Fix srcset-polluted URLs from butlerd, sanitize cover and screenshot URLs

butlerd sometimes passes itch.io coverUrl/stillCoverUrl with trailing
srcset descriptors (e.g. ' 1x, h' or ' 1'). These polluted URLs cause
HTTPImageFetcher to fail because the URL is invalid and the extension
extraction breaks.

Changes:
- Add _sanitize_url() to strip srcset artifacts from URLs
- Add _is_gif_url() helper that works with both clean and polluted URLs
- Fix SCREENSHOT_REGEX_PATTERN to use [\w+/=]* instead of [^"]* to
  avoid crossing srcset boundaries in HTML
- Replace all .ends_with('.gif') checks with _is_gif_url()
- Sanitize all URLs from game metadata and HTML scraping
- Add IMAGE_EXTS constant for extension validation

Also bump to v0.1.7.

core/boxart_itch.gd

@@ -13,7 +13,8 @@ extends BoxArtProvider
 const BOXART_DIR := "user://boxart/itch"
 const META_CACHE_FILE := "itch_art_meta.json"
 const HTTP_TIMEOUT_MS := 15000
-const SCREENSHOT_REGEX_PATTERN := "https://img\\.itch\\.zone/aW1hZ2Uv[^\"]*/original/[^\"]*"
+const SCREENSHOT_REGEX_PATTERN := "https://img\\.itch\\.zone/aW1hZ2Uv[\\w+/=]*/original/[\\w+/=]*\\.[a-z]+"
+const IMAGE_EXTS := ["png", "jpg", "jpeg", "gif", "webp", "bmp"]
 const CONNECT_POLL_DELAY_MS := 50
 const READ_POLL_DELAY_MS := 10
 const LOG_URL_MAX_LEN := 80
@@ -45,6 +46,27 @@ func _ready() -> void:
 	add_child(http_image)
 
 
+## Strips HTML srcset artifacts from a URL. Butlerd sometimes passes through
+## itch.io cover URLs with trailing srcset descriptors like " 1x, h" or " 2x".
+## These cause HTTPImageFetcher to fail because the URL is no longer valid.
+func _sanitize_url(url: String) -> String:
+	if url.is_empty():
+		return url
+	# Srcset descriptors are separated by spaces — take only the URL portion.
+	var clean: String = url.split(" ")[0]
+	# Validate that the cleaned URL ends with a known image extension.
+	var ext: String = clean.get_extension().to_lower()
+	if ext not in IMAGE_EXTS:
+		logger.warn("Sanitized URL has unexpected extension '%s': %s" % [ext, clean.left(LOG_URL_MAX_LEN)])
+	return clean
+
+
+## Returns whether a URL points to a GIF image. Handles both clean URLs and
+## URLs contaminated with srcset artifacts (e.g. "image.gif 1x, h").
+func _is_gif_url(url: String) -> bool:
+	return _sanitize_url(url).to_lower().ends_with(".gif")
+
+
 func get_boxart(item: LibraryItem, kind: LAYOUT) -> Texture2D:
 	if not kind in layout_map:
 		logger.error("Unsupported boxart layout: {0}".format([kind]))
@@ -81,15 +103,15 @@ func get_boxart(item: LibraryItem, kind: LAYOUT) -> Texture2D:
 	else:
 		logger.warn("No URL in game dict for: %s" % title)
 
-	var screenshots: Array = meta.get("screenshots", [])
+	var screenshots: Array = meta.get("screenshots", []).map(_sanitize_url)
 	# Filter out GIF screenshots — Godot can't decode them.
 	var non_gif_screenshots: Array = screenshots.filter(
-		func(url): return not url.to_lower().ends_with(".gif")
+		func(url): return not _is_gif_url(url)
 	)
 	if screenshots.size() != non_gif_screenshots.size():
 		logger.info("Filtered %d GIF screenshots for: %s" % [screenshots.size() - non_gif_screenshots.size(), title])
 
-	var cover_url := _best_portrait_url(game, non_gif_screenshots)
+	var cover_url := _sanitize_url(_best_portrait_url(game, non_gif_screenshots))
 	if cover_url.is_empty():
 		logger.warn("No cover URL for: %s (id=%d)" % [title, game_id])
 		return null
@@ -150,14 +172,14 @@ func _url_for_layout(kind: LAYOUT, cover_url: String, screenshots: Array) -> Str
 		LAYOUT.GRID_LANDSCAPE:
 			# Prefer a non-GIF screenshot
 			for url in screenshots:
-				if not url.to_lower().ends_with(".gif"):
+				if not _is_gif_url(url):
 					return url
 			return cover_url
 		LAYOUT.BANNER:
 			# Single screenshot fallback (two-screenshot case handled in get_boxart).
 			# Prefer a non-GIF screenshot.
 			for url in screenshots:
-				if not url.to_lower().ends_with(".gif"):
+				if not _is_gif_url(url):
 					return url
 			return cover_url
 	return ""
@@ -169,15 +191,15 @@ func _url_for_layout(kind: LAYOUT, cover_url: String, screenshots: Array) -> Str
 func _best_portrait_url(game: Dictionary, screenshots: Array) -> String:
 	# Prefer stillCoverUrl if it's not a GIF
 	var still: String = game.get("stillCoverUrl", "")
-	if not still.is_empty() and not still.to_lower().ends_with(".gif"):
+	if not still.is_empty() and not _is_gif_url(still):
 		return still
 	# Try coverUrl if it's not a GIF
 	var cover: String = game.get("coverUrl", "")
-	if not cover.is_empty() and not cover.to_lower().ends_with(".gif"):
+	if not cover.is_empty() and not _is_gif_url(cover):
 		return cover
 	# Try the first non-GIF screenshot
 	for url in screenshots:
-		if not url.to_lower().ends_with(".gif"):
+		if not _is_gif_url(url):
 			return url
 	# Last resort: use whatever we have (GIF)
 	return still if not still.is_empty() else cover
@@ -274,8 +296,8 @@ func _fetch_game_page_metadata(game: Dictionary) -> Dictionary:
 	shot_re.compile(SCREENSHOT_REGEX_PATTERN)
 	var screenshots: Array = []
 	for m in shot_re.search_all(html):
-		var url: String = m.get_string(0)
-		if url.to_lower().ends_with(".gif"):
+		var url: String = _sanitize_url(m.get_string(0))
+		if _is_gif_url(url):
 			logger.debug("Skipping GIF screenshot: " + url)
 			continue
 		screenshots.append(url)

plugin.json

@@ -1,7 +1,7 @@
 {
   "plugin.id": "itch-artprovider",
   "plugin.name": "itch.io Art Provider",
-  "plugin.version": "0.1.6",
+  "plugin.version": "0.1.7",
   "plugin.min-api-version": "1.1.0",
   "plugin.link": "https://forge.thergic.ar/jose/itchio-opengamepadui-artprovider-plugin",
   "plugin.source": "https://forge.thergic.ar/jose/itchio-opengamepadui-artprovider-plugin",