Better handling of Java escaped Strings, remove commons-text dependency

Convex-Dev · Oct 5, 2024 · 3c8477b · 3c8477b
1 parent 1231dba
commit 3c8477b
Show file tree

Hide file tree

Showing 7 changed files with 134 additions and 17 deletions.
diff --git a/convex-core/src/main/java/convex/core/lang/reader/AntlrReader.java b/convex-core/src/main/java/convex/core/lang/reader/AntlrReader.java
@@ -72,6 +72,7 @@
 import convex.core.lang.reader.antlr.ConvexParser.TagContext;
 import convex.core.lang.reader.antlr.ConvexParser.TaggedFormContext;
 import convex.core.lang.reader.antlr.ConvexParser.VectorContext;
+import convex.core.text.Text;
 import convex.core.util.Utils;
 
 public class AntlrReader {
@@ -446,7 +447,7 @@ public void exitString(StringContext ctx) {
 			String s=ctx.getStop().getText();
 			int n=s.length();
 			s=s.substring(1, n-1); // skip surrounding double quotes
-			s=ReaderUtils.unescapeString(s);
+			s=Text.unescapeJava(s);
 			push(Strings.create(s));
 		}
 

diff --git a/convex-core/src/main/java/convex/core/lang/reader/ReaderUtils.java b/convex-core/src/main/java/convex/core/lang/reader/ReaderUtils.java
@@ -2,8 +2,6 @@
 
 import java.util.HashMap;
 
-import org.apache.commons.text.StringEscapeUtils; // TODO: Kill dependency?
-
 import convex.core.data.ACell;
 import convex.core.data.AHashMap;
 import convex.core.data.AMap;
@@ -58,18 +56,7 @@ public static Symbol getQuotingSymbol(String s) {
 		return quotingSymbols.get(s);
 	}
 
-	/**
-	 * Unescapes a string according to Java rules
-	 * @param s String to unescape
-	 * @return Unescaped string
-	 */
-	public static String unescapeString(String s) {
-		return StringEscapeUtils.unescapeJava(s);
-	}
 
-	public static String escapeString(String s) {
-		return StringEscapeUtils.escapeJava(s);
-	}
 
 	private static final HashMap<String,ACell> specialLiterals=Maps.hashMapOf(
 			"##NaN",CVMDouble.NaN,

diff --git a/convex-core/src/main/java/convex/core/text/Text.java b/convex-core/src/main/java/convex/core/text/Text.java
@@ -10,6 +10,8 @@
 import convex.core.Coin;
 import convex.core.data.prim.CVMDouble;
 import convex.core.data.util.BlobBuilder;
+import convex.core.exceptions.TODOException;
+import convex.core.util.Utils;
 
 public class Text {
 	private static final int WHITESPACE_LENGTH = 32;
@@ -191,7 +193,90 @@ public static String repeat(char c, int count) {
 		Arrays.fill(cs,c);
 		return new String(cs);
 	}
+
+	public static String escapeJava(String s) {
+		throw new TODOException();
+	}
 
+	/**
+	 * Unescapes a Java string
+	 * @param st String to unescape
+	 * @return Unescaped string, or null if not a valid Java String
+	 */
+	public static String unescapeJava(String st) {
+	    StringBuilder sb = null;
+	    int n=st.length();
 
+	    for (int i = 0; i < n; i++) {
+	    	int startPos=i;
+	        char ch = st.charAt(i);
+	        if ((ch == '\\')&&(i+1<n)) {
+	            char nextChar = st.charAt(i + 1);
+
+	            // Check for octal escape, consumes 1-3 octal chars greedily
+	            if (nextChar >= '0' && nextChar <= '7') {
+	                int code=Utils.octalVal(nextChar);
+	                for (int j=i+2; j<i+4; j++) {
+	                	if (j>=n) break;
+	                	int v=Utils.octalVal(st.charAt(j));
+	                	if (v<0) break; // no more octal
+	                	if (code>=32) break; // wouldn't be valid, JLS maximum octal value of 377
+	                	code=code*8+v;
+	                	i++;
+	                }
+	                ch=(char) code;
+	            } else {
+		            switch (nextChar) {
+		            case '\\':
+		                ch = '\\';
+		                break;
+		            case 'b':
+		                ch = '\b';
+		                break;
+		            case 'f':
+		                ch = '\f';
+		                break;
+		            case 'n':
+		                ch = '\n';
+		                break;
+		            case 'r':
+		                ch = '\r';
+		                break;
+		            case 't':
+		                ch = '\t';
+		                break;
+		            case '\"':
+		                ch = '\"';
+		                break;
+		            case '\'':
+		                ch = '\'';
+		                break;
+		            // Hex Unicode: u????
+		            case 'u':
+		                if (i+6 > n) {
+		                	return null; // insufficient chars for unicode
+		                }
+		                int cp=0;
+		                for (int j=0; j<4; j++) {
+		                	int v=Utils.hexVal(st.charAt(i+j+2));
+		                	if (v<0) { // not a hex value
+		                		return null;
+		                	}
+		                	cp=cp*16+Utils.hexVal(st.charAt(i+j+2));
+		                }
+		    	        if (sb==null) sb=new StringBuilder(st.substring(0, startPos));
+		                sb.append(Character.toChars(cp));
+		                i += 5; // skip extra 5 chars on top of loop increment
+		                continue;
+		            }
+	            }
+	            i++; // skip a char, since we consumed nextChar
+	        }
+	        // We are appending a single char
+	        if (sb==null) sb=new StringBuilder(st.substring(0, startPos));
+	        sb.append(ch);
+ 	    }
+	    return (sb==null)?st:sb.toString();
+	}
 
 }
diff --git a/convex-core/src/main/java/convex/core/util/Utils.java b/convex-core/src/main/java/convex/core/util/Utils.java
@@ -286,7 +286,7 @@ public static BigInteger hexToBigInt(String hex) {
 	/**
 	 * Gets the value of a single hex character e.g. hexVal('c') => 12
 	 *
-	 * @param c Character representing a hex digit
+	 * @param c Character possibly representing a hex digit
 	 * @return int in the range 0..15 inclusive, or -1 if not a hex char
 	 */
 	public static int hexVal(char c) {
@@ -303,6 +303,20 @@ public static int hexVal(char c) {
 
 		return -1;
 	}
+
+	/**
+	 * Gets the value of a single octal character e.g. octalVal('6') => 6
+	 *
+	 * @param c Character possibly representing an octal digit
+	 * @return int in the range 0..7 inclusive, or -1 if not an octal char
+	 */
+	public static int octalVal(char c) {
+		int v = (int) c;
+
+		if ((v<48)||(v>55)) return -1; // out of possible range
+
+		return v-48;
+	}
 
 	/**
 	 * Converts a byte array of length N to a hex string of length 2N

diff --git a/convex-core/src/main/java/module-info.java b/convex-core/src/main/java/module-info.java
@@ -22,7 +22,6 @@
 	exports convex.dlfs;
 
 	requires transitive org.antlr.antlr4.runtime;
-	requires org.apache.commons.text;
 	requires org.bouncycastle.pkix;
 	requires transitive org.bouncycastle.provider;
 	requires org.bouncycastle.util;

diff --git a/convex-core/src/test/java/convex/core/lang/ReaderTest.java b/convex-core/src/test/java/convex/core/lang/ReaderTest.java
@@ -32,6 +32,7 @@
 import convex.core.data.prim.CVMDouble;
 import convex.core.data.prim.CVMLong;
 import convex.core.exceptions.ParseException;
+import convex.core.text.Text;
 import convex.test.Samples;
 
 /**
@@ -243,7 +244,6 @@ public void testStrings() {
 
 		// Multi-line String
 		assertEquals(Strings.create("\n"), Reader.read("\"\n\""));
-
 	}
 
 	@Test
@@ -412,6 +412,17 @@ public void doIdempotencyTest(ACell cell) {
 		doReadPrintTest("^{} 0xa89e59cc8ab9fc6a13785a37938c85b306b24663415effc01063a6e25ef52ebcd3647d3a77e0a33908a372146fdccab6");
 	}
 
+	/**
+	 * Test cases for strings with Java escapes
+	 */
+	@Test public void testJavaEscapes() {
+		doEscapeTest("!0\\","\\410\\");
+	}
+
+	private void doEscapeTest(String raw, String escaped) {
+		assertEquals(raw,Text.unescapeJava(escaped));
+	}
+
 	/**
 	 * Test cases that should read and print identically
 	 */

diff --git a/convex-core/src/test/java/convex/core/text/TextTest.java b/convex-core/src/test/java/convex/core/text/TextTest.java
@@ -1,6 +1,7 @@
 package convex.core.text;
 
 import static org.junit.Assert.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertNull;
 
 import java.text.ParseException;
 
@@ -21,6 +22,25 @@ public void testWhiteSpace() {
 		checkWhiteSpace(97);
 		checkWhiteSpace(100);
 	}
+
+	@Test
+	public void testUnescapeJava() {
+		assertEquals("foo",Text.unescapeJava("foo"));
+
+		assertEquals("\\",Text.unescapeJava("\\"));
+
+		assertEquals("zzAzzB",Text.unescapeJava("zz\\u0041zz\\u0042"));
+		assertNull(Text.unescapeJava("\\u"));
+		assertNull(Text.unescapeJava("\\u0x0x")); // not valid unicode
+		assertNull(Text.unescapeJava("\\u012")); // not valid unicode (only 3 chars)
+
+		// octal escapes
+		assertEquals("a%b",Text.unescapeJava("a\\45b")); 
+		assertEquals("\19",Text.unescapeJava("\\19")); 
+		assertEquals("\0\0",Text.unescapeJava("\\0\\0")); 
+		assertEquals("\1\11\111",Text.unescapeJava("\\1\\11\\111")); 
+		assertEquals("!0",Text.unescapeJava("\\410")); // Yeah, JLS has max octal value of 377. Don't ask why...
+	}
 
 	private void checkWhiteSpace(int len) {
 		String s = Text.whiteSpace(len);