Text to speech in reader mode added.

browser/speedreader/page_distiller.cc

@@ -46,6 +46,20 @@ void PageDistiller::GetDistilledText(DistillContentCallback callback) {
                               weak_factory_.GetWeakPtr(), std::move(callback)));
 }
 
+void PageDistiller::GetTextToSpeak(TextToSpeechContentCallback callback) {
+  if (state_ != State::kDistilled) {
+    return std::move(callback).Run(base::Value());
+  }
+
+  constexpr const char16_t kGetTextToSpeak[] = uR"js( extractTextToSpeak() )js";
+
+  web_contents_->GetPrimaryMainFrame()->ExecuteJavaScriptInIsolatedWorld(
+      kGetTextToSpeak,
+      base::BindOnce(&PageDistiller::OnGetTextToSpeak,
+                     weak_factory_.GetWeakPtr(), std::move(callback)),
+      ISOLATED_WORLD_ID_BRAVE_INTERNAL);
+}
+
 void PageDistiller::UpdateState(State state) {
   state_ = state;
   for (auto& observer : observers_) {
@@ -62,22 +76,17 @@ void PageDistiller::StartDistill(DistillContentCallback callback) {
     return std::move(callback).Run(false, {});
   }
 
-  if (state_ == State::kDistilled) {
-    constexpr const char16_t kScript[] = uR"js( extractText() )js";
-    web_contents_->GetPrimaryMainFrame()->ExecuteJavaScriptInIsolatedWorld(
-        kScript,
-        base::BindOnce(&PageDistiller::OnGetText, weak_factory_.GetWeakPtr(),
-                       std::move(callback)),
-        ISOLATED_WORLD_ID_BRAVE_INTERNAL);
-  } else {
-    constexpr const char16_t kScript[] =
-        uR"js( document.documentElement.outerHTML )js";
-    web_contents_->GetPrimaryMainFrame()->ExecuteJavaScriptInIsolatedWorld(
-        kScript,
-        base::BindOnce(&PageDistiller::OnGetOuterHTML,
-                       weak_factory_.GetWeakPtr(), std::move(callback)),
-        ISOLATED_WORLD_ID_BRAVE_INTERNAL);
-  }
+  constexpr const char16_t kGetDocumentSource[] =
+      uR"js( document.documentElement.outerHTML )js";
+
+  constexpr const char16_t kGetBodySource[] =
+      uR"js( document.body.outerHTML )js";
+
+  web_contents_->GetPrimaryMainFrame()->ExecuteJavaScriptInIsolatedWorld(
+      (state_ != State::kDistilled) ? kGetDocumentSource : kGetBodySource,
+      base::BindOnce(&PageDistiller::OnGetOuterHTML, weak_factory_.GetWeakPtr(),
+                     std::move(callback)),
+      ISOLATED_WORLD_ID_BRAVE_INTERNAL);
 }
 
 void PageDistiller::OnGetOuterHTML(DistillContentCallback callback,
@@ -106,14 +115,12 @@ void PageDistiller::OnGetOuterHTML(DistillContentCallback callback,
   }
 }
 
-void PageDistiller::OnGetText(DistillContentCallback callback,
-                              base::Value result) {
-  if (!web_contents_ || !result.is_dict() ||
-      !result.GetDict().FindString("content")) {
-    return std::move(callback).Run(false, {});
+void PageDistiller::OnGetTextToSpeak(TextToSpeechContentCallback callback,
+                                     base::Value result) {
+  if (!result.is_dict()) {
+    return std::move(callback).Run(base::Value());
   }
-  std::move(callback).Run(true,
-                          std::move(*result.GetDict().FindString("content")));
+  std::move(callback).Run(std::move(result));
 }
 
 void PageDistiller::OnPageDistilled(DistillContentCallback callback,
@@ -149,6 +156,7 @@ void PageDistiller::ExtractText(DistillContentCallback callback,
     return std::move(callback).Run(false, {});
   }
 
+  re2::RE2::GlobalReplace(&html_content, "<[^>]*>", " ");
   std::move(callback).Run(true, html_content);
 }
 

browser/speedreader/page_distiller.h

@@ -41,6 +41,7 @@ class PageDistiller {
 
   using DistillContentCallback =
       base::OnceCallback<void(bool success, std::string content)>;
+  using TextToSpeechContentCallback = base::OnceCallback<void(base::Value)>;
 
   State GetState() const;
 
@@ -49,6 +50,7 @@ class PageDistiller {
 
   void GetDistilledHTML(DistillContentCallback callback);
   void GetDistilledText(DistillContentCallback callback);
+  void GetTextToSpeak(TextToSpeechContentCallback callback);
 
  protected:
   explicit PageDistiller(content::WebContents* web_contents);
@@ -60,7 +62,8 @@ class PageDistiller {
  private:
   void StartDistill(DistillContentCallback callback);
   void OnGetOuterHTML(DistillContentCallback callback, base::Value result);
-  void OnGetText(DistillContentCallback callback, base::Value result);
+  void OnGetTextToSpeak(TextToSpeechContentCallback callback,
+                        base::Value result);
   void OnPageDistilled(DistillContentCallback callback,
                        DistillationResult result,
                        std::string original_data,

browser/speedreader/speedreader_tab_helper.cc

@@ -594,14 +594,14 @@ void SpeedreaderTabHelper::SetDocumentAttribute(const std::string& attribute,
 
 void SpeedreaderTabHelper::OnGetDocumentSource(bool success, std::string html) {
   DCHECK(single_shot_next_request_);
-  if (!success) {
+  if (!success || html.empty()) {
     // TODO(boocmp): Show error dialog [Distillation failed on this page].
     SetNextRequestState(DistillState::kPageProbablyReadable);
     UpdateUI();
     return;
   }
 
-  single_show_content_.swap(html);
+  single_show_content_ = std::move(html);
   ReloadContents();
 }
 

browser/ui/webui/speedreader/speedreader_toolbar_data_handler_impl.cc

@@ -8,6 +8,7 @@
 #include <memory>
 #include <utility>
 
+#include "base/strings/string_number_conversions.h"
 #include "brave/browser/speedreader/speedreader_service_factory.h"
 #include "brave/browser/speedreader/speedreader_tab_helper.h"
 #include "brave/browser/ui/brave_browser_window.h"
@@ -22,6 +23,7 @@
 #include "chrome/browser/ui/browser_window.h"
 #include "chrome/browser/ui/color/chrome_color_id.h"
 #include "chrome/browser/ui/tabs/tab_strip_model.h"
+#include "chrome/common/chrome_isolated_world_ids.h"
 #include "ui/color/color_provider.h"
 
 namespace {
@@ -31,14 +33,13 @@ class TtsPlayerDelegate : public speedreader::TtsPlayer::Delegate {
 
   void RequestReadingContent(
       content::WebContents* web_contents,
-      base::OnceCallback<void(bool success, std::string content)> result_cb)
-      override {
+      base::OnceCallback<void(base::Value content)> result_cb) override {
     auto* page_distiller =
         speedreader::SpeedreaderTabHelper::GetPageDistiller(web_contents);
     if (page_distiller) {
-      page_distiller->GetDistilledText(std::move(result_cb));
+      page_distiller->GetTextToSpeak(std::move(result_cb));
     } else {
-      std::move(result_cb).Run(false, {});
+      std::move(result_cb).Run(base::Value());
     }
   }
 };
@@ -232,9 +233,24 @@ void SpeedreaderToolbarDataHandlerImpl::OnReadingStop(
 
 void SpeedreaderToolbarDataHandlerImpl::OnReadingProgress(
     content::WebContents* web_contents,
-    const std::string& element_id,
+    int paragraph_index,
     int char_index,
-    int length) {}
+    int length) {
+  if (!web_contents) {
+    return;
+  }
+
+  constexpr const char16_t kHighlight[] = uR"js( highlightText($1, $2, $3) )js";
+
+  const auto script = base::ReplaceStringPlaceholders(
+      kHighlight,
+      {base::NumberToString16(paragraph_index),
+       base::NumberToString16(char_index), base::NumberToString16(length)},
+      nullptr);
+
+  web_contents->GetPrimaryMainFrame()->ExecuteJavaScriptInIsolatedWorld(
+      script, base::DoNothing(), ISOLATED_WORLD_ID_BRAVE_INTERNAL);
+}
 
 void SpeedreaderToolbarDataHandlerImpl::OnTabStripModelChanged(
     TabStripModel* tab_strip_model,

browser/ui/webui/speedreader/speedreader_toolbar_data_handler_impl.h

@@ -6,8 +6,6 @@
 #ifndef BRAVE_BROWSER_UI_WEBUI_SPEEDREADER_SPEEDREADER_TOOLBAR_DATA_HANDLER_IMPL_H_
 #define BRAVE_BROWSER_UI_WEBUI_SPEEDREADER_SPEEDREADER_TOOLBAR_DATA_HANDLER_IMPL_H_
 
-#include <string>
-
 #include "base/scoped_observation.h"
 #include "brave/components/speedreader/common/speedreader_toolbar.mojom.h"
 #include "brave/components/speedreader/speedreader_service.h"
@@ -94,7 +92,7 @@ class SpeedreaderToolbarDataHandlerImpl
   void OnReadingStart(content::WebContents* web_contents) override;
   void OnReadingStop(content::WebContents* web_contents) override;
   void OnReadingProgress(content::WebContents* web_contents,
-                         const std::string& element_id,
+                         int paragraph_index,
                          int char_index,
                          int length) override;
 

components/speedreader/resources/panel/components/lists/index.tsx

@@ -30,7 +30,6 @@ const mainButtonsOptions = [
     id: 'tts',
     type: MainButtonType.TextToSpeech,
     iconName: 'headphones',
-    hidden: true,  // TODO(boocmp): Enable in future PR.
     title: getLocale('braveReaderModeTextToSpeech')
   },
   {

components/speedreader/resources/speedreader-desktop.css

@@ -548,6 +548,7 @@ iframe {
 
 html,
 html[data-theme='light'] {
+  position: relative; !important;
   background-color: var(--background-color);
 
   --article-background-color: #FFFFFF;
@@ -643,4 +644,33 @@ html[data-font-family='dyslexic'] {
 
 html[data-content-style='text-only'] img {
   display: none !important;
-}
+}
+
+.tts-highlighted, .tts-highlighted * {
+  position: relative;
+}
+
+@keyframes tts-fade-in {
+  0% { opacity: 0; }
+  100% { opacity: 1; }
+}
+
+.tts-highlighted::after {
+  animation: tts-fade-in 0.75s;
+
+  content: '';
+  background: linear-gradient(90deg, rgba(168, 168, 168, 0.5), rgba(128, 128, 128, 0.1));
+  mix-blend-mode: luminosity;
+
+  background-size: 100%;
+  border-radius: 10px;
+
+  position: absolute;
+
+  left: -1.5rem;
+  right: -1.5rem;
+  top: -0.5rem;
+  bottom: -0.5rem;
+
+  pointer-events: none;
+}

components/speedreader/resources/speedreader-desktop.js

@@ -45,12 +45,61 @@ const defaultSpeedreaderData = {
     minutesText: 'min. read',
 }
 
-const extractText = () => {
+const extractTextToSpeak = () => {
+    const textTags = ['P', 'DIV', 'MAIN', 'ARTICLE']
+
+    const extractParagraphs = (node) => {
+        let paragraphs = []
+        if (!node) {
+            return paragraphs
+        }
+        for (const child of node.children) {
+            if (textTags.indexOf(child.tagName) >= 0) {
+                const childParagraphs = extractParagraphs(child)
+                if (childParagraphs.length == 0) {
+                    paragraphs.push(child)
+                } else {
+                    paragraphs = paragraphs.concat(childParagraphs)
+                }
+            }
+        }
+        return paragraphs
+    }
+
+    const paragraphs = extractParagraphs($(contentDivId))
+
+    const textToSpeak = []
+    for (const p of paragraphs) {
+        const text = p.innerText.replace(/\n|\r +/g, ' ').trim()
+        if (text) {
+            p.setAttribute('tts-paragraph-index', textToSpeak.length)
+            textToSpeak.push(p.innerText.replace(/\n|\r +/g, ' '))
+        }
+    }
+
     return {
         title: document.title,
         author: $(metaDataDivId)?.querySelector('.author')?.textContent,
         desciption: $(metaDataDivId)?.querySelector('.subhead')?.textContent,
-        content: $(contentDivId)?.innerText.replace(/\n|\r +/g, ' ')
+        paragraphs: textToSpeak
+    }
+}
+
+const highlightText = (ttsParagraphIndex, charIndex, length) => {
+    document.querySelectorAll('.tts-highlighted').forEach((e) => {
+        if (e.getAttribute('tts-paragraph-index') != ttsParagraphIndex) {
+            e.classList.remove('tts-highlighted')
+        }
+    })
+
+    const paragraph = document.querySelector(
+        '[tts-paragraph-index="' + ttsParagraphIndex + '"]')
+    if (paragraph) {
+        paragraph.classList.add('tts-highlighted')
+        const ttsContent = document.createElement('tts-content')
+        ttsContent.append(...paragraph.childNodes)
+        paragraph.append(ttsContent)
+        document.documentElement.style.setProperty('--tts-highlight-progress', (charIndex / paragraph.textContent.length) * 100 + '%')
     }
 }