From 99053ecd32630e85eab61714f16a7dd404a404c9 Mon Sep 17 00:00:00 2001
From: xeruf <27jf@pm.me>
Date: Tue, 31 Jan 2023 18:13:32 +0100
Subject: [PATCH] bin: extend scripts for duplicate detection

---
 .local/bin/scripts/dupcheck | 17 +++++++++++++++++
 .local/bin/scripts/duploc   | 12 ++++++++----
 2 files changed, 25 insertions(+), 4 deletions(-)
 create mode 100755 .local/bin/scripts/dupcheck

diff --git a/.local/bin/scripts/dupcheck b/.local/bin/scripts/dupcheck
new file mode 100755
index 0000000..64bd8fd
--- /dev/null
+++ b/.local/bin/scripts/dupcheck
@@ -0,0 +1,17 @@
+#!/bin/sh
+# Check the first file against all other ones given and print duplicates.
+# Checks first size then diff.
+# TODO diff initial bytes
+if ! test $# -lt 2
+then echo "Need at least 2 files to compare!" 2>/dev/null
+     exit 2
+fi
+target="$1"
+shift
+for existing
+do test "$(realpath "$target")" != "$(realpath "$existing")" -a -f "$existing" &&
+    test "$(stat --format %s "$target")" -eq "$(stat --format %s "$existing")" &&
+    diff -q "$target" "$existing" >/dev/null &&
+    echo "$existing"
+    #test "$(md5sum "$existing" | cut -d\  -f1)" = "$(md5sum "$filepath" | cut -d\  -f1)" &&
+done
diff --git a/.local/bin/scripts/duploc b/.local/bin/scripts/duploc
index e01c9f8..0014d3b 100755
--- a/.local/bin/scripts/duploc
+++ b/.local/bin/scripts/duploc
@@ -1,11 +1,15 @@
 #!/bin/sh
-# Delete files under current or given path which exist elsewhere as listed in the locate database
-# Matches first by name, then by checksum (currently inefficiently via md5)
+# Delete files under current or given path
+# which exist elsewhere as listed by locate.
+# Args: [threshold (MB)] <filepaths...>
+# OptDepends: synct (for checking against original filename)
 case $1 in ([0-9]|[0-9][0-9]) threshold=$1; shift;; esac
 find "$@" -size +${threshold:-50}M -type f -exec sh -c "IFS=$'\n'"'
   filepath="{}"
-  target="$(synct-unarchive "$filepath")"
+  target="$(synct-unarchive "$filepath" || echo "$filepath")"
   highlight "$filepath"
   for existing in $(locate -b "$target")
-  do test "$(realpath "$target")" != "$(realpath "$existing")" -a -f "$existing" && test "$(md5sum "$existing" | cut -d\  -f1)" = "$(md5sum "$filepath" | cut -d\  -f1)" && echo "Found duplicate at $existing" && rm -vi "$filepath" && break
+  do test -n "$(dupcheck "$target" "$existing") &&
+     echo "Duplicate found at $existing"
+     rm -vi "$filepath" && break
   done' \;