class PostImportPreviewer FIELDS = [ 'title', 'thumbnail_base', 'original_created_from', 'original_created_before', 'duration', 'tags', 'parent_post_ids', ].freeze def initialize existing: 'skip' @existing = existing == 'error' ? 'error' : 'skip' end 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) 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_warnings = [] 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 normal_url.present? && existing_post if @existing == 'error' validation_errors[:url] = ['既存投稿です.'] else validation_warnings << '既存投稿のためスキップします.' end end if row[:metadata_url].present? && row[:metadata_url] != url_for_metadata clear_automatic_values!(attributes, provenance, tag_sources) end fetch_warnings = Array(row[:fetch_warnings]).dup should_fetch = should_fetch_metadata?(fetch_metadata, row[:source_row].to_i) if should_fetch && !(existing_post && @existing == 'skip') metadata, fetch_warnings = metadata_for(url_for_metadata, metadata_cache, fetch_warnings) apply_metadata!(attributes, provenance, tag_sources, metadata) end attributes['url'] = url validate_basic_data(attributes, validation_errors) validate_preview_tags( merged_tags(tag_sources, provenance['tags']), validation_errors, validation_warnings, ) validate_parents(attributes['parent_post_ids'], validation_errors) attributes.delete('url') attributes['tags'] = merged_tags(tag_sources, provenance['tags']) { source_row: row[:source_row], url: normal_url || url, attributes:, provenance:, tag_sources:, metadata_url: url_for_metadata, fetch_warnings:, validation_warnings:, warnings: fetch_warnings + validation_warnings, validation_errors:, status: validation_errors.present? \ ? 'error' \ : ((fetch_warnings + validation_warnings).present? ? 'warning' : 'ready'), existing: existing_post, } 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 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 metadata_for url, cache, previous_warnings cache[url] ||= fetch_metadata(url) end def fetch_metadata url return [{}, ['URL が空です.']] if url.blank? data = PostMetadataFetcher.fetch(url).stringify_keys warnings = [] warnings << 'タイトルを取得できませんでした.' if data['title'].blank? warnings << 'サムネールを取得できませんでした.' if data['thumbnail_base'].blank? [data.compact, 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 }", ) [{}, ['メタデータを取得できませんでした.']] 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].blank? attributes[field] = value provenance[field] = 'automatic' end 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, 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 warnings << "新規タグを作成します: #{ new_tags.join(' ') }" if new_tags.present? 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