Bedrock support: Use home URLs instead of site URL or paths

CHANGELOG.md

@@ -21,6 +21,9 @@
  - [#837](https://github.com/leonstafford/wp2static/pull/837): Require PHP 7.4 or later; bump dependencies @leonstafford
  - [#811](https://github.com/leonstafford/wp2static/issues/811): Optimize FilesHelper::getListOfLocalFilesByDir @bookwyrm
  - [#805](https://github.com/leonstafford/wp2static/issues/805): Fix warning on cache page when there are no deployment namespaces @john-shaffer
+ - [#840](https://github.com/leonstafford/wp2static/pull/840): Add preliminary support for [Bedrock](https://roots.io/bedrock/) by using home URL instead of site URL in several places. This is a potentially breaking change for anyone with `WP_HOME` differing from `WP_SITEURL`. @oriolarcas, @john-shaffer
+   - Use `wp2static_set_wordpress_home_url` filter in post-processing instead of `wp2static_set_wordpress_site_url`.
+   - Add integration tests for Bedrock.
 
 ## WP2Static 7.1.7 (2021-09-04)
 

integration-tests/clj/wp2static_test/main.clj

@@ -92,9 +92,18 @@
                              (str "bedrock.sh exited with code " code)
                              {:code code}))))
                      bedrock)
-     :name "Bedrock Initializer"
+     :name "Bedrock"
      :join? true
-     :open-f (fn [_] (popen ["bash" "bedrock.sh"]))}))
+     :open-f (fn [_] (popen ["bash" "bedrock.sh"]))
+
+     :features {:sitemaps? false}
+     :paths {:cli "bedrock/web/wp"
+             :doc-root "bedrock/web"
+             :plugins "bedrock/web/app/plugins"
+             :site "/wp/"
+             :uploads "bedrock/web/app/uploads"
+             :wp-content "bedrock/web/wp/wp-content"}
+     :urls {:home "http://localhost:7001"}}))
 
 (defn wordpress []
   (shell-process
@@ -106,9 +115,18 @@
                              (str "wordpress.sh exited with code " code)
                              {:code code}))))
                      wordpress)
-     :name "WordPress Initializer"
+     :name "WordPress"
      :join? true
-     :open-f (fn [_] (popen ["bash" "wordpress.sh"]))}))
+     :open-f (fn [_] (popen ["bash" "wordpress.sh"]))
+
+     :features {:sitemaps? true}
+     :paths {:cli "wordpress"
+             :doc-root "wordpress"
+             :plugins "wordpress/wp-content/plugins"
+             :site "/"
+             :uploads "wordpress/wp-content/uploads"
+             :wp-content "wordpress/wp-content"}
+     :urls {:home "http://localhost:7000"}}))
 
 (defn system-map []
   (component/system-map
@@ -117,6 +135,7 @@
     :nginx (component/using (nginx) [:wordpress])
     :php-fpm (component/using (php-fpm) [:mariadb])
     :wordpress (component/using (wordpress) [:mariadb :php-fpm])
+    :wordpresses (component/using {} [:bedrock :wordpress])
     :wp2static (component/using (wp2static) [:bedrock :wordpress])))
 
 (defonce system (atom nil))

integration-tests/clj/wp2static_test/test.clj

@@ -29,7 +29,7 @@
     result))
 
 (defn wp-cli! [opts & args]
-  (apply sh! opts "wp" (concat args ["--path=wordpress"])))
+  (apply sh! opts "wp" (concat args [(str "--path=" (:path opts "wordpress"))])))
 
 (defn test [opts]
   ;; https://clojureverse.org/t/why-doesnt-my-program-exit/3754/8
@@ -42,3 +42,8 @@
   `(let [~name (main/start!)]
      (core/clean-wp2static-cache!)
      ~@body))
+
+(defmacro testing [[wp string] & body]
+  `(clojure.test/testing (str "[" (:name ~wp) "] " ~string)
+     ~@body))
+

integration-tests/test/wp2static_test/crawl_test.clj

@@ -3,12 +3,17 @@
             [clojure.test :refer :all]
             [wp2static-test.test :as test]))
 
-(defn get-crawled-file [path]
-  (slurp (str "wordpress/wp-content/uploads/wp2static-crawled-site/" path)))
+(defn get-crawled-file [wp path]
+  (slurp (str (get-in wp [:paths :uploads]) "/wp2static-crawled-site/" path)))
 
 (deftest test-crawled-site
-  (test/with-test-system [_]
-    (test/wp-cli! {} "wp2static" "detect")
-    (test/wp-cli! {} "wp2static" "crawl")
-    (is (str/includes? (get-crawled-file "index.html") "Welcome to WordPress"))
-    (is (str/includes? (get-crawled-file "robots.txt") "Sitemap: http://localhost:7000/wp-sitemap.xml"))))
+  (test/with-test-system [system {}]
+    (doseq [wp (vals (:wordpresses system))
+            :let [wp-cli! #(apply test/wp-cli! {:path (get-in wp [:paths :cli])} %&)]]
+      (test/testing [wp "Crawling works"]
+        (wp-cli! "wp2static" "detect")
+        (wp-cli! "wp2static" "crawl")
+        (is (str/includes? (get-crawled-file wp "index.html")
+              "Welcome to WordPress"))
+        (is (str/includes? (get-crawled-file wp "robots.txt")
+              (str "Disallow: " (get-in wp [:paths :site]) "wp-admin/")))))))

integration-tests/test/wp2static_test/detect_test.clj

@@ -3,8 +3,16 @@
             [clojure.test :refer :all]
             [wp2static-test.test :as test]))
 
-(defn get-crawled-file [path]
-  (slurp (str "wordpress/wp-content/uploads/wp2static-crawled-site/" path)))
+(defn get-crawled-file [wp path]
+  (slurp (str (get-in wp [:paths :uploads]) "/wp2static-crawled-site/" path)))
+
+(defmacro with-robots-txt [wp s & body]
+  `(let [path# (str (get-in ~wp [:paths :doc-root]) "/robots.txt")]
+     (try
+       (spit path# ~s)
+       (do ~@body)
+       (finally
+         (test/sh! {} "rm" "-f" path#)))))
 
 (def robots-sitemap-404
   "User-agent: *
@@ -20,16 +28,21 @@ Sitemap: http://localhost:7000/does-not-exist.xml")
 <sitemapindex xmlns=\"http://www.sitemaps.org/schemas/sitemap/0.9\"><sitemap><loc>http://localhost:7000/does-not-exist.xml</loc></sitemap></sitemapindex>")
 
 (deftest test-robots-404
-  (testing "robots.txt sitemap URLs that return 404s are ignored"
-    (test/with-test-system [_]
-      (try
-        (spit "wordpress/robots.txt" robots-sitemap-404)
-        (spit "wordpress/wp-content/sitemap.xml" sitemap-with-404)
-        (is (zero? (:exit (test/wp-cli!
-                            {:expect-warnings {#".*Got 404 for sitemap.*" 1}}
-                            "wp2static" "detect"))))
-        (finally
-          (test/sh! {} "rm" "wordpress/robots.txt" "wordpress/wp-content/sitemap.xml"))))))
+  (test/with-test-system [system {}]
+    (doseq [wp (vals (:wordpresses system))
+            :let [wp-cli! #(apply test/wp-cli! {:path (get-in wp [:paths :cli])} %&)]]
+      (when (get-in wp [:features :sitemaps?])
+        (test/testing [wp "robots.txt sitemap URLs that return 404s are ignored"]
+          (with-robots-txt wp robots-sitemap-404
+            (let [sitemap-path (str (get-in wp [:paths :wp-content]) "/sitemap.xml")]
+              (try
+                (spit sitemap-path sitemap-with-404)
+                (is (zero? (:exit (test/wp-cli!
+                                    {:expect-warnings {#".*Got 404 for sitemap.*" 1}
+                                     :path (get-in wp [:paths :cli])}
+                                    "wp2static" "detect"))))
+                (finally
+                  (test/sh! {} "rm" "-f" sitemap-path))))))))))
 
 (def robots-sitemap-slashes
   "User-agent: *
@@ -39,12 +52,13 @@ Allow: /wp-admin/admin-ajax.php
 Sitemap: http://localhost:7000//wp-sitemap.xml")
 
 (deftest test-robots-sitemap-slashes
-  (testing "robots.txt sitemap URLs with double slashes are processed"
-    (test/with-test-system [_]
-      (try
-        (spit "wordpress/robots.txt" robots-sitemap-slashes)
-        (test/wp-cli! {} "wp2static" "detect")
-        (test/wp-cli! {} "wp2static" "crawl")
-        (is (str/includes? (get-crawled-file "wp-sitemap-posts-post-1.xml") "http://localhost:7000/hello-world/"))
-        (finally
-          (test/sh! {} "rm" "wordpress/robots.txt"))))))
+  (test/with-test-system [system {}]
+    (doseq [wp (vals (:wordpresses system))
+            :let [wp-cli! #(apply test/wp-cli! {:path (get-in wp [:paths :cli])} %&)]]
+      (when (get-in wp [:features :sitemaps?])
+        (test/testing [wp "robots.txt sitemap URLs with double slashes are processed"]
+          (with-robots-txt wp robots-sitemap-slashes
+            (wp-cli! "wp2static" "detect")
+            (wp-cli! "wp2static" "crawl")
+            (is (str/includes? (get-crawled-file wp "wp-sitemap-posts-post-1.xml")
+                  (str (get-in wp [:paths :home] "hello-world/"))))))))))

integration-tests/test/wp2static_test/options_test.clj

@@ -20,19 +20,24 @@ function processQueueImmediately_filter ( $val ) {
 add_filter( 'wp2static_option_processQueueImmediately', 'processQueueImmediately_filter' );")
 
 (deftest test-option-filters
-  (test/with-test-system [_]
-    (testing "wp2static_option_* filters work"
-      (is (= "0" (-> (test/wp-cli! {} "wp2static" "options" "get" "processQueueImmediately")
-                   :out
-                   str/trim)))
-      (try
-        (test/sh! {} "mkdir" "wordpress/wp-content/plugins/options-test")
-        (spit "wordpress/wp-content/plugins/options-test/options-test.php"
-          options-test-filters)
-        (test/wp-cli! {} "plugin" "activate" "options-test")
-        (is (= "1" (-> (test/wp-cli! {} "wp2static" "options" "get" "processQueueImmediately")
+  (test/with-test-system [system {}]
+    (doseq [wp (vals (:wordpresses system))
+            :let [wp-cli! #(apply test/wp-cli! {:path (get-in wp [:paths :cli])} %&)
+                  plugins-dir (get-in wp [:paths :plugins])]]
+      (test/testing [wp "wp2static_option_* filters work"]
+        (is (= "0" (-> (test/wp-cli! {} "wp2static" "options" "get" "processQueueImmediately")
                      :out
                      str/trim)))
-        (test/wp-cli! {} "plugin" "deactivate" "options-test")
-        (finally
-          (test/sh! {} "rm" "-rf" "wordpress/wp-content/plugins/options-test"))))))
+        (try
+          (test/sh! {} "mkdir" "options-test"
+            :dir plugins-dir)
+          (spit (str plugins-dir "/options-test/options-test.php")
+            options-test-filters)
+          (wp-cli! "plugin" "activate" "options-test")
+          (is (= "1" (-> (wp-cli! "wp2static" "options" "get" "processQueueImmediately")
+                       :out
+                       str/trim)))
+          (wp-cli! "plugin" "deactivate" "options-test")
+          (finally
+            (test/sh! {} "rm" "-rf" "options-test"
+              :dir plugins-dir)))))))

integration-tests/test/wp2static_test/post_process_test.clj

@@ -3,16 +3,20 @@
             [clojure.test :refer :all]
             [wp2static-test.test :as test]))
 
-(defn get-processed-file [path]
-  (slurp (str "wordpress/wp-content/uploads/wp2static-processed-site/" path)))
+(defn get-processed-file [wp path]
+  (slurp (str (get-in wp [:paths :uploads]) "/wp2static-processed-site/" path)))
 
 (deftest test-processed-site
-  (test/with-test-system [_]
-    (test/wp-cli! {} "wp2static" "detect")
-    (test/wp-cli! {} "wp2static" "crawl")
-    (test/wp-cli! {} "wp2static" "post_process")
-    (let [index (get-processed-file "index.html")]
-      (is (str/includes? index "Welcome to WordPress"))
-      (testing "Rewrites work"
-        (is (str/includes? index "<a href=\"https://example.com/hello-world/\">"))      
-        (is (str/includes? (get-processed-file "robots.txt") "Sitemap: https://example.com/wp-sitemap.xml"))))))
+  (test/with-test-system [system {}]
+    (doseq [wp (vals (:wordpresses system))
+            :let [wp-cli! #(apply test/wp-cli! {:path (get-in wp [:paths :cli])} %&)]]
+      (test/testing [wp "Post-processing works"]
+        (wp-cli! "wp2static" "detect")
+        (wp-cli! "wp2static" "crawl")
+        (wp-cli! "wp2static" "post_process")
+        (let [index (get-processed-file wp "index.html")]
+          (is (str/includes? index "Welcome to WordPress"))
+          (testing "& URL rewriting works"
+            (is (str/includes? index "<a href=\"https://example.com/hello-world/\">"))
+            (when (get-in wp [:features :sitemaps?])
+              (is (str/includes? (get-processed-file wp "robots.txt") "Sitemap: https://example.com/wp-sitemap.xml")))))))))

src/Crawler.php

@@ -113,7 +113,9 @@ public function crawlSite( string $static_site_path ) : void {
         $urls = [];
 
         foreach ( $crawlable_paths as $root_relative_path ) {
-            $absolute_uri = new URL( $this->site_path . $root_relative_path );
+            $absolute_uri = new URL(
+                rtrim( SiteInfo::getURL( 'home' ), '/' ) . $root_relative_path
+            );
             $urls[] = [
                 'url' => $absolute_uri->get(),
                 'path' => $root_relative_path,

src/DetectSitemapsURLs.php

@@ -14,7 +14,7 @@ class DetectSitemapsURLs {
      * @return string[] list of URLs
      * @throws WP2StaticException
      */
-    public static function detect( string $wp_site_url ) : array {
+    public static function detect() : array {
         $sitemaps_urls = [];
         $parser = new SitemapParser(
             'WP2Static.com',
@@ -27,14 +27,13 @@ public static function detect( string $wp_site_url ) : array {
             ]
         );
 
-        $site_path = rtrim( SiteInfo::getURL( 'site' ), '/' );
-
         $port_override = apply_filters(
             'wp2static_curl_port',
             null
         );
 
-        $base_uri = $site_path;
+        $wp_site_url = SiteInfo::getURL( 'home' );
+        $base_uri = rtrim( $wp_site_url, '/' );
 
         if ( $port_override ) {
             $base_uri = "{$base_uri}:{$port_override}";

src/DetectThemeAssets.php

@@ -43,11 +43,18 @@ public static function detect( string $theme_type ) : array {
 
                 $detected_filename =
                     str_replace(
-                        $site_path,
-                        '/',
+                        $template_path,
+                        $template_url,
                         $filename
                     );
 
+                $detected_filename =
+                    str_replace(
+                        get_home_url(),
+                        '/',
+                        $detected_filename
+                    );
+
                 if ( $path_crawlable ) {
                     if ( is_string( $detected_filename ) ) {
                         array_push(

src/FilesHelper.php

@@ -39,21 +39,17 @@ public static function deleteDirWithFiles( string $dir ) : void {
      * Get public URLs for all files in a local directory.
      *
      * @param string $dir
+     * @param string $base_url
      * @param array<string> $filenames_to_ignore
      * @param array<string> $file_extensions_to_ignore
      * @return string[] list of relative, urlencoded URLs
      */
     public static function getListOfLocalFilesByDir(
         string $dir,
+        string $base_url,
         array $filenames_to_ignore,
         array $file_extensions_to_ignore
     ) : array {
-        $site_path = SiteInfo::getPath( 'site' );
-
-        if ( ! is_string( $site_path ) ) {
-            return [];
-        }
-
         $files = [];
 
         if ( is_dir( $dir ) ) {
@@ -72,7 +68,19 @@ public static function getListOfLocalFilesByDir(
                 );
 
                 if ( $path_crawlable ) {
-                    $url = str_replace( $site_path, '/', $filename );
+                    $url =
+                        str_replace(
+                            $dir,
+                            $base_url,
+                            $filename
+                        );
+
+                    $url =
+                        str_replace(
+                            get_home_url(),
+                            '/',
+                            $url
+                        );
 
                     if ( is_string( $url ) ) {
                         $files[] = $url;

src/SimpleRewriter.php

@@ -50,21 +50,21 @@ public static function rewriteFileContents( string $file_contents ) : string
             CoreOptions::getValue( 'deploymentURL' )
         );
 
-        $wordpress_site_url = apply_filters(
-            'wp2static_set_wordpress_site_url',
-            untrailingslashit( SiteInfo::getUrl( 'site' ) )
+        $wordpress_home_url = apply_filters(
+            'wp2static_set_wordpress_home_url',
+            untrailingslashit( SiteInfo::getUrl( 'home' ) )
         );
 
-        $wordpress_site_url = untrailingslashit( $wordpress_site_url );
+        $wordpress_home_url = untrailingslashit( $wordpress_home_url );
         $destination_url = untrailingslashit( $destination_url );
         $destination_url_rel = URLHelper::getProtocolRelativeURL( $destination_url );
         $destination_url_rel_c = addcslashes( $destination_url_rel, '/' );
 
         $replacement_patterns = [
-            $wordpress_site_url => $destination_url,
-            URLHelper::getProtocolRelativeURL( $wordpress_site_url ) =>
+            $wordpress_home_url => $destination_url,
+            URLHelper::getProtocolRelativeURL( $wordpress_home_url ) =>
                 URLHelper::getProtocolRelativeURL( $destination_url ),
-            addcslashes( URLHelper::getProtocolRelativeURL( $wordpress_site_url ), '/' ) =>
+            addcslashes( URLHelper::getProtocolRelativeURL( $wordpress_home_url ), '/' ) =>
                 addcslashes( URLHelper::getProtocolRelativeURL( $destination_url ), '/' ),
         ];
 

src/URLDetector.php

@@ -82,6 +82,7 @@ public static function detectURLs() : string {
             $arrays_to_merge[] =
                 FilesHelper::getListOfLocalFilesByDir(
                     SiteInfo::getPath( 'uploads' ),
+                    SiteInfo::getUrl( 'uploads' ),
                     $filenames_to_ignore,
                     $file_extensions_to_ignore
                 );
@@ -90,7 +91,7 @@ public static function detectURLs() : string {
         $detect_sitemaps = apply_filters( 'wp2static_detect_sitemaps', 1 );
 
         if ( $detect_sitemaps ) {
-            $arrays_to_merge[] = DetectSitemapsURLs::detect( SiteInfo::getURL( 'site' ) );
+            $arrays_to_merge[] = DetectSitemapsURLs::detect();
         }
 
         $detect_parent_theme = apply_filters( 'wp2static_detect_parent_theme', 1 );