#306

backend/app/controllers/material_versions_controller.rb

@@ -35,6 +35,11 @@ class MaterialVersionsController < ApplicationController
         tag_name: row.tag_name,
         tag_category: row.tag_category,
         url: row.url,
+        source_kind: row.source_kind,
+        source_uri: row.source_uri,
+        source_path: row.source_path,
+        source_file_id: row.source_file_id,
+        normalized_source_key: row.normalized_source_key,
         file_blob_id: row.file_blob_id,
         file_filename: row.file_filename,
         file_content_type: row.file_content_type,

backend/app/models/material_sync_source.rb

@@ -1,11 +1,13 @@
 class MaterialSyncSource < ApplicationRecord
+  SOURCE_KINDS = ['uri', 'google_drive_path', 'google_drive_file', 'legacy_drive_path'].freeze
+
   belongs_to :created_by_user, class_name: 'User', optional: true
   belongs_to :updated_by_user, class_name: 'User', optional: true
 
   validates :name, presence: true
   validates :source_kind,
             presence: true,
-            inclusion: { in: MaterialSyncSuppression::SOURCE_KINDS }
+            inclusion: { in: SOURCE_KINDS }
   validates :profile, presence: true, inclusion: { in: MaterialExportItem::VALID_PROFILES }
   validate :source_value_must_be_present
 
@@ -33,7 +35,7 @@ class MaterialSyncSource < ApplicationRecord
     when 'google_drive_file'
       source_file_id.present? || source_uri.present?
     when 'google_drive_path'
-      source_file_id.present? || source_uri.present? || source_path.present?
+      source_file_id.present? || source_uri.present?
     when 'legacy_drive_path'
       source_path.present?
     else

backend/app/services/google_drive/api_client.rb

@@ -11,6 +11,7 @@ module GoogleDrive
     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'],
@@ -27,6 +28,7 @@ module GoogleDrive
       files = []
       walk_folder(folder_id, nil) do |entry, relative_path|
         next if entry['mimeType'] == FOLDER_MIME_TYPE
+        next if native_file?(entry['mimeType'])
 
         files << build_file_entry(entry, relative_path)
       end
@@ -35,6 +37,8 @@ module GoogleDrive
 
     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
 
@@ -85,6 +89,7 @@ module GoogleDrive
         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
@@ -97,7 +102,8 @@ module GoogleDrive
         response =
             request_json('/files', {
                 q: "'#{ folder_id }' in parents and trashed = false",
-                fields: 'nextPageToken,files(id,name,mimeType,webViewLink,webContentLink)',
+                fields: 'nextPageToken,files(id,name,mimeType,sha256Checksum,' \
+                        'webViewLink,webContentLink)',
                 orderBy: 'folder,name',
                 pageSize: 1000,
                 supportsAllDrives: true,
@@ -113,7 +119,7 @@ module GoogleDrive
 
     def get_file file_id
       request_json("/files/#{ file_id }",
-                   fields: 'id,name,mimeType,webViewLink,webContentLink',
+                   fields: 'id,name,mimeType,sha256Checksum,webViewLink,webContentLink',
                    supportsAllDrives: true)
     end
 
@@ -127,13 +133,14 @@ module GoogleDrive
     end
 
     def request_binary path, params = {}
-      response = request(:get, path, params:)
-      unless response.is_a?(Net::HTTPSuccess)
-        raise "Google Drive download error: #{ response.code } #{ response.body }"
-      end
+      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
+        response.read_body do |chunk|
+          yield chunk
+        end
       end
     end
 
@@ -156,7 +163,13 @@ module GoogleDrive
                       use_ssl: true,
                       open_timeout: 10,
                       read_timeout: 60) do |http|
-        http.request(request)
+        if block_given?
+          http.request(request) do |response|
+            return yield response
+          end
+        else
+          http.request(request)
+        end
       end
     end
 
@@ -205,5 +218,9 @@ module GoogleDrive
 
       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

backend/app/services/material_sync_importer.rb

@@ -28,6 +28,13 @@ class MaterialSyncImporter
         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
@@ -43,7 +50,7 @@ class MaterialSyncImporter
         MaterialVersionRecorder.ensure_snapshot!(material,
                                                  created_by_user: @attributes[:updated_by_user])
       end
-      material.assign_attributes(material_attributes.merge(source))
+      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!
@@ -81,10 +88,28 @@ class MaterialSyncImporter
                                                  source_file_id: source[:source_file_id])
   end
 
-  def material_attributes
-    @attributes.except(:source_kind, :source_uri, :source_path, :source_file_id,
-                       :file_blob, :file_tempfile, :filename, :content_type,
-                       :file_sha256, :export_path, :profile)
+  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
@@ -100,7 +125,7 @@ class MaterialSyncImporter
   def uploaded_blob!
     return @attributes[:file_blob] if @attributes[:file_blob]
 
-    tempfile = @attributes[:file_tempfile]
+    tempfile = upload_tempfile
     return nil unless tempfile
 
     tempfile.rewind
@@ -115,18 +140,22 @@ class MaterialSyncImporter
     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 = @attributes[:export_path].to_s.strip
+    export_path = resolved_export_path(material)
     return if export_path.blank?
 
-    profile = @attributes[:profile].presence || 'legacy_drive'
-    path = export_path
-    if export_path_taken_by_other?(material, profile, path)
-      path = MaterialSyncExportPath.uniquify(path, @attributes[:source_file_id])
-    end
-
-    item = material.material_export_items.find_or_initialize_by(profile:)
-    item.export_path = path
+    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!
@@ -140,11 +169,78 @@ class MaterialSyncImporter
   end
 
   def close_tempfile!
-    tempfile = @attributes[:file_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 material.file.blob.metadata['sha256'] == 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

backend/app/services/material_sync_runner.rb

@@ -1,13 +1,13 @@
-require 'digest'
-
 class MaterialSyncRunner
-  Result = Struct.new(:imported, :updated, :suppressed, :failed, :errors, keyword_init: true)
+  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))
@@ -18,9 +18,12 @@ class MaterialSyncRunner
   end
 
   def sync!
-    result = Result.new(imported: 0, updated: 0, suppressed: 0, failed: 0, errors: [])
+    result = Result.new(imported: 0, updated: 0, unchanged: 0,
+                        suppressed: 0, failed: 0, errors: [])
 
     candidates.each do |candidate|
+      next if candidate.blank?
+
       sync_candidate!(candidate, result)
     end
 
@@ -108,8 +111,10 @@ class MaterialSyncRunner
   end
 
   def google_drive_file_candidate
-    build_google_drive_candidate(drive_client.fetch_material_file(google_drive_file_id),
-                                 single_file: true)
+    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
@@ -124,10 +129,6 @@ class MaterialSyncRunner
                                      relative_path:,
                                      filename: entry[:name],
                                      source_file_id: entry[:id])
-    tempfile =
-        drive_client.download_to_tempfile(entry[:id], filename: entry[:name])
-    file_sha256 = Digest::SHA256.file(tempfile.path).hexdigest
-    tempfile.rewind
 
     { source_kind: 'google_drive_file',
       source_uri: entry[:web_view_link] || google_drive_file_url(entry[:id]),
@@ -135,8 +136,10 @@ class MaterialSyncRunner
       source_file_id: entry[:id],
       filename: entry[:name],
       content_type: entry[:mime_type],
-      file_tempfile: tempfile,
-      file_sha256:,
+      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,
@@ -171,13 +174,17 @@ class MaterialSyncRunner
 
   def google_drive_folder_id
     google_drive_target_id.tap do |id|
-      raise ArgumentError, 'google_drive_path source_file_id or source_uri is required' if id.blank?
+      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|
-      raise ArgumentError, 'google_drive_file source_file_id or source_uri is required' if id.blank?
+      if id.blank?
+        raise ArgumentError, 'google_drive_file source_file_id or source_uri is required'
+      end
     end
   end
 
@@ -201,6 +208,7 @@ class MaterialSyncRunner
   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

backend/app/services/material_version_recorder.rb

@@ -33,6 +33,11 @@ class MaterialVersionRecorder < VersionRecorder
       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],