itchio-opengamepadui-artprovider-plugin/core/boxart_itch.gd

extends BoxArtProvider

## itch.io box art provider for OpenGamepadUI
##
## Fetches cover art and screenshots from itch.io game pages. Works alongside
## the itch.io library plugin to provide per-layout artwork: the cover for
## portrait/logo, full-res screenshots for landscape/banner.
##
## Enriches game metadata by scraping the itch.io page HTML for screenshots
## (the itch.io API does not expose them), caches the result in
## itch_art_meta.json, and uses HTTPImageFetcher for async image downloads.

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[\\w+/=]*/original/[\\w+/=]*\\.[a-z]+"
const IMAGE_EXTS := ["png", "jpg", "jpeg", "gif", "webp", "bmp"]
const FFMPEG_BUILDS_URL := "https://github.com/BtbN/FFmpeg-Builds/releases/download/latest"
const CONNECT_POLL_DELAY_MS := 50
const READ_POLL_DELAY_MS := 10
const LOG_URL_MAX_LEN := 80
const CACHE_DIR := "images"
const BANNER_CACHE_DIR := "banners"
const GIF_STATIC_DIR := "gif_static"

@export var use_caching: bool = true
var http_image := HTTPImageFetcher.new()
var _meta_cache_dir := "itch_art"
var _enriched_meta: Dictionary = {}

var layout_map: Dictionary = {
	LAYOUT.GRID_PORTRAIT: "-portrait",
	LAYOUT.GRID_LANDSCAPE: "-landscape",
	LAYOUT.BANNER: "-banner",
	LAYOUT.LOGO: "-logo",
}

## Resolved path to the ffmpeg binary (system or bundled). Populated lazily.
var _ffmpeg_bin: String = ""


func _init() -> void:
	super()
	var globalized := ProjectSettings.globalize_path(BOXART_DIR)
	DirAccess.make_dir_recursive_absolute(globalized)
	provider_id = "itch"
	logger_name = "BoxArtItch"


func _ready() -> void:
	super()
	_enriched_meta = _load_enriched_meta()
	logger.info("itch.io Art Provider loaded")
	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".
func _sanitize_url(url: String) -> String:
	if url.is_empty():
		return url
	var clean: String = url.split(" ")[0]
	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.
func _is_gif_url(url: String) -> bool:
	return _sanitize_url(url).to_lower().ends_with(".gif")


## Returns whether a URL points to a JPEG image.
func _is_jpg_url(url: String) -> bool:
	var clean := _sanitize_url(url).to_lower()
	return clean.ends_with(".jpg") or clean.ends_with(".jpeg")


## Removes and frees an HTTPRequest node.
func _remove_http(http: HTTPRequest) -> void:
	remove_child(http)
	http.queue_free()


## Downloads an image and detects the actual format from content-type headers.
## Used as a fallback when HTTPImageFetcher fails due to extension mismatch.
func _fetch_image_with_format_detection(url: String, cache_flags: int) -> Texture2D:
	var http := HTTPRequest.new()
	http.timeout = HTTP_TIMEOUT_MS / 1000.0
	add_child.call_deferred(http)
	await http.ready

	if http.request(url) != OK:
		_remove_http(http)
		return null

	var args: Array = await http.request_completed
	var result: int = args[0]
	var response_code: int = args[1]
	var headers: PackedStringArray = args[2]
	var body: PackedByteArray = args[3]
	_remove_http(http)

	if result != HTTPRequest.RESULT_SUCCESS or response_code != 200:
		return null

	var content_type := ""
	for h in headers:
		if h.to_lower().begins_with("content-type:"):
			content_type = h.split(":", true, 1)[1].strip_edges().to_lower()
			break

	var image := Image.new()
	var err: int = ERR_INVALID_DATA
	if content_type.find("png") != -1:
		err = image.load_png_from_buffer(body)
	elif content_type.find("jpeg") != -1 or content_type.find("jpg") != -1:
		err = image.load_jpg_from_buffer(body)
	elif content_type.find("webp") != -1:
		err = image.load_webp_from_buffer(body)
	else:
		err = image.load_png_from_buffer(body)
		if err != OK:
			err = image.load_jpg_from_buffer(body)

	if err != OK:
		logger.warn("Format detection failed for %s (content-type: %s)" % [url.left(LOG_URL_MAX_LEN), content_type])
		return null

	var texture := ImageTexture.create_from_image(image)
	if cache_flags & Cache.FLAGS.SAVE:
		Cache.save_image(CACHE_DIR, url, texture)
	return texture


func get_boxart(item: LibraryItem, kind: LAYOUT) -> Texture2D:
	if not kind in layout_map:
		logger.error("Unsupported boxart layout: {0}".format([kind]))
		return null

	var game := _find_game_dict(item)
	if game.is_empty():
		logger.warn("No itch.io game metadata found for: " + item.name)
		return null

	var game_id := int(game.get("id", 0))
	var title: String = game.get("title", "")
	logger.info("get_boxart: title=%s id=%d layout=%s" % [title, game_id, str(kind)])
	if title.is_empty():
		logger.warn("Game title is empty for id %d" % game_id)
		return null

	var meta := await _enrich_metadata(game, game_id, title)
	var screenshots: Array = meta.get("screenshots", []).map(_sanitize_url)
	var cover_url := _sanitize_url(_best_portrait_url(game, screenshots))
	if cover_url.is_empty():
		logger.warn("No cover URL for: %s (id=%d)" % [title, game_id])
		return null
	logger.info("cover_url=%s" % cover_url.left(LOG_URL_MAX_LEN))

	var cache_flags := Cache.FLAGS.NONE
	if use_caching:
		cache_flags = Cache.FLAGS.LOAD | Cache.FLAGS.SAVE

	if kind == LAYOUT.BANNER and screenshots.size() >= 2:
		return await _fetch_banner_textures(item, screenshots, cache_flags)

	var url := _url_for_layout(kind, cover_url, screenshots)
	if url.is_empty():
		logger.warn("URL for layout %s is empty (cover=%s)" % [str(kind), cover_url])
		return null

	logger.info("Fetching itch.io box art for: %s layout=%s url=%s" % [item.name, str(kind), url.left(LOG_URL_MAX_LEN)])
	var texture: Texture2D = await _fetch_image(url, cache_flags)
	if texture == null:
		logger.warn("Image download returned null for: %s url=%s" % [item.name, url.left(LOG_URL_MAX_LEN)])
	return texture


## Enriches game metadata by scraping the itch.io page for screenshots.
func _enrich_metadata(game: Dictionary, game_id: int, title: String) -> Dictionary:
	var meta: Dictionary = _enriched_meta.get(game_id, {})
	if meta.is_empty() and game.has("url"):
		logger.info("Enriching metadata for: %s (url=%s)" % [title, game.get("url", "")])
		meta = await _fetch_game_page_metadata(game)
		_enriched_meta[game_id] = meta
		_save_enriched_meta(game_id, meta)
		if meta.is_empty():
			logger.info("No screenshots available for: %s — will use cover art" % title)
		else:
			logger.info("Scraped %d screenshots for: %s" % [meta.get("screenshots", []).size(), title])
	elif not meta.is_empty():
		logger.info("Using cached metadata for: %s (%d screenshots)" % [title, meta.get("screenshots", []).size()])
	else:
		logger.warn("No URL in game dict for: %s" % title)
	return meta


## Fetches two screenshots and combines them side-by-side for banner layout.
func _fetch_banner_textures(item: LibraryItem, screenshots: Array, cache_flags: int) -> Texture2D:
	var banner_key: String = screenshots[0] + "||" + screenshots[1]

	if cache_flags & Cache.FLAGS.LOAD and Cache.is_cached(BANNER_CACHE_DIR, banner_key):
		var cached := Cache.get_image(BANNER_CACHE_DIR, banner_key)
		if cached != null:
			logger.info("Banner: loaded cached stitched banner for %s" % item.name)
			return cached

	logger.info("Fetching itch.io banner (2 screenshots) for: " + item.name)
	var tex_a := await _fetch_image(screenshots[0], cache_flags)
	var tex_b := await _fetch_image(screenshots[1], cache_flags)
	if tex_a != null and tex_b != null:
		var stitched := _combine_side_by_side(tex_a, tex_b)
		if cache_flags & Cache.FLAGS.SAVE and stitched != null:
			Cache.save_image(BANNER_CACHE_DIR, banner_key, stitched)
		return stitched
	logger.warn("Banner screenshot fetch failed: tex_a=%s tex_b=%s" % [str(tex_a != null), str(tex_b != null)])
	if tex_a != null:
		return tex_a
	if tex_b != null:
		return tex_b
	return null


## Finds the itch.io game dict from the library item's launch items.
func _find_game_dict(item: LibraryItem) -> Dictionary:
	if item.launch_items.is_empty():
		logger.warn("Item has no launch items: " + item.name)
		return {}
	for launch_item in item.launch_items:
		if launch_item._provider_id != "itch":
			continue
		var game: Dictionary = launch_item.metadata.get("game", {})
		if not game.is_empty():
			return game
	logger.warn("No itch.io launch item found for: %s (providers: %s)" % [
		item.name,
		PackedStringArray(item.launch_items.map(func(l): return l._provider_id))
	])
	return {}


func _url_for_layout(kind: LAYOUT, cover_url: String, screenshots: Array) -> String:
	match kind:
		LAYOUT.GRID_PORTRAIT, LAYOUT.LOGO:
			return cover_url
		LAYOUT.GRID_LANDSCAPE:
			return screenshots[0] if not screenshots.is_empty() else cover_url
		LAYOUT.BANNER:
			return screenshots[0] if not screenshots.is_empty() else cover_url
	return ""


## Returns the best portrait/logo URL from the game data.
## Prefers stillCoverUrl (the still/non-animated version of the cover),
## then coverUrl. Screenshots are never used as capsule art.
func _best_portrait_url(game: Dictionary, _screenshots: Array) -> String:
	var still: String = game.get("stillCoverUrl", "")
	if not still.is_empty():
		return still
	return game.get("coverUrl", "")


## Ensures an ffmpeg binary is available. Checks the system PATH first,
## then falls back to a bundled copy under user://boxart/itch/.
func _ensure_ffmpeg() -> String:
	if not _ffmpeg_bin.is_empty():
		return _ffmpeg_bin
	var probe: Array = []
	OS.execute("ffmpeg", ["-version"], probe, true)
	if probe.size() > 0 and probe[0].find("ffmpeg version") != -1:
		_ffmpeg_bin = "ffmpeg"
		logger.info("Using system ffmpeg")
		return _ffmpeg_bin
	var bundled: String = ProjectSettings.globalize_path(BOXART_DIR) + "/ffmpeg"
	if FileAccess.file_exists(bundled):
		_ffmpeg_bin = bundled
		logger.info("Using bundled ffmpeg: " + _ffmpeg_bin)
		return _ffmpeg_bin
	logger.info("ffmpeg not found — downloading static build")
	if await _install_ffmpeg():
		_ffmpeg_bin = ProjectSettings.globalize_path(BOXART_DIR) + "/ffmpeg"
		return _ffmpeg_bin
	logger.warn("ffmpeg unavailable — GIF covers will not load")
	return ""


## Downloads a static ffmpeg build.
func _install_ffmpeg() -> bool:
	var os_name := _detect_os()
	var arch_name := _detect_arch()
	var archive_url := _build_ffmpeg_url(os_name, arch_name)
	var body := await _download_url_bytes(archive_url)
	if body.is_empty():
		return false
	return _extract_ffmpeg(body, os_name)


func _detect_os() -> String:
	if OS.get_name() == "Windows":
		return "win"
	if OS.get_name() == "macOS":
		return "macos"
	return "linux"


func _detect_arch() -> String:
	if Engine.has_method("get_architecture_name"):
		var arch: String = Engine.get_architecture_name()
		if "arm64" in arch or "aarch64" in arch:
			return "arm64"
	return "64"


func _build_ffmpeg_url(os_name: String, arch_name: String) -> String:
	var platform_slug := os_name + arch_name
	var archive_name := "ffmpeg-master-latest-%s-gpl" % platform_slug
	var ext := "zip" if os_name == "win" else "tar.xz"
	return "%s/%s.%s" % [FFMPEG_BUILDS_URL, archive_name, ext]


## Downloads bytes from a URL. Returns empty PackedByteArray on failure.
func _download_url_bytes(url: String) -> PackedByteArray:
	var http := HTTPRequest.new()
	http.timeout = HTTP_TIMEOUT_MS / 1000.0
	add_child.call_deferred(http)
	await http.ready

	if http.request(url) != OK:
		logger.error("Error requesting: " + url)
		remove_child(http)
		http.queue_free()
		return PackedByteArray()

	var args: Array = await http.request_completed
	var result: int = args[0]
	var response_code: int = args[1]
	var body: PackedByteArray = args[3]
	remove_child(http)
	http.queue_free()

	if result != HTTPRequest.RESULT_SUCCESS or response_code != 200:
		logger.error("Download failed: HTTP %d for %s" % [response_code, url])
		return PackedByteArray()
	return body


## Extracts the ffmpeg binary from the downloaded archive.
func _extract_ffmpeg(body: PackedByteArray, os_name: String) -> bool:
	var globalized_dir := ProjectSettings.globalize_path(BOXART_DIR)
	DirAccess.make_dir_recursive_absolute(globalized_dir)
	var is_win: bool = os_name == "win"
	var ext := "zip" if is_win else "tar.xz"
	var archive_path := globalized_dir + "/ffmpeg." + ext
	var file := FileAccess.open(archive_path, FileAccess.WRITE)
	if file == null:
		logger.error("Cannot write ffmpeg archive to " + archive_path)
		return false
	file.store_buffer(body)
	file.close()

	var out := []
	if is_win:
		OS.execute("unzip", ["-o", archive_path, "-d", globalized_dir], out)
		OS.execute("chmod", ["+x", globalized_dir + "/ffmpeg.exe"], out)
	else:
		OS.execute("tar", ["xf", archive_path, "-C", globalized_dir,
			"--strip-components=1", "--wildcards", "*/ffmpeg"], out)
		OS.execute("chmod", ["+x", globalized_dir + "/ffmpeg"], out)
	DirAccess.remove_absolute(archive_path)
	logger.info("ffmpeg installed to " + globalized_dir)
	return true


## Unified image fetch. GIFs go through ffmpeg for first-frame extraction.
## Non-GIF URLs go through HTTPImageFetcher, with a format-detection
## fallback for .jpg URLs that are actually served as PNG by itch.zone.
func _fetch_image(url: String, cache_flags: int) -> Texture2D:
	if _is_gif_url(url):
		return await _fetch_gif_as_texture(url, cache_flags)
	var texture := await http_image.fetch(url, cache_flags)
	if texture == null and _is_jpg_url(url):
		texture = await _fetch_image_with_format_detection(url, cache_flags)
	return texture


## Downloads a GIF and extracts the first frame as a static ImageTexture.
## Results are cached on disk under user://boxart/itch/gif_static/<hash>.png
## and in memory for the session.
func _fetch_gif_as_texture(url: String, cache_flags: int) -> Texture2D:
	var globalized_dir := ProjectSettings.globalize_path(BOXART_DIR)
	DirAccess.make_dir_recursive_absolute(globalized_dir + "/" + GIF_STATIC_DIR)
	var url_hash := url.sha256_text().left(16)
	var cached_png := globalized_dir + "/" + GIF_STATIC_DIR + "/" + url_hash + ".png"

	# Check disk cache.
	if cache_flags & Cache.FLAGS.LOAD and FileAccess.file_exists(cached_png):
		var image := Image.load_from_file(cached_png)
		if image != null:
			logger.info("GIF fetch: loaded cached static frame for %s" % url.left(LOG_URL_MAX_LEN))
			return ImageTexture.create_from_image(image)

	var ffmpeg_bin := await _ensure_ffmpeg()
	if ffmpeg_bin.is_empty():
		return null

	var gif_path := globalized_dir + "/_tmp_%s.gif" % url_hash

	var body := await _download_url_bytes(url)
	if body.is_empty():
		return null

	var gif_file := FileAccess.open(gif_path, FileAccess.WRITE)
	if gif_file == null:
		logger.warn("GIF fetch: cannot write temp gif")
		return null
	gif_file.store_buffer(body)
	gif_file.close()

	# Extract first frame via ffmpeg.
	var ffmpeg_args: PackedStringArray = [
		"-y", "-i", gif_path,
		"-frames:v", "1",
		"-f", "image2", cached_png,
	]
	var ret: Array = []
	OS.execute(ffmpeg_bin, ffmpeg_args, ret)

	DirAccess.remove_absolute(gif_path)

	if not FileAccess.file_exists(cached_png):
		logger.warn("GIF fetch: ffmpeg did not produce output for %s" % url.left(LOG_URL_MAX_LEN))
		return null

	var image := Image.load_from_file(cached_png)
	if image == null:
		logger.warn("GIF fetch: failed to load extracted PNG")
		return null

	logger.info("GIF fetch: extracted first frame from %s" % url.left(LOG_URL_MAX_LEN))
	return ImageTexture.create_from_image(image)


## Combines two textures side-by-side into a single banner image.
func _combine_side_by_side(left: Texture2D, right: Texture2D) -> Texture2D:
	var img_a := left.get_image()
	var img_b := right.get_image()
	if img_a == null or img_b == null:
		return left

	var target_h: int = max(img_a.get_height(), img_b.get_height())
	if img_a.get_height() != target_h:
		var scale: float = float(target_h) / float(img_a.get_height())
		img_a.resize(int(img_a.get_width() * scale), target_h, Image.INTERPOLATE_BILINEAR)
	if img_b.get_height() != target_h:
		var scale: float = float(target_h) / float(img_b.get_height())
		img_b.resize(int(img_b.get_width() * scale), target_h, Image.INTERPOLATE_BILINEAR)

	var combined := Image.create(img_a.get_width() + img_b.get_width(), target_h, false, img_a.get_format())
	combined.blit_rect(img_a, Rect2i(0, 0, img_a.get_width(), target_h), Vector2i.ZERO)
	combined.blit_rect(img_b, Rect2i(0, 0, img_b.get_width(), target_h), Vector2i(img_a.get_width(), 0))

	return ImageTexture.create_from_image(combined)


## Scrapes the itch.io game page for screenshots.
func _fetch_game_page_metadata(game: Dictionary) -> Dictionary:
	var page_url: String = game.get("url", "")
	if page_url.is_empty():
		logger.warn("No page URL in game dict")
		return {}

	logger.info("Scraping page for screenshots: " + page_url)
	var body := await _download_url_bytes(page_url)
	if body.is_empty():
		return {}
	var html := body.get_string_from_utf8()
	if html.is_empty():
		logger.warn("Empty HTML response for: " + page_url)
		return {}
	if _is_cloudflare_challenge(html):
		logger.warn("Cloudflare challenge detected for: %s — falling back to cover art" % page_url)
		return {}
	return _extract_screenshots_from_html(html, page_url)


func _is_cloudflare_challenge(html: String) -> bool:
	return html.find("challenge-platform") != -1 or html.find("cf-browser-verification") != -1


func _extract_screenshots_from_html(html: String, page_url: String) -> Dictionary:
	var shot_re := RegEx.new()
	shot_re.compile(SCREENSHOT_REGEX_PATTERN)
	var screenshots: Array = []
	for m in shot_re.search_all(html):
		var url: String = _sanitize_url(m.get_string(0))
		screenshots.append(url)
	logger.info("Found %d screenshots for page %s" % [screenshots.size(), page_url])
	return {"screenshots": screenshots}


func _load_enriched_meta() -> Dictionary:
	var raw: Variant = Cache.get_json(_meta_cache_dir, META_CACHE_FILE)
	if typeof(raw) != TYPE_DICTIONARY:
		return {}
	var out := {}
	for key in raw:
		out[int(key)] = raw[key]
	return out


func _save_enriched_meta(game_id: int, meta: Dictionary) -> void:
	var raw: Variant = Cache.get_json(_meta_cache_dir, META_CACHE_FILE)
	if typeof(raw) != TYPE_DICTIONARY:
		raw = {}
	raw[str(game_id)] = meta
	Cache.save_json(_meta_cache_dir, META_CACHE_FILE, raw)