Files
agentic-coding-workflow/internal/infra/llm/tokenizer_test.go
T

28 lines
904 B
Go

package llm
import (
"testing"
"github.com/stretchr/testify/require"
)
func TestEstimateTokens_AsciiAndCJK(t *testing.T) {
require.Equal(t, 0, EstimateTokens(""))
require.InDelta(t, 1, EstimateTokens("hi"), 1) // 2 chars / 4 ≈ 0~1
require.InDelta(t, 25, EstimateTokens(string(make([]byte, 100))), 1) // 100 chars / 4 = 25
// CJK 按 rune 数计算
require.InDelta(t, 2, EstimateTokens("你好"), 1) // 2 runes / 4 ≈ 0~1,向上取整 1
}
func TestEstimateRequestTokens_SumsAllBlocks(t *testing.T) {
req := Request{
SystemPrompt: "You are helpful.", // 16 chars / 4 = 4
Messages: []Message{
{Role: RoleUser, Blocks: []ContentBlock{{Type: "text", Text: "hello world"}}}, // 11 chars / 4 = 3
{Role: RoleAssistant, Blocks: []ContentBlock{{Type: "text", Text: "hi"}}}, // 2 chars / 4 = 1
},
}
got := EstimateRequestTokens(req)
require.InDelta(t, 8, got, 2)
}