Don't break CDATA titles for local feeds (#720)

Fixes a parser issue where CDATA was parsed as escaped UTF-8 characters instead of plaintext. The fix is to use Jsoup.clean instead of Android's textutil implementation
jocmp · Jan 14, 2025 · e555963 · e555963
1 parent 4bc6440
commit e555963
Show file tree

Hide file tree

Showing 5 changed files with 25 additions and 59 deletions.
diff --git a/app/src/main/java/com/capyreader/app/notifications/NotificationHelper.kt b/app/src/main/java/com/capyreader/app/notifications/NotificationHelper.kt
@@ -35,11 +35,11 @@ class NotificationHelper(
             return
         }
 
+        sendGroupNotification()
+
         notifications.forEach {
             sendNotification(it)
         }
-
-        sendGroupNotification()
     }
 
     fun dismissNotifications(ids: List<String>) {

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
@@ -40,9 +40,7 @@ import androidx.compose.ui.layout.ContentScale
 import androidx.compose.ui.platform.LocalDensity
 import androidx.compose.ui.platform.LocalHapticFeedback
 import androidx.compose.ui.res.stringResource
-import androidx.compose.ui.text.AnnotatedString
 import androidx.compose.ui.text.font.FontWeight
-import androidx.compose.ui.text.fromHtml
 import androidx.compose.ui.text.style.TextDirection
 import androidx.compose.ui.text.style.TextOverflow
 import androidx.compose.ui.tooling.preview.Preview
@@ -107,7 +105,7 @@ fun ArticleRow(
             ArticleListItem(
                 headlineContent = {
                     Text(
-                        AnnotatedString.fromHtml(article.title),
+                        article.title,
                         fontWeight = FontWeight.Bold,
                     )
                 },

diff --git a/capy/src/main/java/com/jocmp/capy/accounts/local/ParsedItem.kt b/capy/src/main/java/com/jocmp/capy/accounts/local/ParsedItem.kt
@@ -1,6 +1,5 @@
 package com.jocmp.capy.accounts.local
 
-import com.jocmp.capy.common.escapingHTMLCharacters
 import com.jocmp.rssparser.model.RssItem
 import org.jsoup.Jsoup
 import org.jsoup.safety.Safelist
@@ -35,7 +34,7 @@ internal class ParsedItem(private val item: RssItem, private val siteURL: String
         }
 
     val title: String
-        get() = Jsoup.parse(item.title.orEmpty()).text().escapingHTMLCharacters
+        get() = Jsoup.clean(item.title.orEmpty(), Safelist.none())
 
     val imageURL: String?
         get() = cleanedURL(item.image)?.toString()

diff --git a/capy/src/main/java/com/jocmp/capy/common/StringCharactersExt.kt b/capy/src/main/java/com/jocmp/capy/common/StringCharactersExt.kt
@@ -29,51 +29,3 @@ val String.unescapingHTMLCharacters: String
             .replace("&lt;", "<")
             .replace("&gt;", ">")
     }
-
-/**
- * Returns an HTML escaped representation of the given plain text.
- *
- * Copied from android.text.Html
- */
-val String.escapingHTMLCharacters: String
-    get() {
-        val out = java.lang.StringBuilder()
-        withinStyle(out, this, 0, this.length)
-        return out.toString()
-    }
-
-private fun withinStyle(out: StringBuilder, text: CharSequence, start: Int, end: Int) {
-    var i: Int = start
-    while (i < end) {
-        val c: Char = text.get(i)
-
-        if (c == '<') {
-            out.append("&lt;")
-        } else if (c == '>') {
-            out.append("&gt;")
-        } else if (c == '&') {
-            out.append("&amp;")
-        } else if (c.code in 0xD800..0xDFFF) {
-            if (c.code < 0xDC00 && i + 1 < end) {
-                val d: Char = text.get(i + 1)
-                if (d.code in 0xDC00..0xDFFF) {
-                    i++
-                    val codepoint = 0x010000 or (c.code - 0xD800 shl 10) or d.code - 0xDC00
-                    out.append("&#").append(codepoint).append(";")
-                }
-            }
-        } else if (c.code > 0x7E || c < ' ') {
-            out.append("&#").append(c.code).append(";")
-        } else if (c == ' ') {
-            while (i + 1 < end && text.get(i + 1) == ' ') {
-                out.append("&nbsp;")
-                i++
-            }
-
-            out.append(' ')
-        } else {
-            out.append(c)
-        }
-        i++
-    }
-}
diff --git a/capy/src/test/java/com/jocmp/capy/accounts/local/ParsedItemTest.kt b/capy/src/test/java/com/jocmp/capy/accounts/local/ParsedItemTest.kt
@@ -26,7 +26,8 @@ class ParsedItemTest {
 
     @Test
     fun title_withNestedHTML() {
-        val title = "<mark>The `&lt;details&gt;` and `&lt;summary&gt;` elements are getting an upgrade</mark>"
+        val title =
+            "<mark>The `&lt;details&gt;` and `&lt;summary&gt;` elements are getting an upgrade</mark>"
 
         val item = RssItem.Builder().title(title).build()
         val parsedItem = ParsedItem(item, siteURL = "")
@@ -37,6 +38,19 @@ class ParsedItemTest {
         )
     }
 
+    @Test
+    fun title_withNestedHTMLAndNonAsciiText() {
+        val title = "<![CDATA[ 分析：美國五角大樓將騰訊列入涉軍名單的影響及信號 ]]>"
+
+        val item = RssItem.Builder().title(title).build()
+        val parsedItem = ParsedItem(item, siteURL = "")
+
+        assertEquals(
+            expected = "分析：美國五角大樓將騰訊列入涉軍名單的影響及信號",
+            actual = parsedItem.title,
+        )
+    }
+
     @Test
     fun title_whenNull() {
         val item = RssItem.Builder().title(null).build()
@@ -122,11 +136,14 @@ class ParsedItemTest {
 
     @Test
     fun url_withGoogleAlertsFeed() {
-        val articleURL = "https://www.androidcentral.com/apps-software/google-squashes-a-few-pixel-bugs-in-android-15-qpr2-beta-2-1"
-        val link = "https://www.google.com/url?rct=j&sa=t&url=$articleURL&ct=ga&cd=CAIyGmNmNDdiZGVhOWNiNDUxZTA6Y29tOmVuOlVT&usg=AOvVaw0NIyLHLSRUIwSMg9anVWrG"
+        val articleURL =
+            "https://www.androidcentral.com/apps-software/google-squashes-a-few-pixel-bugs-in-android-15-qpr2-beta-2-1"
+        val link =
+            "https://www.google.com/url?rct=j&sa=t&url=$articleURL&ct=ga&cd=CAIyGmNmNDdiZGVhOWNiNDUxZTA6Y29tOmVuOlVT&usg=AOvVaw0NIyLHLSRUIwSMg9anVWrG"
 
         val item = RssItem.Builder().link(link).build()
-        val parsedItem = ParsedItem(item, siteURL = "https://www.google.com/alerts/feeds/12345/12345")
+        val parsedItem =
+            ParsedItem(item, siteURL = "https://www.google.com/alerts/feeds/12345/12345")
 
         assertEquals(expected = articleURL, actual = parsedItem.url)
     }