forked from OpenNeo/impress
Fix inconsistent item string encoding
We discovered a previous string encoding bug fix that was causing crashes for some items, was casuing *other* items to get reencoded incorrectly. In this change, we make the reencoding conditional, only if parsing as UTF-8 is failing. We also include a temporary repair script, to run in production then delete—but held here in git history for posterity.
This commit is contained in:
parent
7430e12655
commit
83281591b3
2 changed files with 166 additions and 19 deletions
|
|
@ -30,24 +30,17 @@ module RocketAMFExtensions
|
||||||
raise RocketAMF::AMFError.new(first_message_data)
|
raise RocketAMF::AMFError.new(first_message_data)
|
||||||
end
|
end
|
||||||
|
|
||||||
# HACK: It seems to me that these messages come back with Windows-1250
|
# HACK: Older items in Neopets' database have Windows-1250 encoding,
|
||||||
# (or similar) encoding on the strings? I'm basing this on the
|
# while newer items use proper UTF-8. We detect which encoding was used
|
||||||
# Patchwork Staff item, whose description arrives as:
|
# by checking if the string is valid UTF-8, and only re-encode if needed.
|
||||||
#
|
#
|
||||||
# "That staff is cute, but dont use it as a walking stick \x96 I " +
|
# Example of Windows-1250 item: Patchwork Staff (57311), whose
|
||||||
# "dont think it will hold you up!"
|
# description contains byte 0x96 (en-dash in Windows-1250).
|
||||||
#
|
#
|
||||||
# And the `\x96` is meant to represent an endash, which it doesn't in
|
# Example of UTF-8 item: Carnival Party Décor (80042), whose name
|
||||||
# UTF-8 or in most extended ASCII encodings, but *does* in Windows's
|
# contains proper UTF-8 bytes [195, 169] for the é character.
|
||||||
# specific extended ASCII.
|
|
||||||
#
|
|
||||||
# Idk if this is something to do with the AMFPHP spec or how the AMFPHP
|
|
||||||
# server code they use serializes strings (I couldn't find any
|
|
||||||
# reference to it?), or just their internal database encoding being
|
|
||||||
# passed along as-is, or what? But this seems to be the most correct
|
|
||||||
# interpretation I know how to do, so, let's do it!
|
|
||||||
result.messages[0].data.body.tap do |body|
|
result.messages[0].data.body.tap do |body|
|
||||||
reencode_strings! body, "Windows-1250", "UTF-8"
|
reencode_strings_if_needed! body, "Windows-1250", "UTF-8"
|
||||||
end
|
end
|
||||||
end
|
end
|
||||||
|
|
||||||
|
|
@ -92,13 +85,17 @@ module RocketAMFExtensions
|
||||||
end
|
end
|
||||||
end
|
end
|
||||||
|
|
||||||
def reencode_strings!(target, from, to)
|
def reencode_strings_if_needed!(target, from, to)
|
||||||
if target.is_a? String
|
if target.is_a? String
|
||||||
target.force_encoding(from).encode!(to)
|
# Only re-encode if the string is not valid UTF-8
|
||||||
|
# (indicating it's in the old Windows-1250 encoding)
|
||||||
|
unless target.valid_encoding?
|
||||||
|
target.force_encoding(from).encode!(to)
|
||||||
|
end
|
||||||
elsif target.is_a? Array
|
elsif target.is_a? Array
|
||||||
target.each { |x| reencode_strings!(x, from, to) }
|
target.each { |x| reencode_strings_if_needed!(x, from, to) }
|
||||||
elsif target.is_a? Hash
|
elsif target.is_a? Hash
|
||||||
target.values.each { |x| reencode_strings!(x, from, to) }
|
target.values.each { |x| reencode_strings_if_needed!(x, from, to) }
|
||||||
end
|
end
|
||||||
end
|
end
|
||||||
end
|
end
|
||||||
|
|
|
||||||
150
lib/tasks/fix_encoding.rake
Normal file
150
lib/tasks/fix_encoding.rake
Normal file
|
|
@ -0,0 +1,150 @@
|
||||||
|
namespace :db do
|
||||||
|
desc "Fix double-encoded UTF-8 strings in item names and descriptions"
|
||||||
|
task fix_double_encoding: :environment do
|
||||||
|
puts "=" * 80
|
||||||
|
puts "Fix Double-Encoded Strings in Database"
|
||||||
|
puts "=" * 80
|
||||||
|
puts
|
||||||
|
|
||||||
|
# Define the double-encoding patterns and their fixes
|
||||||
|
# Each pattern maps: double-encoded string -> correct UTF-8 string
|
||||||
|
# Using byte arrays to avoid encoding issues in the source file itself
|
||||||
|
encoding_fixes = {
|
||||||
|
# Common accented characters (Ă© => é, etc.)
|
||||||
|
"\xC4\x82\xC2\xA9".force_encoding('UTF-8') => "\xC3\xA9".force_encoding('UTF-8'), # é
|
||||||
|
"\xC4\x82\xC2\xB1".force_encoding('UTF-8') => "\xC3\xB1".force_encoding('UTF-8'), # ñ
|
||||||
|
"\xC4\x82\xC2\xAD".force_encoding('UTF-8') => "\xC3\xAD".force_encoding('UTF-8'), # í
|
||||||
|
"\xC4\x82\xC2\xA1".force_encoding('UTF-8') => "\xC3\xA1".force_encoding('UTF-8'), # á
|
||||||
|
"\xC4\x82\xC2\xB3".force_encoding('UTF-8') => "\xC3\xB3".force_encoding('UTF-8'), # ó
|
||||||
|
"\xC4\x82\xC2\xBA".force_encoding('UTF-8') => "\xC3\xBA".force_encoding('UTF-8'), # ú
|
||||||
|
|
||||||
|
# Smart quotes and apostrophes
|
||||||
|
"\xC3\xA2\xE2\x82\xAC\xE2\x84\xA2".force_encoding('UTF-8') => "\xE2\x80\x99".force_encoding('UTF-8'), # '
|
||||||
|
"\xC3\xA2\xE2\x82\xAC\xC5\x93".force_encoding('UTF-8') => "\xE2\x80\x9C".force_encoding('UTF-8'), # "
|
||||||
|
"\xC3\xA2\xE2\x82\xAC\xC2\x9D".force_encoding('UTF-8') => "\xE2\x80\x9D".force_encoding('UTF-8'), # "
|
||||||
|
"\xC3\xA2\xE2\x82\xAC\xCB\x9C".force_encoding('UTF-8') => "\xE2\x80\x98".force_encoding('UTF-8'), # '
|
||||||
|
|
||||||
|
# Other punctuation
|
||||||
|
"\xC3\xA2\xE2\x82\xAC\xE2\x80\x9C".force_encoding('UTF-8') => "\xE2\x80\x93".force_encoding('UTF-8'), # –
|
||||||
|
"\xC3\xA2\xE2\x82\xAC\xE2\x80\x9D".force_encoding('UTF-8') => "\xE2\x80\x94".force_encoding('UTF-8'), # —
|
||||||
|
"\xC3\xA2\xE2\x82\xAC\xC2\xA6".force_encoding('UTF-8') => "\xE2\x80\xA6".force_encoding('UTF-8'), # …
|
||||||
|
|
||||||
|
# Non-breaking space
|
||||||
|
"\xC3\x82\xC2\xA0".force_encoding('UTF-8') => "\xC2\xA0".force_encoding('UTF-8'),
|
||||||
|
}
|
||||||
|
|
||||||
|
puts "Will fix the following patterns:"
|
||||||
|
encoding_fixes.each do |bad, good|
|
||||||
|
puts " #{bad.inspect} → #{good.inspect}"
|
||||||
|
end
|
||||||
|
puts
|
||||||
|
|
||||||
|
# Find affected items by actually checking for the pattern in Ruby
|
||||||
|
# (MySQL LIKE queries give false positives with multi-byte UTF-8)
|
||||||
|
puts "Scanning items for double-encoding patterns..."
|
||||||
|
|
||||||
|
items_by_pattern = {}
|
||||||
|
total_affected = Set.new
|
||||||
|
count_by_pattern = Hash.new { |h, k| h[k] = { name: 0, description: 0 } }
|
||||||
|
|
||||||
|
Item.find_each do |item|
|
||||||
|
encoding_fixes.each_key do |pattern|
|
||||||
|
if item.name.include?(pattern)
|
||||||
|
items_by_pattern[pattern] ||= { name: [], description: [] }
|
||||||
|
items_by_pattern[pattern][:name] << item.id
|
||||||
|
total_affected << item.id
|
||||||
|
count_by_pattern[pattern][:name] += 1
|
||||||
|
end
|
||||||
|
|
||||||
|
if item.description.include?(pattern)
|
||||||
|
items_by_pattern[pattern] ||= { name: [], description: [] }
|
||||||
|
items_by_pattern[pattern][:description] << item.id
|
||||||
|
total_affected << item.id
|
||||||
|
count_by_pattern[pattern][:description] += 1
|
||||||
|
end
|
||||||
|
end
|
||||||
|
end
|
||||||
|
|
||||||
|
puts
|
||||||
|
count_by_pattern.each do |pattern, counts|
|
||||||
|
puts "#{pattern.inspect}: #{counts[:name]} names, #{counts[:description]} descriptions"
|
||||||
|
end
|
||||||
|
|
||||||
|
puts
|
||||||
|
puts "Total affected items: #{total_affected.size}"
|
||||||
|
puts
|
||||||
|
|
||||||
|
if total_affected.empty?
|
||||||
|
puts "No items need fixing!"
|
||||||
|
next
|
||||||
|
end
|
||||||
|
|
||||||
|
# Show some examples
|
||||||
|
puts "Example affected items:"
|
||||||
|
Item.where(id: total_affected.to_a.first(5)).each do |item|
|
||||||
|
puts " #{item.id}: #{item.name}"
|
||||||
|
end
|
||||||
|
puts " ... and #{total_affected.size - 5} more" if total_affected.size > 5
|
||||||
|
puts
|
||||||
|
|
||||||
|
# Ask for confirmation
|
||||||
|
print "Fix these items by replacing double-encoded characters? (y/N): "
|
||||||
|
response = STDIN.gets.chomp
|
||||||
|
unless response.downcase == 'y'
|
||||||
|
puts "Aborted."
|
||||||
|
next
|
||||||
|
end
|
||||||
|
|
||||||
|
puts
|
||||||
|
puts "Fixing items..."
|
||||||
|
puts "-" * 80
|
||||||
|
|
||||||
|
fixed_count = 0
|
||||||
|
no_change_count = 0
|
||||||
|
|
||||||
|
Item.where(id: total_affected.to_a).find_each do |item|
|
||||||
|
original_name = item.name
|
||||||
|
original_description = item.description
|
||||||
|
|
||||||
|
# Apply all fixes to name and description
|
||||||
|
new_name = original_name.dup
|
||||||
|
new_description = original_description.dup
|
||||||
|
|
||||||
|
encoding_fixes.each do |bad, good|
|
||||||
|
new_name.gsub!(bad, good)
|
||||||
|
new_description.gsub!(bad, good)
|
||||||
|
end
|
||||||
|
|
||||||
|
# Only save if something changed
|
||||||
|
if new_name != original_name || new_description != original_description
|
||||||
|
item.name = new_name
|
||||||
|
item.description = new_description
|
||||||
|
item.save!(validate: false) # Skip validations to avoid potential issues
|
||||||
|
|
||||||
|
if new_name != original_name
|
||||||
|
puts "#{item.id}: #{original_name.inspect} → #{new_name.inspect}"
|
||||||
|
elsif new_description != original_description
|
||||||
|
puts "#{item.id}: Updated description only"
|
||||||
|
end
|
||||||
|
|
||||||
|
fixed_count += 1
|
||||||
|
else
|
||||||
|
no_change_count += 1
|
||||||
|
end
|
||||||
|
end
|
||||||
|
|
||||||
|
puts
|
||||||
|
puts "-" * 80
|
||||||
|
puts "Complete!"
|
||||||
|
puts " ✓ Fixed: #{fixed_count}"
|
||||||
|
puts " ⊘ No changes needed: #{no_change_count}"
|
||||||
|
puts
|
||||||
|
|
||||||
|
# Show a sample of fixed items
|
||||||
|
puts "Sample of fixed items:"
|
||||||
|
Item.where(id: total_affected.to_a.first(5)).each do |item|
|
||||||
|
puts " #{item.id}: #{item.name}"
|
||||||
|
end
|
||||||
|
puts
|
||||||
|
end
|
||||||
|
end
|
||||||
Loading…
Reference in a new issue