mirror of
https://github.com/basecamp/once-campfire.git
synced 2026-09-27 18:46:23 +09:00
Stanko K.R.
eab554aa4b
Main strips a javascript: href from a preview node before it's parsed, so tell legacy Trix attachments apart by their filename instead. escapeHTML moved to the string helpers, and the unfurling system test now drives the Lexxy editor.
109 lines
3.7 KiB
Ruby
109 lines
3.7 KiB
Ruby
class ActionText::Attachment::OpengraphEmbed
|
|
include ActiveModel::Model
|
|
|
|
OPENGRAPH_EMBED_CONTENT_TYPE = "application/vnd.actiontext.opengraph-embed"
|
|
TWITTER_AVATAR_URL_PREFIX = "https://pbs.twimg.com/profile_images"
|
|
|
|
class << self
|
|
def from_node(node)
|
|
if node["content-type"]
|
|
if matches = node["content-type"].match(OPENGRAPH_EMBED_CONTENT_TYPE)
|
|
attachment = new(attributes_from_node(node))
|
|
attachment if attachment.valid?
|
|
end
|
|
end
|
|
end
|
|
|
|
private
|
|
# Trix serialized the embed's details as attributes of the
|
|
# <action-text-attachment> node. Lexxy only serializes the sgid, content
|
|
# and content-type, so newer attachments carry the details in their
|
|
# content markup instead.
|
|
def attributes_from_node(node)
|
|
if node["filename"].present?
|
|
{
|
|
href: web_url(node["href"]),
|
|
url: web_url(node["url"]),
|
|
filename: node["filename"],
|
|
description: node["caption"]
|
|
}
|
|
else
|
|
attributes_from_content(node["content"].to_s)
|
|
end
|
|
end
|
|
|
|
def attributes_from_content(content)
|
|
fragment = Nokogiri::HTML.fragment(content)
|
|
title = fragment.at_css(".og-embed__title")
|
|
link = title&.at_css("a")
|
|
|
|
{
|
|
href: web_url(link&.[]("href")),
|
|
url: web_url(fragment.at_css(".og-embed__image img")&.[]("src")),
|
|
filename: (link || title)&.text&.strip,
|
|
description: fragment.at_css(".og-embed__description")&.text&.strip
|
|
}
|
|
end
|
|
|
|
# A link preview points at what we unfurled: an absolute http or https URL
|
|
# on some other host. Drop anything else a message body asks for, so it
|
|
# can't aim the preview's link or its image at this Campfire and have every
|
|
# reader's browser fetch it with their session attached.
|
|
def web_url(value)
|
|
return if value.blank?
|
|
|
|
parsed = URI.parse(value)
|
|
value if parsed.is_a?(URI::HTTP) && elsewhere?(parsed.host)
|
|
rescue URI::InvalidURIError
|
|
nil
|
|
end
|
|
|
|
# "https:/rooms/1" parses as HTTPS with no host at all, and a browser
|
|
# resolves both that and our own hostname against the origin Campfire is
|
|
# served from. A percent-escape hides our hostname from this comparison
|
|
# while a browser still unescapes it back to us, so an escaped host is out
|
|
# too, and neither case is anything an unfurl could have produced.
|
|
def elsewhere?(host)
|
|
return false unless named_host?(host)
|
|
|
|
canonical_host(host) != canonical_host(Current.request_host.to_s)
|
|
end
|
|
|
|
# A preview names a page on the public internet, so its host is a domain
|
|
# name, written plainly. A bare address is not one, and a browser rewrites
|
|
# the many spellings of an address ("2130706433", "0x7f.0.0.1") into a
|
|
# single one before it fetches, which is a race a comparison here loses.
|
|
def named_host?(host)
|
|
host.present? && host.exclude?("%") && host.include?(".") && domain_ending?(host.split(".").last)
|
|
end
|
|
|
|
# What keeps a name from reading as an address is its last label, which is
|
|
# a word: never a number, and never the hexadecimal spelling of one.
|
|
def domain_ending?(label)
|
|
label.match?(/[a-z]/i) && !label.match?(/\A0x/i)
|
|
end
|
|
|
|
def canonical_host(host)
|
|
host.downcase.delete_suffix(".")
|
|
end
|
|
end
|
|
|
|
attr_accessor :href, :url, :filename, :description
|
|
|
|
def twitter_avatar?
|
|
url.to_s.start_with?(TWITTER_AVATAR_URL_PREFIX)
|
|
end
|
|
|
|
def attachable_content_type
|
|
OPENGRAPH_EMBED_CONTENT_TYPE
|
|
end
|
|
|
|
def attachable_plain_text_representation(caption)
|
|
""
|
|
end
|
|
|
|
def to_partial_path
|
|
"action_text/attachables/opengraph_embed"
|
|
end
|
|
end
|