discourse/lib/cooked_post_processor.rb

242 lines
6.8 KiB
Ruby
Raw Normal View History

# Post processing that we can do after a post has already been cooked.
2013-07-07 19:39:08 -04:00
# For example, inserting the onebox content, or image sizes/thumbnails.
2013-02-05 14:16:51 -05:00
2013-11-05 13:04:47 -05:00
require_dependency "oneboxer"
2013-02-05 14:16:51 -05:00
class CookedPostProcessor
2013-06-21 12:29:40 -04:00
include ActionView::Helpers::NumberHelper
2013-02-19 01:57:14 -05:00
2013-02-05 14:16:51 -05:00
def initialize(post, opts={})
@dirty = false
@opts = opts
@post = post
@doc = Nokogiri::HTML::fragment(post.cooked)
2013-02-19 01:57:14 -05:00
@size_cache = {}
2013-02-05 14:16:51 -05:00
end
def post_process
2013-11-05 13:04:47 -05:00
keep_reverse_index_up_to_date
post_process_images
post_process_oneboxes
2013-11-05 13:04:47 -05:00
optimize_urls
pull_hotlinked_images
2013-02-05 14:16:51 -05:00
end
2013-11-05 13:04:47 -05:00
def keep_reverse_index_up_to_date
upload_ids = Set.new
2013-10-14 08:27:41 -04:00
2013-11-05 13:04:47 -05:00
@doc.search("a").each do |a|
href = a["href"].to_s
2013-07-10 16:55:37 -04:00
if upload = Upload.get_from_url(href)
2013-11-05 13:04:47 -05:00
upload_ids << upload.id
end
end
@doc.search("img").each do |img|
src = img["src"].to_s
if upload = Upload.get_from_url(src)
upload_ids << upload.id
end
end
values = upload_ids.map{ |u| "(#{@post.id},#{u})" }.join(",")
PostUpload.transaction do
PostUpload.delete_all(post_id: @post.id)
if upload_ids.length > 0
PostUpload.exec_sql("INSERT INTO post_uploads (post_id, upload_id) VALUES #{values}")
2013-07-10 16:55:37 -04:00
end
end
end
2013-02-25 11:42:20 -05:00
def post_process_images
2013-07-07 19:39:08 -04:00
images = extract_images
return if images.blank?
2013-02-05 14:16:51 -05:00
2013-02-19 01:57:14 -05:00
images.each do |img|
2013-11-05 13:04:47 -05:00
src, width, height = img["src"], img["width"], img["height"]
limit_size!(img)
convert_to_link!(img)
@dirty |= (src != img["src"]) || (width.to_i != img["width"].to_i) || (height.to_i != img["height"].to_i)
2013-04-13 10:31:20 -04:00
end
2013-02-19 01:57:14 -05:00
2013-11-05 13:04:47 -05:00
update_topic_image(images)
end
2013-04-13 10:31:20 -04:00
2013-07-07 19:39:08 -04:00
def extract_images
2013-11-05 13:04:47 -05:00
# do not extract images inside oneboxes or quotes
2013-07-07 19:39:08 -04:00
@doc.css("img") - @doc.css(".onebox-result img") - @doc.css(".quote img")
end
2013-11-05 13:04:47 -05:00
def limit_size!(img)
w, h = get_size_from_image_sizes(img["src"], @opts[:image_sizes]) || get_size(img["src"])
# limit the size of the thumbnail
img["width"], img["height"] = ImageSizer.resize(w, h)
2013-07-07 19:39:08 -04:00
end
2013-11-05 13:04:47 -05:00
def get_size_from_image_sizes(src, image_sizes)
return unless image_sizes.present?
image_sizes.each do |image_size|
url, size = image_size[0], image_size[1]
return [size["width"], size["height"]] if url.include?(src)
end
end
2013-02-20 20:07:36 -05:00
2013-11-05 13:04:47 -05:00
def get_size(url)
absolute_url = url
absolute_url = Discourse.base_url_no_prefix + absolute_url if absolute_url =~ /^\/[^\/]/
# FastImage fails when there's no scheme
absolute_url = (SiteSetting.use_ssl? ? "https:" : "http:") + absolute_url if absolute_url.start_with?("//")
return unless is_valid_image_url?(absolute_url)
# we can *always* crawl our own images
return unless SiteSetting.crawl_images? || Discourse.store.has_been_uploaded?(url)
@size_cache[url] ||= FastImage.size(absolute_url)
rescue Zlib::BufError # FastImage.size raises BufError for some gifs
2013-06-17 16:46:48 -04:00
end
2013-11-05 13:04:47 -05:00
def is_valid_image_url?(url)
uri = URI.parse(url)
%w(http https).include? uri.scheme
rescue URI::InvalidURIError
2013-02-19 01:57:14 -05:00
end
2013-11-05 13:04:47 -05:00
def convert_to_link!(img)
2013-02-19 01:57:14 -05:00
src = img["src"]
return unless src.present?
2013-02-19 01:57:14 -05:00
width, height = img["width"].to_i, img["height"].to_i
2013-04-13 10:31:20 -04:00
original_width, original_height = get_size(src)
2013-02-19 01:57:14 -05:00
2013-07-07 19:39:08 -04:00
return if original_width.to_i <= width && original_height.to_i <= height
2013-08-25 18:24:24 -04:00
return if original_width.to_i <= SiteSetting.max_image_width && original_height.to_i <= SiteSetting.max_image_height
2013-11-05 13:04:47 -05:00
return if is_a_hyperlink?(img)
2013-07-07 19:39:08 -04:00
2013-11-05 13:04:47 -05:00
if upload = Upload.get_from_url(src)
2013-09-27 04:55:50 -04:00
upload.create_thumbnail!(width, height)
2013-07-07 19:39:08 -04:00
# TODO: optimize_image!(img)
end
2013-02-19 01:57:14 -05:00
2013-07-07 19:39:08 -04:00
add_lightbox!(img, original_width, original_height, upload)
@dirty = true
end
2013-11-05 13:04:47 -05:00
def is_a_hyperlink?(img)
2013-02-19 01:57:14 -05:00
parent = img.parent
while parent
return if parent.name == "a"
break unless parent.respond_to? :parent
parent = parent.parent
end
2013-07-07 19:39:08 -04:00
end
2013-02-19 01:57:14 -05:00
2013-07-07 19:39:08 -04:00
def add_lightbox!(img, original_width, original_height, upload=nil)
2013-06-25 20:44:20 -04:00
# first, create a div to hold our lightbox
2013-07-07 19:39:08 -04:00
lightbox = Nokogiri::XML::Node.new("div", @doc)
img.add_next_sibling(lightbox)
lightbox.add_child(img)
2013-06-25 20:44:20 -04:00
# then, the link to our larger image
2013-07-07 19:39:08 -04:00
a = Nokogiri::XML::Node.new("a", @doc)
2013-02-19 01:57:14 -05:00
img.add_next_sibling(a)
2013-11-05 13:04:47 -05:00
a["href"] = img["src"]
2013-02-19 01:57:14 -05:00
a["class"] = "lightbox"
a.add_child(img)
2013-07-07 19:39:08 -04:00
# replace the image by its thumbnail
2013-11-05 13:04:47 -05:00
w, h = img["width"].to_i, img["height"].to_i
img["src"] = upload.thumbnail(w, h).url if upload && upload.has_thumbnail?(w, h)
2013-07-07 19:39:08 -04:00
2013-06-25 20:44:20 -04:00
# then, some overlay informations
2013-07-07 19:39:08 -04:00
meta = Nokogiri::XML::Node.new("div", @doc)
2013-06-25 20:44:20 -04:00
meta["class"] = "meta"
2013-07-07 19:39:08 -04:00
img.add_next_sibling(meta)
2013-06-21 12:29:40 -04:00
2013-11-05 13:04:47 -05:00
filename = get_filename(upload, img["src"])
2013-06-21 12:29:40 -04:00
informations = "#{original_width}x#{original_height}"
2013-07-24 03:24:28 -04:00
informations << " #{number_to_human_size(upload.filesize)}" if upload
2013-06-21 12:29:40 -04:00
2013-06-25 20:44:20 -04:00
meta.add_child create_span_node("filename", filename)
meta.add_child create_span_node("informations", informations)
meta.add_child create_span_node("expand")
2013-06-21 12:29:40 -04:00
end
2013-02-19 01:57:14 -05:00
def get_filename(upload, src)
return File.basename(src) unless upload
return upload.original_filename unless upload.original_filename =~ /^blob(\.png)?$/i
2013-11-05 13:04:47 -05:00
return I18n.t("upload.pasted_image_filename")
end
2013-06-21 12:29:40 -04:00
def create_span_node(klass, content=nil)
2013-07-07 19:39:08 -04:00
span = Nokogiri::XML::Node.new("span", @doc)
2013-06-21 12:29:40 -04:00
span.content = content if content
2013-11-05 13:04:47 -05:00
span["class"] = klass
2013-06-21 12:29:40 -04:00
span
2013-02-05 14:16:51 -05:00
end
2013-11-05 13:04:47 -05:00
def update_topic_image(images)
2013-07-07 19:39:08 -04:00
if @post.post_number == 1
img = images.first
2013-11-05 13:04:47 -05:00
@post.topic.update_column(:image_url, img["src"]) if img["src"].present?
2013-07-07 19:39:08 -04:00
end
end
2013-11-05 13:04:47 -05:00
def post_process_oneboxes
args = {
post_id: @post.id,
invalidate_oneboxes: !!@opts[:invalidate_oneboxes],
}
result = Oneboxer.apply(@doc) do |url, element|
Oneboxer.onebox(url, args)
end
@dirty |= result.changed?
2013-02-05 14:16:51 -05:00
end
2013-11-05 13:04:47 -05:00
def optimize_urls
@doc.search("a").each do |a|
href = a["href"].to_s
2013-11-06 10:59:11 -05:00
a["href"] = schemaless absolute(href) if is_local(href)
2013-11-05 13:04:47 -05:00
end
@doc.search("img").each do |img|
src = img["src"].to_s
2013-11-06 10:59:11 -05:00
img["src"] = schemaless absolute(src) if is_local(src)
2013-11-05 13:04:47 -05:00
end
2013-02-05 14:16:51 -05:00
end
2013-11-06 10:59:11 -05:00
def is_local(url)
Discourse.store.has_been_uploaded?(url) || url =~ /^\/assets\//
end
def absolute(url)
2013-11-05 13:04:47 -05:00
url =~ /^\/[^\/]/ ? (Discourse.asset_host || Discourse.base_url_no_prefix) + url : url
end
def schemaless(url)
url.gsub(/^https?:/, "")
end
2013-11-05 13:04:47 -05:00
def pull_hotlinked_images
# we don't want to run the job if we're not allowed to crawl images
return unless SiteSetting.crawl_images?
# we only want to run the job whenever it's changed by a user
return if @post.updated_by == Discourse.system_user
# make sure no other job is scheduled
Jobs.cancel_scheduled_job(:pull_hotlinked_images, post_id: @post.id)
# schedule the job
delay = SiteSetting.ninja_edit_window + 1
Jobs.enqueue_in(delay.seconds.to_i, :pull_hotlinked_images, post_id: @post.id)
2013-07-10 16:55:37 -04:00
end
def dirty?
@dirty
end
def html
@doc.try(:to_html)
2013-02-05 14:16:51 -05:00
end
end