Merge pull request 'Reject pseudo-binary Office writes and index extracted text' (#34) from feature/binary-doc-index into main

internal/runmgr/exec_tool_artifact_test.go

@@ -12,6 +12,7 @@ import (
 	"testing"
 
 	"agent-system-backend/internal/bus"
+	"agent-system-backend/internal/docgen"
 	"agent-system-backend/internal/settings"
 	"agent-system-backend/internal/store"
 	"agent-system-backend/internal/tools"
@@ -113,3 +114,33 @@ func TestExecuteToolNoArtifactByDefault(t *testing.T) {
 		t.Errorf("无产物工具不应携带 Artifact: %+v", res)
 	}
 }
+
+func TestGeneratedDocxIsSearchable(t *testing.T) {
+	c, st := newExecEnv(t)
+	const marker = "zebraindexword"
+	docx, err := docgen.Build("纪要", "结论："+marker)
+	if err != nil {
+		t.Fatal(err)
+	}
+	mime := "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+	if _, err := c.WriteArtifact("note.docx", "document", mime, docx); err != nil {
+		t.Fatal(err)
+	}
+	hits, err := st.SearchKnowledge("alice", core.RetrievalQuery{Text: marker, ProjectID: c.run.ProjectID, TopK: 5})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(hits) == 0 {
+		t.Fatal("generated docx text was not indexed")
+	}
+	if _, err := c.WriteArtifact("bad.docx", "document", mime, []byte("notazipmarker")); err != nil {
+		t.Fatal(err)
+	}
+	hits, err = st.SearchKnowledge("alice", core.RetrievalQuery{Text: "notazipmarker", ProjectID: c.run.ProjectID, TopK: 5})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(hits) != 0 {
+		t.Fatalf("unparseable docx bytes should not be indexed, got %d hits", len(hits))
+	}
+}

internal/runmgr/runmgr.go

@@ -13,6 +13,7 @@ import (
 
 	"agent-system-backend/internal/bus"
 	"agent-system-backend/internal/cardsvc"
+	"agent-system-backend/internal/ingest"
 	"agent-system-backend/internal/profiles"
 	"agent-system-backend/internal/runtime"
 	"agent-system-backend/internal/settings"
@@ -352,14 +353,15 @@ func (c *runCtx) WriteArtifact(name, kind, mime string, content []byte) (core.Ar
 	if _, err := c.mgr.emit(c.run.ConversationID, c.run.ID, core.EvArtifactCreated, core.ArtifactPayload{Artifact: art}); err != nil {
 		return art, err
 	}
-	// 工件文本进检索索引（管线分块+全文+可选向量；未注入回落整文档单块）。
-	// 二进制 Office/PDF 不进字符串索引（zip 字节当文本是垃圾数据）；
-	// docx 的可检索文本由 ingest 管线在上传路径解析，生成路径暂不索引。
-	if kind == "document" && !binaryDocMime(mime) && len(content) < 200000 {
-		if c.mgr.indexer != nil {
-			_ = c.mgr.indexer.IndexDocument(c.run.Owner, c.run.ProjectID, "artifact", name, art.ID, string(content))
-		} else {
-			_ = c.mgr.st.AddKnowledge(core.KnowledgeDoc{Owner: c.run.Owner, ProjectID: c.run.ProjectID, Source: "artifact", Title: name, Ref: art.ID, Chunk: string(content)})
+	// 工件文本进检索索引。二进制 Office/PDF 先经 ingest 抽出纯文本，抽到才入库；
+	// 抽不到则跳过，避免把 ZIP 字节当文本索引。
+	if kind == "document" && len(content) < 200000 {
+		if text := indexTextOf(name, mime, content); text != "" {
+			if c.mgr.indexer != nil {
+				_ = c.mgr.indexer.IndexDocument(c.run.Owner, c.run.ProjectID, "artifact", name, art.ID, text)
+			} else {
+				_ = c.mgr.st.AddKnowledge(core.KnowledgeDoc{Owner: c.run.Owner, ProjectID: c.run.ProjectID, Source: "artifact", Title: name, Ref: art.ID, Chunk: text})
+			}
 		}
 	}
 	return art, nil
@@ -381,6 +383,19 @@ func binaryDocMime(mime string) bool {
 	return false
 }
 
+// indexTextOf 决定工件进检索索引的文本。普通文本原样返回；二进制文档 mime
+// 先走 ingest 抽文本，解析失败或为空则返回空串（跳过索引）。
+func indexTextOf(name, mime string, content []byte) string {
+	if !binaryDocMime(mime) {
+		return string(content)
+	}
+	doc := ingest.Parse(name, mime, content)
+	if len(doc.Warnings) > 0 || len(doc.Blocks) == 0 {
+		return ""
+	}
+	return strings.TrimSpace(doc.PlainText())
+}
+
 func (c *runCtx) Retrieve(q core.RetrievalQuery) ([]core.RetrievalHit, error) {
 	if q.TopK <= 0 {
 		q.TopK = int(c.eff.GetNumber("retrieval.topK", 5))

internal/tools/fs.go

@@ -113,6 +113,9 @@ func execFSWrite(rc ToolExecCtx, args json.RawMessage) (string, error) {
 	if a.Path == "" {
 		return "", fmt.Errorf("fs_write_file: path required")
 	}
+	if err := rejectBinaryOfficeWrite(a.Path); err != nil {
+		return "", err
+	}
 	if len(a.Content) > fsMaxWrite {
 		return "", fmt.Errorf("fs_write_file: 内容超过 2MB 上限")
 	}
@@ -218,3 +221,14 @@ func mimeOfExt(name string) string {
 		return "text/plain; charset=utf-8"
 	}
 }
+
+// rejectBinaryOfficeWrite 拒绝把字符串直接写成伪二进制 Office/PDF。
+// 带排版的文件由 run_python 生成后再 register_artifact 登记。
+func rejectBinaryOfficeWrite(name string) error {
+	switch strings.ToLower(filepath.Ext(name)) {
+	case ".doc", ".docx", ".xls", ".xlsx", ".xlsm", ".ppt", ".pptx", ".pdf":
+		return fmt.Errorf("%s 是二进制文档扩展名，不能把字符串直接写成该文件。请用 run_python 生成到工作区后，用 register_artifact 登记", filepath.Ext(name))
+	default:
+		return nil
+	}
+}

internal/tools/fs_test.go

@@ -161,3 +161,21 @@ func TestFSUnconfiguredRoot(t *testing.T) {
 		t.Errorf("unconfigured root should fail: %s", res.Output)
 	}
 }
+
+func TestBinaryOfficeWriteRejected(t *testing.T) {
+	r := newFSRegistry(t)
+	rc := &fakeFSCtx{root: t.TempDir(), owner: "o1", conv: "c1", fsToolsOn: true}
+	for _, args := range []struct{ name, body string }{
+		{"fs_write_file", `{"path":"a.docx","content":"plain"}`},
+		{"fs_write_file", `{"path":"a.pdf","content":"plain"}`},
+		{"write_artifact", `{"name":"a.xlsx","content":"plain"}`},
+	} {
+		res := call(t, r, rc, args.name, args.body)
+		if res.Status != "error" || !strings.Contains(res.Output, "run_python") {
+			t.Errorf("%s %s: %s %s", args.name, args.body, res.Status, res.Output)
+		}
+	}
+	if res := call(t, r, rc, "write_artifact", `{"name":"a.md","content":"plain"}`); res.Status != "ok" {
+		t.Fatalf("markdown write should stay allowed: %s", res.Output)
+	}
+}

internal/tools/registry.go

@@ -164,6 +164,9 @@ func (r *Registry) RegisterBuiltin() error {
 		if a.Name == "" {
 			return "", fmt.Errorf("write_artifact: name required")
 		}
+		if err := rejectBinaryOfficeWrite(a.Name); err != nil {
+			return "", err
+		}
 		kind := a.Kind
 		if kind == "" {
 			kind = "document"