HTTP server: add experimental DOM-dump mode

Simply dumps the contents of the DOM
browsh-org · Jun 19, 2019 · d6b5951 · d6b5951
1 parent 81f41b7
commit d6b5951
Show file tree

Hide file tree

Showing 7 changed files with 29 additions and 4 deletions.
diff --git a/interfacer/src/browsh/raw_text_server.go b/interfacer/src/browsh/raw_text_server.go
@@ -225,6 +225,7 @@ func isProductionHTTP(r *http.Request) bool {
 
 // 'PLAIN' mode returns raw text without any HTML whatsoever.
 // 'HTML' mode returns some basic HTML tags for things like anchor links.
+// 'DOM' mode returns a simple dump of the DOM.
 func getRawTextMode(r *http.Request) string {
 	var mode = "HTML"
 	if strings.Contains(r.Host, "text.") {
@@ -233,6 +234,9 @@ func getRawTextMode(r *http.Request) string {
 	if r.Header.Get("X-Browsh-Raw-Mode") == "PLAIN" {
 		mode = "PLAIN"
 	}
+	if r.Header.Get("X-Browsh-Raw-Mode") == "DOM" {
+		mode = "DOM"
+	}
 	return mode
 }
 

diff --git a/interfacer/src/browsh/version.go b/interfacer/src/browsh/version.go
@@ -1,3 +1,3 @@
 package browsh
 
-var browshVersion = "1.6.1"
+var browshVersion = "1.6.2"
diff --git a/interfacer/test/http-server/server_test.go b/interfacer/test/http-server/server_test.go
@@ -29,6 +29,12 @@ var _ = Describe("HTTP Server", func() {
 			"<a href=\"/http://localhost:4444/smorgasbord/another.html\">Another page</a>"))
 	})
 
+	It("should return the DOM", func() {
+		response := getPath("/smorgasbord", "dom")
+		Expect(response).To(ContainSubstring(
+			"<div class=\"big_middle\">"))
+	})
+
 	It("should return a background image", func() {
 		response := getPath("/smorgasbord", "html")
 		Expect(response).To(ContainSubstring("background-image: url(data:image/jpeg"))

diff --git a/interfacer/test/http-server/setup.go b/interfacer/test/http-server/setup.go
@@ -51,6 +51,9 @@ func getPath(path string, mode string) string {
 	if mode == "plain" {
 		request.Header.Add("X-Browsh-Raw-Mode", "PLAIN")
 	}
+	if mode == "dom" {
+		request.Header.Add("X-Browsh-Raw-Mode", "DOM")
+	}
 	response, err := client.Do(request)
 	if err != nil {
 		panic(fmt.Sprintf("%s", err))

diff --git a/webext/src/dom/commands_mixin.js b/webext/src/dom/commands_mixin.js
@@ -50,7 +50,7 @@ export default MixinBase =>
 
     _launch() {
       const mode = this.config.http_server_mode_type;
-      if (mode === "raw_text_plain" || mode === "raw_text_html") {
+      if (mode.includes("raw_text_")) {
         this._is_raw_text_mode = true;
         this._is_interactive_mode = false;
         this._raw_mode_type = mode;

diff --git a/webext/src/dom/serialise_mixin.js b/webext/src/dom/serialise_mixin.js
@@ -289,8 +289,14 @@ export default MixinBase =>
     }
 
     _sendRawText() {
+      let body;
+      if (this._raw_mode_type == "raw_text_dom") {
+        body = document.getElementsByTagName("body")[0].innerHTML;
+      } else {
+        body = this._serialiseRawText();
+      }
       let payload = {
-        body: this._serialiseRawText(),
+        body: body,
         page_load_duration: this.config.page_load_duration,
         parsing_duration: this._parsing_duration
       };

diff --git a/webext/src/dom/text_builder.js b/webext/src/dom/text_builder.js
@@ -30,7 +30,13 @@ export default class extends utils.mixins(CommonMixin, SerialiseMixin) {
   sendRawText(type) {
     this._raw_mode_type = type;
     this._parse_start_time = performance.now();
-    this.buildFormattedText(this._sendRawText.bind(this));
+    if (type == "raw_text_dom") {
+      setTimeout(() => {
+        this._sendRawText();
+      }, this.config["http-server"].render_delay);
+    } else {
+      this.buildFormattedText(this._sendRawText.bind(this));
+    }
   }
 
   buildFormattedText(callback) {