hyperguild/ingestion/internal/mcp/tools_context_test.go

package mcp

import (
	"context"
	"encoding/json"
	"os"
	"path/filepath"
	"sort"
	"testing"

	"github.com/mathiasbq/hyperguild/ingestion/internal/graph"
	"github.com/mathiasbq/hyperguild/ingestion/internal/graphstore"
	"github.com/stretchr/testify/assert"
	"github.com/stretchr/testify/require"
)

// fakeGraph implements graphsync.Store + graphReader so it can be
// assigned to Server.graph and downcast by brainContext. Only Subgraph
// is exercised by brain_context today; the rest are no-op satisfiers.
type fakeGraph struct {
	subgraph map[string][]graphstore.Neighbor
}

func (f *fakeGraph) UpsertEntity(_ context.Context, _ graph.Entity) error { return nil }
func (f *fakeGraph) ReplaceEdgesForDoc(_ context.Context, _ string, _ []graph.Edge) error {
	return nil
}
func (f *fakeGraph) DeleteByDoc(_ context.Context, _ string) error { return nil }

func (f *fakeGraph) Neighbors(_ context.Context, slug, _ string, _ int) ([]graphstore.Neighbor, error) {
	return f.subgraph[slug], nil
}

func (f *fakeGraph) Subgraph(_ context.Context, origin string, _ int) ([]graphstore.Neighbor, error) {
	return f.subgraph[origin], nil
}

func (f *fakeGraph) Path(_ context.Context, _, _ string, _ int) ([]graphstore.PathStep, error) {
	return nil, nil
}

func writeNote(t *testing.T, brainDir, relPath, title, body string) {
	t.Helper()
	full := filepath.Join(brainDir, filepath.FromSlash(relPath))
	require.NoError(t, os.MkdirAll(filepath.Dir(full), 0o755))
	content := "---\ntitle: " + title + "\n---\n\n" + body
	require.NoError(t, os.WriteFile(full, []byte(content), 0o644))
}

// callContext runs brainContext directly and decodes the JSON response.
func callContext(t *testing.T, s *Server, args map[string]any) map[string]any {
	t.Helper()
	raw, err := json.Marshal(args)
	require.NoError(t, err)
	out, err := s.brainContext(context.Background(), raw)
	require.NoError(t, err)
	var resp map[string]any
	require.NoError(t, json.Unmarshal(out, &resp))
	return resp
}

func sortedSlugs(entries []any) []string {
	slugs := make([]string, 0, len(entries))
	for _, e := range entries {
		slugs = append(slugs, e.(map[string]any)["slug"].(string))
	}
	sort.Strings(slugs)
	return slugs
}

func TestBrainContext_RejectsMissingProjectRoot(t *testing.T) {
	s := NewServer(t.TempDir(), nil, nil, nil)
	_, err := s.brainContext(context.Background(), json.RawMessage(`{}`))
	assert.Error(t, err)
}

func TestBrainContext_RejectsUnusableBasename(t *testing.T) {
	s := NewServer(t.TempDir(), nil, nil, nil)
	_, err := s.brainContext(context.Background(), json.RawMessage(`{"project_root":"/"}`))
	assert.Error(t, err)
}

func TestBrainContext_BM25Only_NoGraph(t *testing.T) {
	brainDir := t.TempDir()
	// Two notes whose body contains the hyphenated project name. BM25
	// uses literal substring matching after whitespace tokenisation, so
	// the bodies must carry "azure-tiger" verbatim, not "Azure tiger".
	writeNote(t, brainDir, "wiki/finance/decisions/azure-tiger-routing.md",
		"Azure Tiger Routing", "azure-tiger payment routing decisions.")
	writeNote(t, brainDir, "wiki/finance/facts/iso20022.md",
		"Azure Tiger ISO 20022 fields", "azure-tiger maps invoice fields to ISO 20022.")

	s := NewServer(brainDir, nil, nil, nil)
	// graph is nil — only BM25 hits should appear.

	resp := callContext(t, s, map[string]any{
		"project_root": "/home/mathias/dev/QKX/azure-tiger",
	})
	entries := resp["entries"].([]any)
	require.NotEmpty(t, entries, "expected at least one BM25 hit on project name")

	for _, e := range entries {
		entry := e.(map[string]any)
		assert.Equal(t, "bm25", entry["edge_type"], "no graph configured, every entry must be BM25")
		assert.NotEmpty(t, entry["slug"])
		assert.NotEmpty(t, entry["doc_path"])
	}
}

func TestBrainContext_BM25PlusGraphExpansion(t *testing.T) {
	brainDir := t.TempDir()
	// BM25 seed — body carries the hyphenated project name verbatim.
	writeNote(t, brainDir, "wiki/finance/decisions/azure-tiger-routing.md",
		"Azure Tiger Routing", "azure-tiger payment routing decisions.")
	// Graph neighbour — does NOT match BM25 on "azure-tiger" so it can
	// only arrive via the graph subgraph traversal.
	writeNote(t, brainDir, "wiki/finance/facts/sepa-clearing.md",
		"SEPA Clearing", "SEPA payment clearing rules and timing windows.")

	graphFake := &fakeGraph{
		subgraph: map[string][]graphstore.Neighbor{
			"azure-tiger-routing": {
				{
					Slug:     "sepa-clearing",
					Title:    "SEPA Clearing",
					DocPath:  "wiki/finance/facts/sepa-clearing.md",
					EdgeType: "wikilink",
					Distance: 1,
				},
			},
		},
	}
	s := NewServer(brainDir, nil, nil, nil)
	s.graph = graphFake

	resp := callContext(t, s, map[string]any{
		"project_root": "/home/mathias/dev/QKX/azure-tiger",
	})
	entries := resp["entries"].([]any)
	require.GreaterOrEqual(t, len(entries), 2, "expected BM25 seed plus graph neighbour")

	slugs := sortedSlugs(entries)
	assert.Contains(t, slugs, "azure-tiger-routing", "BM25 seed must appear")
	assert.Contains(t, slugs, "sepa-clearing", "graph neighbour must appear")

	// Verify the graph-only entry carries edge_type="graph".
	var sepaEntry map[string]any
	for _, e := range entries {
		m := e.(map[string]any)
		if m["slug"] == "sepa-clearing" {
			sepaEntry = m
			break
		}
	}
	require.NotNil(t, sepaEntry)
	assert.Equal(t, "graph", sepaEntry["edge_type"])
	assert.NotEmpty(t, sepaEntry["excerpt"], "excerpt should be loaded from disk for graph neighbours")
}

func TestBrainContext_LimitClamps(t *testing.T) {
	brainDir := t.TempDir()
	// Five notes all matching "azure-tiger".
	for i, name := range []string{"a", "b", "c", "d", "e"} {
		writeNote(t, brainDir,
			"wiki/finance/decisions/azure-tiger-"+name+".md",
			"Azure Tiger "+name,
			"azure-tiger note "+name+" with index "+string(rune('0'+i)))
	}
	s := NewServer(brainDir, nil, nil, nil)
	resp := callContext(t, s, map[string]any{
		"project_root": "/home/mathias/dev/QKX/azure-tiger",
		"limit":        2,
	})
	entries := resp["entries"].([]any)
	assert.LessOrEqual(t, len(entries), 2)
}

func TestBrainContext_RecentFilesBoost(t *testing.T) {
	brainDir := t.TempDir()
	// Both notes BM25-match the project name, but azure-tiger-z has
	// twice the term frequency so it naturally ranks above azure-tiger-a.
	// The recent_files boost on azure-tiger-a should pull it level on
	// score; the alphabetical slug tiebreaker (a < z) then promotes it
	// to the top — exercising both the boost and the deterministic
	// tiebreak.
	writeNote(t, brainDir, "wiki/finance/decisions/azure-tiger-a.md",
		"A", "azure-tiger note about a.")
	writeNote(t, brainDir, "wiki/finance/decisions/azure-tiger-z.md",
		"Z", "azure-tiger azure-tiger note about z.")

	s := NewServer(brainDir, nil, nil, nil)

	// Baseline ranking: azure-tiger-z must lead (higher term frequency).
	baseline := callContext(t, s, map[string]any{
		"project_root": "/home/mathias/dev/QKX/azure-tiger",
	})
	baselineEntries := baseline["entries"].([]any)
	require.GreaterOrEqual(t, len(baselineEntries), 2)
	baselineTop := baselineEntries[0].(map[string]any)
	require.Equal(t, "azure-tiger-z", baselineTop["slug"],
		"sanity: higher tf must rank first without a boost")

	// With boost on azure-tiger-a — boosted entry must now lead.
	boosted := callContext(t, s, map[string]any{
		"project_root": "/home/mathias/dev/QKX/azure-tiger",
		"recent_files": []string{"/some/where/azure-tiger-a.md"},
	})
	entries := boosted["entries"].([]any)
	require.GreaterOrEqual(t, len(entries), 2)
	top := entries[0].(map[string]any)
	assert.Equal(t, "azure-tiger-a", top["slug"], "recent_files boost must promote the matching doc")
}