From dbf6670da5a78404c0b898259e9124c9040ddc98 Mon Sep 17 00:00:00 2001
From: Lukas Richter <53831711+Batorian@users.noreply.github.com>
Date: Mon, 29 Apr 2024 18:03:16 +0200
Subject: [PATCH] fix: NU chapter titles,update: ElloTL `parseChapter`  (#917)

* refactored and updated Novel Updates' parseChapter()

* added `ippotranslations` and `shanghaifantasy` to Novel Updates

* small changes

* removed switch cases and reverted back to `let` and `if`/`else if` structure

* added lots of translation group support

* added more site support and tried cleaning chapter content where possible

* quick fix

* quick fix

* logic overhaul; implementation of switch cases (again)

* small changes to `bloatClasses` and `chapterContent`

* added Anomalously Creative (`anotivereads`) to outliers and added new `bloatClasses` to Wuxia World (`wuxiaworld`)

* edited ElloTL (`ellotl.ts`) to match Novel Updates chapter content and slightly reduced title size in Novel Updates (`novelupdates.ts`) chapters; also added Novel Plex (`novelplex`) as supported site for Novel Updates

* ElloTL fix
---
 plugins/english/ellotl.ts       | 57 +++++++++++++++++++++------------
 plugins/english/novelupdates.ts | 38 ++++++++++++----------
 2 files changed, 58 insertions(+), 37 deletions(-)

diff --git a/plugins/english/ellotl.ts b/plugins/english/ellotl.ts
index ba99a79..7c058e2 100644
--- a/plugins/english/ellotl.ts
+++ b/plugins/english/ellotl.ts
@@ -273,50 +273,67 @@ class ElloTL implements Plugin.PluginBase {
     return fetch(this.site + chapterPath)
       .then(res => res.text())
       .then(html => {
+        let isParsingTitleBlock = false;
+        let isParsingTitle = false;
+        let isParsingSubtitle = false;
         let isParsingContent = false;
         let chapterContent = '';
         let notContent = false;
+
         const parser = new Parser({
           onopentag(name, attribs) {
-            if (attribs['class']?.includes('epcontent' || 'entry-content')) {
+            if (attribs['class']?.includes('epheader')) {
+              chapterContent += `<h2>`;
+              isParsingTitleBlock = true;
+              isParsingTitle = true;
+            }
+            if (isParsingSubtitle) {
+              chapterContent += ' | ';
+            }
+            if (attribs['class']?.includes('entry-content')) {
               isParsingContent = true;
-              chapterContent += '<' + name;
-              for (let attrib in attribs) {
-                chapterContent += ` ${attrib}="${attribs[attrib]}"`;
-              }
-              chapterContent += '>';
-            }
-            if (
-              attribs['class']?.includes(
-                'wp-block-buttons' || 'wp-block-spacer' || 'wp-element-button',
-              )
-            ) {
-              notContent = true;
-            }
-            if (name === 'span' && attribs['class']?.includes('maxbutton')) {
-              notContent = true;
             }
             if (isParsingContent && !notContent) {
-              chapterContent += '<' + name;
+              chapterContent += `<${name}`;
               for (let attrib in attribs) {
                 chapterContent += ` ${attrib}="${attribs[attrib]}"`;
               }
-              chapterContent += '>';
+              chapterContent += `>`;
+            }
+            if (attribs['class']?.includes('mb-center')) {
+              notContent = true;
             }
           },
           ontext(data) {
+            if (isParsingTitle) {
+              chapterContent += data;
+            }
+            if (isParsingSubtitle) {
+              chapterContent += data;
+            }
             if (isParsingContent && !notContent) {
               chapterContent += data;
             }
           },
           onclosetag(name) {
+            if (isParsingTitle && name === 'h1') {
+              isParsingTitle = false;
+              isParsingSubtitle = true;
+            }
+            if (isParsingSubtitle && name === 'div') {
+              isParsingSubtitle = false;
+            }
+            if (isParsingTitleBlock && !isParsingTitle && !isParsingSubtitle) {
+              chapterContent += `</h2><br>`;
+              isParsingTitleBlock = false;
+            }
             if (isParsingContent && !notContent) {
               chapterContent += `</${name}>`;
             }
-            if (notContent) {
+            if (notContent && name === 'span') {
               notContent = false;
             }
-            if (isParsingContent && !notContent && name === 'div') {
+            if (isParsingContent && name === 'div') {
               isParsingContent = false;
             }
           },
diff --git a/plugins/english/novelupdates.ts b/plugins/english/novelupdates.ts
index bb96a14..a3e3c6d 100644
--- a/plugins/english/novelupdates.ts
+++ b/plugins/english/novelupdates.ts
@@ -182,24 +182,24 @@ class NovelUpdates implements Plugin.PluginBase {
       case 'anotivereads':
         chapterTitle = loadedCheerio('#comic-nav-name').first().text()!;
         chapterContent = loadedCheerio('#spliced-comic').html()!;
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'asuratls':
         let titleElementAsura = loadedCheerio('.post-body div b').first();
         chapterTitle = titleElementAsura.first().text();
         titleElementAsura.remove();
         chapterContent = loadedCheerio('.post-body').html()!;
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'hiraethtranslation':
         chapterTitle = loadedCheerio('li.active').first().text()!;
         chapterContent = loadedCheerio('.text-left').html()!;
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'hostednovel':
         chapterTitle = loadedCheerio('#chapter-title').first().text()!;
         chapterContent = loadedCheerio('#chapter-content').html()!;
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'inoveltranslation':
         const link_inovel = 'https://api.' + url.slice(8);
@@ -212,13 +212,18 @@ class NovelUpdates implements Plugin.PluginBase {
             '<br><hr><br>TL Notes:<br>' +
             json_inovel.notes.replace(/\n/g, '<br>');
         }
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'ko-fi':
         chapterText = loadedCheerio('script:contains("shadowDom.innerHTML")')
           .html()
           ?.match(/shadowDom\.innerHTML \+= '(<div.*?)';/)![1]!;
         break;
+      case 'novelplex':
+        chapterTitle = loadedCheerio('.halChap--jud').first().text()!;
+        chapterContent = loadedCheerio('.halChap--kontenInner ').html()!;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
+        break;
       case 'novelworldtranslations':
         bloatClasses = ['.separator', 'p[dir="ltr"]'];
         bloatClasses.map(tag => loadedCheerio(tag).remove());
@@ -228,7 +233,7 @@ class NovelUpdates implements Plugin.PluginBase {
           return paragraph.replace(/&nbsp;/g, '').trim() + '<br>';
         });
         chapterContent = chapterContentArray.join('').trim();
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'raeitranslations':
         const parts = url.split('/');
@@ -246,7 +251,7 @@ class NovelUpdates implements Plugin.PluginBase {
           `<br><hr><br>Translator's Note:<br>` +
           json.currentChapter.note;
         chapterContent = chapterContent.replace(/\n/g, '<br>');
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'rainofsnow':
         loadedCheerio('.responsivevoice-button').remove();
@@ -257,12 +262,12 @@ class NovelUpdates implements Plugin.PluginBase {
         bloatClasses.map(tag => loadedCheerio(tag).remove());
         chapterTitle = loadedCheerio('.entry-title').first().text()!;
         chapterContent = loadedCheerio('.entry-content').html()!;
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'scribblehub':
         chapterTitle = loadedCheerio('.chapter-title').first().text()!;
         chapterContent = loadedCheerio('div.chp_raw').html()!;
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'stabbingwithasyringe':
         /**
@@ -281,7 +286,7 @@ class NovelUpdates implements Plugin.PluginBase {
           chapterTitle = titleElementSyringe.first().text();
           titleElementSyringe.remove();
           chapterContent = loadedCheerioSyringe('.entry-content').html()!;
-          chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+          chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         } else {
           chapterText = chapterContent;
         }
@@ -296,7 +301,7 @@ class NovelUpdates implements Plugin.PluginBase {
         chapterTitle = loadedCheerio('.title-content').first().text()!;
         loadedCheerio('.title-content').first().remove();
         chapterContent = loadedCheerio('.content').html()!;
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'tumblr':
         chapterText = loadedCheerio('.post').html()!;
@@ -304,7 +309,7 @@ class NovelUpdates implements Plugin.PluginBase {
       case 'wattpad':
         chapterTitle = loadedCheerio('.h2').first().text()!;
         chapterContent = loadedCheerio('.part-content pre').html()!;
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'webnovel':
         chapterTitle = loadedCheerio('.cha-tit .pr .dib').first().text()!;
@@ -312,14 +317,14 @@ class NovelUpdates implements Plugin.PluginBase {
         if (!chapterContent) {
           chapterContent = loadedCheerio('._content').html()!;
         }
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
       case 'wuxiaworld':
         bloatClasses = ['.MuiLink-root'];
         bloatClasses.map(tag => loadedCheerio(tag).remove());
         chapterTitle = loadedCheerio('h4 span').first().text()!;
         chapterContent = loadedCheerio('.chapter-content').html()!;
-        chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+        chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
         break;
     }
     return chapterText;
@@ -423,7 +428,7 @@ class NovelUpdates implements Plugin.PluginBase {
       chapterContent =
         loadedCheerio('.entry-content').html() ||
         loadedCheerio('.post-body').html()!;
-      chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+      chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
     } else if (isWordPress) {
       bloatClasses = [
         '.ad',
@@ -439,7 +444,6 @@ class NovelUpdates implements Plugin.PluginBase {
         '.sharedaddy',
         '.sidebar',
         '.wp-block-buttons',
-        '.wp-block-columns',
         '.wp-block-image',
         '.wp-dark-mode-switcher',
         '.wp-next-post-navi',
@@ -476,7 +480,7 @@ class NovelUpdates implements Plugin.PluginBase {
       if (chapterSubtitle) {
         chapterTitle = `${chapterTitle} | ${chapterSubtitle}`;
       }
-      chapterText = `<h1>${chapterTitle}</h1><br>${chapterContent}`;
+      chapterText = `<h2>${chapterTitle}</h2><br>${chapterContent}`;
     }
 
     if (!chapterText) {