find-duplicate-packages neu

1 files changed, 53 insertions, 0 deletions

diff --git a/find-duplicate-packages b/find-duplicate-packages
new file mode 100755
index 0000000..babbcac
--- /dev/null
+++ b/find-duplicate-packages
@@ -0,0 +1,53 @@
+#!/bin/sh
+
+set +e
+
+trap 'rm -rf --one-file-system "${tmp_dir:?}"' EXIT
+tmp_dir=$(mktemp -d)
+
+for repo in \
+  'archlinuxewe' \
+  'community' \
+  'community-staging' \
+  'community-testing' \
+  'core' \
+  'extra' \
+  'quarry' \
+  'staging' \
+  'testing'; do
+  >&2 printf '%s ...' "${repo}"
+  {
+    curl -Ss 'https://arch.eckner.net/merged/'"${repo}"'/os/x86_64/'"${repo}"'.db.tar.gz' | \
+      zcat 2>/dev/null || \
+    {
+      curl -Ss 'https://arch.eckner.net/merged/'"${repo}"'/os/x86_64/'"${repo}"'.db.tar.xz' | \
+        xzcat 2>/dev/null
+    }
+  } | \
+    tar -t --wildcards '*/desc' | \
+    sed '
+      s,-[^-]\+-[^-]\+/desc$,,
+    ' > \
+    "${tmp_dir}/${repo}.db"
+  >&2 printf ' OK\n'
+done
+
+find "${tmp_dir}" -type f \
+  -execdir grep -HF '' '{}' \; | \
+  sed '
+    s,^\./\([^/:]\+\)\.db:,\1 ,
+    y/_/-/
+  ' | \
+  sort -k2,2 | \
+  uniq -Df1 | \
+  sed -n '
+    /^archlinuxewe /{
+      :a
+      $!N
+      s/\( \S\+\)\n\(\S\+\)\1/ \2\1/
+      ta
+      P
+      D
+    }
+    d
+  '