fix(search): harden authorized results and ordering
This commit is contained in:
@@ -1,7 +1,10 @@
|
||||
package search
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
"sort"
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
"gorm.io/gorm"
|
||||
"senlinai-agent/backend/internal/models"
|
||||
@@ -11,7 +14,15 @@ type Service struct {
|
||||
db *gorm.DB
|
||||
}
|
||||
|
||||
const maxSearchResults = 50
|
||||
const (
|
||||
maxSearchResults = 50
|
||||
maxSearchSnippetRunes = 240
|
||||
)
|
||||
|
||||
type rankedSearchResult struct {
|
||||
result SearchResultDTO
|
||||
updatedAt time.Time
|
||||
}
|
||||
|
||||
// SearchResultDTO 是搜索接口的稳定结果,所有关联均使用公开 identity。
|
||||
type SearchResultDTO struct {
|
||||
@@ -42,84 +53,110 @@ func (s *Service) Search(userID uint, query string) ([]SearchResultDTO, error) {
|
||||
if query == "" {
|
||||
return []SearchResultDTO{}, nil
|
||||
}
|
||||
if s.database().Dialector.Name() == "postgres" {
|
||||
return s.searchPostgres(userID, query)
|
||||
}
|
||||
|
||||
like := containsPattern(query)
|
||||
results := []SearchResultDTO{}
|
||||
operator := "LIKE"
|
||||
if s.database().Dialector.Name() == "postgres" {
|
||||
operator = "ILIKE"
|
||||
}
|
||||
|
||||
var projects []models.SenlinAgentProject
|
||||
if err := s.database().Where("owner_id = ? AND (name LIKE ? ESCAPE '!' OR description LIKE ? ESCAPE '!')", userID, like, like).Limit(maxSearchResults).Find(&projects).Error; err != nil {
|
||||
projectFilter := fmt.Sprintf("owner_id = ? AND (name %s ? ESCAPE '!' OR description %s ? ESCAPE '!')", operator, operator)
|
||||
if err := s.database().Where(projectFilter, userID, like, like).
|
||||
Order("updated_at DESC").Order("identity ASC").
|
||||
Limit(maxSearchResults).Find(&projects).Error; err != nil {
|
||||
return nil, err
|
||||
}
|
||||
projectResults := make([]rankedSearchResult, 0, len(projects))
|
||||
for _, project := range projects {
|
||||
results = append(results, SearchResultDTO{Type: "project", ID: project.Identity, ProjectID: project.Identity, Title: project.Name, Snippet: project.Description})
|
||||
projectResults = append(projectResults, rankedSearchResult{
|
||||
result: SearchResultDTO{Type: "project", ID: project.Identity, ProjectID: project.Identity, Title: project.Name, Snippet: project.Description},
|
||||
updatedAt: project.UpdatedAt,
|
||||
})
|
||||
}
|
||||
|
||||
var tasks []models.SenlinAgentTask
|
||||
taskFilter := fmt.Sprintf("(senlin_agent_projects.owner_id = ? OR senlin_agent_tasks.assignee_id = ?) AND (senlin_agent_tasks.title %s ? ESCAPE '!' OR senlin_agent_tasks.description %s ? ESCAPE '!')", operator, operator)
|
||||
if err := s.database().Select("senlin_agent_tasks.*").Joins("JOIN senlin_agent_projects ON senlin_agent_projects.id = senlin_agent_tasks.project_id").
|
||||
Where("(senlin_agent_projects.owner_id = ? OR senlin_agent_tasks.assignee_id = ?) AND (senlin_agent_tasks.title LIKE ? ESCAPE '!' OR senlin_agent_tasks.description LIKE ? ESCAPE '!')", userID, userID, like, like).
|
||||
Where(taskFilter, userID, userID, like, like).
|
||||
Order("senlin_agent_tasks.updated_at DESC").Order("senlin_agent_tasks.identity ASC").
|
||||
Limit(maxSearchResults).
|
||||
Find(&tasks).Error; err != nil {
|
||||
return nil, err
|
||||
}
|
||||
taskResults := make([]rankedSearchResult, 0, len(tasks))
|
||||
for _, task := range tasks {
|
||||
results = append(results, SearchResultDTO{Type: "task", ID: task.Identity, ProjectID: task.ProjectIdentity, Title: task.Title, Snippet: task.Description})
|
||||
taskResults = append(taskResults, rankedSearchResult{
|
||||
result: SearchResultDTO{Type: "task", ID: task.Identity, ProjectID: task.ProjectIdentity, Title: task.Title, Snippet: task.Description},
|
||||
updatedAt: task.UpdatedAt,
|
||||
})
|
||||
}
|
||||
|
||||
var notes []models.SenlinAgentNote
|
||||
if err := s.database().Select("senlin_agent_notes.*").Joins("JOIN senlin_agent_projects ON senlin_agent_projects.id = senlin_agent_notes.project_id").
|
||||
Where(`(senlin_agent_projects.owner_id = ? OR EXISTS (
|
||||
noteFilter := fmt.Sprintf(`(senlin_agent_projects.owner_id = ? OR EXISTS (
|
||||
SELECT 1 FROM senlin_agent_task_shares
|
||||
JOIN senlin_agent_tasks ON senlin_agent_tasks.id = senlin_agent_task_shares.task_id
|
||||
WHERE senlin_agent_task_shares.object_type = 'note'
|
||||
AND senlin_agent_task_shares.object_id = senlin_agent_notes.id
|
||||
AND senlin_agent_tasks.project_id = senlin_agent_notes.project_id
|
||||
AND senlin_agent_tasks.assignee_id = ?
|
||||
)) AND (senlin_agent_notes.title LIKE ? ESCAPE '!' OR senlin_agent_notes.markdown LIKE ? ESCAPE '!')`, userID, userID, like, like).
|
||||
)) AND (senlin_agent_notes.title %s ? ESCAPE '!' OR senlin_agent_notes.markdown %s ? ESCAPE '!')`, operator, operator)
|
||||
if err := s.database().Select("senlin_agent_notes.*").Joins("JOIN senlin_agent_projects ON senlin_agent_projects.id = senlin_agent_notes.project_id").
|
||||
Where(noteFilter, userID, userID, like, like).
|
||||
Order("senlin_agent_notes.updated_at DESC").Order("senlin_agent_notes.identity ASC").
|
||||
Limit(maxSearchResults).
|
||||
Find(¬es).Error; err != nil {
|
||||
return nil, err
|
||||
}
|
||||
noteResults := make([]rankedSearchResult, 0, len(notes))
|
||||
for _, note := range notes {
|
||||
results = append(results, SearchResultDTO{Type: "note", ID: note.Identity, ProjectID: note.ProjectIdentity, Title: note.Title, Snippet: note.Markdown})
|
||||
noteResults = append(noteResults, rankedSearchResult{
|
||||
result: SearchResultDTO{Type: "note", ID: note.Identity, ProjectID: note.ProjectIdentity, Title: note.Title, Snippet: note.Markdown},
|
||||
updatedAt: note.UpdatedAt,
|
||||
})
|
||||
}
|
||||
if len(results) > maxSearchResults {
|
||||
results = results[:maxSearchResults]
|
||||
}
|
||||
return results, nil
|
||||
|
||||
return stableFairLimit(projectResults, taskResults, noteResults), nil
|
||||
}
|
||||
|
||||
func (s *Service) searchPostgres(userID uint, query string) ([]SearchResultDTO, error) {
|
||||
var results []SearchResultDTO
|
||||
like := containsPattern(query)
|
||||
err := s.database().Raw(`
|
||||
SELECT 'project' AS type, p.identity AS id, p.identity AS project_id, p.name AS title, p.description AS snippet
|
||||
FROM senlin_agent_projects p
|
||||
WHERE p.owner_id = ?
|
||||
AND (coalesce(p.name, '') ILIKE ? ESCAPE '!' OR coalesce(p.description, '') ILIKE ? ESCAPE '!')
|
||||
UNION ALL
|
||||
SELECT 'task' AS type, t.identity AS id, p.identity AS project_id, t.title, t.description AS snippet
|
||||
FROM senlin_agent_tasks t
|
||||
JOIN senlin_agent_projects p ON p.id = t.project_id
|
||||
WHERE (p.owner_id = ? OR t.assignee_id = ?)
|
||||
AND (coalesce(t.title, '') ILIKE ? ESCAPE '!' OR coalesce(t.description, '') ILIKE ? ESCAPE '!')
|
||||
UNION ALL
|
||||
SELECT 'note' AS type, n.identity AS id, p.identity AS project_id, n.title, n.markdown AS snippet
|
||||
FROM senlin_agent_notes n
|
||||
JOIN senlin_agent_projects p ON p.id = n.project_id
|
||||
WHERE (p.owner_id = ? OR EXISTS (
|
||||
SELECT 1 FROM senlin_agent_task_shares ts
|
||||
JOIN senlin_agent_tasks t ON t.id = ts.task_id
|
||||
WHERE ts.object_type = 'note'
|
||||
AND ts.object_id = n.id
|
||||
AND t.project_id = n.project_id
|
||||
AND t.assignee_id = ?
|
||||
))
|
||||
AND (coalesce(n.title, '') ILIKE ? ESCAPE '!' OR coalesce(n.markdown, '') ILIKE ? ESCAPE '!')
|
||||
LIMIT ?
|
||||
`, userID, like, like, userID, userID, like, like, userID, userID, like, like, maxSearchResults).Scan(&results).Error
|
||||
return results, err
|
||||
func stableFairLimit(buckets ...[]rankedSearchResult) []SearchResultDTO {
|
||||
for _, bucket := range buckets {
|
||||
sort.Slice(bucket, func(left, right int) bool {
|
||||
if bucket[left].updatedAt.Equal(bucket[right].updatedAt) {
|
||||
return bucket[left].result.ID < bucket[right].result.ID
|
||||
}
|
||||
return bucket[left].updatedAt.After(bucket[right].updatedAt)
|
||||
})
|
||||
}
|
||||
|
||||
results := make([]SearchResultDTO, 0, maxSearchResults)
|
||||
for position := 0; len(results) < maxSearchResults; position++ {
|
||||
added := false
|
||||
for _, bucket := range buckets {
|
||||
if position >= len(bucket) {
|
||||
continue
|
||||
}
|
||||
result := bucket[position].result
|
||||
result.Snippet = boundedSnippet(result.Snippet)
|
||||
results = append(results, result)
|
||||
added = true
|
||||
if len(results) == maxSearchResults {
|
||||
return results
|
||||
}
|
||||
}
|
||||
if !added {
|
||||
return results
|
||||
}
|
||||
}
|
||||
return results
|
||||
}
|
||||
|
||||
func boundedSnippet(value string) string {
|
||||
runes := []rune(value)
|
||||
if len(runes) <= maxSearchSnippetRunes {
|
||||
return value
|
||||
}
|
||||
return string(runes[:maxSearchSnippetRunes])
|
||||
}
|
||||
|
||||
func containsPattern(query string) string {
|
||||
|
||||
@@ -1,8 +1,11 @@
|
||||
package search
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
"os"
|
||||
"strings"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/stretchr/testify/require"
|
||||
"gorm.io/driver/postgres"
|
||||
@@ -45,6 +48,65 @@ func TestPostgresSearchKeepsAssignmentAndExplicitShareBoundaries(t *testing.T) {
|
||||
require.ElementsMatch(t, []string{assigned.Identity, sharedNote.Identity}, []string{results[0].ID, results[1].ID})
|
||||
}
|
||||
|
||||
func TestPostgresSearchUsesStableFairLimitAndRuneBoundedSnippets(t *testing.T) {
|
||||
database := newPostgresSearchTestDB(t)
|
||||
owner := createSearchUser(t, database, "postgres-search-order@example.com")
|
||||
baseTime := time.Date(2026, time.July, 21, 12, 0, 0, 0, time.UTC)
|
||||
for index := 0; index < 60; index++ {
|
||||
project := models.SenlinAgentProject{
|
||||
Identity: fmt.Sprintf("00000000-0000-7001-8000-%012d", index),
|
||||
OwnerID: owner.ID,
|
||||
Name: fmt.Sprintf("稳定排序项目 %02d", index),
|
||||
Identifier: fmt.Sprintf("PG-SORT-%02d", index),
|
||||
UpdatedAt: baseTime.Add(time.Duration(index) * time.Minute),
|
||||
}
|
||||
require.NoError(t, database.Create(&project).Error)
|
||||
require.NoError(t, database.Create(&models.SenlinAgentTask{
|
||||
Identity: fmt.Sprintf("00000000-0000-7002-8000-%012d", index),
|
||||
ProjectID: project.ID,
|
||||
CreatedBy: owner.ID,
|
||||
Title: fmt.Sprintf("稳定排序任务 %02d", index),
|
||||
Status: "open",
|
||||
UpdatedAt: baseTime.Add(time.Duration(index) * time.Minute),
|
||||
}).Error)
|
||||
require.NoError(t, database.Create(&models.SenlinAgentNote{
|
||||
Identity: fmt.Sprintf("00000000-0000-7003-8000-%012d", index),
|
||||
ProjectID: project.ID,
|
||||
CreatedBy: owner.ID,
|
||||
Title: fmt.Sprintf("稳定排序笔记 %02d", index),
|
||||
Markdown: strings.Repeat("森", 300) + "尾",
|
||||
UpdatedAt: baseTime.Add(time.Duration(index) * time.Minute),
|
||||
}).Error)
|
||||
}
|
||||
|
||||
service := NewService(database)
|
||||
results, err := service.Search(owner.ID, "稳定排序")
|
||||
require.NoError(t, err)
|
||||
require.Len(t, results, maxSearchResults)
|
||||
require.Equal(t,
|
||||
[]string{"project", "task", "note", "project", "task", "note", "project", "task", "note"},
|
||||
resultTypes(results[:9]),
|
||||
)
|
||||
require.Equal(t, "00000000-0000-7001-8000-000000000059", results[0].ID)
|
||||
require.Equal(t, "00000000-0000-7002-8000-000000000059", results[1].ID)
|
||||
require.Equal(t, "00000000-0000-7003-8000-000000000059", results[2].ID)
|
||||
counts := map[string]int{}
|
||||
for _, result := range results {
|
||||
counts[result.Type]++
|
||||
}
|
||||
require.Equal(t, map[string]int{"project": 17, "task": 17, "note": 16}, counts)
|
||||
|
||||
again, err := service.Search(owner.ID, "稳定排序")
|
||||
require.NoError(t, err)
|
||||
require.Equal(t, results, again)
|
||||
|
||||
noteResults, err := service.Search(owner.ID, "森")
|
||||
require.NoError(t, err)
|
||||
require.NotEmpty(t, noteResults)
|
||||
require.Len(t, []rune(noteResults[0].Snippet), 240)
|
||||
require.NotContains(t, noteResults[0].Snippet, "尾")
|
||||
}
|
||||
|
||||
func newPostgresSearchTestDB(t *testing.T) *gorm.DB {
|
||||
t.Helper()
|
||||
dsn := os.Getenv("DATABASE_URL")
|
||||
|
||||
@@ -2,7 +2,9 @@ package search
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
"strings"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/glebarez/sqlite"
|
||||
"github.com/stretchr/testify/require"
|
||||
@@ -26,6 +28,91 @@ func TestSearchFindsNoteBody(t *testing.T) {
|
||||
require.Equal(t, "note", results[0].Type)
|
||||
}
|
||||
|
||||
func TestSearchAppliesStableFairGlobalLimitAcrossResultTypes(t *testing.T) {
|
||||
database, err := gorm.Open(sqlite.Open(fmt.Sprintf("file:%s?mode=memory&cache=shared", t.Name())), &gorm.Config{})
|
||||
require.NoError(t, err)
|
||||
require.NoError(t, models.AutoMigrate(database))
|
||||
models.DBService = database
|
||||
|
||||
baseTime := time.Date(2026, time.July, 21, 12, 0, 0, 0, time.UTC)
|
||||
for index := 0; index < 60; index++ {
|
||||
project := models.SenlinAgentProject{
|
||||
Identity: fmt.Sprintf("00000000-0000-7001-8000-%012d", index),
|
||||
OwnerID: 7,
|
||||
Name: fmt.Sprintf("稳定排序项目 %02d", index),
|
||||
Identifier: fmt.Sprintf("SORT-%02d", index),
|
||||
UpdatedAt: baseTime.Add(time.Duration(index) * time.Minute),
|
||||
}
|
||||
require.NoError(t, database.Create(&project).Error)
|
||||
require.NoError(t, database.Create(&models.SenlinAgentTask{
|
||||
Identity: fmt.Sprintf("00000000-0000-7002-8000-%012d", index),
|
||||
ProjectID: project.ID,
|
||||
CreatedBy: 7,
|
||||
Title: fmt.Sprintf("稳定排序任务 %02d", index),
|
||||
Status: "open",
|
||||
UpdatedAt: baseTime.Add(time.Duration(index) * time.Minute),
|
||||
}).Error)
|
||||
require.NoError(t, database.Create(&models.SenlinAgentNote{
|
||||
Identity: fmt.Sprintf("00000000-0000-7003-8000-%012d", index),
|
||||
ProjectID: project.ID,
|
||||
CreatedBy: 7,
|
||||
Title: fmt.Sprintf("稳定排序笔记 %02d", index),
|
||||
UpdatedAt: baseTime.Add(time.Duration(index) * time.Minute),
|
||||
}).Error)
|
||||
}
|
||||
|
||||
service := NewService(database)
|
||||
results, err := service.Search(7, "稳定排序")
|
||||
require.NoError(t, err)
|
||||
require.Len(t, results, maxSearchResults)
|
||||
|
||||
counts := map[string]int{}
|
||||
for _, result := range results {
|
||||
counts[result.Type]++
|
||||
}
|
||||
require.Equal(t, map[string]int{"project": 17, "task": 17, "note": 16}, counts)
|
||||
require.Equal(t,
|
||||
[]string{"project", "task", "note", "project", "task", "note", "project", "task", "note"},
|
||||
resultTypes(results[:9]),
|
||||
)
|
||||
require.Equal(t, "00000000-0000-7001-8000-000000000059", results[0].ID)
|
||||
require.Equal(t, "00000000-0000-7002-8000-000000000059", results[1].ID)
|
||||
require.Equal(t, "00000000-0000-7003-8000-000000000059", results[2].ID)
|
||||
|
||||
again, err := service.Search(7, "稳定排序")
|
||||
require.NoError(t, err)
|
||||
require.Equal(t, results, again)
|
||||
}
|
||||
|
||||
func TestSearchBoundsChineseSnippetsByRune(t *testing.T) {
|
||||
database, err := gorm.Open(sqlite.Open(fmt.Sprintf("file:%s?mode=memory&cache=shared", t.Name())), &gorm.Config{})
|
||||
require.NoError(t, err)
|
||||
require.NoError(t, models.AutoMigrate(database))
|
||||
models.DBService = database
|
||||
project := models.SenlinAgentProject{OwnerID: 7, Name: "摘要项目", Identifier: "SNIPPET"}
|
||||
require.NoError(t, database.Create(&project).Error)
|
||||
require.NoError(t, database.Create(&models.SenlinAgentNote{
|
||||
ProjectID: project.ID,
|
||||
CreatedBy: 7,
|
||||
Title: "中文摘要",
|
||||
Markdown: strings.Repeat("森", 300) + "尾",
|
||||
}).Error)
|
||||
|
||||
results, err := NewService(database).Search(7, "森")
|
||||
require.NoError(t, err)
|
||||
require.Len(t, results, 1)
|
||||
require.Len(t, []rune(results[0].Snippet), 240)
|
||||
require.NotContains(t, results[0].Snippet, "尾")
|
||||
}
|
||||
|
||||
func resultTypes(results []SearchResultDTO) []string {
|
||||
types := make([]string, 0, len(results))
|
||||
for _, result := range results {
|
||||
types = append(types, result.Type)
|
||||
}
|
||||
return types
|
||||
}
|
||||
|
||||
func TestSearchOnlyReturnsNavigableProjectTaskAndNoteResults(t *testing.T) {
|
||||
database, err := gorm.Open(sqlite.Open(fmt.Sprintf("file:%s?mode=memory&cache=shared", t.Name())), &gorm.Config{})
|
||||
require.NoError(t, err)
|
||||
|
||||
Reference in New Issue
Block a user