From 29feb9b0a2313da6fbe4350433c38dc2f521332c Mon Sep 17 00:00:00 2001
From: Can Wong <can.wong@emerson.com>
Date: Tue, 5 May 2026 09:12:32 -0500
Subject: [PATCH] build.extra: Update to handle SPDX folders

Update build.extra to de-deupe SPDX entries via
delete-duplicate-spdx script

Signed-off-by: Can Wong <can.wong@emerson.com>
---
 scripts/pipelines/build.extra-feeds.sh     |  18 ++++
 scripts/pipelines/delete-duplicate-spdx.sh | 119 +++++++++++++++++++++
 2 files changed, 137 insertions(+)
 create mode 100644 scripts/pipelines/delete-duplicate-spdx.sh

diff --git a/scripts/pipelines/build.extra-feeds.sh b/scripts/pipelines/build.extra-feeds.sh
index 1149c7b48..3d2cee229 100755
--- a/scripts/pipelines/build.extra-feeds.sh
+++ b/scripts/pipelines/build.extra-feeds.sh
@@ -2,7 +2,9 @@
 set -euo pipefail
 
 SCRIPT_ROOT=$(realpath $(dirname $BASH_SOURCE))
+SPDX_VERSION="2.2"
 DELETE_DUPLICATE_IPKS="bash ${SCRIPT_ROOT}/delete-duplicate-ipks.sh"
+DELETE_DUPLICATE_SPDX="bash ${SCRIPT_ROOT}/delete-duplicate-spdx.sh"
 
 ## ARGUMENT PARSING
 usage() {
@@ -23,6 +25,8 @@ remove any packages from the extras feed which is already in core/.
 # Positional Arguments
 CORE_FEED_PATH
   Filepath to the root of the NILRT core/ IPK feed.
+CORE_SPDX_PATH
+  Filepath to the root of the NILRT core/ SPDX feed.
 EOF
 	exit ${1:-2}
 }
@@ -30,6 +34,7 @@ EOF
 desirable_only=false
 skip_package_index=false
 core_feed_path=""
+core_spdx_path=""
 
 positionals=()
 while [ $# -ge 1 ]; do case "$1" in
@@ -61,6 +66,9 @@ esac; done
 if [ ${#positionals[@]} -gt 0 ]; then
 	core_feed_path="$(realpath ${positionals[0]})"
 fi
+if [ ${#positionals[@]} -gt 1 ]; then
+	core_spdx_path="$(realpath ${positionals[1]})"
+fi
 
 
 ## MAIN
@@ -86,6 +94,16 @@ if [ -n "${core_feed_path}" ]; then
 		"./tmp-glibc/deploy/ipk"
 fi
 
+# If the user provided a core/ spdx path, dedeupe against it.
+if [ -n "${core_spdx_path}" ]; then
+	[ -d "$core_spdx_path" ] || (echo "ERROR: core spdx path $core_spdx_path is not a directory." >&2; exit 1)
+
+	echo "Pruning all SPDX from the extras feed which are already in core."
+	$DELETE_DUPLICATE_SPDX \
+		"${core_spdx_path}/${SPDX_VERSION}" \
+		"./tmp-glibc/deploy/spdx/${SPDX_VERSION}"
+fi
+
 # Package index generation must happen after we have deduped IPKs.
 if [ "$skip_package_index" != true ]; then
 	echo "INFO: Generating extra/ feed indexes."
diff --git a/scripts/pipelines/delete-duplicate-spdx.sh b/scripts/pipelines/delete-duplicate-spdx.sh
new file mode 100644
index 000000000..126bd2d11
--- /dev/null
+++ b/scripts/pipelines/delete-duplicate-spdx.sh
@@ -0,0 +1,119 @@
+#!/usr/bin/env bas
+# This script compares the SPDX contents of two feeds trees and removes any SPDXs
+# in the suboordinate feed which are already present in the superordinate feed.
+set -euo pipefail
+
+
+function usage() {
+	local returncode=${1:-2}
+	local helptext=$(cat <<EOF
+$(basename $0) [-h|--help] [-n|--dry-run] [-v|--verbose] superordinate subordinate
+
+Compares the SPDX contents of two feed trees, optionally removing duplicate SPDXs
+from the subordinate feed.
+
+Opts:
+-h|--help     Print this help test and exit (stdout).
+-n|--dry-run  Do everything except actually deleting the common SPDXs.
+-v|--verbose  Enable verbose reporting of which files are going to be deleted
+			  (stdout).
+
+Positionals:
+(super|sub)oridinate   Path to the feed tree root for each feed in comparison.
+EOF
+)
+
+	if [ "$returncode" == 2 ]; then
+		echo "${helptext}" >&1
+	else
+		echo "${helptext}" >&2
+	fi
+	exit $returncode
+}
+
+[ $# -lt 1 ] && usage
+
+## OPTIONS PARSING ##
+opt_dry_run=false
+opt_verbose=false
+
+opt_pos=()
+while [ $# -gt 0 ]; do
+	case "$1" in
+		-h|--help)
+			usage 0
+			shift
+			;;
+		-n|--dry-run)
+			opt_dry_run=true
+			shift
+			;;
+		-v|--verbose)
+			opt_verbose=true
+			shift
+			;;
+		-*)
+			usage 2
+			;;
+		*)
+			opt_pos+=("$1")
+			shift;;
+	esac
+done
+if [ "${#opt_pos[@]}" -ne 2 ]; then
+	usage 2
+fi
+feed_super=${opt_pos[0]}
+feed_sub=${opt_pos[1]}
+
+echo "Comparing feeds:"
+echo "Superordinate: ${feed_super}"
+echo "Subordinate:   ${feed_sub}"
+$opt_dry_run && echo "!! DRY RUN !! No files will be removed."
+
+if [ "${MACHINE}" = "x64" ]; then
+	tune="core2-64"
+else
+	tune="cortexa9-vfpv3"
+fi
+
+folder_list="all ${MACHINE} ${tune} x86_64"
+
+for folder in ${folder_list}; do
+	## SANITY CHECKS ##
+	# Since this script involves deleting many files, do a couple sanity checks.
+	# CHECK 1 - Both feed paths are actually directories
+	test -d "${feed_super}/${folder}" || (echo "ERROR: ${feed_super}/${folder} is not a real directory path." && exit 1)
+	test -d "${feed_sub}/${folder}" || (echo "ERROR: ${feed_sub}/${folder} is not a real directory path." && exit 1)
+	# CHECK 2 - The feeds are not the same. (If they are, then we would delete every SPDX.)
+	if [ "$(realpath ${feed_super}/${folder})" = "$(realpath ${feed_sub}/${folder})" ]; then
+		echo "ERROR: Supplied feed paths are actually the same directory." && exit 1
+	fi
+done
+
+
+count=0
+for folder in ${folder_list}; do
+	# Collect all the SPDX paths from each feed
+	find_exec="-printf %P\n"
+
+	spdx_super=$(find ${feed_super}/${folder} -name '*.spdx.json' ${find_exec} | sort)
+	spdx_sub=$(find ${feed_sub}/${folder} -name '*.spdx.json' ${find_exec} | sort)
+
+
+	# find entries common to both feeds
+	common=$(comm -12 <(cat <<<$spdx_super) <(cat <<<$spdx_sub))
+
+	# process each entry
+	for entry in $common; do
+		sub_entry_path="${feed_sub}/${folder}/${entry}"
+		if $opt_dry_run; then
+			$opt_verbose && echo "${sub_entry_path}"
+		else
+			rm $(if $opt_verbose; then echo --verbose; fi) ${sub_entry_path}
+		fi
+		count=$(($count + 1))
+	done
+done
+
+echo "Processed $count common entries."