Take the giant Google badge out of the readerHEADmaster

"Remove giant Google icon", reported on a West Hawaii Today article. It is the
"Add as a preferred source on Google" badge that sits beside the byline. Giant
because the page caps it at 120px with its own stylesheet and the reader drops
the page's CSS, so a small badge on the site becomes a full-width image above
the first paragraph.

Matched on the link's address rather than the class, and that is forced rather
than preferred: running the vendored Readability.js over the real page shows
the wrapper's preferred_source_badge class does not survive extraction at all,
so a class rule would match nothing. The anchor does survive, and its address
is Google's own and identical for every publisher in the programme -- where the
badge's image filename sits under this newspaper's theme directory and is not.
So the address is both the only signal left and the one that generalises.

A link with words in it is never touched. An article may discuss the programme
and link to it in a sentence, and that sentence is the author's.

Both cleaners now run from one entry point. There are two places a body is
cleaned -- a fresh extraction on the way out, a cached one on the way back in --
and a cached article is cleaned in place rather than discarded, so a cleaner
wired into only the fresh path would leave every already-read article carrying
litter nothing would ever remove. Peter's report says the article was cached,
which is exactly that case.

Pinned against the real thing: the reported article, put through the real
extractor, loses the badge and keeps its text and byline.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018AwnALZL6bD4VBtATDVv3B

4 files changed, 96 insertions, 3 deletions

diff --git a/app/src/main/java/org/terst/feedreader/data/CachedExtraction.kt b/app/src/main/java/org/terst/feedreader/data/CachedExtraction.kt
index 63614d1..39f39a5 100644
--- a/app/src/main/java/org/terst/feedreader/data/CachedExtraction.kt
+++ b/app/src/main/java/org/terst/feedreader/data/CachedExtraction.kt
@@ -1,7 +1,7 @@
 package org.terst.feedreader.data
 
 import org.terst.feedreader.readability.ConsentNotice
-import org.terst.feedreader.readability.RecirculationWidgets
+import org.terst.feedreader.readability.ArticleLitter
 import org.terst.feedreader.readability.UnrenderedTex
 
 /** What a cached `articles.extractedContent` value turned out to be worth. */
@@ -64,7 +64,7 @@ suspend fun loadCachedExtraction(
         // article carrying the publisher's promo blocks is a good article with litter in it,
         // not a poisoned row: cleaning it in place shows the right thing immediately, where
         // discarding it would throw away a usable body and fetch the identical one back.
-        content = RecirculationWidgets.strip(cached),
+        content = ArticleLitter.strip(cached),
         discardedConsentNotice = false
     )
 
diff --git a/app/src/main/java/org/terst/feedreader/readability/ArticleLitter.kt b/app/src/main/java/org/terst/feedreader/readability/ArticleLitter.kt
new file mode 100644
index 0000000..decddc6
--- /dev/null
+++ b/app/src/main/java/org/terst/feedreader/readability/ArticleLitter.kt
@@ -0,0 +1,24 @@
+package org.terst.feedreader.readability
+
+/**
+ * Everything that gets cleaned out of an extracted article body, in one call.
+ *
+ * There are two places a body is cleaned -- a fresh extraction on its way out of [EvalDecision],
+ * and a cached one on its way back in through `loadCachedExtraction` -- and they must stay in
+ * step. A cached article is cleaned in place rather than discarded, so a cleaner wired into only
+ * the fresh path would leave every already-read article carrying litter that nothing would ever
+ * remove: the body is good, so it is never re-fetched.
+ *
+ * Hence one entry point rather than a list of calls repeated at both sites. Adding a cleaner here
+ * reaches both; adding one to a call site reaches one, and which one it misses would depend on
+ * which file was open at the time.
+ *
+ * None of these discard an article. They remove things from inside it, which is why they sit
+ * apart from [ConsentNotice] and [UnrenderedTex] -- those two decide a body is not worth keeping.
+ */
+object ArticleLitter {
+
+    /** [html] with the publisher's promo furniture removed. */
+    fun strip(html: String?): String? =
+        PreferredSourceBadge.strip(RecirculationWidgets.strip(html))
+}
diff --git a/app/src/main/java/org/terst/feedreader/readability/EvalDecision.kt b/app/src/main/java/org/terst/feedreader/readability/EvalDecision.kt
index dd77014..d0bd1cd 100644
--- a/app/src/main/java/org/terst/feedreader/readability/EvalDecision.kt
+++ b/app/src/main/java/org/terst/feedreader/readability/EvalDecision.kt
@@ -44,7 +44,7 @@ internal fun decideEvalOutcome(
         // never reach `articles.extractedContent` either.
         usable -> EvalDecision.Done(
             ExtractionResult.Success(
-                article!!.copy(content = RecirculationWidgets.strip(article.content))
+                article!!.copy(content = ArticleLitter.strip(article.content))
             )
         )
 
diff --git a/app/src/main/java/org/terst/feedreader/readability/PreferredSourceBadge.kt b/app/src/main/java/org/terst/feedreader/readability/PreferredSourceBadge.kt
new file mode 100644
index 0000000..4ea1370
--- /dev/null
+++ b/app/src/main/java/org/terst/feedreader/readability/PreferredSourceBadge.kt
@@ -0,0 +1,69 @@
+package org.terst.feedreader.readability
+
+/**
+ * Removes the "Add as a preferred source on Google" badge publishers put beside the byline.
+ *
+ * Reported live on a West Hawaii Today article: "remove giant Google icon". Giant because the
+ * page caps it with its own stylesheet -- `.preferred_source_badge { max-width: 120px }` -- and
+ * the reader drops the page's CSS, so what is a small badge on the site becomes a full-width
+ * image in the reader, above the first paragraph.
+ *
+ * The signal is the link's address, not the class or the image filename, and that is forced
+ * rather than chosen. Confirmed by running the vendored Readability.js over the real page: the
+ * wrapper's `preferred_source_badge` class does not survive extraction at all -- the only class
+ * left in the output is Readability's own -- so a class rule would match nothing. What does
+ * survive is the anchor, intact:
+ *
+ * ```
+ * <p><a href="https://www.google.com/preferences/source?q=westhawaiitoday.com" target="_blank"
+ *      rel="noopener"> <img src=".../google_preferred_source_badge_light_en.png"
+ *      alt="Add as a preferred source on Google"> </a> </p>
+ * ```
+ *
+ * The address is Google's own and identical for every publisher in the programme; the image
+ * filename sits under the site's theme directory and is not. So matching the address is both the
+ * only thing available and the one that generalises past this one newspaper.
+ *
+ * A link with real words in it is never touched, which is the same safety valve [ConsentNotice]
+ * and [RecirculationWidgets] use: an article may legitimately link to that page and say so in
+ * writing, and that sentence is the author's, not litter. Only a link whose entire content is an
+ * image goes.
+ */
+object PreferredSourceBadge {
+
+    /** Google's address for the programme. Not the badge image, which is per-publisher. */
+    private const val BADGE_LINK = "google.com/preferences/source"
+
+    private val ANCHOR = Regex("""<a\b[^>]*>.*?</a\s*>""", setOf(RegexOption.IGNORE_CASE, RegexOption.DOT_MATCHES_ALL))
+    private val TAG = Regex("<[^>]*>")
+    private val EMPTY_PARAGRAPH = Regex("""<p>\s*</p>""", RegexOption.IGNORE_CASE)
+
+    /** [html] with any preferred-source badge removed. */
+    fun strip(html: String?): String? =
+        if (html.isNullOrBlank() || !html.contains(BADGE_LINK, ignoreCase = true)) {
+            html
+        } else {
+            removeBadges(html)
+        }
+
+    private fun removeBadges(html: String): String {
+        val withoutBadges = ANCHOR.replace(html) { match ->
+            if (isBadge(match.value)) "" else match.value
+        }
+        // The badge is normally the only thing in its paragraph, and a <p> holding nothing but
+        // the whitespace around the image it used to wrap still takes up a line on the screen.
+        return EMPTY_PARAGRAPH.replace(withoutBadges, "")
+    }
+
+    /**
+     * A badge is a link to the programme whose whole content is an image.
+     *
+     * Both halves are required. The address alone would take a sentence linking to the page with
+     * it; "contains an image" alone would take every photo credit on the site.
+     */
+    private fun isBadge(anchor: String): Boolean {
+        val opening = anchor.substringBefore('>')
+        if (!opening.contains(BADGE_LINK, ignoreCase = true)) return false
+        return TAG.replace(anchor, "").isBlank()
+    }
+}