249 行
8.9 KiB
Ruby
249 行
8.9 KiB
Ruby
class PostImportPreviewer
|
|
FIELDS = [
|
|
'title',
|
|
'thumbnail_base',
|
|
'original_created_from',
|
|
'original_created_before',
|
|
'duration',
|
|
'tags',
|
|
'parent_post_ids',
|
|
].freeze
|
|
|
|
def initialize parsed:, url_column:, mappings: { }, existing: 'skip'
|
|
@parsed = parsed
|
|
@url_column = url_column.to_i
|
|
@mappings = mappings.stringify_keys
|
|
@existing = existing == 'error' ? 'error' : 'skip'
|
|
end
|
|
|
|
def preview
|
|
preview_rows(rows: @parsed[:rows])
|
|
end
|
|
|
|
def preview_rows rows:, fetch_metadata: true, metadata_cache: { }
|
|
urls = { }
|
|
rows.map do |row|
|
|
row = row.symbolize_keys
|
|
values = row[:values] || []
|
|
attributes = row[:attributes]&.stringify_keys || FIELDS.to_h { |field|
|
|
[field, mapped_value(field, values)]
|
|
}
|
|
provenance = row[:provenance]&.stringify_keys || mapped_provenance(attributes)
|
|
tag_sources =
|
|
row[:tag_sources]&.stringify_keys || initial_tag_sources(attributes, provenance)
|
|
tag_sources['manual'] = attributes['tags'].to_s if provenance['tags'] == 'manual'
|
|
url = row[:url].presence || values[@url_column].to_s.strip
|
|
provenance['url'] ||= row[:url].present? ? 'manual' : 'mapped'
|
|
url_for_metadata = normalised_url(url) || url
|
|
existing_post = normalised_url(url).present? && Post.exists?(url: normalised_url(url))
|
|
if existing_post && @existing == 'skip'
|
|
next {
|
|
source_row: row[:source_row],
|
|
url: normalised_url(url) || url,
|
|
attributes:,
|
|
provenance:,
|
|
tag_sources:,
|
|
metadata_url: url_for_metadata,
|
|
fetch_warnings: [],
|
|
validation_warnings: ['既存投稿のためスキップします．'],
|
|
warnings: ['既存投稿のためスキップします．'],
|
|
validation_errors: { },
|
|
status: 'warning',
|
|
existing: true,
|
|
}
|
|
end
|
|
url_changed = row[:metadata_url].present? && row[:metadata_url] != url_for_metadata
|
|
clear_automatic_values!(attributes, provenance, tag_sources) if url_changed
|
|
attributes['url'] = url
|
|
fetch_warnings = Array(row[:fetch_warnings]).dup
|
|
should_fetch =
|
|
case fetch_metadata
|
|
when true then true
|
|
when Integer then fetch_metadata == row[:source_row].to_i
|
|
when false, nil then false
|
|
else raise ArgumentError, 'メタデータ取得対象が不正です．'
|
|
end
|
|
metadata, fetch_warnings =
|
|
metadata_for(url_for_metadata, should_fetch, metadata_cache, fetch_warnings)
|
|
metadata.each do |field, value|
|
|
if field == 'tags' && provenance[field] != 'manual'
|
|
tag_sources['automatic'] = value.to_s
|
|
attributes[field] = merged_tags(tag_sources)
|
|
next
|
|
end
|
|
next unless provenance[field] == 'automatic' || attributes[field].blank?
|
|
|
|
attributes[field] = value
|
|
provenance[field] = 'automatic'
|
|
end
|
|
normal_url = normalised_url(url)
|
|
errors = { }
|
|
errors[:url] = ['URL が不正です．'] if normal_url.blank?
|
|
validation_warnings = []
|
|
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'
|
|
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
|
|
validate_basic_data(attributes, errors)
|
|
validate_preview_tags(
|
|
merged_tags(tag_sources, provenance['tags']),
|
|
errors,
|
|
validation_warnings,
|
|
)
|
|
validate_parents(attributes['parent_post_ids'], 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: errors,
|
|
status: errors.present? \
|
|
? 'error' \
|
|
: ((fetch_warnings + validation_warnings).present? ? 'warning' : 'ready'),
|
|
existing: normal_url.present? && Post.exists?(url: normal_url),
|
|
}
|
|
end
|
|
end
|
|
|
|
def metadata_for url, fetch_metadata, cache, previous_warnings
|
|
return [{ }, previous_warnings] unless fetch_metadata
|
|
|
|
cache[url] ||= fetch_metadata(url)
|
|
end
|
|
|
|
def normalised_url value
|
|
PostUrlNormaliser.normalise(value)
|
|
end
|
|
|
|
private
|
|
|
|
def mapped_value field, values
|
|
mapping = @mappings[field]
|
|
return '' if mapping.blank?
|
|
return mapping['value'].to_s if mapping['kind'] == 'fixed'
|
|
|
|
Array(mapping['columns']).filter_map { |index|
|
|
values[index.to_i].to_s.presence
|
|
}.join(field == 'tags' ? ' ' : '')
|
|
end
|
|
|
|
def mapped_provenance attributes
|
|
attributes.to_h.to_h { |field, value|
|
|
mapping = @mappings[field.to_s]
|
|
origin =
|
|
if value.present? && mapping&.[]('kind') == 'fixed'
|
|
'fixed'
|
|
elsif value.present?
|
|
'mapped'
|
|
else
|
|
'automatic'
|
|
end
|
|
[field.to_s, origin]
|
|
}
|
|
end
|
|
|
|
def initial_tag_sources attributes, provenance
|
|
origin = provenance['tags'] == 'fixed' ? 'fixed' : 'mapped'
|
|
{ 'automatic' => '', 'mapped' => origin == 'mapped' ? attributes['tags'].to_s : '',
|
|
'fixed' => origin == 'fixed' ? attributes['tags'].to_s : '', 'manual' => '' }
|
|
end
|
|
|
|
def merged_tags sources, origin = nil
|
|
return sources['manual'].to_s if origin == 'manual'
|
|
|
|
['automatic', 'mapped', 'fixed', 'manual'].flat_map {
|
|
sources[_1].to_s.split
|
|
}.uniq.join(' ')
|
|
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'] = ''
|
|
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 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
|