From 9edb1e040d7215baa1452bded61fb70731861b02 Mon Sep 17 00:00:00 2001
From: Wirawan Purwanto <wpurwant@turing.hpc.odu.edu>
Date: Tue, 29 Jun 2021 13:29:56 -0400
Subject: [PATCH] * Imported tools to dump information about files in a
 container.   The "dump-info" tool works only for Debian-based distro for now.

---
 containers/simg-dump-info-Debian.sh | 68 +++++++++++++++++++++++++++++
 containers/simg-list-all-files.sh   | 29 ++++++++++++
 2 files changed, 97 insertions(+)
 create mode 100755 containers/simg-dump-info-Debian.sh
 create mode 100755 containers/simg-list-all-files.sh

diff --git a/containers/simg-dump-info-Debian.sh b/containers/simg-dump-info-Debian.sh
new file mode 100755
index 0000000..767d62c
--- /dev/null
+++ b/containers/simg-dump-info-Debian.sh
@@ -0,0 +1,68 @@
+#!/bin/bash
+#
+# Tool to dump info from a singularity image
+# Run this as "root" from _inside_ the container image,
+# because it will interrogate the system using the image's
+# contained files.
+#
+# This is for Debian-based image
+#
+# Usage:
+#    simg-dump-info.sh  INFO_DIR  CONT_NAME  [IMAGE_ROOT_DIR]
+#
+# INFO_DIR is the location to dump the info files
+# CONT_NAME is the name of the container (to be used as the basenames)
+#
+# Info dumped so far:
+# - ${CONT_NAME}.files - actual list of files
+# - ${CONT_NAME}.files-dpkg - registration of files according to dpkg
+# - ${CONT_NAME}.lists-dpkg.tar - registration of files according to dpkg
+#   (individual list files)
+# - ${CONT_NAME}.dpkg-l - list of installed packages
+
+
+_simg_list_all_files () {
+    local ROOT_DIR="${1:-.}"
+
+    find "$ROOT_DIR" -mount \
+        | sort  \
+        | awk -v ROOT="$ROOT_DIR" '
+              BEGIN {
+                  ROOTX = ROOT
+                  gsub(/\/+$/, "", ROOTX)
+                  ROOTX = (ROOTX "/")
+                  len_ROOTX = length(ROOTX)
+              }
+              $0 == ROOT { next }
+              (substr($0, 1, len_ROOTX) == ROOTX) {
+                  print(substr($0, len_ROOTX))
+                  next
+              }
+          '
+}
+
+
+set -e
+
+MYSELF=$0
+MYDIR=$(dirname "$MYSELF")
+
+INFO_DIR=${1:?INFO_DIR required as arg 1}
+CONT_NAME=${2:?CONT_NAME required as arg 2}
+CONT_IMAGE_ROOT_DIR=${3:-/}
+
+
+# Get the view of the files *ACTUALLY* residing in the container
+_simg_list_all_files "${CONT_IMAGE_ROOT_DIR}" > "${INFO_DIR}/${CONT_NAME}.files"
+
+dpkg -l > "${INFO_DIR}/${CONT_NAME}.dpkg-l"
+
+# Get the view of files registered to dpkg
+# (In some containers the files do not exist there because they
+# were somehow deleted, e.g. doc files, man pages, ...)
+(
+    cd /var/lib/dpkg/info
+    tar cf "${INFO_DIR}/${CONT_NAME}.lists-dpkg.tar" *.list
+    cat *.list | sort | uniq > "${INFO_DIR}/${CONT_NAME}.dpkg-files"
+)
+
diff --git a/containers/simg-list-all-files.sh b/containers/simg-list-all-files.sh
new file mode 100755
index 0000000..fe40efd
--- /dev/null
+++ b/containers/simg-list-all-files.sh
@@ -0,0 +1,29 @@
+#!/bin/bash
+
+# Dumps all the filenames on the container.
+# By default, run this on the root directory of the container.
+# This tool does NOT assume that the root directory is "/"
+# to allow access to sandbox _outside_ the container itself.
+# All external mount points are ignored.
+
+_simg_list_all_files () {
+    local ROOT_DIR="${1:-.}"
+
+    find "$ROOT_DIR" -mount \
+        | sort  \
+        | awk -v ROOT="$ROOT_DIR" '
+              BEGIN {
+                  ROOTX = ROOT
+                  gsub(/\/+$/, "", ROOTX)
+                  ROOTX = (ROOTX "/")
+                  len_ROOTX = length(ROOTX)
+              }
+              $0 == ROOT { next }
+              (substr($0, 1, len_ROOTX) == ROOTX) {
+                  print(substr($0, len_ROOTX))
+                  next
+              }
+          '
+}
+
+_simg_list_all_files "$@"