異步 imap 獲取郵件 python3

Question

我正在使用 asyncio 模塊進行測試，但是我需要提示/建議如何以異步方式獲取大型電子郵件。

我有一個包含郵件帳戶用戶名和密碼的列表。

data = [
    {'usern': 'foo@bar.de', 'passw': 'x'},
    {'usern': 'foo2@bar.de', 'passw': 'y'},
    {'usern': 'foo3@bar.de', 'passw': 'z'} (...)
]

我想過：

loop = asyncio.get_event_loop()
loop.run_until_complete(asyncio.wait([get_attachment(d) for d in data]))
loop.close()

但是，較長的部分是下載電子郵件附件。

電子郵件：

@asyncio.coroutine
def get_attachment(d):
    username = d['usern']
    password = d['passw']

    connection = imaplib.IMAP4_SSL('imap.bar.de')
    connection.login(username, password)
    connection.select()

    # list all available mails
    typ, data = connection.search(None, 'ALL')

    for num in data[0].split():
        # fetching each mail
        typ, data = connection.fetch(num, '(RFC822)')
        raw_string = data[0][1].decode('utf-8')
        msg = email.message_from_string(raw_string)
        for part in msg.walk():
            if part.get_content_maintype() == 'multipart':
                continue

            if part.get('Content-Disposition') is None:
                continue

            if part.get_filename():
                body = part.get_payload(decode=True)
                # do something with the body, async?

    connection.close()
    connection.logout()

如何以異步方式處理所有（下載附件）郵件？

Answer 1

如果您沒有基於異步 I/O 的 imap 庫，您可以使用concurrent.futures.ThreadPoolExecutor在線程中執行 I/O。 Python 將在 I/O 期間釋放 GIL，因此您將獲得真正的並發性：

def init_connection(d):    
    username = d['usern']
    password = d['passw']

    connection = imaplib.IMAP4_SSL('imap.bar.de')
    connection.login(username, password)
    connection.select()
    return connection

local = threading.local() # We use this to get a different connection per thread
def do_fetch(num, d, rfc):
    try:
        connection = local.connection
    except AttributeError:
        connnection = local.connection = init_connection(d)
    return connnection.fetch(num, rfc)

@asyncio.coroutine
def get_attachment(d, pool):
    connection = init_connection(d)    
    # list all available mails
    typ, data = connection.search(None, 'ALL')

    # Kick off asynchronous tasks for all the fetches
    loop = asyncio.get_event_loop()
    futs = [asyncio.create_task(loop.run_in_executor(pool, do_fetch, num, d, '(RFC822)'))
                for num in data[0].split()]

    # Process each fetch as it completes
    for fut in asyncio.as_completed(futs):
        typ, data = yield from fut
        raw_string = data[0][1].decode('utf-8')
        msg = email.message_from_string(raw_string)
        for part in msg.walk():
            if part.get_content_maintype() == 'multipart':
                continue

            if part.get('Content-Disposition') is None:
                continue

            if part.get_filename():
                body = part.get_payload(decode=True)
                # do something with the body, async?

    connection.close()
    connection.logout()    


loop = asyncio.get_event_loop()
pool = ThreadPoolExecutor(max_workers=5)  # You can probably increase max_workers, because the threads are almost exclusively doing I/O.
loop.run_until_complete(asyncio.wait([get_attachment(d, pool) for d in data]))
loop.close()

這不如真正的基於異步 I/O 的解決方案好，因為您仍然有創建線程的開銷，這限制了可擴展性並增加了額外的內存開銷。 由於所有代碼都包裝了實際的 I/O 調用，因此您也確實會遇到一些GIL 減慢。 盡管如此，如果您處理的郵件少於數千封，它應該仍然可以正常運行。

我們使用run_in_executor使用ThreadPoolExecutor作為ASYNCIO事件循環的一部分， asyncio.async包裹在返回的協程對象asyncio.Future ，並as_completed迭代通過的順序是期貨，就完成了。

編輯：

看來imaplib不是線程安全的。 我已經編輯了我的答案以通過threading.local使用線程本地存儲，這允許我們為每個線程創建一個連接對象，可以在線程的整個生命周期中重復使用（意味着您只創建num_workers連接對象，而不是每次fetch建立一個新連接）。

Answer 2

我有同樣的需求：使用 python 3 完全異步獲取電子郵件。 如果這里的其他人感興趣，我在這里推送了一個 asyncio IMAP 庫： https : //github.com/bamthomas/aioimaplib

你可以這樣使用它：

import asyncio
from aioimaplib import aioimaplib

@asyncio.coroutine
def wait_for_new_message(host, user, password):
    imap_client = aioimaplib.IMAP4(host=host)
    yield from imap_client.wait_hello_from_server()

    yield from imap_client.login(user, password)
    yield from imap_client.select()

    asyncio.async(imap_client.idle())
    id = 0
    while True:
        msg = yield from imap_client.wait_server_push()
        print('--> received from server: %s' % msg)
        if 'EXISTS' in msg:
            id = msg.split()[0]
            imap_client.idle_done()
            break

    result, data = yield from imap_client.fetch(id, '(RFC822)')
    email_message = email.message_from_bytes(data[0])

    attachments = []
    body = ''
    for part in email_message.walk():
        if part.get_content_maintype() == 'multipart':
            continue
        if part.get_content_maintype() == 'text' and 'attachment' not in part.get('Content-Disposition', ''):
            body = part.get_payload(decode=True).decode(part.get_param('charset', 'ascii')).strip()
        else:
            attachments.append(
                {'type': part.get_content_type(), 'filename': part.get_filename(), 'size': len(part.as_bytes())})

    print('attachments : %s' % attachments)
    print('body : %s' % body)
    yield from imap_client.logout()



if __name__ == '__main__':
    loop = asyncio.get_event_loop()
    loop.run_until_complete(wait_for_new_message('my.imap.server', 'user', 'pass'))

帶有附件的大型電子郵件也可以使用 asyncio 下載。