Improve scrubbing of sensitive identifiers

Fixes #1189

man/signal-cli.1.adoc

@@ -44,6 +44,7 @@ If `--verbose` is also given, the detailed logs will only be written to the log
 
 *--scrub-log*::
 Scrub possibly sensitive information from the log, like phone numbers and UUIDs.
+Doesn't work reliably on dbus logs with very verbose logging (`-vvv`)
 
 *--config* CONFIG::
 Set the path, where to store the config.

src/main/java/org/asamk/signal/logging/LogConfigurator.java

@@ -52,6 +52,8 @@ public class LogConfigurator extends ContextAwareBase implements Configurator {
         rootLogger.addAppender(consoleAppender);
 
         lc.getLogger("org.asamk").setLevel(verboseLevel > 1 ? Level.ALL : verboseLevel > 0 ? Level.DEBUG : Level.INFO);
+        lc.getLogger("org.asamk.Signal")
+                .setLevel(verboseLevel > 2 ? Level.ALL : verboseLevel > 1 ? Level.INFO : Level.WARN);
         lc.getLogger("com.zaxxer.hikari.pool.PoolBase")
                 .setLevel(verboseLevel > 2 ? Level.ALL : verboseLevel > 1 ? Level.INFO : Level.WARN);
         lc.getLogger("org.sqlite.core.NativeDB")

src/main/java/org/asamk/signal/logging/Scrubber.java

@@ -37,9 +37,17 @@ public final class Scrubber {
      * Supposedly, the shortest international phone numbers in use contain seven digits.
      * Handles URL encoded +, %2B
      */
-    private static final Pattern E164_PATTERN = Pattern.compile("(\\+|%2B|/org/asamk/Signal/_)(\\d{5,13})(\\d{2})");
+    private static final Pattern E164_PATTERN = Pattern.compile("(\\+|%2B|_)(\\d{5,13})(\\d{2})");
     private static final String E164_CENSOR = "*************";
 
+    private static final Pattern GROUP_V1_ID_PATTERN = Pattern.compile(
+            "(/org/asamk/Signal/.*Groups/[a-zA-Z0-9/_+-]{2}|[a-zA-Z0-9/_+-]{2})([a-zA-Z0-9/_+-]{18})([a-zA-Z0-9/_+-]{2})(==|__)");
+    private static final String GROUP_V1_ID_CENSOR = "*".repeat(18);
+
+    private static final Pattern GROUP_V2_ID_PATTERN = Pattern.compile(
+            "(/org/asamk/Signal/.*Groups/[a-zA-Z0-9/_+-]{2}|[a-zA-Z0-9/_+-]{2})([a-zA-Z0-9/_+-]{39})([a-zA-Z0-9/_+-]{2})([=_])");
+    private static final String GROUP_V2_ID_CENSOR = "*".repeat(39);
+
     /**
      * The second group will be censored.
      */
@@ -172,9 +180,11 @@ public final class Scrubber {
 
     public static CharSequence scrub(CharSequence in) {
 
+        in = scrubUuids(in);
         in = scrubE164(in);
         in = scrubEmail(in);
-        in = scrubUuids(in);
+        in = scrubGroupV2Ids(in);
+        in = scrubGroupV1Ids(in);
         in = scrubDomains(in);
         in = scrubIpv4(in);
 
@@ -189,6 +199,22 @@ public final class Scrubber {
                         .append(matcher.group(3)));
     }
 
+    private static CharSequence scrubGroupV1Ids(CharSequence in) {
+        return scrub(in,
+                GROUP_V1_ID_PATTERN,
+                (matcher, output) -> output.append(matcher.group(1))
+                        .append(GROUP_V1_ID_CENSOR, 0, matcher.group(2).length())
+                        .append(matcher.group(3)));
+    }
+
+    private static CharSequence scrubGroupV2Ids(CharSequence in) {
+        return scrub(in,
+                GROUP_V2_ID_PATTERN,
+                (matcher, output) -> output.append(matcher.group(1))
+                        .append(GROUP_V2_ID_CENSOR, 0, matcher.group(2).length())
+                        .append(matcher.group(3)));
+    }
+
     private static CharSequence scrubEmail(CharSequence in) {
         return scrub(in,
                 CRUDE_EMAIL_PATTERN,