Fix .onion link linkification.

Fixes #11458.

app/src/main/java/org/thoughtcrime/securesms/conversation/ConversationItem.java

@@ -106,7 +106,6 @@ import org.thoughtcrime.securesms.jobs.MmsSendJob;
 import org.thoughtcrime.securesms.jobs.SmsSendJob;
 import org.thoughtcrime.securesms.keyvalue.SignalStore;
 import org.thoughtcrime.securesms.linkpreview.LinkPreview;
-import org.thoughtcrime.securesms.linkpreview.LinkPreviewUtil;
 import org.thoughtcrime.securesms.mms.GlideRequests;
 import org.thoughtcrime.securesms.mms.ImageSlide;
 import org.thoughtcrime.securesms.mms.PartAuthority;
@@ -123,6 +122,7 @@ import org.thoughtcrime.securesms.recipients.RecipientId;
 import org.thoughtcrime.securesms.revealable.ViewOnceMessageView;
 import org.thoughtcrime.securesms.util.DateUtils;
 import org.thoughtcrime.securesms.util.InterceptableLongClickCopyLinkSpan;
+import org.thoughtcrime.securesms.util.LinkUtil;
 import org.thoughtcrime.securesms.util.LongClickMovementMethod;
 import org.thoughtcrime.securesms.util.MessageRecordUtil;
 import org.thoughtcrime.securesms.util.PlaceholderURLSpan;
@@ -1362,7 +1362,7 @@ public final class ConversationItem extends RelativeLayout implements BindableCo
 
     if (hasLinks) {
       Stream.of(messageBody.getSpans(0, messageBody.length(), URLSpan.class))
-            .filterNot(url -> LinkPreviewUtil.isLegalUrl(url.getURL()))
+            .filterNot(url -> LinkUtil.isLegalUrl(url.getURL()))
             .forEach(messageBody::removeSpan);
 
       URLSpan[] urlSpans = messageBody.getSpans(0, messageBody.length(), URLSpan.class);

app/src/main/java/org/thoughtcrime/securesms/groups/v2/GroupDescriptionUtil.java

@@ -19,7 +19,7 @@ import com.annimon.stream.Stream;
 
 import org.thoughtcrime.securesms.R;
 import org.thoughtcrime.securesms.components.emoji.EmojiTextView;
-import org.thoughtcrime.securesms.linkpreview.LinkPreviewUtil;
+import org.thoughtcrime.securesms.util.LinkUtil;
 import org.thoughtcrime.securesms.util.LongClickCopySpan;
 
 public final class GroupDescriptionUtil {
@@ -43,7 +43,7 @@ public final class GroupDescriptionUtil {
 
       if (hasLinks) {
         Stream.of(descriptionSpannable.getSpans(0, descriptionSpannable.length(), URLSpan.class))
-              .filterNot(url -> LinkPreviewUtil.isLegalUrl(url.getURL()))
+              .filterNot(url -> LinkUtil.isLegalUrl(url.getURL()))
               .forEach(descriptionSpannable::removeSpan);
 
         URLSpan[] urlSpans = descriptionSpannable.getSpans(0, descriptionSpannable.length(), URLSpan.class);

app/src/main/java/org/thoughtcrime/securesms/linkpreview/LinkPreviewRepository.java

@@ -11,6 +11,7 @@ import androidx.core.util.Consumer;
 
 import com.bumptech.glide.load.engine.DiskCacheStrategy;
 
+import org.signal.core.util.Hex;
 import org.signal.core.util.concurrent.SignalExecutors;
 import org.signal.core.util.logging.Log;
 import org.signal.libsignal.protocol.InvalidMessageException;
@@ -45,8 +46,8 @@ import org.thoughtcrime.securesms.stickers.StickerUrl;
 import org.thoughtcrime.securesms.util.AvatarUtil;
 import org.thoughtcrime.securesms.util.BitmapDecodingException;
 import org.thoughtcrime.securesms.util.ByteUnit;
-import org.signal.core.util.Hex;
 import org.thoughtcrime.securesms.util.ImageCompressionUtil;
+import org.thoughtcrime.securesms.util.LinkUtil;
 import org.thoughtcrime.securesms.util.MediaUtil;
 import org.thoughtcrime.securesms.util.OkHttpUtil;
 import org.whispersystems.signalservice.api.SignalServiceMessageReceiver;
@@ -95,7 +96,7 @@ public class LinkPreviewRepository {
 
     CompositeRequestController compositeController = new CompositeRequestController();
 
-    if (!LinkPreviewUtil.isValidPreviewUrl(url)) {
+    if (!LinkUtil.isValidPreviewUrl(url)) {
       Log.w(TAG, "Tried to get a link preview for a non-whitelisted domain.");
       callback.onError(Error.PREVIEW_NOT_AVAILABLE);
       return compositeController;
@@ -164,7 +165,7 @@ public class LinkPreviewRepository {
         Optional<String> imageUrl    = openGraph.getImageUrl();
         long             date        = openGraph.getDate();
 
-        if (imageUrl.isPresent() && !LinkPreviewUtil.isValidPreviewUrl(imageUrl.get())) {
+        if (imageUrl.isPresent() && !LinkUtil.isValidPreviewUrl(imageUrl.get())) {
           Log.i(TAG, "Image URL was invalid or for a non-whitelisted domain. Skipping.");
           imageUrl = Optional.empty();
         }

app/src/main/java/org/thoughtcrime/securesms/linkpreview/LinkPreviewUtil.java

@@ -2,7 +2,6 @@ package org.thoughtcrime.securesms.linkpreview;
 
 import android.annotation.SuppressLint;
 import android.text.SpannableString;
-import android.text.TextUtils;
 import android.text.style.URLSpan;
 import android.text.util.Linkify;
 
@@ -14,10 +13,8 @@ import androidx.core.text.util.LinkifyCompat;
 import com.annimon.stream.Collectors;
 import com.annimon.stream.Stream;
 
-import org.signal.core.util.logging.Log;
-import org.thoughtcrime.securesms.stickers.StickerUrl;
 import org.thoughtcrime.securesms.util.DateUtils;
-import org.signal.core.util.SetUtil;
+import org.thoughtcrime.securesms.util.LinkUtil;
 import org.thoughtcrime.securesms.util.Util;
 import org.whispersystems.signalservice.api.util.OptionalUtil;
 
@@ -34,12 +31,6 @@ import okhttp3.HttpUrl;
 
 public final class LinkPreviewUtil {
 
-  private static final String TAG = Log.tag(LinkPreviewUtil.class);
-
-  private static final Pattern DOMAIN_PATTERN             = Pattern.compile("^(https?://)?([^/]+).*$");
-  private static final Pattern ALL_ASCII_PATTERN          = Pattern.compile("^[\\x00-\\x7F]*$");
-  private static final Pattern ALL_NON_ASCII_PATTERN      = Pattern.compile("^[^\\x00-\\x7F]*$");
-  private static final Pattern ILLEGAL_CHARACTERS_PATTERN = Pattern.compile("[\u202C\u202D\u202E\u2500-\u25FF]");
   private static final Pattern OPEN_GRAPH_TAG_PATTERN     = Pattern.compile("<\\s*meta[^>]*property\\s*=\\s*\"\\s*og:([^\"]+)\"[^>]*/?\\s*>");
   private static final Pattern ARTICLE_TAG_PATTERN        = Pattern.compile("<\\s*meta[^>]*property\\s*=\\s*\"\\s*article:([^\"]+)\"[^>]*/?\\s*>");
   private static final Pattern OPEN_GRAPH_CONTENT_PATTERN = Pattern.compile("content\\s*=\\s*\"([^\"]*)\"");
@@ -47,8 +38,6 @@ public final class LinkPreviewUtil {
   private static final Pattern FAVICON_PATTERN            = Pattern.compile("<\\s*link[^>]*rel\\s*=\\s*\".*icon.*\"[^>]*>");
   private static final Pattern FAVICON_HREF_PATTERN       = Pattern.compile("href\\s*=\\s*\"([^\"]*)\"");
 
-  private static final Set<String> INVALID_TOP_LEVEL_DOMAINS = SetUtil.newHashSet("onion", "i2p");
-
   public static @Nullable String getTopLevelDomain(@Nullable String urlString) {
     if (!Util.isEmpty(urlString)) {
       HttpUrl url = HttpUrl.parse(urlString);
@@ -61,7 +50,7 @@ public final class LinkPreviewUtil {
   }
 
   /**
-   * @return All whitelisted URLs in the source text.
+   * @return All URLs allowed as previews in the source text.
    */
   public static @NonNull Links findValidPreviewUrls(@NonNull String text) {
     SpannableString spannable = new SpannableString(text);
@@ -73,47 +62,10 @@ public final class LinkPreviewUtil {
 
     return new Links(Stream.of(spannable.getSpans(0, spannable.length(), URLSpan.class))
                            .map(span -> new Link(span.getURL(), spannable.getSpanStart(span)))
-                           .filter(link -> isValidPreviewUrl(link.getUrl()))
+                           .filter(link -> LinkUtil.isValidPreviewUrl(link.getUrl()))
                            .toList());
   }
 
-  /**
-   * @return True if the host is present in the link whitelist.
-   */
-  public static boolean isValidPreviewUrl(@Nullable String linkUrl) {
-    if (linkUrl == null)                      return false;
-    if (StickerUrl.isValidShareLink(linkUrl)) return true;
-
-    HttpUrl url = HttpUrl.parse(linkUrl);
-    return url != null                                   &&
-           !TextUtils.isEmpty(url.scheme())              &&
-           "https".equals(url.scheme())                  &&
-           isLegalUrl(linkUrl);
-  }
-
-  public static boolean isLegalUrl(@NonNull String url) {
-    if (ILLEGAL_CHARACTERS_PATTERN.matcher(url).find()) {
-      return false;
-    }
-
-    Matcher matcher = DOMAIN_PATTERN.matcher(url);
-
-    if (matcher.matches()) {
-      String domain         = matcher.group(2);
-      String cleanedDomain  = domain.replaceAll("\\.", "");
-      String topLevelDomain = parseTopLevelDomain(domain);
-
-      boolean validCharacters = ALL_ASCII_PATTERN.matcher(cleanedDomain).matches() ||
-                                ALL_NON_ASCII_PATTERN.matcher(cleanedDomain).matches();
-
-      boolean validTopLevelDomain = !INVALID_TOP_LEVEL_DOMAINS.contains(topLevelDomain);
-
-      return validCharacters &&  validTopLevelDomain;
-    } else {
-      return false;
-    }
-  }
-
   public static @NonNull OpenGraph parseOpenGraphFields(@Nullable String html) {
     if (html == null) {
       return new OpenGraph(Collections.emptyMap(), null, null);
@@ -169,16 +121,6 @@ public final class LinkPreviewUtil {
     return new OpenGraph(openGraphTags, htmlTitle, faviconUrl);
   }
 
-  private static @Nullable String parseTopLevelDomain(@NonNull String domain) {
-    int periodIndex = domain.lastIndexOf(".");
-
-    if (periodIndex >= 0 && periodIndex < domain.length() - 1) {
-      return domain.substring(periodIndex + 1);
-    } else {
-      return null;
-    }
-  }
-
   private static @NonNull String fromDoubleEncoded(@NonNull String html) {
     return HtmlCompat.fromHtml(HtmlCompat.fromHtml(html, 0).toString(), 0).toString();
   }

app/src/main/java/org/thoughtcrime/securesms/longmessage/LongMessageFragment.java

@@ -26,8 +26,8 @@ import org.thoughtcrime.securesms.components.FullScreenDialogFragment;
 import org.thoughtcrime.securesms.components.emoji.EmojiTextView;
 import org.thoughtcrime.securesms.conversation.colors.ColorizerView;
 import org.thoughtcrime.securesms.keyvalue.SignalStore;
-import org.thoughtcrime.securesms.linkpreview.LinkPreviewUtil;
 import org.thoughtcrime.securesms.recipients.Recipient;
+import org.thoughtcrime.securesms.util.LinkUtil;
 import org.thoughtcrime.securesms.util.Projection;
 import org.thoughtcrime.securesms.util.ThemeUtil;
 import org.thoughtcrime.securesms.util.views.Stub;
@@ -152,7 +152,7 @@ public class LongMessageFragment extends FullScreenDialogFragment {
 
     if (hasLinks) {
       Stream.of(messageBody.getSpans(0, messageBody.length(), URLSpan.class))
-            .filterNot(url -> LinkPreviewUtil.isLegalUrl(url.getURL()))
+            .filterNot(url -> LinkUtil.isLegalUrl(url.getURL()))
             .forEach(messageBody::removeSpan);
     }
     return messageBody;

app/src/main/java/org/thoughtcrime/securesms/messages/MessageContentProcessor.java

@@ -128,6 +128,7 @@ import org.thoughtcrime.securesms.util.Base64;
 import org.thoughtcrime.securesms.util.FeatureFlags;
 import org.thoughtcrime.securesms.util.GroupUtil;
 import org.thoughtcrime.securesms.util.IdentityUtil;
+import org.thoughtcrime.securesms.util.LinkUtil;
 import org.thoughtcrime.securesms.util.MediaUtil;
 import org.thoughtcrime.securesms.util.RemoteDeleteUtil;
 import org.thoughtcrime.securesms.util.TextSecurePreferences;
@@ -2702,7 +2703,7 @@ public final class MessageContentProcessor {
       Optional<String>     description   = Optional.ofNullable(preview.getDescription());
       boolean              hasTitle      = !TextUtils.isEmpty(title.orElse(""));
       boolean              presentInBody = url.isPresent() && urlsInMessage.containsUrl(url.get());
-      boolean              validDomain   = url.isPresent() && LinkPreviewUtil.isValidPreviewUrl(url.get());
+      boolean              validDomain   = url.isPresent() && LinkUtil.isValidPreviewUrl(url.get());
 
       if (hasTitle && (presentInBody || isStoryEmbed) && validDomain) {
         LinkPreview linkPreview = new LinkPreview(url.get(), title.orElse(""), description.orElse(""), preview.getDate(), thumbnail);

app/src/main/java/org/thoughtcrime/securesms/util/LinkUtil.java

@@ -0,0 +1,87 @@
+package org.thoughtcrime.securesms.util;
+
+import android.text.TextUtils;
+
+import androidx.annotation.NonNull;
+import androidx.annotation.Nullable;
+
+import org.signal.core.util.SetUtil;
+import org.thoughtcrime.securesms.stickers.StickerUrl;
+
+import java.util.Objects;
+import java.util.Set;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+import okhttp3.HttpUrl;
+
+public final class LinkUtil {
+
+  private static final Pattern DOMAIN_PATTERN             = Pattern.compile("^(https?://)?([^/]+).*$");
+  private static final Pattern ALL_ASCII_PATTERN          = Pattern.compile("^[\\x00-\\x7F]*$");
+  private static final Pattern ALL_NON_ASCII_PATTERN      = Pattern.compile("^[^\\x00-\\x7F]*$");
+  private static final Pattern ILLEGAL_CHARACTERS_PATTERN = Pattern.compile("[\u202C\u202D\u202E\u2500-\u25FF]");
+
+  private static final Set<String> INVALID_TOP_LEVEL_DOMAINS = SetUtil.newHashSet("onion", "i2p");
+
+  private LinkUtil() {}
+
+  /**
+   * @return True if URL is valid for link previews.
+   */
+  public static boolean isValidPreviewUrl(@Nullable String linkUrl) {
+    if (linkUrl == null) {
+      return false;
+    }
+
+    if (StickerUrl.isValidShareLink(linkUrl)) {
+      return true;
+    }
+
+    HttpUrl url = HttpUrl.parse(linkUrl);
+    return url != null &&
+           !TextUtils.isEmpty(url.scheme()) &&
+           "https".equals(url.scheme()) &&
+           isLegalUrl(linkUrl, false);
+  }
+
+  /**
+   * @return True if URL is valid, mostly useful for linkifying.
+   */
+  public static boolean isLegalUrl(@NonNull String url) {
+    return isLegalUrl(url, true);
+  }
+
+  private static boolean isLegalUrl(@NonNull String url, boolean skipTopLevelDomainValidation) {
+    if (ILLEGAL_CHARACTERS_PATTERN.matcher(url).find()) {
+      return false;
+    }
+
+    Matcher matcher = DOMAIN_PATTERN.matcher(url);
+
+    if (matcher.matches()) {
+      String domain         = Objects.requireNonNull(matcher.group(2));
+      String cleanedDomain  = domain.replaceAll("\\.", "");
+      String topLevelDomain = parseTopLevelDomain(domain);
+
+      boolean validCharacters = ALL_ASCII_PATTERN.matcher(cleanedDomain).matches() ||
+                                ALL_NON_ASCII_PATTERN.matcher(cleanedDomain).matches();
+
+      boolean validTopLevelDomain = skipTopLevelDomainValidation || !INVALID_TOP_LEVEL_DOMAINS.contains(topLevelDomain);
+
+      return validCharacters && validTopLevelDomain;
+    } else {
+      return false;
+    }
+  }
+
+  private static @Nullable String parseTopLevelDomain(@NonNull String domain) {
+    int periodIndex = domain.lastIndexOf(".");
+
+    if (periodIndex >= 0 && periodIndex < domain.length() - 1) {
+      return domain.substring(periodIndex + 1);
+    } else {
+      return null;
+    }
+  }
+}

app/src/test/java/org/thoughtcrime/securesms/util/LinkUtilTest_isLegal.java

@@ -1,4 +1,4 @@
-package org.thoughtcrime.securesms.linkpreview;
+package org.thoughtcrime.securesms.util;
 
 import org.junit.Test;
 import org.junit.runner.RunWith;
@@ -10,7 +10,7 @@ import java.util.Collection;
 import static junit.framework.TestCase.assertEquals;
 
 @RunWith(Parameterized.class)
-public class LinkPreviewUtilTest_isLegal {
+public class LinkUtilTest_isLegal {
 
   private final String  input;
   private final boolean output;
@@ -24,12 +24,12 @@ public class LinkPreviewUtilTest_isLegal {
         { "https://foo.google.com/some/path.html", true  },
         { "кц.рф",                                 true  },
         { "https://кц.рф/some/path",               true  },
+        { "https://abcdefg.onion",                 true  },
+        { "https://abcdefg.i2p",                   true  },
         { "http://кц.com",                         false },
         { "кц.com",                                false },
         { "http://asĸ.com",                        false },
         { "http://foo.кц.рф",                      false },
-        { "https://abcdefg.onion",                 false },
-        { "https://abcdefg.i2p",                   false },
         { "кц.рф\u202C",                           false },
         { "кц.рф\u202D",                           false },
         { "кц.рф\u202E",                           false },
@@ -40,13 +40,13 @@ public class LinkPreviewUtilTest_isLegal {
     });
   }
 
-  public LinkPreviewUtilTest_isLegal(String input, boolean output) {
+  public LinkUtilTest_isLegal(String input, boolean output) {
     this.input  = input;
     this.output = output;
   }
 
   @Test
   public void isLegal() {
-    assertEquals(output, LinkPreviewUtil.isLegalUrl(input));
+    assertEquals(output, LinkUtil.isLegalUrl(input));
   }
 }