feat: 主 Agent 对话支持文件上传(方式1:落盘 + Read/Bash 读取)

- 前置:task 工作目录 <workDir>/<taskID> → <workDir>/tasks/<taskID>(改中心
  ensureRunDir,与 sessions/<id> 对称;三 agent 全走它)。
- 上传接口(方案B,一个/两 scope):POST /api/chat/upload?scope=task|session&id=X,
  落到 tasks|sessions/<id>/uploads/,路径安全化 + 同名加序号,返回 {name,path,size}。
- 后端注入(task main agent):chat handler 收 attachments;user 活动 Detail 存
  {text,attachments} JSON 供消息渲染;发给 agent 的清单用【绝对路径】(baseDir+相对,
  与落盘/CWD 完全对齐)保证一定 Read 得到。
- 前端(会话与任务共用 transcript 渲染):
  · api.chatUpload + chat(attachments) + ChatAttachment 类型;
  · transcript UserRow 解析 Detail JSON,气泡上方渲染附件卡片(乐观回显内联 detail);
  · sessions-tab 输入区加回形针上传按钮 + 发送前 chip 预览(可移除)+ 带附件发送。
- 说明:文本/代码/二进制经 Read/Bash 处理;图片视觉不支持(norma 无 image block)。
  会话(ChatAgent)面暂只共用了渲染,后端注入 + Composer 上传待后续补。
This commit is contained in:
Autumn-27 committed 2026-08-17 11:10:03 -04:00
1 parent 6348ec9d21
commit 071414063e
9 files changed
+337 -39

No files matched your search

+143
View File
@@ -0,0 +1,143 @@
package server
import (
"encoding/json"
"fmt"
"net/http"
"os"
"path/filepath"
"regexp"
"strings"
"github.com/Autumn-27/artex/db"
)
// maxChatUpload caps a single chat-attachment upload request (memory + spill).
const maxChatUpload = 128 << 20 // 128 MiB
// safeChatID guards the {id} path segment against traversal — task ids are numeric,
// session ids are alnum/_/- ; anything with "/" or ".." is rejected.
var safeChatID = regexp.MustCompile(`^[A-Za-z0-9_-]+$`)
// chatAttachment is one uploaded file as the frontend + agent see it. Path is relative
// to the chat's working dir (e.g. "uploads/report.txt"), which is the agent's CWD, so
// it can Read/Bash the file directly; Name/Size drive the UI card.
type chatAttachment struct {
Name string `json:"name"`
Path string `json:"path"`
Size int64 `json:"size"`
}
// chatUpload implements method-1 file support: it saves one or more files into a chat's
// working dir under uploads/, so the agent opens them with its existing Read/Bash tools
// and the sent message carries their paths. No LLM-layer change, no multimodal.
//
// POST /api/chat/upload?scope=task|session&id=<id>, multipart field "file" (repeatable).
// Returns {attachments:[{name,path,size}]}. Target dir mirrors the agent CWD layout:
//
// scope=task → <workDir>/tasks/<id>/uploads/
// scope=session → <workDir>/sessions/<id>/uploads/
func (s *Server) chatUpload(w http.ResponseWriter, r *http.Request) {
var sub string
switch r.URL.Query().Get("scope") {
case "task":
sub = "tasks"
case "session":
sub = "sessions"
default:
writeErr(w, 400, "scope 必须是 task 或 session")
return
}
id := r.URL.Query().Get("id")
if !safeChatID.MatchString(id) {
writeErr(w, 400, "非法 id")
return
}
dir := filepath.Join(s.m.dir, sub, id, "uploads")
if err := os.MkdirAll(dir, 0o755); err != nil {
writeErr(w, 500, "建目录失败: "+err.Error())
return
}
r.Body = http.MaxBytesReader(w, r.Body, maxChatUpload)
if err := r.ParseMultipartForm(32 << 20); err != nil {
writeErr(w, 400, "解析上传失败或超出大小限制: "+err.Error())
return
}
files := r.MultipartForm.File["file"]
if len(files) == 0 {
writeErr(w, 400, "缺少上传文件(表单字段 file)")
return
}
out := make([]chatAttachment, 0, len(files))
for _, hdr := range files {
name := filepath.Base(hdr.Filename) // strip any path component
if name == "" || name == "." || name == ".." || strings.ContainsAny(name, `/\`) {
continue
}
dest := uniqueUploadPath(dir, name)
if err := saveUpload(hdr, dest); err != nil {
writeErr(w, 500, "保存失败: "+err.Error())
return
}
base := filepath.Base(dest)
out = append(out, chatAttachment{Name: base, Path: "uploads/" + base, Size: hdr.Size})
}
writeJSON(w, 200, map[string]any{"attachments": out})
}
// uniqueUploadPath returns dir/name, or dir/name-1, dir/name-2… when it already exists,
// so re-uploading the same filename never clobbers a prior attachment.
func uniqueUploadPath(dir, name string) string {
dest := filepath.Join(dir, name)
if _, err := os.Stat(dest); os.IsNotExist(err) {
return dest
}
ext := filepath.Ext(name)
stem := strings.TrimSuffix(name, ext)
for i := 1; ; i++ {
cand := filepath.Join(dir, fmt.Sprintf("%s-%d%s", stem, i, ext))
if _, err := os.Stat(cand); os.IsNotExist(err) {
return cand
}
}
}
// composeAgentMessage appends an attachment manifest to the user's message so the agent
// knows which files were uploaded and where to Read them. baseDir is the agent's working
// dir (its CWD); we emit ABSOLUTE paths (baseDir + relative) so the agent can Read/Bash
// them unambiguously regardless of how it interprets relative paths.
func composeAgentMessage(msg string, atts []chatAttachment, baseDir string) string {
if len(atts) == 0 {
return msg
}
var b strings.Builder
b.WriteString(msg)
b.WriteString("\n\n【用户上传的附件】(绝对路径,需要时用 Read/Bash 查看):")
for _, a := range atts {
fmt.Fprintf(&b, "\n- %s(%s)", filepath.Join(baseDir, a.Path), humanBytes(a.Size))
}
return b.String()
}
// userActivityWithAttachments builds the persisted 'user' activity. With attachments,
// Detail holds JSON {text, attachments} so the transcript renders text + attachment
// cards; Summary stays the plain text (the list payload omits Detail, lazy-loaded).
func userActivityWithAttachments(worker, text string, atts []chatAttachment) db.Activity {
a := db.Activity{Worker: worker, Kind: "user", Summary: text}
if len(atts) > 0 {
blob, _ := json.Marshal(map[string]any{"text": text, "attachments": atts})
a.Detail = string(blob)
}
return a
}
func humanBytes(n int64) string {
switch {
case n >= 1<<20:
return fmt.Sprintf("%.1f MB", float64(n)/(1<<20))
case n >= 1<<10:
return fmt.Sprintf("%.1f KB", float64(n)/(1<<10))
default:
return fmt.Sprintf("%d B", n)
}
}
+12 -4
View File
@@ -628,6 +628,7 @@ func (s *Server) Handler() http.Handler {
mux.HandleFunc("POST /api/settings/web-search/test", s.testWebSearch)
mux.HandleFunc("GET /api/report", s.getReport)
mux.HandleFunc("POST /api/chat", s.chat)
mux.HandleFunc("POST /api/chat/upload", s.chatUpload) // 方式1 文件上传:落到会话/任务工作目录 uploads/
mux.HandleFunc("POST /api/tasks/{id}/chat/stop", s.stopChat)
// --- 管理后台 API (PostgreSQL 数据源; 新版数据库与管理后台方案) ---
@@ -2057,7 +2058,8 @@ func (s *Server) chat(w http.ResponseWriter, r *http.Request) {
return
}
var req struct {
Message string `json:"message"`
Message string `json:"message"`
Attachments []chatAttachment `json:"attachments,omitempty"` // 方式1 上传的文件(路径相对任务工作目录)
}
if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
writeErr(w, 400, err.Error())
@@ -2065,8 +2067,10 @@ func (s *Server) chat(w http.ResponseWriter, r *http.Request) {
}
// Persist + broadcast the human turn so the 主 Agent 编排会话 survives page
// reloads and updates live: the conversation lives in the activity stream as
// worker="mainagent" (the per-task activity table, replayed via SSE).
s.engine.emitActivity(t, db.Activity{Worker: "mainagent", Kind: "user", Summary: req.Message})
// worker="mainagent" (the per-task activity table, replayed via SSE). With
// attachments, the activity's Detail carries {text, attachments} so the transcript
// renders attachment cards.
s.engine.emitActivity(t, userActivityWithAttachments("mainagent", req.Message, req.Attachments))
if ma := s.mainAgentRef(); ma != nil {
// Serialize per task: one main-agent run at a time so concurrent messages
// don't corrupt the shared exp<id>-main transcript. If a prior turn is still
@@ -2103,7 +2107,11 @@ func (s *Server) chat(w http.ResponseWriter, r *http.Request) {
emit := func(rec db.Activity) { s.engine.emitActivity(t, rec) }
maTaskID, _ := strconv.ParseInt(t.ID, 10, 64)
resume := func() { s.reviveTask(t) } // set_goals 新增目标 → 把任务拉回 running
if _, err := ma.Chat(ctx, maTaskID, s.m.Assets(), t.Store, t.Goal, req.Message, emit, t.Notify, resume, t.NotifyGoal); err != nil && ctx.Err() == nil {
// 把上传附件的【绝对路径】清单拼进发给 agent 的消息,它据此用 Read/Bash 打开文件。
// taskDir = agent 的工作目录(CWD),与 chatUpload 落盘、ensureRunDir 一致。
taskDir := filepath.Join(s.m.dir, "tasks", t.ID)
agentMsg := composeAgentMessage(req.Message, req.Attachments, taskDir)
if _, err := ma.Chat(ctx, maTaskID, s.m.Assets(), t.Store, t.Goal, agentMsg, emit, t.Notify, resume, t.NotifyGoal); err != nil && ctx.Err() == nil {
s.engine.emitActivity(t, db.Activity{Worker: "mainagent", Kind: "text", IsError: true, Summary: "(主 Agent 出错:" + err.Error() + ")"})
}
}()