1 package net.pterodactylus.sone.core
3 import com.google.common.base.Ticker
4 import com.google.common.cache.Cache
5 import com.google.common.cache.CacheBuilder
6 import freenet.keys.FreenetURI
8 import org.jsoup.nodes.Document
9 import org.jsoup.nodes.TextNode
10 import java.io.ByteArrayInputStream
11 import java.net.URLDecoder
12 import java.nio.charset.Charset
13 import java.text.Normalizer
14 import java.util.concurrent.TimeUnit.MINUTES
15 import javax.activation.MimeType
16 import javax.imageio.ImageIO
17 import javax.inject.Inject
20 * [ElementLoader] implementation that uses a simple Guava [com.google.common.cache.Cache].
22 class DefaultElementLoader(private val freenetInterface: FreenetInterface, ticker: Ticker): ElementLoader {
24 @Inject constructor(freenetInterface: FreenetInterface): this(freenetInterface, Ticker.systemTicker())
26 private val loadingLinks: Cache<String, Boolean> = CacheBuilder.newBuilder().build()
27 private val failureCache: Cache<String, Boolean> = CacheBuilder.newBuilder().ticker(ticker).expireAfterWrite(30, MINUTES).build()
28 private val elementCache: Cache<String, LinkedElement> = CacheBuilder.newBuilder().build()
29 private val callback = object: FreenetInterface.BackgroundFetchCallback {
30 override fun shouldCancel(uri: FreenetURI, mimeType: String, size: Long): Boolean {
31 return (size > 2097152) || (!mimeType.startsWith("image/") && !mimeType.startsWith("text/html"))
34 override fun loaded(uri: FreenetURI, mimeTypeText: String, data: ByteArray) {
35 MimeType(mimeTypeText).also { mimeType ->
37 mimeType.primaryType == "image" -> {
38 ByteArrayInputStream(data).use {
41 elementCache.get(uri.toString().decode().normalize()) {
42 LinkedElement(uri.toString(), properties = mapOf("type" to "image", "size" to data.size, "sizeHuman" to data.size.human))
46 mimeType.baseType == "text/html" -> {
47 val document = Jsoup.parse(data.toString(Charset.forName(mimeType.getParameter("charset") ?: "UTF-8")))
48 elementCache.get(uri.toString().decode().normalize()) {
49 LinkedElement(uri.toString(), properties = mapOf(
50 "type" to "html", "size" to data.size, "sizeHuman" to data.size.human,
51 "title" to document.title().emptyToNull,
52 "description" to (document.metaDescription ?: document.firstNonHeadingParagraph)
57 removeLoadingLink(uri)
61 private val String?.emptyToNull get() = if (this == "") null else this
63 private val Document.metaDescription: String?
64 get() = head().getElementsByTag("meta")
65 .map { it.attr("name") to it.attr("content") }
66 .firstOrNull { it.first == "description" }
69 private val Document.firstNonHeadingParagraph: String?
70 get() = body().children()
71 .filter { it.children().all { it is TextNode } }
72 .map { it to it.text() }
73 .filterNot { it.second == "" }
74 .firstOrNull { !it.first.tagName().startsWith("h", ignoreCase = true) }
77 private val Int.human get() = when (this) {
78 in 0..1023 -> "$this B"
79 in 1024..1048575 -> "${this / 1024} KiB"
80 in 1048576..1073741823 -> "${this / 1048576} MiB"
81 else -> "${this / 1073741824} GiB"
84 override fun failed(uri: FreenetURI) {
85 failureCache.put(uri.toString().decode().normalize(), true)
86 removeLoadingLink(uri)
89 private fun removeLoadingLink(uri: FreenetURI) {
90 synchronized(loadingLinks) {
91 loadingLinks.invalidate(uri.toString().decode().normalize())
96 override fun loadElement(link: String): LinkedElement {
97 val normalizedLink = link.decode().normalize()
98 synchronized(loadingLinks) {
99 elementCache.getIfPresent(normalizedLink)?.run {
102 failureCache.getIfPresent(normalizedLink)?.run {
103 return LinkedElement(link, failed = true)
105 if (loadingLinks.getIfPresent(normalizedLink) == null) {
106 loadingLinks.put(normalizedLink, true)
107 freenetInterface.startFetch(FreenetURI(link), callback)
110 return LinkedElement(link, loading = true)
113 private fun String.decode() = URLDecoder.decode(this, "UTF-8")!!
114 private fun String.normalize() = Normalizer.normalize(this, Normalizer.Form.NFC)!!