From bd395d0d94306d35601fa4382e9544251439aa73 Mon Sep 17 00:00:00 2001
From: Josiah Campbell <9521010+jocmp@users.noreply.github.com>
Date: Thu, 2 Jan 2025 21:43:58 -0600
Subject: [PATCH] Improve FreshRSS image preview parsing (#669)

* Unescape &amp; in image URLs

* Parse FreshRSS content for first image
---
 .../capyreader/app/ui/articles/ArticleRow.kt    |  1 +
 .../capy/accounts/reader/ItemParsedImageUrl.kt  | 17 +++++++++++++++++
 .../accounts/reader/ReaderAccountDelegate.kt    |  2 +-
 ...LCharactersExt.kt => StringCharactersExt.kt} |  8 ++++++++
 4 files changed, 27 insertions(+), 1 deletion(-)
 create mode 100644 capy/src/main/java/com/jocmp/capy/accounts/reader/ItemParsedImageUrl.kt
 rename capy/src/main/java/com/jocmp/capy/common/{StringXMLCharactersExt.kt => StringCharactersExt.kt} (76%)

diff --git a/app/src/main/java/com/capyreader/app/ui/articles/ArticleRow.kt b/app/src/main/java/com/capyreader/app/ui/articles/ArticleRow.kt
index 92583650..2c2e7b9b 100644
--- a/app/src/main/java/com/capyreader/app/ui/articles/ArticleRow.kt
+++ b/app/src/main/java/com/capyreader/app/ui/articles/ArticleRow.kt
@@ -58,6 +58,7 @@ import com.capyreader.app.ui.fixtures.ArticleSample
 import com.capyreader.app.ui.theme.CapyTheme
 import com.jocmp.capy.Article
 import com.jocmp.capy.MarkRead
+import com.jocmp.capy.common.escapingSpecialXMLCharacters
 import java.net.URL
 import java.time.LocalDateTime
 import java.time.ZoneOffset
diff --git a/capy/src/main/java/com/jocmp/capy/accounts/reader/ItemParsedImageUrl.kt b/capy/src/main/java/com/jocmp/capy/accounts/reader/ItemParsedImageUrl.kt
new file mode 100644
index 00000000..d4af768b
--- /dev/null
+++ b/capy/src/main/java/com/jocmp/capy/accounts/reader/ItemParsedImageUrl.kt
@@ -0,0 +1,17 @@
+package com.jocmp.capy.accounts.reader
+
+import com.jocmp.capy.common.unescapingHTMLCharacters
+import com.jocmp.readerclient.Item
+import org.jsoup.Jsoup
+
+internal fun parsedImageURL(item: Item): String? {
+    val imageHref = item.image?.href
+
+    if (imageHref != null) {
+        return imageHref.unescapingHTMLCharacters
+    }
+
+    val content = item.summary.content
+
+    return Jsoup.parse(content).selectFirst("img")?.attr("src")
+}
diff --git a/capy/src/main/java/com/jocmp/capy/accounts/reader/ReaderAccountDelegate.kt b/capy/src/main/java/com/jocmp/capy/accounts/reader/ReaderAccountDelegate.kt
index eef5142f..858d5fad 100644
--- a/capy/src/main/java/com/jocmp/capy/accounts/reader/ReaderAccountDelegate.kt
+++ b/capy/src/main/java/com/jocmp/capy/accounts/reader/ReaderAccountDelegate.kt
@@ -348,7 +348,7 @@ internal class ReaderAccountDelegate(
                     extracted_content_url = null,
                     summary = Jsoup.parse(item.summary.content).text(),
                     url = item.canonical.firstOrNull()?.href,
-                    image_url = item.image?.href,
+                    image_url = parsedImageURL(item),
                     published_at = item.published
                 )
 
diff --git a/capy/src/main/java/com/jocmp/capy/common/StringXMLCharactersExt.kt b/capy/src/main/java/com/jocmp/capy/common/StringCharactersExt.kt
similarity index 76%
rename from capy/src/main/java/com/jocmp/capy/common/StringXMLCharactersExt.kt
rename to capy/src/main/java/com/jocmp/capy/common/StringCharactersExt.kt
index dfe0af02..898b3103 100644
--- a/capy/src/main/java/com/jocmp/capy/common/StringXMLCharactersExt.kt
+++ b/capy/src/main/java/com/jocmp/capy/common/StringCharactersExt.kt
@@ -21,3 +21,11 @@ val String.escapingSpecialXMLCharacters: String
 
         return escaped
     }
+
+val String.unescapingHTMLCharacters: String
+    get() {
+        return this
+            .replace("&amp;", "&")
+            .replace("&lt;", "<")
+            .replace("&gt;", ">")
+    }