def handle(self, *app_labels, **options): from my_django.db.models import get_app, get_apps, get_model format = options.get('format') indent = options.get('indent') using = options.get('database') excludes = options.get('exclude') show_traceback = options.get('traceback') use_natural_keys = options.get('use_natural_keys') use_base_manager = options.get('use_base_manager') excluded_apps = set() excluded_models = set() for exclude in excludes: if '.' in exclude: app_label, model_name = exclude.split('.', 1) model_obj = get_model(app_label, model_name) if not model_obj: raise CommandError('Unknown model in excludes: %s' % exclude) excluded_models.add(model_obj) else: try: app_obj = get_app(exclude) excluded_apps.add(app_obj) except ImproperlyConfigured: raise CommandError('Unknown app in excludes: %s' % exclude) if len(app_labels) == 0: app_list = SortedDict( (app, None) for app in get_apps() if app not in excluded_apps) else: app_list = SortedDict() for label in app_labels: try: app_label, model_label = label.split('.') try: app = get_app(app_label) except ImproperlyConfigured: raise CommandError("Unknown application: %s" % app_label) if app in excluded_apps: continue model = get_model(app_label, model_label) if model is None: raise CommandError("Unknown model: %s.%s" % (app_label, model_label)) if app in app_list.keys(): if app_list[app] and model not in app_list[app]: app_list[app].append(model) else: app_list[app] = [model] except ValueError: # This is just an app - no model qualifier app_label = label try: app = get_app(app_label) except ImproperlyConfigured: raise CommandError("Unknown application: %s" % app_label) if app in excluded_apps: continue app_list[app] = None # Check that the serialization format exists; this is a shortcut to # avoid collating all the objects and _then_ failing. if format not in serializers.get_public_serializer_formats(): raise CommandError("Unknown serialization format: %s" % format) try: serializers.get_serializer(format) except KeyError: raise CommandError("Unknown serialization format: %s" % format) # Now collate the objects to be serialized. objects = [] for model in sort_dependencies(app_list.items()): if model in excluded_models: continue if not model._meta.proxy and router.allow_syncdb(using, model): if use_base_manager: objects.extend(model._base_manager.using(using).all()) else: objects.extend(model._default_manager.using(using).all()) try: return serializers.serialize(format, objects, indent=indent, use_natural_keys=use_natural_keys) except Exception, e: if show_traceback: raise raise CommandError("Unable to serialize database: %s" % e)
class CachedFilesMixin(object): patterns = ( ("*.css", ( r"""(url\(['"]{0,1}\s*(.*?)["']{0,1}\))""", r"""(@import\s*["']\s*(.*?)["'])""", )), ) def __init__(self, *args, **kwargs): super(CachedFilesMixin, self).__init__(*args, **kwargs) try: self.cache = get_cache('staticfiles') except InvalidCacheBackendError: # Use the default backend self.cache = default_cache self._patterns = SortedDict() for extension, patterns in self.patterns: for pattern in patterns: compiled = re.compile(pattern) self._patterns.setdefault(extension, []).append(compiled) def hashed_name(self, name, content=None): parsed_name = urlsplit(unquote(name)) clean_name = parsed_name.path.strip() if content is None: if not self.exists(clean_name): raise ValueError("The file '%s' could not be found with %r." % (clean_name, self)) try: content = self.open(clean_name) except IOError: # Handle directory paths and fragments return name path, filename = os.path.split(clean_name) root, ext = os.path.splitext(filename) # Get the MD5 hash of the file md5 = hashlib.md5() for chunk in content.chunks(): md5.update(chunk) md5sum = md5.hexdigest()[:12] hashed_name = os.path.join(path, u"%s.%s%s" % (root, md5sum, ext)) unparsed_name = list(parsed_name) unparsed_name[2] = hashed_name # Special casing for a @font-face hack, like url(myfont.eot?#iefix") # http://www.fontspring.com/blog/the-new-bulletproof-font-face-syntax if '?#' in name and not unparsed_name[3]: unparsed_name[2] += '?' return urlunsplit(unparsed_name) def cache_key(self, name): return u'staticfiles:%s' % hashlib.md5(smart_str(name)).hexdigest() def url(self, name, force=False): """ Returns the real URL in DEBUG mode. """ if settings.DEBUG and not force: hashed_name, fragment = name, '' else: clean_name, fragment = urldefrag(name) if urlsplit(clean_name).path.endswith('/'): # don't hash paths hashed_name = name else: cache_key = self.cache_key(name) hashed_name = self.cache.get(cache_key) if hashed_name is None: hashed_name = self.hashed_name(clean_name).replace('\\', '/') # set the cache if there was a miss # (e.g. if cache server goes down) self.cache.set(cache_key, hashed_name) final_url = super(CachedFilesMixin, self).url(hashed_name) # Special casing for a @font-face hack, like url(myfont.eot?#iefix") # http://www.fontspring.com/blog/the-new-bulletproof-font-face-syntax query_fragment = '?#' in name # [sic!] if fragment or query_fragment: urlparts = list(urlsplit(final_url)) if fragment and not urlparts[4]: urlparts[4] = fragment if query_fragment and not urlparts[3]: urlparts[2] += '?' final_url = urlunsplit(urlparts) return unquote(final_url) def url_converter(self, name): """ Returns the custom URL converter for the given file name. """ def converter(matchobj): """ Converts the matched URL depending on the parent level (`..`) and returns the normalized and hashed URL using the url method of the storage. """ matched, url = matchobj.groups() # Completely ignore http(s) prefixed URLs, # fragments and data-uri URLs if url.startswith(('#', 'http:', 'https:', 'data:')): return matched name_parts = name.split(os.sep) # Using posix normpath here to remove duplicates url = posixpath.normpath(url) url_parts = url.split('/') parent_level, sub_level = url.count('..'), url.count('/') if url.startswith('/'): sub_level -= 1 url_parts = url_parts[1:] if parent_level or not url.startswith('/'): start, end = parent_level + 1, parent_level else: if sub_level: if sub_level == 1: parent_level -= 1 start, end = parent_level, 1 else: start, end = 1, sub_level - 1 joined_result = '/'.join(name_parts[:-start] + url_parts[end:]) hashed_url = self.url(unquote(joined_result), force=True) file_name = hashed_url.split('/')[-1:] relative_url = '/'.join(url.split('/')[:-1] + file_name) # Return the hashed version to the file return 'url("%s")' % unquote(relative_url) return converter def post_process(self, paths, dry_run=False, **options): """ Post process the given list of files (called from collectstatic). Processing is actually two separate operations: 1. renaming files to include a hash of their content for cache-busting, and copying those files to the target storage. 2. adjusting files which contain references to other files so they refer to the cache-busting filenames. If either of these are performed on a file, then that file is considered post-processed. """ # don't even dare to process the files if we're in dry run mode if dry_run: return # delete cache of all handled paths self.cache.delete_many([self.cache_key(path) for path in paths]) # build a list of adjustable files matches = lambda path: matches_patterns(path, self._patterns.keys()) adjustable_paths = [path for path in paths if matches(path)] # then sort the files by the directory level path_level = lambda name: len(name.split(os.sep)) for name in sorted(paths.keys(), key=path_level, reverse=True): # use the original, local file, not the copied-but-unprocessed # file, which might be somewhere far away, like S3 storage, path = paths[name] with storage.open(path) as original_file: # generate the hash with the original content, even for # adjustable files. hashed_name = self.hashed_name(name, original_file) # then get the original's file content.. if hasattr(original_file, 'seek'): original_file.seek(0) hashed_file_exists = self.exists(hashed_name) processed = False # ..to apply each replacement pattern to the content if name in adjustable_paths: content = original_file.read() converter = self.url_converter(name) for patterns in self._patterns.values(): for pattern in patterns: content = pattern.sub(converter, content) if hashed_file_exists: self.delete(hashed_name) # then save the processed result content_file = ContentFile(smart_str(content)) saved_name = self._save(hashed_name, content_file) hashed_name = force_unicode(saved_name.replace('\\', '/')) processed = True else: # or handle the case in which neither processing nor # a change to the original file happened if not hashed_file_exists: processed = True saved_name = self._save(hashed_name, original_file) hashed_name = force_unicode(saved_name.replace('\\', '/')) # and then set the cache accordingly self.cache.set(self.cache_key(name), hashed_name) yield name, hashed_name, processed