src/main/kotlin/net/pterodactylus/sone/core/DefaultElementLoader.kt

   1 package net.pterodactylus.sone.core
   2
   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
   7 import org.jsoup.Jsoup
   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
  18
  19 /**
  20  * [ElementLoader] implementation that uses a simple Guava [com.google.common.cache.Cache].
  21  */
  22 class DefaultElementLoader(private val freenetInterface: FreenetInterface, ticker: Ticker): ElementLoader {
  23
  24         @Inject constructor(freenetInterface: FreenetInterface): this(freenetInterface, Ticker.systemTicker())
  25
  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"))
  32                 }
  33
  34                 override fun loaded(uri: FreenetURI, mimeTypeText: String, data: ByteArray) {
  35                         MimeType(mimeTypeText).also { mimeType ->
  36                                 when {
  37                                         mimeType.primaryType == "image" -> {
  38                                                 ByteArrayInputStream(data).use {
  39                                                         ImageIO.read(it)
  40                                                 }?.let {
  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))
  43                                                         }
  44                                                 }
  45                                         }
  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)
  53                                                         ))
  54                                                 }
  55                                         }
  56                                 }
  57                                 removeLoadingLink(uri)
  58                         }
  59                 }
  60
  61                 override fun failed(uri: FreenetURI) {
  62                         failureCache.put(uri.toString().decode().normalize(), true)
  63                         removeLoadingLink(uri)
  64                 }
  65
  66                 private fun removeLoadingLink(uri: FreenetURI) {
  67                         synchronized(loadingLinks) {
  68                                 loadingLinks.invalidate(uri.toString().decode().normalize())
  69                         }
  70                 }
  71         }
  72
  73         override fun loadElement(link: String): LinkedElement {
  74                 val normalizedLink = link.decode().normalize()
  75                 synchronized(loadingLinks) {
  76                         elementCache.getIfPresent(normalizedLink)?.run {
  77                                 return this
  78                         }
  79                         failureCache.getIfPresent(normalizedLink)?.run {
  80                                 return LinkedElement(link, failed = true)
  81                         }
  82                         if (loadingLinks.getIfPresent(normalizedLink) == null) {
  83                                 loadingLinks.put(normalizedLink, true)
  84                                 freenetInterface.startFetch(FreenetURI(link), callback)
  85                         }
  86                 }
  87                 return LinkedElement(link, loading = true)
  88         }
  89
  90 }
  91
  92 private fun String.decode() = URLDecoder.decode(this, "UTF-8")!!
  93 private fun String.normalize() = Normalizer.normalize(this, Normalizer.Form.NFC)!!
  94 private val String?.emptyToNull get() = if (this == "") null else this
  95
  96 private val Document.metaDescription: String?
  97         get() = head().getElementsByTag("meta")
  98                 .map { it.attr("name") to it.attr("content") }
  99                 .firstOrNull { it.first == "description" }
 100                 ?.second
 101
 102 private val Document.firstNonHeadingParagraph: String?
 103         get() = body().children()
 104                 .filter { it.children().all { it is TextNode } }
 105                 .map { it to it.text() }
 106                 .filterNot { it.second == "" }
 107                 .firstOrNull { !it.first.tagName().startsWith("h", ignoreCase = true) }
 108                 ?.second
 109
 110 private val Int.human get() = when (this) {
 111         in 0..1023 -> "$this B"
 112         in 1024..1048575 -> "${this / 1024} KiB"
 113         in 1048576..1073741823 -> "${this / 1048576} MiB"
 114         else -> "${this / 1073741824} GiB"
 115 }