Mercurial > public > sg101

annotate core/management/commands/ssl_images.py @ 866:98adae6e13a1

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
More WIP on ssl_images command.
author Brian Neal <bgneal@gmail.com>
date Thu, 04 Dec 2014 20:32:43 -0600
parents 0ffdb434d2dd
children 64a5acb83937
rev   line source
bgneal@859 1 """
bgneal@859 2 ssl_images is a custom manage.py command to convert forum post and comment
bgneal@859 3 images to https. It does this by rewriting the markup:
bgneal@859 4 - Images with src = http://surfguitar101.com/something are rewritten to be
bgneal@859 5 /something.
bgneal@859 6 - Non SG101 images that use http: are downloaded, resized, and uploaded to
bgneal@859 7 an S3 bucket. The src attribute is replaced with the new S3 URL.
bgneal@859 8 """
bgneal@859 9 import logging
bgneal@859 10 from optparse import make_option
bgneal@859 11 import os.path
bgneal@863 12 import re
bgneal@863 13 import signal
bgneal@859 14
bgneal@859 15 from django.core.management.base import NoArgsCommand, CommandError
bgneal@859 16 from django.conf import settings
bgneal@863 17 import markdown.inlinepatterns
bgneal@859 18
bgneal@860 19 from comments.models import Comment
bgneal@860 20 from forums.models import Post
bgneal@860 21
bgneal@860 22
bgneal@859 23 LOGFILE = os.path.join(settings.PROJECT_PATH, 'logs', 'ssl_images.log')
bgneal@859 24 logger = logging.getLogger(__name__)
bgneal@859 25
bgneal@863 26 IMAGE_LINK_RE = re.compile(markdown.inlinepatterns.IMAGE_LINK_RE)
bgneal@863 27 IMAGE_REF_RE = re.compile(markdown.inlinepatterns.IMAGE_REFERENCE_RE)
bgneal@863 28
bgneal@866 29 MODEL_CHOICES = ['comments', 'posts']
bgneal@866 30
bgneal@863 31 quit_flag = False
bgneal@863 32
bgneal@863 33
bgneal@863 34 def signal_handler(signum, frame):
bgneal@863 35 """SIGINT signal handler"""
bgneal@863 36 global quit_flag
bgneal@863 37 quit_flag = True
bgneal@863 38
bgneal@859 39
bgneal@859 40 def _setup_logging():
bgneal@859 41 logger.setLevel(logging.DEBUG)
bgneal@859 42 logger.propagate = False
bgneal@859 43 handler = logging.FileHandler(filename=LOGFILE, encoding='utf-8')
bgneal@859 44 formatter = logging.Formatter('%(asctime)s %(levelname)s %(message)s')
bgneal@859 45 handler.setFormatter(formatter)
bgneal@859 46 logger.addHandler(handler)
bgneal@859 47
bgneal@859 48
bgneal@866 49 def replace_image_markup(match):
bgneal@866 50 return match.group(0)
bgneal@860 51
bgneal@860 52
bgneal@866 53 def process_post(text):
bgneal@863 54 """Process the post object:
bgneal@863 55
bgneal@863 56 A regex substitution is run on the post's text field. This fixes up image
bgneal@863 57 links, getting rid of plain old http sources; either converting to https
bgneal@863 58 or relative style links (if the link is to SG101).
bgneal@863 59
bgneal@863 60 We also do a search for Markdown image reference markup. We aren't expecting
bgneal@863 61 these, but we will log something if we see any.
bgneal@863 62
bgneal@863 63 """
bgneal@866 64 return IMAGE_LINK_RE.sub(replace_image_markup, text)
bgneal@863 65
bgneal@863 66
bgneal@859 67 class Command(NoArgsCommand):
bgneal@859 68 help = "Rewrite forum posts and comments to not use http for images"
bgneal@859 69 option_list = NoArgsCommand.option_list + (
bgneal@866 70 make_option('-m', '--model',
bgneal@866 71 choices=MODEL_CHOICES,
bgneal@866 72 help="which model to update; must be one of {{{}}}".format(
bgneal@866 73 ', '.join(MODEL_CHOICES))),
bgneal@860 74 make_option('-i', '--i',
bgneal@859 75 type='int',
bgneal@863 76 help="optional first slice index; the i in [i:j]"),
bgneal@860 77 make_option('-j', '--j',
bgneal@859 78 type='int',
bgneal@863 79 help="optional second slice index; the j in [i:j]"),
bgneal@859 80 )
bgneal@859 81
bgneal@859 82 def handle_noargs(self, **options):
bgneal@859 83 _setup_logging()
bgneal@860 84 logger.info("Starting; arguments received: %s", options)
bgneal@859 85
bgneal@866 86 if options['model'] not in MODEL_CHOICES:
bgneal@866 87 raise CommandError('Please choose a --model option')
bgneal@859 88
bgneal@866 89 if options['model'] == 'comments':
bgneal@860 90 qs = Comment.objects.all()
bgneal@866 91 text_attr = 'comment'
bgneal@860 92 else:
bgneal@860 93 qs = Post.objects.all()
bgneal@866 94 text_attr = 'body'
bgneal@860 95
bgneal@860 96 i, j = options['i'], options['j']
bgneal@860 97
bgneal@860 98 if i is not None and i < 0:
bgneal@860 99 raise CommandError("-i must be >= 0")
bgneal@860 100 if j is not None and j < 0:
bgneal@860 101 raise CommandError("-j must be >= 0")
bgneal@860 102 if j is not None and i is not None and j <= i:
bgneal@860 103 raise CommandError("-j must be > -i")
bgneal@860 104
bgneal@860 105 if i is not None and j is not None:
bgneal@860 106 qs = qs[i:j]
bgneal@860 107 elif i is not None and j is None:
bgneal@860 108 qs = qs[i:]
bgneal@860 109 elif i is None and j is not None:
bgneal@860 110 qs = qs[:j]
bgneal@860 111
bgneal@863 112 # Install signal handler for ctrl-c
bgneal@863 113 signal.signal(signal.SIGINT, signal_handler)
bgneal@863 114
bgneal@860 115 s = []
bgneal@860 116 for model in qs.iterator():
bgneal@863 117 if quit_flag:
bgneal@863 118 logger.warning("SIGINT received, exiting")
bgneal@866 119 txt = getattr(model, text_attr)
bgneal@866 120 new_txt = process_post(txt)
bgneal@866 121 s.append(new_txt)
bgneal@860 122
bgneal@860 123 import pprint
bgneal@860 124 pprint.pprint(s)