[apple/icu.git] / icuSources / tools / genren / genren.pl

#!/usr/bin/perl 
#*
#*******************************************************************************
#*   Copyright (C) 2001-2010, International Business Machines
#*   Corporation and others.  All Rights Reserved.
#*******************************************************************************
#*
#*   file name:  genren.pl
#*   encoding:   US-ASCII
#*   tab size:   8 (not used)
#*   indentation:4
#*
#*   Created by: Vladimir Weinstein
#*   07/19/2001
#*
#*  Used to generate renaming headers.
#*  Run on UNIX platforms (linux) in order to catch all the exports

use POSIX qw(strftime);

$headername = 'urename.h';

$path = substr($0, 0, rindex($0, "/")+1)."../../common/unicode/uversion.h";

$nmopts = '-Cg -f s';
$post = '';

$mode = 'POSIX';

(-e $path) || die "Cannot find uversion.h";

open(UVERSION, $path);

while(<UVERSION>) {
    if(/\#define U_ICU_VERSION_SUFFIX/) {
        chop;
        s/\#define U_ICU_VERSION_SUFFIX //;
        $U_ICU_VERSION_SUFFIX = "$_";
        last;
    }
}

while($ARGV[0] =~ /^-/) { # detects whether there are any arguments
    $_ = shift @ARGV;      # extracts the argument for processing
    /^-v/ && ($VERBOSE++, next);                      # verbose
    /^-h/ && (&printHelpMsgAndExit, next);               # help
    /^-o/ && (($headername = shift (@ARGV)), next);   # output file
    /^-n/ && (($nmopts = shift (@ARGV)), next);   # nm opts
    /^-p/ && (($post = shift (@ARGV)), next);   # nm opts
    /^-x/ && (($mode = shift (@ARGV)), next);   # nm opts
    /^-S/ && (($U_ICU_VERSION_SUFFIX = shift(@ARGV)), next); # pick the suffix
    warn("Invalid option $_\n");
    &printHelpMsgAndExit;
}

unless(@ARGV > 0) {
    warn "No libraries, exiting...\n";
    &printHelpMsgAndExit;
}

#$headername = "uren".substr($ARGV[0], 6, index(".", $ARGV[0])-7).".h";
    
$HEADERDEF = uc($headername);  # this is building the constant for #define
$HEADERDEF =~ s/\./_/;

    
    open HEADER, ">$headername"; # opening a header file

#We will print our copyright here + warnings

$YEAR = strftime "%Y",localtime;

print HEADER <<"EndOfHeaderComment";
/*
*******************************************************************************
*   Copyright (C) 2002-$YEAR, International Business Machines
*   Corporation and others.  All Rights Reserved.
*******************************************************************************
*
*   file name:  $headername
*   encoding:   US-ASCII
*   tab size:   8 (not used)
*   indentation:4
*
*   Created by: Perl script written by Vladimir Weinstein
*
*  Contains data for renaming ICU exports.
*  Gets included by umachine.h
*
*  THIS FILE IS MACHINE-GENERATED, DON'T PLAY WITH IT IF YOU DON'T KNOW WHAT
*  YOU ARE DOING, OTHERWISE VERY BAD THINGS WILL HAPPEN!
*/

#ifndef $HEADERDEF
#define $HEADERDEF

/* Uncomment the following line to disable renaming on platforms
   that do not use Autoconf. */
/* #define U_DISABLE_RENAMING 1 */

#if !U_DISABLE_RENAMING

/* We need the U_ICU_ENTRY_POINT_RENAME definition. There's a default one in unicode/uvernum.h we can use, but we will give
   the platform a chance to define it first.
   Normally (if utypes.h or umachine.h was included first) this will not be necessary as it will already be defined.
 */
#ifndef U_ICU_ENTRY_POINT_RENAME
#include "unicode/umachine.h"
#endif

/* If we still don't have U_ICU_ENTRY_POINT_RENAME use the default. */
#ifndef U_ICU_ENTRY_POINT_RENAME
#include "unicode/uvernum.h"
#endif

/* Error out before the following defines cause very strange and unexpected code breakage */
#ifndef U_ICU_ENTRY_POINT_RENAME
#error U_ICU_ENTRY_POINT_RENAME is not defined - cannot continue. Consider defining U_DISABLE_RENAMING if renaming should not be used.
#endif

EndOfHeaderComment

$fileCount = 0;
$itemCount = 0;
$symbolCount = 0;

for(;@ARGV; shift(@ARGV)) {
    $fileCount++;
    @NMRESULT = `nm $nmopts $ARGV[0] $post`;
    if($?) {
        warn "Couldn't do 'nm' for $ARGV[0], continuing...\n";
        next; # Couldn't do nm for the file
    }
    if($mode =~ /POSIX/) {
        splice @NMRESULT, 0, 6;
    } elsif ($mode =~ /Mach-O/) {
#        splice @NMRESULT, 0, 10;
    }
    foreach (@NMRESULT) { # Process every line of result and stuff it in $_
        $itemCount++;
        if($mode =~ /POSIX/) {
            ($_, $address, $type) = split(/\|/);
        } elsif ($mode =~ /Mach-O/) {
            if(/^(?:[0-9a-fA-F]){8} ([A-Z]) (?:_)?(.*)$/) {
                ($_, $type) = ($2, $1);
            } else {
                next;
            }
        } else {
            die "Unknown mode $mode";
        }
        &verbose( "type: \"$type\" ");
        if(!($type =~ /[UAwW?]/)) {
            if(/@@/) { # These would be imports
                &verbose( "Import: $_ \"$type\"\n");
                &verbose( "C++ method: $_\n");
            } elsif (/^[^\(]*::/) { # C++ methods, stuff class name in associative array
	        ##  DON'T match    ...  (   foo::bar ...   want :: to be to the left of paren
                ## icu_2_0::CharString::~CharString(void) -> CharString
                @CppName = split(/::/); ## remove scope stuff
                if(@CppName>1) {
                    ## MessageFormat virtual table -> MessageFormat
                    @CppName = split(/ /, $CppName[1]); ## remove debug stuff
                }
                ## ures_getUnicodeStringByIndex(UResourceBundle -> ures_getUnicodeStringByIndex
                @CppName = split(/\(/, $CppName[0]); ## remove function args
                if($CppName[0] =~ /^operator/) {
                    &verbose ("Skipping C++ function: $_\n");
                } elsif($CppName[0] =~ /^~/) {
                    &verbose ("Skipping C++ destructor: $_\n");
                } else {
		    &verbose( " Class: '$CppName[0]': $_ \n");
                    $CppClasses{$CppName[0]}++;
		    $symbolCount++;
                }
	    } elsif ( my ($cfn) = m/^([A-Za-z0-9_]*)\(.*/ ) {
		&verbose ( "$ARGV[0]:  got global C++ function  $cfn with '$_'\n" );
                $CFuncs{$cfn}++;
		$symbolCount++;
            } elsif ( /\(/) { # These are strange functions
                print STDERR "$ARGV[0]: Not sure what to do with '$_'\n";
	    } elsif ( /^_init/ ) {
		&verbose( "$ARGV[0]: Skipped initializer $_\n" );
	    } elsif ( /^_fini/ ) {
		&verbose( "$ARGV[0]: Skipped finilizer $_\n" );
            } elsif ( /icu_/) {
                print STDERR "$ARGV[0]: Skipped strange mangled function $_\n";
            } elsif ( /^vtable for /) {
                print STDERR "$ARGV[0]: Skipped vtable $_\n";
            } elsif ( /^typeinfo for /) {
                print STDERR "$ARGV[0]: Skipped typeinfo $_\n";
            } elsif ( /operator\+/ ) {
                print STDERR "$ARGV[0]: Skipped ignored function $_\n";
            } else { # This is regular C function 
                &verbose( "C func: $_\n");
                @funcname = split(/[\(\s+]/);
                $CFuncs{$funcname[0]}++;
		$symbolCount++;
            }
        } else {
            &verbose( "Skipped: $_ $1\n");
        }
    }
}

if( $fileCount == 0 ) {
  die "Error: $itemCount lines from $fileCount files processed, but $symbolCount symbols were found.\n";
}

if( $symbolCount == 0 ) {
  die "Error: $itemCount lines from $fileCount files processed, but $symbolCount symbols were found.\n";
}

print " Loaded $symbolCount symbols from $itemCount lines in $fileCount files.\n";

print HEADER "\n/* C exports renaming data */\n\n";
foreach(sort keys(%CFuncs)) {
    print HEADER "#define $_ U_ICU_ENTRY_POINT_RENAME($_)\n";
#    print HEADER "#define $_ $_$U_ICU_VERSION_SUFFIX\n";
}

print HEADER "\n\n";
print HEADER "/* C++ class names renaming defines */\n\n";
print HEADER "#ifdef XP_CPLUSPLUS\n";
print HEADER "#if !U_HAVE_NAMESPACE\n\n";
foreach(sort keys(%CppClasses)) {
    print HEADER "#define $_ U_ICU_ENTRY_POINT_RENAME($_)\n";
}
print HEADER "\n#endif\n";
print HEADER "#endif\n";
print HEADER "\n#endif\n";
print HEADER "\n#endif\n";

close HEADER;

sub verbose {
    if($VERBOSE) {
        print STDERR @_;
    }
}


sub printHelpMsgAndExit {
    print STDERR <<"EndHelpText";
Usage: $0 [OPTIONS] LIBRARY_FILES
  Options: 
    -v - verbose
    -h - help
    -o - output file name (defaults to 'urename.h'
    -S - suffix (defaults to _MAJOR_MINOR of current ICU version)
Will produce a renaming .h file

EndHelpText

    exit 0;

}
Commit	Line	Data
b75a7d8f A	1	#!/usr/bin/perl
	2	#*
	3	#*******************************************************************************
729e4ab9	4	#* Copyright (C) 2001-2010, International Business Machines
b75a7d8f A	5	#* Corporation and others. All Rights Reserved.
	6	#*******************************************************************************
	7	#*
	8	#* file name: genren.pl
	9	#* encoding: US-ASCII
	10	#* tab size: 8 (not used)
	11	#* indentation:4
	12	#*
	13	#* Created by: Vladimir Weinstein
	14	#* 07/19/2001
	15	#*
	16	#* Used to generate renaming headers.
	17	#* Run on UNIX platforms (linux) in order to catch all the exports
	18
729e4ab9 A	19	use POSIX qw(strftime);
729e4ab9 A	20
b75a7d8f A	21	$headername = 'urename.h';
	22
	23	$path = substr($0, 0, rindex($0, "/")+1)."../../common/unicode/uversion.h";
	24
46f4442e A	25	$nmopts = '-Cg -f s';
	26	$post = '';
	27
729e4ab9	28	$mode = 'POSIX';
46f4442e	29
b75a7d8f A	30	(-e $path) \|\| die "Cannot find uversion.h";
	31
	32	open(UVERSION, $path);
	33
	34	while(<UVERSION>) {
	35	if(/\#define U_ICU_VERSION_SUFFIX/) {
	36	chop;
	37	s/\#define U_ICU_VERSION_SUFFIX //;
	38	$U_ICU_VERSION_SUFFIX = "$_";
	39	last;
	40	}
	41	}
	42
	43	while($ARGV[0] =~ /^-/) { # detects whether there are any arguments
	44	$_ = shift @ARGV; # extracts the argument for processing
	45	/^-v/ && ($VERBOSE++, next); # verbose
	46	/^-h/ && (&printHelpMsgAndExit, next); # help
	47	/^-o/ && (($headername = shift (@ARGV)), next); # output file
46f4442e A	48	/^-n/ && (($nmopts = shift (@ARGV)), next); # nm opts
	49	/^-p/ && (($post = shift (@ARGV)), next); # nm opts
	50	/^-x/ && (($mode = shift (@ARGV)), next); # nm opts
b75a7d8f A	51	/^-S/ && (($U_ICU_VERSION_SUFFIX = shift(@ARGV)), next); # pick the suffix
	52	warn("Invalid option $_\n");
	53	&printHelpMsgAndExit;
	54	}
	55
	56	unless(@ARGV > 0) {
	57	warn "No libraries, exiting...\n";
	58	&printHelpMsgAndExit;
	59	}
	60
	61	#$headername = "uren".substr($ARGV[0], 6, index(".", $ARGV[0])-7).".h";
	62
	63	$HEADERDEF = uc($headername); # this is building the constant for #define
	64	$HEADERDEF =~ s/\./_/;
	65
	66
	67	open HEADER, ">$headername"; # opening a header file
	68
	69	#We will print our copyright here + warnings
729e4ab9 A	70
	71	$YEAR = strftime "%Y",localtime;
	72
b75a7d8f A	73	print HEADER <<"EndOfHeaderComment";
	74	/*
	75	*******************************************************************************
729e4ab9	76	* Copyright (C) 2002-$YEAR, International Business Machines
b75a7d8f A	77	* Corporation and others. All Rights Reserved.
	78	*******************************************************************************
	79	*
	80	* file name: $headername
	81	* encoding: US-ASCII
	82	* tab size: 8 (not used)
	83	* indentation:4
	84	*
	85	* Created by: Perl script written by Vladimir Weinstein
	86	*
	87	* Contains data for renaming ICU exports.
	88	* Gets included by umachine.h
	89	*
	90	* THIS FILE IS MACHINE-GENERATED, DON'T PLAY WITH IT IF YOU DON'T KNOW WHAT
	91	* YOU ARE DOING, OTHERWISE VERY BAD THINGS WILL HAPPEN!
	92	*/
	93
	94	#ifndef $HEADERDEF
	95	#define $HEADERDEF
	96
	97	/* Uncomment the following line to disable renaming on platforms
	98	that do not use Autoconf. */
	99	/* #define U_DISABLE_RENAMING 1 */
	100
	101	#if !U_DISABLE_RENAMING
729e4ab9 A	102
	103	/* We need the U_ICU_ENTRY_POINT_RENAME definition. There's a default one in unicode/uvernum.h we can use, but we will give
	104	the platform a chance to define it first.
	105	Normally (if utypes.h or umachine.h was included first) this will not be necessary as it will already be defined.
	106	*/
	107	#ifndef U_ICU_ENTRY_POINT_RENAME
	108	#include "unicode/umachine.h"
	109	#endif
	110
	111	/* If we still don't have U_ICU_ENTRY_POINT_RENAME use the default. */
	112	#ifndef U_ICU_ENTRY_POINT_RENAME
	113	#include "unicode/uvernum.h"
	114	#endif
	115
	116	/* Error out before the following defines cause very strange and unexpected code breakage */
	117	#ifndef U_ICU_ENTRY_POINT_RENAME
	118	#error U_ICU_ENTRY_POINT_RENAME is not defined - cannot continue. Consider defining U_DISABLE_RENAMING if renaming should not be used.
	119	#endif
	120
b75a7d8f A	121	EndOfHeaderComment
b75a7d8f A	122
729e4ab9 A	123	$fileCount = 0;
	124	$itemCount = 0;
	125	$symbolCount = 0;
	126
b75a7d8f	127	for(;@ARGV; shift(@ARGV)) {
729e4ab9	128	$fileCount++;
46f4442e	129	@NMRESULT = `nm $nmopts $ARGV[0] $post`;
b75a7d8f A	130	if($?) {
	131	warn "Couldn't do 'nm' for $ARGV[0], continuing...\n";
	132	next; # Couldn't do nm for the file
	133	}
46f4442e A	134	if($mode =~ /POSIX/) {
	135	splice @NMRESULT, 0, 6;
	136	} elsif ($mode =~ /Mach-O/) {
	137	# splice @NMRESULT, 0, 10;
	138	}
b75a7d8f	139	foreach (@NMRESULT) { # Process every line of result and stuff it in $_
729e4ab9	140	$itemCount++;
46f4442e A	141	if($mode =~ /POSIX/) {
	142	($_, $address, $type) = split(/\\|/);
	143	} elsif ($mode =~ /Mach-O/) {
	144	if(/^(?:[0-9a-fA-F]){8} ([A-Z]) (?:_)?(.*)$/) {
	145	($_, $type) = ($2, $1);
	146	} else {
	147	next;
	148	}
	149	} else {
	150	die "Unknown mode $mode";
	151	}
b75a7d8f A	152	&verbose( "type: \"$type\" ");
	153	if(!($type =~ /[UAwW?]/)) {
	154	if(/@@/) { # These would be imports
	155	&verbose( "Import: $_ \"$type\"\n");
b75a7d8f	156	&verbose( "C++ method: $_\n");
729e4ab9 A	157	} elsif (/^[^\(]*::/) { # C++ methods, stuff class name in associative array
729e4ab9 A	158	## DON'T match ... ( foo::bar ... want :: to be to the left of paren
b75a7d8f A	159	## icu_2_0::CharString::~CharString(void) -> CharString
	160	@CppName = split(/::/); ## remove scope stuff
	161	if(@CppName>1) {
	162	## MessageFormat virtual table -> MessageFormat
	163	@CppName = split(/ /, $CppName[1]); ## remove debug stuff
	164	}
	165	## ures_getUnicodeStringByIndex(UResourceBundle -> ures_getUnicodeStringByIndex
	166	@CppName = split(/\(/, $CppName[0]); ## remove function args
46f4442e A	167	if($CppName[0] =~ /^operator/) {
	168	&verbose ("Skipping C++ function: $_\n");
	169	} elsif($CppName[0] =~ /^~/) {
	170	&verbose ("Skipping C++ destructor: $_\n");
	171	} else {
729e4ab9	172	&verbose( " Class: '$CppName[0]': $_ \n");
46f4442e	173	$CppClasses{$CppName[0]}++;
729e4ab9	174	$symbolCount++;
46f4442e	175	}
729e4ab9 A	176	} elsif ( my ($cfn) = m/^([A-Za-z0-9_])\(./ ) {
	177	&verbose ( "$ARGV[0]: got global C++ function $cfn with '$_'\n" );
	178	$CFuncs{$cfn}++;
	179	$symbolCount++;
b75a7d8f	180	} elsif ( /\(/) { # These are strange functions
729e4ab9 A	181	print STDERR "$ARGV[0]: Not sure what to do with '$_'\n";
	182	} elsif ( /^_init/ ) {
	183	&verbose( "$ARGV[0]: Skipped initializer $_\n" );
	184	} elsif ( /^_fini/ ) {
	185	&verbose( "$ARGV[0]: Skipped finilizer $_\n" );
b75a7d8f	186	} elsif ( /icu_/) {
729e4ab9	187	print STDERR "$ARGV[0]: Skipped strange mangled function $_\n";
46f4442e	188	} elsif ( /^vtable for /) {
729e4ab9	189	print STDERR "$ARGV[0]: Skipped vtable $_\n";
46f4442e	190	} elsif ( /^typeinfo for /) {
729e4ab9	191	print STDERR "$ARGV[0]: Skipped typeinfo $_\n";
46f4442e	192	} elsif ( /operator\+/ ) {
729e4ab9	193	print STDERR "$ARGV[0]: Skipped ignored function $_\n";
b75a7d8f A	194	} else { # This is regular C function
	195	&verbose( "C func: $_\n");
	196	@funcname = split(/[\(\s+]/);
	197	$CFuncs{$funcname[0]}++;
729e4ab9	198	$symbolCount++;
b75a7d8f A	199	}
	200	} else {
	201	&verbose( "Skipped: $_ $1\n");
	202	}
	203	}
	204	}
	205
729e4ab9 A	206	if( $fileCount == 0 ) {
	207	die "Error: $itemCount lines from $fileCount files processed, but $symbolCount symbols were found.\n";
	208	}
	209
	210	if( $symbolCount == 0 ) {
	211	die "Error: $itemCount lines from $fileCount files processed, but $symbolCount symbols were found.\n";
	212	}
	213
	214	print " Loaded $symbolCount symbols from $itemCount lines in $fileCount files.\n";
	215
b75a7d8f A	216	print HEADER "\n/* C exports renaming data */\n\n";
b75a7d8f A	217	foreach(sort keys(%CFuncs)) {
729e4ab9 A	218	print HEADER "#define $_ U_ICU_ENTRY_POINT_RENAME($_)\n";
729e4ab9 A	219	# print HEADER "#define $_ $_$U_ICU_VERSION_SUFFIX\n";
b75a7d8f A	220	}
b75a7d8f A	221
729e4ab9	222	print HEADER "\n\n";
b75a7d8f A	223	print HEADER "/* C++ class names renaming defines */\n\n";
	224	print HEADER "#ifdef XP_CPLUSPLUS\n";
	225	print HEADER "#if !U_HAVE_NAMESPACE\n\n";
	226	foreach(sort keys(%CppClasses)) {
729e4ab9	227	print HEADER "#define $_ U_ICU_ENTRY_POINT_RENAME($_)\n";
b75a7d8f A	228	}
	229	print HEADER "\n#endif\n";
	230	print HEADER "#endif\n";
	231	print HEADER "\n#endif\n";
	232	print HEADER "\n#endif\n";
	233
	234	close HEADER;
	235
	236	sub verbose {
	237	if($VERBOSE) {
	238	print STDERR @_;
	239	}
	240	}
	241
	242
	243	sub printHelpMsgAndExit {
	244	print STDERR <<"EndHelpText";
	245	Usage: $0 [OPTIONS] LIBRARY_FILES
	246	Options:
	247	-v - verbose
	248	-h - help
	249	-o - output file name (defaults to 'urename.h'
	250	-S - suffix (defaults to _MAJOR_MINOR of current ICU version)
	251	Will produce a renaming .h file
	252
	253	EndHelpText
	254
	255	exit 0;
	256
	257	}
	258