jaculabilis/danbooru
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

nokogiri: switch to the nokogumbo-based html5 parser.

Browse Source 
https://github.com/sparklemotion/nokogiri/blob/main/CHANGELOG.md#1120--2021-08-02
This commit is contained in:
evazion
2021-08-30 21:21:03 -05:00
parent 49d18e64e8
commit 38c9559fe8
3 changed files with 7 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
app/logical/d_text.rb
View File

@@ -49,7 +49,7 @@ class DText
# @param artists [Array<Artist>] # @param artists [Array<Artist>]
# @return [String] the HTML output # @return [String] the HTML output
def self.postprocess(html, wiki_pages, tags, artists) def self.postprocess(html, wiki_pages, tags, artists)
fragment = Nokogiri::HTML.fragment(html) fragment = Nokogiri::HTML5.fragment(html)
fragment.css("a.dtext-wiki-link").each do |node| fragment.css("a.dtext-wiki-link").each do |node|
path = Addressable::URI.parse(node["href"]).path path = Addressable::URI.parse(node["href"]).path
@@ -174,7 +174,7 @@ class DText
# @return [Array<String>] the list of wiki page names # @return [Array<String>] the list of wiki page names
def self.parse_wiki_titles(text) def self.parse_wiki_titles(text)
html = DTextRagel.parse(text) html = DTextRagel.parse(text)
fragment = Nokogiri::HTML.fragment(html) fragment = Nokogiri::HTML5.fragment(html)
titles = fragment.css("a.dtext-wiki-link").map do |node| titles = fragment.css("a.dtext-wiki-link").map do |node|
title = node["href"][%r{\A/wiki_pages/(.*)\z}i, 1] title = node["href"][%r{\A/wiki_pages/(.*)\z}i, 1]
@@ -191,7 +191,7 @@ class DText
# @return [Array<String>] the list of external URLs # @return [Array<String>] the list of external URLs
def self.parse_external_links(text) def self.parse_external_links(text)
html = DTextRagel.parse(text) html = DTextRagel.parse(text)
fragment = Nokogiri::HTML.fragment(html) fragment = Nokogiri::HTML5.fragment(html)
links = fragment.css("a.dtext-external-link").map { |node| node["href"] } links = fragment.css("a.dtext-external-link").map { |node| node["href"] }
links.uniq links.uniq
@@ -326,7 +326,7 @@ class DText
# @param html [String] the HTML input # @param html [String] the HTML input
# @return [String] the Markdown output # @return [String] the Markdown output
def self.html_to_markdown(html) def self.html_to_markdown(html)
html = Nokogiri::HTML.fragment(html) html = Nokogiri::HTML5.fragment(html)
html.children.map do |node| html.children.map do |node|
case node.name case node.name
@@ -349,7 +349,7 @@ class DText
# @param inline [Boolean] if true, convert <img> tags to plaintext # @param inline [Boolean] if true, convert <img> tags to plaintext
# @return [String] the DText output # @return [String] the DText output
def self.from_html(text, inline: false, &block) def self.from_html(text, inline: false, &block)
html = Nokogiri::HTML.fragment(text) html = Nokogiri::HTML5.fragment(text)
dtext = html.children.map do |element| dtext = html.children.map do |element|
block.call(element) if block.present? block.call(element) if block.present?

2
app/logical/sources/strategies/deviant_art.rb
View File

@@ -269,7 +269,7 @@ module Sources
resp = http.cache(1.minute).get(page_url_from_image_url, follow: {max_hops: 1}) resp = http.cache(1.minute).get(page_url_from_image_url, follow: {max_hops: 1})
if resp.status.success? if resp.status.success?
Nokogiri::HTML(resp.body.to_s) resp.parse
# the work was deleted # the work was deleted
elsif resp.code == 404 elsif resp.code == 404
nil nil

2
app/logical/sources/strategies/tumblr.rb
View File

@@ -183,7 +183,7 @@ module Sources::Strategies
end end
def inline_images def inline_images
html = Nokogiri::HTML.fragment(artist_commentary_desc) html = Nokogiri::HTML5.fragment(artist_commentary_desc)
html.css("img").map { |node| node["src"] } html.css("img").map { |node| node["src"] }
end end