Code Duplication - milvus-io/milvus - Measure and Improve Code Quality continuously with Scrutinizer

Code Duplication Length = 22-23 lines in 2 locations

tests/milvus_groundtruth/milvus_ground_truth.py 2 locations


    return dist


def get_ground_truth_l2(topk, idx, vct_nq):
    filenames = os.listdir(BASE_FOLDER_NAME)
    filenames.sort()
    no_dist = {}
    k = 0
    for filename in filenames:
        vec_list = load_vec_list(BASE_FOLDER_NAME + '/' + filename)
        for j in range(len(vec_list)):
            dist = calEuclideanDistance(vct_nq, vec_list[j])
            num_j = "%01d%03d%06d" % (8, k, j)
            if k==0 and j<topk :
                no_dist[num_j] = dist
            else:
                # sorted by values
                max_key = max(no_dist, key=no_dist.get)
                max_value = no_dist[max_key]
                if dist < max_value:
                    m = no_dist.pop(max_key)
                    no_dist[num_j] = dist
        k += 1
    no_dist = sorted(no_dist.items(), key=lambda x: x[1])
    print(no_dist)
    save_gt_file(no_dist, idx)


def get_ground_truth_ip(topk, idx, vct_nq):

    save_gt_file(no_dist, idx)


def get_ground_truth_ip(topk, idx, vct_nq):
    filenames = os.listdir(BASE_FOLDER_NAME)  # get the whole file names
    filenames.sort()
    no_dist = {}
    k = 0
    for filename in filenames:
        vec_list = load_vec_list(BASE_FOLDER_NAME + '/' + filename)
        for j in range(len(vec_list)):
            dist = calInnerDistance(vct_nq, vec_list[j])
            num_j = "%03d%06d" % (k, j)
            if k==0 and j<topk :
                no_dist[num_j] = dist
            else:
                min_key = min(no_dist, key=no_dist.get)
                min_value = no_dist[min_key]
                if dist > min_value:
                    m = no_dist.pop(min_key)
                    no_dist[num_j] = dist
        k += 1
    no_dist = sorted(no_dist.items(), key=lambda x: x[1], reverse=True)
    print(no_dist)
    save_gt_file(no_dist, idx)


def get_ground_truth_tanimoto(topk, idx, vec_nq):

		@@ 90-112 (lines=23) @@
87		return dist
88
89
90		def get_ground_truth_l2(topk, idx, vct_nq):
91		filenames = os.listdir(BASE_FOLDER_NAME)
92		filenames.sort()
93		no_dist = {}
94		k = 0
95		for filename in filenames:
96		vec_list = load_vec_list(BASE_FOLDER_NAME + '/' + filename)
97		for j in range(len(vec_list)):
98		dist = calEuclideanDistance(vct_nq, vec_list[j])
99		num_j = "%01d%03d%06d" % (8, k, j)
100		if k==0 and j<topk :
101		no_dist[num_j] = dist
102		else:
103		# sorted by values
104		max_key = max(no_dist, key=no_dist.get)
105		max_value = no_dist[max_key]
106		if dist < max_value:
107		m = no_dist.pop(max_key)
108		no_dist[num_j] = dist
109		k += 1
110		no_dist = sorted(no_dist.items(), key=lambda x: x[1])
111		print(no_dist)
112		save_gt_file(no_dist, idx)
113
114
115		def get_ground_truth_ip(topk, idx, vct_nq):
		@@ 115-136 (lines=22) @@
112		save_gt_file(no_dist, idx)
113
114
115		def get_ground_truth_ip(topk, idx, vct_nq):
116		filenames = os.listdir(BASE_FOLDER_NAME) # get the whole file names
117		filenames.sort()
118		no_dist = {}
119		k = 0
120		for filename in filenames:
121		vec_list = load_vec_list(BASE_FOLDER_NAME + '/' + filename)
122		for j in range(len(vec_list)):
123		dist = calInnerDistance(vct_nq, vec_list[j])
124		num_j = "%03d%06d" % (k, j)
125		if k==0 and j<topk :
126		no_dist[num_j] = dist
127		else:
128		min_key = min(no_dist, key=no_dist.get)
129		min_value = no_dist[min_key]
130		if dist > min_value:
131		m = no_dist.pop(min_key)
132		no_dist[num_j] = dist
133		k += 1
134		no_dist = sorted(no_dist.items(), key=lambda x: x[1], reverse=True)
135		print(no_dist)
136		save_gt_file(no_dist, idx)
137
138
139		def get_ground_truth_tanimoto(topk, idx, vec_nq):

milvus-io / milvus

Code Duplication Length = 22-23 lines in 2 locations

tests/milvus_groundtruth/milvus_ground_truth.py 2 locations