Merged 'rt28345': new git-based merge_copyrights

util/copyrights

@@ -3236,6 +3236,7 @@
 ./util/check-pullups.pl				PERL	2001,2002,2003,2004,2007
 ./util/check-sources.pl				PERL	2000,2001,2004,2007
 ./util/copyrights				X	1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,2008,2009,2010,2011,2012
+./util/file_year.sh				SH	2012
 ./util/kit.sh					SH	2000,2001,2002,2003,2004,2007,2008,2009,2010,2012
 ./util/mandoc2docbook.pl			PERL	2001,2004,2007
 ./util/mdnbuildtest.sh				SH	2000,2001,2004,2007
@@ -3246,6 +3247,7 @@
 ./util/nanny.pl					PERL	2000,2001,2004,2007
 ./util/new-func					PERL	2005,2007
 ./util/nt-kit					SH	1999,2000,2001,2004,2007
+./util/recent_changes.sh			SH	2012
 ./util/spacewhack.pl				PERL	2000,2001,2004,2007
 ./util/tabify-changes				SH	2004,2007
 ./util/update-drafts.pl				PERL	2000,2001,2004,2007

util/file_year.sh

@@ -0,0 +1,7 @@
+#!/bin/sh
+# Given a file in the currently checked-out branch of the Git
+# repository, find out in what year it was most recently committed.
+# Used by merge_copyrights.
+
+rev=`git rev-list HEAD -- "$1" | head -n 1`
+git show --pretty=format:%ai $rev | head -n 1 | sed 's;-.*;;'

util/merge_copyrights

@@ -17,29 +17,42 @@
 
 # $Id$
 
-%file_types = ();
+use strict;
-%file_years = ();
+
+my %file_types = ();
+my %file_years = ();
+my %exists = ();
 
 open(COPYRIGHTS, "<util/copyrights") || die "can't open ./util/copyrights: $!";
 while (<COPYRIGHTS>) {
     chomp;
-    ($file, $type, $years) = split;
+    my ($file, $type, $years) = split;
     $file_types{$file} = $type;
     $file_years{$file} = $years;
 }
 close(COPYRIGHTS);
 
-($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = localtime(time());
+my ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = localtime(time());
+$sec = $min = $hour = $mday = $mon = $wday = $yday = $isdst = 0;
 $year += 1900;
 
-$find = "find . -type f -print";
+open(FILES, "git ls-files | sed 's;^;./;' |") || die "git ls-files: $!";
-
-open(FILES, "$find | sort |") || die "can't start \"$find\": $!";
 while (<FILES>) {
     chomp;
+    $exists{$_} = 1;
+}
+close(FILES);
+
+open(CHANGES, "sh util/recent_changes.sh |") || die "recent_changes.sh: $!";
+while (<CHANGES>) {
+    chomp;
+
+    # this file isn't in the repository now
+    next unless ($exists{$_});
 
     next if (m%/\.\# |		# CVS old conflict file
                /CVS/ |		# CVS directory
+               /.git/ |		# git directory
                util/newcopyrights | # our output
                \.bak$ |		# created by update_copyrights
                /(dnssafe|openssl)/.*\.[ch]$ |	# imported
@@ -49,6 +62,7 @@ while (<FILES>) {
     if (!$file_types{$_}) {
 	# Strip any .in extension to find out the file's real type.
 	# .in files are processed by configure to produce the target file.
+	my $base;
 	($base = $_) =~ s/\.in$//;
 
 	# Contributed code should maintain its own copyright.
@@ -95,24 +109,13 @@ while (<FILES>) {
 	} else {
 	    $file_types{$_} = "?";
 	}
-        ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,$atime,$mtime,$ctime,
+	my $m_year = int(`sh util/file_year.sh $_`);
-	 $blksize,$blocks)
+	if ($m_year != $year) {
-	    = stat($_);
-	($sec,$min,$hour,$mday,$mon,$c_year,$wday,$yday,$isdst) =
-	    localtime($ctime);
-	($sec,$min,$hour,$mday,$mon,$m_year,$wday,$yday,$isdst) =
-	    localtime($mtime);
-	$c_year += 1900;
-	$m_year += 1900;
-	if ($m_year != $year || $c_year != $year) {
 	    print "$_: must set copyright year(s) manually\n";
 	    $file_years{$_} = "????";
 	} else {
 	    $file_years{$_} = "$year";
 	}
-	# keep perl from issuing warnings about "used only once"
-	$dev = $ino = $mode = $nlink = $uid = $gid = $rdev = $size = 0;
-	$atime = $blksize = $blocks = 0;
     } else {
         if (! defined($file_years{$_}) || $file_years{$_} eq "????") {
             print "$_: must set copyright year(s) manually\n";
@@ -126,45 +129,42 @@ while (<FILES>) {
 	next if ($file_years{$_} =~ /^PARENT:/);
 	next if ($file_years{$_} eq "DOCBOOK");
 
-	@years = split(/,/, $file_years{$_});
+	my @years = split(/,/, $file_years{$_});
-	$has_current = 0;
+	my $has_current = 0;
-	foreach $fyear (@years) {
+	foreach my $fyear (@years) {
 	    if ($fyear == $year) {
 		$has_current = 1;
 	    }
 	}
 	if (!$has_current) {
-	    ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,$atime,$mtime,$ctime,
+	    $file_years{$_} .= ",$year";
-	     $blksize,$blocks)
-		= stat($_);
-	    ($sec,$min,$hour,$mday,$mon,$m_year,$wday,$yday,$isdst) =
-		localtime($mtime);
-	    $m_year += 1900;
-	    if ($m_year == $year) {
-		$file_years{$_} .= ",$year";
-	    }
 	}
     }
 }
-close(FILES);
+close(CHANGES);
 
 open(NEWCOPYRIGHTS, ">util/newcopyrights") ||
     die "can't open newcopyrights: $!";
-foreach $file (sort(keys(%file_types))) {
+foreach my $file (sort(keys(%file_types))) {
     print NEWCOPYRIGHTS "$file";
-    $len = length($file);
+    my $len = length($file);
+    my $tabs = 0;
     if ($len >= 48) {
 	$tabs = 1;
     } else {
-	$needed = int (48 - $len);
+	my $needed = int (48 - $len);
 	$tabs = int ($needed / 8);
 	if ($needed % 8 != 0) {
 	    $tabs++;
 	}
     }
-    for ($i = 0; $i < $tabs; $i++) {
+    for (my $i = 0; $i < $tabs; $i++) {
 	printf NEWCOPYRIGHTS "\t";
     }
     printf NEWCOPYRIGHTS "%s\t%s\n", $file_types{$file}, $file_years{$file};
+
+    if (($file_years{$file} eq "????") || ($file_types{$file} eq "?")) {
+        print "Unknown file type or year: $file\n";
+    }
 }
 close(NEWCOPYRIGHTS);

util/recent_changes.sh

@@ -0,0 +1,12 @@
+#!/bin/sh
+# Find the list of files that have been touched in the Git repository
+# during the current calendar year.  This is done by walking backwards
+# through the output of "git whatchanged" until a year other than the
+# current one is seen.  Used by merge_copyrights.
+
+thisyear=`date +%Y`
+git whatchanged --pretty="date %ai" --date=iso8601 | awk -vre="${thisyear}-" '
+    $1 == "date" && $2 !~ re { exit(0); }
+    $1 == "date" { next; }
+    NF == 0 { next; }
+    $(NF-1) ~ /[AM]/ { print "./" $NF }' | sort | uniq