encodings.pl: New file.

* scripts/encodings.pl: New file. * Makefile.in: Rebuilt. * Makefile.am (convert_source_files): Added IOConverter.java. * gnu/gcj/convert/UnicodeToBytes.java (UnicodeToBytes): Extend IOConverter. (getDefaultDecodingClass): Canonicalize default encoding name. (getEncoder): Likewise. * gnu/gcj/convert/BytesToUnicode.java (BytesToUnicode): Extend IOConverter. (getDefaultDecodingClass): Canonicalize default encoding name. (getDecoder): Likewise. * gnu/gcj/convert/IOConverter.java: New file. From-SVN: r35432
2024-11-30 15:24:07 +08:00 · 2000-08-02 19:56:53 +00:00 · 2000-08-02 19:56:53 +00:00 · f9427d1704
commit f9427d1704
parent 5f51a7528f
7 changed files with 186 additions and 30 deletions
--- a/libjava/ChangeLog
+++ b/libjava/ChangeLog
@ -1,3 +1,18 @@
+2000-08-02  Tom Tromey  <tromey@cygnus.com>
+
+	* scripts/encodings.pl: New file.
+	* Makefile.in: Rebuilt.
+	* Makefile.am (convert_source_files): Added IOConverter.java.
+	* gnu/gcj/convert/UnicodeToBytes.java (UnicodeToBytes): Extend
+	IOConverter.
+	(getDefaultDecodingClass): Canonicalize default encoding name.
+	(getEncoder): Likewise.
+	* gnu/gcj/convert/BytesToUnicode.java (BytesToUnicode): Extend
+	IOConverter.
+	(getDefaultDecodingClass): Canonicalize default encoding name.
+	(getDecoder): Likewise.
+	* gnu/gcj/convert/IOConverter.java: New file.
+
 2000-08-02  Bryce McKinlay  <bryce@albatross.co.nz>

        * interpret.cc (_Jv_InterpMethod::continue1): Type of `fun' changed
--- a/libjava/Makefile.am
+++ b/libjava/Makefile.am
@ -473,6 +473,7 @@ gnu/gcj/convert/Input_JavaSrc.java \
 gnu/gcj/convert/Input_SJIS.java \
 gnu/gcj/convert/Input_UTF8.java	\
 gnu/gcj/convert/Input_iconv.java \
+gnu/gcj/convert/IOConverter.java \
 gnu/gcj/convert/Output_8859_1.java \
 gnu/gcj/convert/Output_EUCJIS.java \
 gnu/gcj/convert/Output_JavaSrc.java \
@ -494,8 +495,8 @@ java/awt/AWTEventMulticaster.java \
 java/awt/AWTException.java \
 java/awt/ActiveEvent.java \
 java/awt/Adjustable.java \
-java/awt/Button.java \
 java/awt/BorderLayout.java \
+java/awt/Button.java \
 java/awt/CheckboxGroup.java \
 java/awt/Color.java \
 java/awt/Component.java	\
--- a/libjava/Makefile.in
+++ b/libjava/Makefile.in
@ -115,31 +115,48 @@ here = @here@
 libgcj_basedir = @libgcj_basedir@

 AUTOMAKE_OPTIONS = foreign no-installinfo
-@TESTSUBDIR_TRUE@SUBDIRS = @TESTSUBDIR_TRUE@$(DIRLTDL) testsuite gcj include
-@TESTSUBDIR_FALSE@SUBDIRS = @TESTSUBDIR_FALSE@$(DIRLTDL) gcj include
-@USE_LIBDIR_TRUE@toolexeclibdir = @USE_LIBDIR_TRUE@$(libdir)$(MULTISUBDIR)
-@USE_LIBDIR_FALSE@toolexeclibdir = @USE_LIBDIR_FALSE@$(toolexecdir)/lib$(MULTISUBDIR)
-@USE_LIBDIR_FALSE@toolexecdir = @USE_LIBDIR_FALSE@$(exec_prefix)/$(target_alias)
+@TESTSUBDIR_TRUE@SUBDIRS = \
+@TESTSUBDIR_TRUE@$(DIRLTDL) testsuite gcj include
+@TESTSUBDIR_FALSE@SUBDIRS = \
+@TESTSUBDIR_FALSE@$(DIRLTDL) gcj include
+@USE_LIBDIR_TRUE@toolexeclibdir = \
+@USE_LIBDIR_TRUE@$(libdir)$(MULTISUBDIR)
+@USE_LIBDIR_FALSE@toolexeclibdir = \
+@USE_LIBDIR_FALSE@$(toolexecdir)/lib$(MULTISUBDIR)
+@USE_LIBDIR_FALSE@toolexecdir = \
+@USE_LIBDIR_FALSE@$(exec_prefix)/$(target_alias)

 toolexeclib_LTLIBRARIES = libgcj.la
 toolexeclib_DATA = libgcj.spec
 data_DATA = libgcj.zip

-@NEEDS_DATA_START_TRUE@toolexeclib_LIBRARIES = @NEEDS_DATA_START_TRUE@libgcjdata.a
-@NEEDS_DATA_START_TRUE@libgcjdata_a_SOURCES = @NEEDS_DATA_START_TRUE@libgcjdata.c
+@NEEDS_DATA_START_TRUE@toolexeclib_LIBRARIES = \
+@NEEDS_DATA_START_TRUE@libgcjdata.a
+@NEEDS_DATA_START_TRUE@libgcjdata_a_SOURCES = \
+@NEEDS_DATA_START_TRUE@libgcjdata.c

-@NATIVE_TRUE@bin_PROGRAMS = @NATIVE_TRUE@jv-convert gij
+@NATIVE_TRUE@bin_PROGRAMS = \
+@NATIVE_TRUE@jv-convert gij

 bin_SCRIPTS = addr2name.awk
-@CANADIAN_TRUE@@NULL_TARGET_TRUE@GCJ = @CANADIAN_TRUE@@NULL_TARGET_TRUE@gcj
-@CANADIAN_TRUE@@NULL_TARGET_FALSE@GCJ = @CANADIAN_TRUE@@NULL_TARGET_FALSE@$(target_alias)-gcj
-@CANADIAN_FALSE@GCJ = @CANADIAN_FALSE@$(expanded)/gcj$(EXEEXT) -B$(expanded)/
-@CANADIAN_TRUE@@NULL_TARGET_TRUE@ZIP = @CANADIAN_TRUE@@NULL_TARGET_TRUE@$(MULTIBUILDTOP)../$(COMPPATH)/zip/zip$(EXEEXT)
-@CANADIAN_TRUE@@NULL_TARGET_FALSE@ZIP = @CANADIAN_TRUE@@NULL_TARGET_FALSE@zip
-@CANADIAN_FALSE@ZIP = @CANADIAN_FALSE@$(MULTIBUILDTOP)../$(COMPPATH)/zip/zip$(EXEEXT)
-@CANADIAN_TRUE@GCJH = @CANADIAN_TRUE@gcjh
-@CANADIAN_FALSE@GCJH = @CANADIAN_FALSE@$(MULTIBUILDTOP)../$(COMPPATH)/gcc/gcjh$(EXEEXT)
-@CANADIAN_FALSE@expanded = @CANADIAN_FALSE@`cd $(MULTIBUILDTOP)../$(COMPPATH)/gcc && pwd`
+@CANADIAN_TRUE@@NULL_TARGET_TRUE@GCJ = \
+@CANADIAN_TRUE@@NULL_TARGET_TRUE@gcj
+@CANADIAN_TRUE@@NULL_TARGET_FALSE@GCJ = \
+@CANADIAN_TRUE@@NULL_TARGET_FALSE@$(target_alias)-gcj
+@CANADIAN_FALSE@GCJ = \
+@CANADIAN_FALSE@$(expanded)/gcj$(EXEEXT) -B$(expanded)/
+@CANADIAN_TRUE@@NULL_TARGET_TRUE@ZIP = \
+@CANADIAN_TRUE@@NULL_TARGET_TRUE@$(MULTIBUILDTOP)../$(COMPPATH)/zip/zip$(EXEEXT)
+@CANADIAN_TRUE@@NULL_TARGET_FALSE@ZIP = \
+@CANADIAN_TRUE@@NULL_TARGET_FALSE@zip
+@CANADIAN_FALSE@ZIP = \
+@CANADIAN_FALSE@$(MULTIBUILDTOP)../$(COMPPATH)/zip/zip$(EXEEXT)
+@CANADIAN_TRUE@GCJH = \
+@CANADIAN_TRUE@gcjh
+@CANADIAN_FALSE@GCJH = \
+@CANADIAN_FALSE@$(MULTIBUILDTOP)../$(COMPPATH)/gcc/gcjh$(EXEEXT)
+@CANADIAN_FALSE@expanded = \
+@CANADIAN_FALSE@`cd $(MULTIBUILDTOP)../$(COMPPATH)/gcc && pwd`

 GCJCOMPILE = CLASSPATH=$(here) $(LIBTOOL) --mode=compile $(GCJ) -fassume-compiled -L$(here) $(JC1FLAGS) -c
 GCJLINK = $(LIBTOOL) --mode=link $(GCJ) -L$(here) $(JC1FLAGS) $(LDFLAGS) -o $@
@ -153,8 +170,10 @@ WARNINGS = -W -Wall
 AM_CXXFLAGS = -fno-rtti -fvtable-thunks -fasynchronous-exceptions \
 	@LIBGCJ_CXXFLAGS@ @EXCEPTIONSPEC@ $(WARNINGS) -D_GNU_SOURCE

-@USING_GCC_TRUE@AM_CFLAGS = @USING_GCC_TRUE@@LIBGCJ_CFLAGS@ $(WARNINGS)
-@USING_GCC_FALSE@AM_CFLAGS = @USING_GCC_FALSE@@LIBGCJ_CFLAGS@
+@USING_GCC_TRUE@AM_CFLAGS = \
+@USING_GCC_TRUE@@LIBGCJ_CFLAGS@ $(WARNINGS)
+@USING_GCC_FALSE@AM_CFLAGS = \
+@USING_GCC_FALSE@@LIBGCJ_CFLAGS@

 JCFLAGS = -g
 JC1FLAGS = -g @LIBGCJ_JAVAFLAGS@
@ -206,7 +225,8 @@ extra_headers = java/lang/Object.h java/lang/Class.h

 NM = nm

-@NATIVE_TRUE@@MAINTAINER_MODE_TRUE@noinst_PROGRAMS = @NATIVE_TRUE@@MAINTAINER_MODE_TRUE@gen-from-JIS
+@NATIVE_TRUE@@MAINTAINER_MODE_TRUE@noinst_PROGRAMS = \
+@NATIVE_TRUE@@MAINTAINER_MODE_TRUE@gen-from-JIS

 CONVERT_DIR = gnu/gcj/convert

@ -250,6 +270,7 @@ gnu/gcj/convert/Input_JavaSrc.java \
 gnu/gcj/convert/Input_SJIS.java \
 gnu/gcj/convert/Input_UTF8.java	\
 gnu/gcj/convert/Input_iconv.java \
+gnu/gcj/convert/IOConverter.java \
 gnu/gcj/convert/Output_8859_1.java \
 gnu/gcj/convert/Output_EUCJIS.java \
 gnu/gcj/convert/Output_JavaSrc.java \
@ -270,8 +291,8 @@ java/awt/AWTEventMulticaster.java \
 java/awt/AWTException.java \
 java/awt/ActiveEvent.java \
 java/awt/Adjustable.java \
-java/awt/Button.java \
 java/awt/BorderLayout.java \
+java/awt/Button.java \
 java/awt/CheckboxGroup.java \
 java/awt/Color.java \
 java/awt/Component.java	\
@ -992,7 +1013,8 @@ DEP_FILES =  .deps/$(srcdir)/$(CONVERT_DIR)/gen-from-JIS.P \
 .deps/defineclass.P .deps/exception.P .deps/gij.P \
 .deps/gnu/gcj/RawData.P .deps/gnu/gcj/awt/BitMaskExtent.P \
 .deps/gnu/gcj/awt/Buffers.P .deps/gnu/gcj/convert/BytesToUnicode.P \
-.deps/gnu/gcj/convert/Convert.P .deps/gnu/gcj/convert/Input_8859_1.P \
+.deps/gnu/gcj/convert/Convert.P .deps/gnu/gcj/convert/IOConverter.P \
+.deps/gnu/gcj/convert/Input_8859_1.P \
 .deps/gnu/gcj/convert/Input_EUCJIS.P \
 .deps/gnu/gcj/convert/Input_JavaSrc.P \
 .deps/gnu/gcj/convert/Input_SJIS.P .deps/gnu/gcj/convert/Input_UTF8.P \
@ -1765,7 +1787,7 @@ distdir: $(DISTFILES)
 	@for file in $(DISTFILES); do \
 	  d=$(srcdir); \
 	  if test -d $$d/$$file; then \
-	    cp -pr $$d/$$file $(distdir)/$$file; \
+	    cp -pr $$/$$file $(distdir)/$$file; \
 	  else \
 	    test -f $(distdir)/$$file \
 	    || ln $$d/$$file $(distdir)/$$file 2> /dev/null \
--- a/libjava/gnu/gcj/convert/BytesToUnicode.java
+++ b/libjava/gnu/gcj/convert/BytesToUnicode.java
@ -8,7 +8,7 @@ details.  */

 package gnu.gcj.convert;

-public abstract class BytesToUnicode
+public abstract class BytesToUnicode extends IOConverter
 {
  /** Buffer to read bytes from.
   * The characters inbuffer[inpos] ... inbuffer[inlength-1] are available. */
@ -25,7 +25,7 @@ public abstract class BytesToUnicode
    // Test (defaultDecodingClass == null) again in case of race condition.
    if (defaultDecodingClass == null)
      {
-	String encoding = System.getProperty("file.encoding");
+	String encoding = canonicalize (System.getProperty("file.encoding"));
 	String className = "gnu.gcj.convert.Input_"+encoding;
 	try
 	  {
@ -60,7 +60,7 @@ public abstract class BytesToUnicode
  public static BytesToUnicode getDecoder (String encoding)
    throws java.io.UnsupportedEncodingException
  {
-    String className = "gnu.gcj.convert.Input_"+encoding;
+    String className = "gnu.gcj.convert.Input_" + canonicalize (encoding);
    Class decodingClass;
    try 
      { 
@ -71,6 +71,8 @@ public abstract class BytesToUnicode
      { 
 	try
 	  {
+	    // We pass the original name to iconv and let it handle
+	    // its own aliasing.
 	    return new Input_iconv (encoding);
 	  }
 	catch (Throwable _)
--- a/libjava/gnu/gcj/convert/IOConverter.java
+++ b/libjava/gnu/gcj/convert/IOConverter.java
@ -0,0 +1,52 @@
+/* Copyright (C) 2000  Free Software Foundation
+
+   This file is part of libgcj.
+
+This software is copyrighted work licensed under the terms of the
+Libgcj License.  Please consult the file "LIBGCJ_LICENSE" for
+details.  */
+
+/* This is a base class that handles aliasing issues for
+   UnicodeToBytes to BytesToUnicode.  */
+
+package gnu.gcj.convert;
+
+import java.util.Hashtable;
+
+public abstract class IOConverter
+{
+  // Map encoding aliases to our canonical form.
+  static private Hashtable hash = new Hashtable ();
+
+  static
+  {
+    // Manually maintained aliases.  Note that the value must be our
+    // canonical name.
+    hash.put ("ISO-Latin-1", "8859_1");
+    // All aliases after this point are automatically generated by the
+    // `encodings.pl' script.  Run it to make any corrections.
+    hash.put ("ISO_8859-1:1987", "8859_1");
+    hash.put ("iso-ir-100", "8859_1");
+    hash.put ("ISO_8859-1", "8859_1");
+    hash.put ("ISO-8859-1", "8859_1");
+    hash.put ("latin1", "8859_1");
+    hash.put ("l1", "8859_1");
+    hash.put ("IBM819", "8859_1");
+    hash.put ("CP819", "8859_1");
+    hash.put ("csISOLatin1", "8859_1");
+    hash.put ("UTF-8", "UTF8");
+    hash.put ("Shift_JIS", "SJIS");
+    hash.put ("MS_Kanji", "SJIS");
+    hash.put ("csShiftJIS", "SJIS");
+    hash.put ("Extended_UNIX_Code_Packed_Format_for_Japanese", "EUCJIS");
+    hash.put ("csEUCPkdFmtJapanese", "EUCJIS");
+    hash.put ("EUC-JP", "EUCJIS");
+  }
+
+  // Turn an alias into the canonical form.
+  protected static final String canonicalize (String name)
+  {
+    String c = (String) hash.get (name);
+    return c == null ? name : c;
+  }
+}
--- a/libjava/gnu/gcj/convert/UnicodeToBytes.java
+++ b/libjava/gnu/gcj/convert/UnicodeToBytes.java
@ -8,7 +8,7 @@ details.  */

 package gnu.gcj.convert; 
 
-public abstract class UnicodeToBytes
+public abstract class UnicodeToBytes extends IOConverter
 {
  /** Buffer to emit bytes to.
   * The locations buf[count] ... buf[buf.length-1] are available. */
@ -22,7 +22,7 @@ public abstract class UnicodeToBytes
    // Test (defaultEncodingClass == null) again in case of race condition.
    if (defaultEncodingClass == null)
      {
-	String encoding = System.getProperty("file.encoding");
+	String encoding = canonicalize (System.getProperty("file.encoding"));
 	String className = "gnu.gcj.convert.Output_"+encoding;
 	try
 	  {
@ -58,7 +58,7 @@ public abstract class UnicodeToBytes
  public static UnicodeToBytes getEncoder (String encoding)
    throws java.io.UnsupportedEncodingException
  {
-    String className = "gnu.gcj.convert.Output_"+encoding;
+    String className = "gnu.gcj.convert.Output_" + canonicalize (encoding);
    Class encodingClass;
    try 
      { 
@ -69,6 +69,8 @@ public abstract class UnicodeToBytes
      { 
 	try
 	  {
+	    // We pass the original name to iconv and let it handle
+	    // its own aliasing.
 	    return new Output_iconv (encoding);
 	  }
 	catch (Throwable _)
--- a/libjava/scripts/encodings.pl
+++ b/libjava/scripts/encodings.pl
@ -0,0 +1,62 @@
+# encodings.pl - Download IANA text and compute alias list.
+# Assumes you are running this program from gnu/gcj/convert/.
+# Output suitable for direct inclusion in IOConverter.java.
+
+# Map IANA canonical names onto our canonical names.
+%map = (
+	'ISO_8859-1:1987' => '8859_1',
+	'UTF-8' => 'UTF8',
+	'Shift_JIS' => 'SJIS',
+	'Extended_UNIX_Code_Packed_Format_for_Japanese' => 'EUCJIS'
+	);
+
+if ($ARGV[0] eq '')
+{
+    $file = 'character-sets';
+    if (! -f $file)
+    {
+	# Too painful to figure out how to get Perl to do it.
+	system 'wget -o .wget-log http://www.isi.edu/in-notes/iana/assignments/character-sets';
+    }
+}
+else
+{
+    $file = $ARGV[0];
+}
+
+open (INPUT, "< $file") || die "couldn't open $file: $!";
+
+$body = 0;
+$current = '';
+while (<INPUT>)
+{
+    chop;
+    $body = 1 if /^Name:/;
+    next unless $body;
+
+    if (/^$/)
+    {
+	$current = '';
+	next;
+    }
+
+    ($type, $name) = split (/\s+/);
+    if ($type eq 'Name:')
+    {
+	$current = $map{$name};
+	if ($current)
+	{
+	    print "    hash.put (\"$name\", \"$current\");\n";
+	}
+    }
+    elsif ($type eq 'Alias:')
+    {
+	# The IANA list has some ugliness.
+	if ($name ne '' && $name ne 'NONE' && $current)
+	{
+	    print "    hash.put (\"$name\", \"$current\");\n";
+	}
+    }
+}
+
+close (INPUT);