Merge remote-tracking branch 'origin/main' into feature/351

このコミットが含まれているのは:
2026-07-02 01:55:35 +09:00
コミット 46de995f8d
98個のファイルの変更6862行の追加1175行の削除
+233
ファイルの表示
@@ -0,0 +1,233 @@
require 'json'
require 'jwt'
require 'net/http'
require 'openssl'
require 'tempfile'
require 'uri'
module GoogleDrive
class ApiClient
DRIVE_ENDPOINT = 'https://www.googleapis.com/drive/v3'
TOKEN_ENDPOINT = 'https://oauth2.googleapis.com/token'
FOLDER_MIME_TYPE = 'application/vnd.google-apps.folder'
NATIVE_FILE_MIME_TYPE_PREFIX = 'application/vnd.google-apps.'
DRIVE_SCOPE = 'https://www.googleapis.com/auth/drive.readonly'
def initialize service_account_email: ENV['GOOGLE_DRIVE_SERVICE_ACCOUNT_EMAIL'],
private_key: ENV['GOOGLE_DRIVE_PRIVATE_KEY'],
private_key_path: ENV['GOOGLE_DRIVE_PRIVATE_KEY_PATH'],
subject: ENV['GOOGLE_DRIVE_SUBJECT']
@service_account_email = service_account_email.to_s
@private_key = load_private_key(private_key, private_key_path)
@subject = subject.to_s.presence
@access_token = nil
end
def list_material_files_under_folder folder_id
files = []
each_material_file_under_folder(folder_id) { |entry| files << entry }
files
end
def each_material_file_under_folder folder_id
return enum_for(__method__, folder_id) unless block_given?
walk_folder(folder_id, nil) do |entry, relative_path|
next if entry['mimeType'] == FOLDER_MIME_TYPE
next if native_file?(entry['mimeType'])
yield build_file_entry(entry, relative_path)
end
end
def fetch_material_file file_id
metadata = get_file(file_id)
return nil if native_file?(metadata['mimeType'])
build_file_entry(metadata, metadata['name'])
end
def download_to_tempfile file_id, filename:
tempfile = Tempfile.new(['material-sync', File.extname(filename.to_s)])
tempfile.binmode
request_binary("/files/#{ file_id }",
{ alt: 'media', supportsAllDrives: true }) do |chunk|
tempfile.write(chunk.b)
end
tempfile.rewind
tempfile
rescue StandardError
tempfile&.close!
raise
end
def extract_file_id value
raw = value.to_s.strip
return nil if raw.blank?
return raw unless raw.include?('/')
uri = URI.parse(raw)
return uri.query.to_s[%r{(?:^|&)id=([^&]+)}, 1] if uri.query.present?
return uri.path[%r{/folders/([^/]+)}, 1] if uri.path.include?('/folders/')
uri.path[%r{/d/([^/]+)}, 1]
rescue URI::InvalidURIError
nil
end
private
def walk_folder folder_id, prefix, &block
list_children(folder_id).each do |entry|
relative_path = MaterialSyncExportPath.build(prefix:,
relative_path: entry['name'])
if entry['mimeType'] == FOLDER_MIME_TYPE
walk_folder(entry['id'], relative_path, &block)
next
end
block.call(entry, relative_path)
end
end
def build_file_entry entry, relative_path
{ id: entry['id'],
name: entry['name'],
mime_type: entry['mimeType'],
relative_path: MaterialSyncExportPath.normalize_path(relative_path),
sha256_checksum: entry['sha256Checksum'],
web_view_link: entry['webViewLink'],
web_content_link: entry['webContentLink'] }
end
def list_children folder_id
files = []
page_token = nil
loop do
response =
request_json('/files', {
q: "'#{ folder_id }' in parents and trashed = false",
fields: 'nextPageToken,files(id,name,mimeType,sha256Checksum,' \
'webViewLink,webContentLink)',
orderBy: 'folder,name',
pageSize: 1000,
supportsAllDrives: true,
includeItemsFromAllDrives: true,
pageToken: page_token }.compact)
files.concat(response.fetch('files'))
page_token = response['nextPageToken']
break if page_token.blank?
end
files
end
def get_file file_id
request_json("/files/#{ file_id }",
fields: 'id,name,mimeType,sha256Checksum,webViewLink,webContentLink',
supportsAllDrives: true)
end
def request_json path, params = {}
response = request(:get, path, params:)
unless response.is_a?(Net::HTTPSuccess)
raise "Google Drive API error: #{ response.code } #{ response.body }"
end
JSON.parse(response.body)
end
def request_binary path, params = {}
request(:get, path, params:) do |response|
unless response.is_a?(Net::HTTPSuccess)
raise "Google Drive download error: #{ response.code } #{ response.body }"
end
response.read_body do |chunk|
yield chunk.b
end
end
end
def request method, path, params: nil
uri = URI(DRIVE_ENDPOINT + path)
uri.query = URI.encode_www_form(params) if params.present?
klass =
case method
when :get
Net::HTTP::Get
else
raise ArgumentError, "Unsupported Google Drive request method: #{ method }"
end
request = klass.new(uri)
request['Authorization'] = "Bearer #{ access_token }"
Net::HTTP.start(uri.host,
uri.port,
use_ssl: true,
open_timeout: 10,
read_timeout: 60) do |http|
if block_given?
http.request(request) do |response|
return yield response
end
else
http.request(request)
end
end
end
def access_token
@access_token ||= fetch_access_token
end
def fetch_access_token
raise 'GOOGLE_DRIVE_SERVICE_ACCOUNT_EMAIL is required' if @service_account_email.blank?
raise 'Google Drive private key is required' if @private_key.blank?
payload = { iss: @service_account_email,
scope: DRIVE_SCOPE,
aud: TOKEN_ENDPOINT,
exp: 1.hour.from_now.to_i,
iat: Time.current.to_i }
payload[:sub] = @subject if @subject.present?
assertion = JWT.encode(payload, @private_key, 'RS256')
uri = URI(TOKEN_ENDPOINT)
request = Net::HTTP::Post.new(uri)
request.set_form_data(
grant_type: 'urn:ietf:params:oauth:grant-type:jwt-bearer',
assertion:)
response =
Net::HTTP.start(uri.host,
uri.port,
use_ssl: true,
open_timeout: 10,
read_timeout: 30) do |http|
http.request(request)
end
unless response.is_a?(Net::HTTPSuccess)
raise "Google OAuth error: #{ response.code } #{ response.body }"
end
JSON.parse(response.body).fetch('access_token')
end
def load_private_key private_key, private_key_path
raw = private_key.to_s
raw = raw.gsub('\n', "\n") if raw.present?
raw = File.read(private_key_path) if raw.blank? && private_key_path.present?
return nil if raw.blank?
OpenSSL::PKey::RSA.new(raw)
end
def native_file? mime_type
mime_type.to_s.start_with?(NATIVE_FILE_MIME_TYPE_PREFIX)
end
end
end
+57
ファイルの表示
@@ -0,0 +1,57 @@
require 'digest'
require 'json'
class MaterialFileSha256
def self.blob_metadata blob
metadata = blob.metadata
return metadata if metadata.is_a?(Hash)
return { } if metadata.blank?
JSON.parse(metadata)
rescue JSON::ParserError
{ }
end
def self.metadata_sha256 blob
blob_metadata(blob)['sha256'].presence
end
def self.assign_metadata_sha256! blob, sha256
return if sha256.blank?
metadata = blob_metadata(blob)
metadata['sha256'] = sha256
blob.metadata = metadata
blob.save! if blob.changed?
end
def self.from_blob blob, allow_download: false
sha256 = metadata_sha256(blob)
return sha256 if sha256.present?
return nil unless allow_download
begin
blob.open do |file|
sha256 = Digest::SHA256.file(file.path).hexdigest
assign_metadata_sha256!(blob, sha256)
sha256
end
rescue ActiveStorage::FileNotFoundError, ArgumentError => error
Rails.logger.warn(
"MaterialFileSha256.from_blob failed for blob_id=#{blob.id}: " \
"#{error.class}: #{error.message}",
)
nil
end
end
def self.from_upload upload
tempfile = upload&.tempfile
return nil unless tempfile
tempfile.rewind
Digest::SHA256.file(tempfile.path).hexdigest.tap do
tempfile.rewind
end
end
end
+7
ファイルの表示
@@ -0,0 +1,7 @@
class MaterialImportBlockMatcher
def self.match_for_sha256 sha256
return nil if sha256.blank?
MaterialImportBlock.find_by(match_kind: 'sha256', sha256:)
end
end
+56
ファイルの表示
@@ -0,0 +1,56 @@
class MaterialSyncExportPath
class << self
def build prefix: nil, relative_path: nil, filename: nil, source_file_id: nil
path = normalize_path(relative_path)
path = fallback_filename(filename, source_file_id) if path.blank?
normalized_prefix = normalize_path(prefix)
return path if normalized_prefix.blank?
return normalized_prefix if path.blank?
[normalized_prefix, path].join('/')
end
def normalize_path path
raw = path.to_s.delete("\0").tr('\\', '/').strip
raw = raw.sub(%r{\A[A-Za-z]:/+}, '')
raw = raw.sub(%r{\A/+}, '')
return nil if raw.blank?
segments = raw.split('/').filter_map do |segment|
normalized = normalize_segment(segment)
normalized.presence
end
return nil if segments.empty?
segments.join('/')
end
def uniquify path, source_file_id
normalized = normalize_path(path)
return normalized if normalized.blank? || source_file_id.blank?
ext = File.extname(normalized)
base = ext.present? ? normalized.delete_suffix(ext) : normalized
"#{ base }--#{ source_file_id }#{ ext }"
end
private
def fallback_filename filename, source_file_id
name = normalize_path(filename)
return name if name.present?
return nil if source_file_id.blank?
source_file_id.to_s
end
def normalize_segment segment
cleaned = segment.to_s.delete("\0").tr('\\/', '_').strip
return nil if cleaned.blank? || cleaned == '.'
return '__' if cleaned == '..'
cleaned
end
end
end
+259
ファイルの表示
@@ -0,0 +1,259 @@
require 'digest'
class MaterialSyncImporter
Result = Struct.new(:material, :action, :suppressed, :suppression, keyword_init: true)
def self.import! attributes
new(attributes).import!
end
def initialize attributes
@attributes = attributes
end
def import!
source = source_attributes
key = normalized_source_key(source)
raise ArgumentError, 'normalized_source_key is required for material sync' if key.blank?
suppression = suppression_for(source)
if suppression
return Result.new(material: nil,
action: :suppressed,
suppressed: true,
suppression:)
end
material =
Material
.unscoped
.find_or_initialize_by(normalized_source_key: key)
if unchanged?(material, source)
return Result.new(material:,
action: :unchanged,
suppressed: false,
suppression: nil)
end
event_type =
if material.new_record?
:create
elsif material.discarded?
:restore
else
:update
end
uploaded_blob = uploaded_blob!
if @import_block
return Result.new(material: nil,
action: :suppressed,
suppressed: true,
suppression: @import_block)
end
Material.transaction do
if material.persisted?
MaterialVersionRecorder.ensure_snapshot!(material,
created_by_user: @attributes[:updated_by_user])
end
material.assign_attributes(material_attributes_for(material).merge(source))
material.discarded_at = nil if material.respond_to?(:discarded_at=)
material.file.attach(uploaded_blob) if uploaded_blob
material.save!
upsert_export_item!(material)
MaterialVersionRecorder.record!(material:,
event_type:,
created_by_user: @attributes[:updated_by_user])
end
MaterialThumbnailGenerator.generate!(material)
Result.new(material:,
action: event_type == :create ? :imported : :updated,
suppressed: false,
suppression: nil)
rescue StandardError
uploaded_blob&.purge_later
raise
ensure
close_tempfile!
end
private
def source_attributes
{ source_kind: @attributes[:source_kind],
source_uri: @attributes[:source_uri],
source_path: @attributes[:source_path],
source_file_id: @attributes[:source_file_id] }
end
def normalized_source_key source
MaterialSyncSuppression.normalize_source_key(source_kind: source[:source_kind],
source_uri: source[:source_uri],
source_path: source[:source_path],
source_file_id: source[:source_file_id])
end
def material_attributes_for material
attrs =
@attributes
.except(:source_kind, :source_uri, :source_path, :source_file_id,
:file_blob, :file_tempfile, :file_downloader,
:filename, :content_type,
:file_sha256, :export_path, :profile, :tag, :url)
if material.new_record?
attrs[:tag] = @attributes[:tag] if @attributes.key?(:tag)
attrs[:url] = @attributes[:url] if @attributes.key?(:url)
return attrs
end
if @attributes.key?(:tag) && @attributes[:tag].present?
attrs[:tag] = @attributes[:tag]
end
if @attributes.key?(:url) && @attributes[:url].present?
attrs[:url] = @attributes[:url]
end
attrs
end
def suppression_for source
if source[:source_kind] == 'google_drive_file' && source[:source_path].present?
return MaterialSyncSuppressionMatcher.match_google_drive_candidate(
drive_file_id: source[:source_file_id],
relative_path: source[:source_path])
end
MaterialSyncSuppressionMatcher.match(**source)
end
def uploaded_blob!
return @attributes[:file_blob] if @attributes[:file_blob]
tempfile = upload_tempfile
return nil unless tempfile
tempfile.rewind
file_sha256 = @attributes[:file_sha256] || Digest::SHA256.file(tempfile.path).hexdigest
@attributes[:file_sha256] = file_sha256
@import_block = MaterialImportBlockMatcher.match_for_sha256(file_sha256)
if @import_block
tempfile.rewind
return nil
end
blob = ActiveStorage::Blob.create_and_upload!(
io: tempfile,
filename: @attributes[:filename],
content_type: @attributes[:content_type])
MaterialFileSha256.assign_metadata_sha256!(blob, file_sha256)
tempfile.rewind
blob
end
def upload_tempfile
return @upload_tempfile if defined?(@upload_tempfile)
@upload_tempfile = @attributes[:file_tempfile]
if @upload_tempfile.blank? && @attributes[:file_downloader]
@upload_tempfile = @attributes[:file_downloader].call
end
@upload_tempfile
end
def upsert_export_item! material
export_path = resolved_export_path(material)
return if export_path.blank?
item = material.material_export_items.find_or_initialize_by(profile: effective_profile)
item.export_path = export_path
item.enabled = true
item.created_by_user ||= @attributes[:created_by_user]
item.save!
end
def export_path_taken_by_other? material, profile, export_path
MaterialExportItem
.where(profile:, export_path:)
.where.not(material_id: material.id)
.exists?
end
def close_tempfile!
tempfile = upload_tempfile
return unless tempfile
tempfile.close! unless tempfile.closed?
rescue StandardError
nil
end
def unchanged? material, source
return false if material.new_record? || material.discarded?
return false unless same_source_attributes?(material, source)
return false unless same_export_path?(material)
return false unless same_material_attributes?(material)
same_file_snapshot?(material)
end
def same_source_attributes? material, source
material.source_kind == source[:source_kind] \
&& material.source_uri == source[:source_uri] \
&& material.source_path == source[:source_path] \
&& material.source_file_id == source[:source_file_id]
end
def same_export_path? material
expected = resolved_export_path(material)
current = material.material_export_items.find { |item| item.profile == effective_profile }
return current.blank? if expected.blank?
current&.enabled && current.export_path == expected
end
def same_material_attributes? material
same_tag_attribute?(material) && same_url_attribute?(material)
end
def same_file_snapshot? material
expected_sha256 = @attributes[:file_sha256].to_s.presence
if expected_sha256.present?
return false unless material.file.attached?
return MaterialFileSha256.metadata_sha256(material.file.blob) == expected_sha256
end
!material.file.attached?
end
def effective_profile
@attributes[:profile].presence || 'legacy_drive'
end
def resolved_export_path material
export_path = @attributes[:export_path].to_s.strip
return export_path if export_path.blank?
return export_path unless export_path_taken_by_other?(material, effective_profile, export_path)
MaterialSyncExportPath.uniquify(export_path, @attributes[:source_file_id])
end
def same_tag_attribute? material
return true unless @attributes.key?(:tag)
return true if @attributes[:tag].blank?
material.tag_id == @attributes[:tag].id
end
def same_url_attribute? material
return true unless @attributes.key?(:url)
return true if @attributes[:url].blank?
material.url == @attributes[:url]
end
end
+251
ファイルの表示
@@ -0,0 +1,251 @@
class MaterialSyncRunner
Result = Struct.new(:imported, :updated, :unchanged, :suppressed, :failed,
:errors, keyword_init: true)
def self.sync_enabled!
results = MaterialSyncSource.enabled.order(:id).map { |source| new(source).sync! }
Result.new(imported: results.sum(&:imported),
updated: results.sum(&:updated),
unchanged: results.sum(&:unchanged),
suppressed: results.sum(&:suppressed),
failed: results.sum(&:failed),
errors: results.flat_map(&:errors))
end
def initialize source
@source = source
end
def sync!
result = Result.new(imported: 0, updated: 0, unchanged: 0,
suppressed: 0, failed: 0, errors: [])
if @source.source_kind == 'google_drive_path'
sync_google_drive_path!(result)
else
candidates.each do |candidate|
next if candidate.blank?
sync_candidate!(candidate, result)
end
end
@source.update!(last_synced_at: Time.current)
log_result(result)
result
rescue NotImplementedError, StandardError => e
result.failed += 1
result.errors << { source_id: @source.id, error: e.message }
log_result(result)
result
end
private
def candidates
case @source.source_kind
when 'uri'
[uri_candidate]
when 'google_drive_file'
[google_drive_file_candidate]
when 'legacy_drive_path'
raise NotImplementedError, 'legacy_drive_path material sync is not implemented'
else
raise NotImplementedError, "Unsupported material sync source_kind: #{ @source.source_kind }"
end
end
def uri_candidate
{ source_kind: @source.source_kind,
source_uri: @source.source_uri,
source_path: @source.source_path,
source_file_id: @source.source_file_id,
url: @source.source_uri,
tag: nil,
created_by_user: @source.created_by_user,
updated_by_user: @source.updated_by_user || @source.created_by_user }
end
def sync_candidate! candidate, result
block = MaterialImportBlockMatcher.match_for_sha256(candidate[:file_sha256])
if block
result.suppressed += 1
Rails.logger.info(
material_sync_log(action: 'suppressed',
reason: block.reason,
normalized_source_key: candidate_normalized_source_key(candidate)))
return
end
suppression = suppression_for(candidate)
if suppression
result.suppressed += 1
Rails.logger.info(
material_sync_log(action: 'suppressed',
normalized_source_key: suppression.normalized_source_key))
return
end
import = MaterialSyncImporter.import!(candidate)
result.public_send("#{ import.action }=", result.public_send(import.action) + 1)
rescue StandardError => e
result.failed += 1
result.errors << { source_id: @source.id,
normalized_source_key: candidate_normalized_source_key(candidate),
error: e.message }
Rails.logger.warn(
material_sync_log(action: 'failed',
error: e.message,
normalized_source_key: candidate_normalized_source_key(candidate)))
ensure
close_candidate_tempfile(candidate)
end
def google_drive_path_candidates
folder_id = google_drive_folder_id
Enumerator.new do |entries|
drive_client.each_material_file_under_folder(folder_id).each do |entry|
entries << build_google_drive_candidate(entry)
end
end
end
def sync_google_drive_path! result
folder_id = google_drive_folder_id
scanned_count = 0
drive_client.each_material_file_under_folder(folder_id) do |entry|
scanned_count += 1
sync_candidate!(build_google_drive_candidate(entry), result)
log_google_drive_progress(folder_id, scanned_count, result) if progress_log_scan_count?(scanned_count)
end
log_google_drive_progress(folder_id, scanned_count, result, summary: true)
end
def google_drive_file_candidate
entry = drive_client.fetch_material_file(google_drive_file_id)
return nil unless entry
build_google_drive_candidate(entry, single_file: true)
end
def build_google_drive_candidate entry, single_file: false
relative_path =
if single_file
nil
else
entry[:relative_path]
end
export_path =
MaterialSyncExportPath.build(prefix: @source.export_path_prefix,
relative_path:,
filename: entry[:name],
source_file_id: entry[:id])
{ source_kind: 'google_drive_file',
source_uri: entry[:web_view_link] || google_drive_file_url(entry[:id]),
source_path: relative_path.presence || MaterialSyncExportPath.normalize_path(entry[:name]),
source_file_id: entry[:id],
filename: entry[:name],
content_type: entry[:mime_type],
file_downloader: lambda {
drive_client.download_to_tempfile(entry[:id], filename: entry[:name])
},
file_sha256: entry[:sha256_checksum],
export_path:,
profile: @source.profile,
tag: nil,
url: nil,
created_by_user: @source.created_by_user,
updated_by_user: @source.updated_by_user || @source.created_by_user }
end
def candidate_normalized_source_key candidate
MaterialSyncSuppression.normalize_source_key(source_kind: candidate[:source_kind],
source_uri: candidate[:source_uri],
source_path: candidate[:source_path],
source_file_id: candidate[:source_file_id])
end
def suppression_for candidate
if candidate[:source_kind] == 'google_drive_file' && candidate[:source_path].present?
return MaterialSyncSuppressionMatcher.match_google_drive_candidate(
drive_file_id: candidate[:source_file_id],
relative_path: candidate[:source_path])
end
MaterialSyncSuppressionMatcher.match(source_kind: candidate[:source_kind],
source_uri: candidate[:source_uri],
source_path: candidate[:source_path],
source_file_id: candidate[:source_file_id])
end
def drive_client
@drive_client ||= GoogleDrive::ApiClient.new
end
def google_drive_folder_id
google_drive_target_id.tap do |id|
if id.blank?
raise ArgumentError, 'google_drive_path source_file_id or source_uri is required'
end
end
end
def google_drive_file_id
google_drive_target_id.tap do |id|
if id.blank?
raise ArgumentError, 'google_drive_file source_file_id or source_uri is required'
end
end
end
def google_drive_target_id
@source.source_file_id.presence || drive_client.extract_file_id(@source.source_uri)
end
def google_drive_file_url file_id
"https://drive.google.com/file/d/#{ file_id }/view"
end
def close_candidate_tempfile candidate
tempfile = candidate[:file_tempfile]
return unless tempfile
tempfile.close! unless tempfile.closed?
rescue StandardError
nil
end
def log_result result
Rails.logger.info(material_sync_log(imported: result.imported,
updated: result.updated,
unchanged: result.unchanged,
suppressed: result.suppressed,
failed: result.failed))
end
def progress_log_scan_count? scanned_count
scanned_count == 1 || (scanned_count % 50).zero?
end
def log_google_drive_progress folder_id, scanned_count, result, summary: false
Rails.logger.info(
material_sync_log(folder_id:,
scanned_count:,
imported: result.imported,
updated: result.updated,
unchanged: result.unchanged,
suppressed: result.suppressed,
failed: result.failed,
progress: summary ? 'summary' : 'scan'))
end
def material_sync_log fields
{ material_sync_source_id: @source.id,
material_sync_source_name: @source.name }.merge(fields).to_json
end
end
+68
ファイルの表示
@@ -0,0 +1,68 @@
class MaterialSyncSuppressionMatcher
def self.match source_kind:,
source_uri: nil,
drive_path: nil,
drive_file_id: nil,
source_path: nil,
source_file_id: nil
key = MaterialSyncSuppression.normalize_source_key(source_kind:,
source_uri:,
drive_path:,
drive_file_id:,
source_path:,
source_file_id:)
return nil if key.blank?
MaterialSyncSuppression.find_by(normalized_source_key: key)
end
def self.match_google_drive_candidate drive_file_id:, relative_path:
normalized_path = validate_relative_path!(relative_path)
exact = [
normalize_key('google_drive_file', drive_file_id),
normalize_key('google_drive_path', normalized_path),
normalize_key('legacy_drive_path', normalized_path),
].compact
suppression =
MaterialSyncSuppression.find_by(normalized_source_key: exact)
return suppression if suppression
MaterialSyncSuppression
.where(source_kind: ['google_drive_path_prefix', 'legacy_drive_path_prefix'])
.where('drive_path = :path OR :path LIKE CONCAT(drive_path, "/%")', path: normalized_path)
.order(:id)
.first
end
def self.suppressed?(...)
match(...).present?
end
def self.validate_relative_path! relative_path
raw = relative_path.to_s.delete("\0").tr('\\', '/').strip
raise ArgumentError, 'relative_path is required' if raw.blank?
raise ArgumentError, 'relative_path must be relative' if raw.start_with?('/')
raise ArgumentError, 'relative_path must be relative' if raw.match?(/\A[A-Za-z]:\//)
if raw.start_with?('My Drive/', 'マイドライブ/')
raise ArgumentError, 'relative_path must be relative to source folder'
end
raise ArgumentError, 'relative_path must not contain //' if raw.include?('//')
raise ArgumentError, 'relative_path must not end with /' if raw.end_with?('/')
parts = raw.split('/')
if parts.any? { |part| part.in?(['.', '..']) }
raise ArgumentError, 'relative_path must not contain dot segments'
end
raw
end
def self.normalize_key source_kind, value
return nil if value.blank?
MaterialSyncSuppression.normalize_source_key(source_kind:, source_path: value,
source_file_id: value)
end
end
+51
ファイルの表示
@@ -0,0 +1,51 @@
class MaterialSyncSuppressionRegistrar
def self.create! attributes, created_by_user:
new(attributes, created_by_user:).create!
end
def initialize attributes, created_by_user:
@attributes = attributes
@created_by_user = created_by_user
end
def create!
suppression = nil
MaterialSyncSuppression.transaction do
suppression = MaterialSyncSuppression.create!(
@attributes.merge(created_by_user: @created_by_user))
discard_existing_materials!(suppression)
end
suppression
end
private
def discard_existing_materials! suppression
matching_materials(suppression).find_each do |material|
MaterialVersionRecorder.ensure_snapshot!(material, created_by_user: @created_by_user)
material.discard!
MaterialVersionRecorder.record!(material:,
event_type: :discard,
created_by_user: @created_by_user)
end
end
def matching_materials suppression
materials = Material.unscoped.kept
case suppression.source_kind
when 'google_drive_path', 'legacy_drive_path'
materials.where(source_path: suppression.drive_path)
when 'google_drive_path_prefix', 'legacy_drive_path_prefix'
path = suppression.drive_path.to_s
materials.where(
'source_path = :path OR source_path LIKE :prefix',
path:,
prefix: "#{ path }/%")
else
materials.where(normalized_source_key: suppression.normalized_source_key)
end
end
end
+126
ファイルの表示
@@ -0,0 +1,126 @@
# frozen_string_literal: true
require 'mini_magick'
require 'open3'
require 'tempfile'
class MaterialThumbnailGenerator
SIZE = '180x180'
class << self
def generate! material
new(material).generate!
end
end
def initialize material
@material = material
@ffmpeg_stderr = []
end
def generate!
return log_result(:no_file) unless @material.file.attached?
return log_result(:unsupported_content_type) unless image? || video?
@material.file.blob.open do |file|
thumbnail = image? ? image_thumbnail(file.path) : video_thumbnail(file.path)
return log_result(:generation_failed) unless thumbnail
return attach_thumbnail(thumbnail)
end
rescue ActiveStorage::FileNotFoundError => e
log_result(:file_not_found, error: e)
rescue MiniMagick::Error => e
log_result(:mini_magick_error, error: e)
rescue ArgumentError, StandardError => e
log_result(:generation_failed, error: e)
end
private
def image? = content_type.start_with?('image/')
def video? = content_type.start_with?('video/')
def content_type
return nil unless @material.file.attached?
@material.file.blob.content_type.to_s
end
def image_thumbnail path
image = MiniMagick::Image.open(path)
image.resize(SIZE)
image.format('jpg')
image
end
def video_thumbnail path
[1, 0].each do |seconds|
tempfile = Tempfile.new(['material-thumbnail', '.jpg'])
tempfile.close
ok = extract_video_frame(path, tempfile.path, seconds)
next unless ok && File.size?(tempfile.path)
return image_thumbnail(tempfile.path)
ensure
tempfile&.unlink
end
nil
end
def extract_video_frame input_path, output_path, seconds
_stdout, stderr, status =
Open3.capture3('ffmpeg',
'-y',
'-ss', seconds.to_s,
'-i', input_path,
'-frames:v', '1',
'-f', 'image2',
output_path)
@ffmpeg_stderr << stderr if stderr.present?
status.success?
rescue Errno::ENOENT => e
@ffmpeg_stderr << "ffmpeg unavailable: #{ e.message }"
false
end
def attach_thumbnail image
blob = nil
File.open(image.path) do |io|
blob = ActiveStorage::Blob.create_and_upload!(
io:,
filename: 'material-thumbnail.jpg',
content_type: 'image/jpeg')
end
@material.thumbnail.attach(blob)
log_result(:attached)
rescue StandardError => e
blob&.purge_later
log_result(:attach_failed, error: e)
end
def log_result result, error: nil
log_payload = { material_id: @material.id,
file_blob_id: file_blob_id,
content_type:,
result:,
error_class: error&.class&.name,
error_message: error&.message,
ffmpeg_stderr: @ffmpeg_stderr.join("\n").presence }
if [:attached, :no_file, :unsupported_content_type].include?(result)
Rails.logger.info("Material thumbnail generation: #{ log_payload.to_json }")
else
Rails.logger.warn("Material thumbnail generation: #{ log_payload.to_json }")
end
result
end
def file_blob_id
return nil unless @material.file.attached?
@material.file.blob.id
end
end
+73
ファイルの表示
@@ -0,0 +1,73 @@
class MaterialVersionRecorder < VersionRecorder
EVENT_TYPES = ['create', 'update', 'discard', 'restore'].freeze
def self.record! material:, event_type:, created_by_user:, file_snapshot: nil
new(material:, event_type:, created_by_user:, file_snapshot:).record!
end
def initialize material:, event_type:, created_by_user:, file_snapshot: nil
@file_snapshot = file_snapshot
super(record: material, event_type:, created_by_user:)
end
def self.ensure_snapshot! material, created_by_user:
return if material.material_versions.exists?
record!(material:, event_type: :create,
created_by_user: material.created_by_user || created_by_user)
end
private
def version_class = MaterialVersion
def version_association = :material_versions
def record_key = :material
def snapshot_attributes
blob = @record.file.attached? ? @record.file.blob : nil
file_snapshot = build_file_snapshot(blob)
{ url: @record.url,
parent: @record.parent,
tag: @record.tag,
tag_name: @record.tag&.name,
tag_category: @record.tag&.category,
source_kind: @record.source_kind,
source_uri: @record.source_uri,
source_path: @record.source_path,
source_file_id: @record.source_file_id,
normalized_source_key: @record.normalized_source_key,
export_paths_json: @record.snapshot_export_paths,
discarded_at: @record.discarded_at,
file_blob_id: file_snapshot[:file_blob_id],
file_filename: file_snapshot[:file_filename],
file_content_type: file_snapshot[:file_content_type],
file_byte_size: file_snapshot[:file_byte_size],
file_checksum: file_snapshot[:file_checksum],
file_sha256: file_snapshot[:file_sha256] }
end
def build_file_snapshot blob
return @file_snapshot if @file_snapshot
return empty_file_snapshot unless blob
{ file_blob_id: blob.id,
file_filename: blob.filename.to_s,
file_content_type: blob.content_type,
file_byte_size: blob.byte_size,
file_checksum: blob.checksum,
file_sha256: MaterialFileSha256.metadata_sha256(blob) }
end
def empty_file_snapshot
{ file_blob_id: nil,
file_filename: nil,
file_content_type: nil,
file_byte_size: nil,
file_checksum: nil,
file_sha256: nil }
end
def event_types = self.class::EVENT_TYPES
end
+148
ファイルの表示
@@ -0,0 +1,148 @@
require 'stringio'
require 'zlib'
# Initial implementation keeps every file payload and the final ZIP in memory.
# Keep this service boundary stable so job/cached export paths can replace it later.
class MaterialZipExporter
Entry = Struct.new(:path, :data, :mtime, keyword_init: true)
MissingFile = Struct.new(:material_id, :export_path, :blob_id, :filename, keyword_init: true)
class EmptyExportError < StandardError; end
class DuplicatePathError < StandardError; end
class MissingFileError < StandardError
attr_reader :missing_files
def initialize missing_files
@missing_files = missing_files
super("Missing files: #{missing_files.map(&:export_path).join(', ')}")
end
end
def initialize profile: 'legacy_drive', tag_id: nil
@profile = profile.presence || 'legacy_drive'
@tag_id = tag_id.presence
end
def export
entries = build_entries
raise EmptyExportError if entries.empty?
ZipWriter.write(entries)
end
private
def build_entries
rows = MaterialExportItem
.enabled
.includes(material: { file_attachment: :blob })
.joins(:material)
.merge(Material.kept)
.where(profile: @profile)
.order(:export_path)
rows = rows.where(materials: { tag_id: @tag_id }) if @tag_id
missing_files = []
entries = rows.filter_map do |item|
material = item.material
next unless material.file.attached?
data = download_blob(item, missing_files)
next unless data
Entry.new(path: item.export_path,
data:,
mtime: material.updated_at || Time.current)
end
raise MissingFileError.new(missing_files) if missing_files.any?
paths = entries.map(&:path)
duplicated = paths.find { |path| paths.count(path) > 1 }
raise DuplicatePathError, duplicated if duplicated
entries
end
def download_blob item, missing_files
blob = item.material.file.blob
blob.download
rescue ActiveStorage::FileNotFoundError
missing_files << MissingFile.new(
material_id: item.material_id,
export_path: item.export_path,
blob_id: blob.id,
filename: blob.filename.to_s,
)
nil
end
class ZipWriter
VERSION_NEEDED = 20
GP_FLAG = 0x0800
COMPRESSION_STORE = 0
def self.write entries
new(entries).write
end
def initialize entries
@entries = entries
@central_directory = []
end
def write
io = StringIO.new(''.b)
@entries.each do |entry|
write_entry(io, entry)
end
central_start = io.pos
@central_directory.each { |header| io.write(header) }
central_size = io.pos - central_start
io.write([0x06054b50, 0, 0, @entries.size, @entries.size,
central_size, central_start, 0].pack('VvvvvVVv'))
io.string
end
private
def write_entry io, entry
path = entry.path.b
data = entry.data.b
crc32 = Zlib.crc32(data)
dos_time, dos_date = dos_timestamp(entry.mtime)
offset = io.pos
local_header = [0x04034b50, VERSION_NEEDED, GP_FLAG, COMPRESSION_STORE,
dos_time, dos_date, crc32, data.bytesize, data.bytesize,
path.bytesize, 0].pack('VvvvvvVVVvv')
io.write(local_header)
io.write(path)
io.write(data)
@central_directory << central_header(path:, crc32:, size: data.bytesize,
dos_time:, dos_date:, offset:)
end
def central_header path:, crc32:, size:, dos_time:, dos_date:, offset:
[0x02014b50, VERSION_NEEDED, VERSION_NEEDED, GP_FLAG, COMPRESSION_STORE,
dos_time, dos_date, crc32, size, size, path.bytesize, 0, 0, 0, 0, 0,
offset].pack('VvvvvvvVVVvvvvvVV') + path
end
def dos_timestamp time
local = time.to_time
dos_time = (local.hour << 11) | (local.min << 5) | (local.sec / 2)
dos_date = ((local.year - 1980) << 9) | (local.month << 5) | local.day
[dos_time, dos_date]
end
end
end
+2 -1
ファイルの表示
@@ -73,7 +73,7 @@ class VersionRecorder
end
def validate_event_type!
return if EVENT_TYPES.include?(@event_type)
return if event_types.include?(@event_type)
raise ArgumentError, "Invalid event_type: #{ @event_type }"
end
@@ -84,4 +84,5 @@ class VersionRecorder
def snapshot_attributes = raise NotImplementedError
def record_class = @record.class
def event_types = self.class::EVENT_TYPES
end