fix requireValidLength rejecting line-wrapped MIME Base64

Author: nklapste

src/main/java/jadx/plugins/stringdecoder/B64Detector.java

@@ -22,6 +22,8 @@ public final class B64Detector {
 	// Allow embedded \n/\r (PEM/MIME line-wrapped Base64); = padding and trailing newline are optional.
 	private static final Pattern BASE64_STANDARD = Pattern.compile("^[A-Za-z0-9+/\\n\\r]*=*[\\n\\r]*$");
 	private static final Pattern BASE64_URL_SAFE = Pattern.compile("^[A-Za-z0-9_\\-\\n\\r]*=*[\\n\\r]*$");
+	// Line-break chars permitted as PEM/MIME wrapping above; stripped before the length-quantum check.
+	private static final Pattern LINE_BREAKS = Pattern.compile("[\\n\\r]");
 	// Identifier shapes — only checked below IDENTIFIER_FILTER_MAX_LEN.
 	private static final Pattern CAMEL_CASE = Pattern.compile("^[a-z]+([A-Z][a-z]+)+$");
 	private static final Pattern PASCAL_CASE = Pattern.compile("^[A-Z][a-z]+([A-Z][a-z0-9]+)+$");
@@ -57,7 +59,7 @@ private interface InputFilter {
 	 */
 	private static final InputFilter[] INPUT_FILTERS = {
 			(s, o) -> B64FalsePositives.contains(s),
-			(s, o) -> o.isRequireValidLength() && s.length() % 4 != 0,
+			(s, o) -> o.isRequireValidLength() && significantLength(s) % 4 != 0,
 			(s, o) -> isLikelyIdentifier(s, o),
 			(s, o) -> !hasValidBase64Charset(s)
 	};
@@ -116,6 +118,18 @@ private static boolean hasValidBase64Charset(String str) {
 		return BASE64_STANDARD.matcher(str).matches() || BASE64_URL_SAFE.matcher(str).matches();
 	}
 
+	/**
+	 * Length excluding {@code \n}/{@code \r} — line wraps don't count toward the 4-char Base64
+	 * quantum, so {@code requireValidLength} must ignore them or it rejects line-wrapped strings.
+	 *
+	 * <p>TODO: full MIME also skips spaces/tabs; supporting it means stripping all {@code \s} here
+	 * and relaxing {@link #hasValidBase64Charset}. Only worth it with a solid false-positive
+	 * defense, else prose like {@code "Java Code"} decodes.
+	 */
+	private static int significantLength(String str) {
+		return LINE_BREAKS.matcher(str).replaceAll("").length();
+	}
+
 	private static Variant[] variantsFor(String str) {
 		boolean hasLineBreaks = str.indexOf('\n') >= 0 || str.indexOf('\r') >= 0;
 		return hasLineBreaks ? Variant.values() : new Variant[] { Variant.STANDARD, Variant.URL };

src/test/java/jadx/plugins/stringdecoder/B64DetectorTest.java

@@ -5,6 +5,7 @@
 import java.nio.charset.CharacterCodingException;
 import java.nio.charset.CodingErrorAction;
 import java.nio.charset.StandardCharsets;
+import java.util.Base64;
 import java.util.Map;
 
 import static org.assertj.core.api.Assertions.assertThat;
@@ -229,6 +230,28 @@ public void detectRequireValidLengthFilterTest() {
 		assertThat(r.getDecoded()).isEqualTo("hello");
 	}
 
+	@Test
+	public void detectLineWrappedMimeB64NotDivisibleByFourTest() {
+		// One \n makes the raw length 21 (21 % 4 == 1), which the old length filter rejected outright.
+		// significantLength() ignores the line break, so the 20 real Base64 chars (20 % 4 == 0) now
+		// pass the default requireValidLength=true filter and decode via the MIME variant.
+		B64Result r = B64Detector.detect("SGVsbG8s\nIFdvcmxkIQ==", defaultOptions());
+		assertThat(r).isNotNull();
+		assertThat(r.commentText()).isEqualTo("b64(mime): Hello, World!");
+	}
+
+	@Test
+	public void detectSpaceDelimitedMimeB64CurrentlyFailsTest() throws CharacterCodingException {
+		// A space-delimited blob is genuinely valid MIME Base64 — the MIME decoder skips the space:
+		String spaceDelimited = "SGVsbG8s IFdvcmxkIQ==";
+		byte[] bytes = Base64.getMimeDecoder().decode(spaceDelimited);
+		assertThat(B64Detector.decodeUtf8(bytes, CodingErrorAction.REPORT)).isEqualTo("Hello, World!");
+		// ...but B64Detector rejects it: the charset patterns permit only \n/\r as whitespace, not spaces.
+		// Known limitation; flip this to a successful-decode assertion if full-MIME whitespace tolerance
+		// is ever added (see significantLength()).
+		assertThat(B64Detector.detect(spaceDelimited, defaultOptions())).isNull();
+	}
+
 	@Test
 	public void detectMinDecodedLengthFilterTest() {
 		// "aGVs" decodes to "hel" (3 chars) — below default minDecodedLength=4