michael@0: #!/usr/bin/perl 
michael@0: #*
michael@0: #*******************************************************************************
michael@0: #*   Copyright (C) 2001-2012, International Business Machines
michael@0: #*   Corporation and others.  All Rights Reserved.
michael@0: #*******************************************************************************
michael@0: #*
michael@0: #*   file name:  genren.pl
michael@0: #*   encoding:   US-ASCII
michael@0: #*   tab size:   8 (not used)
michael@0: #*   indentation:4
michael@0: #*
michael@0: #*   Created by: Vladimir Weinstein
michael@0: #*   07/19/2001
michael@0: #*
michael@0: #*  Used to generate renaming headers.
michael@0: #*  Run on UNIX platforms (linux) in order to catch all the exports
michael@0: 
michael@0: use POSIX qw(strftime);
michael@0: 
michael@0: $headername = 'urename.h';
michael@0: 
michael@0: $path = substr($0, 0, rindex($0, "/")+1)."../../common/unicode/uversion.h";
michael@0: 
michael@0: $nmopts = '-Cg -f s';
michael@0: $post = '';
michael@0: 
michael@0: $mode = 'POSIX';
michael@0: 
michael@0: (-e $path) || die "Cannot find uversion.h";
michael@0: 
michael@0: open(UVERSION, $path);
michael@0: 
michael@0: while(<UVERSION>) {
michael@0:     if(/\#define U_ICU_VERSION_SUFFIX/) {
michael@0:         chop;
michael@0:         s/\#define U_ICU_VERSION_SUFFIX //;
michael@0:         $U_ICU_VERSION_SUFFIX = "$_";
michael@0:         last;
michael@0:     }
michael@0: }
michael@0: 
michael@0: while($ARGV[0] =~ /^-/) { # detects whether there are any arguments
michael@0:     $_ = shift @ARGV;      # extracts the argument for processing
michael@0:     /^-v/ && ($VERBOSE++, next);                      # verbose
michael@0:     /^-h/ && (&printHelpMsgAndExit, next);               # help
michael@0:     /^-o/ && (($headername = shift (@ARGV)), next);   # output file
michael@0:     /^-n/ && (($nmopts = shift (@ARGV)), next);   # nm opts
michael@0:     /^-p/ && (($post = shift (@ARGV)), next);   # nm opts
michael@0:     /^-x/ && (($mode = shift (@ARGV)), next);   # nm opts
michael@0:     /^-S/ && (($U_ICU_VERSION_SUFFIX = shift(@ARGV)), next); # pick the suffix
michael@0:     warn("Invalid option $_\n");
michael@0:     &printHelpMsgAndExit;
michael@0: }
michael@0: 
michael@0: unless(@ARGV > 0) {
michael@0:     warn "No libraries, exiting...\n";
michael@0:     &printHelpMsgAndExit;
michael@0: }
michael@0: 
michael@0: #$headername = "uren".substr($ARGV[0], 6, index(".", $ARGV[0])-7).".h";
michael@0:     
michael@0: $HEADERDEF = uc($headername);  # this is building the constant for #define
michael@0: $HEADERDEF =~ s/\./_/;
michael@0: 
michael@0:     
michael@0:     open HEADER, ">$headername"; # opening a header file
michael@0: 
michael@0: #We will print our copyright here + warnings
michael@0: 
michael@0: $YEAR = strftime "%Y",localtime;
michael@0: 
michael@0: print HEADER <<"EndOfHeaderComment";
michael@0: /*
michael@0: *******************************************************************************
michael@0: *   Copyright (C) 2002-$YEAR, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *******************************************************************************
michael@0: *
michael@0: *   file name:  $headername
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   Created by: Perl script tools/genren.pl written by Vladimir Weinstein
michael@0: *
michael@0: *  Contains data for renaming ICU exports.
michael@0: *  Gets included by umachine.h
michael@0: *
michael@0: *  THIS FILE IS MACHINE-GENERATED, DON'T PLAY WITH IT IF YOU DON'T KNOW WHAT
michael@0: *  YOU ARE DOING, OTHERWISE VERY BAD THINGS WILL HAPPEN!
michael@0: */
michael@0: 
michael@0: #ifndef $HEADERDEF
michael@0: #define $HEADERDEF
michael@0: 
michael@0: /* U_DISABLE_RENAMING can be defined in the following ways:
michael@0:  *   - when running configure, e.g.
michael@0:  *        runConfigureICU Linux --disable-renaming
michael@0:  *   - by changing the default setting of U_DISABLE_RENAMING in uconfig.h
michael@0:  */
michael@0: 
michael@0: #include "unicode/uconfig.h"
michael@0: 
michael@0: #if !U_DISABLE_RENAMING
michael@0: 
michael@0: /* 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
michael@0:    the platform a chance to define it first.
michael@0:    Normally (if utypes.h or umachine.h was included first) this will not be necessary as it will already be defined.
michael@0:  */
michael@0: 
michael@0: #ifndef U_ICU_ENTRY_POINT_RENAME
michael@0: #include "unicode/umachine.h"
michael@0: #endif
michael@0: 
michael@0: /* If we still don't have U_ICU_ENTRY_POINT_RENAME use the default. */
michael@0: #ifndef U_ICU_ENTRY_POINT_RENAME
michael@0: #include "unicode/uvernum.h"
michael@0: #endif
michael@0: 
michael@0: /* Error out before the following defines cause very strange and unexpected code breakage */
michael@0: #ifndef U_ICU_ENTRY_POINT_RENAME
michael@0: #error U_ICU_ENTRY_POINT_RENAME is not defined - cannot continue. Consider defining U_DISABLE_RENAMING if renaming should not be used.
michael@0: #endif
michael@0: 
michael@0: EndOfHeaderComment
michael@0: 
michael@0: $fileCount = 0;
michael@0: $itemCount = 0;
michael@0: $symbolCount = 0;
michael@0: 
michael@0: for(;@ARGV; shift(@ARGV)) {
michael@0:     $fileCount++;
michael@0:     @NMRESULT = `nm $nmopts $ARGV[0] $post`;
michael@0:     if($?) {
michael@0:         warn "Couldn't do 'nm' for $ARGV[0], continuing...\n";
michael@0:         next; # Couldn't do nm for the file
michael@0:     }
michael@0:     if($mode =~ /POSIX/) {
michael@0:         splice @NMRESULT, 0, 6;
michael@0:     } elsif ($mode =~ /Mach-O/) {
michael@0: #        splice @NMRESULT, 0, 10;
michael@0:     }
michael@0:     foreach (@NMRESULT) { # Process every line of result and stuff it in $_
michael@0:         $itemCount++;
michael@0:         if($mode =~ /POSIX/) {
michael@0:             &verbose("  $_");
michael@0:             ($_, $address, $type) = split(/\|/);
michael@0:             chop $qtype;
michael@0:         } elsif ($mode =~ /Mach-O/) {
michael@0:             ($address, $type, $_) = split(/ /);
michael@0:             if(/^_(.*)$/) {
michael@0:                 $_ = $1;
michael@0:             } else {
michael@0:                 next;
michael@0:             }
michael@0:         } else {
michael@0:             die "Unknown mode $mode";
michael@0:         }
michael@0:         &verbose( "type: \"$type\" ");
michael@0:         if(!($type =~ /[UAwW?]/)) {
michael@0:             if(/@@/) { # These would be imports
michael@0:                 &verbose( "Import: $_ \"$type\"\n");
michael@0:                 &verbose( "C++ method: $_\n");
michael@0:             } elsif (/^[^\(]*::/) { # C++ methods, stuff class name in associative array
michael@0: 	        ##  DON'T match    ...  (   foo::bar ...   want :: to be to the left of paren
michael@0:                 ## icu::CharString::~CharString(void) -> CharString
michael@0:                 @CppName = split(/::/); ## remove scope stuff
michael@0:                 
michael@0:                 if(@CppName>1) {
michael@0:                     ## MessageFormat virtual table -> MessageFormat
michael@0:                     if(! ($CppName[0] =~ /icu/ )) {
michael@0:                         # *** WARNING Bad namespace (not 'icu') on ShoeSize::ShoeSize()
michael@0:                         warn "*** WARNING Bad namespace (not 'icu') on $_\n";
michael@0:                         next;
michael@0:                     }
michael@0:                     &verbose ( "(Chopping scope $CppName[0] )");
michael@0:                     @CppName = split(/ /, $CppName[1]); ## remove debug stuff
michael@0:                 }
michael@0:                 ## ures_getUnicodeStringByIndex(UResourceBundle -> ures_getUnicodeStringByIndex
michael@0:                 @CppName = split(/\(/, $CppName[0]); ## remove function args
michael@0:                 if($CppName[0] =~ /^operator/) {
michael@0:                     &verbose ("Skipping C++ function: $_\n");
michael@0:                 } elsif($CppName[0] =~ /^~/) {
michael@0:                     &verbose ("Skipping C++ destructor: $_\n");
michael@0:                 } else {
michael@0:                     &verbose( "Skipping C++ class: '$CppName[0]': $_ \n");
michael@0:                     # $CppClasses{$CppName[0]}++;
michael@0:                     # $symbolCount++;
michael@0:                 }
michael@0: 	    } elsif ( my ($cfn) = m/^([A-Za-z0-9_]*)\(.*/ ) {
michael@0: 		&verbose ( "$ARGV[0]:  got global C++ function  $cfn with '$_'\n" );
michael@0:                 $CFuncs{$cfn}++;
michael@0: 		$symbolCount++;
michael@0:             } elsif ( /\(/) { # These are strange functions
michael@0:                 print STDERR "$ARGV[0]: Not sure what to do with '$_'\n";
michael@0: 	    } elsif ( /^_init/ ) {
michael@0: 		&verbose( "$ARGV[0]: Skipped initializer $_\n" );
michael@0: 	    } elsif ( /^_fini/ ) {
michael@0: 		&verbose( "$ARGV[0]: Skipped finilizer $_\n" );
michael@0:             } elsif ( /icu_/) {
michael@0:                 print STDERR "$ARGV[0]: Skipped strange mangled function $_\n";
michael@0:             } elsif ( /^vtable for /) {
michael@0:                 print STDERR "$ARGV[0]: Skipped vtable $_\n";
michael@0:             } elsif ( /^typeinfo/) {
michael@0:                 print STDERR "$ARGV[0]: Skipped typeinfo $_\n";
michael@0:             } elsif ( /operator\+/ ) {
michael@0:                 print STDERR "$ARGV[0]: Skipped ignored function $_\n";
michael@0:             } else { # This is regular C function 
michael@0:                 &verbose( "C func: $_\n");
michael@0:                 @funcname = split(/[\(\s+]/);
michael@0:                 $CFuncs{$funcname[0]}++;
michael@0: 		$symbolCount++;
michael@0:             }
michael@0:         } else {
michael@0:             &verbose( "Skipped: $_ $1\n");
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: if( $fileCount == 0 ) {
michael@0:   die "Error: $itemCount lines from $fileCount files processed, but $symbolCount symbols were found.\n";
michael@0: }
michael@0: 
michael@0: if( $symbolCount == 0 ) {
michael@0:   die "Error: $itemCount lines from $fileCount files processed, but $symbolCount symbols were found.\n";
michael@0: }
michael@0: 
michael@0: print " Loaded $symbolCount symbols from $itemCount lines in $fileCount files.\n";
michael@0: 
michael@0: print HEADER "\n/* C exports renaming data */\n\n";
michael@0: foreach(sort keys(%CFuncs)) {
michael@0:     print HEADER "#define $_ U_ICU_ENTRY_POINT_RENAME($_)\n";
michael@0: #    print HEADER "#define $_ $_$U_ICU_VERSION_SUFFIX\n";
michael@0: }
michael@0: 
michael@0: print HEADER "\n#endif\n";
michael@0: print HEADER "\n#endif\n";
michael@0: 
michael@0: close HEADER;
michael@0: 
michael@0: sub verbose {
michael@0:     if($VERBOSE) {
michael@0:         print STDERR @_;
michael@0:     }
michael@0: }
michael@0: 
michael@0: 
michael@0: sub printHelpMsgAndExit {
michael@0:     print STDERR <<"EndHelpText";
michael@0: Usage: $0 [OPTIONS] LIBRARY_FILES
michael@0:   Options: 
michael@0:     -v - verbose
michael@0:     -h - help
michael@0:     -o - output file name (defaults to 'urename.h'
michael@0:     -S - suffix (defaults to _MAJOR_MINOR of current ICU version)
michael@0: Will produce a renaming .h file
michael@0: 
michael@0: EndHelpText
michael@0: 
michael@0:     exit 0;
michael@0: 
michael@0: }