X-Git-Url: https://git.pterodactylus.net/?a=blobdiff_plain;f=src%2Fmain%2Fkotlin%2Fnet%2Fpterodactylus%2Fsone%2Fcore%2FDefaultElementLoader.kt;h=e497fdcb18f1493bba9f14d56a46f1b7f5518ac9;hb=refs%2Fheads%2Fnext;hp=ebdae49b4b9c22d57b455c96f0bc15ff4c36157e;hpb=27a0934b7e68315756020e0de0bc2564355be90a;p=Sone.git

diff --git a/src/main/kotlin/net/pterodactylus/sone/core/DefaultElementLoader.kt b/src/main/kotlin/net/pterodactylus/sone/core/DefaultElementLoader.kt
index ebdae49..e497fdc 100644
--- a/src/main/kotlin/net/pterodactylus/sone/core/DefaultElementLoader.kt
+++ b/src/main/kotlin/net/pterodactylus/sone/core/DefaultElementLoader.kt
@@ -6,12 +6,12 @@ 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 java.util.logging.Logger
 import javax.activation.MimeType
 import javax.imageio.ImageIO
 import javax.inject.Inject
@@ -23,23 +23,35 @@ class DefaultElementLoader(private val freenetInterface: FreenetInterface, ticke
 
 	@Inject constructor(freenetInterface: FreenetInterface): this(freenetInterface, Ticker.systemTicker())
 
-	private val loadingLinks: Cache<String, Boolean> = CacheBuilder.newBuilder().build<String, Boolean>()
-	private val failureCache: Cache<String, Boolean> = CacheBuilder.newBuilder().ticker(ticker).expireAfterWrite(30, MINUTES).build<String, Boolean>()
-	private val elementCache: Cache<String, LinkedElement> = CacheBuilder.newBuilder().build<String, LinkedElement>()
+	private val loadingLinks: Cache<String, Boolean> = CacheBuilder.newBuilder().build()
+	private val failureCache: Cache<String, Boolean> = CacheBuilder.newBuilder().ticker(ticker).expireAfterWrite(30, MINUTES).build()
+	private val elementCache: Cache<String, LinkedElement> = CacheBuilder.newBuilder().build()
+	private val logger = Logger.getLogger(DefaultElementLoader::class.qualifiedName)
 	private val callback = object: FreenetInterface.BackgroundFetchCallback {
 		override fun shouldCancel(uri: FreenetURI, mimeType: String, size: Long): Boolean {
-			return (size > 2097152) || (!mimeType.startsWith("image/") && !mimeType.startsWith("text/html"))
+			if (size > 2097152) {
+				logger.fine { "Canceling download of $uri because it’s > 2 MiB." }
+				return true
+			}
+			if (!mimeType.startsWith("image/") && !mimeType.startsWith("text/html")) {
+				logger.fine { "Canceling download of $uri because of its MIME type, $mimeType." }
+				return true
+			}
+			return false
 		}
 
-		override fun loaded(uri: FreenetURI, mimeType: String, data: ByteArray) {
-			MimeType(mimeType).also { mimeType ->
+		override fun loaded(uri: FreenetURI, mimeTypeText: String, data: ByteArray) {
+			MimeType(mimeTypeText).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))
+								LinkedElement(uri.toString(), properties = mapOf("type" to "image", "size" to data.size, "sizeHuman" to data.size.human))
+									.apply {
+										logger.fine("Downloaded image from $link: size=${properties["size"]}.")
+									}
 							}
 						}
 					}
@@ -47,10 +59,12 @@ class DefaultElementLoader(private val freenetInterface: FreenetInterface, ticke
 						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,
+									"type" to "html", "size" to data.size, "sizeHuman" to data.size.human,
 									"title" to document.title().emptyToNull,
 									"description" to (document.metaDescription ?: document.firstNonHeadingParagraph)
-							))
+							)).apply {
+								logger.fine { "Extracted information from $link: title=${properties["title"]}, description=${properties["description"]}." }
+							}
 						}
 					}
 				}
@@ -58,32 +72,10 @@ class DefaultElementLoader(private val freenetInterface: FreenetInterface, ticke
 			}
 		}
 
-		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"
-			in 1048576..1073741823 -> "${this / 1048576} MiB"
-			else -> "${this / 1073741824} GiB"
-		}
-
 		override fun failed(uri: FreenetURI) {
 			failureCache.put(uri.toString().decode().normalize(), true)
 			removeLoadingLink(uri)
+			logger.fine { "Download failed for $uri." }
 		}
 
 		private fun removeLoadingLink(uri: FreenetURI) {
@@ -110,7 +102,28 @@ class DefaultElementLoader(private val freenetInterface: FreenetInterface, ticke
 		return LinkedElement(link, loading = true)
 	}
 
-	private fun String.decode() = URLDecoder.decode(this, "UTF-8")!!
-	private fun String.normalize() = Normalizer.normalize(this, Normalizer.Form.NFC)!!
+}
+
+private fun String.decode() = try { URLDecoder.decode(this, "UTF-8")!! } catch (e: RuntimeException) { freenet.support.Logger.error(DefaultElementLoader::class.java, "Could not decode %s!".format(this), e); throw e }
+private fun String.normalize() = Normalizer.normalize(this, Normalizer.Form.NFC)!!
+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().select("div, p")
+		.filter { it.textNodes().isNotEmpty() }
+		.map { it to it.text() }
+		.firstOrNull { it.second != "" }
+		?.second
 
+private val Int.human get() = when (this) {
+	in 0..1023 -> "$this B"
+	in 1024..1048575 -> "${this / 1024} KiB"
+	in 1048576..1073741823 -> "${this / 1048576} MiB"
+	else -> "${this / 1073741824} GiB"
 }