ZDDC/zddc/internal/handler/archivehandler_test.go

package handler

import (
	"context"
	"encoding/json"
	"net/http"
	"net/http/httptest"
	"net/url"
	"os"
	"path/filepath"
	"strings"
	"testing"

	"codeberg.org/VARASYS/ZDDC/zddc/internal/archive"
	"codeberg.org/VARASYS/ZDDC/zddc/internal/config"
	"codeberg.org/VARASYS/ZDDC/zddc/internal/listing"
	"codeberg.org/VARASYS/ZDDC/zddc/internal/zddc"
)

// archiveTestRoot lays down a two-project tree so listings exercise project
// scoping, ACL cascading, and the per-project bucket boundary. ACLs are
// written per-test in the helper that calls this.
//
//	<root>/
//	  ProjectA/
//	    2025-01-01_T1 (IFR) - Title/100_~A (IFR) - Title.pdf
//	    2025-01-01_T1 (IFR) - Title/100_A  (IFC) - Title.pdf
//	    2025-02-01_T2 (RTN) - Comments/100_~A+C1 (RTN) - Comments.pdf
//	  ProjectB/
//	    2025-01-01_T3 (IFR) - Title/200_0 (IFR) - Other.pdf
func archiveTestRoot(t *testing.T) (string, *archive.Index) {
	t.Helper()
	root := t.TempDir()

	mk := func(rel string) {
		path := filepath.Join(root, filepath.FromSlash(rel))
		if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
			t.Fatalf("mkdir: %v", err)
		}
		if err := os.WriteFile(path, []byte("x"), 0o644); err != nil {
			t.Fatalf("write %s: %v", path, err)
		}
	}

	mk("ProjectA/2025-01-01_T1 (IFR) - Title/100_~A (IFR) - Title.pdf")
	mk("ProjectA/2025-01-01_T1 (IFR) - Title/100_A (IFC) - Title.pdf")
	mk("ProjectA/2025-02-01_T2 (RTN) - Comments/100_~A+C1 (RTN) - Comments.pdf")
	mk("ProjectB/2025-01-01_T3 (IFR) - Title/200_0 (IFR) - Other.pdf")

	idx, err := archive.BuildIndex(root)
	if err != nil {
		t.Fatalf("BuildIndex: %v", err)
	}
	return root, idx
}

// writeZddc writes a .zddc YAML at <root>/<rel>/.zddc and clears the
// per-directory policy cache so a previous test's permissive .zddc doesn't
// bleed into this one.
func writeZddc(t *testing.T, root, rel, body string) {
	t.Helper()
	dir := filepath.Join(root, filepath.FromSlash(rel))
	if err := os.MkdirAll(dir, 0o755); err != nil {
		t.Fatalf("mkdir %s: %v", dir, err)
	}
	if err := os.WriteFile(filepath.Join(dir, ".zddc"), []byte(body), 0o644); err != nil {
		t.Fatalf("write .zddc: %v", err)
	}
	zddc.InvalidateCache(dir)
}

func archiveCfg(root string) config.Config {
	return config.Config{Root: root, EmailHeader: "X-Auth-Request-Email", IndexPath: ".archive"}
}

func callArchive(t *testing.T, cfg config.Config, idx *archive.Index, email, contextPath, filename string) *httptest.ResponseRecorder {
	t.Helper()
	// Build a syntactically valid URL by escaping each segment of the
	// contextPath and filename. The handler receives the decoded
	// contextPath/filename arguments directly (as the dispatcher would have
	// decoded them); the URL itself just needs to parse for httptest.
	urlPath := encodePath(contextPath) + "/" + cfg.IndexPath
	if filename != "" {
		urlPath += "/" + url.PathEscape(filename)
	} else {
		urlPath += "/"
	}
	req := httptest.NewRequest(http.MethodGet, urlPath, nil)
	req = req.WithContext(context.WithValue(req.Context(), EmailKey, email))
	rec := httptest.NewRecorder()
	ServeArchive(cfg, idx, rec, req, contextPath, filename)
	return rec
}

// encodePath URL-escapes each non-empty slash-separated segment of p so
// special characters like spaces and parens don't break NewRequest's URL
// parser. A leading slash is preserved; an empty input becomes "/".
func encodePath(p string) string {
	trimmed := strings.Trim(p, "/")
	if trimmed == "" {
		return ""
	}
	parts := strings.Split(trimmed, "/")
	for i, s := range parts {
		parts[i] = url.PathEscape(s)
	}
	return "/" + strings.Join(parts, "/")
}

func decodeListing(t *testing.T, body []byte) []listing.FileInfo {
	t.Helper()
	var out []listing.FileInfo
	if err := json.Unmarshal(body, &out); err != nil {
		t.Fatalf("invalid JSON: %v\n%s", err, body)
	}
	return out
}

func names(entries []listing.FileInfo) []string {
	out := make([]string, 0, len(entries))
	for _, e := range entries {
		out = append(out, e.Name)
	}
	return out
}

func contains(xs []string, x string) bool {
	for _, v := range xs {
		if v == x {
			return true
		}
	}
	return false
}

// /.archive/ at the very root has no project segment to scope by, so it's a
// hard 404 — even for an admin. Stable references must include the project
// directory; otherwise cross-project tracking-number collisions would silently
// pick a winner.
func TestServeArchive_RootHasNoProjectScope404(t *testing.T) {
	root, idx := archiveTestRoot(t)
	writeZddc(t, root, ".", `acl:
  allow: ["*"]
`)
	cfg := archiveCfg(root)

	for _, ctx := range []string{"/", ""} {
		t.Run("ctx="+ctx, func(t *testing.T) {
			rec := callArchive(t, cfg, idx, "alice@example.com", ctx, "")
			if rec.Code != http.StatusNotFound {
				t.Errorf("listing at root: status %d, want 404; body = %s", rec.Code, rec.Body.String())
			}
			rec = callArchive(t, cfg, idx, "alice@example.com", ctx, "100.html")
			if rec.Code != http.StatusNotFound {
				t.Errorf("resolve at root: status %d, want 404", rec.Code)
			}
		})
	}
}

// .archive listings are scoped to the contextPath's first segment (the
// project). Each project sees only its own tracking numbers; cross-project
// entries are invisible. Subdirectory contextPaths still resolve to the
// top-level project's bucket — a request from /ProjectA/sub/sub/.archive/
// shows ProjectA's entries with that deeper URL prefix.
func TestServeArchive_ListingScopedToProject(t *testing.T) {
	root, idx := archiveTestRoot(t)
	writeZddc(t, root, ".", `acl:
  allow: ["*"]
`)
	cfg := archiveCfg(root)
	const email = "alice@example.com"

	cases := []struct {
		name        string
		contextPath string
		urlPrefix   string
		wantNames   []string
		denyNames   []string
	}{
		{
			"ProjectA top level",
			"/ProjectA",
			"/ProjectA/.archive/",
			[]string{"100.html", "100_A.html", "100_~A.html"},
			[]string{"200.html", "200_0.html"},
		},
		{
			"ProjectA deeper subpath",
			"/ProjectA/2025-01-01_T1 (IFR) - Title",
			"/ProjectA/2025-01-01_T1 (IFR) - Title/.archive/",
			[]string{"100.html", "100_A.html", "100_~A.html"},
			[]string{"200.html", "200_0.html"},
		},
		{
			"ProjectB top level",
			"/ProjectB",
			"/ProjectB/.archive/",
			[]string{"200.html", "200_0.html"},
			[]string{"100.html", "100_A.html", "100_~A.html"},
		},
	}

	for _, c := range cases {
		t.Run(c.name, func(t *testing.T) {
			rec := callArchive(t, cfg, idx, email, c.contextPath, "")
			if rec.Code != http.StatusOK {
				t.Fatalf("status = %d, want 200; body = %s", rec.Code, rec.Body.String())
			}
			got := decodeListing(t, rec.Body.Bytes())
			gotNames := names(got)
			for _, want := range c.wantNames {
				if !contains(gotNames, want) {
					t.Errorf("missing %q at %s; got %v", want, c.contextPath, gotNames)
				}
			}
			for _, deny := range c.denyNames {
				if contains(gotNames, deny) {
					t.Errorf("unexpected cross-project entry %q at %s; got %v", deny, c.contextPath, gotNames)
				}
			}
			for _, e := range got {
				if !strings.HasPrefix(e.URL, c.urlPrefix) {
					t.Errorf("entry %q URL = %q, want %s prefix", e.Name, e.URL, c.urlPrefix)
				}
			}
		})
	}
}

// Listing endpoint is gated by the contextPath ACL: callers who can't reach
// the directory the .archive virtually sits in get 403 (the directory is
// known to exist; just not accessible).
func TestServeArchive_ListingDeniedByContextPathACL(t *testing.T) {
	root, idx := archiveTestRoot(t)
	writeZddc(t, root, ".", `acl:
  allow: ["alice@example.com"]
`)
	writeZddc(t, root, "ProjectA", `acl:
  deny: ["mallory@example.com"]
  allow: ["alice@example.com"]
`)
	cfg := archiveCfg(root)

	rec := callArchive(t, cfg, idx, "mallory@example.com", "/ProjectA", "")
	if rec.Code != http.StatusForbidden {
		t.Errorf("denied caller got status %d, want 403; body = %s", rec.Code, rec.Body.String())
	}

	rec = callArchive(t, cfg, idx, "alice@example.com", "/ProjectA", "")
	if rec.Code != http.StatusOK {
		t.Errorf("allowed caller got status %d, want 200; body = %s", rec.Code, rec.Body.String())
	}
}

// Listing entries are filtered per-target by ACL: a caller denied at a
// subtree's transmittal directory sees no entries whose target lives there.
// Excluding a user from a subdir requires an explicit deny there (the
// cascade is "first explicit match wins, bottom-up", so a child allow list
// doesn't narrow a parent's allow:["*"]).
func TestServeArchive_ListingFiltersEntriesByPerTargetACL(t *testing.T) {
	root, idx := archiveTestRoot(t)
	writeZddc(t, root, ".", `acl:
  allow: ["*"]
`)
	// Deny alice on the transmittal folder where 100_~A+C1 lives, so her
	// listing of /ProjectA/.archive/ drops that entry — but other ProjectA
	// entries stay visible. (A blanket /ProjectA deny would 403 the
	// listing entirely; that's covered by the previous test.)
	writeZddc(t, root, "ProjectA/2025-02-01_T2 (RTN) - Comments", `acl:
  deny: ["alice@example.com"]
`)
	cfg := archiveCfg(root)

	rec := callArchive(t, cfg, idx, "alice@example.com", "/ProjectA", "")
	if rec.Code != http.StatusOK {
		t.Fatalf("status = %d, want 200; body = %s", rec.Code, rec.Body.String())
	}
	gotNames := names(decodeListing(t, rec.Body.Bytes()))

	for _, want := range []string{"100.html", "100_A.html", "100_~A.html"} {
		if !contains(gotNames, want) {
			t.Errorf("alice missing accessible entry %q; got %v", want, gotNames)
		}
	}

	// Bob has no per-target denials in either project.
	rec = callArchive(t, cfg, idx, "bob@example.com", "/ProjectB", "")
	if rec.Code != http.StatusOK {
		t.Fatalf("bob ProjectB listing: status %d, want 200", rec.Code)
	}
	gotNames = names(decodeListing(t, rec.Body.Bytes()))
	if !contains(gotNames, "200.html") {
		t.Errorf("bob should see ProjectB entry 200.html; got %v", gotNames)
	}
}

// Direct redirect requests for a tracking number whose target the caller
// can't read return 404 (not 403, not 302) — the file's existence must not
// leak across the ACL boundary. Cross-project tracking-number requests also
// 404 because each project's bucket is separate.
func TestServeArchive_ResolveACLDeniedReturns404(t *testing.T) {
	root, idx := archiveTestRoot(t)
	writeZddc(t, root, ".", `acl:
  allow: ["*"]
`)
	writeZddc(t, root, "ProjectB", `acl:
  deny: ["alice@example.com"]
`)
	cfg := archiveCfg(root)

	// 200 doesn't even live in ProjectA, so the resolver itself returns 404
	// regardless of ACL — project scoping comes first.
	rec := callArchive(t, cfg, idx, "alice@example.com", "/ProjectA", "200.html")
	if rec.Code != http.StatusNotFound {
		t.Errorf("alice → /ProjectA/.archive/200.html: status %d, want 404 (cross-project)", rec.Code)
	}

	// Alice in /ProjectA can resolve all of ProjectA's entries.
	for _, fn := range []string{"100.html", "100_A.html", "100_~A.html", "100_~A+C1.html"} {
		rec := callArchive(t, cfg, idx, "alice@example.com", "/ProjectA", fn)
		if rec.Code != http.StatusFound {
			t.Errorf("alice → /ProjectA/.archive/%s: status %d, want 302; body = %s", fn, rec.Code, rec.Body.String())
		}
	}

	// Alice attempting ProjectB directly is denied at the contextPath ACL.
	rec = callArchive(t, cfg, idx, "alice@example.com", "/ProjectB", "200.html")
	if rec.Code != http.StatusForbidden {
		t.Errorf("alice → /ProjectB/.archive/200.html: status %d, want 403 (denied at contextPath)", rec.Code)
	}

	// Bob has no denies — he can pull 200.html from /ProjectB.
	rec = callArchive(t, cfg, idx, "bob@example.com", "/ProjectB", "200.html")
	if rec.Code != http.StatusFound {
		t.Errorf("bob → /ProjectB/.archive/200.html: status %d, want 302", rec.Code)
	}
}

// Cascade direction sanity check: a denial at the subtree wins over an
// allow at the parent, AND a target-level allow can rescue a user the
// parent didn't mention. Both directions must be exercised so future
// refactors of the per-target ACL helper can't silently break one.
func TestServeArchive_CascadeDirectionsBothEnforced(t *testing.T) {
	root, idx := archiveTestRoot(t)
	// Root: deny default — only bob is on the list. ProjectA: explicitly
	// allow alice. So alice is rescued at ProjectA, mallory stays out
	// everywhere, bob stays in everywhere. Per-target ACL on resolved files
	// doesn't kick in here — both projects allow bob via the root rule.
	writeZddc(t, root, ".", `acl:
  allow: ["bob@example.com"]
`)
	writeZddc(t, root, "ProjectA", `acl:
  allow: ["alice@example.com"]
`)
	cfg := archiveCfg(root)

	cases := []struct {
		email       string
		contextPath string
		filename    string
		wantStatus  int
		why         string
	}{
		{"bob@example.com", "/ProjectA", "100.html", http.StatusFound, "bob allowed at root → reaches ProjectA target"},
		{"bob@example.com", "/ProjectB", "200.html", http.StatusFound, "bob allowed at root → reaches ProjectB target"},
		{"alice@example.com", "/ProjectA", "100.html", http.StatusFound, "alice rescued by ProjectA allow"},
		{"alice@example.com", "/ProjectB", "200.html", http.StatusForbidden, "alice not in ProjectB chain → 403 at contextPath"},
		// mallory denied everywhere; the contextPath gate fires first.
		{"mallory@example.com", "/ProjectA", "100.html", http.StatusForbidden, "mallory blocked at contextPath"},
	}
	for _, c := range cases {
		t.Run(c.email+"_"+c.contextPath+"_"+c.filename, func(t *testing.T) {
			rec := callArchive(t, cfg, idx, c.email, c.contextPath, c.filename)
			if rec.Code != c.wantStatus {
				t.Errorf("%s @ %s → %s: status %d, want %d (%s)", c.email, c.contextPath, c.filename, rec.Code, c.wantStatus, c.why)
			}
		})
	}
}

// Resolved redirect Location header is the absolute path to the actual file
// under cfg.Root. From any depth within the same project, the resolver
// returns the same target — `/ProjectA/.archive/100.html` and
// `/ProjectA/2025-01-01_T1 (IFR) - Title/.archive/100.html` 302 to the same
// file because both look up project ProjectA.
func TestServeArchive_ResolveLocationStableAcrossDepthWithinProject(t *testing.T) {
	root, idx := archiveTestRoot(t)
	writeZddc(t, root, ".", `acl:
  allow: ["*"]
`)
	cfg := archiveCfg(root)

	wantLocPrefix := "/ProjectA/2025-01-01_T1 (IFR) - Title/100_A"
	for _, ctx := range []string{
		"/ProjectA",
		"/ProjectA/2025-01-01_T1 (IFR) - Title",
		"/ProjectA/2025-02-01_T2 (RTN) - Comments",
	} {
		rec := callArchive(t, cfg, idx, "alice@example.com", ctx, "100.html")
		if rec.Code != http.StatusFound {
			t.Errorf("ctx=%s status=%d body=%s", ctx, rec.Code, rec.Body.String())
			continue
		}
		loc := rec.Header().Get("Location")
		if !strings.HasPrefix(loc, wantLocPrefix) {
			t.Errorf("ctx=%s Location=%q, want prefix %q", ctx, loc, wantLocPrefix)
		}
	}
}

// Cross-project: same tracking number issued under two projects. Each
// project's .archive/ resolves to its own copy, never the other's.
func TestServeArchive_CrossProjectSameTrackingNoLeak(t *testing.T) {
	root := t.TempDir()
	mk := func(rel string) {
		path := filepath.Join(root, filepath.FromSlash(rel))
		if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
			t.Fatalf("mkdir: %v", err)
		}
		if err := os.WriteFile(path, []byte("x"), 0o644); err != nil {
			t.Fatalf("write %s: %v", path, err)
		}
	}
	mk("ProjectA/2025-01-01_T1 (IFR) - Title/123_A (IFR) - Title.pdf")
	mk("ProjectB/2025-06-01_T9 (IFR) - Other Title/123_A (IFR) - Other Title.pdf")
	idx, err := archive.BuildIndex(root)
	if err != nil {
		t.Fatalf("BuildIndex: %v", err)
	}

	writeZddc(t, root, ".", `acl:
  allow: ["*"]
`)
	cfg := archiveCfg(root)
	const email = "alice@example.com"

	recA := callArchive(t, cfg, idx, email, "/ProjectA", "123.html")
	if recA.Code != http.StatusFound {
		t.Fatalf("ProjectA 123.html status=%d body=%s", recA.Code, recA.Body.String())
	}
	locA := recA.Header().Get("Location")
	if !strings.HasPrefix(locA, "/ProjectA/") {
		t.Errorf("ProjectA Location=%q, want /ProjectA/ prefix", locA)
	}

	recB := callArchive(t, cfg, idx, email, "/ProjectB", "123.html")
	if recB.Code != http.StatusFound {
		t.Fatalf("ProjectB 123.html status=%d body=%s", recB.Code, recB.Body.String())
	}
	locB := recB.Header().Get("Location")
	if !strings.HasPrefix(locB, "/ProjectB/") {
		t.Errorf("ProjectB Location=%q, want /ProjectB/ prefix", locB)
	}

	if locA == locB {
		t.Errorf("cross-project leak: same Location for both projects: %q", locA)
	}

	// Listing each project shows only its own.
	for _, c := range []struct{ ctx, mustHave, mustNot string }{
		{"/ProjectA", "ProjectA", "ProjectB"},
		{"/ProjectB", "ProjectB", "ProjectA"},
	} {
		rec := callArchive(t, cfg, idx, email, c.ctx, "")
		if rec.Code != http.StatusOK {
			t.Fatalf("listing %s: status %d", c.ctx, rec.Code)
		}
		got := decodeListing(t, rec.Body.Bytes())
		for _, e := range got {
			if !strings.Contains(e.URL, "/"+c.mustHave+"/") {
				t.Errorf("ctx=%s entry URL %q lacks /%s/ segment", c.ctx, e.URL, c.mustHave)
			}
		}
	}
}

// Default-deny: as soon as ANY .zddc exists in the chain, an unmatched
// caller is denied. Verify this applies to listing entries too — a target
// in a directory with a restrictive .zddc is not surfaced to outsiders even
// though the file exists.
func TestServeArchive_DefaultDenyOnceZddcExists(t *testing.T) {
	root, idx := archiveTestRoot(t)
	// Root .zddc allows alice only. No "*" — so anyone else is default-denied.
	writeZddc(t, root, ".", `acl:
  allow: ["alice@example.com"]
`)
	cfg := archiveCfg(root)

	// alice sees everything she's allowed to in ProjectA.
	rec := callArchive(t, cfg, idx, "alice@example.com", "/ProjectA", "")
	if rec.Code != http.StatusOK {
		t.Fatalf("alice listing: status %d, want 200", rec.Code)
	}
	if len(decodeListing(t, rec.Body.Bytes())) == 0 {
		t.Errorf("alice listing was empty, want entries")
	}

	// Charlie isn't on any list → default-deny → 403 even for the listing.
	rec = callArchive(t, cfg, idx, "charlie@example.com", "/ProjectA", "")
	if rec.Code != http.StatusForbidden {
		t.Errorf("charlie listing: status %d, want 403", rec.Code)
	}

	// Direct resolve: contextPath ACL fires first → 403.
	rec = callArchive(t, cfg, idx, "charlie@example.com", "/ProjectA", "100.html")
	if rec.Code != http.StatusForbidden {
		t.Errorf("charlie resolve: status %d, want 403 (denied at contextPath)", rec.Code)
	}
}

// Empty email never matches — even an `allow: ["*"]` policy denies it,
// which is the existing zddc package contract. .archive must honor it.
func TestServeArchive_EmptyEmailDeniedEvenWithStarAllow(t *testing.T) {
	root, idx := archiveTestRoot(t)
	writeZddc(t, root, ".", `acl:
  allow: ["*@example.com"]
`)
	cfg := archiveCfg(root)

	rec := callArchive(t, cfg, idx, "", "/ProjectA", "")
	if rec.Code != http.StatusForbidden {
		t.Errorf("anonymous listing: status %d, want 403", rec.Code)
	}
}

// projectFromContextPath is the canonical place to derive the project key
// from the .archive contextPath. Pin the edge cases.
func TestProjectFromContextPath(t *testing.T) {
	cases := []struct {
		ctx  string
		want string
	}{
		{"/ProjectA", "ProjectA"},
		{"/ProjectA/", "ProjectA"},
		{"/ProjectA/sub/sub", "ProjectA"},
		{"/", ""},
		{"", ""},
		{"ProjectA/sub", "ProjectA"},
	}
	for _, c := range cases {
		got := projectFromContextPath(c.ctx)
		if got != c.want {
			t.Errorf("projectFromContextPath(%q) = %q, want %q", c.ctx, got, c.want)
		}
	}
}