Use canonical EPUB HREFs

readium/navigator/src/main/java/org/readium/r2/navigator/epub/HtmlInjector.kt

@@ -15,6 +15,7 @@ import org.readium.r2.shared.publication.services.isProtected
 import org.readium.r2.shared.util.AbsoluteUrl
 import org.readium.r2.shared.util.Try
 import org.readium.r2.shared.util.Url
+import org.readium.r2.shared.util.data.ReadError
 import org.readium.r2.shared.util.mediatype.MediaType
 import org.readium.r2.shared.util.resource.Resource
 import org.readium.r2.shared.util.resource.TransformingResource
@@ -42,7 +43,12 @@ internal fun Resource.injectHtml(
         val injectables = mutableListOf<String>()
 
         if (publication.metadata.presentation.layout == EpubLayout.REFLOWABLE) {
-            content = css.injectHtml(content)
+            content = try {
+                css.injectHtml(content)
+            } catch (e: Exception) {
+                return@TransformingResource Try.failure(ReadError.Decoding(e))
+            }
+
             injectables.add(
                 script(
                     baseHref.resolve(Url("readium/scripts/readium-reflowable.js")!!)

readium/shared/src/main/java/org/readium/r2/shared/extensions/String.kt

@@ -101,3 +101,9 @@ internal fun String.percentEncodedPath(): String =
  */
 internal fun String.percentEncodedQuery(): String =
     Uri.encode(this, "$+,/?:=@")
+
+/**
+ * Returns whether the String receiver contains only printable ASCII characters.
+ */
+internal fun String.isPrintableAscii(): Boolean =
+    all { it.code in 0x20..0x7F }

readium/shared/src/main/java/org/readium/r2/shared/publication/Locator.kt

@@ -19,6 +19,7 @@ import org.readium.r2.shared.JSONable
 import org.readium.r2.shared.extensions.*
 import org.readium.r2.shared.toJSON
 import org.readium.r2.shared.util.Url
+import org.readium.r2.shared.util.fromLegacyHref
 import org.readium.r2.shared.util.logging.WarningLogger
 import org.readium.r2.shared.util.logging.log
 import org.readium.r2.shared.util.mediatype.MediaType

readium/shared/src/main/java/org/readium/r2/shared/publication/epub/EpubEncryptionParser.kt

@@ -10,6 +10,7 @@ import org.readium.r2.shared.InternalReadiumApi
 import org.readium.r2.shared.publication.encryption.Encryption
 import org.readium.r2.shared.publication.protection.ContentProtection
 import org.readium.r2.shared.util.Url
+import org.readium.r2.shared.util.fromEpubHref
 import org.readium.r2.shared.util.xml.ElementNode
 
 @InternalReadiumApi
@@ -29,7 +30,7 @@ public object EpubEncryptionParser {
     private fun parseEncryptedData(node: ElementNode): Pair<Url, Encryption>? {
         val resourceURI = node.getFirst("CipherData", Namespaces.ENC)
             ?.getFirst("CipherReference", Namespaces.ENC)?.getAttr("URI")
-            ?.let { Url(it) ?: Url.fromDecodedPath(it) }
+            ?.let { Url.fromEpubHref(it) }
             ?: return null
         val retrievalMethod = node.getFirst("KeyInfo", Namespaces.SIG)
             ?.getFirst("RetrievalMethod", Namespaces.SIG)?.getAttr("URI")

readium/shared/src/main/java/org/readium/r2/shared/util/Url.kt

@@ -15,6 +15,7 @@ import java.net.URL
 import kotlinx.parcelize.Parcelize
 import org.readium.r2.shared.DelicateReadiumApi
 import org.readium.r2.shared.InternalReadiumApi
+import org.readium.r2.shared.extensions.isPrintableAscii
 import org.readium.r2.shared.extensions.percentEncodedPath
 import org.readium.r2.shared.extensions.tryOrNull
 
@@ -43,19 +44,6 @@ public sealed class Url : Parcelable {
             return invoke(Uri.parse(url))
         }
 
-        /**
-         * Creates an [Url] from a legacy HREF.
-         *
-         * For example, if it is a relative path such as `/dir/my chapter.html`, it will be
-         * converted to the valid relative URL `dir/my%20chapter.html`.
-         *
-         * Only use this API when you are upgrading to Readium 3.x and migrating the HREFs stored in
-         * your database. See the 3.0 migration guide for more information.
-         */
-        @DelicateReadiumApi
-        public fun fromLegacyHref(href: String): Url? =
-            AbsoluteUrl(href) ?: fromDecodedPath(href.removePrefix("/"))
-
         internal operator fun invoke(uri: Uri): Url? =
             if (uri.isAbsolute) {
                 AbsoluteUrl(uri)
@@ -306,6 +294,32 @@ public class RelativeUrl private constructor(override val uri: Uri) : Url() {
     }
 }
 
+/**
+ * Creates an [Url] from a legacy HREF.
+ *
+ * For example, if it is a relative path such as `/dir/my chapter.html`, it will be
+ * converted to the valid relative URL `dir/my%20chapter.html`.
+ *
+ * Only use this API when you are upgrading to Readium 3.x and migrating the HREFs stored in
+ * your database. See the 3.0 migration guide for more information.
+ */
+@DelicateReadiumApi
+public fun Url.Companion.fromLegacyHref(href: String): Url? =
+    AbsoluteUrl(href) ?: Url.fromDecodedPath(href.removePrefix("/"))
+
+/**
+ * According to the EPUB specification, the HREFs in the EPUB package must be valid URLs (so
+ * percent-encoded). Unfortunately, many EPUBs don't follow this rule, and use invalid HREFs such
+ * as `my chapter.html` or `/dir/my chapter.html`.
+ *
+ * As a workaround, we assume the HREFs are valid percent-encoded URLs, and fallback to decoded paths
+ * if we can't parse the URL.
+ */
+@InternalReadiumApi
+public fun Url.Companion.fromEpubHref(href: String): Url? {
+    return (Url(href) ?: Url.fromDecodedPath(href))
+}
+
 public fun File.toUrl(): AbsoluteUrl =
     checkNotNull(AbsoluteUrl(Uri.fromFile(this)))
 
@@ -353,8 +367,9 @@ private fun Uri.addFileAuthority(): Uri =
     }
 
 private fun String.isValidUrl(): Boolean =
-    // Uri.parse doesn't really validate the URL, it could contain invalid characters.
-    isNotBlank() && tryOrNull { URI(this) } != null
+    // Uri.parse doesn't really validate the URL, it could contain invalid characters, so we use
+    // URI. However, URI allows some non-ASCII characters.
+    isNotBlank() && isPrintableAscii() && tryOrNull { URI(this) } != null
 
 @JvmInline
 public value class FileExtension(

readium/shared/src/test/java/org/readium/r2/shared/extensions/StringTest.kt

@@ -0,0 +1,22 @@
+/*
+ * Copyright 2024 Readium Foundation. All rights reserved.
+ * Use of this source code is governed by the BSD-style license
+ * available in the top-level LICENSE file of the project.
+ */
+
+package org.readium.r2.shared.extensions
+
+import kotlin.test.assertFalse
+import kotlin.test.assertTrue
+import org.junit.Test
+
+class StringTest {
+
+    @Test
+    fun `checks if a string is made of printable ASCII characters`() {
+        assertTrue("".isPrintableAscii())
+        assertTrue(" foo/@bar".isPrintableAscii())
+        assertFalse("école".isPrintableAscii())
+        assertFalse("\u0001 non printable".isPrintableAscii())
+    }
+}

readium/shared/src/test/java/org/readium/r2/shared/util/UrlTest.kt

@@ -23,12 +23,15 @@ class UrlTest {
         assertNull(Url(""))
         assertNull(Url("     "))
         assertNull(Url("invalid character"))
+        assertNull(Url("école"))
 
         assertNull(AbsoluteUrl("  "))
         assertNull(AbsoluteUrl("invalid character"))
+        assertNull(AbsoluteUrl("école"))
 
         assertNull(RelativeUrl("   "))
         assertNull(RelativeUrl("invalid character"))
+        assertNull(RelativeUrl("école"))
     }
 
     @Test

readium/streamer/src/main/java/org/readium/r2/streamer/parser/epub/EpubParser.kt

@@ -27,6 +27,7 @@ import org.readium.r2.shared.util.data.decodeXml
 import org.readium.r2.shared.util.data.readDecodeOrElse
 import org.readium.r2.shared.util.data.readDecodeOrNull
 import org.readium.r2.shared.util.format.EpubSpecification
+import org.readium.r2.shared.util.fromEpubHref
 import org.readium.r2.shared.util.getOrElse
 import org.readium.r2.shared.util.logging.WarningLogger
 import org.readium.r2.shared.util.mediatype.MediaType
@@ -35,7 +36,6 @@ import org.readium.r2.shared.util.resource.TransformingContainer
 import org.readium.r2.shared.util.use
 import org.readium.r2.shared.util.xml.ElementNode
 import org.readium.r2.streamer.parser.PublicationParser
-import org.readium.r2.streamer.parser.epub.extensions.fromEpubHref
 
 /**
  * Parses a Publication from an EPUB publication.

readium/streamer/src/main/java/org/readium/r2/streamer/parser/epub/MetadataParser.kt

@@ -8,9 +8,9 @@ package org.readium.r2.streamer.parser.epub
 
 import org.readium.r2.shared.publication.Href
 import org.readium.r2.shared.util.Url
+import org.readium.r2.shared.util.fromEpubHref
 import org.readium.r2.shared.util.mediatype.MediaType
 import org.readium.r2.shared.util.xml.ElementNode
-import org.readium.r2.streamer.parser.epub.extensions.fromEpubHref
 
 internal class MetadataParser(
     private val prefixMap: Map<String, String>

readium/streamer/src/main/java/org/readium/r2/streamer/parser/epub/NavigationDocumentParser.kt

@@ -8,8 +8,8 @@ package org.readium.r2.streamer.parser.epub
 
 import org.readium.r2.shared.publication.Link
 import org.readium.r2.shared.util.Url
+import org.readium.r2.shared.util.fromEpubHref
 import org.readium.r2.shared.util.xml.ElementNode
-import org.readium.r2.streamer.parser.epub.extensions.fromEpubHref
 
 internal object NavigationDocumentParser {
 

readium/streamer/src/main/java/org/readium/r2/streamer/parser/epub/NcxParser.kt

@@ -8,8 +8,8 @@ package org.readium.r2.streamer.parser.epub
 
 import org.readium.r2.shared.publication.Link
 import org.readium.r2.shared.util.Url
+import org.readium.r2.shared.util.fromEpubHref
 import org.readium.r2.shared.util.xml.ElementNode
-import org.readium.r2.streamer.parser.epub.extensions.fromEpubHref
 
 internal object NcxParser {
 

readium/streamer/src/main/java/org/readium/r2/streamer/parser/epub/PackageDocument.kt

@@ -8,8 +8,8 @@ package org.readium.r2.streamer.parser.epub
 
 import org.readium.r2.shared.publication.ReadingProgression
 import org.readium.r2.shared.util.Url
+import org.readium.r2.shared.util.fromEpubHref
 import org.readium.r2.shared.util.xml.ElementNode
-import org.readium.r2.streamer.parser.epub.extensions.fromEpubHref
 
 internal data class PackageDocument(
     val path: Url,

readium/streamer/src/main/java/org/readium/r2/streamer/parser/epub/SmilParser.kt

@@ -9,8 +9,8 @@ package org.readium.r2.streamer.parser.epub
 import org.readium.r2.shared.MediaOverlayNode
 import org.readium.r2.shared.MediaOverlays
 import org.readium.r2.shared.util.Url
+import org.readium.r2.shared.util.fromEpubHref
 import org.readium.r2.shared.util.xml.ElementNode
-import org.readium.r2.streamer.parser.epub.extensions.fromEpubHref
 
 internal object SmilParser {
     /* According to https://www.w3.org/publishing/epub3/epub-mediaoverlays.html#sec-overlays-content-conf