fix(ai): Ask P0 禁 CoT、可澄清、禁 UPDATE_TIME 终答
prompt 禁止推理外露与编造列；信息不足允许一句澄清。 新增 SanitizeAssistantContent 剥回复中的推理草稿；业务「最近 N 天」不得默认用 information_schema.UPDATE_TIME。
This commit is contained in:
logan committed
1 parent
dc7089857b
commit
4c68539db6
5 files changed
+154
-3
No files matched your search
@@ -118,6 +118,12 @@ func TestAskChatsWithoutClaimingDatabaseExecution(t *testing.T) {
|
||||
if !strings.Contains(generate[0].Content, "生成一条 mysql 方言的 SQL") {
|
||||
t.Fatalf("SQL generation lost its dedicated instruction: %s", generate[0].Content)
|
||||
}
|
||||
if !strings.Contains(generate[0].Content, "澄清") || strings.Contains(generate[0].Content, "只输出可执行 SQL") {
|
||||
t.Fatalf("generate must allow clarification and not force executable-only: %s", generate[0].Content)
|
||||
}
|
||||
if !strings.Contains(generate[0].Content, "禁止编造") || !strings.Contains(generate[0].Content, "UPDATE_TIME") {
|
||||
t.Fatalf("system rules missing invent/UPDATE_TIME guards: %s", generate[0].Content)
|
||||
}
|
||||
offline := chatStub(request)
|
||||
if offline.SQL != "" || !strings.Contains(offline.Content, "离线体验无法回答") {
|
||||
t.Fatalf("offline Ask must not pretend to generate or query: %+v", offline)
|
||||
|
||||
@@ -14,13 +14,13 @@ func BuildMessages(req ChatReq) []Message {
|
||||
if driver == "" {
|
||||
driver = "mysql"
|
||||
}
|
||||
sys := "你是资深数据库工程师助手,服务于数据库客户端 OpsxDB。回答使用中文,SQL 代码放在 ```sql 代码块中。"
|
||||
sys := "你是资深数据库工程师助手,服务于数据库客户端 OpsxDB。回答使用中文。SQL 放在 ```sql 代码块中。\n规则：\n- 禁止输出思考过程、推理草稿、内心独白或逐步分析；只给用户最终答案。\n- 禁止编造上下文中未出现的列名或表结构。\n- 缺表/缺时间列/意图不清时：用一两句中文澄清（可问具体表与时间字段），不要为了「可执行」硬造 SQL。\n- 「最近 N 天数据变化」默认指业务行时间过滤；不得默认用 information_schema.TABLES.UPDATE_TIME 充当终答（除非用户明确要表元数据更新时间）。"
|
||||
var task string
|
||||
switch strings.ToLower(req.Action) {
|
||||
case "generate":
|
||||
task = fmt.Sprintf("根据用户描述生成一条 %s 方言的 SQL。只输出可执行 SQL 与极简说明。", driver)
|
||||
task = fmt.Sprintf("根据用户描述生成一条 %s 方言的 SQL。结构足够时：极简说明（≤2 行）+ 一条可执行 SQL；结构不足或缺时间列时：只输出一两句澄清问题，不要编造列名，不要输出思考过程。", driver)
|
||||
case "ask":
|
||||
task = "回答用户关于当前数据库、数据或 SQL 的问题；只有用户明确要求编写 SQL 时才给出 SQL。你没有调用数据库工具，不要声称已经查询或修改数据。"
|
||||
task = "回答用户关于当前数据库、数据或 SQL 的问题；只有用户明确要求编写 SQL 时才给出 SQL（极简说明≤2 行 + SQL 块）。结构不足时一句澄清即可。你没有调用数据库工具，不要声称已经查询或修改数据。禁止输出思考过程。"
|
||||
case "explain":
|
||||
task = "逐部分解释用户给出的 SQL:访问路径、索引使用、潜在代价。"
|
||||
case "optimize":
|
||||
|
||||
@@ -191,6 +191,7 @@ func (p *RemoteProvider) Complete(ctx context.Context, req ChatReq) (ChatRes, er
|
||||
if err != nil {
|
||||
return ChatRes{}, err
|
||||
}
|
||||
content = SanitizeAssistantContent(content)
|
||||
return ChatRes{Content: content, SQL: extractSQL(content)}, nil
|
||||
}
|
||||
|
||||
|
||||
@@ -0,0 +1,111 @@
|
||||
package ai
|
||||
|
||||
import (
|
||||
"regexp"
|
||||
"strings"
|
||||
"unicode/utf8"
|
||||
)
|
||||
|
||||
var (
|
||||
anyFence = regexp.MustCompile("(?s)```[a-zA-Z]*\\s*\n.*?```")
|
||||
cotCue = regexp.MustCompile(`(?i)(我们需要|作为助手|指令要求|让我们看看|最终答案|chain of thought|step by step|I need to|as an assistant|the instruction)`)
|
||||
sqlFenceFind = regexp.MustCompile("(?s)```sql\\s*\n(.+?)```")
|
||||
)
|
||||
|
||||
// SanitizeAssistantContent 去掉 Ask/Generate 回复里的推理草稿，只保留短说明 + SQL 块（或短澄清）。
|
||||
func SanitizeAssistantContent(content string) string {
|
||||
s := strings.TrimSpace(content)
|
||||
if s == "" {
|
||||
return s
|
||||
}
|
||||
fences := anyFence.FindAllStringIndex(s, -1)
|
||||
if len(fences) == 0 {
|
||||
if utf8.RuneCountInString(s) <= 240 && !cotCue.MatchString(s) {
|
||||
return s
|
||||
}
|
||||
return keepLeadingLines(s, 2)
|
||||
}
|
||||
var b strings.Builder
|
||||
prev := 0
|
||||
for i, loc := range fences {
|
||||
prose := strings.TrimSpace(s[prev:loc[0]])
|
||||
if prose != "" {
|
||||
if i == 0 {
|
||||
prose = trimProse(prose)
|
||||
} else {
|
||||
prose = keepLeadingLines(prose, 2)
|
||||
}
|
||||
if prose != "" {
|
||||
if b.Len() > 0 {
|
||||
b.WriteByte('\n')
|
||||
}
|
||||
b.WriteString(prose)
|
||||
}
|
||||
}
|
||||
if b.Len() > 0 {
|
||||
b.WriteString("\n\n")
|
||||
}
|
||||
b.WriteString(strings.TrimSpace(s[loc[0]:loc[1]]))
|
||||
prev = loc[1]
|
||||
}
|
||||
tail := strings.TrimSpace(s[prev:])
|
||||
if tail != "" {
|
||||
tail = keepLeadingLines(tail, 2)
|
||||
if tail != "" {
|
||||
b.WriteString("\n\n")
|
||||
b.WriteString(tail)
|
||||
}
|
||||
}
|
||||
out := strings.TrimSpace(b.String())
|
||||
if out == "" {
|
||||
if m := sqlFenceFind.FindString(s); m != "" {
|
||||
return strings.TrimSpace(m)
|
||||
}
|
||||
return keepLeadingLines(s, 2)
|
||||
}
|
||||
return out
|
||||
}
|
||||
|
||||
func trimProse(prose string) string {
|
||||
lines := nonEmptyLines(prose)
|
||||
if len(lines) == 0 {
|
||||
return ""
|
||||
}
|
||||
var notes []string
|
||||
for _, ln := range lines {
|
||||
if strings.HasPrefix(ln, "说明") || strings.HasPrefix(ln, "注意") || strings.HasPrefix(ln, "提示") {
|
||||
notes = append(notes, ln)
|
||||
}
|
||||
}
|
||||
if len(notes) > 0 {
|
||||
if len(notes) > 2 {
|
||||
notes = notes[len(notes)-2:]
|
||||
}
|
||||
return strings.Join(notes, "\n")
|
||||
}
|
||||
joined := strings.Join(lines, "\n")
|
||||
if cotCue.MatchString(joined) || utf8.RuneCountInString(joined) > 160 {
|
||||
return ""
|
||||
}
|
||||
return keepLeadingLines(joined, 2)
|
||||
}
|
||||
|
||||
func keepLeadingLines(s string, n int) string {
|
||||
lines := nonEmptyLines(s)
|
||||
if len(lines) > n {
|
||||
lines = lines[:n]
|
||||
}
|
||||
return strings.Join(lines, "\n")
|
||||
}
|
||||
|
||||
func nonEmptyLines(s string) []string {
|
||||
raw := strings.Split(s, "\n")
|
||||
out := make([]string, 0, len(raw))
|
||||
for _, ln := range raw {
|
||||
ln = strings.TrimSpace(ln)
|
||||
if ln != "" {
|
||||
out = append(out, ln)
|
||||
}
|
||||
}
|
||||
return out
|
||||
}
|
||||
@@ -0,0 +1,33 @@
|
||||
package ai
|
||||
|
||||
import (
|
||||
"strings"
|
||||
"testing"
|
||||
)
|
||||
|
||||
func TestSanitizeAssistantContentStripsCoTKeepsSQL(t *testing.T) {
|
||||
raw := `我们需要回答中文。用户说查询最近7天。作为助手我们只能根据现有信息。指令要求生成SQL。
|
||||
|
||||
说明：因未提供业务时间字段，以表更新时间近似。
|
||||
|
||||
` + "```sql\nSELECT TABLE_NAME, UPDATE_TIME\nFROM information_schema.TABLES\nWHERE TABLE_SCHEMA = 'demo';\n```"
|
||||
out := SanitizeAssistantContent(raw)
|
||||
if strings.Contains(out, "我们需要回答中文") || strings.Contains(out, "作为助手") {
|
||||
t.Fatalf("CoT leaked: %s", out)
|
||||
}
|
||||
if !strings.Contains(out, "```sql") || !strings.Contains(out, "UPDATE_TIME") {
|
||||
t.Fatalf("SQL fence lost: %s", out)
|
||||
}
|
||||
// prose should be short
|
||||
prose := strings.Split(out, "```")[0]
|
||||
if len([]rune(strings.TrimSpace(prose))) > 80 {
|
||||
t.Fatalf("prose too long: %q", prose)
|
||||
}
|
||||
}
|
||||
|
||||
func TestSanitizeAssistantContentKeepsShortClarify(t *testing.T) {
|
||||
in := "当前结构未给出时间列，请说明要查哪张表以及时间字段名。"
|
||||
if got := SanitizeAssistantContent(in); got != in {
|
||||
t.Fatalf("got %q", got)
|
||||
}
|
||||
}
|
||||
Reference in new issue
Block a user