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 java.io.ByteArrayInputStream
  10 import java.net.URLDecoder
  11 import java.nio.charset.Charset
  12 import java.text.Normalizer
  13 import java.util.concurrent.TimeUnit.MINUTES
  14 import javax.activation.MimeType
  15 import javax.imageio.ImageIO
  16 import javax.inject.Inject
  17
  18 /**
  19  * [ElementLoader] implementation that uses a simple Guava [com.google.common.cache.Cache].
  20  */
  21 class DefaultElementLoader(private val freenetInterface: FreenetInterface, ticker: Ticker): ElementLoader {
  22
  23         @Inject constructor(freenetInterface: FreenetInterface): this(freenetInterface, Ticker.systemTicker())
  24
  25         private val loadingLinks: Cache<String, Boolean> = CacheBuilder.newBuilder().build()
  26         private val failureCache: Cache<String, Boolean> = CacheBuilder.newBuilder().ticker(ticker).expireAfterWrite(30, MINUTES).build()
  27         private val elementCache: Cache<String, LinkedElement> = CacheBuilder.newBuilder().build()
  28         private val callback = object: FreenetInterface.BackgroundFetchCallback {
  29                 override fun shouldCancel(uri: FreenetURI, mimeType: String, size: Long): Boolean {
  30                         return (size > 2097152) || (!mimeType.startsWith("image/") && !mimeType.startsWith("text/html"))
  31                 }
  32
  33                 override fun loaded(uri: FreenetURI, mimeTypeText: String, data: ByteArray) {
  34                         MimeType(mimeTypeText).also { mimeType ->
  35                                 when {
  36                                         mimeType.primaryType == "image" -> {
  37                                                 ByteArrayInputStream(data).use {
  38                                                         ImageIO.read(it)
  39                                                 }?.let {
  40                                                         elementCache.get(uri.toString().decode().normalize()) {
  41                                                                 LinkedElement(uri.toString(), properties = mapOf("type" to "image", "size" to data.size, "sizeHuman" to data.size.human))
  42                                                         }
  43                                                 }
  44                                         }
  45                                         mimeType.baseType == "text/html" -> {
  46                                                 val document = Jsoup.parse(data.toString(Charset.forName(mimeType.getParameter("charset") ?: "UTF-8")))
  47                                                 elementCache.get(uri.toString().decode().normalize()) {
  48                                                         LinkedElement(uri.toString(), properties = mapOf(
  49                                                                         "type" to "html", "size" to data.size, "sizeHuman" to data.size.human,
  50                                                                         "title" to document.title().emptyToNull,
  51                                                                         "description" to (document.metaDescription ?: document.firstNonHeadingParagraph)
  52                                                         ))
  53                                                 }
  54                                         }
  55                                 }
  56                                 removeLoadingLink(uri)
  57                         }
  58                 }
  59
  60                 override fun failed(uri: FreenetURI) {
  61                         failureCache.put(uri.toString().decode().normalize(), true)
  62                         removeLoadingLink(uri)
  63                 }
  64
  65                 private fun removeLoadingLink(uri: FreenetURI) {
  66                         synchronized(loadingLinks) {
  67                                 loadingLinks.invalidate(uri.toString().decode().normalize())
  68                         }
  69                 }
  70         }
  71
  72         override fun loadElement(link: String): LinkedElement {
  73                 val normalizedLink = link.decode().normalize()
  74                 synchronized(loadingLinks) {
  75                         elementCache.getIfPresent(normalizedLink)?.run {
  76                                 return this
  77                         }
  78                         failureCache.getIfPresent(normalizedLink)?.run {
  79                                 return LinkedElement(link, failed = true)
  80                         }
  81                         if (loadingLinks.getIfPresent(normalizedLink) == null) {
  82                                 loadingLinks.put(normalizedLink, true)
  83                                 freenetInterface.startFetch(FreenetURI(link), callback)
  84                         }
  85                 }
  86                 return LinkedElement(link, loading = true)
  87         }
  88
  89 }
  90
  91 private fun String.decode() = URLDecoder.decode(this, "UTF-8")!!
  92 private fun String.normalize() = Normalizer.normalize(this, Normalizer.Form.NFC)!!
  93 private val String?.emptyToNull get() = if (this == "") null else this
  94
  95 private val Document.metaDescription: String?
  96         get() = head().getElementsByTag("meta")
  97                 .map { it.attr("name") to it.attr("content") }
  98                 .firstOrNull { it.first == "description" }
  99                 ?.second
 100
 101 private val Document.firstNonHeadingParagraph: String?
 102         get() = body().select("div, p")
 103                 .filter { it.textNodes().isNotEmpty() }
 104                 .map { it to it.text() }
 105                 .firstOrNull { it.second != "" }
 106                 ?.second
 107
 108 private val Int.human get() = when (this) {
 109         in 0..1023 -> "$this B"
 110         in 1024..1048575 -> "${this / 1024} KiB"
 111         in 1048576..1073741823 -> "${this / 1048576} MiB"
 112         else -> "${this / 1073741824} GiB"
 113 }