logan
4c68539db6
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。
112 lines
2.5 KiB
Go
112 lines
2.5 KiB
Go
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
|
|
}
|