Mercurial > public > sg101

annotate core/management/commands/ssl_images.py @ 888:deef1536a54a

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
See if file available at https: by doing HEAD request.
author Brian Neal <bgneal@gmail.com>
date Mon, 09 Feb 2015 20:25:13 -0600
parents 9a15f7c27526
children ae146e30d588
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@882 9 import base64
bgneal@888 10 import httplib
bgneal@859 11 import logging
bgneal@859 12 from optparse import make_option
bgneal@888 13 import os
bgneal@863 14 import re
bgneal@863 15 import signal
bgneal@881 16 import socket
bgneal@881 17 import urllib
bgneal@868 18 import urlparse
bgneal@881 19 import uuid
bgneal@859 20
bgneal@859 21 from django.core.management.base import NoArgsCommand, CommandError
bgneal@859 22 from django.conf import settings
bgneal@863 23 import markdown.inlinepatterns
bgneal@881 24 from PIL import Image
bgneal@859 25
bgneal@860 26 from comments.models import Comment
bgneal@860 27 from forums.models import Post
bgneal@881 28 from core.s3 import S3Bucket
bgneal@860 29
bgneal@860 30
bgneal@859 31 LOGFILE = os.path.join(settings.PROJECT_PATH, 'logs', 'ssl_images.log')
bgneal@859 32 logger = logging.getLogger(__name__)
bgneal@859 33
bgneal@871 34 IMAGE_LINK_RE = re.compile(markdown.inlinepatterns.IMAGE_LINK_RE,
bgneal@871 35 re.DOTALL | re.UNICODE)
bgneal@871 36 IMAGE_REF_RE = re.compile(markdown.inlinepatterns.IMAGE_REFERENCE_RE,
bgneal@871 37 re.DOTALL | re.UNICODE)
bgneal@863 38
bgneal@868 39 SG101_HOSTS = set(['www.surfguitar101.com', 'surfguitar101.com'])
bgneal@866 40 MODEL_CHOICES = ['comments', 'posts']
bgneal@866 41
bgneal@881 42 PHOTO_MAX_SIZE = (660, 720)
bgneal@881 43 PHOTO_BASE_URL = 'https://s3.amazonaws.com/'
bgneal@881 44 PHOTO_BUCKET_NAME = 'sg101.forum.photos'
bgneal@881 45
bgneal@863 46 quit_flag = False
bgneal@881 47 opener = None
bgneal@881 48 bucket = None
bgneal@881 49 url_cache = {}
bgneal@863 50
bgneal@863 51
bgneal@863 52 def signal_handler(signum, frame):
bgneal@863 53 """SIGINT signal handler"""
bgneal@863 54 global quit_flag
bgneal@863 55 quit_flag = True
bgneal@863 56
bgneal@859 57
bgneal@859 58 def _setup_logging():
bgneal@859 59 logger.setLevel(logging.DEBUG)
bgneal@859 60 logger.propagate = False
bgneal@859 61 handler = logging.FileHandler(filename=LOGFILE, encoding='utf-8')
bgneal@859 62 formatter = logging.Formatter('%(asctime)s %(levelname)s %(message)s')
bgneal@859 63 handler.setFormatter(formatter)
bgneal@859 64 logger.addHandler(handler)
bgneal@859 65
bgneal@859 66
bgneal@881 67 class ImageURLopener(urllib.FancyURLopener):
bgneal@881 68 """Our URL opener. Handles redirects as per FancyURLopener. But all other
bgneal@881 69 errors and authentication requests will raise an IOError.
bgneal@881 70 """
bgneal@881 71 HANDLED_ERRORS = set([302, 301, 303, 307])
bgneal@881 72
bgneal@881 73 def http_error_default(self, url, fp, errcode, errmsg, headers):
bgneal@881 74 return urllib.URLopener.http_error_default(self, url, fp, errcode,
bgneal@881 75 errmsg, headers)
bgneal@881 76
bgneal@881 77 def http_error(self, url, fp, errcode, errmsg, headers, data=None):
bgneal@881 78 """Handle http errors.
bgneal@881 79 We let FancyURLopener handle the redirects, but any other error we want
bgneal@881 80 to let fail.
bgneal@881 81 """
bgneal@881 82 if errcode in self.HANDLED_ERRORS:
bgneal@881 83 name = 'http_error_%d' % errcode
bgneal@881 84 method = getattr(self, name)
bgneal@881 85 if data is None:
bgneal@881 86 result = method(url, fp, errcode, errmsg, headers)
bgneal@881 87 else:
bgneal@881 88 result = method(url, fp, errcode, errmsg, headers, data)
bgneal@881 89 if result:
bgneal@881 90 return result
bgneal@881 91 return self.http_error_default(url, fp, errcode, errmsg, headers)
bgneal@881 92
bgneal@881 93
bgneal@881 94 def download_image(src):
bgneal@881 95 """Downloads the image file from the given source URL.
bgneal@881 96
bgneal@881 97 If successful returns the path to the downloaded file. Otherwise None is
bgneal@881 98 returned.
bgneal@881 99 """
bgneal@881 100 logger.info("Retrieving %s", src)
bgneal@881 101 try:
bgneal@881 102 fn, hdrs = opener.retrieve(src)
bgneal@881 103 except IOError as ex:
bgneal@881 104 args = ex.args
bgneal@881 105 if len(args) == 4 and args[0] == 'http error':
bgneal@881 106 logger.error("http error: %d - %s", args[1], args[2])
bgneal@881 107 else:
bgneal@881 108 logger.error("%s", ex)
bgneal@881 109 return None
bgneal@881 110
bgneal@881 111 # Does it look like an image?
bgneal@881 112 content_type = hdrs.get('content-type')
bgneal@881 113 if not content_type:
bgneal@881 114 logger.error("No content-type header found")
bgneal@881 115 return None
bgneal@881 116
bgneal@888 117 file_size = os.stat(fn).st_size
bgneal@888 118 logger.info("Retrieved: %s bytes; content-type: %s", file_size, content_type)
bgneal@881 119
bgneal@881 120 parts = content_type.split('/')
bgneal@881 121 if len(parts) < 2 or parts[0] != 'image':
bgneal@881 122 logger.error("Unknown content-type: %s", content_type)
bgneal@881 123 return None
bgneal@881 124
bgneal@881 125 return fn
bgneal@881 126
bgneal@881 127
bgneal@881 128 def resize_image(img_path):
bgneal@881 129 """Resizes the image found at img_path if necessary."""
bgneal@881 130 image = Image.open(img_path)
bgneal@881 131 if image.size > PHOTO_MAX_SIZE:
bgneal@881 132 logger.info('Resizing from %s to %s', image.size, PHOTO_MAX_SIZE)
bgneal@881 133 image.thumbnail(PHOTO_MAX_SIZE, Image.ANTIALIAS)
bgneal@881 134 image.save(img_path)
bgneal@881 135
bgneal@881 136
bgneal@882 137 def gen_key():
bgneal@882 138 """Return a random key."""
bgneal@882 139 return base64.b64encode(uuid.uuid4().bytes, '-_').rstrip('=')
bgneal@882 140
bgneal@882 141
bgneal@881 142 def upload_image(img_path):
bgneal@881 143 """Upload image file located at img_path to our S3 bucket.
bgneal@881 144
bgneal@881 145 Returns the URL of the image in the bucket or None if an error occurs.
bgneal@881 146 """
bgneal@881 147 logger.info("upload_image starting")
bgneal@881 148 # Make a unique name for the image in the bucket
bgneal@881 149 ext = os.path.splitext(img_path)[1]
bgneal@882 150 file_key = gen_key() + ext
bgneal@881 151 try:
bgneal@881 152 return bucket.upload_from_filename(file_key, img_path, public=True)
bgneal@881 153 except IOError as ex:
bgneal@881 154 logger.error("Error uploading file: %s", ex)
bgneal@881 155 return None
bgneal@881 156
bgneal@881 157
bgneal@888 158 def convert_to_ssl(parsed_url):
bgneal@888 159 """Top-level function for moving an image to SSL."""
bgneal@888 160
bgneal@888 161 src = parsed_url.geturl()
bgneal@888 162
bgneal@888 163 # Check the cache first
bgneal@888 164 new_url = url_cache.get(src)
bgneal@888 165 if new_url:
bgneal@888 166 logger.info("Found URL in cache: %s => %s", src, new_url)
bgneal@888 167 return new_url
bgneal@888 168
bgneal@888 169 # It has been observed that at least 2 different services
bgneal@888 170 # serve up the same image on https: with the URL otherwise the same.
bgneal@888 171 # Check to see if the image is available via https first.
bgneal@888 172 new_url = check_https_availability(parsed_url)
bgneal@888 173 if new_url:
bgneal@888 174 return new_url
bgneal@888 175
bgneal@888 176 # If none of the above worked, try to download and upload to our S3 bucket
bgneal@888 177 return save_image_to_cloud(src)
bgneal@888 178
bgneal@888 179
bgneal@888 180 def check_https_availability(parsed_url):
bgneal@888 181 """Given a urlparse.urlparse() result, perform a HEAD request over https
bgneal@888 182 using the same net location and path. If we get a response that indicates an
bgneal@888 183 image is available, return the url of the image over https. Otherwise return
bgneal@888 184 None.
bgneal@888 185 """
bgneal@888 186 logger.info("Checking https availability for %s", parsed_url.geturl())
bgneal@888 187 con = httplib.HTTPSConnection(parsed_url.netloc)
bgneal@888 188 try:
bgneal@888 189 con.request('HEAD', parsed_url.path)
bgneal@888 190 except (httplib.HTTPException, socket.timeout) as ex:
bgneal@888 191 logger.info("https HEAD request failed: %s", ex)
bgneal@888 192 return None
bgneal@888 193
bgneal@888 194 content_type = None
bgneal@888 195 response = con.getresponse()
bgneal@888 196 if response.status == 200:
bgneal@888 197 content_type = response.getheader('content-type')
bgneal@888 198 if content_type:
bgneal@888 199 parts = content_type.split('/')
bgneal@888 200 if len(parts) >= 2 and parts[0] == 'image':
bgneal@888 201 url = urlparse.urlunparse(('https', ) + parsed_url[1:])
bgneal@888 202 logger.info("Image is available at %s", url)
bgneal@888 203 return url
bgneal@888 204
bgneal@888 205 logger.info('https HEAD request failed; status = %d, content-type = %s',
bgneal@888 206 response.status, content_type)
bgneal@888 207 return None
bgneal@888 208
bgneal@888 209
bgneal@868 210 def save_image_to_cloud(src):
bgneal@881 211 """Downloads an image at a given source URL. Uploads it to cloud storage.
bgneal@881 212
bgneal@881 213 Returns the new URL or None if unsuccessful.
bgneal@881 214 """
bgneal@881 215 fn = download_image(src)
bgneal@881 216 if fn:
bgneal@881 217 resize_image(fn)
bgneal@881 218 new_url = upload_image(fn)
bgneal@881 219 if new_url:
bgneal@881 220 url_cache[src] = new_url
bgneal@881 221 return new_url
bgneal@881 222 return None
bgneal@868 223
bgneal@868 224
bgneal@866 225 def replace_image_markup(match):
bgneal@870 226 src_parts = match.group(8).split()
bgneal@868 227 if src_parts:
bgneal@868 228 src = src_parts[0]
bgneal@868 229 if src[0] == "<" and src[-1] == ">":
bgneal@868 230 src = src[1:-1]
bgneal@868 231 else:
bgneal@868 232 src = ''
bgneal@868 233
bgneal@868 234 title = ''
bgneal@868 235 if len(src_parts) > 1:
bgneal@868 236 title = " ".join(src_parts[1:])
bgneal@870 237 alt = match.group(1)
bgneal@868 238
bgneal@871 239 new_src = None
bgneal@868 240 if src:
bgneal@868 241 r = urlparse.urlparse(src)
bgneal@871 242 if r.hostname in SG101_HOSTS:
bgneal@871 243 new_src = r.path # convert to relative path
bgneal@871 244 elif r.scheme == 'http':
bgneal@888 245 # Try a few things to get this on ssl:
bgneal@888 246 new_src = convert_to_ssl(r)
bgneal@868 247 elif r.scheme == 'https':
bgneal@868 248 new_src = src # already https, accept it as-is
bgneal@868 249
bgneal@868 250 if new_src:
bgneal@868 251 if title:
bgneal@871 252 s = u'![{alt}]({src} {title})'.format(alt=alt, src=new_src, title=title)
bgneal@868 253 else:
bgneal@868 254 s = u'![{alt}]({src})'.format(alt=alt, src=new_src)
bgneal@868 255 else:
bgneal@868 256 # something's messed up, convert to a link using original src
bgneal@868 257 s = u'[{alt}]({src})'.format(alt=alt, src=src)
bgneal@868 258
bgneal@868 259 return s
bgneal@860 260
bgneal@860 261
bgneal@887 262 def warn_if_image_refs(text, model_name, pk):
bgneal@887 263 """Search text for Markdown image reference markup.
bgneal@887 264
bgneal@887 265 We aren't expecting these, but we will log something if we see any.
bgneal@887 266 """
bgneal@887 267 if IMAGE_REF_RE.search(text):
bgneal@887 268 logger.warning("Image reference found in %s pk = #%d", model_name, pk)
bgneal@887 269
bgneal@887 270
bgneal@866 271 def process_post(text):
bgneal@863 272 """Process the post object:
bgneal@863 273
bgneal@863 274 A regex substitution is run on the post's text field. This fixes up image
bgneal@863 275 links, getting rid of plain old http sources; either converting to https
bgneal@863 276 or relative style links (if the link is to SG101).
bgneal@863 277
bgneal@863 278 """
bgneal@866 279 return IMAGE_LINK_RE.sub(replace_image_markup, text)
bgneal@863 280
bgneal@863 281
bgneal@859 282 class Command(NoArgsCommand):
bgneal@859 283 help = "Rewrite forum posts and comments to not use http for images"
bgneal@859 284 option_list = NoArgsCommand.option_list + (
bgneal@866 285 make_option('-m', '--model',
bgneal@866 286 choices=MODEL_CHOICES,
bgneal@866 287 help="which model to update; must be one of {{{}}}".format(
bgneal@866 288 ', '.join(MODEL_CHOICES))),
bgneal@860 289 make_option('-i', '--i',
bgneal@859 290 type='int',
bgneal@863 291 help="optional first slice index; the i in [i:j]"),
bgneal@860 292 make_option('-j', '--j',
bgneal@859 293 type='int',
bgneal@863 294 help="optional second slice index; the j in [i:j]"),
bgneal@859 295 )
bgneal@859 296
bgneal@859 297 def handle_noargs(self, **options):
bgneal@859 298 _setup_logging()
bgneal@860 299 logger.info("Starting; arguments received: %s", options)
bgneal@859 300
bgneal@866 301 if options['model'] not in MODEL_CHOICES:
bgneal@866 302 raise CommandError('Please choose a --model option')
bgneal@859 303
bgneal@866 304 if options['model'] == 'comments':
bgneal@860 305 qs = Comment.objects.all()
bgneal@866 306 text_attr = 'comment'
bgneal@881 307 model_name = 'Comment'
bgneal@860 308 else:
bgneal@860 309 qs = Post.objects.all()
bgneal@866 310 text_attr = 'body'
bgneal@881 311 model_name = 'Post'
bgneal@860 312
bgneal@860 313 i, j = options['i'], options['j']
bgneal@860 314
bgneal@860 315 if i is not None and i < 0:
bgneal@860 316 raise CommandError("-i must be >= 0")
bgneal@860 317 if j is not None and j < 0:
bgneal@860 318 raise CommandError("-j must be >= 0")
bgneal@860 319 if j is not None and i is not None and j <= i:
bgneal@860 320 raise CommandError("-j must be > -i")
bgneal@860 321
bgneal@860 322 if i is not None and j is not None:
bgneal@860 323 qs = qs[i:j]
bgneal@860 324 elif i is not None and j is None:
bgneal@860 325 qs = qs[i:]
bgneal@860 326 elif i is None and j is not None:
bgneal@860 327 qs = qs[:j]
bgneal@860 328
bgneal@881 329 # Set global socket timeout
bgneal@881 330 socket.setdefaulttimeout(30)
bgneal@881 331
bgneal@863 332 # Install signal handler for ctrl-c
bgneal@863 333 signal.signal(signal.SIGINT, signal_handler)
bgneal@863 334
bgneal@881 335 # Create URL opener to download photos
bgneal@881 336 global opener
bgneal@881 337 opener = ImageURLopener()
bgneal@881 338
bgneal@881 339 # Create bucket to upload photos
bgneal@881 340 global bucket
bgneal@881 341 bucket = S3Bucket(access_key=settings.USER_PHOTOS_ACCESS_KEY,
bgneal@881 342 secret_key=settings.USER_PHOTOS_SECRET_KEY,
bgneal@881 343 base_url=PHOTO_BASE_URL,
bgneal@881 344 bucket_name=PHOTO_BUCKET_NAME)
bgneal@887 345
bgneal@887 346 if i is None:
bgneal@887 347 i = 0
bgneal@887 348
bgneal@881 349 for n, model in enumerate(qs.iterator()):
bgneal@863 350 if quit_flag:
bgneal@863 351 logger.warning("SIGINT received, exiting")
bgneal@881 352 break
bgneal@881 353 logger.info("Processing %s #%d (pk = %d)", model_name, n + i, model.pk)
bgneal@866 354 txt = getattr(model, text_attr)
bgneal@887 355 warn_if_image_refs(txt, model_name, model.pk)
bgneal@866 356 new_txt = process_post(txt)
bgneal@881 357 if txt != new_txt:
bgneal@887 358 logger.info("Content changed on %s #%d (pk= %d)",
bgneal@887 359 model_name, n + i, model.pk)
bgneal@881 360 logger.debug("original: %s", txt)
bgneal@881 361 logger.debug("changed: %s", new_txt)
bgneal@887 362 setattr(model, text_attr, new_txt)
bgneal@887 363 model.save()
bgneal@860 364
bgneal@887 365 logger.info("ssl_images exiting")