| 146 | |
| 147 | |
| 148 | def pre_multi_process(file_list, q, id): |
| 149 | # print("{} start".format(id)) |
| 150 | sizes = [] |
| 151 | for i, sets_file in enumerate(file_list): |
| 152 | try: |
| 153 | df = pd.read_csv(sets_file, dtype='str', lineterminator='\n').dropna() |
| 154 | except pd.errors.ParserError as e: |
| 155 | print(sets_file) |
| 156 | data = df.values.T.tolist() |
| 157 | for vals in data: |
| 158 | # 需要对value去重 |
| 159 | sizes.append(len(set(vals))) |
| 160 | if id==0: |
| 161 | sys.stdout.write("\rId 0 Process Read and pre {}/{} files".format(i+1, len(file_list))) |
| 162 | if id==0: |
| 163 | sys.stdout.write("\n") |
| 164 | q.put((sizes,)) |
| 165 | # print("{} end,size={}".format(id,len(sizes))) |
| 166 | |
| 167 | def minhash_multi_process(file_list, q, param_dic, id): |
| 168 | # print("{} start".format(id)) |