fix: URL patterns should not be included in the sitemap

apps/builder/app/builder/shared/url-pattern.ts

@@ -1,4 +1,6 @@
+import { matchPathnameParams } from "@webstudio-is/sdk";
 import { URLPattern } from "urlpattern-polyfill";
+export { isPathnamePattern } from "@webstudio-is/sdk";
 
 const baseUrl = "http://url";
 
@@ -21,23 +23,12 @@ type Token =
   | { type: "fragment"; value: string }
   | { type: "param"; name: string; optional: boolean; splat: boolean };
 
-// /:slug -> { name: "slug", modifier: "" }
-// /:slug* -> { name: "slug", modifier: "*" }
-// /:slug? -> { name: "slug", modifier: "?" }
-// /* -> { wildcard: "*" }
-const tokenRegex = /:(?<name>\w+)(?<modifier>[?*]?)|(?<wildcard>(?<!:\w+)\*)/;
-// use separate regex from matchAll because regex.test is stateful when used with g flag
-const tokenRegexGlobal = new RegExp(tokenRegex.source, "g");
-
-export const isPathnamePattern = (pathname: string) =>
-  tokenRegex.test(pathname);
-
 export const tokenizePathnamePattern = (pathname: string) => {
   const tokens: Token[] = [];
   let lastCursor = 0;
   let lastWildcard = -1;
 
-  for (const match of pathname.matchAll(tokenRegexGlobal)) {
+  for (const match of matchPathnameParams(pathname)) {
     const cursor = match.index ?? 0;
     if (lastCursor < cursor) {
       tokens.push({

packages/sdk/src/index.ts

@@ -19,3 +19,4 @@ export * from "./expression";
 export * from "./forms-generator";
 export * from "./resources-generator";
 export * from "./page-meta-generator";
+export * from "./url-pattern";

packages/sdk/src/page-utils.ts

@@ -1,5 +1,6 @@
 import { executeExpression } from "./expression";
 import type { Folder, Page, Pages } from "./schema/pages";
+import { isPathnamePattern } from "./url-pattern";
 
 export const ROOT_FOLDER_ID = "root";
 
@@ -85,6 +86,7 @@ export const getStaticSiteMapXml = (pages: Pages, updatedAt: string) => {
       .filter(
         (page) => executeExpression(page.meta.excludePageFromSearch) !== true
       )
+      .filter((page) => false === isPathnamePattern(page.path))
       .map((page) => ({
         path: getPagePath(page.id, pages),
         lastModified: updatedAt.split("T")[0],

packages/sdk/src/url-pattern.test.ts

@@ -0,0 +1,14 @@
+import { expect, test } from "@jest/globals";
+import { isPathnamePattern } from "./url-pattern";
+
+test("check pathname is pattern", () => {
+  expect(isPathnamePattern("/:name")).toEqual(true);
+  expect(isPathnamePattern("/:slug*")).toEqual(true);
+  expect(isPathnamePattern("/:id?")).toEqual(true);
+  expect(isPathnamePattern("/*")).toEqual(true);
+
+  expect(isPathnamePattern("")).toEqual(false);
+  expect(isPathnamePattern("/")).toEqual(false);
+  expect(isPathnamePattern("/blog")).toEqual(false);
+  expect(isPathnamePattern("/blog/post-name")).toEqual(false);
+});

packages/sdk/src/url-pattern.ts

@@ -0,0 +1,15 @@
+// /:slug -> { name: "slug", modifier: "" }
+// /:slug* -> { name: "slug", modifier: "*" }
+// /:slug? -> { name: "slug", modifier: "?" }
+// /* -> { wildcard: "*" }
+const tokenRegex = /:(?<name>\w+)(?<modifier>[?*]?)|(?<wildcard>(?<!:\w+)\*)/;
+
+export const isPathnamePattern = (pathname: string) =>
+  tokenRegex.test(pathname);
+
+// use separate regex from matchAll because regex.test is stateful when used with g flag
+const tokenRegexGlobal = new RegExp(tokenRegex.source, "g");
+
+export const matchPathnameParams = (pathname: string) => {
+  return pathname.matchAll(tokenRegexGlobal);
+};