Agent-System/internal/server/chat_document_reader.go

package server

import (
	"context"
	"fmt"
	"path/filepath"
	"strings"
	"time"
	"unicode"

	"chat-agent-system/internal/auth"
	"chat-agent-system/internal/bizdev"
	"chat-agent-system/internal/chat"
	"chat-agent-system/internal/coordinator"
	"chat-agent-system/internal/docgov"
	"chat-agent-system/internal/intent"
	"chat-agent-system/internal/projectinstance"
	"chat-agent-system/internal/textextract"
	"chat-agent-system/internal/storage"
	"chat-agent-system/internal/stream"
	"chat-agent-system/internal/workflowscope"
)

// bindSessionDocumentSummaryReader 将正文访问绑定到已认证的用户与项目。
// 读取前先应用与文档列表/正文 API 相同的阶段权限过滤。
func (s *Server) bindSessionDocumentSummaryReader(sess *chat.Session, user *auth.UserAccount) {
	if sess == nil || sess.UserCtx == nil {
		return
	}
	sess.UserCtx.DocumentSummaryReader = nil
	projectID := sess.Project()
	userID := int64(0)
	if user != nil {
		userID = user.ID
	}
	if projectID <= 0 || userID <= 0 || s.gov() == nil {
		return
	}

	sess.UserCtx.DocumentSummaryReader = func(ctx context.Context, target string) (*coordinator.ReadableDocument, error) {
		if s.authorizer() == nil {
			return nil, fmt.Errorf("project authorization unavailable")
		}
		if err := s.authorizer().RequireProjectAccessOrAdmin(ctx, projectID, userID, s.isSystemAdminUser(ctx, user)); err != nil {
			return nil, err
		}
		if strings.TrimSpace(target) == intent.TargetDocLatestUpload {
			return s.resolveLatestUploadDocument(ctx, projectID)
		}
		listed, err := s.gov().ListWorkflowDocumentsByProject(ctx, projectID)
		if err != nil {
			return nil, err
		}
		docs := s.filterWorkflowDocsByStageAccess(ctx, projectID, userID, listed)
		// 工作流作用域：仅在激活工作流的文档池内解析正文；跨流点名返回可解释的引导错误。
		// 激活键由 enrichSessionWorkflowStage 每轮刷新；任务集合按 live 实例快照解析。
		workflowKey, scopeTaskKeys := s.activeWorkflowScopeForDocs(ctx, projectID)
		if sess.UserCtx != nil && sess.UserCtx.Project.ActiveWorkflowKey != "" {
			workflowKey = sess.UserCtx.Project.ActiveWorkflowKey
		}
		if workflowscope.CurrentMode() == workflowscope.ModeWeighted && len(scopeTaskKeys) > 0 {
			// 权重模式把被阶段过滤掉的他流台账加回候选，胜出后再做查看权限校验。
			docs = append(append([]docgov.WorkflowDocumentSummary{}, docs...), filterDocsOutsideTaskKeys(listed, scopeTaskKeys)...)
		}
		doc, err := selectReadableWorkflowDocument(docs, target, workflowKey, scopeTaskKeys)
		if err != nil {
			return nil, err
		}
		owner := intent.TaskKeyWorkflowOwner(doc.TaskKey)
		cross := workflowKey != "" && owner != "" && owner != workflowKey &&
			(len(scopeTaskKeys) == 0 || !projectinstance.TaskKeyInScope(scopeTaskKeys, doc.TaskKey))
		if cross && !s.crossWorkflowViewAllowed(ctx, projectID, userID, doc.TaskKey) {
			title := doc.Title
			if title == "" {
				title = doc.TaskKey
			}
			return nil, &coordinator.ErrDocViewDenied{
				Message: fmt.Sprintf("您无权查看【%s】的《%s》。", intent.WorkflowDisplayName(owner), title),
			}
		}
		content, err := s.loadControlledDocumentContent(ctx, doc.DocID)
		if err != nil {
			return nil, err
		}
		return &coordinator.ReadableDocument{
			DocID:          doc.DocID,
			Title:          doc.Title,
			TaskKey:        doc.TaskKey,
			LifecycleState: string(doc.LifecycleState),
			Version:        doc.Version,
			Markdown:       content.Markdown,
			IsTemplateOnly: doc.IsTemplateOnly || content.IsTemplateOnly,
			WorkflowKey:    owner,
			ActiveWorkflow: workflowKey,
			CrossWorkflow:  cross,
		}, nil
	}
}

// resolveLatestUploadDocument 读取项目内最新上传的 CAS 资产正文（供 @latest_upload 哨兵）。
func (s *Server) resolveLatestUploadDocument(ctx context.Context, projectID int64) (*coordinator.ReadableDocument, error) {
	if s == nil || s.cas() == nil {
		return nil, &coordinator.ErrNoUploadAsset{
			Message: "当前项目还没有上传过附件。请先在对话中上传文件后再试总结。",
		}
	}
	refs, err := s.cas().ListProjectRefs(ctx, projectID)
	if err != nil {
		return nil, err
	}
	latest := pickLatestFileReference(refs)
	if latest == nil {
		return nil, &coordinator.ErrNoUploadAsset{
			Message: "当前项目还没有上传过附件。请先在对话中上传文件后再试总结。",
		}
	}

	matIdx := bizdev.Snapshot(s.svc()).MaterialIndex
	state := ""
	if matIdx != nil {
		state = strings.TrimSpace(matIdx.GetParseState(ctx, latest.BlobSHA256))
	}
	switch state {
	case "failed":
		return nil, &coordinator.ErrUploadParseState{
			State:    state,
			Filename: latest.FileName,
		}
	case "pending", "running":
		limit := 90 * time.Second
		if s.cfg != nil && s.cfg.MineruTimeout > 0 {
			limit = s.cfg.MineruTimeout
		}
		if matIdx != nil {
			if w := matIdx.ParseWaitLimit(); w > 0 {
				limit = w
			}
		}
		finalState := waitForUploadParse(ctx, func() string {
			if matIdx == nil {
				return ""
			}
			return strings.TrimSpace(matIdx.GetParseState(ctx, latest.BlobSHA256))
		}, latest.FileName, limit)
		switch finalState {
		case "done":
			// continue to load content
		case "failed":
			return nil, &coordinator.ErrUploadParseState{
				State:    "failed",
				Filename: latest.FileName,
			}
		case "slow", "pending", "running":
			return nil, &coordinator.ErrUploadParseState{
				State:    "slow",
				Filename: latest.FileName,
				Message: fmt.Sprintf(
					"《%s》解析耗时较长（仍在后台进行），请稍后再让我总结。",
					defaultUploadTitle(latest.FileName),
				),
			}
		default:
			if ctx.Err() != nil {
				return nil, ctx.Err()
			}
			// unknown / empty after wait: fall through and try EnsureParsed
		}
	}

	data, ref, err := s.cas().ReadRefBytes(ctx, latest.ID)
	if err != nil || ref == nil {
		return nil, fmt.Errorf("load latest upload asset: %w", err)
	}
	markdown := ""
	if matIdx != nil {
		md, _, parseErr := matIdx.EnsureParsed(ctx, ref.FileName, ref.BlobSHA256, data)
		if parseErr != nil {
			return nil, parseErr
		}
		markdown = strings.TrimSpace(md)
	} else {
		md, extractErr := textextract.ExtractTextWithAnchors(ref.FileName, data)
		if extractErr != nil {
			return nil, extractErr
		}
		markdown = strings.TrimSpace(md)
	}
	if markdown == "" {
		return nil, &coordinator.ErrUploadParseState{
			State:    "empty",
			Filename: ref.FileName,
			Message:  fmt.Sprintf("《%s》暂无可提取的正文，无法总结。", defaultUploadTitle(ref.FileName)),
		}
	}

	title := defaultUploadTitle(ref.FileName)
	return &coordinator.ReadableDocument{
		DocID:          fmt.Sprintf("asset:%d", ref.ID),
		Title:          title,
		TaskKey:        "uploaded_material",
		LifecycleState: "uploaded",
		Version:        1,
		Markdown:       markdown,
		IsTemplateOnly: false,
	}, nil
}

var uploadParsePollInterval = 2 * time.Second

// waitForUploadParse 在同轮对话内有界等待 MinerU 解析，并周期性发 document_parsing phase。
// stateFn 返回当前解析状态；终态 "done"/"failed" 立即返回；超时返回 "slow"。
func waitForUploadParse(ctx context.Context, stateFn func() string, fileName string, limit time.Duration) string {
	if limit <= 0 {
		limit = 90 * time.Second
	}
	title := defaultUploadTitle(fileName)
	started := time.Now()
	emitUploadParsePhase(ctx, title, 0, 5)

	ticker := time.NewTicker(uploadParsePollInterval)
	defer ticker.Stop()
	deadline := time.NewTimer(limit)
	defer deadline.Stop()

	for {
		state := ""
		if stateFn != nil {
			state = strings.TrimSpace(stateFn())
		}
		switch state {
		case "done", "failed":
			return state
		}

		select {
		case <-ctx.Done():
			return state
		case <-deadline.C:
			final := ""
			if stateFn != nil {
				final = strings.TrimSpace(stateFn())
			}
			if final == "done" || final == "failed" {
				return final
			}
			return "slow"
		case <-ticker.C:
			elapsed := time.Since(started)
			progress := int(elapsed.Seconds() / limit.Seconds() * 100)
			if progress > 95 {
				progress = 95
			}
			if progress < 5 {
				progress = 5
			}
			emitUploadParsePhase(ctx, title, int(elapsed.Seconds()), progress)
		}
	}
}

func emitUploadParsePhase(ctx context.Context, title string, elapsedSec, progress int) {
	st := stream.FromContext(ctx)
	if st == nil {
		return
	}
	msg := fmt.Sprintf("《%s》已排队解析，正在等待 MinerU…", title)
	if elapsedSec > 0 {
		msg = fmt.Sprintf("《%s》正在解析中…已等待 %ds", title, elapsedSec)
	}
	st.EmitPhase("document_parsing", msg, "assistant", 0, 0, progress)
}

func pickLatestFileReference(refs []*storage.FileReference) *storage.FileReference {
	var latest *storage.FileReference
	for _, ref := range refs {
		if ref == nil {
			continue
		}
		if latest == nil {
			latest = ref
			continue
		}
		if ref.CreatedAt.After(latest.CreatedAt) ||
			(ref.CreatedAt.Equal(latest.CreatedAt) && ref.ID > latest.ID) {
			latest = ref
		}
	}
	return latest
}

func defaultUploadTitle(fileName string) string {
	base := filepath.Base(strings.TrimSpace(fileName))
	if base == "" || base == "." {
		return "上传附件"
	}
	stem := strings.TrimSuffix(base, filepath.Ext(base))
	if strings.TrimSpace(stem) == "" {
		return base
	}
	return stem
}

// activeWorkflowScopeForDocs 读取项目激活工作流键与任务集合；失败时返回空（= 不过滤）。
func (s *Server) activeWorkflowScopeForDocs(ctx context.Context, projectID int64) (string, map[string]bool) {
	if s == nil || s.instSvc() == nil || !s.instSvc().Available() || projectID <= 0 {
		return "", nil
	}
	workflowKey, taskKeys, err := s.instSvc().ActiveWorkflowScope(ctx, projectID)
	if err != nil {
		return "", nil
	}
	return workflowKey, taskKeys
}

// taskKeyOutsideActiveWorkflow 判断任务键是否落在激活工作流之外。
// 未激活 / 任务键未知归属时放行（存量兼容）；跨流时返回引导话术。
func (s *Server) taskKeyOutsideActiveWorkflow(ctx context.Context, projectID int64, taskKey string) (bool, string) {
	taskKey = strings.TrimSpace(taskKey)
	if taskKey == "" || projectID <= 0 {
		return false, ""
	}
	workflowKey, scopeTaskKeys := s.activeWorkflowScopeForDocs(ctx, projectID)
	if workflowKey == "" {
		return false, ""
	}
	if len(scopeTaskKeys) > 0 {
		if projectinstance.TaskKeyInScope(scopeTaskKeys, taskKey) {
			return false, ""
		}
		owner := intent.TaskKeyWorkflowOwner(taskKey)
		if workflowscope.BlocksWrite(workflowKey, owner, taskKey) {
			return true, intent.WorkflowScopeGuidanceQuestion(taskKey, []string{owner}, workflowKey)
		}
		return false, ""
	}
	owner := intent.TaskKeyWorkflowOwner(taskKey)
	if workflowscope.BlocksWrite(workflowKey, owner, taskKey) {
		return true, intent.WorkflowScopeGuidanceQuestion(taskKey, []string{owner}, workflowKey)
	}
	return false, ""
}

// workflowKeyForDocument 为列表项标注归属工作流：优先激活流任务集合，其次任务键前缀推断。
func workflowKeyForDocument(taskKey, activeKey string, scopeTaskKeys map[string]bool) string {
	if activeKey != "" && len(scopeTaskKeys) > 0 && projectinstance.TaskKeyInScope(scopeTaskKeys, taskKey) {
		return activeKey
	}
	if owner := intent.TaskKeyWorkflowOwner(taskKey); owner != "" {
		return owner
	}
	return activeKey
}

// errDocWrongWorkflow 目标文档存在但不属于当前激活工作流。
type errDocWrongWorkflow struct {
	Title          string
	TaskKey        string
	OwnerWorkflow  string
	ActiveWorkflow string
}

func (e *errDocWrongWorkflow) Error() string {
	if e == nil {
		return "document belongs to another workflow"
	}
	owners := []string{}
	if e.OwnerWorkflow != "" {
		owners = []string{e.OwnerWorkflow}
	}
	doc := e.Title
	if doc == "" {
		doc = e.TaskKey
	}
	return intent.WorkflowScopeGuidanceQuestion(doc, owners, e.ActiveWorkflow)
}

func (e *errDocWrongWorkflow) AsCoordinatorErr() error {
	return &coordinator.ErrWrongWorkflow{Message: e.Error()}
}

// selectReadableWorkflowDocument 在可读文档池中解析目标。
// scopeTaskKeys 非空时先在激活流池内匹配；未命中但精确/归一化匹配到被剔除的他流文档时返回 ErrWrongWorkflow。
// scopeTaskKeys 为空（未激活 / planning）时保持现状：在全量池内匹配。
func selectReadableWorkflowDocument(
	docs []docgov.WorkflowDocumentSummary,
	target string,
	activeWorkflow string,
	scopeTaskKeys map[string]bool,
) (*docgov.WorkflowDocumentSummary, error) {
	if workflowscope.CurrentMode() == workflowscope.ModeWeighted && activeWorkflow != "" && len(scopeTaskKeys) > 0 && strings.TrimSpace(target) != "" {
		return selectReadableWeighted(docs, target, activeWorkflow, scopeTaskKeys)
	}
	return selectReadableStrict(docs, target, activeWorkflow, scopeTaskKeys)
}

func selectReadableWeighted(
	docs []docgov.WorkflowDocumentSummary,
	target string,
	activeWorkflow string,
	scopeTaskKeys map[string]bool,
) (*docgov.WorkflowDocumentSummary, error) {
	candidates := make([]workflowscope.Candidate, 0, len(docs))
	byID := make(map[string]docgov.WorkflowDocumentSummary, len(docs))
	for _, doc := range docs {
		quality := workflowscope.BestQuality(target, doc.DocID, doc.TaskKey, doc.Title)
		if quality <= 0 {
			continue
		}
		owner := ""
		if projectinstance.TaskKeyInScope(scopeTaskKeys, doc.TaskKey) {
			owner = activeWorkflow
		} else if inferred := intent.TaskKeyWorkflowOwner(doc.TaskKey); inferred != "" {
			owner = inferred
		}
		id := doc.DocID
		if id == "" {
			id = doc.TaskKey
		}
		candidates = append(candidates, workflowscope.Candidate{
			ID:      id,
			Label:   doc.Title,
			Owner:   owner,
			Quality: quality,
		})
		byID[id] = doc
	}
	decision := workflowscope.Decide(candidates, activeWorkflow, workflowscope.ActionRead)
	switch decision.Kind {
	case workflowscope.KindInScope, workflowscope.KindCrossAllow:
		doc := byID[decision.Winner.ID]
		return &doc, nil
	case workflowscope.KindCrossBlock:
		doc := byID[decision.Winner.ID]
		owner := decision.Winner.Owner
		if owner == "" {
			owner = intent.TaskKeyWorkflowOwner(doc.TaskKey)
		}
		return nil, (&errDocWrongWorkflow{
			Title:          doc.Title,
			TaskKey:        doc.TaskKey,
			OwnerWorkflow:  owner,
			ActiveWorkflow: activeWorkflow,
		}).AsCoordinatorErr()
	default:
		return nil, fmt.Errorf("document not found in readable scope")
	}
}

func selectReadableStrict(
	docs []docgov.WorkflowDocumentSummary,
	target string,
	activeWorkflow string,
	scopeTaskKeys map[string]bool,
) (*docgov.WorkflowDocumentSummary, error) {
	if len(docs) == 0 {
		return nil, fmt.Errorf("no readable workflow documents")
	}
	targetKey := normalizeDocumentLookupKey(target)
	if targetKey == "" {
		inScope := filterDocsByTaskKeys(docs, scopeTaskKeys)
		pool := inScope
		if len(scopeTaskKeys) == 0 {
			pool = docs
		}
		if len(pool) == 1 {
			return &pool[0], nil
		}
		return nil, fmt.Errorf("document target required")
	}

	inScope := filterDocsByTaskKeys(docs, scopeTaskKeys)
	pool := inScope
	if len(scopeTaskKeys) == 0 {
		pool = docs
	}
	if hit := matchDocumentInPool(pool, target, targetKey); hit != nil {
		return hit, nil
	}
	// 激活流池未命中：检查是否精确/归一化命中被剔除的他流文档。
	if len(scopeTaskKeys) > 0 {
		outOfScope := filterDocsOutsideTaskKeys(docs, scopeTaskKeys)
		if hit := matchDocumentInPool(outOfScope, target, targetKey); hit != nil {
			owner := intent.TaskKeyWorkflowOwner(hit.TaskKey)
			return nil, (&errDocWrongWorkflow{
				Title:          hit.Title,
				TaskKey:        hit.TaskKey,
				OwnerWorkflow:  owner,
				ActiveWorkflow: activeWorkflow,
			}).AsCoordinatorErr()
		}
	}
	return nil, fmt.Errorf("document not found in readable scope")
}

func filterDocsByTaskKeys(docs []docgov.WorkflowDocumentSummary, taskKeys map[string]bool) []docgov.WorkflowDocumentSummary {
	if len(taskKeys) == 0 {
		return docs
	}
	out := make([]docgov.WorkflowDocumentSummary, 0, len(docs))
	for _, d := range docs {
		if projectinstance.TaskKeyInScope(taskKeys, d.TaskKey) {
			out = append(out, d)
		}
	}
	return out
}

func filterDocsOutsideTaskKeys(docs []docgov.WorkflowDocumentSummary, taskKeys map[string]bool) []docgov.WorkflowDocumentSummary {
	if len(taskKeys) == 0 {
		return nil
	}
	out := make([]docgov.WorkflowDocumentSummary, 0)
	for _, d := range docs {
		if !projectinstance.TaskKeyInScope(taskKeys, d.TaskKey) {
			out = append(out, d)
		}
	}
	return out
}

func matchDocumentInPool(docs []docgov.WorkflowDocumentSummary, target, targetKey string) *docgov.WorkflowDocumentSummary {
	for i := range docs {
		if target == docs[i].DocID || target == docs[i].TaskKey || target == docs[i].Title {
			return &docs[i]
		}
	}
	for i := range docs {
		keys := []string{
			normalizeDocumentLookupKey(docs[i].DocID),
			normalizeDocumentLookupKey(docs[i].TaskKey),
			normalizeDocumentLookupKey(docs[i].Title),
		}
		for _, key := range keys {
			if key != "" && (key == targetKey || strings.Contains(key, targetKey) || strings.Contains(targetKey, key)) {
				return &docs[i]
			}
		}
	}
	return nil
}

func normalizeDocumentLookupKey(value string) string {
	value = strings.ToLower(strings.TrimSpace(value))
	value = strings.NewReplacer("收集", "", "采集", "", "正文", "", "文档", "", "《", "", "》", "").Replace(value)
	return strings.Map(func(r rune) rune {
		if unicode.IsSpace(r) || unicode.IsPunct(r) {
			return -1
		}
		return r
	}, value)
}