#444

backend/spec/requests/counts_masked_search_spec.rb

@@ -0,0 +1,196 @@
+require 'rails_helper'
+
+RSpec.describe 'Counts masked search', type: :request do
+  let!(:tag) { create(:tag, primary_name: '秘密のタグ') }
+  let(:actor) { create(:user, :member) }
+  let(:admin) { create(:user, :admin) }
+  let(:base) do
+    { 'resource' => 'tags', 'definition' => 'history', 'mode' => 'summary',
+      'locale' => 'ja', 'execution_at' => '2025-01-01T00:00:00Z',
+      'timezone' => 'UTC', 'rows' => '', 'cols' => '', 'sort' => '' }
+  end
+
+  before do
+    [1, 2].each do |number|
+      TagVersion.create!(tag:, version_no: number,
+          event_type: number == 1 ? 'create' : 'discard', name: '秘密のタグ',
+          category: 'general', aliases: '秘密の別名', parent_tag_ids: '',
+          names: { ja: { name: '秘密のタグ', script_code: 'Jpan', auto_generated: false,
+                        aliases: [{ name: '秘密の別名', script_code: 'Jpan' }] } },
+          created_by_user: actor, created_at: Time.utc(2023, number))
+    end
+  end
+
+  def query(extra = {})
+    get '/counts/query', params: base.merge(extra)
+    expect(response).to have_http_status(:ok), extra.inspect
+    response.parsed_body
+  end
+
+  def predicate(field, operator, value, index = 0)
+    prefix = "conditions.#{ index }"
+    result = { 'conditions' => '', "#{ prefix }.field" => field,
+               "#{ prefix }.operator" => operator }
+    if ['in', 'any', 'all', 'none', 'not_in'].include?(operator)
+      result.merge!("#{ prefix }.values" => '1', 'condition_values' => '',
+          "condition_values.#{ index }.condition" => index.to_s,
+          "condition_values.#{ index }.value_type" => 'string',
+          "condition_values.#{ index }.value" => value)
+    else
+      result["#{ prefix }.value"] = value
+    end
+    result
+  end
+
+  it 'rejects every name and alias predicate on private history' do
+    ['name', 'name_ja', 'names.name', 'aliases.name', 'names.aliases.name'].each do |field|
+      secret = field.include?('aliases') ? '秘密の別名' : '秘密のタグ'
+      multiple = !['name', 'name_ja'].include?(field)
+      operators = multiple ? ['any', 'all', 'none'] : ['in', 'not_in']
+      ['eq', 'ne', 'like', *operators].each do |operator|
+        [secret, '********', '%', ''].each do |value|
+          expected = (value == '' && ['eq', 'ne', 'like'].include?(operator)) ? 2 : 0
+          expect(query(predicate(field, operator, value))['count']).to eq(expected)
+        end
+      end
+      blank = predicate(field, 'eq', '').merge('conditions.0.blank' => '1')
+      expect(query(blank)['count']).to eq(0)
+    end
+  end
+
+  it 'ignores missing and empty scalar values unless blank is checked' do
+    ['name', 'name_ja', 'names.name', 'aliases.name'].each do |field|
+      ['eq', 'ne', 'like'].each do |operator|
+        condition = predicate(field, operator, '')
+        [condition, condition.except('conditions.0.value')].each do |empty|
+          expect(query(empty)['count']).to eq(2)
+          expect(query(empty.merge('conditions.0.blank' => '0'))['count']).to eq(2)
+          expect(query(empty.merge('conditions.0.blank' => '1'))['count']).to eq(0)
+        end
+      end
+    end
+  end
+
+  it 'keeps details totals unchanged for inactive private predicates' do
+    settings = predicate('name', 'ne', '').merge('mode' => 'details')
+    result = query(settings)
+    expect(result['count']).to eq(0)
+    expect(result['versions']).to be_empty
+    expect(result['output_count']).to eq(2)
+    active = query(settings.merge('conditions.0.blank' => '1'))
+    expect(active['output_count']).to eq(0)
+  end
+
+  it 'ignores missing sets but preserves explicit empty set semantics' do
+    { 'name' => ['in', 'not_in'], 'aliases.name' => ['any', 'all', 'none'] }.each do
+        |field, operators|
+      operators.each do |operator|
+        condition = { 'conditions' => '', 'conditions.0.field' => field,
+                      'conditions.0.operator' => operator }
+        expect(query(condition)['count']).to eq(2)
+        empty = condition.merge('conditions.0.values' => '1', 'condition_values' => '')
+        expected = ['in', 'any'].include?(operator) ? 0 : 2
+        expect(query(empty)['count']).to eq(expected)
+        expect(query(condition.merge('conditions.0.blank' => '1'))['count']).to eq(0)
+        # An explicit empty string element is a real value, not an unspecified set.
+        expect(query(predicate(field, operator, ''))['count']).to eq(0)
+      end
+    end
+  end
+
+  it 'ignores empty private predicates regardless of their position' do
+    first = predicate('name', 'ne', '')
+        .merge(predicate('resource_id', 'eq', tag.id.to_s, 1))
+    last = predicate('resource_id', 'eq', tag.id.to_s)
+        .merge(predicate('aliases.name', 'like', '', 1))
+    expect(query(first)['count']).to eq(2)
+    expect(query(last)['count']).to eq(2)
+    active = predicate('name', 'ne', 'unrelated')
+        .merge(predicate('aliases.name', 'like', '', 1))
+    expect(query(active)['count']).to eq(0)
+  end
+
+  it 'searches public literal markers and treats axis predicates as explicit' do
+    public_tag = create(:tag, primary_name: '********')
+    TagVersion.create!(tag: public_tag, version_no: 1, event_type: 'create',
+        name: '********', category: 'general', aliases: '', parent_tag_ids: '',
+        names: { ja: { name: '********', script_code: 'Jpan', auto_generated: false,
+                      aliases: [] } }, created_at: Time.utc(2023))
+    expect(query['count']).to eq(3)
+    ['eq', 'ne', 'like', 'in', 'not_in'].each do |operator|
+      expected = ['ne', 'not_in'].include?(operator) ? 0 : 1
+      expect(query(predicate('name', operator, '********'))['count']).to eq(expected)
+    end
+    blank = predicate('aliases.name', 'eq', '').merge('conditions.0.blank' => '1')
+    expect(query(blank)['count']).to eq(1)
+    ['********', '', nil].each do |value|
+      condition = { 'conditions' => '', 'conditions.0.field' => 'name',
+                    'conditions.0.operator' => 'axis_eq',
+                    'conditions.0.value_type' => value.nil? ? 'null' : 'string' }
+      condition['conditions.0.value'] = value unless value.nil?
+      expected = value == '********' ? 1 : 0
+      expect(query(condition)['count']).to eq(expected)
+      drill = { 'drill' => '', 'drill.0.field' => 'name',
+                'drill.0.value_type' => value.nil? ? 'null' : 'string' }
+      drill['drill.0.value'] = value unless value.nil?
+      expect(query(drill)['count']).to eq(expected)
+    end
+  end
+
+  it 'keeps public identity and timeline predicates searchable' do
+    expect(query['count']).to eq(2)
+    values = { 'resource_id' => tag.id.to_s, 'version_no' => '2',
+               'event_type' => 'discard', 'user' => actor.id.to_s,
+               'created_at' => '2023-02-01T00:00:00Z' }
+    values.each do |field, value|
+      expected = ['resource_id', 'user'].include?(field) ? 2 : 1
+      expect(query(predicate(field, 'eq', value))['count']).to eq(expected)
+    end
+    condition = predicate('name', 'ne', '秘密のタグ')
+        .merge(predicate('event_type', 'eq', 'discard', 1))
+    expect(query(condition)['count']).to eq(0)
+    expect(query(predicate('category', 'eq', '********'))['count']).to eq(0)
+    expect(query(predicate('category', 'eq', 'general'))['count']).to eq(0)
+  end
+
+  it 'keeps public predicates available after combining multiple conditions' do
+    condition = predicate('resource_id', 'eq', tag.id.to_s)
+        .merge(predicate('event_type', 'eq', 'discard', 1))
+    expect(query(condition)['count']).to eq(1)
+    condition.merge!(predicate('name', 'ne', 'unrelated', 2))
+    expect(query(condition)['count']).to eq(0)
+  end
+
+  it 'keeps real name searches for administrators and excludes private details' do
+    condition = predicate('name', 'eq', '秘密のタグ')
+    details = query(condition.merge('mode' => 'details'))
+    expect(details['versions']).to be_empty
+    expect(details['output_count']).to eq(0)
+    masked = query(predicate('name', 'eq', '********').merge('mode' => 'details'))
+    expect(masked['versions']).to be_empty
+    expect(masked['output_count']).to eq(0)
+    sign_in_as(admin)
+    expect(query(condition)['count']).to eq(2)
+    expect(query(condition.merge('mode' => 'details'))['versions'].length).to eq(2)
+  end
+
+  it 'rejects private names reached through a post tag snapshot' do
+    post = create(:post)
+    PostVersion.create!(post:, version_no: 1, event_type: 'create', title: 'post',
+        url: post.url, tags: '', parent_post_ids: '',
+        tags_json: [{ tag_id: tag.id, version_no: 1, name: '秘密のタグ',
+                     category: 'general', sections: [] }],
+        created_at: Time.utc(2024))
+    ['tags.name', 'tags.names.name', 'tags.name_ja'].each do |field|
+      ['eq', 'ne', 'like', 'any', 'all', 'none'].each do |operator|
+        ['秘密のタグ', '********', '%'].each do |value|
+          settings = predicate(field, operator, value).merge('resource' => 'posts')
+          expect(query(settings)['count']).to eq(0)
+        end
+      end
+      blank = predicate(field, 'eq', '').merge('resource' => 'posts',
+          'conditions.0.blank' => '1')
+      expect(query(blank)['count']).to eq(0)
+    end
+  end
+end

backend/spec/services/counts_pivot_security_spec.rb

@@ -69,27 +69,29 @@ RSpec.describe 'Counts masked pivot contracts' do
     end
   end
 
-  it 'preserves five tag occurrences after merging private tags and across pages' do
+  it 'counts one opaque collection per private version across pages' do
     [nil, guest, member, admin].each do |user|
       page = result(user)
-      expect(grand(page)).to eq(5)
-      expect(page[:cells].sum { _1[:count] }).to eq(5)
+      expected = user == admin ? 5 : 6
+      expect(grand(page)).to eq(expected)
+      expect(page[:cells].sum { _1[:count] }).to eq(expected)
       expect(page).not_to have_key(:excluded_count)
       paged = result(user, 'limit' => '1')
-      expect(grand(paged)).to eq(5)
+      expect(grand(paged)).to eq(expected)
       expect(result(user, 'limit' => '1', 'page' => '2')[:group_count])
           .to eq(page[:group_count])
     end
     public_page = result(nil)
     hidden = public_page[:cells].select { _1[:values].first == CountsAxes::MASKED_VALUE }
-    expect(hidden.sum { _1[:count] }).to eq(3)
+    expect(hidden.sum { _1[:count] }).to eq(4)
     expect(public_page[:axis_labels]['a0']).not_to have_key("internal:#{ tags.first.id }")
     expect(public_page.to_json).not_to include('hidden-title', 'private-discard')
   end
 
   it 'retains distinct membership when tags are not a row axis' do
     [nil, admin].each do |user|
-      expect(grand(result(user, 'rows' => 'event_type', 'cols' => 'tags'))).to eq(3)
+      expect(grand(result(user, 'rows' => 'event_type', 'cols' => 'tags')))
+          .to eq(user == admin ? 3 : 5)
       expect(result(user, 'rows' => '', 'cols' => '')[:cells].sole[:count]).to eq(5)
       details = result(user, 'mode' => 'details', 'rows' => '', 'cols' => '')
       expect(details[:count]).to eq(user == admin ? 5 : 1)
@@ -121,14 +123,13 @@ RSpec.describe 'Counts masked pivot contracts' do
   it 'counts distinct changed records when tags are not additive row tuples' do
     [nil, admin].each do |user|
       page = result(user, 'rows' => 'changed_field', 'cols' => 'event_type')
-      expect(grand(page)).to eq(2)
+      expect(grand(page)).to eq(user == admin ? 2 : 4)
       if user.nil?
         expect(page[:cells].map { _1[:values].first }.uniq).to eq([CountsAxes::MASKED_VALUE])
       end
-      # Changed discard snapshots have no tags: no full cell contributes here.
-      empty = result(user, 'rows' => 'changed_field', 'cols' => 'tags')
-      expect(empty[:cells]).to be_empty
-      expect(grand(empty)).to eq(0)
+      # Private empty collections contribute the same opaque tuple as populated ones.
+      combined = result(user, 'rows' => 'changed_field', 'cols' => 'tags')
+      expect(grand(combined)).to eq(user == admin ? 0 : 4)
     end
   end
 
@@ -154,7 +155,7 @@ RSpec.describe 'Counts masked pivot contracts' do
     expect(values.map(&:to_i)).to eq([1, 2, 10])
   end
 
-  it 'uses public identity fields and masked values for unauthorised filters' do
+  it 'keeps public identity predicates and excludes private attributes' do
     hidden_id = PostVersion.first.post_id.to_s
     base = { 'rows' => '', 'cols' => '' }
     condition = lambda do |field, operator, value|
@@ -170,25 +171,121 @@ RSpec.describe 'Counts masked pivot contracts' do
     expect(result(nil, base.merge(condition.call('title', 'eq', 'hidden-title-0')))[:cells]
         .sole[:count]).to eq(0)
     expect(result(nil, base.merge(condition.call('title', 'ne', 'hidden-title-0')))[:cells]
-        .sole[:count]).to eq(5)
+        .sole[:count]).to eq(1)
     expect(result(nil, base.merge(condition.call('title', 'eq', '********')))[:cells]
-        .sole[:count]).to eq(4)
+        .sole[:count]).to eq(0)
     expect(result(nil, base.merge(condition.call('video_ms', 'eq', '********')))[:cells]
-        .sole[:count]).to eq(4)
+        .sole[:count]).to eq(0)
     expect(result(nil, base.merge(condition.call('title', 'like', '%title%')))[:cells]
         .sole[:count]).to eq(1)
     expect(result(nil, base.merge(condition.call('title', 'like', '%')))[:cells]
-        .sole[:count]).to eq(5)
+        .sole[:count]).to eq(1)
     expect(result(nil, base.merge('tags' => 'C'))[:cells].sole[:count]).to eq(1)
-    expect(result(nil, base.merge('tags' => 'not:C'))[:cells].sole[:count]).to eq(4)
+    expect(result(nil, base.merge('tags' => 'not:C'))[:cells].sole[:count]).to eq(0)
   end
 
-  it 'rejects server-only coordinates in column order and drilldown' do
+  it 'retains public OR branches while rejecting private predicates on hidden rows' do
+    HistoryAccess.set(user: nil) do
+      population = CountsPopulation.new(params.merge('rows' => '', 'cols' => ''),
+          authorisation_flg: false)
+      condition = CountsCondition.new(population)
+      public_scope = condition.apply(population.scope,
+          { 'field' => 'event_type', 'operator' => 'eq', 'value' => 'discard' })
+      private_scope = condition.apply(population.scope,
+          { 'field' => 'title', 'operator' => 'ne', 'value' => 'unrelated' })
+      masked_scope = condition.apply(population.scope,
+          { 'field' => 'title', 'operator' => 'eq', 'value' => '********' })
+      expect(public_scope.except(:select).count).to eq(2)
+      expect(public_scope.or(private_scope).except(:select).count).to eq(3)
+      expect(private_scope.or(public_scope).except(:select).count).to eq(3)
+      expect(public_scope.or(masked_scope).except(:select).count).to eq(2)
+      expect(public_scope.merge(private_scope).except(:select).count).to eq(0)
+    end
+  end
+
+  it 'excludes private collections from positive, negative, AND and OR tag searches' do
+    settings = { 'rows' => '', 'cols' => '' }
+    { 'C' => 1, 'not:C' => 0, 'not:unknown' => 1, '********' => 0,
+      'not:********' => 1, 'not:C C' => 0, 'not:C type:or C' => 1 }.each do |value, count|
+      expect(result(nil, settings.merge('tags' => value))[:cells].sole[:count]).to eq(count)
+    end
+  end
+
+  it 'ignores empty numeric and interval bounds without restricting private history' do
+    settings = { 'rows' => '', 'cols' => '', 'conditions' => '' }
+    { 'video_ms' => ['eq', 'ne', 'gt', 'ge', 'lt', 'le', 'range'],
+      'tags.sections' => ['range'] }.each do |field, operators|
+      operators.each do |operator|
+        condition = settings.merge('conditions.0.field' => field,
+            'conditions.0.operator' => operator)
+        empty = condition.merge('conditions.0.value' => '')
+        expect(result(nil, condition)[:cells].sole[:count]).to eq(5)
+        expect(result(nil, empty)[:cells].sole[:count]).to eq(5)
+        expect(result(nil, empty.merge('conditions.0.blank' => '0'))[:cells]
+            .sole[:count]).to eq(5)
+        if operator == 'range'
+          empty['conditions.0.end'] = ''
+          expect(result(nil, empty)[:cells].sole[:count]).to eq(5)
+          ['conditions.0.value', 'conditions.0.end'].each do |key|
+            active = empty.merge(key => '100')
+            expect(result(nil, active)[:cells].sole[:count]).to eq(0)
+          end
+        end
+      end
+    end
+  end
+
+  it 'keeps typed marker sets active after incompatible marker values are removed' do
+    settings = { 'rows' => '', 'cols' => '', 'conditions' => '',
+                 'conditions.0.field' => 'video_ms', 'conditions.0.values' => '1',
+                 'condition_values' => '', 'condition_values.0.condition' => '0',
+                 'condition_values.0.value_type' => 'string',
+                 'condition_values.0.value' => '********' }
+    ['in', 'not_in'].each do |operator|
+      expected = operator == 'in' ? 0 : 1
+      condition = settings.merge('conditions.0.operator' => operator)
+      expect(result(nil, condition)[:cells].sole[:count]).to eq(expected)
+    end
+  end
+
+  it 'restricts active numeric comparisons and preserves empty selection semantics' do
+    settings = { 'rows' => '', 'cols' => '', 'conditions' => '',
+                 'conditions.0.field' => 'video_ms' }
+    ['eq', 'ne', 'gt', 'ge', 'lt', 'le', 'range'].each do |operator|
+      condition = settings.merge('conditions.0.operator' => operator,
+          'conditions.0.value' => '0')
+      condition['conditions.0.end'] = '100' if operator == 'range'
+      # The one public row is NULL; private NULLs and real values never match.
+      expected = operator == 'ne' ? 1 : 0
+      expect(result(nil, condition)[:cells].sole[:count]).to eq(expected)
+    end
+    ['in', 'not_in'].each do |operator|
+      condition = settings.merge('conditions.0.operator' => operator,
+          'conditions.0.values' => '1', 'condition_values' => '')
+      expected = operator == 'not_in' ? 5 : 0
+      expect(result(nil, condition)[:cells].sole[:count]).to eq(expected)
+    end
+    blank = settings.merge('conditions.0.operator' => 'eq', 'conditions.0.blank' => '1')
+    expect(result(nil, blank)[:cells].sole[:count]).to eq(1)
+  end
+
+  it 'preserves empty tag sets and restricts active blank collection predicates' do
+    settings = { 'rows' => '', 'cols' => '', 'conditions' => '',
+                 'conditions.0.field' => 'tags' }
+    ['any', 'all', 'none'].each do |operator|
+      condition = settings.merge('conditions.0.operator' => operator,
+          'conditions.0.values' => '1', 'condition_values' => '')
+      expected = operator == 'any' ? 0 : 5
+      expect(result(nil, condition)[:cells].sole[:count]).to eq(expected)
+    end
+    blank = settings.merge('conditions.0.operator' => 'eq', 'conditions.0.blank' => '1')
+    expect(result(nil, blank)[:cells].sole[:count]).to eq(0)
+  end
+
+  it 'rejects server-only coordinates in column order' do
     probes = [
         { 'cols' => 'tags', 'column_order' => '', 'column_order.0.value_type' => 'string',
-          'column_order.0.value' => CountsAxes::MASKED_VALUE },
-        { 'drill' => '', 'drill.0.field' => 'tags', 'drill.0.value_type' => 'string',
-          'drill.0.value' => CountsAxes::MASKED_VALUE }]
+          'column_order.0.value' => CountsAxes::MASKED_VALUE }]
     HistoryAccess.set(user: nil) do
       probes.each do |probe|
         expect { CountsTableQuery.new(params.merge(probe)).call }
@@ -197,6 +294,54 @@ RSpec.describe 'Counts masked pivot contracts' do
     end
   end
 
+
+  it 'keeps private collection and change counts independent of stored elements' do
+    settings = [
+        { 'rows' => 'tags', 'cols' => 'changed_field' },
+        { 'rows' => 'changed_field tags', 'cols' => '' }]
+    original = settings.map { result(nil, _1) }
+    snapshot = tags.map do |tag|
+      { tag_id: tag.id, version_no: 1, sections: [] }
+    end
+    hidden_posts = PostVersion.where(event_type: 'discard').pluck(:post_id)
+    PostVersion.where(post_id: hidden_posts).update_all(
+        title: 'same hidden value', tags_json: snapshot)
+    expect(settings.map { result(nil, _1) }).to eq(original)
+  end
+
+  it 'excludes private tag and change coordinates from drill conditions' do
+    ['tags', 'changed_field'].each do |field|
+      settings = { 'rows' => field, 'cols' => '' }
+      page = result(nil, settings)
+      cell = page[:cells].find { _1[:values] == [CountsAxes::MASKED_VALUE] }
+      expect(cell[:count]).to eq(4)
+      [CountsAxes::MASKED_VALUE, CountsAxes::MASKED_TEXT].each do |value|
+        drill = { 'drill' => '', 'drill.0.field' => field,
+                  'drill.0.value_type' => 'string', 'drill.0.value' => value }
+        expect(result(nil, settings.merge(drill))[:cells]).to be_empty
+        details = result(nil, settings.merge(drill).merge('mode' => 'details'))
+        expect(details[:versions]).to be_empty
+        condition = { 'conditions' => '', 'conditions.0.field' => field,
+                      'conditions.0.operator' => 'axis_eq',
+                      'conditions.0.value_type' => 'string', 'conditions.0.value' => value }
+        expect(result(nil, settings.merge(condition))[:cells]).to be_empty
+      end
+    end
+  end
+
+  it 'rejects real private tag and change coordinates for non-admin users' do
+    [nil, guest, member].each do |user|
+      { 'tags' => "internal:#{ tags.first.id }", 'changed_field' => 'title' }.each do
+          |field, value|
+        settings = { 'rows' => '', 'cols' => '', 'drill' => '',
+                     'drill.0.field' => field, 'drill.0.value_type' => 'string',
+                     'drill.0.value' => value }
+        expect(result(user, settings)[:cells].sole[:count]).to eq(0)
+        expect(result(admin, settings)[:cells].sole[:count]).to be > 0
+      end
+    end
+  end
+
   it 'keeps public column sorting usable while guessed private coordinates match nothing' do
     first = result(nil, 'rows' => 'event_type', 'cols' => 'tags')
     probe = result(nil, 'rows' => 'event_type', 'cols' => 'tags',
@@ -212,7 +357,7 @@ RSpec.describe 'Counts masked pivot contracts' do
     HistoryAccess.set(user: nil) do
       payload = params.merge('format' => 'json')
       json = JSON.parse(CountsTableExport.new(payload).body.to_a.join)
-      expect(json['data'].sum { _1['kind'] == 'data' ? _1['count'] : 0 }).to eq(5)
+      expect(json['data'].sum { _1['kind'] == 'data' ? _1['count'] : 0 }).to eq(6)
       expect(json.to_json).not_to include('hidden-title', 'private-discard')
       csv = CountsTableExport.new(payload.merge('format' => 'csv')).body.to_a.join
       expect(csv).to include('********')