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

#!/usr/bin/perl 
#*
#*******************************************************************************
#*   Copyright (C) 2001-2012, 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 tools/genren.pl 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

/* U_DISABLE_RENAMING can be defined in the following ways:
 *   - when running configure, e.g.
 *        runConfigureICU Linux --disable-renaming
 *   - by changing the default setting of U_DISABLE_RENAMING in uconfig.h
 */

#include "unicode/uconfig.h"

#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/) {
            &verbose("  $_");
            ($_, $address, $type) = split(/\|/);
            chop $qtype;
        } elsif ($mode =~ /Mach-O/) {
            ($address, $type, $_) = split(/ /);
            if(/^_(.*)$/) {
                $_ = $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::CharString::~CharString(void) -> CharString
                @CppName = split(/::/); ## remove scope stuff
                
                if(@CppName>1) {
                    ## MessageFormat virtual table -> MessageFormat
                    if(! ($CppName[0] =~ /icu/ )) {
                        # *** WARNING Bad namespace (not 'icu') on ShoeSize::ShoeSize()
                        warn "*** WARNING Bad namespace (not 'icu') on $_\n";
                        next;
                    }
                    &verbose ( "(Chopping scope $CppName[0] )");
                    @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: $_\n");
                } else {
                    &verbose( "Skipping C++ class: '$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 $_\n";
            } elsif ( /^typeinfo/) {
                print STDERR "$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#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	#*******************************************************************************
51004dcb	4	#* Copyright (C) 2001-2012, 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	*
51004dcb	85	* Created by: Perl script tools/genren.pl written by Vladimir Weinstein
b75a7d8f A	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
51004dcb A	97	/* U_DISABLE_RENAMING can be defined in the following ways:
	98	* - when running configure, e.g.
	99	* runConfigureICU Linux --disable-renaming
	100	* - by changing the default setting of U_DISABLE_RENAMING in uconfig.h
	101	*/
	102
	103	#include "unicode/uconfig.h"
b75a7d8f A	104
b75a7d8f A	105	#if !U_DISABLE_RENAMING
729e4ab9 A	106
	107	/* 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
	108	the platform a chance to define it first.
	109	Normally (if utypes.h or umachine.h was included first) this will not be necessary as it will already be defined.
	110	*/
4388f060	111
729e4ab9 A	112	#ifndef U_ICU_ENTRY_POINT_RENAME
	113	#include "unicode/umachine.h"
	114	#endif
	115
	116	/* If we still don't have U_ICU_ENTRY_POINT_RENAME use the default. */
	117	#ifndef U_ICU_ENTRY_POINT_RENAME
	118	#include "unicode/uvernum.h"
	119	#endif
	120
	121	/* Error out before the following defines cause very strange and unexpected code breakage */
	122	#ifndef U_ICU_ENTRY_POINT_RENAME
	123	#error U_ICU_ENTRY_POINT_RENAME is not defined - cannot continue. Consider defining U_DISABLE_RENAMING if renaming should not be used.
	124	#endif
	125
b75a7d8f A	126	EndOfHeaderComment
b75a7d8f A	127
729e4ab9 A	128	$fileCount = 0;
	129	$itemCount = 0;
	130	$symbolCount = 0;
	131
b75a7d8f	132	for(;@ARGV; shift(@ARGV)) {
729e4ab9	133	$fileCount++;
46f4442e	134	@NMRESULT = `nm $nmopts $ARGV[0] $post`;
b75a7d8f A	135	if($?) {
	136	warn "Couldn't do 'nm' for $ARGV[0], continuing...\n";
	137	next; # Couldn't do nm for the file
	138	}
46f4442e A	139	if($mode =~ /POSIX/) {
	140	splice @NMRESULT, 0, 6;
	141	} elsif ($mode =~ /Mach-O/) {
	142	# splice @NMRESULT, 0, 10;
	143	}
b75a7d8f	144	foreach (@NMRESULT) { # Process every line of result and stuff it in $_
729e4ab9	145	$itemCount++;
46f4442e	146	if($mode =~ /POSIX/) {
4388f060	147	&verbose(" $_");
46f4442e	148	($_, $address, $type) = split(/\\|/);
4388f060	149	chop $qtype;
46f4442e	150	} elsif ($mode =~ /Mach-O/) {
51004dcb A	151	($address, $type, $_) = split(/ /);
	152	if(/^_(.*)$/) {
	153	$_ = $1;
46f4442e A	154	} else {
	155	next;
	156	}
	157	} else {
	158	die "Unknown mode $mode";
	159	}
b75a7d8f A	160	&verbose( "type: \"$type\" ");
	161	if(!($type =~ /[UAwW?]/)) {
	162	if(/@@/) { # These would be imports
	163	&verbose( "Import: $_ \"$type\"\n");
b75a7d8f	164	&verbose( "C++ method: $_\n");
729e4ab9 A	165	} elsif (/^[^\(]*::/) { # C++ methods, stuff class name in associative array
729e4ab9 A	166	## DON'T match ... ( foo::bar ... want :: to be to the left of paren
4388f060	167	## icu::CharString::~CharString(void) -> CharString
b75a7d8f	168	@CppName = split(/::/); ## remove scope stuff
4388f060	169
b75a7d8f A	170	if(@CppName>1) {
b75a7d8f A	171	## MessageFormat virtual table -> MessageFormat
4388f060 A	172	if(! ($CppName[0] =~ /icu/ )) {
	173	# *** WARNING Bad namespace (not 'icu') on ShoeSize::ShoeSize()
	174	warn "*** WARNING Bad namespace (not 'icu') on $_\n";
	175	next;
	176	}
	177	&verbose ( "(Chopping scope $CppName[0] )");
b75a7d8f A	178	@CppName = split(/ /, $CppName[1]); ## remove debug stuff
	179	}
	180	## ures_getUnicodeStringByIndex(UResourceBundle -> ures_getUnicodeStringByIndex
	181	@CppName = split(/\(/, $CppName[0]); ## remove function args
46f4442e A	182	if($CppName[0] =~ /^operator/) {
	183	&verbose ("Skipping C++ function: $_\n");
	184	} elsif($CppName[0] =~ /^~/) {
	185	&verbose ("Skipping C++ destructor: $_\n");
	186	} else {
4388f060 A	187	&verbose( "Skipping C++ class: '$CppName[0]': $_ \n");
	188	# $CppClasses{$CppName[0]}++;
	189	# $symbolCount++;
46f4442e	190	}
729e4ab9 A	191	} elsif ( my ($cfn) = m/^([A-Za-z0-9_])\(./ ) {
	192	&verbose ( "$ARGV[0]: got global C++ function $cfn with '$_'\n" );
	193	$CFuncs{$cfn}++;
	194	$symbolCount++;
b75a7d8f	195	} elsif ( /\(/) { # These are strange functions
729e4ab9 A	196	print STDERR "$ARGV[0]: Not sure what to do with '$_'\n";
	197	} elsif ( /^_init/ ) {
	198	&verbose( "$ARGV[0]: Skipped initializer $_\n" );
	199	} elsif ( /^_fini/ ) {
	200	&verbose( "$ARGV[0]: Skipped finilizer $_\n" );
b75a7d8f	201	} elsif ( /icu_/) {
729e4ab9	202	print STDERR "$ARGV[0]: Skipped strange mangled function $_\n";
46f4442e	203	} elsif ( /^vtable for /) {
729e4ab9	204	print STDERR "$ARGV[0]: Skipped vtable $_\n";
4388f060	205	} elsif ( /^typeinfo/) {
729e4ab9	206	print STDERR "$ARGV[0]: Skipped typeinfo $_\n";
46f4442e	207	} elsif ( /operator\+/ ) {
729e4ab9	208	print STDERR "$ARGV[0]: Skipped ignored function $_\n";
b75a7d8f A	209	} else { # This is regular C function
	210	&verbose( "C func: $_\n");
	211	@funcname = split(/[\(\s+]/);
	212	$CFuncs{$funcname[0]}++;
729e4ab9	213	$symbolCount++;
b75a7d8f A	214	}
	215	} else {
	216	&verbose( "Skipped: $_ $1\n");
	217	}
	218	}
	219	}
	220
729e4ab9 A	221	if( $fileCount == 0 ) {
	222	die "Error: $itemCount lines from $fileCount files processed, but $symbolCount symbols were found.\n";
	223	}
	224
	225	if( $symbolCount == 0 ) {
	226	die "Error: $itemCount lines from $fileCount files processed, but $symbolCount symbols were found.\n";
	227	}
	228
	229	print " Loaded $symbolCount symbols from $itemCount lines in $fileCount files.\n";
	230
b75a7d8f A	231	print HEADER "\n/* C exports renaming data */\n\n";
b75a7d8f A	232	foreach(sort keys(%CFuncs)) {
729e4ab9 A	233	print HEADER "#define $_ U_ICU_ENTRY_POINT_RENAME($_)\n";
729e4ab9 A	234	# print HEADER "#define $_ $_$U_ICU_VERSION_SUFFIX\n";
b75a7d8f A	235	}
b75a7d8f A	236
b75a7d8f A	237	print HEADER "\n#endif\n";
	238	print HEADER "\n#endif\n";
	239
	240	close HEADER;
	241
	242	sub verbose {
	243	if($VERBOSE) {
	244	print STDERR @_;
	245	}
	246	}
	247
	248
	249	sub printHelpMsgAndExit {
	250	print STDERR <<"EndHelpText";
	251	Usage: $0 [OPTIONS] LIBRARY_FILES
	252	Options:
	253	-v - verbose
	254	-h - help
	255	-o - output file name (defaults to 'urename.h'
	256	-S - suffix (defaults to _MAJOR_MINOR of current ICU version)
	257	Will produce a renaming .h file
	258
	259	EndHelpText
	260
	261	exit 0;
	262
	263	}