Attempt to use ByteList 2.0 with proper CharSequence logic.

The changes here move away from using CharSequence methods from ByteList when the intent is to traverse bytes rather than decoded characters. Places where a ByteList is not typed as such in code (type is CharSequence or Object) or where it is concatenated into a String using `+` were not detected by my search and may still be incorrect.
jruby · Apr 4, 2018 · 3a08ac3 · 3a08ac3
1 parent e75b913
commit 3a08ac3
Showing 48 changed files with 241 additions and 248 deletions.
diff --git a/core/pom.rb b/core/pom.rb
@@ -52,7 +52,7 @@
   jar 'com.github.jnr:jffi:${jffi.version}:native'
 
   jar 'org.jruby.joni:joni:2.1.15'
-  jar 'org.jruby.extras:bytelist:1.0.15'
+  jar 'org.jruby.extras:bytelist:2.0.0-SNAPSHOT'
   jar 'org.jruby.jcodings:jcodings:1.0.27'
   jar 'org.jruby:dirgra:0.3'
 

diff --git a/core/pom.xml b/core/pom.xml
@@ -177,7 +177,7 @@ DO NOT MODIFIY - GENERATED CODE
     <dependency>
       <groupId>org.jruby.extras</groupId>
       <artifactId>bytelist</artifactId>
-      <version>1.0.15</version>
+      <version>2.0.0-SNAPSHOT</version>
     </dependency>
     <dependency>
       <groupId>org.jruby.jcodings</groupId>

diff --git a/core/src/main/java/org/jruby/RubyArgsFile.java b/core/src/main/java/org/jruby/RubyArgsFile.java
@@ -199,7 +199,7 @@ public boolean next_argv(ThreadContext context) {
 
         private static boolean filenameEqlDash(final RubyString filename) {
             final ByteList filenameBytes = filename.getByteList();
-            return filenameBytes.length() == 1 && filenameBytes.get(0) == '-';
+            return filenameBytes.byteLength() == 1 && filenameBytes.get(0) == '-';
         }
 
         public static ArgsFileData getArgsFileData(Ruby runtime) {
@@ -805,8 +805,8 @@ public static IRubyObject read(ThreadContext context, IRubyObject recv, IRubyObj
                     continue;
                 }
             } else if(args.length >= 1) {
-                if (((RubyString)str).getByteList().length() < len) {
-                    len -= ((RubyString)str).getByteList().length();
+                if (((RubyString)str).getByteList().byteLength() < len) {
+                    len -= ((RubyString)str).getByteList().byteLength();
                     args[0] = runtime.newFixnum(len);
                     continue;
                 }

diff --git a/core/src/main/java/org/jruby/RubyArray.java b/core/src/main/java/org/jruby/RubyArray.java
@@ -1981,7 +1981,7 @@ public IRubyObject join19(final ThreadContext context, IRubyObject sep) {
                 return joinAny(context, sepString, i, joinStrings(sepString, i, result));
             }
 
-            len += ((RubyString) tmp).getByteList().length();
+            len += ((RubyString) tmp).getByteList().byteLength();
         }
 
         return joinStrings(sepString, realLength, (RubyString) RubyString.newStringLight(runtime, len).infectBy(this));

diff --git a/core/src/main/java/org/jruby/RubyComplex.java b/core/src/main/java/org/jruby/RubyComplex.java
@@ -1121,7 +1121,7 @@ private static RubyNumeric convertString(ThreadContext context, final IRubyObjec
     // MRI: string_to_c_strict
     private static RubyNumeric str_to_c_strict(ThreadContext context, RubyString str) {
         IRubyObject[] ary = str_to_c_internal(context, str);
-        if (ary[0] == context.nil || ary[1].convertToString().getByteList().length() > 0) {
+        if (ary[0] == context.nil || ary[1].convertToString().getByteList().byteLength() > 0) {
             throw context.runtime.newArgumentError("invalid value for convert(): " + str.inspect(context.runtime));
         }
         return (RubyNumeric) ary[0]; // (RubyComplex)

diff --git a/core/src/main/java/org/jruby/RubyIO.java b/core/src/main/java/org/jruby/RubyIO.java
@@ -2625,7 +2625,7 @@ private static void putsSingle(ThreadContext context, Ruby runtime, IRubyObject
             write(context, maybeIO, line);
         }
 
-        if (line.length() == 0 || !line.endsWith(separator.getByteList())) {
+        if (line.byteLength() == 0 || !line.endsWith(separator.getByteList())) {
             write(context, maybeIO, separator);
         }
     }

diff --git a/core/src/main/java/org/jruby/RubyKernel.java b/core/src/main/java/org/jruby/RubyKernel.java
@@ -1406,9 +1406,9 @@ public static IRubyObject test(ThreadContext context, IRubyObject recv, IRubyObj
         if (args[0] instanceof RubyFixnum) {
             cmd = (int)((RubyFixnum) args[0]).getLongValue();
         } else if (args[0] instanceof RubyString &&
-                ((RubyString) args[0]).getByteList().length() > 0) {
+                ((RubyString) args[0]).getByteList().byteLength() > 0) {
             // MRI behavior: use first byte of string value if len > 0
-            cmd = ((RubyString) args[0]).getByteList().charAt(0);
+            cmd = ((RubyString) args[0]).getByteList().get(0);
         } else {
             cmd = (int) args[0].convertToInteger().getLongValue();
         }

diff --git a/core/src/main/java/org/jruby/RubyMarshal.java b/core/src/main/java/org/jruby/RubyMarshal.java
@@ -127,7 +127,7 @@ public static IRubyObject load(ThreadContext context, IRubyObject recv, IRubyObj
             if (str != context.nil) {
                 tainted = in.isTaint();
                 ByteList bytes = ((RubyString) str).getByteList();
-                rawInput = new ByteArrayInputStream(bytes.getUnsafeBytes(), bytes.begin(), bytes.length());
+                rawInput = new ByteArrayInputStream(bytes.getUnsafeBytes(), bytes.begin(), bytes.byteLength());
             } else if (sites(context).respond_to_getc.respondsTo(context, in, in) &&
                         sites(context).respond_to_read.respondsTo(context, in, in)) {
                 tainted = true;

diff --git a/core/src/main/java/org/jruby/RubyMatchData.java b/core/src/main/java/org/jruby/RubyMatchData.java
@@ -732,7 +732,7 @@ public IRubyObject post_match(ThreadContext context) {
         check();
         if (begin == -1) return context.nil;
 
-        final int strLen = str.getByteList().length();
+        final int strLen = str.getByteList().byteLength();
         return makeShared(context.runtime, str, end, strLen - end).infectBy(this);
     }
 

diff --git a/core/src/main/java/org/jruby/RubyRational.java b/core/src/main/java/org/jruby/RubyRational.java
@@ -1378,7 +1378,7 @@ static IRubyObject[] str_to_r_internal(final ThreadContext context, final RubySt
 
             if (si != nil) {
                 ByteList siBytes = si.convertToString().getByteList();
-                if (siBytes.length() > 0 && siBytes.get(0) == '-') v = f_negate(context, v); 
+                if (siBytes.byteLength() > 0 && siBytes.get(0) == '-') v = f_negate(context, v);
             }
 
             if (exp != nil) {
@@ -1395,7 +1395,7 @@ static IRubyObject[] str_to_r_internal(final ThreadContext context, final RubySt
 
     private static RubyNumeric str_to_r_strict(ThreadContext context, RubyString str) {
         IRubyObject[] ary = str_to_r_internal(context, str);
-        if (ary[0] == context.nil || ary[1].convertToString().getByteList().length() > 0) {
+        if (ary[0] == context.nil || ary[1].convertToString().getByteList().byteLength() > 0) {
             throw context.runtime.newArgumentError("invalid value for convert(): " + str.inspect(context.runtime));
         }
         return (RubyNumeric) ary[0]; // (RubyRational)

diff --git a/core/src/main/java/org/jruby/RubyString.java b/core/src/main/java/org/jruby/RubyString.java
@@ -1005,7 +1005,7 @@ public final void resize(int length) {
         if (value.getRealSize() > length) {
             modify();
             value.setRealSize(length);
-        } else if (value.length() < length) {
+        } else if (value.byteLength() < length) {
             modify();
             value.length(length);
         }
@@ -1053,7 +1053,7 @@ public static String bytesToString(byte[] bytes, int beg, int len) {
     }
 
     public static String byteListToString(ByteList bytes) {
-        return bytesToString(bytes.getUnsafeBytes(), bytes.begin(), bytes.length());
+        return bytesToString(bytes.getUnsafeBytes(), bytes.begin(), bytes.byteLength());
     }
 
     public static String bytesToString(byte[] bytes) {
@@ -2498,7 +2498,7 @@ public RubyBoolean empty_p(ThreadContext context) {
     }
 
     public boolean isEmpty() {
-        return value.length() == 0;
+        return value.byteLength() == 0;
     }
 
     /** rb_str_append
@@ -2669,7 +2669,7 @@ public RubyString crypt(ThreadContext context, IRubyObject other) {
         otherStr.modify();
         otherStr.associateEncoding(ascii8bit);
         ByteList otherBL = otherStr.getByteList();
-        if (otherBL.length() < 2) {
+        if (otherBL.byteLength() < 2) {
             throw context.runtime.newArgumentError("salt too short (need >=2 bytes)");
         }
 
@@ -3192,7 +3192,7 @@ public final IRubyObject substr(int beg, int len) {
 
     /* rb_str_substr */
     public final IRubyObject substr(Ruby runtime, int beg, int len) {
-        int length = value.length();
+        int length = value.byteLength();
         if (len < 0 || beg > length) return runtime.getNil();
 
         if (beg < 0) {
@@ -3206,7 +3206,7 @@ public final IRubyObject substr(Ruby runtime, int beg, int len) {
 
     /* str_byte_substr */
     private IRubyObject byteSubstr(Ruby runtime, int beg, int len) {
-        int length = value.length();
+        int length = value.byteLength();
 
         if (len < 0 || beg > length) return runtime.getNil();
 
@@ -3228,7 +3228,7 @@ private IRubyObject byteARef(Ruby runtime, IRubyObject idx) {
         final int index;
 
         if (idx instanceof RubyRange){
-            int[] begLen = ((RubyRange) idx).begLenInt(getByteList().length(), 0);
+            int[] begLen = ((RubyRange) idx).begLenInt(getByteList().byteLength(), 0);
             return begLen == null ? runtime.getNil() : byteSubstr(runtime, begLen[0], begLen[1]);
         } else if (idx instanceof RubyFixnum) {
             index = RubyNumeric.fix2int((RubyFixnum)idx);
@@ -3238,15 +3238,15 @@ private IRubyObject byteARef(Ruby runtime, IRubyObject idx) {
             if (RubyRange.isRangeLike(context, idx, sites.respond_to_begin, sites.respond_to_end)) {
                 RubyRange range = RubyRange.rangeFromRangeLike(context, idx, sites.begin, sites.end, sites.exclude_end);
 
-                int[] begLen = range.begLenInt(getByteList().length(), 0);
+                int[] begLen = range.begLenInt(getByteList().byteLength(), 0);
                 return begLen == null ? runtime.getNil() : byteSubstr(runtime, begLen[0], begLen[1]);
             } else {
                 index = RubyNumeric.num2int(idx);
             }
         }
 
         IRubyObject obj = byteSubstr(runtime, index, 1);
-        if (obj.isNil() || ((RubyString)obj).getByteList().length() == 0) return runtime.getNil();
+        if (obj.isNil() || ((RubyString)obj).getByteList().byteLength() == 0) return runtime.getNil();
         return obj;
     }
 
@@ -3652,7 +3652,7 @@ final IRubyObject uptoCommon(ThreadContext context, IRubyObject arg, boolean exc
             IRubyObject b = stringToInum(10);
             IRubyObject e = end.stringToInum(10);
 
-            RubyArray argsArr = RubyArray.newArray(runtime, RubyFixnum.newFixnum(runtime, value.length()), context.nil);
+            RubyArray argsArr = RubyArray.newArray(runtime, RubyFixnum.newFixnum(runtime, value.byteLength()), context.nil);
 
             if (b instanceof RubyFixnum && e instanceof RubyFixnum) {
                 long bl = RubyNumeric.fix2long(b);
@@ -3661,7 +3661,7 @@ final IRubyObject uptoCommon(ThreadContext context, IRubyObject arg, boolean exc
                 while (bl <= el) {
                     if (excl && bl == el) break;
                     argsArr.eltSetOk(1, RubyFixnum.newFixnum(runtime, bl));
-                    ByteList to = new ByteList(value.length() + 5);
+                    ByteList to = new ByteList(value.byteLength() + 5);
                     Sprintf.sprintf(to, "%.*d", argsArr);
                     RubyString str = RubyString.newStringNoCopy(runtime, to, USASCIIEncoding.INSTANCE, CR_7BIT);
                     block.yield(context, asSymbol ? runtime.newSymbol(str.toString()) : str);
@@ -3673,7 +3673,7 @@ final IRubyObject uptoCommon(ThreadContext context, IRubyObject arg, boolean exc
 
                 while (op.call(context, b, b, e).isTrue()) {
                     argsArr.eltSetOk(1, b);
-                    ByteList to = new ByteList(value.length() + 5);
+                    ByteList to = new ByteList(value.byteLength() + 5);
                     Sprintf.sprintf(to, "%.*d", argsArr);
                     RubyString str = RubyString.newStringNoCopy(runtime, to, USASCIIEncoding.INSTANCE, CR_7BIT);
                     block.yield(context, asSymbol ? runtime.newSymbol(str.toString()) : str);
@@ -3702,7 +3702,7 @@ private IRubyObject uptoCommonNoDigits(ThreadContext context, RubyString end, bo
             if (next == null) break;
             current = next.convertToString();
             if (excl && current.op_equal19(context, end).isTrue()) break;
-            if (current.getByteList().length() > end.getByteList().length() || current.getByteList().length() == 0) break;
+            if (current.getByteList().byteLength() > end.getByteList().length() || current.getByteList().length() == 0) break;
         }
         return this;
     }
@@ -4933,19 +4933,19 @@ public IRubyObject count(ThreadContext context, IRubyObject arg) {
         final ByteList countValue = countStr.getByteList();
         final Encoding enc = checkEncoding(countStr);
 
-        if ( countValue.length() == 1 && enc.isAsciiCompatible() ) {
+        if ( countValue.byteLength() == 1 && enc.isAsciiCompatible() ) {
             final byte[] countBytes = countValue.unsafeBytes();
-            final int begin = countValue.begin(), size = countValue.length();
+            final int begin = countValue.begin(), size = countValue.byteLength();
             if ( enc.isReverseMatchAllowed(countBytes, begin, begin + size) && ! isCodeRangeBroken() ) {
-                if ( value.length() == 0 ) return RubyFixnum.zero(runtime);
+                if ( value.byteLength() == 0 ) return RubyFixnum.zero(runtime);
 
                 int n = 0;
                 int[] len_p = {0};
                 int c = EncodingUtils.encCodepointLength(runtime, countBytes, begin, begin + size, len_p, enc);
 
                 final byte[] bytes = value.unsafeBytes();
                 int i = value.begin();
-                final int end = i + value.length();
+                final int end = i + value.byteLength();
                 while ( i < end ) {
                     if ( ( bytes[i++] & 0xff ) == c ) n++;
                 }
@@ -4963,7 +4963,7 @@ public IRubyObject count(ThreadContext context, IRubyObject arg) {
     public IRubyObject count(ThreadContext context, IRubyObject[] args) {
         final Ruby runtime = context.runtime;
 
-        if ( value.length() == 0 ) return RubyFixnum.zero(runtime);
+        if ( value.byteLength() == 0 ) return RubyFixnum.zero(runtime);
 
         RubyString countStr = args[0].convertToString();
         Encoding enc = checkEncoding(countStr);
@@ -5365,7 +5365,7 @@ public RubyString each_byte(ThreadContext context, Block block) {
         Ruby runtime = context.runtime;
         // Check the length every iteration, since
         // the block can modify this string.
-        for (int i = 0; i < value.length(); i++) {
+        for (int i = 0; i < value.byteLength(); i++) {
             block.yield(context, runtime.newFixnum(value.get(i) & 0xFF));
         }
         return this;
@@ -5690,7 +5690,7 @@ private RubySymbol checkSpecialCasesIntern(ByteList value) {
 
         for (int i = 0; i < opTable.length; i++) {
             String op = opTable[i][1];
-            if (value.toString().equals(op)) {
+            if (value.toByteString().equals(op)) {
                 return getRuntime().getSymbolTable().getSymbol(opTable[i][0]);
             }
         }

diff --git a/core/src/main/java/org/jruby/RubyStruct.java b/core/src/main/java/org/jruby/RubyStruct.java
@@ -184,7 +184,7 @@ public static RubyClass newInstance(IRubyObject recv, IRubyObject[] args, Block
         if (args.length > 0) {
             IRubyObject firstArgAsString = args[0].checkStringType();
             if (!firstArgAsString.isNil()) {
-                name = ((RubyString)firstArgAsString).getByteList().toString();
+                name = ((RubyString)firstArgAsString).getByteList().toByteString();
             } else if (args[0].isNil()) {
                 nilName = true;
             }

diff --git a/core/src/main/java/org/jruby/RubySymbol.java b/core/src/main/java/org/jruby/RubySymbol.java
@@ -757,7 +757,7 @@ public RubySymbol getSymbol(ByteList bytes, boolean hard) {
 
             if (symbol == null) {
                 bytes = bytes.dup();
-                symbol = createSymbol(bytes.toString(), bytes, hash, hard);
+                symbol = createSymbol(bytes.toByteString(), bytes, hash, hard);
             }
 
             return symbol;
@@ -1009,7 +1009,7 @@ private static int javaStringHashCode(String str) {
 
     private static int javaStringHashCode(ByteList iso8859) {
         int h = 0;
-        int length = iso8859.length();
+        int length = iso8859.byteLength();
         if (length > 0) {
             byte val[] = iso8859.getUnsafeBytes();
             int begin = iso8859.begin();
@@ -1047,9 +1047,9 @@ public static String objectToSymbolString(IRubyObject object) {
             return ((RubySymbol) object).toString();
         }
         if (object instanceof RubyString) {
-            return ((RubyString) object).getByteList().toString();
+            return ((RubyString) object).getByteList().toByteString();
         }
-        return object.convertToString().getByteList().toString();
+        return object.convertToString().getByteList().toByteString();
     }
 
     private static final class SymbolProcBody extends ContextAwareBlockBody {

diff --git a/core/src/main/java/org/jruby/ext/digest/RubyDigest.java b/core/src/main/java/org/jruby/ext/digest/RubyDigest.java
@@ -159,7 +159,7 @@ public static RubyString hexencode(IRubyObject self, IRubyObject arg) {
     @JRubyMethod(name = "bubblebabble", required = 1, meta = true)
     public static RubyString bubblebabble(IRubyObject recv, IRubyObject arg) {
         final ByteList bytes = arg.convertToString().getByteList();
-        return RubyString.newString(recv.getRuntime(), BubbleBabble.bubblebabble(bytes.unsafeBytes(), bytes.begin(), bytes.length()));
+        return RubyString.newString(recv.getRuntime(), BubbleBabble.bubblebabble(bytes.unsafeBytes(), bytes.begin(), bytes.byteLength()));
     }
 
     private static class Metadata {

diff --git a/core/src/main/java/org/jruby/ext/ffi/AbstractMemory.java b/core/src/main/java/org/jruby/ext/ffi/AbstractMemory.java
@@ -1853,22 +1853,22 @@ public IRubyObject read_array_of_string(ThreadContext context, IRubyObject rbLen
     public IRubyObject put_string(ThreadContext context, IRubyObject offArg, IRubyObject strArg) {
         long off = getOffset(offArg);
         ByteList bl = strArg.convertToString().getByteList();
-        getMemoryIO().putZeroTerminatedByteArray(off, bl.getUnsafeBytes(), bl.begin(), bl.length());
+        getMemoryIO().putZeroTerminatedByteArray(off, bl.getUnsafeBytes(), bl.begin(), bl.byteLength());
         return this;
     }
 
     @JRubyMethod(name = "write_string")
     public IRubyObject write_string(ThreadContext context, IRubyObject strArg) {
         ByteList bl = strArg.convertToString().getByteList();
-        getMemoryIO().put(0, bl.getUnsafeBytes(), bl.begin(), bl.length());
+        getMemoryIO().put(0, bl.getUnsafeBytes(), bl.begin(), bl.byteLength());
         return this;
     }
 
     @JRubyMethod(name = "write_string")
     public IRubyObject write_string(ThreadContext context, IRubyObject strArg, IRubyObject lenArg) {
         ByteList bl = strArg.convertToString().getByteList();
         getMemoryIO().put(0, bl.getUnsafeBytes(), bl.begin(),
-                Math.min(bl.length(), (int) org.jruby.RubyInteger.num2long(lenArg)));
+                Math.min(bl.byteLength(), (int) org.jruby.RubyInteger.num2long(lenArg)));
         return this;
     }
 
@@ -1879,11 +1879,11 @@ public IRubyObject get_bytes(ThreadContext context, IRubyObject offArg, IRubyObj
     }
 
     private IRubyObject putBytes(ThreadContext context, long off, ByteList bl, int idx, int len) {
-        if (idx < 0 || idx > bl.length()) {
+        if (idx < 0 || idx > bl.byteLength()) {
             throw context.runtime.newRangeError("invalid string index");
         }
 
-        if (len < 0 || len > (bl.length() - idx)) {
+        if (len < 0 || len > (bl.byteLength() - idx)) {
             throw context.runtime.newRangeError("invalid length");
         }
         getMemoryIO().put(off, bl.getUnsafeBytes(), bl.begin() + idx, len);
@@ -1895,7 +1895,7 @@ private IRubyObject putBytes(ThreadContext context, long off, ByteList bl, int i
     public IRubyObject put_bytes(ThreadContext context, IRubyObject[] args) {
         ByteList bl = args[1].convertToString().getByteList();
         int idx = args.length > 2 ? Util.int32Value(args[2]) : 0;
-        int len = args.length > 3 ? Util.int32Value(args[3]) : (bl.length() - idx);
+        int len = args.length > 3 ? Util.int32Value(args[3]) : (bl.byteLength() - idx);
 
         return putBytes(context, getOffset(args[0]), bl, idx, len);
     }
@@ -1909,7 +1909,7 @@ public IRubyObject read_bytes(ThreadContext context, IRubyObject lenArg) {
     public IRubyObject write_bytes(ThreadContext context, IRubyObject[] args) {
         ByteList bl = args[0].convertToString().getByteList();
         int idx = args.length > 1 ? Util.int32Value(args[1]) : 0;
-        int len = args.length > 2 ? Util.int32Value(args[2]) : (bl.length() - idx);
+        int len = args.length > 2 ? Util.int32Value(args[2]) : (bl.byteLength() - idx);
         return putBytes(context, 0, bl, idx, len);
     }