smallwebwaf/internal/coreruleset/gitea_test.go

package coreruleset_test

import (
	"net/http"
	"slices"
	"testing"

	"sneak.berlin/go/smallwebwaf/internal/coreruleset"
)

// giteaSearches are searches of the pages gitea and Forgejo share whose q
// skips the lists of system files and command names, in the eighth change.
// gitea serves its admin pages under /-/admin/ and Forgejo under /admin/.
//
//nolint:gochecknoglobals // a constant cannot be a list
var giteaSearches = []string{
	"/notifications/watching",
	"/org/acme/milestones/developers",
	"/org/acme/teams/-/search",
	"/alice/-/projects",
	"/alice/-/packages/generic/tools-cli/versions",
	"/alice/-/packages/npm/@alice%2Ftools-cli/versions",
	"/alice/tools/milestones",
	"/alice/tools/projects",
	"/alice/tools/issues/suggestions",
	"/alice/tools/settings/actions/runners",
	"/org/acme/settings/actions/runners",
	"/user/settings/actions/runners",
	"/-/admin/actions/runners",
	"/-/admin/users",
	"/-/admin/orgs",
	"/-/admin/repos",
	"/-/admin/repos/unadopted",
	"/-/admin/emails",
	"/-/admin/emails/activate",
	"/admin/actions/runners",
	"/admin/users",
	"/admin/repos/unadopted",
	"/admin/emails/activate",
}

// giteaTagAndPackageSearches are the searches of tags and packages, whose q
// also lets /bin/sh through, since a tag or package name can hold a slash.
//
//nolint:gochecknoglobals // a constant cannot be a list
var giteaTagAndPackageSearches = []string{
	"/alice/tools/tags",
	"/alice/tools/packages",
	"/-/admin/packages",
	"/admin/packages",
}

func TestGiteaAdminFormsLetHardValuesThroughInTheBody(t *testing.T) {
	t.Parallel()

	crs := readingBodies(t)

	// The forms that adopt a repository and activate an email send the
	// admin page's search in q, as gitea 1.27.3 and Forgejo 17.0.1 do.
	for _, value := range hardValues {
		adopt := "id=alice%2Flegacy-site&action=adopt&q=" + value + "&page=1"

		for _, path := range []string{"/-/admin/repos/unadopted", "/admin/repos/unadopted"} {
			r := request{http.MethodPost, path, []string{formData}}
			wantBody(t, crs, r, adopt, coreruleset.Result{}, adopt)
		}

		activate := "sort=email&q=" + value + "&is_primary=&is_activated=" +
			"&uid=2&email=bob%40example.com&primary=1&activate=0"

		for _, path := range []string{"/-/admin/emails/activate", "/admin/emails/activate"} {
			r := request{http.MethodPost, path, []string{formData}}
			wantBody(t, crs, r, activate, coreruleset.Result{}, activate)
		}
	}
}

func TestGiteaAttacksAreRefusedInItsSearches(t *testing.T) {
	t.Parallel()

	const threshold = 5 // SWW_WAF_CORE_RULE_SET_ANOMALY_THRESHOLD's default

	defaults, bodies := atDefaults(t), readingBodies(t)

	for _, path := range slices.Concat(giteaSearches, giteaTagAndPackageSearches) {
		for _, attack := range []string{
			sqlInjection, pathTraversal, scriptInjection, commandInjection,
		} {
			if got := inspect(t, defaults, get(path+"?q="+attack)); got.Score < threshold {
				t.Errorf("GET %s?q=%s let through: %+v", path, attack, got)
			}

			// The same field in a form data body.
			r := request{http.MethodPost, path, []string{formData}}
			if got, _ := inspectBody(t, bodies, r, "q="+attack); got.Score < threshold {
				t.Errorf("POST %s q=%s let through: %+v", path, attack, got)
			}
		}
	}
}

func TestGiteaSearchExclusionsAreNarrow(t *testing.T) {
	t.Parallel()

	const threshold = 5 // SWW_WAF_CORE_RULE_SET_ANOMALY_THRESHOLD's default

	crs := atDefaults(t)

	// /bin/sh stays refused in q on the searches that are not of tags or
	// packages.
	for _, path := range giteaSearches {
		if got := inspect(t, crs, get(path+"?q=%2Fbin%2Fsh")); got.Score < threshold {
			t.Errorf("%s?q=%%2Fbin%%2Fsh let through: %+v", path, got)
		}
	}

	attacks := []string{"%2Fbin%2Fsh", commandInjection, pathTraversal}

	// /bin/sh, a shell command after a ";", a path traversal and the hard
	// values stay refused in q on paths near the searches, and in another
	// field on them.
	for _, target := range []string{
		"/org/acme/teams?q=",
		"/org/acme/teams/-/search/x?q=",
		"/org/acme/issues/developers/x?q=",
		"/alice/-/projects/1?q=",
		"/alice/-/packages/generic/tools-cli/1.0.0?q=",
		"/alice/tools/milestones/new?q=",
		"/alice/tools/milestone/new?q=",
		"/alice/tools/projects/1?q=",
		"/alice/tools/packages/x?q=",
		"/alice/tools/tags/v1.0?q=",
		"/alice/tools/issues/1?q=",
		"/alice/tools/settings/actions?q=",
		"/user/settings/actions?q=",
		"/-/admin/users/1/edit?q=",
		"/-/admin/emails/delete?q=",
		"/-/admin/actions/runners/1?q=",
		"/-/admin/packages/delete?q=",
		"/alice/tools/milestones?other=",
		"/alice/tools/milestone/1?other=",
		"/alice/tools/tags?other=",
		"/org/acme/issues/developers?other=",
		"/org/acme/teams/-/search?other=",
		"/-/admin/users?other=",
		"/-/admin/packages?other=",
	} {
		for _, value := range slices.Concat(attacks, hardValues) {
			if got := inspect(t, crs, get(target+value)); got.Score < threshold {
				t.Errorf("%s%s let through: %+v", target, value, got)
			}
		}
	}

	// Under /admin/, where Forgejo serves its admin pages, only the hard
	// values are left out: the tenth change lets them through in q on every
	// path there, for Drupal.
	for _, target := range []string{
		"/admin/packages/delete?q=",
		"/admin/packages?other=",
	} {
		for _, value := range attacks {
			if got := inspect(t, crs, get(target+value)); got.Score < threshold {
				t.Errorf("%s%s let through: %+v", target, value, got)
			}
		}
	}

	// Referer keeps its rules on a page opened from a path near the
	// searches of one team and of one milestone.
	for _, path := range []string{
		"/org/acme/issues/developers/x",
		"/alice/tools/milestone/new",
	} {
		for _, attack := range []string{scriptTag, injection} {
			referer := "Referer: http://example.com" + path + "?q=" + attack
			if got := inspect(t, crs, get("/", referer)); got.Score < threshold {
				t.Errorf("%s let through: %+v", referer, got)
			}
		}
	}
}