#!/usr/bin/perl
#  IBM_PROLOG_BEGIN_TAG
#  This is an automatically generated prolog.
#
#  $Source: src/build/tools/addCopyright.pl $
#
#  IBM CONFIDENTIAL
#
#  COPYRIGHT International Business Machines Corp. 2011
#
#  p1
#
#  Object Code Only (OCO) source materials
#  Licensed Internal Code Source Materials
#  IBM HostBoot Licensed Internal Code
#
#  The source code for this program is not published or other-
#  wise divested of its trade secrets, irrespective of what has
#  been deposited with the U.S. Copyright Office.
#
#  Origin: 30
#
#  IBM_PROLOG_END

# Forked from:
# Author: Mark Jerde (mjerde@us.ibm.com)
# Date: Fri Mar 19 17:40:32 2010 UTC
#
# CVS Information:
# $Revision: 1.55 $
# $Date: 2011/01/10 19:23:05 $
# $RCSfile: addCopyright.pl,v $

use strict;
use warnings;

#################################### ABOUT ####################################
# addCopyright.pl will automatically insert appropriate copyright statements  #
# in source files following the IBM copyright guidelines (and templates)      #
# referenced below :                                                          #
#   FSP ClearCase Architecture                                                #
#   Version 1.9                                                               #
#   10/12/2010                                                                #
#   Editor: Alan Hlava                                                        #
#                                                                             #
#   Section 3.14.1 has templates for different files                          #
#                                                                             #
#   NOTE: FSP uses the phrase "OCO Source materials" in their copyright       #
#           block, which is classified as 'p1' .  We will use the same        #
#           classification here.                                              #
#   NOTE: to list all files in src EXCEPT the build dir, run:                 #
#   make clean      # remove autogenerated files                              #
#   find src -path 'src/build' -prune  -o ! -type d -print | tr '\n' ' '      #
#                                                                             #
#       addCopyright.pl does not support piping, but you can send these       #
#       to a file, add "addCopyright.pl update" to the beginning of the line, #
#       and run the file to update all                                        #
###############################################################################

# Project-specific settings.
#
## my $releaseYear = "2011";
my $releaseYear = `date +%Y`;
chomp( $releaseYear );
my $copyrightSymbol = "";
# my $copyrightSymbol = "(C)";  # Uncomment if unable to use  character.

my  $projectName            =   "HostBoot";
my  $DELIMITER_END          =   "IBM_PROLOG_END";
my  $DELIMITER_BEGIN        =   "IBM_PROLOG_BEGIN_TAG";
my  $DELIMITER_WARNING_TEXT =   "This is an automatically generated prolog.";
my  $PVALUE                 =   "p1";       ## my best guess

my  $PVALUE_TEXT            =
"Object Code Only (OCO) source materials
  Licensed Internal Code Source Materials
  IBM $projectName Licensed Internal Code

  The source code for this program is not published or other-
  wise divested of its trade secrets, irrespective of what has
  been deposited with the U.S. Copyright Office.";
my  $ORIGIN                 =   "30";


# End Project-specific settings

#######################################################################
# Main
#######################################################################

if (scalar(@ARGV) < 2)
{
    usage();
}

##
my $operation = shift;
if (not ($operation =~ m/update/i) and
    not ($operation =~ m/validate/i))
{
    print "Operation update or validate is required\n";
    usage();
}

##  check for debug mode
my  $debug  =   shift( @ARGV );
# print   "$debug\n";
if ( $debug =~ m/debug/i )
{
    print   STDERR  "debug mode\n";
    $debug  =   1;
}
else
{
    print   STDERR "normal mode\n";
    unshift( @ARGV , $debug );
    $debug  =   0;
}

# Read files and process.
while (defined($_ = shift))
{
    my $filetype = filetype($_);
    print "File $_: Type $filetype\n";



    ##
    ##  validate
    ##      Exit 1 for unknown filetype
    ##      Exit 2 if has old copyright block
    ##      Exit 3 if no copyright block
    ##
    if ( $operation =~  m/validate/i )
    {
        if  ("Unknown" eq $filetype)
        {
            print "ERROR: File $_ :Unknown Filetype: $_";
            exit 1;
        }
        if ( hasoldcopyrightblock( $_, $filetype )    )
        {
            print "ERROR File $_: has old copyright block, please fix\n";
            exit 2;
        }
        
        if ( ! hascopyrightblock( $_, $filetype )  )
        {
            print "ERROR File $_: Missing Copyright notice!!!\n";
            exit 3;
        }
        
        #   good file
        next;
    }

    ##
    ##  update
    ##  Continue with a warning for unknown filetype - 
    ##
    if ($operation =~ m/update/i)
    {
        if  ("Unknown" eq $filetype)
        {
            print "!!!!! WARNING: File $_ :Unknown Filetype: $_\n";
            next;
        }
        ##  do a check for the old  "$IBMCopyright:  $"  block
        if ( hasoldcopyrightblock ( $_, $filetype )  )
        {
            print "ERROR File $_: has old copyright block, please fix\n";
            exit 2;
        }

        if ( ! hascopyrightblock( $_, $filetype ) )
        {
            if ($debug) { print "Add empty copyright block\n";    }
            addEmptyCopyrightBlock( $_, $filetype );
            ##  check that the first part worked
            if ( ! hascopyrightblock( $_, $filetype ) )
            {
                print STDERR "Failed to create Copyright block, please fix\n";
                exit 2;
            }

            fillInEmptyCopyrightBlock( $_, $filetype );
        }
    }

}   #   endwhile



#########################################################################
##  Subroutines
#########################################################################

#######################################
##  usage:  print usage and quit
#######################################
sub usage
{
    print "Usage: addCopyright.pl { update | validate } [ debug ] file1 ...\n\n";
    exit 1;
}

#####################################
##  log unknown file type - unfinished
#####################################
sub logUnknownFiletype
{
    my $filename = shift;
    my $fileinfo = shift;

    # TODO
    print "Unknown filetype: $filename $fileinfo\n";
}

#####################################
##  Analyze file and return a text string of the file type
#####################################
sub filetype
{
    my $filename = shift;
    my $fileinfo = `file $filename | sed 's/^.*: //'`;
    chomp $fileinfo;

    # Sorted by anticipated frequency of occurrence.
    if ( ( $filename =~ m/\.[cht]$/i )
       ||( $filename =~ m/\.[cht]\+\+$/i )
       ||( $filename =~ m/\.[cht]pp$/i )
       ||( $fileinfo =~ m/c program text/i )
       ||( $fileinfo =~ m/c\+\+ program text/i )
       )
    {
        return "C";
    }
    if ( ( $filename =~ m/\.pl$/ )
       ||( $filename =~ m/\.perl$/ )
       ||( $filename =~ m/\.pm$/ )
       ||( $fileinfo =~ m/perl.*script text executable/i) )
    {
        return "Perl";
    }
    if ($filename =~ m/\.s$/i)
    {
        return "Assembly";
    }
    if (($filename =~ m/Makefile$/i) or
        ($filename =~ m/\.mk$/i))
    {
        return "Makefile";
    }
    if ( $filename =~ m/\.am$/i )
    {
        return "Automake";
    }
    if ( ($filename =~ m/configure\.ac$/i)
       ||($filename =~ m/Makefile\.in$/i) )
    {
        return "Autoconf";
    }
    if ( ( $filename =~ m/\.[kc]{0,1}sh$/i )
       ||( $filename =~ m/\.bash$/i )
       ||( $fileinfo =~ m/shell script/i )
       ||( $fileinfo =~ m/^a \/bin\/[af]sh( -x|) *script text( executable|)$/ )
       ||( $fileinfo eq "Bourne shell script text")
       ||( $fileinfo eq "Bourne shell script text executable")
       ||( $fileinfo eq "Bourne-Again shell script text")
       ||( $fileinfo eq "Bourne-Again shell script text executable") )
    {
        return "Shellscript";
    }
    if ( $filename =~ m/\.py$/ )
    {
        return "Python";
    }
    if ( $filename =~ m/\.tcl$/ )
    {
        return "Tcl";
    }
    if ( $filename =~ m/\.x$/ )
    {
        return "RPC";
    }
    if ( ($filename =~ m/^commitinfo$/)
       ||($filename =~ m/^checkoutlist$/)
       ||($filename =~ m/^loginfo$/) )
    {
        return "CVS";
    }
    if ( $filename =~ m/\.emx$/ )
    {
        # Used by Rational Software Architect.  modelling file.
        return "EmxFile";
    }
    if ( $filename =~ m/\.mof$/ )
    {
        return "MofFile";
    }
    if ( $filename =~ m/\.ld$/ )
    {
        return "LinkerScript";
    }
    if ( $filename =~ m/\.xml$/i )
    {
        return "xml"
    }
    if ( ( $filename =~ m/\.emx$/i )
       ||( $filename =~ m/\.odt$/i )
       ||( $filename =~ m/\.gitignore$/i )
       ||( $filename =~ m/\.conf$/i )
       ||( $filename =~ m/\.lidhdr$/i )
       ||( $filename =~ m/\.vpdinfo$/i )
       ||( $filename =~ m/\.pdf$/i )
       
       )
    {
        # Known, but we can't deal with it so call it unknown.
        return "Unknown";
    }
    if ( -f $filename )
    {
        my $type = `grep "\\\$Filetype:.*\\\$" $filename`;
        if ( $type =~ m/\$Filetype:([^\$]*)\$/ )
        {
            $type = $1;
        }
        $type =~ s/^\s*//;
        $type =~ s/\s*$//;
        my %knownTypes = qw/Assembly Assembly Automake Automake Autoconf Autoconf C C CVS CVS EmxFile EmxFile LinkerScript LinkerScript Makefile Makefile MofFile MofFile Perl Perl Python Python RPC RPC Shellscript Shellscript Tcl Tcl/;
        return $type if defined($knownTypes{$type});
    }
    { # Other random files containing non-printable characters.
        my $file = `cat $filename`;
        if ( $file =~ m/([^\x20-\x7E\s])/ )
        {
            return "Unknown";
        }
    }
    logUnknownFiletype($filename,$fileinfo);
    return "Unknown";
}

######################################
##  Figure out the first copyright year for the file
##      return as a string
######################################
sub copyrightFirstYearForFile
{
    my $file = shift;
    my $fileContent = shift;
    my $year = $releaseYear;
    my @logstrings;
    # TODO find multiple copyright statements and pick the earliest one
    # Search for earlier years from prior release.
    if ($fileContent =~ m/Copyright IBM Corp. (\d{4})/)
    {
        $year = $1 if ($1 < $year);
    }

    my  $cmd = "git log -- $file | grep Date: | tail --line=1";
    ## print "run $cmd\n";
    @logstrings = split( " ", `$cmd` );

    #   new files will not have a log, so the "git log" call above will
    #       return nothing.  In this case, just return the old calculated
    #       value.
    if ( (scalar(@logstrings) < 5)  or ($year lt $logstrings[5]) )
    {
        return $year;
    }
    else
    {
        return $logstrings[5];
    }
}

########################################
##  Figure out the last copyright year for file
##      Return as a string
########################################
sub copyrightLastYearForFile
{
    return $releaseYear;
}



#######################################
##  Check for old $IBMCopyrightBlock signature
##      return 1 if it's there, otherwise 0
#######################################
sub hasoldcopyrightblock
{
    my ( $filename, $filetype ) =   @_;
    my $file = `head -180 $filename`;       # Must appear in first 180 lines.

    if ( $file =~ m/\$IBMCopyrightBlock:.*\$/s )
    {
        return  1;
    }

    return  0;
}

#######################################
##  Check for new IBM_PROLOG_BEGIN_TAG signature
##      return 1 if it's there, otherwise 0
#######################################
sub hascopyrightblock
{
    my ( $filename, $filetype ) =   @_;
    my $file = `head -180 $filename`;       # Must appear in first 180 lines.
    
    #print $filename;
    #print $file;
    #print   $DELIMITER_BEGIN;
    #print   $DELIMITER_END;


    if ( $file =~ m/$DELIMITER_BEGIN.*$DELIMITER_END/s )
    {
        return  1;
    }

    return  0;
}
###################################
##  Add an empty copyright block to the file, for example (C/C++ files):
##
##  // IBM_PROLOG_BEGIN_TAG IBM_PROLOG_END_TAG
##
##  - The block will be filled-in in the next step.
##  - Makes up a temporary file called "<filename>.cvsCPYRT"
##################################
sub addEmptyCopyrightBlock
{
    my ( $filename, $filetype)  =   @_;
    my $line;
    open(INPUT, $_) or exit;
    system "touch $_.cvsCPYRT";
    open(OUTPUT, ">$_.cvsCPYRT") or exit;

    if ("Assembly" eq $filetype)
    {
        ## print OUTPUT ";// \$IBMCopyrightBlock: \$\n";
        print OUTPUT "#  $DELIMITER_BEGIN $DELIMITER_END\n";
    }
    elsif (("Autoconf" eq $filetype) or
        ("Automake" eq $filetype) or
        ("CVS" eq $filetype) or
        ("Makefile" eq $filetype) or
        ("Perl" eq $filetype) or
        ("Python" eq $filetype) or
        ("Shellscript" eq $filetype) or
        ("Tcl" eq $filetype))
    {
        $line = <INPUT>;
        # Keep the '#!' line at the top.
        if ($line =~ m/^#!/)
        {
            print OUTPUT $line;
        }
        ## print OUTPUT "# \$IBMCopyrightBlock: \$\n";
        print OUTPUT "# $DELIMITER_BEGIN $DELIMITER_END\n";
        unless ($line =~ m/^#!/)
        {
            print OUTPUT $line;
        }
    }
    elsif ( ("C" eq $filetype)
           )
    {
        ##print OUTPUT " * \$IBMCopyrightBlock: \$\n";
        print OUTPUT "// $DELIMITER_BEGIN $DELIMITER_END\n";
    }
    elsif ( ("RPC" eq $filetype) or
            ("LinkerScript" eq $filetype)
          )
    {
        # ld stubbornly refuses to use modern comment lines.
        print OUTPUT "/* $DELIMITER_BEGIN $DELIMITER_END\n */\n";
    }
    elsif ("MofFile" eq $filetype)
    {
        ##print OUTPUT "// \$IBMCopyrightBlock: \$\n";
        print OUTPUT "// $DELIMITER_BEGIN $DELIMITER_END\n";
    }
    elsif ("xml" eq $filetype )
    {
        print   OUTPUT "<!-- $DELIMITER_BEGIN $DELIMITER_END -->\n";
    }
    else
    {
        print   STDERR  "Can\'t handle filetype:  $filetype\n";
        exit    1;
    }

    # Copy rest of file
    while (defined($line = <INPUT>))
    {
        print OUTPUT $line;
    }
    close OUTPUT;
    close INPUT;
    system "cat $_.cvsCPYRT > $_";
    system "rm -f $_.cvsCPYRT";
    system "touch $_";

}

############################################
##  fill in the empty copyright block
## Copyright guidelines from:
##   FSP ClearCase Architecture
##   Version 1.9
##   10/12/2010
##   Editor: Alan Hlava
##
##   Section 3.14.1 has templates for different files
##
############################################
sub fillInEmptyCopyrightBlock
{
    my  ( $filename, $filetype )    =   @_;
    my $file = `cat $_`;
    my $firstYear = copyrightFirstYearForFile($_, $file);
    my $lastYear = copyrightLastYearForFile($_);
    my $copyrightYear = $firstYear;
    $copyrightYear .= " - $lastYear" unless ( "$firstYear" eq "$lastYear" );
    ##  define the final copyright block template here.
    my $IBMCopyrightBlock =
" $DELIMITER_BEGIN
  $DELIMITER_WARNING_TEXT

  \$Source: $filename \$

  IBM CONFIDENTIAL

  COPYRIGHT International Business Machines Corp. $copyrightYear

  $PVALUE

  $PVALUE_TEXT

  Origin: $ORIGIN

  $DELIMITER_END";


    if (($file =~ m/$DELIMITER_BEGIN*.$DELIMITER_END/s ) and
        ("Unknown" ne $filetype))
    {
        open(INPUT, $_ ) or exit;
        system "touch $_.cvsCPYRT";
        open(OUTPUT, ">$_.cvsCPYRT") or exit;

        my $newline;
        my $lines = "";
        while ( defined($newline = <INPUT>) ) { $lines .= $newline; }

        if ("Assembly" eq $filetype)
        {
            ## assembly
            $IBMCopyrightBlock =~ s/\n/\n# /sg;
        }
        if (("Autoconf" eq $filetype) or
            ("Automake" eq $filetype) or
            ("CVS" eq $filetype) or
            ("Makefile" eq $filetype) or
            ("Perl" eq $filetype) or
            ("Python" eq $filetype) or
            ("Shellscript" eq $filetype) or
            ("Tcl" eq $filetype))
        {
            $IBMCopyrightBlock =~ s/\n/\n#/sg;
        }
        if (("C" eq $filetype)
           )
        {
            # Use C++ style comments
            $IBMCopyrightBlock =~ s|\n|\n//|sg;
        }
        if (("RPC" eq $filetype) or
            ("LinkerScript" eq $filetype)
           )
        {
            $IBMCopyrightBlock =~ s|\n|\n *|sg;
        }
        if ("EmxFile" eq $filetype)
        {
            # Not yet formatted correctly for EmxFile needs, but should coexist.
            $IBMCopyrightBlock = "$DELIMITER_BEGIN IBM Confidential OCO Source Materials (C) Copyright IBM Corp. $copyrightYear The source code for this program is not published or otherwise divested of its trade secrets, irrespective of what has been deposited with the U.S. Copyright Office. $DELIMITER_END";
        }
        if ("MofFile" eq $filetype)
        {
            $IBMCopyrightBlock =~ s|\n|\n//|sg;
        }
        if ( "xml" eq $filetype)
        {
            $IBMCopyrightBlock =~ s|\n|\n   |sg;
        }

        # Replace existing block with the current content.
        # $lines =~ s/\$IBMCopyrightBlock:[^\$]*\$/$IBMCopyrightBlock/s;
        $lines =~ s/$DELIMITER_BEGIN[^\$]*$DELIMITER_END/$IBMCopyrightBlock/s;

        print OUTPUT $lines;
        close OUTPUT;
        close INPUT;

        system "cat $_.cvsCPYRT > $_";
        system "rm -f $_.cvsCPYRT";
        system "touch $_";
    }
    else
    {
        print "Cannot find empty copyright block\n";
        exit 1;
    }
}