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@882: import base64
bgneal@895: import datetime
bgneal@899: import json
bgneal@859: import logging
bgneal@859: from optparse import make_option
bgneal@888: import os
bgneal@863: import re
bgneal@863: import signal
bgneal@868: import urlparse
bgneal@881: import uuid
bgneal@859:
bgneal@859: from django.core.management.base import NoArgsCommand, CommandError
bgneal@859: from django.conf import settings
bgneal@894: from lxml import etree
bgneal@987: import lxml.html
bgneal@863: import markdown.inlinepatterns
bgneal@881: from PIL import Image
bgneal@979: import requests
bgneal@859:
bgneal@860: from comments.models import Comment
bgneal@860: from forums.models import Post
bgneal@979: from core.download import download_file
bgneal@979: from core.functions import remove_file
bgneal@881: from core.s3 import S3Bucket
bgneal@987: from news.models import Story
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@871: IMAGE_LINK_RE = re.compile(markdown.inlinepatterns.IMAGE_LINK_RE,
bgneal@871: re.DOTALL | re.UNICODE)
bgneal@871: IMAGE_REF_RE = re.compile(markdown.inlinepatterns.IMAGE_REFERENCE_RE,
bgneal@871: re.DOTALL | re.UNICODE)
bgneal@863:
bgneal@868: SG101_HOSTS = set(['www.surfguitar101.com', 'surfguitar101.com'])
bgneal@963: WHITELIST_HOSTS = set(settings.USER_IMAGES_SOURCES)
bgneal@987: MODEL_CHOICES = ['comments', 'posts', 'news']
bgneal@866:
bgneal@881: PHOTO_MAX_SIZE = (660, 720)
bgneal@979: PHOTO_BASE_URL = settings.HOT_LINK_PHOTOS_BASE_URL
bgneal@979: PHOTO_BUCKET_NAME = settings.HOT_LINK_PHOTOS_BUCKET
bgneal@881:
bgneal@899: CACHE_FILENAME = 'ssl_images_cache.json'
bgneal@899:
bgneal@863: quit_flag = False
bgneal@881: bucket = None
bgneal@881: url_cache = {}
bgneal@899: bad_hosts = set()
bgneal@980: request_timeout = None
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@981: requests_log = logging.getLogger("requests.packages.urllib3")
bgneal@981: requests_log.setLevel(logging.INFO)
bgneal@981: requests_log.propagate = True
bgneal@981: requests_log.addHandler(handler)
bgneal@981:
bgneal@982: dl_log = logging.getLogger("core.download")
bgneal@982: dl_log.setLevel(logging.INFO)
bgneal@982: dl_log.propagate = True
bgneal@982: dl_log.addHandler(handler)
bgneal@982:
bgneal@859:
bgneal@979: def resize_image(img_path):
bgneal@979: """Resizes the image found at img_path if necessary.
bgneal@979:
bgneal@979: Returns True if the image was resized or resizing wasn't necessary.
bgneal@979: Returns False if the image could not be read or processed.
bgneal@881: """
bgneal@979: try:
bgneal@979: image = Image.open(img_path)
bgneal@979: except IOError as ex:
bgneal@979: logger.error("Error opening %s: %s", img_path, ex)
bgneal@979: return False
bgneal@881:
bgneal@881: if image.size > PHOTO_MAX_SIZE:
bgneal@881: logger.info('Resizing from %s to %s', image.size, PHOTO_MAX_SIZE)
bgneal@981: try:
bgneal@981: image.thumbnail(PHOTO_MAX_SIZE, Image.ANTIALIAS)
bgneal@981: image.save(img_path)
bgneal@981: except IOError as ex:
bgneal@981: logger.error("Error resizing image from %s: %s", img_path, ex)
bgneal@981: return False
bgneal@881:
bgneal@979: return True
bgneal@979:
bgneal@881:
bgneal@882: def gen_key():
bgneal@882: """Return a random key."""
bgneal@882: return base64.b64encode(uuid.uuid4().bytes, '-_').rstrip('=')
bgneal@882:
bgneal@882:
bgneal@881: def upload_image(img_path):
bgneal@881: """Upload image file located at img_path to our S3 bucket.
bgneal@881:
bgneal@881: Returns the URL of the image in the bucket or None if an error occurs.
bgneal@881: """
bgneal@881: logger.info("upload_image starting")
bgneal@881: # Make a unique name for the image in the bucket
bgneal@881: ext = os.path.splitext(img_path)[1]
bgneal@882: file_key = gen_key() + ext
bgneal@881: try:
bgneal@881: return bucket.upload_from_filename(file_key, img_path, public=True)
bgneal@881: except IOError as ex:
bgneal@881: logger.error("Error uploading file: %s", ex)
bgneal@881: return None
bgneal@881:
bgneal@881:
bgneal@888: def convert_to_ssl(parsed_url):
bgneal@888: """Top-level function for moving an image to SSL."""
bgneal@888:
bgneal@888: src = parsed_url.geturl()
bgneal@888:
bgneal@899: if parsed_url.hostname in bad_hosts:
bgneal@899: logger.info("Host known to be bad, skipping: %s", src)
bgneal@899: return None
bgneal@899:
bgneal@899: # Check the cache
bgneal@897: try:
bgneal@897: new_url = url_cache[src]
bgneal@897: except KeyError:
bgneal@897: # cache miss, try to get the file
bgneal@899: new_url = save_image_to_cloud(parsed_url)
bgneal@897: url_cache[src] = new_url
bgneal@897: else:
bgneal@897: if new_url:
bgneal@897: logger.info("Found URL in cache: %s => %s", src, new_url)
bgneal@897: else:
bgneal@897: logger.info("URL known to be bad, skipping: %s", src)
bgneal@888:
bgneal@889: return new_url
bgneal@888:
bgneal@888:
bgneal@899: def save_image_to_cloud(parsed_url):
bgneal@881: """Downloads an image at a given source URL. Uploads it to cloud storage.
bgneal@881:
bgneal@881: Returns the new URL or None if unsuccessful.
bgneal@881: """
bgneal@979: url = parsed_url.geturl()
bgneal@979: fn = None
bgneal@979: try:
bgneal@980: fn = download_file(url, timeout=request_timeout)
bgneal@979: except requests.ConnectionError as ex:
bgneal@979: logger.error("ConnectionError, ignoring host %s", parsed_url.hostname)
bgneal@979: bad_hosts.add(parsed_url.hostname)
bgneal@979: except requests.RequestException as ex:
bgneal@979: logger.error("%s", ex)
bgneal@979: except Exception as ex:
bgneal@979: logger.exception("%s", ex)
bgneal@979:
bgneal@881: if fn:
bgneal@979: with remove_file(fn):
bgneal@979: if resize_image(fn):
bgneal@979: return upload_image(fn)
bgneal@881: return None
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@871: new_src = None
bgneal@868: if src:
bgneal@986: try:
bgneal@986: r = urlparse.urlparse(src)
bgneal@986: except ValueError:
bgneal@986: return u'{bad image}'
bgneal@986:
bgneal@871: if r.hostname in SG101_HOSTS:
bgneal@871: new_src = r.path # convert to relative path
bgneal@871: elif r.scheme == 'http':
bgneal@888: # Try a few things to get this on ssl:
bgneal@888: new_src = convert_to_ssl(r)
bgneal@868: elif r.scheme == 'https':
bgneal@963: if r.hostname in WHITELIST_HOSTS:
bgneal@963: new_src = src # already in whitelist
bgneal@963: else:
bgneal@963: new_src = convert_to_ssl(r)
bgneal@868:
bgneal@868: if new_src:
bgneal@868: if title:
bgneal@871: s = u''.format(alt=alt, src=new_src, title=title)
bgneal@868: else:
bgneal@868: s = u''.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@887: def warn_if_image_refs(text, model_name, pk):
bgneal@887: """Search text for Markdown image reference markup.
bgneal@887:
bgneal@887: We aren't expecting these, but we will log something if we see any.
bgneal@887: """
bgneal@887: if IMAGE_REF_RE.search(text):
bgneal@887: logger.warning("Image reference found in %s pk = #%d", model_name, pk)
bgneal@887:
bgneal@887:
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: """
bgneal@866: return IMAGE_LINK_RE.sub(replace_image_markup, text)
bgneal@863:
bgneal@863:
bgneal@987: def process_html(html):
bgneal@987: """Process the html fragment, converting to https where needed."""
bgneal@987: s = html.strip()
bgneal@987: if not s:
bgneal@987: return s
bgneal@987:
bgneal@987: changed = False
bgneal@987: root = lxml.html.fragment_fromstring(s, create_parent=True)
bgneal@987: for img in root.iter('img'):
bgneal@987: src = img.get('src')
bgneal@987: src = src.strip() if src else ''
bgneal@987: if src:
bgneal@987: try:
bgneal@987: r = urlparse.urlparse(src)
bgneal@987: except ValueError:
bgneal@987: logger.warning("Bad url? Should not happen; skipping...")
bgneal@987: continue
bgneal@987:
bgneal@987: new_src = None
bgneal@987: if r.hostname in SG101_HOSTS:
bgneal@987: new_src = r.path # convert to relative path
bgneal@987: elif ((r.scheme == 'http') or
bgneal@987: (r.scheme == 'https' and r.hostname not in WHITELIST_HOSTS)):
bgneal@987: new_src = convert_to_ssl(r)
bgneal@987: if not new_src:
bgneal@987: # failed to convert to https; convert to a link
bgneal@987: tail = img.tail
bgneal@987: img.clear()
bgneal@987: img.tag = 'a'
bgneal@987: img.set('href', src)
bgneal@987: img.text = 'Image'
bgneal@987: img.tail = tail
bgneal@987: changed = True
bgneal@987:
bgneal@987: if new_src:
bgneal@987: img.set('src', new_src)
bgneal@987: changed = True
bgneal@987:
bgneal@987: if changed:
bgneal@987: result = lxml.html.tostring(root, encoding='utf-8')
bgneal@988: result = result[5:-6] # strip off parent div we added
bgneal@988: return result.decode('utf-8')
bgneal@987: return html
bgneal@987:
bgneal@987:
bgneal@894: def html_check(html):
bgneal@894: """Return True if the given HTML fragment has <img> tags with src attributes
bgneal@894: that use http, and False otherwise.
bgneal@894: """
bgneal@894: if not html:
bgneal@894: return False
bgneal@894:
bgneal@894: root = etree.HTML(html)
bgneal@894: for img in root.iter('img'):
bgneal@894: src = img.get('src')
bgneal@894: if src and src.lower().startswith('http:'):
bgneal@894: return True
bgneal@894: return False
bgneal@894:
bgneal@894:
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@898: make_option('-t', '--timeout',
bgneal@980: type='float',
bgneal@979: help="optional socket timeout (secs)",
bgneal@980: default=30.0),
bgneal@859: )
bgneal@859:
bgneal@859: def handle_noargs(self, **options):
bgneal@895: time_started = datetime.datetime.now()
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@987: text_attrs = ['comment']
bgneal@881: model_name = 'Comment'
bgneal@987: elif options['model'] == 'posts':
bgneal@987: qs = Post.objects.all()
bgneal@987: text_attrs = ['body']
bgneal@987: model_name = 'Post'
bgneal@860: else:
bgneal@987: qs = Story.objects.all()
bgneal@987: text_attrs = ['short_text', 'long_text']
bgneal@987: model_name = 'Story'
bgneal@987:
bgneal@987: html_based = options['model'] == 'news'
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@881: # Set global socket timeout
bgneal@980: global request_timeout
bgneal@980: request_timeout = options.get('timeout')
bgneal@980: logger.info("Using socket timeout of %4.2f", request_timeout)
bgneal@881:
bgneal@863: # Install signal handler for ctrl-c
bgneal@863: signal.signal(signal.SIGINT, signal_handler)
bgneal@863:
bgneal@881: # Create bucket to upload photos
bgneal@881: global bucket
bgneal@881: bucket = S3Bucket(access_key=settings.USER_PHOTOS_ACCESS_KEY,
bgneal@881: secret_key=settings.USER_PHOTOS_SECRET_KEY,
bgneal@881: base_url=PHOTO_BASE_URL,
bgneal@881: bucket_name=PHOTO_BUCKET_NAME)
bgneal@887:
bgneal@899: # Load cached info from previous runs
bgneal@899: load_cache()
bgneal@899:
bgneal@887: if i is None:
bgneal@887: i = 0
bgneal@887:
bgneal@895: count = 0
bgneal@881: for n, model in enumerate(qs.iterator()):
bgneal@863: if quit_flag:
bgneal@863: logger.warning("SIGINT received, exiting")
bgneal@881: break
bgneal@881: logger.info("Processing %s #%d (pk = %d)", model_name, n + i, model.pk)
bgneal@987: save_flag = False
bgneal@987: for text_attr in text_attrs:
bgneal@987: txt = getattr(model, text_attr)
bgneal@987:
bgneal@987: if html_based:
bgneal@987: new_txt = process_html(txt)
bgneal@987: else:
bgneal@987: new_txt = process_post(txt)
bgneal@987: warn_if_image_refs(txt, model_name, model.pk)
bgneal@987:
bgneal@987: if txt != new_txt:
bgneal@987: logger.info("Content changed on %s #%d (pk = %d)",
bgneal@987: model_name, n + i, model.pk)
bgneal@987: logger.debug(u"original: %s", txt)
bgneal@987: logger.debug(u"changed: %s", new_txt)
bgneal@987: setattr(model, text_attr, new_txt)
bgneal@987: save_flag = True
bgneal@987: elif not html_based and html_check(model.html):
bgneal@987: # Check for content generated with older smiley code that used
bgneal@987: # absolute URLs for the smiley images. If True, then just save
bgneal@987: # the model again to force updated HTML to be created.
bgneal@987: logger.info("Older Smiley HTML detected, forcing a save")
bgneal@987: save_flag = True
bgneal@987:
bgneal@987: if save_flag:
bgneal@894: model.save()
bgneal@895: count += 1
bgneal@860:
bgneal@895: time_finished = datetime.datetime.now()
bgneal@895: elapsed = time_finished - time_started
bgneal@895: logger.info("ssl_images exiting; number of objects: %d; elapsed: %s",
bgneal@895: count, elapsed)
bgneal@897:
bgneal@897: http_images = len(url_cache)
bgneal@897: https_images = sum(1 for v in url_cache.itervalues() if v)
bgneal@897: bad_images = http_images - https_images
bgneal@897: if http_images > 0:
bgneal@897: pct_saved = float(https_images) / http_images * 100.0
bgneal@897: else:
bgneal@897: pct_saved = 0.0
bgneal@897:
bgneal@897: logger.info("Summary: http: %d; https: %d; lost: %d; saved: %3.1f %%",
bgneal@897: http_images, https_images, bad_images, pct_saved)
bgneal@899:
bgneal@899: save_cache()
bgneal@899: logger.info("ssl_images done")
bgneal@899:
bgneal@899:
bgneal@899: def load_cache():
bgneal@899: """Load cache from previous runs."""
bgneal@899: logger.info("Loading cached information")
bgneal@899: try:
bgneal@899: with open(CACHE_FILENAME, 'r') as fp:
bgneal@899: d = json.load(fp)
bgneal@899: except IOError as ex:
bgneal@899: logger.error("Cache file (%s) IOError: %s", CACHE_FILENAME, ex)
bgneal@899: return
bgneal@899: except ValueError:
bgneal@899: logger.error("Mangled cache file: %s", CACHE_FILENAME)
bgneal@899: return
bgneal@899:
bgneal@899: global bad_hosts, url_cache
bgneal@899: try:
bgneal@899: bad_hosts = set(d['bad_hosts'])
bgneal@899: url_cache = d['url_cache']
bgneal@899: except KeyError:
bgneal@899: logger.error("Malformed cache file: %s", CACHE_FILENAME)
bgneal@899:
bgneal@899:
bgneal@899: def save_cache():
bgneal@899: """Save our cache to a file for future runs."""
bgneal@899: logger.info("Saving cached information")
bgneal@899: d = {'bad_hosts': list(bad_hosts), 'url_cache': url_cache}
bgneal@899: with open(CACHE_FILENAME, 'w') as fp:
bgneal@899: json.dump(d, fp, indent=4)