Squashed commit of the following:

commit d2d17a0e5c81fe0e11e6551a1ab6450fc5d4ee68 Author: Charles Oliver Nutter <headius@headius.com> Date: Tue Aug 1 00:55:34 2017 -0500 Use jcodings 1.0.24. commit 9abc2bbb264335c3d08d6c62eea9f1c412fc8e2c Author: Charles Oliver Nutter <headius@headius.com> Date: Tue Aug 1 00:49:08 2017 -0500 Tweaks for chr fixes. * Raise proper error for out of char range in encMbcput (found by ruby/spec) * Don't use single-arg chr from multi-arg path at all (caused infinite recursion; may impact perf by missing single-byte ByteLists) commit e343686 Author: Charles Oliver Nutter <headius@headius.com> Date: Mon Jul 31 23:44:04 2017 -0500 Various fixes for Integer#chr. * Fixed missing chr logic from MRI * Fixed encoding utility methods that were missing or different * Fixed error handling of invalid values * Updated jcodings to get sjis, single-byte fixes
jruby · Aug 1, 2017 · 0f746f6 · 0f746f6
1 parent ffbad9c
commit 0f746f6
Showing 5 changed files with 80 additions and 25 deletions.
diff --git a/core/pom.rb b/core/pom.rb
@@ -54,7 +54,7 @@
 
   jar 'org.jruby.joni:joni:2.1.12'
   jar 'org.jruby.extras:bytelist:1.0.15'
-  jar 'org.jruby.jcodings:jcodings:1.0.23'
+  jar 'org.jruby.jcodings:jcodings:1.0.24'
   jar 'org.jruby:dirgra:0.3'
 
   jar 'com.headius:invokebinder:1.9'

diff --git a/core/pom.xml b/core/pom.xml
@@ -183,7 +183,7 @@ DO NOT MODIFIY - GENERATED CODE
     <dependency>
       <groupId>org.jruby.jcodings</groupId>
       <artifactId>jcodings</artifactId>
-      <version>1.0.23</version>
+      <version>1.0.24-SNAPSHOT</version>
     </dependency>
     <dependency>
       <groupId>org.jruby</groupId>

diff --git a/core/src/main/java/org/jruby/RubyInteger.java b/core/src/main/java/org/jruby/RubyInteger.java
@@ -50,6 +50,7 @@
 import org.jruby.runtime.builtin.IRubyObject;
 import org.jruby.util.ByteList;
 import org.jruby.util.StringSupport;
+import org.jruby.util.io.EncodingUtils;
 
 import java.math.RoundingMode;
 
@@ -320,18 +321,22 @@ static ByteList singleCharByteList(final byte index) {
     @JRubyMethod(name = "chr")
     public RubyString chr(ThreadContext context) {
         Ruby runtime = context.runtime;
-        int value = (int) getLongValue();
-        if (value >= 0 && value <= 0xFF) {
-            ByteList bytes = SINGLE_CHAR_BYTELISTS[value];
-            return RubyString.newStringShared(runtime, bytes, bytes.getEncoding());
-        } else {
-            Encoding enc = runtime.getDefaultInternalEncoding();
-            if (value > 0xFF && (enc == null || enc == ASCIIEncoding.INSTANCE)) {
+
+        // rb_num_to_uint
+        long i = getLongValue() & 0xFFFFFFFFL;
+        int c = (int) i;
+
+        Encoding enc;
+
+        if (0xff < i) {
+            enc = runtime.getDefaultInternalEncoding();
+            if (enc == null) {
                 throw runtime.newRangeError(toString() + " out of char range");
             }
-            if (enc == null) enc = USASCIIEncoding.INSTANCE;
-            return RubyString.newStringNoCopy(runtime, fromEncodedBytes(runtime, enc, value), enc, 0);
+            return chrCommon(context, c, enc);
         }
+
+        return RubyString.newStringShared(runtime, SINGLE_CHAR_BYTELISTS[c]);
     }
 
     @Deprecated
@@ -342,17 +347,26 @@ public final RubyString chr19(ThreadContext context) {
     @JRubyMethod(name = "chr")
     public RubyString chr(ThreadContext context, IRubyObject arg) {
         Ruby runtime = context.runtime;
-        long value = getLongValue();
+
+        // rb_num_to_uint
+        long i = getLongValue() & 0xFFFFFFFFL;
+
         Encoding enc;
         if (arg instanceof RubyEncoding) {
             enc = ((RubyEncoding)arg).getEncoding();
         } else {
             enc =  arg.convertToString().toEncoding(runtime);
         }
-        if (enc == ASCIIEncoding.INSTANCE && value >= 0x80) {
-            return chr19(context);
+        return chrCommon(context, i, enc);
+    }
+
+    private RubyString chrCommon(ThreadContext context, long value, Encoding enc) {
+        if (value > 0xFFFFFFFFL) {
+            throw context.runtime.newRangeError(this + " out of char range");
         }
-        return RubyString.newStringNoCopy(runtime, fromEncodedBytes(runtime, enc, value), enc, 0);
+        int c = (int) value;
+        if (enc == null) enc = ASCIIEncoding.INSTANCE;
+        return EncodingUtils.encUintChr(context, c, enc);
     }
 
     @Deprecated

diff --git a/core/src/main/java/org/jruby/util/StringSupport.java b/core/src/main/java/org/jruby/util/StringSupport.java
@@ -153,7 +153,7 @@ public static int length(Encoding enc, byte[]bytes, int p, int end) {
 
     // rb_enc_precise_mbclen
     public static int preciseLength(Encoding enc, byte[]bytes, int p, int end) {
-        if (p >= end) return -1 - (1);
+        if (p >= end) return MBCLEN_NEEDMORE(1);
         int n = enc.length(bytes, p, end);
         if (n > end - p) return MBCLEN_NEEDMORE(n - (end - p));
         return n;

diff --git a/core/src/main/java/org/jruby/util/io/EncodingUtils.java b/core/src/main/java/org/jruby/util/io/EncodingUtils.java
@@ -4,6 +4,8 @@
 import org.jcodings.EncodingDB;
 import org.jcodings.Ptr;
 import org.jcodings.ascii.AsciiTables;
+import org.jcodings.exception.EncodingError;
+import org.jcodings.exception.EncodingException;
 import org.jcodings.exception.ErrorCodes;
 import org.jcodings.specific.ASCIIEncoding;
 import org.jcodings.specific.USASCIIEncoding;
@@ -1990,31 +1992,62 @@ public static IRubyObject ioEncStr(Ruby runtime, IRubyObject str, OpenFile fptr)
     }
 
     // rb_enc_uint_chr
-    public static IRubyObject encUintChr(ThreadContext context, int code, Encoding enc) {
+    public static RubyString encUintChr(ThreadContext context, int code, Encoding enc) {
         Ruby runtime = context.runtime;
 
+        long i = code & 0xFFFFFFFFL;
+
         int n;
-        switch (n = enc.codeToMbcLength(code)) {
+        switch (n = EncodingUtils.encCodelen(context, code, enc)) {
             case ErrorCodes.ERR_INVALID_CODE_POINT_VALUE:
-                throw runtime.newRangeError("invalid codepoint " + Integer.toHexString(code) + " in " + enc);
+                throw runtime.newRangeError("invalid codepoint " + Long.toHexString(i) + " in " + enc);
             case ErrorCodes.ERR_TOO_BIG_WIDE_CHAR_VALUE:
             case 0:
-                throw runtime.newRangeError(Integer.toString(code) + " out of char range");
+                throw runtime.newRangeError(Long.toString(i) + " out of char range");
         }
+
         ByteList strBytes = new ByteList(n);
         strBytes.setEncoding(enc);
         strBytes.length(n);
-        enc.codeToMbc(code, strBytes.unsafeBytes(), strBytes.begin());
-        if (enc.length(strBytes.unsafeBytes(), strBytes.begin(), strBytes.realSize()) != n) {
-            throw runtime.newRangeError("invalid codepoint " + Integer.toHexString(code) + " in " + enc);
+        byte[] bytes = strBytes.unsafeBytes();
+        int begin = strBytes.begin();
+        int end = strBytes.realSize();
+
+        encMbcput(context, code, bytes, begin, enc);
+        if (StringSupport.preciseLength(enc, bytes, begin, end) != n) {
+            throw runtime.newRangeError("invalid codepoint " + Long.toHexString(i) + " in " + enc);
         }
+
         return RubyString.newString(runtime, strBytes);
 
     }
 
-    // rb_enc_mbcput
+    // rb_enc_mbcput with Java exception
     public static void encMbcput(int c, byte[] buf, int p, Encoding enc) {
-        enc.codeToMbc(c, buf, p);
+        int len = enc.codeToMbc(c, buf, p);
+        if (len < 0) {
+            throw new EncodingException(EncodingError.fromCode(len));
+        }
+    }
+
+    // rb_enc_mbcput with Ruby exception
+    public static void encMbcput(ThreadContext context, int c, byte[] buf, int p, Encoding enc) {
+        int len = enc.codeToMbc(c, buf, p);
+
+        // in MRI, this check occurs within some of the individual encoding functions, such as the
+        // US-ASCII check for values >= 0x80. In MRI, unlike in JRuby, we can't throw Ruby errors
+        // from within encoding logic, so we try to reproduce the expected results via normal
+        // error codes here.
+        // See MRI's rb_enc_mbcput and related downstream encoding functions.
+        if (len < 0) {
+            switch (len) {
+                case ErrorCodes.ERR_INVALID_CODE_POINT_VALUE:
+                    throw context.runtime.newRangeError("invalid codepoint " + Long.toHexString(c & 0xFFFFFFFFL) + " in " + enc);
+                case ErrorCodes.ERR_TOO_BIG_WIDE_CHAR_VALUE:
+                    throw context.runtime.newRangeError("" + (c & 0xFFFFFFFFL) + " out of char range");
+            }
+            throw context.runtime.newEncodingError(EncodingError.fromCode(len).getMessage());
+        }
     }
 
     // rb_enc_codepoint_len
@@ -2210,4 +2243,12 @@ else if (c < 0x10000) {
         return buf.length;
     }
 
+    public static int encCodelen(ThreadContext context, int c, Encoding enc) {
+        int n = enc.codeToMbcLength(c);
+        if (n == 0) {
+            throw context.runtime.newArgumentError("invalid codepoint " + Long.toHexString(c & 0xFFFFFFFFL) + " in " + enc);
+        }
+        return n;
+    }
+
 }