package coreruleset_test

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

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

// giteaAdminUnadoptedRepos and forgejoAdminUnadoptedRepos are the admin pages
// of unadopted repositories. The forms that adopt or delete one post there.
const (
	giteaAdminUnadoptedRepos   = "/-/admin/repos/unadopted"
	forgejoAdminUnadoptedRepos = "/admin/repos/unadopted"
)

// 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",
	giteaAdminUnadoptedRepos,
	"/-/admin/emails",
	"/-/admin/emails/activate",
	"/admin/actions/runners",
	"/admin/users",
	forgejoAdminUnadoptedRepos,
	"/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{giteaAdminUnadoptedRepos, forgejoAdminUnadoptedRepos} {
			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 TestGiteaFormsThatAdoptARepositoryLetItsNameThrough(t *testing.T) {
	t.Parallel()

	const threshold = 5 // SWW_WAF_CORE_RULE_SET_SCORE_THRESHOLD's default

	crs := readingBodies(t)

	// The admin forms send the repository's owner and name in id, and the
	// form in a user's settings only its name. robin/assets holds bin/as, a
	// shell path.
	ownerAndName := []string{
		"alice%2Fpython3-app",
		"alice%2Fpowershell-scripts",
		"alice%2F.profile",
		"robin%2Fassets",
	}

	for _, form := range []struct {
		path string
		ids  []string
	}{
		{giteaAdminUnadoptedRepos, ownerAndName},
		{forgejoAdminUnadoptedRepos, ownerAndName},
		{
			"/user/settings/repos/unadopted", // gitea's and Forgejo's
			[]string{"python3-app", "powershell-scripts", ".profile"},
		},
	} {
		r := request{http.MethodPost, form.path, []string{formData}}

		for _, action := range []string{"adopt", "delete"} {
			for _, id := range form.ids {
				body := "id=" + id + "&action=" + action
				wantBody(t, crs, r, body, coreruleset.Result{}, body)
			}
		}

		// Another field keeps those rules.
		body := "id=" + form.ids[0] + "&action=adopt&page=python3"
		wantBody(t, crs, r, body, matched(932260), body)

		// Attacks in id are still refused.
		for _, attack := range []string{
			sqlInjection,
			scriptInjection,
			commandInjection,
			"..%2F..%2Fetc%2Fpasswd",
		} {
			body := "id=" + attack + "&action=adopt"
			if got, _ := inspectBody(t, crs, r, body); got.Score < threshold {
				t.Errorf("POST %s %s let through: %+v", form.path, body, got)
			}
		}
	}

	// On another path, id keeps those rules.
	r := request{http.MethodPost, "/-/admin/repos", []string{formData}}
	body := "id=alice%2Fpython3-app&action=adopt&q=&page=1"
	wantBody(t, crs, r, body, matched(932260), body)
}

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

	const threshold = 5 // SWW_WAF_CORE_RULE_SET_SCORE_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_SCORE_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{binSh, 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)
			}
		}
	}
}

// giteaAPISearches are the API's searches that gitea and Forgejo share whose
// field skips the lists of system files and command names. Each ends in the
// field it sends its words in.
//
//nolint:gochecknoglobals // a constant cannot be a list
var giteaAPISearches = []string{
	"/api/v1/users/search?q=",
	"/api/v1/topics/search?q=",
	"/api/v1/orgs/acme/teams/search?q=",
	"/api/v1/admin/emails/search?q=",
	"/api/v1/admin/unadopted?query=",
	"/api/v1/repos/alice/tools/milestones?name=",
}

// giteaAPIPackageSearch is the API's search of an owner's packages, whose q
// also lets /bin/sh through, since a package name can hold a slash.
const giteaAPIPackageSearch = "/api/v1/packages/alice?q="

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

	checkSearchesRefuseAttacks(
		t,
		giteaAPISearches,
		[]string{giteaAPIPackageSearch},
	)
}

// checkSearchesRefuseAttacks checks that SQL injection, a path traversal,
// script injection and a shell command stay refused in the field of each
// search, in the query string and in a form data body. Each search ends in
// its field. /bin/sh stays refused too in the field of each of
// refusingBinSh, but not of lettingBinSh.
func checkSearchesRefuseAttacks(
	t *testing.T, refusingBinSh, lettingBinSh []string,
) {
	t.Helper()

	const threshold = 5 // SWW_WAF_CORE_RULE_SET_SCORE_THRESHOLD's default

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

	for _, search := range slices.Concat(refusingBinSh, lettingBinSh) {
		path, field, _ := strings.Cut(search, "?")

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

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

	for _, search := range refusingBinSh {
		if got := inspect(t, defaults, get(search+binSh)); got.Score < threshold {
			t.Errorf("%s%s let through: %+v", search, binSh, got)
		}
	}
}

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

	// Each search's field on paths near the searches, and another field on
	// them.
	checkValuesStayRefused(t, []string{
		"/api/v1/topics/search/x?q=",
		"/api/v1/orgs/acme/teams?q=",
		"/api/v1/orgs/acme/teams/search/x?q=",
		"/api/v1/admin/emails?q=",
		"/api/v1/admin/unadopted/alice/legacy-site?query=",
		"/api/v1/admin/users?query=",
		"/api/v1/repos/alice/tools/milestones/1?name=",
		"/api/v1/repos/alice/tools/labels?name=",
		"/api/v1/packages/alice/generic/tools-cli?q=",
		"/api/v1/topics/search?other=",
		"/api/v1/orgs/acme/teams/search?other=",
		"/api/v1/admin/emails/search?other=",
		"/api/v1/admin/unadopted?q=",
		"/api/v1/repos/alice/tools/milestones?q=",
		"/api/v1/packages/alice?other=",
	})
}

// checkValuesStayRefused checks that /bin/sh, a shell command after a ";", a
// path traversal and the hard values stay refused at the end of each target.
func checkValuesStayRefused(t *testing.T, targets []string) {
	t.Helper()

	const threshold = 5 // SWW_WAF_CORE_RULE_SET_SCORE_THRESHOLD's default

	crs := atDefaults(t)

	for _, target := range targets {
		for _, value := range slices.Concat(
			[]string{binSh, commandInjection, pathTraversal},
			hardValues,
		) {
			if got := inspect(t, crs, get(target+value)); got.Score < threshold {
				t.Errorf("%s%s let through: %+v", target, value, got)
			}
		}
	}
}

// giteaRegistrySearches are the package registries' searches whose field
// skips the lists of system files and command names. Each ends in the field
// it sends its words in.
//
//nolint:gochecknoglobals // a constant cannot be a list
var giteaRegistrySearches = []string{
	"/api/packages/alice/cargo/api/v1/crates?q=",
	"/api/packages/alice/chef/api/v1/search?q=",
	"/api/packages/alice/chef/api/v1/cookbooks?q=",
	"/api/packages/alice/nuget/query?q=",
	"/api/packages/alice/nuget/Search()?searchTerm=",
	"/api/packages/alice/nuget/Packages()?$filter=",
	"/api/packages/alice/nuget/Search()/$count?searchTerm=",
	"/api/packages/alice/nuget/Packages()/$count?$filter=",
}

// giteaRegistrySearchesLettingBinSh are the package registries' searches
// whose field also lets /bin/sh through, since a Composer, Conan or npm name
// can hold a slash.
//
//nolint:gochecknoglobals // a constant cannot be a list
var giteaRegistrySearchesLettingBinSh = []string{
	"/api/packages/alice/composer/search.json?q=",
	"/api/packages/alice/conan/v1/conans/search?q=",
	"/api/packages/alice/conan/v2/conans/search?q=",
	"/api/packages/alice/npm/-/v1/search?text=",
}

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

	checkSearchesRefuseAttacks(
		t,
		giteaRegistrySearches,
		giteaRegistrySearchesLettingBinSh,
	)
}

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

	// Each search's field on paths near the searches, and another field on
	// them.
	checkValuesStayRefused(t, []string{
		"/api/packages/alice/cargo/api/v1/crates/tools-cli?q=",
		"/api/packages/alice/chef/api/v1/cookbooks/tools?q=",
		"/api/packages/alice/composer/packages.json?q=",
		"/api/packages/alice/conan/v1/conans/tools/1.0/alice/stable/search?q=",
		"/api/packages/alice/nuget/query/x?q=",
		"/api/packages/alice/nuget/FindPackagesById()?id=",
		"/api/packages/alice/nuget/FindPackagesById()/$count?id=",
		"/api/packages/alice/nuget/Search()/$count/x?searchTerm=",
		"/api/packages/alice/npm/tools-cli?text=",
		"/api/packages/alice/cargo/api/v1/crates?other=",
		"/api/packages/alice/chef/api/v1/search?other=",
		"/api/packages/alice/composer/search.json?type=",
		"/api/packages/alice/conan/v2/conans/search?other=",
		"/api/packages/alice/nuget/query?other=",
		"/api/packages/alice/nuget/Search()?targetFramework=",
		"/api/packages/alice/nuget/Packages()?$orderby=",
		"/api/packages/alice/nuget/Search()/$count?targetFramework=",
		"/api/packages/alice/nuget/Packages()/$count?$orderby=",
		"/api/packages/alice/npm/-/v1/search?q=",
	})

	const threshold = 5 // SWW_WAF_CORE_RULE_SET_SCORE_THRESHOLD's default

	crs := atDefaults(t)

	// The rule for MongoDB operators (942290) still refuses the name of
	// another operator on NuGet's searches, and $filter on NuGet's other
	// paths. The rule for SQL injection (942100) still refuses tools/* in a
	// field other than a Conan search's, SQL injection in a second q after a
	// Conan pattern, SQL comments or dots that stand in for spaces, and SQL
	// injection holding + or --.
	for _, target := range []string{
		"/api/packages/alice/nuget/Search()?$where=1",
		"/api/packages/alice/nuget/Packages()?$ne=1",
		"/api/packages/alice/nuget/Search()/$count?$where=1",
		"/api/packages/alice/nuget/FindPackagesById()?$filter=IsLatestVersion",
		"/api/packages/alice/nuget/FindPackagesById()/$count?$filter=IsLatestVersion",
		"/api/packages/alice/composer/search.json?q=tools%2F%2A",
		"/api/packages/alice/conan/v2/conans/search?other=tools%2F%2A",
		"/api/packages/alice/conan/v2/conans/search?q=tools%2F%2A&q=" +
			sqlInjection,
		"/api/packages/alice/conan/v2/conans/search?q=1%2F%2A%2A%2For%2F%2A%2A%2F1",
		"/api/packages/alice/conan/v2/conans/search?q=1%2F%2A%40%2A%2For-1",
		"/api/packages/alice/conan/v2/conans/search?q=1%2Bor%2B1--",
		"/api/packages/alice/conan/v2/conans/search?q=1-or-1--",
		"/api/packages/alice/conan/v2/conans/search?q=x%2F1.or.1",
		"/api/packages/alice/conan/v2/conans/search?q=x%2F1.and.1",
		"/api/packages/alice/conan/v2/conans/search?q=1.having.1",
	} {
		if got := inspect(t, crs, get(target)); got.Score < threshold {
			t.Errorf("%s let through: %+v", target, got)
		}
	}

	// SQL injection in a q in the body is refused, even after a Conan
	// pattern in the query string.
	search := request{
		http.MethodPost,
		"/api/packages/alice/conan/v2/conans/search?q=zlib",
		[]string{formData},
	}

	body := "q=" + sqlInjection
	if got, _ := inspectBody(t, readingBodies(t), search, body); got.Score < threshold {
		t.Errorf("POST %s %s let through: %+v", search.target, body, got)
	}

	// So is SQL injection with dots for spaces, in a q sent only in the body.
	search.target = "/api/packages/alice/conan/v2/conans/search"

	body = "q=x%2F1.or.1"
	if got, _ := inspectBody(t, readingBodies(t), search, body); got.Score < threshold {
		t.Errorf("POST %s %s let through: %+v", search.target, body, got)
	}

	// The rule that lets a Conan pattern through is in phase 2, where it
	// checks tx.app_gitea itself: a hostname that does not serve gitea
	// refuses the pattern.
	pattern := get(
		"/api/packages/alice/conan/v2/conans/search?q=tools%2F%2A",
	)
	apps := servingApps(t, map[string][]string{
		giteaHost: {gitea}, otherHost: {wordpress},
	})
	wantResult(t, apps, withHost(pattern, giteaHost), coreruleset.Result{})
	wantResult(t, apps, withHost(pattern, otherHost), matched(942100))
}