From 225fbdc589079c7d0cc28d708c21ec01c9e9b48e Mon Sep 17 00:00:00 2001
From: Rich Salz <rsalz@akamai.com>
Date: Sat, 20 Jul 2024 16:39:32 -0400
Subject: [PATCH] feat: Basic copyright script and CI integration

This PR adds a script to do basic copyright notice checks, and will be used
to add the necessary features in issue #7690.  It reads the list of files
from argv[] and exits with a useful return status.

This PR also creates a copyright CI workflow
---
 .github/workflows/check-copyright.yml |  19 ++++
 ietf/check-copyright                  | 130 ++++++++++++++++++++++++++
 2 files changed, 149 insertions(+)
 create mode 100644 .github/workflows/check-copyright.yml
 create mode 100755 ietf/check-copyright

diff --git a/.github/workflows/check-copyright.yml b/.github/workflows/check-copyright.yml
new file mode 100644
index 0000000000..e47019ccc0
--- /dev/null
+++ b/.github/workflows/check-copyright.yml
@@ -0,0 +1,19 @@
+# Copyright The IETF Trust 2024. All Rights Reserved.
+name: Copyright years
+on: [pull_request, push]
+permissions:
+  contents: read
+
+jobs:
+  copyright-notice:
+    runs-on: ubuntu-latest
+    name: Test changed-files
+    steps:
+    - name: Get changed files
+      id: changed-files
+      uses: tj-actions/changed-files@v44
+    - uses: actions/checkout@v4
+    - name: Check changed files
+      env:
+        ALL_CHANGED_FILES: ${{ steps.changed-files.outputs.all_changed_files }}
+      run: perl ./ietf/check-copyright -lv -f ${ALL_CHANGED_FILES};
diff --git a/ietf/check-copyright b/ietf/check-copyright
new file mode 100755
index 0000000000..56ce0b6014
--- /dev/null
+++ b/ietf/check-copyright
@@ -0,0 +1,130 @@
+#!/usr/bin/env perl
+# Copyright The IETF Trust 2024. All Rights Reserved.
+
+use strict;
+use warnings;
+use POSIX qw/strftime/;
+use Getopt::Std;
+
+# Number of errors found; exit status
+my $ERRORS = 0;
+
+# Convenience variables, for the copyright year regexp patterns.
+my $this_year  = strftime("%Y", localtime);
+my $some_year  = "[12][0-9][0-9][0-9]";
+my $year_range = "(${some_year})(-${some_year})?";
+my $copyright  = "Copyright The IETF Trust *${year_range}";
+
+# Getopt settings.
+our($opt_h, $opt_f, $opt_v, $opt_c, $opt_l, $opt_m);
+
+sub
+usage()
+{
+    my $retcode = pop();
+
+    print STDERR "Options:\n";
+    print STDERR "    -h   This help message\n";
+    print STDERR "    -f   Read filenames from argv\n";
+    print STDERR "    -v   List files as processed\n";
+    print STDERR "    -c   Modify files that should be changed\n";
+    print STDERR "         (Does not do a git commit)\n";
+    print STDERR "    -l   List files that need to be changed\n";
+    print STDERR "    -m   List files missing copyright\n";
+    exit $retcode;
+}
+
+
+##  Get list of files changed during this year.
+sub
+collect_files
+{
+    # Get last commit of the of the previous year/
+    my $FIRST=`git rev-list -1 --before=$this_year-01-01 HEAD`;
+    chop $FIRST;
+
+    # Get every file changed since then, ignoring deleted files.
+    open(my $FH, "-|", "git diff-tree -r --name-status $FIRST..HEAD")
+        || die "Can't open diff-tree, $!";
+    my @FILES = ();
+    my @FIELDS;
+    while ( <$FH> ) {
+        @FIELDS = split();
+        next if $FIELDS[0] =~ /D/;        # ignore deleted files
+        push(@FILES, $FIELDS[1]);
+    }
+    close($FH) || die "Can't close diff-tree";
+    return @FILES;
+}
+
+##  Process file, notice if copyright is missing or outdated.
+sub
+process()
+{
+    my $NAME = pop();
+    my $NEW = "";
+    my $found = 0;
+    my $changed = 0;
+    my $SAVE;
+
+    print "# Processing $NAME\n" if defined $opt_v;
+
+    # ignore ZIP files
+    return if $NAME =~ /\.zip/;
+
+    # Read the file, copying to $NEW and changing copyright
+    # along the way.
+    open my $FH, '<', $NAME || die "Can't open $NAME, $!";
+    while ( <$FH> ) {
+        $SAVE = $_;
+        if ( /$copyright/io ) {
+            $found = 1;
+            $SAVE =~ s|${year_range}|$1-${this_year}|;
+            $SAVE =~ s|(${some_year})-$1|$1|;
+            $changed = 1 if $SAVE ne $_;
+        }
+        $NEW .= $SAVE;
+    }
+    close($FH) || die "Can't close $NAME, $!";
+
+    # Copyright missing?
+    if ( defined $opt_m ) {
+        print "$NAME\n" if ! $found;
+        $ERRORS++;
+        return;
+    }
+
+    # Unchanged file?
+    return if ! $changed;
+
+    if ( defined $opt_l ) {
+        print "$NAME\n";
+        $ERRORS++;
+    } else {
+        # Write the new file
+        open my $FH, '>', $NAME || die "Can't close-write $NAME, $!";
+        print $FH $NEW;
+        close($FH) || die "Can't close-write $NAME, $!";
+    }
+}
+
+##  Parse JCL.
+getopts('hfvclm') || &usage(1);
+&usage(0) if $opt_h;
+if ( defined($opt_m) + defined($opt_l) + defined($opt_c) != 1 ) {
+    print STDERR "Must have exactly one of m/l/c options\n";
+    exit 1;
+}
+
+##  Do the work.
+my @FILES = ();
+if ( defined($opt_f) ) {
+    @FILES = @ARGV;
+} else {
+    @FILES = &collect_files();
+}
+
+foreach my $F ( @FILES ) {
+    &process($F);
+}
+exit $ERRORS;