argonism/IIR_1_1.4.py

## IIR_1_1.4.py
# p: posting list
# p must be like [docID_1, docID_2, docID_3, ...]
def union(p1, p2):
  answer = []
  p1_idx = 0
  p2_idx = 0
  count = 0
  while len(p1) > p1_idx or len(p2) > p2_idx:
    if len(p1) <= p1_idx:
      answer.append(p2[p2_idx])
      p2_idx += 1
      continue
    elif len(p2) <= p2_idx:
      answer.append(p1[p1_idx])
      p1_idx += 1
      continue

    if p1[p1_idx] == p2[p2_idx]:
      answer.append(p1[p1_idx])
      p1_idx += 1
      p2_idx += 1
    else:
      if p1[p1_idx] < p2[p2_idx]:
        answer.append(p1[p1_idx])
        p1_idx += 1
      else:
        answer.append(p2[p2_idx])
        p2_idx += 1
    count += 1
  # print("count:", count)
  return answer, count

# p: posting list
# p must be like [docID_1, docID_2, docID_3, ...]
def intersect(p1, p2):
  answer = []
  p1_idx = 0
  p2_idx = 0
  count = 0
  while len(p1) > p1_idx and len(p2) > p2_idx:
    if p1[p1_idx] == p2[p2_idx]:
      answer.append(p1[p1_idx])
      p1_idx += 1
      p2_idx += 1
    else:
      if p1[p1_idx] < p2[p2_idx]:
        p1_idx += 1
      else:
        p2_idx += 1
    count += 1
  print("count:", count)
  return answer, count

if __name__ == "__main__":
  def _not(doc_list, p):
    return [ e for e in doc_list if not e in p]

  def print_variables():
    x = len(Brutus)
    y = len(Caesar)
    N = len(D)
    print(f"x: {x}")
    print(f"y: {y}")
    print(f"N: {N}")
    print(f"NOT(Caesar): {NOT_caesar}")
    print(f"NOT(Caesar) length: {len(NOT_caesar)}")
    print(f"N - y > y: {N - y > y}")
    print(f"x + y: {x + y}")
    print(f"loop count: {count}")


  print("\n\n -- test intersect case -- \n\n")


  Brutus = [1, 2, 4, 11, 31, 45, 173, 174]
  D = [ n for n in range(1, 200)]
  print(D)

  print("********* N - y > y のとき *********")
  Caesar = [1, 2, 4, 5, 6, 16, 57, 132, 133]
  NOT_caesar = _not(D, Caesar)
  result, count = intersect(Brutus, NOT_caesar)

  print_variables()

  print("********* N - y < y のとき *********")
  # y > N/2 となるように、Dから3の倍数を除いたものを使う。
  Caesar = [n for n in D if not n % 3 == 0]
  NOT_caesar = _not(D, Caesar)
  result, count = intersect(Brutus, NOT_caesar)

  print_variables()


  print("\n\n -- test union case -- \n\n")


  print("********* N - y > y のとき *********")
  Caesar = [1, 2, 4, 5, 6, 16, 57, 132, 133]
  NOT_caesar = _not(D, Caesar)
  result, count = union(Brutus, NOT_caesar)

  print_variables()

  print("********* N - y < y のとき *********")
  # y > N/2 となるように、Dから3の倍数を除いたものを使う。
  Caesar = [n for n in D if not n % 3 == 0]
  NOT_caesar = _not(D, Caesar)
  result, count = union(Brutus, NOT_caesar)

  print_variables()
	# p: posting list
	# p must be like [docID_1, docID_2, docID_3, ...]
	def union(p1, p2):
	answer = []
	p1_idx = 0
	p2_idx = 0
	count = 0
	while len(p1) > p1_idx or len(p2) > p2_idx:
	if len(p1) <= p1_idx:
	answer.append(p2[p2_idx])
	p2_idx += 1
	continue
	elif len(p2) <= p2_idx:
	answer.append(p1[p1_idx])
	p1_idx += 1
	continue

	if p1[p1_idx] == p2[p2_idx]:
	answer.append(p1[p1_idx])
	p1_idx += 1
	p2_idx += 1
	else:
	if p1[p1_idx] < p2[p2_idx]:
	answer.append(p1[p1_idx])
	p1_idx += 1
	else:
	answer.append(p2[p2_idx])
	p2_idx += 1
	count += 1
	# print("count:", count)
	return answer, count

	# p: posting list
	# p must be like [docID_1, docID_2, docID_3, ...]
	def intersect(p1, p2):
	answer = []
	p1_idx = 0
	p2_idx = 0
	count = 0
	while len(p1) > p1_idx and len(p2) > p2_idx:
	if p1[p1_idx] == p2[p2_idx]:
	answer.append(p1[p1_idx])
	p1_idx += 1
	p2_idx += 1
	else:
	if p1[p1_idx] < p2[p2_idx]:
	p1_idx += 1
	else:
	p2_idx += 1
	count += 1
	print("count:", count)
	return answer, count

	if __name__ == "__main__":
	def _not(doc_list, p):
	return [ e for e in doc_list if not e in p]

	def print_variables():
	x = len(Brutus)
	y = len(Caesar)
	N = len(D)
	print(f"x: {x}")
	print(f"y: {y}")
	print(f"N: {N}")
	print(f"NOT(Caesar): {NOT_caesar}")
	print(f"NOT(Caesar) length: {len(NOT_caesar)}")
	print(f"N - y > y: {N - y > y}")
	print(f"x + y: {x + y}")
	print(f"loop count: {count}")


	print("\n\n -- test intersect case -- \n\n")


	Brutus = [1, 2, 4, 11, 31, 45, 173, 174]
	D = [ n for n in range(1, 200)]
	print(D)

	print("******* N - y > y のとき *******")
	Caesar = [1, 2, 4, 5, 6, 16, 57, 132, 133]
	NOT_caesar = _not(D, Caesar)
	result, count = intersect(Brutus, NOT_caesar)

	print_variables()

	print("******* N - y < y のとき *******")
	# y > N/2 となるように、Dから3の倍数を除いたものを使う。
	Caesar = [n for n in D if not n % 3 == 0]
	NOT_caesar = _not(D, Caesar)
	result, count = intersect(Brutus, NOT_caesar)

	print_variables()


	print("\n\n -- test union case -- \n\n")


	print("******* N - y > y のとき *******")
	Caesar = [1, 2, 4, 5, 6, 16, 57, 132, 133]
	NOT_caesar = _not(D, Caesar)
	result, count = union(Brutus, NOT_caesar)

	print_variables()

	print("******* N - y < y のとき *******")
	# y > N/2 となるように、Dから3の倍数を除いたものを使う。
	Caesar = [n for n in D if not n % 3 == 0]
	NOT_caesar = _not(D, Caesar)
	result, count = union(Brutus, NOT_caesar)

	print_variables()