Python HTTP download with resume and optional MD5 hash checking

download.py

import os.path
import requests
import shutil
import hashlib
import logging

def validate_file(file_path, hash):
    """
    Validates a file against an MD5 hash value
 
    :param file_path: path to the file for hash validation
    :type file_path:  string
    :param hash:      expected hash value of the file
    :type hash:       string -- MD5 hash value
    """
    m = hashlib.md5()
    with open(file_path, 'rb') as f:
        while True:
            chunk = f.read(1000 * 1000) # 1MB
            if not chunk:
                break
            m.update(chunk)
    return m.hexdigest() == hash

def download_with_resume(url, file_path, hash=None, timeout=10):
    """
    Performs a HTTP(S) download that can be restarted if prematurely terminated.
    The HTTP server must support byte ranges.
 
    :param file_path: the path to the file to write to disk
    :type file_path:  string
    :param hash: hash value for file validation
    :type hash:  string (MD5 hash value)
    """
     # don't download if the file exists
    if os.path.exists(file_path):
        return
    block_size = 1000 * 1000 # 1MB
    tmp_file_path = file_path + '.part'
    first_byte = os.path.getsize(tmp_file_path) if os.path.exists(tmp_file_path) else 0
    file_mode = 'ab' if first_byte else 'wb'
    logging.debug('Starting download at %.1fMB' % (first_byte / 1e6))
    file_size = -1
    try:
        file_size = int(requests.head(url).headers['Content-length'])
        logging.debug('File size is %s' % file_size)
        headers = {"Range": "bytes=%s-" % first_byte}
        r = requests.get(url, headers=headers, stream=True, timeout=timeout)
        with open(tmp_file_path, file_mode) as f:
            for chunk in r.iter_content(chunk_size=block_size): 
                if chunk: # filter out keep-alive new chunks
                    f.write(chunk)
    except IOError as e:
        logging.debug('IO Error - %s' % e)
    finally:
        # rename the temp download file to the correct name if fully downloaded
        if file_size == os.path.getsize(tmp_file_path):
            # if there's a hash value, validate the file
            if hash and not validate_file(tmp_file_path, hash):
                raise Exception('Error validating the file against its MD5 hash')
            shutil.move(tmp_file_path, file_path)
        elif file_size == -1:
            raise Exception('Error getting Content-Length from server: %s' % url)

Modified to use 'requests' instead of 'urllib2'.
Significantly increases transfer speed by streaming the file in chunks rather than individual requests for each chunk.

I'm having an issue:

First: no timeout, please change this piece of code so the timeout is really considered:

#r = requests.get(url, headers=headers, stream=True)
r = requests.get(url, headers=headers, stream=True, timeout=timeout)

Second: please try downloading a file of 50mb (or higher)

• start downloading the file
• when you see the '.part' file is created, unplug the network cable
• wait for the timeout to cancel the process
• plug the network cable
• restart downloading the file
• wait for completion
• the file won't be renamed, as the file will be appended to the '.part' and let's say, we had download 20mb, + 50mb the resulting file will be 70mb, maybe there's a problem in range calculation but I don't know

PS: I'm trying to download a file hosted on Azure Blob

Oh duh... will add timeout param. Been a while since I tested this.
I use it in an addon for kodi and haven't noticed any issues. I will try to reproduce your issue to see what is going on.
Thanks for the feedback.