From f23390ed01e14f9679f5b37bedd6edc38a9d4548 Mon Sep 17 00:00:00 2001
From: Jonas Kalderstam <jonas@cowboyprogrammer.org>
Date: Sun, 9 Jun 2024 19:55:50 +0200
Subject: [PATCH] Fixed nested content inside blockquotes

---
 .../feeder/model/html/HtmlLinearizer.kt       | 58 ++++++++++++++-----
 .../feeder/model/html/HtmlLinearizerTest.kt   | 35 +++++++++++
 2 files changed, 78 insertions(+), 15 deletions(-)

diff --git a/app/src/main/java/com/nononsenseapps/feeder/model/html/HtmlLinearizer.kt b/app/src/main/java/com/nononsenseapps/feeder/model/html/HtmlLinearizer.kt
index d3a31e45c..baa35912d 100644
--- a/app/src/main/java/com/nononsenseapps/feeder/model/html/HtmlLinearizer.kt
+++ b/app/src/main/java/com/nononsenseapps/feeder/model/html/HtmlLinearizer.kt
@@ -334,21 +334,49 @@ class HtmlLinearizer {
 
                         "blockquote" -> {
                             finalizeAndAddCurrentElement(blockStyle)
-                            add(
-                                LinearBlockQuote(
-                                    cite = element.attr("cite").ifBlank { null },
-                                    content =
-                                        ListBuilderScope {
-                                            asElement(blockStyle = LinearTextBlockStyle.TEXT) {
-                                                linearizeChildren(
-                                                    element.childNodes(),
-                                                    blockStyle = LinearTextBlockStyle.TEXT,
-                                                    baseUrl = baseUrl,
-                                                )
-                                            }
-                                        }.items,
-                                ),
-                            )
+
+                            val cite = element.attr("cite").ifBlank { null }
+
+                            // Text should be collected into LinearBlockQuote
+                            // Other types should be added separately
+                            ListBuilderScope {
+                                asElement(blockStyle = LinearTextBlockStyle.TEXT) {
+                                    linearizeChildren(
+                                        element.childNodes(),
+                                        blockStyle = LinearTextBlockStyle.TEXT,
+                                        baseUrl = baseUrl,
+                                    )
+                                }
+                            }.items
+                                .fold(mutableListOf<LinearElement>()) { acc, item ->
+                                    if (item is LinearText) {
+                                        acc.add(item)
+                                        acc
+                                    } else {
+                                        if (acc.isNotEmpty()) {
+                                            add(
+                                                LinearBlockQuote(
+                                                    cite = cite,
+                                                    content = acc,
+                                                ),
+                                            )
+                                        }
+                                        add(
+                                            item,
+                                        )
+                                        mutableListOf()
+                                    }
+                                }
+                                .let {
+                                    if (it.isNotEmpty()) {
+                                        add(
+                                            LinearBlockQuote(
+                                                cite = cite,
+                                                content = it,
+                                            ),
+                                        )
+                                    }
+                                }
                         }
 
                         "a" -> {
diff --git a/app/src/test/java/com/nononsenseapps/feeder/model/html/HtmlLinearizerTest.kt b/app/src/test/java/com/nononsenseapps/feeder/model/html/HtmlLinearizerTest.kt
index b08153cbe..63475d793 100644
--- a/app/src/test/java/com/nononsenseapps/feeder/model/html/HtmlLinearizerTest.kt
+++ b/app/src/test/java/com/nononsenseapps/feeder/model/html/HtmlLinearizerTest.kt
@@ -2,6 +2,7 @@ package com.nononsenseapps.feeder.model.html
 
 import com.nononsenseapps.feeder.ui.compose.html.toTableData
 import org.junit.Before
+import org.junit.Ignore
 import kotlin.test.Test
 import kotlin.test.assertEquals
 import kotlin.test.assertTrue
@@ -536,6 +537,40 @@ class HtmlLinearizerTest {
         )
     }
 
+    @Test
+    fun `content inside blockquote`() {
+        val html =
+            """
+            <p><a class="tumblr_blog" href="https://everydaylouie.tumblr.com/post/716429042034802688/challenge">everydaylouie</a>:</p>
+            <blockquote>
+            <div class="npf_row">
+            <figure class="tmblr-full" data-orig-height="624" data-orig-width="1200">
+            <img src="https://64.media.tumblr.com/55359709d06eb4309538800b64745d40/5149f25c974a1479-f7/s640x960/75e2a17fad52257ac5c7326e123d125be18ecd14.png" data-orig-height="624" data-orig-width="1200" srcset="https://64.media.tumblr.com/55359709d06eb4309538800b64745d40/5149f25c974a1479-f7/s75x75_c1/48387a66ca77d9e8af781b47efb733f8e5abe553.png 75w, https://64.media.tumblr.com/55359709d06eb4309538800b64745d40/5149f25c974a1479-f7/s100x200/9387531174b8be87ae89fd6f967e274336b7f5a4.png 100w, https://64.media.tumblr.com/55359709d06eb4309538800b64745d40/5149f25c974a1479-f7/s250x400/435ce8fba2aa5f75a548a497f1157f563245dd93.png 250w, https://64.media.tumblr.com/55359709d06eb4309538800b64745d40/5149f25c974a1479-f7/s400x600/3fc4822b65cab32e05ed0611b6721ab3ab0ba5c8.png 400w, https://64.media.tumblr.com/55359709d06eb4309538800b64745d40/5149f25c974a1479-f7/s500x750/f86d4df43eb0de4695d3261cca41739278c9e6bc.png 500w, https://64.media.tumblr.com/55359709d06eb4309538800b64745d40/5149f25c974a1479-f7/s540x810/64c4028177a81a99d416b8994b07b8d228b25ea5.png 540w, https://64.media.tumblr.com/55359709d06eb4309538800b64745d40/5149f25c974a1479-f7/s640x960/75e2a17fad52257ac5c7326e123d125be18ecd14.png 640w, https://64.media.tumblr.com/55359709d06eb4309538800b64745d40/5149f25c974a1479-f7/s1280x1920/258a6c4e80c6403aa4163c73cf39c09183dd4b4a.png 1200w" sizes="(max-width: 1200px) 100vw, 1200px"/>
+            </figure></div><p>CHALLENGE!</p></blockquote>
+            """.trimIndent()
+
+        val baseUrl = "https://example.com"
+
+        val result = linearizer.linearize(html, baseUrl).elements
+
+        assertEquals(3, result.size, "Expected items: $result")
+
+        assertTrue { result[1] is LinearImage }
+    }
+
+    @Ignore
+    @Test
+    fun `audio in iframe`() {
+        val html =
+            """
+            <iframe 
+            class="tumblr_audio_player tumblr_audio_player_718384010494197760" 
+            src="https://pokemonmysterydungeon.tumblr.com/post/718384010494197760/audio_player_iframe/pokemonmysterydungeon/tumblr_pgyj6gSPFJ1skzors?audio_file=https%3A%2F%2Fa.tumblr.com%2Ftumblr_pgyj6gSPFJ1skzorso1.mp3"
+            frameborder="0" allowtransparency="true" scrolling="no" width="540" height="169">
+            </iframe>
+            """.trimIndent()
+    }
+
     @Test
     fun `video with no sources is ignored`() {
         val html = "<html><body><video controls></video></body></html>"