From d7d1870be318303e4f5679eb4e1a6823b9f5c930 Mon Sep 17 00:00:00 2001
From: Brian Harring <ferringb@gmail.com>
Date: Sat, 13 Oct 2012 18:09:53 -0700
Subject: add usable script

---
 .gitignore           |  1 +
 process_directory.sh | 34 ++++++++++++++++++++++++++++++++++
 script.sh            | 22 ++++++++++++++++++++++
 3 files changed, 57 insertions(+)
 create mode 100755 process_directory.sh
 create mode 100755 script.sh

diff --git a/.gitignore b/.gitignore
index 8b5efc7..3a10075 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,2 +1,3 @@
 cvs-repo
 output
+final
diff --git a/process_directory.sh b/process_directory.sh
new file mode 100755
index 0000000..54c51cf
--- /dev/null
+++ b/process_directory.sh
@@ -0,0 +1,34 @@
+#!/bin/bash
+f() {
+  set -x
+  mkdir -p "${output}"/{git,cvs-repo/gentoo-x86/Attic}
+  ln -s "${cvsroot}" "${output}/cvs-repo/CVSROOT"
+  ln -s "${root}/gentoo-x86/$1" "${output}/cvs-repo/gentoo-x86/$1"
+  #ln -s "${root}/gentoo-x86/Attic" "${output}/cvs-repo/gentoo-x86/Attic"
+  ln -s "$(pwd)/config" "${output}/config"
+  cd "${output}"
+  time cvs2git --options config -vv
+  cd git
+  git init --bare
+  cat ../cvs2svn-tmp/git-{blob,dump}.dat | git fast-import
+  rm -rf "${final}"
+  cd "$root"
+  mv "$output" "${final}"
+  git --git-dir "${final}/git" log --pretty=tformat:"%at %H" > "${final}/git-hashes"
+  set +x
+}
+
+[ $# -ne 1 ] && { echo "need an argument..."; exit 1; }
+
+root="$(pwd)/cvs-repo"
+cvsroot="${root}/CVSROOT"
+repo="${root}/gentoo-x86"
+output="$(pwd)/output/${1%,v}"
+final="$(pwd)/final/$1"
+mkdir -p "$(dirname "${final}")"
+
+rm -rf "${output}"
+mkdir -p "${output}"
+echo "processing ${1%,v} ${1}"
+time f "$1" &> "${output}/"log || { echo "failed $1"; exit 1; }
+echo "processed  $1"
diff --git a/script.sh b/script.sh
new file mode 100755
index 0000000..12c2032
--- /dev/null
+++ b/script.sh
@@ -0,0 +1,22 @@
+#!/bin/bash
+
+proc_count=$(grep -c MHz /proc/cpuinfo)
+[ $proc_count -eq 0 ] && proc_count=1
+
+rm -rf git
+mkdir git
+# Prioritize the larger categories first; they typically will have
+# the most revs, thus start them first.
+time { \
+  find cvs-repo/gentoo-x86 -maxdepth 1 -mindepth 1 -printf '%P\n' | \
+  xargs -n1 -I{} --  du -cs "cvs-repo/gentoo-x86/{}" | grep -v 'total$' | \
+  sort -gr | awk '{print $2;}' | xargs -n1 basename | \
+  xargs -n1 -P${proc_count} ./process_directory.sh | \
+  {
+    cd git;
+    git init &> /dev/null
+    while read l; do
+      git fetch "$(readlink -f "../final/$l/git")" && git merge FETCH_HEAD -m "blah" -q
+    done
+  }
+}
-- 
cgit v1.2.3-65-gdbad