discourse/app/services/search_indexer.rb

251 lines
7.1 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
require_dependency 'search'
class SearchIndexer
2019-05-28 16:34:48 -04:00
INDEX_VERSION = 3
REINDEX_VERSION = 0
def self.disable
@disabled = true
end
def self.enable
@disabled = false
end
2013-02-05 14:16:51 -05:00
def self.scrub_html_for_search(html, strip_diacritics: SiteSetting.search_ignore_accents)
HtmlScrubber.scrub(html, strip_diacritics: strip_diacritics)
2013-02-05 14:16:51 -05:00
end
def self.inject_extra_terms(raw)
# insert some extra words for I.am.a.word so "word" is tokenized
# I.am.a.word becomes I.am.a.word am a word
raw.gsub(/[^[:space:]]*[\.]+[^[:space:]]*/) do |with_dot|
PERF: Improve quality of `PostSearchData#raw_data`. (#7275) This commit fixes the follow quality issue with `PostSearchData#raw_data`: 1. URLs are being tokenized and links with similar href and characters are being duplicated in the raw data. `Post#cooked`: ``` <p><a href=\"https://meta.discourse.org/some.png\" class=\"onebox\" target=\"_blank\" rel=\"nofollow noopener\">https://meta.discourse.org/some.png</a></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png discourse org/some png https://meta.discourse.org/some.png discourse org/some png ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png meta discourse org ``` 2. Ligthbox being included in search pollutes the `PostSearchData#raw_data` unncessarily. From 28 March 2018 to 28 March 2019, searches for the term `image` on `meta.discourse.org` had a click through rate of 2.1%. Non-lightboxed images are not included in indexing for search yet we were indexing content within a lightbox. Also, search for terms like `image` was affected we were using `Pasted image` as the filename for uploads that were pasted. `Post#cooked` ``` <p>Let me see how I can fix this image<br>\n<div class=\"lightbox-wrapper\"><a class=\"lightbox\" href=\"https://meta.discourse.org/some.png\" title=\"some.png\" rel=\"nofollow noopener\"><img src=\"https://meta.discourse.org/some.png\" width=\"275\" height=\"299\"><div class=\"meta\">\n<svg class=\"fa d-icon d-icon-far-image svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#far-image\"></use></svg><span class=\"filename\">some.png</span><span class=\"informations\">1750×2000</span><svg class=\"fa d-icon d-icon-discourse-expand svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#discourse-expand\"></use></svg>\n</div></a></div></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image some.png png https://meta.discourse.org/some.png discourse org/some png some.png png 1750×2000 ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image ``` In terms of indexing performance, we now have to parse the given HTML through nokogiri twice. However performance is not a huge worry here since a string length of 194170 takes only 30ms to scrub plus the indexing takes place in a background job.
2019-03-31 22:14:29 -04:00
split = with_dot.split(/https?:\/\/|[?:;,.\/]/)
if split.length > 1
with_dot + ((+" ") << split[1..-1].reject { |x| x.blank? }.join(" "))
else
with_dot
end
end
end
def self.update_index(table: , id: , raw_data:)
search_data = raw_data.map do |data|
inject_extra_terms(Search.prepare_data(data || "", :index))
end
table_name = "#{table}_search_data"
foreign_key = "#{table}_id"
# for user login and name use "simple" lowercase stemmer
stemmer = table == "user" ? "simple" : Search.ts_config
ranked_index = <<~SQL
setweight(to_tsvector('#{stemmer}', coalesce(:a,'')), 'A') ||
setweight(to_tsvector('#{stemmer}', coalesce(:b,'')), 'B') ||
setweight(to_tsvector('#{stemmer}', coalesce(:c,'')), 'C') ||
setweight(to_tsvector('#{stemmer}', coalesce(:d,'')), 'D')
SQL
indexed_data = search_data.select { |d| d.length > 0 }.join(' ')
params = {
a: search_data[0],
b: search_data[1],
c: search_data[2],
d: search_data[3],
raw_data: indexed_data,
id: id,
locale: SiteSetting.default_locale,
version: INDEX_VERSION
}
# Would be nice to use AR here but not sure how to execut Postgres functions
# when inserting data like this.
rows = DB.exec(<<~SQL, params)
UPDATE #{table_name}
SET
raw_data = :raw_data,
locale = :locale,
search_data = #{ranked_index},
version = :version
WHERE #{foreign_key} = :id
SQL
if rows == 0
DB.exec(<<~SQL, params)
INSERT INTO #{table_name}
(#{foreign_key}, search_data, locale, raw_data, version)
VALUES (:id, #{ranked_index}, :locale, :raw_data, :version)
SQL
2013-02-05 14:16:51 -05:00
end
rescue
# TODO is there any way we can safely avoid this?
# best way is probably pushing search indexer into a dedicated process so it no longer happens on save
# instead in the post processor
2013-02-05 14:16:51 -05:00
end
def self.update_topics_index(topic_id, title, cooked)
scrubbed_cooked = scrub_html_for_search(cooked)[0...Topic::MAX_SIMILAR_BODY_LENGTH]
# a bit inconsitent that we use title as A and body as B when in
# the post index body is C
update_index(table: 'topic', id: topic_id, raw_data: [title, scrubbed_cooked])
end
def self.update_posts_index(post_id, title, category, tags, cooked)
update_index(table: 'post', id: post_id, raw_data: [title, category, tags, scrub_html_for_search(cooked)])
2013-02-05 14:16:51 -05:00
end
def self.update_users_index(user_id, username, name)
update_index(table: 'user', id: user_id, raw_data: [username, name])
2013-02-05 14:16:51 -05:00
end
2013-02-07 10:45:24 -05:00
2013-02-05 14:16:51 -05:00
def self.update_categories_index(category_id, name)
update_index(table: 'category', id: category_id, raw_data: [name])
2013-02-05 14:16:51 -05:00
end
2017-08-25 11:52:18 -04:00
def self.update_tags_index(tag_id, name)
update_index(table: 'tag', id: tag_id, raw_data: [name.downcase])
end
def self.queue_post_reindex(topic_id)
return if @disabled
DB.exec(<<~SQL, topic_id: topic_id, version: REINDEX_VERSION)
UPDATE post_search_data
SET version = :version
FROM posts
WHERE post_search_data.post_id = posts.id
AND posts.topic_id = :topic_id
SQL
2017-08-25 11:52:18 -04:00
end
def self.index(obj, force: false)
return if @disabled
category_name = nil
tag_names = nil
topic = nil
if Topic === obj
topic = obj
elsif Post === obj
topic = obj.topic
end
category_name = topic.category&.name if topic
tag_names = topic.tags.pluck(:name).join(' ') if topic
if Post === obj && obj.raw.present? &&
(
obj.saved_change_to_cooked? ||
obj.saved_change_to_topic_id? ||
force
)
if topic
SearchIndexer.update_posts_index(obj.id, topic.title, category_name, tag_names, obj.cooked)
SearchIndexer.update_topics_index(topic.id, topic.title, obj.cooked) if obj.is_first_post?
2014-05-06 22:35:26 -04:00
end
2013-02-05 14:16:51 -05:00
end
if User === obj && (obj.saved_change_to_username? || obj.saved_change_to_name? || force)
SearchIndexer.update_users_index(obj.id, obj.username_lower || '', obj.name ? obj.name.downcase : '')
2013-02-05 14:16:51 -05:00
end
if Topic === obj && (obj.saved_change_to_title? || force)
2013-02-05 14:16:51 -05:00
if obj.posts
if post = obj.posts.find_by(post_number: 1)
SearchIndexer.update_posts_index(post.id, obj.title, category_name, tag_names, post.cooked)
SearchIndexer.update_topics_index(obj.id, obj.title, post.cooked)
2013-02-05 14:16:51 -05:00
end
end
end
if Category === obj && (obj.saved_change_to_name? || force)
SearchIndexer.update_categories_index(obj.id, obj.name)
2013-02-05 14:16:51 -05:00
end
2017-08-25 11:52:18 -04:00
if Tag === obj && (obj.saved_change_to_name? || force)
2017-08-25 11:52:18 -04:00
SearchIndexer.update_tags_index(obj.id, obj.name)
end
2013-02-05 14:16:51 -05:00
end
class HtmlScrubber < Nokogiri::XML::SAX::Document
2013-02-05 14:16:51 -05:00
attr_reader :scrubbed
def initialize(strip_diacritics: false)
@scrubbed = +""
@strip_diacritics = strip_diacritics
2013-02-05 14:16:51 -05:00
end
def self.scrub(html, strip_diacritics: false)
return +"" if html.blank?
PERF: Improve quality of `PostSearchData#raw_data`. (#7275) This commit fixes the follow quality issue with `PostSearchData#raw_data`: 1. URLs are being tokenized and links with similar href and characters are being duplicated in the raw data. `Post#cooked`: ``` <p><a href=\"https://meta.discourse.org/some.png\" class=\"onebox\" target=\"_blank\" rel=\"nofollow noopener\">https://meta.discourse.org/some.png</a></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png discourse org/some png https://meta.discourse.org/some.png discourse org/some png ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png meta discourse org ``` 2. Ligthbox being included in search pollutes the `PostSearchData#raw_data` unncessarily. From 28 March 2018 to 28 March 2019, searches for the term `image` on `meta.discourse.org` had a click through rate of 2.1%. Non-lightboxed images are not included in indexing for search yet we were indexing content within a lightbox. Also, search for terms like `image` was affected we were using `Pasted image` as the filename for uploads that were pasted. `Post#cooked` ``` <p>Let me see how I can fix this image<br>\n<div class=\"lightbox-wrapper\"><a class=\"lightbox\" href=\"https://meta.discourse.org/some.png\" title=\"some.png\" rel=\"nofollow noopener\"><img src=\"https://meta.discourse.org/some.png\" width=\"275\" height=\"299\"><div class=\"meta\">\n<svg class=\"fa d-icon d-icon-far-image svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#far-image\"></use></svg><span class=\"filename\">some.png</span><span class=\"informations\">1750×2000</span><svg class=\"fa d-icon d-icon-discourse-expand svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#discourse-expand\"></use></svg>\n</div></a></div></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image some.png png https://meta.discourse.org/some.png discourse org/some png some.png png 1750×2000 ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image ``` In terms of indexing performance, we now have to parse the given HTML through nokogiri twice. However performance is not a huge worry here since a string length of 194170 takes only 30ms to scrub plus the indexing takes place in a background job.
2019-03-31 22:14:29 -04:00
document = Nokogiri::HTML("<div>#{html}</div>", nil, Encoding::UTF_8.to_s)
nodes = document.css(
PERF: Improve quality of `PostSearchData#raw_data`. (#7275) This commit fixes the follow quality issue with `PostSearchData#raw_data`: 1. URLs are being tokenized and links with similar href and characters are being duplicated in the raw data. `Post#cooked`: ``` <p><a href=\"https://meta.discourse.org/some.png\" class=\"onebox\" target=\"_blank\" rel=\"nofollow noopener\">https://meta.discourse.org/some.png</a></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png discourse org/some png https://meta.discourse.org/some.png discourse org/some png ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png meta discourse org ``` 2. Ligthbox being included in search pollutes the `PostSearchData#raw_data` unncessarily. From 28 March 2018 to 28 March 2019, searches for the term `image` on `meta.discourse.org` had a click through rate of 2.1%. Non-lightboxed images are not included in indexing for search yet we were indexing content within a lightbox. Also, search for terms like `image` was affected we were using `Pasted image` as the filename for uploads that were pasted. `Post#cooked` ``` <p>Let me see how I can fix this image<br>\n<div class=\"lightbox-wrapper\"><a class=\"lightbox\" href=\"https://meta.discourse.org/some.png\" title=\"some.png\" rel=\"nofollow noopener\"><img src=\"https://meta.discourse.org/some.png\" width=\"275\" height=\"299\"><div class=\"meta\">\n<svg class=\"fa d-icon d-icon-far-image svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#far-image\"></use></svg><span class=\"filename\">some.png</span><span class=\"informations\">1750×2000</span><svg class=\"fa d-icon d-icon-discourse-expand svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#discourse-expand\"></use></svg>\n</div></a></div></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image some.png png https://meta.discourse.org/some.png discourse org/some png some.png png 1750×2000 ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image ``` In terms of indexing performance, we now have to parse the given HTML through nokogiri twice. However performance is not a huge worry here since a string length of 194170 takes only 30ms to scrub plus the indexing takes place in a background job.
2019-03-31 22:14:29 -04:00
"div.#{CookedPostProcessor::LIGHTBOX_WRAPPER_CSS_CLASS}"
)
if nodes.present?
nodes.each do |node|
node.traverse do |child_node|
next if child_node == node
if %w{a img}.exclude?(child_node.name)
child_node.remove
elsif child_node.name == "a"
ATTRIBUTES.each do |attribute|
child_node.remove_attribute(attribute)
end
end
end
end
end
PERF: Improve quality of `PostSearchData#raw_data`. (#7275) This commit fixes the follow quality issue with `PostSearchData#raw_data`: 1. URLs are being tokenized and links with similar href and characters are being duplicated in the raw data. `Post#cooked`: ``` <p><a href=\"https://meta.discourse.org/some.png\" class=\"onebox\" target=\"_blank\" rel=\"nofollow noopener\">https://meta.discourse.org/some.png</a></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png discourse org/some png https://meta.discourse.org/some.png discourse org/some png ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png meta discourse org ``` 2. Ligthbox being included in search pollutes the `PostSearchData#raw_data` unncessarily. From 28 March 2018 to 28 March 2019, searches for the term `image` on `meta.discourse.org` had a click through rate of 2.1%. Non-lightboxed images are not included in indexing for search yet we were indexing content within a lightbox. Also, search for terms like `image` was affected we were using `Pasted image` as the filename for uploads that were pasted. `Post#cooked` ``` <p>Let me see how I can fix this image<br>\n<div class=\"lightbox-wrapper\"><a class=\"lightbox\" href=\"https://meta.discourse.org/some.png\" title=\"some.png\" rel=\"nofollow noopener\"><img src=\"https://meta.discourse.org/some.png\" width=\"275\" height=\"299\"><div class=\"meta\">\n<svg class=\"fa d-icon d-icon-far-image svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#far-image\"></use></svg><span class=\"filename\">some.png</span><span class=\"informations\">1750×2000</span><svg class=\"fa d-icon d-icon-discourse-expand svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#discourse-expand\"></use></svg>\n</div></a></div></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image some.png png https://meta.discourse.org/some.png discourse org/some png some.png png 1750×2000 ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image ``` In terms of indexing performance, we now have to parse the given HTML through nokogiri twice. However performance is not a huge worry here since a string length of 194170 takes only 30ms to scrub plus the indexing takes place in a background job.
2019-03-31 22:14:29 -04:00
document.css("img[class='emoji']").each do |node|
node.remove_attribute("alt")
end
PERF: Improve quality of `PostSearchData#raw_data`. (#7275) This commit fixes the follow quality issue with `PostSearchData#raw_data`: 1. URLs are being tokenized and links with similar href and characters are being duplicated in the raw data. `Post#cooked`: ``` <p><a href=\"https://meta.discourse.org/some.png\" class=\"onebox\" target=\"_blank\" rel=\"nofollow noopener\">https://meta.discourse.org/some.png</a></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png discourse org/some png https://meta.discourse.org/some.png discourse org/some png ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png meta discourse org ``` 2. Ligthbox being included in search pollutes the `PostSearchData#raw_data` unncessarily. From 28 March 2018 to 28 March 2019, searches for the term `image` on `meta.discourse.org` had a click through rate of 2.1%. Non-lightboxed images are not included in indexing for search yet we were indexing content within a lightbox. Also, search for terms like `image` was affected we were using `Pasted image` as the filename for uploads that were pasted. `Post#cooked` ``` <p>Let me see how I can fix this image<br>\n<div class=\"lightbox-wrapper\"><a class=\"lightbox\" href=\"https://meta.discourse.org/some.png\" title=\"some.png\" rel=\"nofollow noopener\"><img src=\"https://meta.discourse.org/some.png\" width=\"275\" height=\"299\"><div class=\"meta\">\n<svg class=\"fa d-icon d-icon-far-image svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#far-image\"></use></svg><span class=\"filename\">some.png</span><span class=\"informations\">1750×2000</span><svg class=\"fa d-icon d-icon-discourse-expand svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#discourse-expand\"></use></svg>\n</div></a></div></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image some.png png https://meta.discourse.org/some.png discourse org/some png some.png png 1750×2000 ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image ``` In terms of indexing performance, we now have to parse the given HTML through nokogiri twice. However performance is not a huge worry here since a string length of 194170 takes only 30ms to scrub plus the indexing takes place in a background job.
2019-03-31 22:14:29 -04:00
document.css("a[href]").each do |node|
if node["href"] == node.text || MENTION_CLASSES.include?(node["class"])
node.remove_attribute("href")
end
PERF: Improve quality of `PostSearchData#raw_data`. (#7275) This commit fixes the follow quality issue with `PostSearchData#raw_data`: 1. URLs are being tokenized and links with similar href and characters are being duplicated in the raw data. `Post#cooked`: ``` <p><a href=\"https://meta.discourse.org/some.png\" class=\"onebox\" target=\"_blank\" rel=\"nofollow noopener\">https://meta.discourse.org/some.png</a></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png discourse org/some png https://meta.discourse.org/some.png discourse org/some png ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png meta discourse org ``` 2. Ligthbox being included in search pollutes the `PostSearchData#raw_data` unncessarily. From 28 March 2018 to 28 March 2019, searches for the term `image` on `meta.discourse.org` had a click through rate of 2.1%. Non-lightboxed images are not included in indexing for search yet we were indexing content within a lightbox. Also, search for terms like `image` was affected we were using `Pasted image` as the filename for uploads that were pasted. `Post#cooked` ``` <p>Let me see how I can fix this image<br>\n<div class=\"lightbox-wrapper\"><a class=\"lightbox\" href=\"https://meta.discourse.org/some.png\" title=\"some.png\" rel=\"nofollow noopener\"><img src=\"https://meta.discourse.org/some.png\" width=\"275\" height=\"299\"><div class=\"meta\">\n<svg class=\"fa d-icon d-icon-far-image svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#far-image\"></use></svg><span class=\"filename\">some.png</span><span class=\"informations\">1750×2000</span><svg class=\"fa d-icon d-icon-discourse-expand svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#discourse-expand\"></use></svg>\n</div></a></div></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image some.png png https://meta.discourse.org/some.png discourse org/some png some.png png 1750×2000 ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image ``` In terms of indexing performance, we now have to parse the given HTML through nokogiri twice. However performance is not a huge worry here since a string length of 194170 takes only 30ms to scrub plus the indexing takes place in a background job.
2019-03-31 22:14:29 -04:00
end
me = new(strip_diacritics: strip_diacritics)
PERF: Improve quality of `PostSearchData#raw_data`. (#7275) This commit fixes the follow quality issue with `PostSearchData#raw_data`: 1. URLs are being tokenized and links with similar href and characters are being duplicated in the raw data. `Post#cooked`: ``` <p><a href=\"https://meta.discourse.org/some.png\" class=\"onebox\" target=\"_blank\" rel=\"nofollow noopener\">https://meta.discourse.org/some.png</a></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png discourse org/some png https://meta.discourse.org/some.png discourse org/some png ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png meta discourse org ``` 2. Ligthbox being included in search pollutes the `PostSearchData#raw_data` unncessarily. From 28 March 2018 to 28 March 2019, searches for the term `image` on `meta.discourse.org` had a click through rate of 2.1%. Non-lightboxed images are not included in indexing for search yet we were indexing content within a lightbox. Also, search for terms like `image` was affected we were using `Pasted image` as the filename for uploads that were pasted. `Post#cooked` ``` <p>Let me see how I can fix this image<br>\n<div class=\"lightbox-wrapper\"><a class=\"lightbox\" href=\"https://meta.discourse.org/some.png\" title=\"some.png\" rel=\"nofollow noopener\"><img src=\"https://meta.discourse.org/some.png\" width=\"275\" height=\"299\"><div class=\"meta\">\n<svg class=\"fa d-icon d-icon-far-image svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#far-image\"></use></svg><span class=\"filename\">some.png</span><span class=\"informations\">1750×2000</span><svg class=\"fa d-icon d-icon-discourse-expand svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#discourse-expand\"></use></svg>\n</div></a></div></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image some.png png https://meta.discourse.org/some.png discourse org/some png some.png png 1750×2000 ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image ``` In terms of indexing performance, we now have to parse the given HTML through nokogiri twice. However performance is not a huge worry here since a string length of 194170 takes only 30ms to scrub plus the indexing takes place in a background job.
2019-03-31 22:14:29 -04:00
Nokogiri::HTML::SAX::Parser.new(me).parse(document.to_html)
me.scrubbed.squish
2013-02-05 14:16:51 -05:00
end
MENTION_CLASSES ||= %w{mention mention-group}
ATTRIBUTES ||= %w{alt title href data-youtube-title}
PERF: Improve quality of `PostSearchData#raw_data`. (#7275) This commit fixes the follow quality issue with `PostSearchData#raw_data`: 1. URLs are being tokenized and links with similar href and characters are being duplicated in the raw data. `Post#cooked`: ``` <p><a href=\"https://meta.discourse.org/some.png\" class=\"onebox\" target=\"_blank\" rel=\"nofollow noopener\">https://meta.discourse.org/some.png</a></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png discourse org/some png https://meta.discourse.org/some.png discourse org/some png ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png meta discourse org ``` 2. Ligthbox being included in search pollutes the `PostSearchData#raw_data` unncessarily. From 28 March 2018 to 28 March 2019, searches for the term `image` on `meta.discourse.org` had a click through rate of 2.1%. Non-lightboxed images are not included in indexing for search yet we were indexing content within a lightbox. Also, search for terms like `image` was affected we were using `Pasted image` as the filename for uploads that were pasted. `Post#cooked` ``` <p>Let me see how I can fix this image<br>\n<div class=\"lightbox-wrapper\"><a class=\"lightbox\" href=\"https://meta.discourse.org/some.png\" title=\"some.png\" rel=\"nofollow noopener\"><img src=\"https://meta.discourse.org/some.png\" width=\"275\" height=\"299\"><div class=\"meta\">\n<svg class=\"fa d-icon d-icon-far-image svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#far-image\"></use></svg><span class=\"filename\">some.png</span><span class=\"informations\">1750×2000</span><svg class=\"fa d-icon d-icon-discourse-expand svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#discourse-expand\"></use></svg>\n</div></a></div></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image some.png png https://meta.discourse.org/some.png discourse org/some png some.png png 1750×2000 ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image ``` In terms of indexing performance, we now have to parse the given HTML through nokogiri twice. However performance is not a huge worry here since a string length of 194170 takes only 30ms to scrub plus the indexing takes place in a background job.
2019-03-31 22:14:29 -04:00
def start_element(_name, attributes = [])
2013-02-05 14:16:51 -05:00
attributes = Hash[*attributes.flatten]
PERF: Improve quality of `PostSearchData#raw_data`. (#7275) This commit fixes the follow quality issue with `PostSearchData#raw_data`: 1. URLs are being tokenized and links with similar href and characters are being duplicated in the raw data. `Post#cooked`: ``` <p><a href=\"https://meta.discourse.org/some.png\" class=\"onebox\" target=\"_blank\" rel=\"nofollow noopener\">https://meta.discourse.org/some.png</a></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png discourse org/some png https://meta.discourse.org/some.png discourse org/some png ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized https://meta.discourse.org/some.png meta discourse org ``` 2. Ligthbox being included in search pollutes the `PostSearchData#raw_data` unncessarily. From 28 March 2018 to 28 March 2019, searches for the term `image` on `meta.discourse.org` had a click through rate of 2.1%. Non-lightboxed images are not included in indexing for search yet we were indexing content within a lightbox. Also, search for terms like `image` was affected we were using `Pasted image` as the filename for uploads that were pasted. `Post#cooked` ``` <p>Let me see how I can fix this image<br>\n<div class=\"lightbox-wrapper\"><a class=\"lightbox\" href=\"https://meta.discourse.org/some.png\" title=\"some.png\" rel=\"nofollow noopener\"><img src=\"https://meta.discourse.org/some.png\" width=\"275\" height=\"299\"><div class=\"meta\">\n<svg class=\"fa d-icon d-icon-far-image svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#far-image\"></use></svg><span class=\"filename\">some.png</span><span class=\"informations\">1750×2000</span><svg class=\"fa d-icon d-icon-discourse-expand svg-icon\" aria-hidden=\"true\"><use xlink:href=\"#discourse-expand\"></use></svg>\n</div></a></div></p> ``` `PostSearchData#raw_data` Before: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image some.png png https://meta.discourse.org/some.png discourse org/some png some.png png 1750×2000 ``` `PostSearchData#raw_data` After: ``` This is a test topic 0 Uncategorized Let me see how I can fix this image ``` In terms of indexing performance, we now have to parse the given HTML through nokogiri twice. However performance is not a huge worry here since a string length of 194170 takes only 30ms to scrub plus the indexing takes place in a background job.
2019-03-31 22:14:29 -04:00
ATTRIBUTES.each do |attribute_name|
if attributes[attribute_name].present? &&
!(
attribute_name == "href" &&
UrlHelper.is_local(attributes[attribute_name])
)
characters(attributes[attribute_name])
2018-09-13 12:53:53 -04:00
end
2018-08-19 20:39:19 -04:00
end
2013-02-05 14:16:51 -05:00
end
def characters(str)
str = Search.strip_diacritics(str) if @strip_diacritics
scrubbed << " #{str} "
2013-02-05 14:16:51 -05:00
end
end
end