bgneal@859: """
bgneal@859: ssl_images is a custom manage.py command to convert forum post and comment
bgneal@859: images to https. It does this by rewriting the markup:
bgneal@859:     - Images with src = http://surfguitar101.com/something are rewritten to be
bgneal@859:       /something.
bgneal@859:     - Non SG101 images that use http: are downloaded, resized, and uploaded to
bgneal@859:       an S3 bucket. The src attribute is replaced with the new S3 URL.
bgneal@859: """
bgneal@859: import logging
bgneal@859: from optparse import make_option
bgneal@859: import os.path
bgneal@863: import re
bgneal@863: import signal
bgneal@868: import urlparse
bgneal@859: 
bgneal@859: from django.core.management.base import NoArgsCommand, CommandError
bgneal@859: from django.conf import settings
bgneal@863: import markdown.inlinepatterns
bgneal@859: 
bgneal@860: from comments.models import Comment
bgneal@860: from forums.models import Post
bgneal@860: 
bgneal@860: 
bgneal@859: LOGFILE = os.path.join(settings.PROJECT_PATH, 'logs', 'ssl_images.log')
bgneal@859: logger = logging.getLogger(__name__)
bgneal@859: 
bgneal@870: IMAGE_LINK_RE = re.compile(markdown.inlinepatterns.IMAGE_LINK_RE, re.UNICODE)
bgneal@870: IMAGE_REF_RE = re.compile(markdown.inlinepatterns.IMAGE_REFERENCE_RE, re.UNICODE)
bgneal@863: 
bgneal@868: SG101_HOSTS = set(['www.surfguitar101.com', 'surfguitar101.com'])
bgneal@866: MODEL_CHOICES = ['comments', 'posts']
bgneal@866: 
bgneal@863: quit_flag = False
bgneal@863: 
bgneal@863: 
bgneal@863: def signal_handler(signum, frame):
bgneal@863:     """SIGINT signal handler"""
bgneal@863:     global quit_flag
bgneal@863:     quit_flag = True
bgneal@863: 
bgneal@859: 
bgneal@859: def _setup_logging():
bgneal@859:     logger.setLevel(logging.DEBUG)
bgneal@859:     logger.propagate = False
bgneal@859:     handler = logging.FileHandler(filename=LOGFILE, encoding='utf-8')
bgneal@859:     formatter = logging.Formatter('%(asctime)s %(levelname)s %(message)s')
bgneal@859:     handler.setFormatter(formatter)
bgneal@859:     logger.addHandler(handler)
bgneal@859: 
bgneal@859: 
bgneal@868: def save_image_to_cloud(src):
bgneal@868:     # TODO
bgneal@868:     return src
bgneal@868: 
bgneal@868: 
bgneal@866: def replace_image_markup(match):
bgneal@870:     src_parts = match.group(8).split()
bgneal@868:     if src_parts:
bgneal@868:         src = src_parts[0]
bgneal@868:         if src[0] == "<" and src[-1] == ">":
bgneal@868:             src = src[1:-1]
bgneal@868:     else:
bgneal@868:         src = ''
bgneal@868: 
bgneal@868:     title = ''
bgneal@868:     if len(src_parts) > 1:
bgneal@868:         title = " ".join(src_parts[1:])
bgneal@870:     alt = match.group(1)
bgneal@868: 
bgneal@868:     new_src = ''
bgneal@868:     if src:
bgneal@868:         r = urlparse.urlparse(src)
bgneal@868:         if r.scheme == 'http':
bgneal@868:             if r.hostname in SG101_HOSTS:
bgneal@868:                 new_src = r.path    # convert to relative path
bgneal@868:             else:
bgneal@868:                 new_src = save_image_to_cloud(src)
bgneal@868:         elif r.scheme == 'https':
bgneal@868:             new_src = src       # already https, accept it as-is
bgneal@868: 
bgneal@868:     if new_src:
bgneal@868:         if title:
bgneal@868:             s = u'![{alt}]({src} "{title}")'.format(alt=alt, src=new_src, title=title)
bgneal@868:         else:
bgneal@868:             s = u'![{alt}]({src})'.format(alt=alt, src=new_src)
bgneal@868:     else:
bgneal@868:         # something's messed up, convert to a link using original src
bgneal@868:         s = u'[{alt}]({src})'.format(alt=alt, src=src)
bgneal@868: 
bgneal@868:     return s
bgneal@860: 
bgneal@860: 
bgneal@866: def process_post(text):
bgneal@863:     """Process the post object:
bgneal@863: 
bgneal@863:     A regex substitution is run on the post's text field. This fixes up image
bgneal@863:     links, getting rid of plain old http sources; either converting to https
bgneal@863:     or relative style links (if the link is to SG101).
bgneal@863: 
bgneal@863:     We also do a search for Markdown image reference markup. We aren't expecting
bgneal@863:     these, but we will log something if we see any.
bgneal@863: 
bgneal@863:     """
bgneal@866:     return IMAGE_LINK_RE.sub(replace_image_markup, text)
bgneal@863: 
bgneal@863: 
bgneal@859: class Command(NoArgsCommand):
bgneal@859:     help = "Rewrite forum posts and comments to not use http for images"
bgneal@859:     option_list = NoArgsCommand.option_list + (
bgneal@866:             make_option('-m', '--model',
bgneal@866:                 choices=MODEL_CHOICES,
bgneal@866:                 help="which model to update; must be one of {{{}}}".format(
bgneal@866:                                                     ', '.join(MODEL_CHOICES))),
bgneal@860:             make_option('-i', '--i',
bgneal@859:                 type='int',
bgneal@863:                 help="optional first slice index; the i in [i:j]"),
bgneal@860:             make_option('-j', '--j',
bgneal@859:                 type='int',
bgneal@863:                 help="optional second slice index; the j in [i:j]"),
bgneal@859:             )
bgneal@859: 
bgneal@859:     def handle_noargs(self, **options):
bgneal@859:         _setup_logging()
bgneal@860:         logger.info("Starting; arguments received: %s", options)
bgneal@859: 
bgneal@866:         if options['model'] not in MODEL_CHOICES:
bgneal@866:             raise CommandError('Please choose a --model option')
bgneal@859: 
bgneal@866:         if options['model'] == 'comments':
bgneal@860:             qs = Comment.objects.all()
bgneal@866:             text_attr = 'comment'
bgneal@860:         else:
bgneal@860:             qs = Post.objects.all()
bgneal@866:             text_attr = 'body'
bgneal@860: 
bgneal@860:         i, j = options['i'], options['j']
bgneal@860: 
bgneal@860:         if i is not None and i < 0:
bgneal@860:             raise CommandError("-i must be >= 0")
bgneal@860:         if j is not None and j < 0:
bgneal@860:             raise CommandError("-j must be >= 0")
bgneal@860:         if j is not None and i is not None and j <= i:
bgneal@860:             raise CommandError("-j must be > -i")
bgneal@860: 
bgneal@860:         if i is not None and j is not None:
bgneal@860:             qs = qs[i:j]
bgneal@860:         elif i is not None and j is None:
bgneal@860:             qs = qs[i:]
bgneal@860:         elif i is None and j is not None:
bgneal@860:             qs = qs[:j]
bgneal@860: 
bgneal@863:         # Install signal handler for ctrl-c
bgneal@863:         signal.signal(signal.SIGINT, signal_handler)
bgneal@863: 
bgneal@860:         s = []
bgneal@860:         for model in qs.iterator():
bgneal@863:             if quit_flag:
bgneal@863:                 logger.warning("SIGINT received, exiting")
bgneal@866:             txt = getattr(model, text_attr)
bgneal@866:             new_txt = process_post(txt)
bgneal@866:             s.append(new_txt)
bgneal@860: 
bgneal@860:         import pprint
bgneal@860:         pprint.pprint(s)