created(self):
file = open('test.txt', 'w')
file.write('Baikal')
file.close()
self.x.prescribe_index('test.txt')
'''
The method listdir() returns a list containing the names of the entries in the directory given by path.
'''
files = os.listdir(path=".")
flag = False
for f in files:
if f.startswith('database.'):
flag = True
elif f.startswith('database'):
if f == 'database':
flag = True
self.assertEqual(flag, True)
db_dict = dict(self.x.db)
standart_result = {'Baikal': {'test.txt': [Position(0, 6, 0)]}}
self.assertEqual(db_dict, standart_result)
os.remove('test.txt')
import unittest
from searchengine import SearchEngine
from indexer import Position
import shelve
import os
file1 = " The Earth is round"
file2 = " The sun is round"
idealdict = {
'sun': {
'file2.txt': [Position(5, 8, 0)]
},
'The': {
'file1.txt': [Position(1, 4, 0)],
'file2.txt': [Position(1, 4, 0)]
},
'Earth': {
'file1.txt': [Position(5, 10, 0)]
},
'round': {
'file1.txt': [Position(14, 19, 0)],
'file2.txt': [Position(12, 17, 0)]
},
'is': {
'file1.txt': [Position(11, 13, 0)],
'file2.txt': [(9, 11, 0)]
}
}
class TestMySearchEngine(unittest.TestCase):
def test_get_window_simple(self):
result = self.x.multiple_search_lim('tree', 0, 1)
res = result['test_window_one.txt'][0]
t = [Position(11, 15, 0)]
ideal = {'test_window_one.txt': t}
self.assertEqual(result, ideal)
def test_get_window_end(self):
result = self.x.multiple_search_lim('good', 0, 1)
res = result['test_window_one.txt'][0]
t = [Position(46, 50, 0)]
ideal = {'test_window_one.txt': t}
self.assertEqual(result, ideal)
def test_get_window_simple2(self):
result = self.x.multiple_search_lim('нашу', 0, 1)
res = result['test_window_two.txt'][0]
t = [Position(13, 17, 0)]
ideal = {'test_window_two.txt': t}
self.assertEqual(result, ideal)
def test_get_window_simple_two_line(self):
result = self.x.multiple_search_lim('Вторая', 0, 1)
res = result['test_window_three.txt'][0]
t = [Position(0, 6, 1)]
ideal = {'test_window_three.txt': t}
self.assertEqual(result, ideal)
def test_search_two_tokens(self):
result = self.x.multiple_search('The sun')
self.assertIsInstance(result, dict)
self.assertEqual(result,
{'file2.txt': [Position(1, 4, 0),
Position(5, 8, 0)]})
import unittest
from searchengine import SearchEngine, TokenWindow
from indexer import Position
import os
idealdict = {
'sun': {
'test_window_one.txt': [Position(0, 3, 0)]
},
'window': {
'test_window_one.txt': [Position(4, 10, 0)]
},
'tree': {
'test_window_one.txt': [Position(11, 15, 0)]
},
'apple': {
'test_window_one.txt': [Position(16, 21, 0)]
},
'juse': {
'test_window_one.txt': [Position(23, 27, 0)]
},
'border': {
'test_window_one.txt': [Position(28, 34, 0)]
},
'films': {
'test_window_one.txt': [Position(35, 40, 0)]
},
'23': {
'test_window_one.txt': [Position(41, 43, 0)]
},
'45': {