def lsi():
if len(sys.argv) == 1:
k = 100 # Default dimensions is 100
elif len(sys.argv) == 2:
k = int(sys.argv[1])
else:
shared.error("11", ["lsi", ""])
return -1
try:
texts, documents = shared.build_texts("lsi")
except:
shared.error("0", ["lsi", k])
return -1
try:
ck = get_lsi(texts, k)
except:
shared.error("4", ["lsi", k])
return -1
try:
shared.write_to_file(ck, documents, LSIFOLDER, "lsi.csv")
except:
shared.error("8", ["lsi", k])
return -1
try:
shared.insert_to_db("lsi", k, "Finished")
except:
shared.error("10", ["lsi", k])
return -1
return 1
distances = []
min_distance = float('inf')
for idx, point1 in enumerate(points):
for point2 in points[idx + 1:]:
if point1 != point2:
distance = Point.distance(point1, point2)
min_distance = distance if distance < min_distance else min_distance
distance_tuple = (distance, point1, point2)
distances.append(distance_tuple)
return format_point_list(distances, min_distance)
def format_point_list(points, min_distance):
closest_points = [p for p in points if p[0] == min_distance]
closest_points.sort(key=lambda point: point[1].x)
closest_points.sort(key=lambda point: point[1].y)
point_string = str(min_distance)
for p in closest_points:
point_string += "\n{0} {1} {2} {3}".format(p[1].x, p[1].y, p[2].x,
p[2].y)
return point_string
# Prevent running if imported as a module
if __name__ == "__main__":
POINT_LIST = point_list_from_file(argv[1])
result = closest_pair_brute(POINT_LIST)
print result
write_to_file('output_bruteforce.txt', result)
p for p in y_sorted_points if left_bound <= p.x <= right_bound
]
return closest_cross_pair(middle_set, closest_pair)
def closest_cross_pair(points, closest_pair):
for idx, point1 in enumerate(points):
for point2 in points[idx + 1:]:
if point2.y - point1.y > closest_pair[0]:
break
distance = Point.distance(point1, point2)
if distance < closest_pair[0]:
closest_pair = (distance, point1, point2)
return closest_pair
def enhanced_closest_pair_dnc_main(points):
x_sorted_points = sorted(points, key=lambda point: point.x)
y_sorted_points = sorted(points, key=lambda point: point.y)
return enhanced_closest_pair_dnc(x_sorted_points, y_sorted_points)
# Prevent running if imported as a module
if __name__ == "__main__":
POINT_LIST = point_list_from_file(argv[1])
result = enhanced_closest_pair_dnc_main(POINT_LIST)
print format_output(result)
write_to_file('output_enhanceddnc.txt', format_output(result))
elif a_shift == D[i][j]:
B[i][j] = Shift.a_shift
elif b_shift == D[i][j]:
B[i][j] = Shift.b_shift
trace = backtrace(sequence_a[1:], sequence_b[1:], B)
final_cost = D[len(sequence_a) - 1][len(sequence_b) - 1]
aligned_sequence_a = trace[0]
aligned_sequence_b = trace[1]
return Solution(aligned_sequence_a, aligned_sequence_b, final_cost)
# Prevent running if imported as a module
if __name__ == "__main__":
# Load from command line arguments or default files, otherwise exit
if len(sys.argv) == 3:
COST_MATRIX = CostMatrix.from_file(sys.argv[1])
SEQUENCES = load_sequences_from_file(sys.argv[2])
elif path.isfile("imp2cost.txt") and path.isfile("imp2input.txt"):
COST_MATRIX = CostMatrix.from_file("imp2cost.txt")
SEQUENCES = load_sequences_from_file("imp2input.txt")
else:
sys.exit("ERROR: No input supplied via files or command line args.\n"
"USAGE: $ python sequence_align.py {COST_FILE} {INPUT_FILE}")
# Map solutions to list, write to file
SOLUTIONS = [str(align_sequences(*sequence)) for sequence in SEQUENCES]
write_to_file('imp2output.txt', '\n'.join(SOLUTIONS))
# Run recursive algorithm on left and right halves
closest_pair = min(closest_pair_dnc(left_set), closest_pair_dnc(right_set))
# Get middle set (technically a list)
left_bound = median - closest_pair[0]
right_bound = median + closest_pair[0]
middle_set = [p for p in points if left_bound <= p.x <= right_bound]
return closest_cross_pair(middle_set, closest_pair)
def closest_cross_pair(points, closest_pair):
points.sort(key=lambda point: point.y)
for idx, point1 in enumerate(points):
for point2 in points[idx + 1:]:
if point2.y - point1.y > closest_pair[0]:
break
distance = Point.distance(point1, point2)
if distance < closest_pair[0]:
closest_pair = (distance, point1, point2)
return closest_pair
# Prevent running if imported as a module
if __name__ == "__main__":
POINT_LIST = point_list_from_file(argv[1])
result = closest_pair_dnc(POINT_LIST)
print format_output(result)
write_to_file('output_divideandconquer.txt', format_output(result))
def test_write_to_file(self):
shared.write_to_file([[1,3],[3,2]], ['doc0','doc1'], 'testing/',\
'tmp.csv')
self.assertTrue(os.path.getsize('testing/tmp.csv') > 0)
os.remove('testing/tmp.csv')