Imported Upstream version 2007.02.18

[youtubedl] / youtube-dl

diff --git a/youtube-dl b/youtube-dl
index dbd40b7bbcde8a7baccbf956eb37d327656ab771..6d53fa3c4d7acbe6c0a49ba25751853ad03943d6 100755 (executable)
--- a/youtube-dl
+++ b/youtube-dl
@@ -25,32 +25,35 @@
 # sale, use or other dealings in this Software without prior written
 # authorization.
 #
 # sale, use or other dealings in this Software without prior written
 # authorization.
 #

-import sys
+import getpass
+import httplib
+import netrc

 import optparse
 import optparse

-import urllib2
+import os

 import re
 import re

-
-# Exit status constants
-exit_failure = 1
-exit_success = 0
+import socket
+import string
+import sys
+import time
+import urllib2

 
 # Global constants
 const_video_url_str = 'http://www.youtube.com/watch?v=%s'
 
 # Global constants
 const_video_url_str = 'http://www.youtube.com/watch?v=%s'

-const_video_url_re = re.compile(r'http://(?:www\.)?youtube\.com/(?:v/|(?:watch)?\?v=)([^&]+).*')
-const_login_url_str = 'http://www.youtube.com/signup?next=/'
-const_login_post_str = 'current_form=loginForm&next=%%2F&username=%s&password=%s&action_login=Log+In'
+const_video_url_re = re.compile(r'(?:http://)?(?:www\d*\.)?youtube\.com/(?:v/|(?:watch(?:\.php)?)?\?v=)([^&]+).*')
+const_login_url_str = 'http://www.youtube.com/login?next=/watch%%3Fv%%3D%s'
+const_login_post_str = 'current_form=loginForm&next=%%2Fwatch%%3Fv%%3D%s&username=%s&password=%s&action_login=Log+In'

 const_age_url_str = 'http://www.youtube.com/verify_age?next_url=/watch%%3Fv%%3D%s'
 const_age_post_str = 'next_url=%%2Fwatch%%3Fv%%3D%s&action_confirm=Confirm'
 const_age_url_str = 'http://www.youtube.com/verify_age?next_url=/watch%%3Fv%%3D%s'
 const_age_post_str = 'next_url=%%2Fwatch%%3Fv%%3D%s&action_confirm=Confirm'

-const_video_url_params_re = re.compile(r'player2\.swf\?video_id=([^&]+)&.*t=([^&]+)&', re.M)
-const_video_url_real_str = 'http://www.youtube.com/get_video?video_id=%s&t=%s'
+const_video_url_params_re = re.compile(r'player2\.swf\?([^"]+)"', re.M)
+const_video_url_real_str = 'http://www.youtube.com/get_video?%s'
+const_video_title_re = re.compile(r'<title>YouTube - ([^<]*)</title>', re.M | re.I)

 const_1k = 1024
 const_1k = 1024

-const_block_size = 10 * const_1k
+const_initial_block_size = 10 * const_1k

 
 # Print error message, followed by standard advice information, and then exit
 def error_advice_exit(error_text):
 
 # Print error message, followed by standard advice information, and then exit
 def error_advice_exit(error_text):

-       global exit_failure

        sys.stderr.write('Error: %s.\n' % error_text)
        sys.stderr.write('Error: %s.\n' % error_text)

-       sys.stderr.write('Try again several times. It may be a temporal problem.\n')
+       sys.stderr.write('Try again several times. It may be a temporary problem.\n')

        sys.stderr.write('Other typical problems:\n\n')
        sys.stderr.write('\tVideo no longer exists.\n')
        sys.stderr.write('\tVideo requires age confirmation but you did not provide an account.\n')
        sys.stderr.write('Other typical problems:\n\n')
        sys.stderr.write('\tVideo no longer exists.\n')
        sys.stderr.write('\tVideo requires age confirmation but you did not provide an account.\n')


@@ -59,16 +62,17 @@ def error_advice_exit(error_text):
        sys.stderr.write('\tYouTube changed their system, and the program no longer works.\n')
        sys.stderr.write('\nTry to confirm you are able to view the video using a web browser.\n')
        sys.stderr.write('Use the same video URL and account information, if needed, with this program.\n')
        sys.stderr.write('\tYouTube changed their system, and the program no longer works.\n')
        sys.stderr.write('\nTry to confirm you are able to view the video using a web browser.\n')
        sys.stderr.write('Use the same video URL and account information, if needed, with this program.\n')

+       sys.stderr.write('When using a proxy, make sure http_proxy has http://host:port format.\n')

        sys.stderr.write('Try again several times and contact me if the problem persists.\n')
        sys.stderr.write('Try again several times and contact me if the problem persists.\n')

-       sys.exit(exit_failure)
+       sys.exit('\n')

 
 # Wrapper to create custom requests with typical headers
 def request_create(url, data=None):
        retval = urllib2.Request(url)
 
 # Wrapper to create custom requests with typical headers
 def request_create(url, data=None):
        retval = urllib2.Request(url)

-       if not data is None:
+       if data is not None:

                retval.add_data(data)
        # Try to mimic Firefox, at least a little bit
                retval.add_data(data)
        # Try to mimic Firefox, at least a little bit

-       retval.add_header('User-Agent', 'Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.0.6) Gecko/20060728 Firefox/1.5.0.6')
+       retval.add_header('User-Agent', 'Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1) Gecko/20061010 Firefox/2.0')

        retval.add_header('Accept-Charset', 'ISO-8859-1,utf-8;q=0.7,*;q=0.7')
        retval.add_header('Accept', 'text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5')
        retval.add_header('Accept-Language', 'en-us,en;q=0.5')
        retval.add_header('Accept-Charset', 'ISO-8859-1,utf-8;q=0.7,*;q=0.7')
        retval.add_header('Accept', 'text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5')
        retval.add_header('Accept-Language', 'en-us,en;q=0.5')


@@ -92,9 +96,64 @@ def cond_print(str):
                sys.stdout.write(str)
                sys.stdout.flush()
 
                sys.stdout.write(str)
                sys.stdout.flush()
 

+# Title string normalization
+def title_string_norm(title):
+       title = ''.join((x in string.ascii_letters or x in string.digits) and x or ' ' for x in title)
+       title = '_'.join(title.split())
+       title = title.lower()
+       return title
+
+# Generic download step
+def download_step(return_data_flag, step_title, step_error, url, post_data=None):
+       try:
+               cond_print('%s... ' % step_title)
+               data = perform_request(url, post_data).read()
+               cond_print('done.\n')
+               if return_data_flag:
+                       return data
+               return None
+
+       except (urllib2.URLError, ValueError, httplib.HTTPException, TypeError, socket.error):
+               cond_print('failed.\n')
+               error_advice_exit(step_error)
+
+       except KeyboardInterrupt:
+               sys.exit('\n')
+
+# Generic extract step
+def extract_step(step_title, step_error, regexp, data):
+       try:
+               cond_print('%s... ' % step_title)
+               match = regexp.search(data)
+               
+               if match is None:
+                       cond_print('failed.\n')
+                       error_advice_exit(step_error)
+               
+               extracted_data = match.group(1)
+               cond_print('done.\n')
+               return extracted_data
+       
+       except KeyboardInterrupt:
+               sys.exit('\n')
+
+# Calculate new block size based on previous block size
+def new_block_size(before, after, bytes):
+       new_min = max(bytes / 2, 1)
+       new_max = max(bytes * 2, 1)
+       dif = after - before
+       if dif < 0.0001:
+               return new_max
+       rate = int(bytes / dif)
+       if rate > new_max:
+               return new_max
+       if rate < new_min:
+               return new_min
+       return rate
+

 # Create the command line options parser and parse command line
 cmdl_usage = 'usage: %prog [options] video_url'
 # Create the command line options parser and parse command line
 cmdl_usage = 'usage: %prog [options] video_url'

-cmdl_version = '2006.08.28'
+cmdl_version = '2007.02.18'

 cmdl_parser = optparse.OptionParser(usage=cmdl_usage, version=cmdl_version, conflict_handler='resolve')
 cmdl_parser.add_option('-h', '--help', action='help', help='print this help text and exit')
 cmdl_parser.add_option('-v', '--version', action='version', help='print program version and exit')
 cmdl_parser = optparse.OptionParser(usage=cmdl_usage, version=cmdl_version, conflict_handler='resolve')
 cmdl_parser.add_option('-h', '--help', action='help', help='print this help text and exit')
 cmdl_parser.add_option('-v', '--version', action='version', help='print program version and exit')


@@ -103,25 +162,71 @@ cmdl_parser.add_option('-p', '--password', dest='password', metavar='PASSWORD',
 cmdl_parser.add_option('-o', '--output', dest='outfile', metavar='FILE', help='output video file name')
 cmdl_parser.add_option('-q', '--quiet', action='store_true', dest='quiet', help='activates quiet mode')
 cmdl_parser.add_option('-s', '--simulate', action='store_true', dest='simulate', help='do not download video')
 cmdl_parser.add_option('-o', '--output', dest='outfile', metavar='FILE', help='output video file name')
 cmdl_parser.add_option('-q', '--quiet', action='store_true', dest='quiet', help='activates quiet mode')
 cmdl_parser.add_option('-s', '--simulate', action='store_true', dest='simulate', help='do not download video')

+cmdl_parser.add_option('-t', '--title', action='store_true', dest='use_title', help='use title in file name')
+cmdl_parser.add_option('-n', '--netrc', action='store_true', dest='use_netrc', help='use .netrc authentication data')

 (cmdl_opts, cmdl_args) = cmdl_parser.parse_args()
 
 # Get video URL
 if len(cmdl_args) != 1:
        cmdl_parser.print_help()
 (cmdl_opts, cmdl_args) = cmdl_parser.parse_args()
 
 # Get video URL
 if len(cmdl_args) != 1:
        cmdl_parser.print_help()

-       sys.exit(exit_failure)
+       sys.exit('\n')

 video_url_cmdl = cmdl_args[0]
 
 # Verify video URL format and convert to "standard" format
 video_url_mo = const_video_url_re.match(video_url_cmdl)
 if video_url_mo is None:
 video_url_cmdl = cmdl_args[0]
 
 # Verify video URL format and convert to "standard" format
 video_url_mo = const_video_url_re.match(video_url_cmdl)
 if video_url_mo is None:

-       sys.exit('Error: URL does not seem to be a youtube video URL. If it is, report a bug.\n')
+       sys.exit('Error: URL does not seem to be a youtube video URL. If it is, report a bug.')

 video_url_id = video_url_mo.group(1)
 video_url = const_video_url_str % video_url_id
 
 # Check conflicting options
 video_url_id = video_url_mo.group(1)
 video_url = const_video_url_str % video_url_id
 
 # Check conflicting options

-if not cmdl_opts.outfile is None and cmdl_opts.simulate:
+if cmdl_opts.outfile is not None and cmdl_opts.simulate:

        sys.stderr.write('Warning: video file name given but will not be used.\n')
 
        sys.stderr.write('Warning: video file name given but will not be used.\n')
 

+if cmdl_opts.outfile is not None and cmdl_opts.use_title:
+       sys.exit('Error: using the video title conflicts with using a given file name.')
+
+if cmdl_opts.use_netrc and cmdl_opts.password is not None:
+       sys.exit('Error: using netrc conflicts with giving command line password.')
+
+# Incorrect option formatting
+if cmdl_opts.username is None and cmdl_opts.password is not None:
+       sys.exit('Error: password give but username is missing.')
+
+# Get account information if any
+account_username = None
+account_password = None
+
+if cmdl_opts.use_netrc:
+       try:
+               info = netrc.netrc().authenticators('youtube')
+               if info is None:
+                       sys.exit('Error: no authenticators for machine youtube.')
+               netrc_username = info[0]
+               netrc_password = info[2]
+       except IOError:
+               sys.exit('Error: unable to read .netrc file.')
+       except netrc.NetrcParseError:
+               sys.exit('Error: unable to parse .netrc file.')
+
+if cmdl_opts.password is not None:
+       account_username = cmdl_opts.username
+       account_password = cmdl_opts.password
+else:
+       if cmdl_opts.username is not None and cmdl_opts.use_netrc:
+               if cmdl_opts.username != netrc_username:
+                       sys.exit('Error: conflicting username from .netrc and command line options.')
+               account_username = cmdl_opts.username
+               account_password = netrc_password
+       elif cmdl_opts.username is not None:
+               account_username = cmdl_opts.username
+               account_password = getpass.getpass('Type YouTube password and press return: ')
+       elif cmdl_opts.use_netrc:
+               if len(netrc_username) == 0:
+                       sys.exit('Error: empty username in .netrc file.')
+               account_username = netrc_username
+               account_password = netrc_password
+

 # Get output file name 
 if cmdl_opts.outfile is None:
        video_filename = '%s.flv' % video_url_id
 # Get output file name 
 if cmdl_opts.outfile is None:
        video_filename = '%s.flv' % video_url_id


@@ -132,11 +237,6 @@ else:
 if not video_filename.lower().endswith('.flv'):
        sys.stderr.write('Warning: video file name does not end in .flv\n')
 
 if not video_filename.lower().endswith('.flv'):
        sys.stderr.write('Warning: video file name does not end in .flv\n')
 

-# Verify both or none present
-if ((cmdl_opts.username is None and not cmdl_opts.password is None) or
-    (not cmdl_opts.username is None and cmdl_opts.password is None)):
-       sys.exit('Error: both username and password must be given, or none.')
-

 # Test writable file
 if not cmdl_opts.simulate:
        try:
 # Test writable file
 if not cmdl_opts.simulate:
        try:


@@ -146,67 +246,30 @@ if not cmdl_opts.simulate:
        except (OSError, IOError):
                sys.exit('Error: unable to open %s for writing.' % video_filename)
 
        except (OSError, IOError):
                sys.exit('Error: unable to open %s for writing.' % video_filename)
 

-# Install cookie handler
+# Install cookie and proxy handlers
+urllib2.install_opener(urllib2.build_opener(urllib2.ProxyHandler()))

 urllib2.install_opener(urllib2.build_opener(urllib2.HTTPCookieProcessor()))
 
 urllib2.install_opener(urllib2.build_opener(urllib2.HTTPCookieProcessor()))
 

-# Login and confirm age if needed
-if not cmdl_opts.username is None:
-       try:
-               # Log in
-               cond_print('Logging in... ')
-               perform_request(const_login_url_str, const_login_post_str % (cmdl_opts.username, cmdl_opts.password)).read()
-               cond_print('done.\n')
-
-       except (urllib2.URLError, ValueError):
-               cond_print('failed.\n')
-               error_advice_exit('unable to login')
-
-       except KeyboardInterrupt:
-               sys.exit(exit_failure)
-
-       try:
-               # Get age confirmation cookie
-               cond_print('Confirming age... ')
-               perform_request(const_age_url_str % video_url_id, const_age_post_str % video_url_id).read()
-               cond_print('done.\n')
+# Log in and confirm age if needed
+if account_username is not None:
+       url = const_login_url_str % video_url_id
+       post = const_login_post_str % (video_url_id, account_username, account_password)
+       download_step(False, 'Logging in', 'unable to log in', url, post)

 
 

-       except (urllib2.URLError, ValueError):
-               cond_print('failed.\n')
-               error_advice_exit('unable to confirm age')
-
-       except KeyboardInterrupt:
-               sys.exit(exit_failure)
+       url = const_age_url_str % video_url_id
+       post = const_age_post_str % video_url_id
+       download_step(False, 'Confirming age', 'unable to confirm age', url, post)

 
 # Retrieve video webpage
 
 # Retrieve video webpage

-try:
-       cond_print('Retrieving video webpage... ')
-       video_webpage = perform_request(video_url).read()
+video_webpage = download_step(True, 'Retrieving video webpage', 'unable to retrieve video webpage', video_url)

 
 

-except (urllib2.URLError, ValueError):
-       cond_print('failed.\n')
-       error_advice_exit('unable to download video webpage')
-
-except KeyboardInterrupt:
-       sys.exit(exit_failure)
-
-cond_print('done.\n')
+# Extract video title if needed
+if cmdl_opts.use_title:
+       video_title = extract_step('Extracting video title', 'unable to extract video title', const_video_title_re, video_webpage)

 
 # Extract needed video URL parameters
 
 # Extract needed video URL parameters

-try:
-       cond_print('Extracting video URL parameters... ')
-       video_url_params_mo = const_video_url_params_re.search(video_webpage)
-       
-       if video_url_params_mo is None:
-               cond_print('failed.\n')
-               error_advice_exit('unable to extract URL parameters')
-       
-       video_real_id = video_url_params_mo.group(1)
-       video_t_param = video_url_params_mo.group(2)
-       video_url_real = const_video_url_real_str % (video_real_id, video_t_param)
-       cond_print('done.\n')
-
-except KeyboardInterrupt:
-       sys.exit(exit_failure)
+video_url_params = extract_step('Extracting video URL parameters', 'unable to extract URL parameters', const_video_url_params_re, video_webpage)
+video_url_real = const_video_url_real_str % video_url_params

 
 # Retrieve video data
 try:
 
 # Retrieve video data
 try:


@@ -217,30 +280,49 @@ try:
        if cmdl_opts.simulate:
                sys.exit()
 
        if cmdl_opts.simulate:
                sys.exit()
 

-       cond_print('Retrieving video data... ')

        video_file = open(video_filename, 'wb')
        try:
        video_file = open(video_filename, 'wb')
        try:

-               video_len_str = '%sk' % to_k(int(video_data.info()['Content-length']))
+               video_len_str = '%sk' % to_k(long(video_data.info()['Content-length']))

        except KeyError:
                video_len_str = '(unknown)'
 
        byte_counter = 0
        except KeyError:
                video_len_str = '(unknown)'
 
        byte_counter = 0

-       video_block = video_data.read(const_block_size)
-       while len(video_block) != 0:
-               byte_counter += len(video_block)
-               video_file.write(video_block)
+       block_size = const_initial_block_size
+       while True:

                cond_print('\rRetrieving video data... %sk of %s ' % (to_k(byte_counter), video_len_str))
                cond_print('\rRetrieving video data... %sk of %s ' % (to_k(byte_counter), video_len_str))

-               video_block = video_data.read(const_block_size)
+               before = time.time()
+               video_block = video_data.read(block_size)
+               after = time.time()
+               dl_bytes = len(video_block)
+               if dl_bytes == 0:
+                       break
+               byte_counter += dl_bytes
+               video_file.write(video_block)
+               block_size = new_block_size(before, after, dl_bytes)

 
        video_file.close()
 
        video_file.close()

+       cond_print('done.\n')
+       cond_print('Video data saved to %s\n' % video_filename)

 
 

-except (urllib2.URLError, ValueError):
+except (urllib2.URLError, ValueError, httplib.HTTPException, TypeError, socket.error):

        cond_print('failed.\n')
        error_advice_exit('unable to download video data')
 
 except KeyboardInterrupt:
        cond_print('failed.\n')
        error_advice_exit('unable to download video data')
 
 except KeyboardInterrupt:

-       sys.exit(exit_failure)
+       sys.exit('\n')
+
+# Rename video file if needed
+if cmdl_opts.use_title:
+       try:
+               final_filename = '%s-%s.flv' % (title_string_norm(video_title), video_url_id)
+               os.rename(video_filename, final_filename)
+               cond_print('Video file renamed to %s\n' % final_filename)
+       
+       except OSError:
+               sys.stderr.write('Warning: unable to rename file.\n')
+
+       except KeyboardInterrupt:
+               sys.exit('\n')

 
 

-cond_print('done.\n')
-cond_print('Video data saved to %s\n' % video_filename)
+# Finish

 sys.exit()
 sys.exit()