266 行
9.3 KiB
Ruby
266 行
9.3 KiB
Ruby
class PostImportPreviewer
|
|
FIELDS = [
|
|
'title',
|
|
'thumbnail_base',
|
|
'original_created_from',
|
|
'original_created_before',
|
|
'duration',
|
|
'tags',
|
|
'parent_post_ids',
|
|
].freeze
|
|
FETCH_WARNING_FIELDS = ['url', 'title', 'thumbnail_base'].freeze
|
|
EXISTING_SKIP_WARNING = '既存投稿のためスキップします．'.freeze
|
|
TITLE_FETCH_WARNING = 'タイトルを取得できませんでした．'.freeze
|
|
THUMBNAIL_FETCH_WARNING = 'サムネールを取得できませんでした．'.freeze
|
|
METADATA_FETCH_WARNING = 'メタデータを取得できませんでした．'.freeze
|
|
|
|
def preview_rows rows:, fetch_metadata: true, metadata_cache: { }
|
|
urls = {}
|
|
rows.map { |row|
|
|
preview_row(row, urls:, fetch_metadata:, metadata_cache:)
|
|
}
|
|
end
|
|
|
|
def normalised_url value
|
|
PostUrlNormaliser.normalise(value)
|
|
end
|
|
|
|
private
|
|
|
|
def preview_row row, urls:, fetch_metadata:, metadata_cache:
|
|
row = row.symbolize_keys
|
|
attributes = initial_attributes(row)
|
|
provenance = initial_provenance(row)
|
|
tag_sources = initial_tag_sources(row, attributes, provenance)
|
|
field_warnings = initial_field_warnings(row)
|
|
base_warnings = initial_base_warnings(row)
|
|
url = row[:url].to_s.strip
|
|
provenance['url'] = 'manual'
|
|
normal_url = normalised_url(url)
|
|
url_for_metadata = normal_url || url
|
|
existing_post = normal_url.present? && Post.exists?(url: normal_url)
|
|
|
|
validation_errors = {}
|
|
validation_errors[:url] = ['URL が不正です．'] if normal_url.blank?
|
|
validation_errors[:url] = ['URL が重複しています．'] if normal_url.present? && urls.key?(normal_url)
|
|
urls[normal_url] = true if normal_url.present?
|
|
|
|
if row[:metadata_url].present? && row[:metadata_url] != url_for_metadata
|
|
clear_automatic_values!(attributes, provenance, tag_sources)
|
|
clear_fetch_warnings!(field_warnings)
|
|
end
|
|
|
|
if normal_url.present? && existing_post
|
|
add_field_warning!(field_warnings, 'url', EXISTING_SKIP_WARNING)
|
|
attributes['tags'] = merged_tags(tag_sources, provenance['tags'])
|
|
warnings_present = field_warnings.values.any?(&:present?) || base_warnings.present?
|
|
return {
|
|
source_row: row[:source_row],
|
|
url: normal_url,
|
|
attributes:,
|
|
provenance:,
|
|
tag_sources:,
|
|
metadata_url: url_for_metadata,
|
|
field_warnings:,
|
|
base_warnings:,
|
|
validation_errors:,
|
|
status: validation_errors.present? ? 'error' : (warnings_present ? 'warning' : 'ready'),
|
|
}
|
|
end
|
|
|
|
should_fetch = should_fetch_metadata?(fetch_metadata, row[:source_row].to_i)
|
|
if should_fetch
|
|
clear_fetch_warnings!(field_warnings)
|
|
metadata = metadata_for(url_for_metadata, metadata_cache)
|
|
apply_metadata!(attributes, provenance, tag_sources, metadata[:data])
|
|
apply_fetch_warnings!(field_warnings, metadata[:warnings])
|
|
end
|
|
|
|
attributes['url'] = url
|
|
validate_basic_data(attributes, validation_errors)
|
|
validate_preview_tags(merged_tags(tag_sources, provenance['tags']),
|
|
validation_errors,
|
|
field_warnings)
|
|
validate_parents(attributes['parent_post_ids'], validation_errors)
|
|
attributes.delete('url')
|
|
attributes['tags'] = merged_tags(tag_sources, provenance['tags'])
|
|
|
|
warnings_present = field_warnings.values.any?(&:present?) || base_warnings.present?
|
|
{
|
|
source_row: row[:source_row],
|
|
url: normal_url || url,
|
|
attributes:,
|
|
provenance:,
|
|
tag_sources:,
|
|
metadata_url: url_for_metadata,
|
|
field_warnings:,
|
|
base_warnings:,
|
|
validation_errors:,
|
|
status: validation_errors.present? ? 'error' : (warnings_present ? 'warning' : 'ready'),
|
|
}
|
|
end
|
|
|
|
def should_fetch_metadata? fetch_metadata, source_row
|
|
case fetch_metadata
|
|
when true then true
|
|
when Integer then fetch_metadata == source_row
|
|
when false, nil then false
|
|
else raise ArgumentError, 'メタデータ取得対象が不正です．'
|
|
end
|
|
end
|
|
|
|
def initial_attributes row
|
|
attributes = row[:attributes]&.stringify_keys || { }
|
|
FIELDS.to_h { |field| [field, attributes[field].to_s] }
|
|
end
|
|
|
|
def initial_provenance row
|
|
provenance = row[:provenance]&.stringify_keys || { }
|
|
FIELDS.to_h { |field| [field, provenance[field].presence || 'automatic'] }
|
|
.merge('url' => 'manual')
|
|
end
|
|
|
|
def initial_tag_sources row, attributes, provenance
|
|
sources = row[:tag_sources]&.stringify_keys || { 'automatic' => '', 'manual' => '' }
|
|
sources['automatic'] = sources['automatic'].to_s
|
|
sources['manual'] =
|
|
provenance['tags'] == 'manual' ? attributes['tags'].to_s : sources['manual'].to_s
|
|
sources
|
|
end
|
|
|
|
def initial_field_warnings row
|
|
(row[:field_warnings] || { }).stringify_keys.transform_values { |value| Array(value).map(&:to_s) }
|
|
end
|
|
|
|
def initial_base_warnings row
|
|
Array(row[:base_warnings]).map(&:to_s)
|
|
end
|
|
|
|
def clear_automatic_values! attributes, provenance, tag_sources
|
|
['title', 'thumbnail_base', 'original_created_from',
|
|
'original_created_before', 'duration'].each do |field|
|
|
attributes[field] = '' if provenance[field] == 'automatic'
|
|
end
|
|
tag_sources['automatic'] = ''
|
|
attributes['tags'] = merged_tags(tag_sources, provenance['tags'])
|
|
end
|
|
|
|
def clear_fetch_warnings! field_warnings
|
|
FETCH_WARNING_FIELDS.each do |field|
|
|
field_warnings.delete(field)
|
|
end
|
|
end
|
|
|
|
def metadata_for url, cache
|
|
cache[url] ||= fetch_metadata(url)
|
|
end
|
|
|
|
def fetch_metadata url
|
|
return { data: { }, warnings: { 'url' => ['URL が空です．'] } } if url.blank?
|
|
|
|
data = PostMetadataFetcher.fetch(url).stringify_keys.compact
|
|
warnings = { }
|
|
add_field_warning!(warnings, 'title', TITLE_FETCH_WARNING) if data['title'].blank?
|
|
add_field_warning!(warnings, 'thumbnail_base', THUMBNAIL_FETCH_WARNING) if data['thumbnail_base'].blank?
|
|
{ data:, warnings: }
|
|
rescue Preview::UrlSafety::UnsafeUrl,
|
|
Preview::HttpFetcher::FetchFailed,
|
|
Preview::HttpFetcher::ResponseTooLarge => e
|
|
Rails.logger.info(
|
|
"post_import_metadata_fetch_failure "\
|
|
"#{ { error: e.class.name, message: e.message }.to_json }",
|
|
)
|
|
{ data: { }, warnings: { 'url' => [METADATA_FETCH_WARNING] } }
|
|
end
|
|
|
|
def apply_metadata! attributes, provenance, tag_sources, metadata
|
|
metadata.each do |field, value|
|
|
if field == 'tags'
|
|
next if provenance['tags'] == 'manual'
|
|
|
|
tag_sources['automatic'] = value.to_s
|
|
attributes['tags'] = merged_tags(tag_sources)
|
|
next
|
|
end
|
|
next unless provenance[field] == 'automatic'
|
|
|
|
attributes[field] = value
|
|
provenance[field] = 'automatic'
|
|
end
|
|
end
|
|
|
|
def apply_fetch_warnings! field_warnings, warnings
|
|
warnings.each do |field, values|
|
|
values.each do |value|
|
|
add_field_warning!(field_warnings, field, value)
|
|
end
|
|
end
|
|
end
|
|
|
|
def add_field_warning! field_warnings, field, message
|
|
field_warnings[field] ||= []
|
|
field_warnings[field] << message unless field_warnings[field].include?(message)
|
|
end
|
|
|
|
def merged_tags sources, origin = nil
|
|
return sources['manual'].to_s if origin == 'manual'
|
|
|
|
sources['automatic'].to_s
|
|
end
|
|
|
|
def validate_preview_tags raw, errors, field_warnings
|
|
names = raw.to_s.split
|
|
return if names.empty?
|
|
if names.any? { _1.downcase.start_with?('nico:') }
|
|
errors[:tags] = ['ニコニコ・タグは直接指定できません．']
|
|
return
|
|
end
|
|
|
|
parsed = names.map { |name| TagName.canonicalise(name.sub(/\[.*\]\z/, '')).first }
|
|
existing = Tag.joins(:tag_name).where(tag_names: { name: parsed }).includes(:tag_name).to_a
|
|
deprecated = existing.select(&:deprecated?).map(&:name)
|
|
errors[:tags] = ["廃止済みタグがあります: #{ deprecated.join(' ') }"] if deprecated.present?
|
|
known = existing.reject(&:deprecated?).map(&:name)
|
|
new_tags = parsed.uniq - known
|
|
if new_tags.present?
|
|
add_field_warning!(field_warnings, 'tags', "新規タグを作成します: #{ new_tags.join(' ') }")
|
|
end
|
|
rescue Tag::SectionLiteralParseError
|
|
errors[:tags] = ['タグ区間の記法が不正です．']
|
|
end
|
|
|
|
def validate_basic_data attributes, errors
|
|
post = Post.new(title: attributes['title'].presence,
|
|
url: attributes['url'],
|
|
thumbnail_base: attributes['thumbnail_base'].presence,
|
|
original_created_from: attributes['original_created_from'].presence,
|
|
original_created_before: attributes['original_created_before'].presence,
|
|
video_ms: parse_duration(attributes['duration'], errors))
|
|
post.valid?
|
|
post.errors.each do |error|
|
|
next if error.attribute == :url && error.type == :taken
|
|
|
|
(errors[error.attribute] ||= []) << error.message
|
|
end
|
|
end
|
|
|
|
def parse_duration value, errors
|
|
return nil if value.blank?
|
|
|
|
value.is_a?(Numeric) ? value.to_i : Tag.time_to_ms!(value.to_s, tag_name: '動画時間')
|
|
rescue Tag::SectionLiteralParseError
|
|
errors[:video_ms] = ['動画時間の記法が不正です．']
|
|
nil
|
|
end
|
|
|
|
def validate_parents raw, errors
|
|
ids = raw.to_s.split.map { Integer(_1, exception: false) }
|
|
return if ids.compact.empty? && raw.to_s.blank?
|
|
|
|
errors[:parent_post_ids] = ['親投稿 Id. が不正です．'] and return if ids.any? { _1.nil? || _1 <= 0 }
|
|
if Post.where(id: ids).count != ids.uniq.length
|
|
errors[:parent_post_ids] = ['存在しない親投稿 Id. があります．']
|
|
end
|
|
end
|
|
end
|