Files
Enterprise-Knowledge-Base/internal/application/service/knowledge_process_config_test.go

307 lines
9.5 KiB
Go

package service
import (
"context"
"testing"
werrors "github.com/Tencent/WeKnora/internal/errors"
"github.com/Tencent/WeKnora/internal/types"
"github.com/stretchr/testify/require"
)
func processConfigBoolPtr(v bool) *bool {
return &v
}
func testKBWithGraphEnabled(enabled bool) *types.KnowledgeBase {
return &types.KnowledgeBase{
IndexingStrategy: types.IndexingStrategy{GraphEnabled: enabled},
ExtractConfig: &types.ExtractConfig{Enabled: enabled},
}
}
func TestResolveProcessConfig_OverridesChunkSize(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
ChunkingConfig: types.ChunkingConfig{ChunkSize: 512, ChunkOverlap: 50},
}
overrides := &types.KnowledgeProcessOverrides{
ChunkingConfig: &types.ChunkingConfig{ChunkSize: 2048},
}
eff := ResolveProcessConfig(kb, overrides)
require.Equal(t, 2048, eff.ChunkingConfig.ChunkSize)
require.Equal(t, 50, eff.ChunkingConfig.ChunkOverlap)
}
func TestResolveProcessConfig_OverrideTogglesParentChild(t *testing.T) {
t.Parallel()
// KB has parent-child on; override snapshot turns it off.
kbOn := &types.KnowledgeBase{
ChunkingConfig: types.ChunkingConfig{ChunkSize: 512, EnableParentChild: true},
}
effOff := ResolveProcessConfig(kbOn, &types.KnowledgeProcessOverrides{
ChunkingConfig: &types.ChunkingConfig{ChunkSize: 512, EnableParentChild: false},
})
require.False(t, effOff.ChunkingConfig.EnableParentChild)
// KB has parent-child off; override snapshot turns it on.
kbOff := &types.KnowledgeBase{
ChunkingConfig: types.ChunkingConfig{ChunkSize: 512, EnableParentChild: false},
}
effOn := ResolveProcessConfig(kbOff, &types.KnowledgeProcessOverrides{
ChunkingConfig: &types.ChunkingConfig{ChunkSize: 512, EnableParentChild: true},
})
require.True(t, effOn.ChunkingConfig.EnableParentChild)
}
func TestResolveProcessConfig_GraphDisabled(t *testing.T) {
t.Parallel()
kb := testKBWithGraphEnabled(true)
overrides := &types.KnowledgeProcessOverrides{GraphEnabled: processConfigBoolPtr(false)}
eff := ResolveProcessConfig(kb, overrides)
require.False(t, eff.GraphEnabled)
}
func TestResolveProcessConfig_GraphRequiresExtractEnabled(t *testing.T) {
t.Parallel()
kb := testKBWithGraphEnabled(true)
overrides := &types.KnowledgeProcessOverrides{
GraphEnabled: processConfigBoolPtr(true),
ExtractConfig: &types.ExtractConfig{Enabled: false},
}
eff := ResolveProcessConfig(kb, overrides)
require.False(t, eff.ExtractConfig.Enabled)
require.False(t, eff.GraphEnabled)
}
func TestResolveProcessConfig_NilOverridesUsesKBDefaults(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
ChunkingConfig: types.ChunkingConfig{ChunkSize: 512, ChunkOverlap: 50},
VLMConfig: types.VLMConfig{Enabled: true, ModelID: "vlm-1"},
ASRConfig: types.ASRConfig{Enabled: true, ModelID: "asr-1"},
QuestionGenerationConfig: &types.QuestionGenerationConfig{
Enabled: true,
QuestionCount: 3,
},
IndexingStrategy: types.IndexingStrategy{GraphEnabled: true},
ExtractConfig: &types.ExtractConfig{Enabled: true, Tags: []string{"tag-a"}},
}
eff := ResolveProcessConfig(kb, nil)
require.Equal(t, 512, eff.ChunkingConfig.ChunkSize)
require.Equal(t, 50, eff.ChunkingConfig.ChunkOverlap)
require.True(t, eff.EnableMultimodel)
require.Equal(t, "vlm-1", eff.VLMConfig.ModelID)
require.Equal(t, "asr-1", eff.ASRConfig.ModelID)
require.True(t, eff.QuestionGenerationConfig.Enabled)
require.Equal(t, 3, eff.QuestionGenerationConfig.QuestionCount)
require.True(t, eff.GraphEnabled)
require.True(t, eff.ExtractConfig.Enabled)
require.Equal(t, []string{"tag-a"}, eff.ExtractConfig.Tags)
}
func TestBuildSplitterConfigFromChunking_UsesEffectiveChunkingConfig(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
ChunkingConfig: types.ChunkingConfig{ChunkSize: 512, ChunkOverlap: 50, Strategy: "token"},
}
overrides := &types.KnowledgeProcessOverrides{
ChunkingConfig: &types.ChunkingConfig{ChunkSize: 1500, ChunkOverlap: 120, Strategy: "character"},
}
eff := ResolveProcessConfig(kb, overrides)
cfg := buildSplitterConfigFromChunking(eff.ChunkingConfig)
require.Equal(t, 1500, cfg.ChunkSize)
require.Equal(t, 120, cfg.ChunkOverlap)
require.Equal(t, "character", cfg.Strategy)
}
func TestEffectiveChunkingConfig_ResolveParserEngineFromOverrides(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
ChunkingConfig: types.ChunkingConfig{
ParserEngineRules: []types.ParserEngineRule{
{FileTypes: []string{"pdf"}, Engine: "builtin"},
},
},
}
overrides := &types.KnowledgeProcessOverrides{
ParserEngineRules: []types.ParserEngineRule{
{FileTypes: []string{"pdf"}, Engine: "mineru"},
},
}
eff := ResolveProcessConfig(kb, overrides)
require.Equal(t, "mineru", eff.ChunkingConfig.ResolveParserEngine("pdf"))
}
func TestResolveProcessConfig_ParserEngineRulesReplaced(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
ChunkingConfig: types.ChunkingConfig{
ParserEngineRules: []types.ParserEngineRule{
{FileTypes: []string{"pdf"}, Engine: "builtin"},
},
},
}
overrides := &types.KnowledgeProcessOverrides{
ParserEngineRules: []types.ParserEngineRule{
{FileTypes: []string{"docx"}, Engine: "custom"},
},
}
eff := ResolveProcessConfig(kb, overrides)
require.Len(t, eff.ChunkingConfig.ParserEngineRules, 1)
require.Equal(t, []string{"docx"}, eff.ChunkingConfig.ParserEngineRules[0].FileTypes)
require.Equal(t, "custom", eff.ChunkingConfig.ParserEngineRules[0].Engine)
}
func TestResolveProcessConfig_EnableMultimodelOverride(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
VLMConfig: types.VLMConfig{Enabled: true, ModelID: "vlm-1"},
}
overrides := &types.KnowledgeProcessOverrides{
EnableMultimodel: processConfigBoolPtr(false),
}
eff := ResolveProcessConfig(kb, overrides)
require.False(t, eff.EnableMultimodel)
}
func TestResolveProcessConfig_ExtractConfigFieldMerge(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
ExtractConfig: &types.ExtractConfig{
Enabled: true,
Text: "base text",
Tags: []string{"base-tag"},
},
}
overrides := &types.KnowledgeProcessOverrides{
ExtractConfig: &types.ExtractConfig{
Enabled: true,
Tags: []string{"override-tag"},
},
}
eff := ResolveProcessConfig(kb, overrides)
require.True(t, eff.ExtractConfig.Enabled)
require.Equal(t, "base text", eff.ExtractConfig.Text)
require.Equal(t, []string{"override-tag"}, eff.ExtractConfig.Tags)
}
func TestValidateProcessOverrides_NilOverrides(t *testing.T) {
t.Parallel()
err := ValidateProcessOverrides(context.Background(), &types.KnowledgeBase{}, nil, []string{"png"})
require.NoError(t, err)
}
func TestValidateProcessOverrides_ImageRequiresVLM(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
VLMConfig: types.VLMConfig{Enabled: false},
}
err := ValidateProcessOverrides(context.Background(), kb, &types.KnowledgeProcessOverrides{}, []string{"png"})
require.Error(t, err)
var badReq *werrors.AppError
require.ErrorAs(t, err, &badReq)
}
func TestValidateProcessOverrides_ImageWithEffectiveVLM(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
VLMConfig: types.VLMConfig{Enabled: false},
}
overrides := &types.KnowledgeProcessOverrides{
VLMConfig: &types.VLMConfig{Enabled: true, ModelID: "vlm-1"},
}
err := ValidateProcessOverrides(context.Background(), kb, overrides, []string{"jpg"})
require.NoError(t, err)
}
func TestValidateProcessOverrides_AudioRequiresASR(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
ASRConfig: types.ASRConfig{Enabled: false},
}
err := ValidateProcessOverrides(context.Background(), kb, &types.KnowledgeProcessOverrides{}, []string{"mp3"})
require.Error(t, err)
}
func TestValidateProcessOverrides_AudioWithEffectiveASR(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{
ASRConfig: types.ASRConfig{Enabled: false},
}
overrides := &types.KnowledgeProcessOverrides{
ASRConfig: &types.ASRConfig{Enabled: true, ModelID: "asr-1"},
}
err := ValidateProcessOverrides(context.Background(), kb, overrides, []string{"wav"})
require.NoError(t, err)
}
func TestValidateProcessOverrides_NonMediaFileTypes(t *testing.T) {
t.Parallel()
kb := &types.KnowledgeBase{}
err := ValidateProcessOverrides(context.Background(), kb, &types.KnowledgeProcessOverrides{}, []string{"pdf", "txt"})
require.NoError(t, err)
}
func TestValidateProcessOverrides_COSIncompleteForImage(t *testing.T) {
t.Parallel()
ctx := context.WithValue(context.Background(), types.TenantInfoContextKey, &types.Tenant{
StorageEngineConfig: &types.StorageEngineConfig{
COS: &types.COSEngineConfig{SecretID: "id"},
},
})
kb := &types.KnowledgeBase{
VLMConfig: types.VLMConfig{Enabled: true, ModelID: "vlm-1"},
}
kb.SetStorageProvider("cos")
err := ValidateProcessOverrides(ctx, kb, &types.KnowledgeProcessOverrides{}, []string{"png"})
require.Error(t, err)
}
func TestMergeParserEngineOverrides(t *testing.T) {
t.Parallel()
// 1. Both nil/empty
merged := MergeParserEngineOverrides(nil, nil)
require.Empty(t, merged)
// 2. Tenant only
merged = MergeParserEngineOverrides(map[string]string{"k1": "v1"}, nil)
require.Equal(t, map[string]string{"k1": "v1"}, merged)
// 3. Upload only
merged = MergeParserEngineOverrides(nil, map[string]string{"k2": "v2"})
require.Equal(t, map[string]string{"k2": "v2"}, merged)
// 4. Overlap priority (upload override should take priority over tenant config)
tenant := map[string]string{"k1": "tenant_val", "k2": "v2"}
upload := map[string]string{"k1": "upload_val", "k3": "v3"}
merged = MergeParserEngineOverrides(tenant, upload)
require.Equal(t, map[string]string{
"k1": "upload_val",
"k2": "v2",
"k3": "v3",
}, merged)
}