debian-mirror-gitlab/lib/gitlab/i18n/po_linter.rb

304 lines
9.8 KiB
Ruby
Raw Normal View History

2019-02-15 15:39:39 +05:30
# frozen_string_literal: true
2018-03-17 18:26:18 +05:30
module Gitlab
module I18n
class PoLinter
2018-11-08 19:23:39 +05:30
include Gitlab::Utils::StrongMemoize
2020-10-24 23:57:45 +05:30
attr_reader :po_path, :translation_entries, :metadata_entry, :locale, :html_todolist
2018-03-17 18:26:18 +05:30
VARIABLE_REGEX = /%{\w*}|%[a-z]/.freeze
2020-10-24 23:57:45 +05:30
def initialize(po_path:, html_todolist:, locale: I18n.locale.to_s)
2018-03-17 18:26:18 +05:30
@po_path = po_path
@locale = locale
2020-10-24 23:57:45 +05:30
@html_todolist = html_todolist
2018-03-17 18:26:18 +05:30
end
def errors
@errors ||= validate_po
end
def validate_po
2020-10-24 23:57:45 +05:30
if (parse_error = parse_po)
2018-03-17 18:26:18 +05:30
return 'PO-syntax errors' => [parse_error]
end
validate_entries
end
def parse_po
entries = SimplePoParser.parse(po_path)
# The first entry is the metadata entry if there is one.
# This is an entry when empty `msgid`
if entries.first[:msgid].empty?
@metadata_entry = Gitlab::I18n::MetadataEntry.new(entries.shift)
else
return 'Missing metadata entry.'
end
@translation_entries = entries.map do |entry_data|
2020-10-24 23:57:45 +05:30
Gitlab::I18n::TranslationEntry.new(
entry_data: entry_data,
nplurals: metadata_entry.expected_forms,
html_allowed: html_todolist.fetch(entry_data[:msgid], false)
)
2018-03-17 18:26:18 +05:30
end
nil
rescue SimplePoParser::ParserError => e
@translation_entries = []
e.message
end
def validate_entries
errors = {}
translation_entries.each do |entry|
errors_for_entry = validate_entry(entry)
2018-11-08 19:23:39 +05:30
errors[entry.msgid] = errors_for_entry if errors_for_entry.any?
2018-03-17 18:26:18 +05:30
end
errors
end
def validate_entry(entry)
errors = []
validate_flags(errors, entry)
validate_variables(errors, entry)
validate_newlines(errors, entry)
validate_number_of_plurals(errors, entry)
validate_unescaped_chars(errors, entry)
2020-10-24 23:57:45 +05:30
validate_html(errors, entry)
2018-11-08 19:23:39 +05:30
validate_translation(errors, entry)
2018-03-17 18:26:18 +05:30
errors
end
def validate_unescaped_chars(errors, entry)
if entry.msgid_contains_unescaped_chars?
errors << 'contains unescaped `%`, escape it using `%%`'
end
if entry.plural_id_contains_unescaped_chars?
errors << 'plural id contains unescaped `%`, escape it using `%%`'
end
if entry.translations_contain_unescaped_chars?
errors << 'translation contains unescaped `%`, escape it using `%%`'
end
end
2020-10-24 23:57:45 +05:30
def validate_html(errors, entry)
common_message = 'contains < or >. Use variables to include HTML in the string, or the &lt; and &gt; codes ' \
'for the symbols. For more info see: https://docs.gitlab.com/ee/development/i18n/externalization.html#html'
if entry.msgid_contains_potential_html? && !entry.msgid_html_allowed?
errors << common_message
end
if entry.plural_id_contains_potential_html? && !entry.plural_id_html_allowed?
errors << 'plural id ' + common_message
end
if entry.translations_contain_potential_html? && !entry.translations_html_allowed?
errors << 'translation ' + common_message
end
end
2018-03-17 18:26:18 +05:30
def validate_number_of_plurals(errors, entry)
2018-11-08 19:23:39 +05:30
return unless metadata_entry&.expected_forms
2018-03-17 18:26:18 +05:30
return unless entry.translated?
2018-11-08 19:23:39 +05:30
if entry.has_plural? && entry.all_translations.size != metadata_entry.expected_forms
errors << "should have #{metadata_entry.expected_forms} "\
"#{'translations'.pluralize(metadata_entry.expected_forms)}"
2018-03-17 18:26:18 +05:30
end
end
def validate_newlines(errors, entry)
2018-11-08 19:23:39 +05:30
if entry.msgid_has_multiple_lines?
2018-03-17 18:26:18 +05:30
errors << 'is defined over multiple lines, this breaks some tooling.'
end
2018-11-08 19:23:39 +05:30
if entry.plural_id_has_multiple_lines?
2018-03-17 18:26:18 +05:30
errors << 'plural is defined over multiple lines, this breaks some tooling.'
end
2018-11-08 19:23:39 +05:30
if entry.translations_have_multiple_lines?
2018-03-17 18:26:18 +05:30
errors << 'has translations defined over multiple lines, this breaks some tooling.'
end
end
def validate_variables(errors, entry)
if entry.has_singular_translation?
2018-11-08 19:23:39 +05:30
validate_variables_in_message(errors, entry.msgid, entry.msgid)
2018-03-17 18:26:18 +05:30
validate_variables_in_message(errors, entry.msgid, entry.singular_translation)
end
if entry.has_plural?
2018-11-08 19:23:39 +05:30
validate_variables_in_message(errors, entry.plural_id, entry.plural_id)
2018-03-17 18:26:18 +05:30
entry.plural_translations.each do |translation|
validate_variables_in_message(errors, entry.plural_id, translation)
end
end
end
def validate_variables_in_message(errors, message_id, message_translation)
required_variables = message_id.scan(VARIABLE_REGEX)
validate_unnamed_variables(errors, required_variables)
validate_variable_usage(errors, message_translation, required_variables)
end
2018-11-08 19:23:39 +05:30
def validate_translation(errors, entry)
Gitlab::I18n.with_locale(locale) do
if entry.has_plural?
translate_plural(entry)
else
translate_singular(entry)
end
end
# `sprintf` could raise an `ArgumentError` when invalid passing something
# other than a Hash when using named variables
#
# `sprintf` could raise `TypeError` when passing a wrong type when using
# unnamed variables
#
# FastGettext::Translation could raise `RuntimeError` (raised as a string),
# or as subclassess `NoTextDomainConfigured` & `InvalidFormat`
#
# `FastGettext::Translation` could raise `ArgumentError` as subclassess
# `InvalidEncoding`, `IllegalSequence` & `InvalidCharacter`
rescue ArgumentError, TypeError, RuntimeError => e
errors << "Failure translating to #{locale}: #{e.message}"
end
def translate_singular(entry)
used_variables = entry.msgid.scan(VARIABLE_REGEX)
2018-03-17 18:26:18 +05:30
variables = fill_in_variables(used_variables)
2018-11-08 19:23:39 +05:30
translation = if entry.msgid.include?('|')
FastGettext::Translation.s_(entry.msgid)
else
FastGettext::Translation._(entry.msgid)
end
2018-03-17 18:26:18 +05:30
2018-11-08 19:23:39 +05:30
translation % variables if used_variables.any?
end
def translate_plural(entry)
numbers_covering_all_plurals.map do |number|
translation = FastGettext::Translation.n_(entry.msgid, entry.plural_id, number)
2019-12-04 20:38:33 +05:30
index = index_for_pluralization(number)
used_variables = index == 0 ? entry.msgid.scan(VARIABLE_REGEX) : entry.plural_id.scan(VARIABLE_REGEX)
variables = fill_in_variables(used_variables)
2018-11-08 19:23:39 +05:30
2019-12-04 20:38:33 +05:30
translation % variables if variables.any?
2018-11-08 19:23:39 +05:30
end
end
def numbers_covering_all_plurals
@numbers_covering_all_plurals ||= calculate_numbers_covering_all_plurals
end
def calculate_numbers_covering_all_plurals
required_numbers = []
discovered_indexes = []
counter = 0
while discovered_indexes.size < metadata_entry.forms_to_test && counter < Gitlab::I18n::MetadataEntry::MAX_FORMS_TO_TEST
index_for_count = index_for_pluralization(counter)
unless discovered_indexes.include?(index_for_count)
discovered_indexes << index_for_count
required_numbers << counter
2018-03-17 18:26:18 +05:30
end
2018-11-08 19:23:39 +05:30
counter += 1
end
required_numbers
end
def index_for_pluralization(counter)
# This calls the C function that defines the pluralization rule, it can
# return a boolean (`false` represents 0, `true` represents 1) or an integer
# that specifies the plural form to be used for the given number
pluralization_result = Gitlab::I18n.with_locale(locale) do
FastGettext.pluralisation_rule.call(counter)
end
case pluralization_result
when false
0
when true
1
else
pluralization_result
2018-03-17 18:26:18 +05:30
end
end
def fill_in_variables(variables)
if variables.empty?
[]
elsif variables.any? { |variable| unnamed_variable?(variable) }
variables.map do |variable|
variable == '%d' ? Random.rand(1000) : Gitlab::Utils.random_string
end
else
variables.inject({}) do |hash, variable|
variable_name = variable[/\w+/]
hash[variable_name] = Gitlab::Utils.random_string
hash
end
end
end
def validate_unnamed_variables(errors, variables)
2018-11-08 19:23:39 +05:30
unnamed_variables, named_variables = variables.partition { |name| unnamed_variable?(name) }
if unnamed_variables.any? && named_variables.any?
errors << 'is combining named variables with unnamed variables'
end
if unnamed_variables.size > 1
2018-03-17 18:26:18 +05:30
errors << 'is combining multiple unnamed variables'
end
end
def validate_variable_usage(errors, translation, required_variables)
# We don't need to validate when the message is empty.
2019-02-15 15:39:39 +05:30
# In this case we fall back to the default, which has all the
2018-03-17 18:26:18 +05:30
# required variables.
return if translation.empty?
found_variables = translation.scan(VARIABLE_REGEX)
missing_variables = required_variables - found_variables
if missing_variables.any?
errors << "<#{translation}> is missing: [#{missing_variables.to_sentence}]"
end
unknown_variables = found_variables - required_variables
if unknown_variables.any?
errors << "<#{translation}> is using unknown variables: [#{unknown_variables.to_sentence}]"
end
end
def unnamed_variable?(variable_name)
!variable_name.start_with?('%{')
end
def validate_flags(errors, entry)
errors << "is marked #{entry.flag}" if entry.flag
end
end
end
end