github-mirror/discourse
github-mirror/discourse
2
0
Fork 0
mirror of https://github.com/discourse/discourse.git synced 2025-03-27 03:25:40 +08:00
Code Issues Actions1 Packages Projects Releases Wiki Activity

DEV: Improve import of attachments from Telligent

Browse Source
This commit is contained in:
Gerhard Schlager 2020-03-26 16:33:38 +01:00
parent d21d80198c
commit c94b63bc75
1 changed files with 9 additions and 3 deletions

12
script/import_scripts/telligent.rb

@ -42,8 +42,13 @@ class ImportScripts::Telligent < ImportScripts::Base
BATCH_SIZE ||= 1000
LOCAL_AVATAR_REGEX ||= /\A~\/.*(?<directory>communityserver-components-(?:selectable)?avatars)\/(?<path>[^\/]+)\/(?<filename>.+)/i
REMOTE_AVATAR_REGEX ||= /\Ahttps?:\/\//i
EMBEDDED_ATTACHMENT_REGEX ||= /<a href="\/cfs-file(?:\.ashx)?\/__key\/(?<directory>[^\/]+)\/(?<path>[^\/]+)\/(?<filename>.+?)".*?>.*?<\/a>/i
EMBEDDED_VIEW_ATTACHMENT_REGEX ||= /\[View:~\/cfs-file(?:\.ashx)?\/__key\/(?<directory>[^\/]+)\/(?<path>[^\/]+)\/(?<filename>.+?)(?:\:[:\d\s]*?)\]/i
ATTACHMENT_REGEXES ||= [
/<a[^>]*\shref="[^"]*?\/cfs-file(?:systemfile)?(?:\.ashx)?\/__key\/(?<directory>[^\/]+)\/(?<path>[^\/]+)\/(?<filename>.+?)".*?>.*?<\/a>/i,
/<img[^>]*\ssrc="[^"]*?\/cfs-file(?:systemfile)?(?:\.ashx)?\/__key\/(?<directory>[^\/]+)\/(?<path>[^\/]+)\/(?<filename>.+?)".*?>/i,
/\[View:[^\]]*?\/cfs-file(?:systemfile)?(?:\.ashx)?\/__key\/(?<directory>[^\/]+)\/(?<path>[^\/]+)\/(?<filename>.+?)(?:\:[:\d\s]*?)?\]/i,
/\[(?<tag>img|url)\][^\[]*?cfs-file(?:systemfile)?(?:\.ashx)?\/__key\/(?<directory>[^\/]+)\/(?<path>[^\/]+)\/(?<filename>.+?)\[\/\k<tag>\]/i,
/\[(?<tag>img|url)=[^\[]*?cfs-file(?:systemfile)?(?:\.ashx)?\/__key\/(?<directory>[^\/]+)\/(?<path>[^\/]+)\/(?<filename>.+?)\][^\[]*?\[\/\k<tag>\]/i
]
PROPERTY_NAMES_REGEX ||= /(?<name>\w+):S:(?<start>\d+):(?<length>\d+):/
CATEGORY_LINK_NORMALIZATION = '/.*?(f\/\d+)$/\1'
@ -560,7 +565,7 @@ class ImportScripts::Telligent < ImportScripts::Base
return [raw, paths, upload_ids] if @filestore_root_directory.blank?
[EMBEDDED_ATTACHMENT_REGEX, EMBEDDED_VIEW_ATTACHMENT_REGEX].each do |regex|
ATTACHMENT_REGEXES.each do |regex|
raw = raw.gsub(regex) do
match_data = Regexp.last_match
@ -579,6 +584,7 @@ class ImportScripts::Telligent < ImportScripts::Base
else
path = File.join(path, match_data[:filename])
STDERR.puts "Could not find file: #{path}"
match_data[0]
end
end
end