Agent-System-Backend/internal/runmgr/material_retrieve_test.go

package runmgr

import (
	"strings"
	"testing"

	core "agent-system-core"
)

func materialRetrievalContext(t *testing.T) *runCtx {
	t.Helper()
	c := checkpointEnvironment(t)
	card, err := c.mgr.st.GetCard(c.run.Owner, c.task.cardID)
	if err != nil || card.Task == nil || len(card.Task.Inputs) != 1 {
		t.Fatalf("trusted material input missing: card=%+v err=%v", card, err)
	}
	input := card.Task.Inputs[0]
	c.attachments = []core.ParsedDocument{{
		Meta: core.DocMeta{Filename: "决定.txt", Parser: "builtin:txt", Origin: &core.DocumentOrigin{
			ArtifactID: core.ID(input.ArtifactID), ContentHash: input.ContentHash, ParserRevision: input.ParserRevision,
		}},
		Blocks: []core.DocBlock{{Type: core.BlockPara, Text: "上线六月"}},
	}}
	return c
}

func TestMaterialAttachmentFallbackQuery(t *testing.T) {
	c := materialRetrievalContext(t)
	hits, err := c.Retrieve(core.RetrievalQuery{Text: "决定.txt 上线 日期"})
	if err != nil || len(hits) != 1 {
		t.Fatalf("current material was not recovered from a combined query: hits=%+v err=%v", hits, err)
	}
	hit := hits[0].Doc
	if hit.Ref != string(c.attachments[0].Meta.Origin.ArtifactID) ||
		hit.ProjectID != c.run.ProjectID || hit.Owner != c.run.Owner ||
		hit.Title != "决定.txt" || !strings.Contains(hit.Chunk, "上线六月") {
		t.Fatalf("fallback lost the current attachment identity or text: %+v", hit)
	}
}

func TestMaterialAttachmentFallbackNeedsExactFilename(t *testing.T) {
	c := materialRetrievalContext(t)
	for _, tc := range []struct {
		query string
		want  int
	}{
		{"决定.txt 上线 日期", 1},
		{`"决定.txt" 上线`, 1},
		{"非决定.txt 上线", 0},
		{"决定.txt备份 上线", 0},
		{"前缀-决定.txt 上线", 0},
	} {
		hits, err := c.materialAttachmentFallback(core.RetrievalQuery{ProjectID: c.run.ProjectID, Text: tc.query})
		if err != nil || len(hits) != tc.want {
			t.Fatalf("query %q matched %d current attachments; want %d (err=%v)", tc.query, len(hits), tc.want, err)
		}
	}
}

func TestMaterialAttachmentFallbackScope(t *testing.T) {
	for _, tc := range []struct {
		name   string
		change func(*runCtx)
	}{
		{"other-subagent", func(c *runCtx) { c.subAgent = "agent.sub.general" }},
		{"other-owner", func(c *runCtx) { c.run.Owner = "another-owner" }},
		{"other-project", func(c *runCtx) { c.run.ProjectID = "another-project" }},
		{"duplicate-name", func(c *runCtx) {
			other := c.attachments[0]
			origin := *other.Meta.Origin
			origin.ArtifactID = "art_other"
			other.Meta.Origin = &origin
			c.attachments = append(c.attachments, other)
		}},
	} {
		t.Run(tc.name, func(t *testing.T) {
			c := materialRetrievalContext(t)
			before, err := c.Retrieve(core.RetrievalQuery{Text: "决定.txt 上线 日期"})
			if err != nil || len(before) != 1 {
				t.Fatalf("control material was unavailable before scope change: hits=%+v err=%v", before, err)
			}
			tc.change(c)
			hits, err := c.Retrieve(core.RetrievalQuery{Text: "决定.txt 上线 日期"})
			if len(hits) != 0 {
				t.Fatalf("out-of-scope attachment leaked: hits=%+v err=%v", hits, err)
			}
		})
	}
}

func TestMaterialAttachmentFallbackVersionDrift(t *testing.T) {
	for _, tc := range []struct {
		name   string
		change func(*testing.T, *runCtx)
	}{
		{"changed-content", func(t *testing.T, c *runCtx) {
			t.Helper()
			if err := c.mgr.st.UpdateArtifactContent(string(c.run.Owner), string(c.attachments[0].Meta.Origin.ArtifactID), []byte("上线七月")); err != nil {
				t.Fatal(err)
			}
		}},
		{"changed-parser-revision", func(_ *testing.T, c *runCtx) {
			c.attachments[0].Meta.Origin.ParserRevision = "different-parser"
		}},
	} {
		t.Run(tc.name, func(t *testing.T) {
			c := materialRetrievalContext(t)
			before, err := c.Retrieve(core.RetrievalQuery{Text: "决定.txt 上线 日期"})
			if err != nil || len(before) != 1 {
				t.Fatalf("control material was unavailable before version change: hits=%+v err=%v", before, err)
			}
			tc.change(t, c)
			hits, err := c.Retrieve(core.RetrievalQuery{Text: "决定.txt 上线 日期"})
			if len(hits) != 0 {
				t.Fatalf("stale material was returned: hits=%+v err=%v", hits, err)
			}
		})
	}
}

func TestMaterialAttachmentFallbackRespectsResourceAllow(t *testing.T) {
	c := materialRetrievalContext(t)
	c.mgr.SetKnowledgeResourceAllow(func(owner string, doc core.KnowledgeDoc) bool {
		return owner == string(c.run.Owner) && doc.Ref == string(c.attachments[0].Meta.Origin.ArtifactID)
	})
	allowed, err := c.Retrieve(core.RetrievalQuery{Text: "决定.txt 上线 日期"})
	if err != nil || len(allowed) != 1 {
		t.Fatalf("authorized current material was unavailable: hits=%+v err=%v", allowed, err)
	}
	c.mgr.SetKnowledgeResourceAllow(func(string, core.KnowledgeDoc) bool { return false })
	denied, err := c.Retrieve(core.RetrievalQuery{Text: "决定.txt 上线 日期"})
	if err != nil || len(denied) != 0 {
		t.Fatalf("resource policy was bypassed by fallback: hits=%+v err=%v", denied, err)
	}
}

func TestMaterialExcerptKeepsMatchedBeginning(t *testing.T) {
	full := "上线六月" + strings.Repeat("附", 20) + "后续结论"
	docs := []core.ParsedDocument{{
		Meta:   core.DocMeta{Filename: "长文.txt", Origin: &core.DocumentOrigin{ArtifactID: "art_long"}},
		Blocks: []core.DocBlock{{Type: core.BlockPara, Text: full}},
	}}
	eff := core.EffectiveSettings{"context": map[string]any{"maxAttachmentChars": float64(4)}}
	got := materialExcerpt(full, "长文.txt 上线", "长文.txt", docs, "art_long", eff, 10)
	if want := "上线六月" + strings.Repeat("附", 6); got != want {
		t.Fatalf("matched start lost or exceeded the result limit: got %q want %q", got, want)
	}
	got = materialExcerpt(full, "长文.txt", "长文.txt", docs, "art_long", eff, 10)
	if want := strings.Repeat("附", 10); got != want {
		t.Fatalf("filename-only retrieval did not start after the inline prefix: got %q want %q", got, want)
	}
}