discourse/lib/cooked_post_processor.rb

241 lines
6.7 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
require_dependency 'oneboxer'
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 = {}
@has_been_uploaded_cache = {}
2013-02-05 14:16:51 -05:00
end
def post_process
2013-07-10 16:55:37 -04:00
post_process_attachments
post_process_images
post_process_oneboxes
2013-02-05 14:16:51 -05:00
end
2013-07-10 16:55:37 -04:00
def post_process_attachments
attachments.each do |attachment|
href = attachment['href']
attachment['href'] = relative_to_absolute(href)
# update reverse index
2013-07-10 16:55:37 -04:00
if upload = Upload.get_from_url(href)
associate_to_post(upload)
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|
2013-07-07 19:39:08 -04:00
if img['src'].present?
# keep track of the original src
src = img['src']
# make sure the src is absolute (when working with locally uploaded files)
img['src'] = relative_to_absolute(src)
# make sure the img has proper width and height attributes
update_dimensions!(img)
2013-06-16 19:00:25 -04:00
# retrieve the associated upload, if any
2013-07-10 16:55:37 -04:00
if upload = Upload.get_from_url(src)
2013-06-17 16:46:48 -04:00
# update reverse index
2013-07-07 19:39:08 -04:00
associate_to_post(upload)
2013-06-16 19:00:25 -04:00
end
2013-07-07 19:39:08 -04:00
# lightbox treatment
convert_to_link!(img, upload)
# mark the post as dirty whenever the src has changed
@dirty |= src != img['src']
2013-02-05 14:16:51 -05:00
end
2013-04-13 10:31:20 -04:00
end
2013-02-19 01:57:14 -05:00
2013-04-13 10:31:20 -04:00
# Extract the first image from the first post and use it as the 'topic image'
2013-07-07 19:39:08 -04:00
extract_topic_image(images)
2013-02-19 01:57:14 -05:00
end
def post_process_oneboxes
args = { post_id: @post.id }
args[:invalidate_oneboxes] = true if @opts[:invalidate_oneboxes]
# bake onebox content into the post
result = Oneboxer.apply(@doc) do |url, element|
Oneboxer.onebox(url, args)
end
# mark the post as dirty whenever a onebox as been baked
@dirty |= result.changed?
end
2013-04-13 10:31:20 -04:00
2013-07-07 19:39:08 -04:00
def extract_images
# do not extract images inside a onebox or a quote
@doc.css("img") - @doc.css(".onebox-result img") - @doc.css(".quote img")
end
def relative_to_absolute(src)
if src =~ /^\/[^\/]/
2013-07-07 19:39:08 -04:00
Discourse.base_url_no_prefix + src
else
src
end
end
def update_dimensions!(img)
return if img['width'].present? && img['height'].present?
w, h = get_size_from_image_sizes(img['src'], @opts[:image_sizes]) || image_dimensions(img['src'])
if w && h
2013-07-07 19:39:08 -04:00
img['width'] = w
img['height'] = h
@dirty = true
end
end
2013-02-20 20:07:36 -05:00
2013-06-17 16:46:48 -04:00
def associate_to_post(upload)
return if PostUpload.where(post_id: @post.id, upload_id: upload.id).count > 0
PostUpload.create(post_id: @post.id, upload_id: upload.id)
2013-06-17 16:46:48 -04:00
rescue ActiveRecord::RecordNotUnique
# do not care if it's already associated
end
2013-07-07 19:39:08 -04:00
def optimize_image!(img)
# TODO
# 1) optimize using image_optim
2013-07-07 19:39:08 -04:00
# 2) .png vs. .jpg (> 1.5x)
2013-02-19 01:57:14 -05:00
end
2013-06-21 12:29:40 -04:00
def convert_to_link!(img, upload=nil)
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
2013-07-07 19:39:08 -04:00
return if is_a_hyperlink(img)
if upload
# create a thumbnail
upload.create_thumbnail!
# optimize image
# 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
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)
2013-07-07 19:39:08 -04:00
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
img['src'] = relative_to_absolute(upload.thumbnail.url) if upload && upload.has_thumbnail?
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
2013-07-07 19:39:08 -04:00
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
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
span['class'] = klass
span
2013-02-05 14:16:51 -05:00
end
2013-07-07 19:39:08 -04:00
def extract_topic_image(images)
if @post.post_number == 1
img = images.first
@post.topic.update_column :image_url, img['src'] if img['src'].present?
end
end
2013-02-19 01:57:14 -05:00
def get_size_from_image_sizes(src, image_sizes)
if image_sizes.present?
if dim = image_sizes[src]
ImageSizer.resize(dim['width'], dim['height'])
end
end
end
2013-02-05 14:16:51 -05:00
# Retrieve the image dimensions for a url
def image_dimensions(url)
w, h = get_size(url)
ImageSizer.resize(w, h) if w && h
2013-02-05 14:16:51 -05:00
end
def get_size(url)
uri = url
# make sure urls have a scheme (otherwise, FastImage will fail)
uri = (SiteSetting.use_ssl? ? "https:" : "http:") + url if url.start_with?("//")
return unless is_valid_image_uri?(uri)
# we can *always* crawl our own images
return unless SiteSetting.crawl_images? || Discourse.store.has_been_uploaded?(url)
@size_cache[url] ||= FastImage.size(uri)
rescue Zlib::BufError # FastImage.size raises BufError for some gifs
2013-02-05 14:16:51 -05:00
end
def is_valid_image_uri?(url)
uri = URI.parse(url)
%w(http https).include? uri.scheme
rescue URI::InvalidURIError
end
2013-07-10 16:55:37 -04:00
def attachments
attachments = @doc.css("a.attachment[href^=\"#{Discourse.store.absolute_base_url}\"]")
attachments += @doc.css("a.attachment[href^=\"#{Discourse.store.relative_base_url}\"]") if Discourse.store.internal?
attachments
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