Question

我需要在praralel中处理一些文件。我正在使用池，但是在保存池处理的文件时遇到了麻烦。这是代码：

... All imports...

def extract(text_lines):

    line_tr01 = []
    line_tr02 = []
    line_tr03 = []
    line_tr03 = []
    for line in text_lines:
        treatment01 = treatment_a(line, args)
        line_tr01.append(treatment01)
        treatment02 = treatment_b(line, args)
        line_tr02.append(treatment02)
        treatment03 = treatment_c(line, args)
        line_tr03.append(treatment03)
        treatment04 = treatment_d(line, args)
        line_tr04.append(treatment04)


for file in folder:
    text_lines = read_file_into_list(file_path)
    chunk_size=len(text_lines)/6
    divided=[]

    divided.append(text_lines[0:chunk_size])
    divided.append(text_lines[chunk_size:2*chunk_size])
    divided.append(text_lines[2*chunk_size:3*chunk_size])
    divided.append(text_lines[3*chunk_size:4*chunk_size])
    divided.append(text_lines[4*chunk_size:5*chunk_size])
    divided.append(text_lines[5*chunk_size:6*chunk_size])

    lines=[]
    p = Pool(6)
    lines.extend(p.map(extract(text_lines),divided))
    p.close()
    p.join()
    p.terminate()

    line_tr01=lines[0]
    with open(pkl_filename, 'wb') as f:
        pickle.dump(line_tr01, f)
    line_tr02=lines[1]
    with open(pkl_filename, 'wb') as f:
        pickle.dump(line_tr02, f)
    line_tr03=lines[2]
    with open(pkl_filename, 'wb') as f:
        pickle.dump(line_tr03, f)
    line_tr04=lines[3]
    with open(pkl_filename, 'wb') as f:
        pickle.dump(line_tr04, f)

任何有关如何停止覆盖文件的提示任何帮助都将受到欢迎。预先感谢

Answer 1

因此，问题在于，当您将内容划分为多个池时，您将不再具有当前正在（滥用）使用的公共全局名称空间。因此，让我们重写一下以正确传递信息。

def extract(text_lines):
    treatments = dict(tr01=[], tr02=[], tr03=[], tr04=[])
    for line in text_lines:
        treatments['tr01'].append(treatment_a(line, args))
        treatments['tr02'].append(treatment_b(line, args))
        treatments['tr03'].append(treatment_c(line, args))
        treatments['tr04'].append(treatment_d(line, args))
    return treatments

def line_gen(lines, chunk_size=1):
    for i in range(0, len(lines), chunk_size):
        yield lines[i:i + chunk_size]

for file in folder:
    text_lines = read_file_into_list(file_path)
    treatments = dict(tr01=[], tr02=[], tr03=[], tr04=[])
    p = Pool(6)
    for treat_data in p.imap(extract, line_gen(text_lines, chunk_size=int(len(text_lines)/6))):
        for tr, data in treat_data.items():
            treatments[tr].extend(data)

    # Do something with all your data in the treatments dict

这应该将所有数据堆积到名为treatments的字典中，因为它从运行extract的子进程中返回数据，然后您可以通过任何方式写出数据你想。

Python麻烦保存多个池处理的文件

1 个答案: