X-Git-Url: https://git.pterodactylus.net/?p=Sone.git;a=blobdiff_plain;f=src%2Fmain%2Fkotlin%2Fnet%2Fpterodactylus%2Fsone%2Fcore%2FDefaultElementLoader.kt;h=f553ae84d2ac87ef41cc4e6fc7334f111e4f0ea6;hp=67d55387a510e7ad3715f138df93b6d9ec8bd16e;hb=71d9b57ce88771258d2d3678b90bb5a55c6e4d9e;hpb=5a1d1f95e7d122752a405bf1730e67cafb0df177 diff --git a/src/main/kotlin/net/pterodactylus/sone/core/DefaultElementLoader.kt b/src/main/kotlin/net/pterodactylus/sone/core/DefaultElementLoader.kt index 67d5538..f553ae8 100644 --- a/src/main/kotlin/net/pterodactylus/sone/core/DefaultElementLoader.kt +++ b/src/main/kotlin/net/pterodactylus/sone/core/DefaultElementLoader.kt @@ -4,10 +4,15 @@ import com.google.common.base.Ticker import com.google.common.cache.Cache import com.google.common.cache.CacheBuilder import freenet.keys.FreenetURI +import org.jsoup.Jsoup +import org.jsoup.nodes.Document +import org.jsoup.nodes.TextNode import java.io.ByteArrayInputStream import java.net.URLDecoder +import java.nio.charset.Charset import java.text.Normalizer import java.util.concurrent.TimeUnit.MINUTES +import javax.activation.MimeType import javax.imageio.ImageIO import javax.inject.Inject @@ -20,26 +25,55 @@ class DefaultElementLoader(private val freenetInterface: FreenetInterface, ticke private val loadingLinks: Cache = CacheBuilder.newBuilder().build() private val failureCache: Cache = CacheBuilder.newBuilder().ticker(ticker).expireAfterWrite(30, MINUTES).build() - private val imageCache: Cache = CacheBuilder.newBuilder().build() + private val elementCache: Cache = CacheBuilder.newBuilder().build() private val callback = object : FreenetInterface.BackgroundFetchCallback { override fun shouldCancel(uri: FreenetURI, mimeType: String, size: Long): Boolean { - return !mimeType.startsWith("image/") || (size > 2097152) + return (size > 2097152) || (!mimeType.startsWith("image/") && !mimeType.startsWith("text/html")) } override fun loaded(uri: FreenetURI, mimeType: String, data: ByteArray) { - if (!mimeType.startsWith("image/")) { - return - } - ByteArrayInputStream(data).use { - ImageIO.read(it) - }?.let { - imageCache.get(uri.toString().decode().normalize()) { - LinkedElement(uri.toString(), properties = mapOf("size" to data.size, "sizeHuman" to data.size.human)) + MimeType(mimeType).also { mimeType -> + when { + mimeType.primaryType == "image" -> { + ByteArrayInputStream(data).use { + ImageIO.read(it) + }?.let { + elementCache.get(uri.toString().decode().normalize()) { + LinkedElement(uri.toString(), properties = mapOf("size" to data.size, "sizeHuman" to data.size.human)) + } + } + } + mimeType.baseType == "text/html" -> { + val document = Jsoup.parse(data.toString(Charset.forName(mimeType.getParameter("charset") ?: "UTF-8"))) + elementCache.get(uri.toString().decode().normalize()) { + LinkedElement(uri.toString(), properties = mapOf( + "size" to data.size, "sizeHuman" to data.size.human, + "title" to document.title().emptyToNull, + "description" to (document.metaDescription ?: document.firstNonHeadingParagraph) + )) + } + } } + removeLoadingLink(uri) } - removeLoadingLink(uri) } + private val String?.emptyToNull get() = if (this == "") null else this + + private val Document.metaDescription: String? + get() = head().getElementsByTag("meta") + .map { it.attr("name") to it.attr("content") } + .firstOrNull { it.first == "description" } + ?.second + + private val Document.firstNonHeadingParagraph: String? + get() = body().children() + .filter { it.children().all { it is TextNode } } + .map { it to it.text() } + .filterNot { it.second == "" } + .firstOrNull { !it.first.tagName().startsWith("h", ignoreCase = true) } + ?.second + private val Int.human get() = when (this) { in 0..1023 -> "$this B" in 1024..1048575 -> "${this / 1024} KiB" @@ -62,7 +96,7 @@ class DefaultElementLoader(private val freenetInterface: FreenetInterface, ticke override fun loadElement(link: String): LinkedElement { val normalizedLink = link.decode().normalize() synchronized(loadingLinks) { - imageCache.getIfPresent(normalizedLink)?.run { + elementCache.getIfPresent(normalizedLink)?.run { return this } failureCache.getIfPresent(normalizedLink)?.run {