Files
ARTEX-chixuweihu/server/credentials.go

461 lines
14 KiB
Go

package server
import (
"context"
"io"
"log"
"net/http"
"net/url"
"regexp"
"sort"
"strconv"
"strings"
"time"
"github.com/Autumn-27/artex/agent"
"github.com/Autumn-27/artex/db"
"github.com/Autumn-27/artex/llmrec"
"github.com/Autumn-27/artex/traffic"
"github.com/Autumn-27/norma/transcript"
)
func (s *Server) seedCredentialProcessor() {
const flag = "credential_processor_seed_v1"
if s.m.pg.GetBool(flag, false) {
return
}
a, err := s.m.pg.GetAgentByKey(db.CredentialAgentKey)
if err != nil {
return
}
if a == nil {
a, err = s.m.pg.CreateAgent(db.CredentialAgentKey, "凭据整理", "Agent 开关控制的凭据整理:拆分账号、密码、Token、Cookie,仅分析已上报内容,不主动登录。可独立绑定模型;每条最多一次模型调用。")
if err != nil {
return
}
if err = s.m.pg.SeedPromptIfEmpty(a.ID, agent.CredentialProcessorPrompt); err != nil {
return
}
_ = s.m.pg.SetAgentMaxTurns(db.CredentialAgentKey, 1)
}
_ = s.m.pg.SetBool(flag, true)
}
func (s *Server) runCredentialProcessor() {
ticker := time.NewTicker(3 * time.Second)
defer ticker.Stop()
for {
select {
case <-s.ctx.Done():
return
case <-ticker.C:
}
if !s.m.pg.GetBool(db.SettingTaskIntelligence, false) || !s.m.pg.GetBool(db.SettingCredentialAgent, false) {
continue
}
a, err := s.m.pg.GetAgentByKey(db.CredentialAgentKey)
if err != nil || a == nil || !a.Enabled {
continue
}
item, err := s.m.pg.ClaimCredentialJob(s.ctx, s.m.pg.GetBool(db.SettingCredentialTraffic, false), 0)
if err != nil {
if s.ctx.Err() == nil {
log.Printf("[credentials] queue: %v", err)
}
continue
}
if item == nil {
continue
}
s.processCredentialJob(s.ctx, item)
}
}
func (s *Server) processCredentialJob(parent context.Context, item *db.Intelligence) {
if !s.m.pg.GetBool(db.SettingTaskIntelligence, false) || !s.m.pg.GetBool(db.SettingCredentialAgent, false) {
_ = s.m.pg.FinishCredentialJob(parent, item, *item.Credentials)
return
}
ctx, cancel := context.WithTimeout(parent, 90*time.Second)
defer cancel()
// Turning the switch off cancels even an in-flight model call.
go func() {
tick := time.NewTicker(500 * time.Millisecond)
defer tick.Stop()
for {
select {
case <-ctx.Done():
return
case <-tick.C:
if !s.m.pg.GetBool(db.SettingTaskIntelligence, false) || !s.m.pg.GetBool(db.SettingCredentialAgent, false) {
cancel()
return
}
}
}
}()
result := *item.Credentials
needsExtract := result.ForceExtract || result.SourceHash != db.CredentialSourceHash(item.Content) || result.Method == "" || (result.Error != "" && len(result.Accounts) == 0)
result.State = "ready"
result.ForceExtract = false
result.Error = ""
result.TrafficNote = ""
sourceHash := db.CredentialSourceHash(item.Content)
if needsExtract {
result.Accounts = nil
if result.SourceHash != sourceHash {
result.Traffic = nil
}
result.InputTokens = 0
result.OutputTokens = 0
result.Model = ""
result.Note = ""
accounts, parsed := agent.ParseCredentialJSON(item.Content)
if parsed {
result.Method = "rules"
result.Accounts = accounts
} else {
result.Method = "agent"
prov, cfg, ok := s.agentBindingProvider(db.CredentialAgentKey)
if !ok {
prov, cfg, ok = s.globalProvider()
}
if !ok {
result.State = "error"
result.Error = "未配置可用模型,请在 Agent 配置中为凭据整理选择模型,或激活全局模型后重试"
} else {
prompt := agent.CredentialProcessorPrompt
if a, err := s.m.pg.GetAgentByKey(db.CredentialAgentKey); err == nil && a != nil {
if p, err := s.m.pg.CurrentPrompt(a.ID); err == nil && strings.TrimSpace(p) != "" {
prompt = p
}
}
callCtx := llmrec.WithTaskID(ctx, strconv.FormatInt(item.TaskID, 10))
callCtx = llmrec.WithWorker(callCtx, db.CredentialAgentKey)
callCtx = transcript.WithSessionID(callCtx, "credential-"+strconv.FormatInt(item.ID, 10)+"-"+sourceHash[:12])
accounts, usage, err := agent.ExtractCredentials(callCtx, prov, prompt, item.Content, cfg.MaxTokens)
result.Model = cfg.Model
result.InputTokens = usage.InputTokens
result.OutputTokens = usage.OutputTokens
if err != nil {
result.State = "error"
result.Error = err.Error()
} else {
result.Accounts = accounts
}
}
}
result.SourceHash = sourceHash
}
if result.State == "ready" && len(result.Accounts) == 100 {
result.Note = "单条最多整理 100 组凭据,完整信息仍可在原始上报中查看"
}
if result.State == "ready" && len(result.Accounts) == 0 {
result.Note = "未识别到可独立展示的凭据,原文已保留"
}
if result.State == "ready" && s.m.pg.GetBool(db.SettingCredentialTraffic, false) && ctx.Err() == nil {
matched, state, note := s.matchCredentialTraffic(ctx, item, result.Accounts)
// Binding is additive: a later retry must not discard previously saved
// evidence just because disposable traffic has since been purged.
seen := map[string]bool{}
merged := []db.CredentialTraffic{}
for _, ref := range append(append([]db.CredentialTraffic{}, result.Traffic...), matched...) {
if !seen[ref.ID] && len(merged) < 12 {
seen[ref.ID] = true
merged = append(merged, ref)
}
}
result.Traffic = merged
result.TrafficState = state
result.TrafficNote = note
if len(merged) > 0 {
result.TrafficState = "ready"
if len(matched) == 0 {
result.TrafficNote = "既有流量引用与匹配片段已保留;原始流量可能已删除或暂不可用"
}
}
} else if !s.m.pg.GetBool(db.SettingCredentialTraffic, false) {
result.TrafficState = "off"
}
if ctx.Err() != nil && result.State == "ready" {
result.State = "error"
result.Error = "整理已中断,原始内容保留,可手动重试"
}
// Use a fresh bounded persistence context so cancellation can release the lease.
finish, stop := context.WithTimeout(context.Background(), 5*time.Second)
defer stop()
if err := s.m.pg.FinishCredentialJob(finish, item, result); err != nil {
log.Printf("[credentials] save record #%d: %v", item.ID, err)
}
}
var credentialURLs = regexp.MustCompile(`https?://[^\s<>"',;))]+`)
var credentialTrafficIDs = regexp.MustCompile(`\b[0-9]{8,}-[0-9]{3,}\b`)
func (s *Server) credentialHosts(item *db.Intelligence) map[string]bool {
hosts := map[string]bool{}
add := func(raw string, children bool) {
raw = strings.TrimSpace(raw)
if raw == "" {
return
}
if !strings.Contains(raw, "://") {
raw = "http://" + raw
}
u, err := url.Parse(raw)
if err == nil && u.Hostname() != "" {
host := strings.ToLower(u.Hostname())
hosts[host] = hosts[host] || children
}
}
if assets, err := s.m.Assets().GetByIDs(item.AssetIDs); err == nil {
for _, a := range assets {
switch a.Type {
case "root_domain":
add(a.Domain, true)
case "service", "endpoint":
add(a.URL, false)
add(a.Domain, false)
add(a.IP, false)
case "subdomain":
add(a.Domain, false)
case "ip":
add(a.IP, false)
}
}
}
for _, raw := range credentialURLs.FindAllString(item.Content+"\n"+item.Evidence, 16) {
add(raw, false)
}
return hosts
}
func matchesCredentialHost(raw string, hosts map[string]bool) bool {
u, err := url.Parse(raw)
if err != nil {
return false
}
host := strings.ToLower(u.Hostname())
for known, children := range hosts {
if host == known || children && strings.HasSuffix(host, "."+known) {
return true
}
}
return false
}
func credentialMatches(raw string, a db.CredentialAccount) bool {
source := agent.CredentialSourceText(raw)
// Tokens and cookies can be returned without a username in the same response.
for _, secret := range []string{a.Token, a.Cookie} {
if len(secret) >= 8 && strings.Contains(source, secret) {
return true
}
}
if len(a.Password) < 6 || !strings.Contains(source, a.Password) {
return false
}
if a.Username != "" {
return strings.Contains(source, a.Username)
}
return len(a.Password) >= 8
}
func credentialExcerpt(raw string, accounts []db.CredentialAccount) string {
index := 0
found := false
for _, a := range accounts {
for _, secret := range []string{a.Token, a.Cookie, a.Password} {
if secret != "" {
if at := strings.Index(raw, secret); at >= 0 {
index = at
found = true
break
}
}
}
if found {
break
}
}
runes := []rune(raw[:index])
start := len(runes) - 100
if start < 0 {
start = 0
}
all := []rune(raw)
end := start + 600
if end > len(all) {
end = len(all)
}
return string(all[start:end])
}
func (s *Server) matchCredentialTraffic(ctx context.Context, item *db.Intelligence, accounts []db.CredentialAccount) ([]db.CredentialTraffic, string, string) {
bound := []db.CredentialTraffic{}
if s.m.traffic == nil {
return bound, "unavailable", "未启用流量采集或没有可用的流量存储"
}
hosts := s.credentialHosts(item)
if len(hosts) == 0 {
return bound, "unavailable", "没有可核对的关联资产或 URL,未猜测流量归属"
}
seen := map[string]bool{}
inspected, queries := 0, 0
inspect := func(id string) {
if seen[id] || inspected >= 32 || len(bound) >= 12 || ctx.Err() != nil {
return
}
seen[id] = true
inspected++
// Reads only captured data. No network requests or authentication attempts.
_ = s.m.traffic.ReadEvidence(ctx, []string{id}, func(e traffic.EvidenceExchange) error {
if !matchesCredentialHost(e.URL, hosts) {
return nil
}
req, err := io.ReadAll(io.LimitReader(e.Request, 128<<10))
if err != nil {
return err
}
resp, err := io.ReadAll(io.LimitReader(e.Response, 128<<10))
if err != nil {
return err
}
head := func(s string) string {
if len(s) > 8192 {
return s[:8192]
}
return s
}
for _, side := range []struct{ name, raw string }{{"request", head(e.ReqHead) + "\n" + string(req)}, {"response", head(e.RespHead) + "\n" + string(resp)}} {
matches := []int{}
for i, a := range accounts {
if credentialMatches(side.raw, a) {
matches = append(matches, i)
}
}
if len(matches) > 0 {
bound = append(bound, db.CredentialTraffic{ID: e.ID, URL: e.URL, Method: e.Method, Status: e.Status, Side: side.name, Excerpt: credentialExcerpt(side.raw, accounts), Accounts: matches})
break
}
}
return nil
})
}
for _, id := range credentialTrafficIDs.FindAllString(item.Evidence, 8) {
inspect(id)
}
names := make([]string, 0, len(hosts))
for host := range hosts {
names = append(names, host)
}
sort.Strings(names)
if len(names) > 8 {
names = names[:8]
}
for _, a := range accounts {
for _, secret := range []string{a.Token, a.Cookie, a.Password} {
if len(secret) < 6 {
continue
}
for _, host := range names {
if queries >= 24 || len(bound) >= 12 || ctx.Err() != nil {
break
}
queries++
rows, _, err := s.m.traffic.Page(traffic.PageQuery{Host: host, Body: secret, RespMin: -1, RespMax: -1}, 0, 4)
if err != nil {
continue
}
for _, row := range rows {
inspect(row.ID)
}
}
}
}
if ctx.Err() != nil {
return bound, "error", "流量核对已中断,可手动重试"
}
if len(bound) == 0 {
return bound, "none", "未找到包含匹配凭据的已抓取流量;不会据此改变凭据验证状态"
}
return bound, "ready", "已核对资产域名及凭据原文;保存流量引用与片段,原始流量删除后仍可查看片段"
}
func (s *Server) retryCredentials(w http.ResponseWriter, r *http.Request) {
if !s.m.pg.GetBool(db.SettingTaskIntelligence, false) || !s.m.pg.GetBool(db.SettingCredentialAgent, false) {
writeErr(w, 409, "请先开启 Agent 开关")
return
}
taskID, ok := pathInt(r, "id")
if !ok {
return
}
id, ok := pathInt(r, "intelID")
if !ok {
return
}
item, err := s.m.pg.GetIntelligence(r.Context(), id)
if err != nil || item.TaskID != taskID || item.Kind != "credential" {
intelligenceError(w, db.ErrIntelligenceNotFound)
return
}
if item.Credentials != nil && item.Credentials.State == "processing" {
writeErr(w, 409, "该凭据正在整理,请稍后再试")
return
}
if err = s.m.pg.RetryCredentialJob(r.Context(), taskID, id); err != nil {
intelligenceError(w, err)
return
}
writeJSON(w, 202, map[string]bool{"queued": true})
}
// Scope checking reuses the intelligence API before exposing a live flow. A
// bound excerpt remains visible even if the original exchange has been purged.
func (s *Server) credentialTrafficDetail(w http.ResponseWriter, r *http.Request) {
f, ok := s.intelligenceFilter(w, r)
if !ok {
return
}
id, ok := pathInt(r, "intelID")
if !ok {
return
}
item, err := s.m.pg.GetIntelligence(r.Context(), id)
if err != nil {
intelligenceError(w, err)
return
}
allowed := false
if f.TaskID > 0 {
err = s.m.pg.QueryRowContext(r.Context(), `SELECT $1::bigint=$2::bigint OR EXISTS(SELECT 1 FROM task_relations WHERE task_id=$1 AND source_task_id=$2)`, f.TaskID, item.TaskID).Scan(&allowed)
} else {
for _, a := range item.AssetIDs {
if a == f.AssetID {
allowed = true
}
}
}
if err != nil || !allowed || item.Credentials == nil {
intelligenceError(w, db.ErrIntelligenceNotFound)
return
}
var ref *db.CredentialTraffic
for i := range item.Credentials.Traffic {
if item.Credentials.Traffic[i].ID == r.PathValue("trafficID") {
ref = &item.Credentials.Traffic[i]
break
}
}
if ref == nil {
intelligenceError(w, db.ErrIntelligenceNotFound)
return
}
w.Header().Set("Cache-Control", "no-store")
if s.m.traffic == nil {
writeJSON(w, 200, map[string]any{"reference": ref, "unavailable": true})
return
}
req, resp, err := s.m.traffic.Get(ref.ID)
if err != nil {
writeJSON(w, 200, map[string]any{"reference": ref, "unavailable": true})
return
}
writeJSON(w, 200, map[string]any{"reference": ref, "req": req, "resp": resp, "unavailable": false})
}