feat(导入): 添加加密文档验证和文档冲突检查

添加对加密文档的验证，禁止导入包含正文检索字段的加密文档
增加文档冲突检查，避免导入重复文档（相同标题+日期）
在导入时跳过已存在的文档并返回警告信息

Author: Wpc-121

app/services/user_import_service.rb

@@ -31,6 +31,7 @@ def validate
       validate_schema_version
       validate_structure
       validate_unique_names
+      validate_encrypted_documents
     rescue SchemaVersionError => e
       @errors << e.message
     rescue ImportError => e
@@ -147,6 +148,19 @@ def validate_unique_names
     check_existing_conflicts(folder_names, tag_names, person_names, state_names)
   end
 
+  def validate_encrypted_documents
+    data = @json_data['data']
+
+    data['documents'].each_with_index do |doc, idx|
+      next unless doc['encrypted_flag']
+
+      if doc.key?('document_text') && doc['document_text'].present?
+        raise ImportError,
+          "Document #{idx} ('#{doc['title']}'): 加密文档禁止导入正文检索字段"
+      end
+    end
+  end
+
   def check_existing_conflicts(folder_names, tag_names, person_names, state_names)
     existing_folders = @user.folders.where(name: folder_names).pluck(:name)
     if existing_folders.any?
@@ -167,6 +181,59 @@ def check_existing_conflicts(folder_names, tag_names, person_names, state_names)
     if existing_states.any?
       @warnings << "States already exist and will be skipped: #{existing_states.join(', ')}"
     end
+
+    check_document_conflicts
+  end
+
+  def check_document_conflicts
+    data = @json_data['data']
+    imported_docs = data['documents']
+
+    doc_keys = Set.new
+    duplicate_in_import = []
+
+    imported_docs.each do |doc|
+      key = "#{doc['title']}|||#{doc['document_date']}"
+      if doc_keys.include?(key)
+        duplicate_in_import << doc['title']
+      end
+      doc_keys.add(key)
+    end
+
+    if duplicate_in_import.any?
+      raise ImportError,
+        "Duplicate documents in import (same title + document_date): #{duplicate_in_import.uniq.join(', ')}"
+    end
+
+    existing_doc_keys = Set.new
+    @user.documents.each do |doc|
+      key = "#{doc.title}|||#{doc.document_date}"
+      existing_doc_keys.add(key)
+    end
+
+    import_dates = imported_docs.pluck('document_date').compact.uniq
+    existing_docs_by_date = {}
+
+    import_dates.each do |date_str|
+      date = parse_date(date_str)
+      docs = @user.documents.where(document_date: date)
+      docs.each do |doc|
+        key = "#{doc.title}|||#{doc.document_date}"
+        existing_docs_by_date[key] = doc
+      end
+    end
+
+    conflicting_titles = []
+    imported_docs.each do |doc|
+      key = "#{doc['title']}|||#{doc['document_date']}"
+      if existing_docs_by_date.key?(key)
+        conflicting_titles << doc['title']
+      end
+    end
+
+    if conflicting_titles.any?
+      @warnings << "Documents already exist (same title + document_date) and will be skipped: #{conflicting_titles.join(', ')}"
+    end
   end
 
   def import_with_transaction
@@ -320,8 +387,28 @@ def import_states(states_data)
 
   def import_documents(documents_data)
     imported = 0
+    skipped = 0
+
+    import_dates = documents_data.pluck('document_date').compact.uniq
+    existing_docs_by_key = {}
+
+    import_dates.each do |date_str|
+      date = parse_date(date_str)
+      docs = @user.documents.where(document_date: date)
+      docs.each do |doc|
+        key = "#{doc.title}|||#{doc.document_date}"
+        existing_docs_by_key[key] = doc
+      end
+    end
 
     documents_data.each do |doc_data|
+      doc_key = "#{doc_data['title']}|||#{doc_data['document_date']}"
+
+      if existing_docs_by_key.key?(doc_key)
+        skipped += 1
+        next
+      end
+
       folder_id = @name_to_id_map[:folders][doc_data['folder_name']]
       state_id = @name_to_id_map[:states][doc_data['state_name']]
       person_id = @name_to_id_map[:people][doc_data['person_name']]
@@ -363,6 +450,7 @@ def import_documents(documents_data)
     end
 
     @imported_data[:documents] = imported
+    @imported_data[:documents_skipped] = skipped if skipped > 0
   end
 
   def parse_time(value)

test/controllers/imports_controller_test.rb

@@ -91,6 +91,36 @@ class ImportsControllerTest < ActionController::TestCase
         documents: []
       }
     }
+
+    @encrypted_with_document_text_data = {
+      schema_version: 1,
+      exported_at: Time.current.iso8601,
+      user: { name: 'Encrypted User', email: 'encrypted@example.com' },
+      data: {
+        folders: [
+          { name: 'Encrypted Folder', parent_folder_name: nil, created_at: 1.day.ago.iso8601, updated_at: 1.day.ago.iso8601 }
+        ],
+        tags: [],
+        people: [],
+        states: [],
+        documents: [
+          {
+            title: 'Encrypted With Text',
+            description: 'Should be rejected',
+            document_date: Date.today.iso8601,
+            version: '1.0',
+            encrypted_flag: true,
+            folder_name: 'Encrypted Folder',
+            state_name: nil,
+            person_name: nil,
+            tag_names: [],
+            document_text: 'This should not be here for encrypted docs',
+            created_at: 1.day.ago.iso8601,
+            updated_at: 1.day.ago.iso8601
+          }
+        ]
+      }
+    }
   end
 
   test 'should preview import data without modifying database' do
@@ -299,4 +329,193 @@ class ImportsControllerTest < ActionController::TestCase
 
     assert_equal 1, json_response['preview']['encrypted_documents']
   end
+
+  test 'should reject encrypted document with document_text' do
+    original_folder_count = @user.folders.count
+    original_doc_count = @user.documents.count
+
+    post :create, params: { import: @encrypted_with_document_text_data }, format: :json
+
+    assert_response :unprocessable_entity
+    json_response = JSON.parse(response.body)
+
+    assert_not json_response['success']
+    assert_not_empty json_response['errors']
+    assert_includes json_response['errors'].first, '加密文档禁止导入正文检索字段'
+    assert_includes json_response['errors'].first, 'Encrypted With Text'
+
+    assert_equal original_folder_count, @user.folders.count
+    assert_equal original_doc_count, @user.documents.count
+  end
+
+  test 'should reject encrypted document with document_text in preview' do
+    post :preview, params: { import: @encrypted_with_document_text_data }, format: :json
+
+    assert_response :success
+    json_response = JSON.parse(response.body)
+
+    assert_not_empty json_response['errors']
+    assert_includes json_response['errors'].first, '加密文档禁止导入正文检索字段'
+  end
+
+  test 'conflict: existing folder should be skipped with warning' do
+    existing_folder = @user.folders.create!(name: 'Imported Work')
+
+    original_folder_count = @user.folders.count
+
+    post :create, params: { import: @valid_export_data }, format: :json
+
+    assert_response :success
+    json_response = JSON.parse(response.body)
+
+    assert json_response['success']
+    assert_not_empty json_response['warnings']
+    assert_includes json_response['warnings'].first, 'Folders already exist'
+    assert_includes json_response['warnings'].first, 'Imported Work'
+
+    assert_equal original_folder_count + 1, @user.folders.count
+    assert @user.folders.exists?(id: existing_folder.id)
+  end
+
+  test 'conflict: existing tag should be skipped with warning' do
+    existing_tag = @user.tags.create!(name: 'Imported Tag', color: '#00ff00')
+
+    original_tag_count = @user.tags.count
+
+    post :create, params: { import: @valid_export_data }, format: :json
+
+    assert_response :success
+    json_response = JSON.parse(response.body)
+
+    assert json_response['success']
+    assert_not_empty json_response['warnings']
+    assert_includes json_response['warnings'].first, 'Tags already exist'
+    assert_includes json_response['warnings'].first, 'Imported Tag'
+
+    assert_equal original_tag_count, @user.tags.count
+    assert @user.tags.exists?(id: existing_tag.id)
+    assert_equal '#00ff00', existing_tag.reload.color
+  end
+
+  test 'conflict: existing person should be skipped with warning' do
+    existing_person = @user.people.create!(name: 'Imported Person')
+
+    original_person_count = @user.people.count
+
+    post :create, params: { import: @valid_export_data }, format: :json
+
+    assert_response :success
+    json_response = JSON.parse(response.body)
+
+    assert json_response['success']
+    assert_not_empty json_response['warnings']
+    assert_includes json_response['warnings'].to_s, 'People already exist'
+    assert_includes json_response['warnings'].to_s, 'Imported Person'
+
+    assert_equal original_person_count, @user.people.count
+    assert @user.people.exists?(id: existing_person.id)
+  end
+
+  test 'conflict: existing document (same title + date) should be skipped with warning' do
+    test_date = Date.today
+    existing_doc = @user.documents.create!(
+      title: 'Imported Document',
+      document_date: test_date,
+      document_url: 'existing.pdf',
+      document_text: 'existing text'
+    )
+
+    original_doc_count = @user.documents.count
+
+    post :create, params: { import: @valid_export_data }, format: :json
+
+    assert_response :success
+    json_response = JSON.parse(response.body)
+
+    assert json_response['success']
+    assert_not_empty json_response['warnings']
+    assert_includes json_response['warnings'].to_s, 'Documents already exist'
+    assert_includes json_response['warnings'].to_s, 'Imported Document'
+
+    assert_equal 0, json_response['imported_data']['documents']
+    assert_equal 1, json_response['imported_data']['documents_skipped']
+
+    assert_equal original_doc_count, @user.documents.count
+    assert @user.documents.exists?(id: existing_doc.id)
+    assert_equal 'existing text', existing_doc.reload.document_text
+  end
+
+  test 'should reject import with duplicate documents in same data (same title + date)' do
+    duplicate_data = @valid_export_data.deep_dup
+    duplicate_data[:data][:documents] << {
+      title: 'Imported Document',
+      description: 'Duplicate',
+      document_date: Date.today.iso8601,
+      version: '2.0',
+      encrypted_flag: false,
+      folder_name: nil,
+      state_name: nil,
+      person_name: nil,
+      tag_names: [],
+      document_text: 'duplicate text',
+      created_at: 1.day.ago.iso8601,
+      updated_at: 1.day.ago.iso8601
+    }
+
+    original_doc_count = @user.documents.count
+
+    post :create, params: { import: duplicate_data }, format: :json
+
+    assert_response :unprocessable_entity
+    json_response = JSON.parse(response.body)
+
+    assert_not json_response['success']
+    assert_not_empty json_response['errors']
+    assert_includes json_response['errors'].first, 'Duplicate documents in import'
+    assert_includes json_response['errors'].first, 'Imported Document'
+
+    assert_equal original_doc_count, @user.documents.count
+  end
+
+  test 'conflict: mixed existing and new documents should import new ones' do
+    test_date = Date.today
+    @user.documents.create!(
+      title: 'Imported Document',
+      document_date: test_date,
+      document_url: 'existing.pdf'
+    )
+
+    data_with_new_doc = @valid_export_data.deep_dup
+    data_with_new_doc[:data][:documents] << {
+      title: 'New Document',
+      description: 'Brand new',
+      document_date: 1.day.ago.iso8601,
+      version: '1.0',
+      encrypted_flag: false,
+      folder_name: 'Imported Work',
+      state_name: nil,
+      person_name: nil,
+      tag_names: [],
+      document_text: 'new document text',
+      created_at: 1.day.ago.iso8601,
+      updated_at: 1.day.ago.iso8601
+    }
+
+    original_doc_count = @user.documents.count
+
+    assert_difference -> { @user.documents.count }, 1 do
+      post :create, params: { import: data_with_new_doc }, format: :json
+    end
+
+    assert_response :success
+    json_response = JSON.parse(response.body)
+
+    assert json_response['success']
+    assert_equal 1, json_response['imported_data']['documents']
+    assert_equal 1, json_response['imported_data']['documents_skipped']
+    assert_not_empty json_response['warnings']
+
+    assert @user.documents.find_by(title: 'Imported Document').present?
+    assert @user.documents.find_by(title: 'New Document').present?
+  end
 end