#!/bin/bash

# This script requires fedora-release-rawhide and perl-XML-Rules

LOCAL_SOURCE_REPO=city-fan.org-rawhide-source
LOCAL_BINARY_REPO=city-fan.org-rawhide
BASE_SOURCE_REPO=rawhide-source
BASE_BINARY_REPO=rawhide
BINARY_ARCHLIST="$(uname -i),$(uname -m),noarch"
PERL_SPEC_URI='https://src.fedoraproject.org/rpms/perl/raw/rawhide/f/perl.spec'
DATA_DIR=$HOME/lib/repo-data
TIMESTAMP_FILE=$DATA_DIR/cfo-build-data-timestamp
EXPIRE_MINUTES=60

# Get primary.xml for a repo
get-primary-xml () {
	repo="$1"
	outfile="$2"
	# Get dnf to give us the baseurl for the repo
	baseurl=$(
		dnf --quiet \
		--disablerepo=\* \
		--enablerepo=$repo \
		repolist -v |
		awk '/^Repo-baseurl / { print $3 }' |
		sed -e 's|/$||' |
		# Horrible hack because curl cannot do rsync
		sed -e 's|^rsync:|http:|'
	)
	# Retrieve the repo's repomd.xml
	repomd=$(mktemp)
	curl --silent --output "$repomd" "$baseurl/repodata/repomd.xml"
	# Get the name of the repo's primary.xml file
	primaryfile=$(
		perl -se '
			use IO::File;
			use XML::Rules;
			my $repomd = new IO::File($xml_repomd);
			my @rules = (
				data => sub {
					print $_[1]->{location}->{href} if $_[1]->{type} eq "primary";
					return;
				},
			);
			my $parser = XML::Rules->new(rules => \@rules);
			$parser->parse($repomd);' -- -xml_repomd="$repomd"
	)
	rm -f "$repomd"
	case "$primaryfile" in
	*.gz)	decompress="gzip -dc";;
	*.xz)	decompress="xz -dc";;
	*.zst)	decompress="zstd -dc";;
	*)	decompress="cat";;
	esac
	timestamp=$(
		curl --silent --head "$baseurl/$primaryfile" |
		awk '/^Last-Modified:/ { print substr($0, 16) }'
	)
	curl --silent "$baseurl/$primaryfile" | $decompress > "$outfile"
	touch --date="$timestamp" "$outfile"
}

# Create data directory if it doesn't already exist
mkdir -p $DATA_DIR

# Check to see if we've updated the metadata recently
if [ -f "$TIMESTAMP_FILE" -a -z "$(find $TIMESTAMP_FILE -cmin +$EXPIRE_MINUTES)" ]; then
	# Data updated recently; no need to refresh
	exit 0
fi

# Generate local metadata caches
echo "================================================================="
echo "Generate local metadata caches"
echo "================================================================="
export http_proxy=http://yum.intra.city-fan.org:3128/
export ftp_proxy=http://yum.intra.city-fan.org:3128/
dnf	--quiet \
	--disablerepo=\* \
	--enablerepo=$LOCAL_SOURCE_REPO \
	--enablerepo=$LOCAL_BINARY_REPO \
	clean metadata
dnf	--quiet \
	--disablerepo=\* \
	--enablerepo=$LOCAL_SOURCE_REPO \
	--enablerepo=$LOCAL_BINARY_REPO \
	--enablerepo=$BASE_SOURCE_REPO \
	--enablerepo=$BASE_BINARY_REPO \
	makecache

# Get primary.xml for repos
get-primary-xml $LOCAL_SOURCE_REPO $DATA_DIR/local-source-primary.xml
get-primary-xml $LOCAL_BINARY_REPO $DATA_DIR/local-binary-primary.xml
get-primary-xml $BASE_BINARY_REPO $DATA_DIR/base-binary-primary.xml

# Get the list of packages in the default buildroot, to be treated as leaf packages
echo "================================================================="
echo "Get full list of packages in the default buildroot"
echo "================================================================="
BUILDROOT_TOP_LIST=$(
        dnf --quiet group info \
                --disablerepo=\* \
                --enablerepo=$BASE_BINARY_REPO \
                buildsys-build | grep '^   '
)
# Expand out to the full buildroot package list
python3 -c '
import sys
import dnf

db = dnf.Base()
db.conf.debuglevel = 0
db.read_all_repos()
base_binary_repo = "'$BASE_BINARY_REPO'"

for repo in db.repos.all():
    if repo.id == base_binary_repo:
        repo.enable()
    else:
        repo.disable()
db.fill_sack(load_system_repo=False)

for pkgname in sys.argv[1:]:
    db.install(pkgname)

db.resolve()

for pkg in sorted(db.transaction.install_set, key=lambda pkg: pkg.name.lower()):
    print("%s" % pkg.name)
' $BUILDROOT_TOP_LIST > $DATA_DIR/buildsys-build-package-list

# Get list of dual-lived perl modules, from base primary.xml
echo "================================================================="
echo "Get list of dual-lived perl modules"
echo "================================================================="
# Can no longer determine dual-lived perl packages by looking for packages
# built from perl srpm because the dual-lived ones aren't built post-bootstrap
curl --silent "$PERL_SPEC_URI" |
	grep '^%package ' |
	sed -e 's/^%package /perl-/' |
	sort > $DATA_DIR/perl-dual-lived-packages

# Touch the timestamp file so we don't repeat this process for a while
touch $TIMESTAMP_FILE