package internal import ( "strings" "testing" "unicode/utf8" ) const truncSuffix = "\n\n... (文件过大,截断显示)" // TestTruncateSkillContentRuneBoundary 验证截断点落在多字节序列中间时 // 逐字节回退到合法 rune 边界,结果不含残缺字节 func TestTruncateSkillContentRuneBoundary(t *testing.T) { limit := 100 * 1024 // 中文各 3 字节:分别构造截断点落在中文字符第 1、2 字节后的场景 for _, cutInChar := range []int{1, 2} { prefix := strings.Repeat("a", limit-cutInChar) content := prefix + "中文内容超出限制" got := truncateSkillContent(content, limit) body := strings.TrimSuffix(got, truncSuffix) if !utf8.ValidString(body) { t.Fatalf("截断点落在字符第 %d 字节后, 结果应为合法 UTF-8", cutInChar) } if !strings.HasPrefix(body, prefix) { t.Fatalf("截断点落在字符第 %d 字节后, 不应吞掉 ASCII 前缀", cutInChar) } if len(body) > limit { t.Fatalf("截断点落在字符第 %d 字节后, body 长度 %d 超限 %d", cutInChar, len(body), limit) } if !strings.HasSuffix(got, truncSuffix) { t.Fatalf("截断后应带省略提示后缀, got %q", got[len(got)-40:]) } } } // TestTruncateSkillContentEdge 验证边界:内容不超限原样返回,恰在边界截断不误删 func TestTruncateSkillContentEdge(t *testing.T) { limit := 100 * 1024 short := "短内容" if got := truncateSkillContent(short, limit); got != short { t.Fatalf("未超限内容应原样返回, got %q", got) } // 恰好 limit 字节后跟多字节字符:截断点在边界上,不应回退删掉合法内容 exact := strings.Repeat("a", limit) got := truncateSkillContent(exact+"中文", limit) body := strings.TrimSuffix(got, truncSuffix) if body != exact { t.Fatalf("边界截断不应删减合法内容, body 长度 = %d, 期望 %d", len(body), limit) } }