From 8ba6fb3c145dc8323280343efc83ae873c8f99ad Mon Sep 17 00:00:00 2001
From: Alberto Fernandez <infjaf@gmail.com>
Date: Fri, 27 Mar 2020 18:57:46 +0100
Subject: [PATCH] Only detet US-ASCII if there are printable characters, new
 lines and tabs #33

---
 .../universalchardet/UniversalDetector.java   | 10 +++++-
 .../Bug33USASCIIToGenerous.java               | 34 +++++++++++++++++++
 2 files changed, 43 insertions(+), 1 deletion(-)
 create mode 100644 src/test/java/org/mozilla/universalchardet/Bug33USASCIIToGenerous.java

diff --git a/src/main/java/org/mozilla/universalchardet/UniversalDetector.java b/src/main/java/org/mozilla/universalchardet/UniversalDetector.java
index 3344577..3cce54c 100644
--- a/src/main/java/org/mozilla/universalchardet/UniversalDetector.java
+++ b/src/main/java/org/mozilla/universalchardet/UniversalDetector.java
@@ -83,6 +83,7 @@ public enum InputState
     private boolean     done;
     private boolean     start;
     private boolean     gotData;
+    private boolean     onlyPrintableASCII = true;
     private byte        lastChar;
     private String      detectedCharset;
 
@@ -192,6 +193,13 @@ public void handleData(final byte[] buf, int offset, int length)
                     (c == 0x1B || (c == 0x7B && this.lastChar == 0x7E))) {
                     this.inputState = InputState.ESC_ASCII;
                 }
+                if (this.inputState == InputState.PURE_ASCII && onlyPrintableASCII) {
+                	onlyPrintableASCII =
+                			(c >= 0x20 && c <= 0x7e) // Printable characters 
+                			|| c == 0x0A  // New Line
+                			|| c == 0x0D  // Carriage return 
+                			|| c== 0x09;  // TAB
+                }
                 this.lastChar = buf[i];
             }
         } // for end
@@ -302,7 +310,7 @@ public void dataEnd()
             }
         } else if (this.inputState == InputState.ESC_ASCII) {
             // do nothing
-        } else if (this.inputState == InputState.PURE_ASCII && this.gotData) {
+        } else if (this.inputState == InputState.PURE_ASCII && this.onlyPrintableASCII) {
         	this.detectedCharset = Constants.CHARSET_US_ASCCI;
         }
         else {
diff --git a/src/test/java/org/mozilla/universalchardet/Bug33USASCIIToGenerous.java b/src/test/java/org/mozilla/universalchardet/Bug33USASCIIToGenerous.java
new file mode 100644
index 0000000..c8db5ee
--- /dev/null
+++ b/src/test/java/org/mozilla/universalchardet/Bug33USASCIIToGenerous.java
@@ -0,0 +1,34 @@
+package org.mozilla.universalchardet;
+
+import java.io.ByteArrayInputStream;
+import java.io.IOException;
+
+import org.junit.Assert;
+import org.junit.Ignore;
+import org.junit.Test;
+
+public class Bug33USASCIIToGenerous {
+	
+	public Bug33USASCIIToGenerous() {
+		super();
+	}
+	
+
+	
+	@Test
+	@Ignore("Not sure")
+	public void testUTF16 () throws IOException {
+		Assert.assertEquals("UTF-16BE", detect("ab".getBytes("UTF-16BE")));
+		Assert.assertEquals("UTF-16LE", detect("ab".getBytes("UTF-16LE")));
+	}
+	@Test
+	public void testZipHeader() throws IOException {
+		byte[] zipHeader = new byte[]{0x50, 0x4b, 0x03, 0x04, 0x14, 0x00, 0x02, 0x00}; 
+		Assert.assertNull(detect(zipHeader));
+	}
+	
+	private String detect(byte[] data) throws IOException {
+		return UniversalDetector.detectCharset(new ByteArrayInputStream(data));
+	}
+
+}