started a script to find duplicates in mirror and database

1 files changed, 79 insertions, 0 deletions

diff --git a/bin/find-double-packages b/bin/find-double-packages
new file mode 100755
index 0000000..1732b5b
--- /dev/null
+++ b/bin/find-double-packages
@@ -0,0 +1,79 @@
+#!/bin/sh
+
+# shellcheck disable=SC2119,SC2120
+
+# shellcheck source=../lib/load-configuration
+. "${0%/*}/../lib/load-configuration"
+
+usage() {
+  >&2 echo ''
+  >&2 echo 'find-double-packages: find packages which are double on the mirror.'
+  >&2 echo ''
+  >&2 echo 'possible options:'
+  >&2 echo '  -h|--help:'
+  >&2 echo '    Show this help and exit.'
+  >&2 echo '  -m|--mirror-dir location of the mirror files'
+  [ -z "$1" ] && exit 1 || exit "$1"
+}
+
+eval set -- "$(
+  getopt -o hm: \
+    --long help \
+    --long mirror: \
+    -n "$(basename "$0")" -- "$@" || \
+  echo usage
+)"
+
+mirror=''
+no_action=false
+wait_for_lock='-n'
+
+while true
+do
+  case "$1" in
+    -h|--help)
+      usage 0
+    ;;
+    -m|--mirror)
+      if [ -n "${mirror}" ]; then
+        >&2 echo 'Already have one -m flag.'
+        usage
+      fi
+      shift
+      mirror="$1"
+    ;;
+    --)
+      shift
+      break
+    ;;
+    *)
+      >&2 echo 'Whoops, forgot to implement option "'"$1"'" internally.'
+      exit 42
+    ;;
+  esac
+  shift
+done
+
+if [ $# -ne 0 ]; then
+  >&2 echo 'Too many arguments.'
+  usage
+fi
+
+if [ -z "${mirror}" ]; then
+  >&2 echo 'Flag -m missing'
+  usage
+fi
+
+pooldir="${mirror}/pool"
+poolfiles="/tmp/files_in_pool"
+find "${pooldir}" -type f -name '*pkg.tar.xz' | \
+	rev | cut -f 1 -d / | rev | sort >"${poolfiles}"
+
+for arch in i486 i686 pentium4; do
+  for repo in core extra community; do
+    packagedir=${mirror}/${arch}/${repo}/
+    find "$packagedir" -type l -name '*pkg.tar.xz' >"/tmp/files_in_repo_${arch}_${repo}"
+  done
+done
+
+