update-ruby-trunk-changes-notes.sh

update-ruby-trunk-changes-notes.sh

#!/bin/sh
# Usage:
#   update-ruby-trunk-changes-notes.sh YYYY-MM-DD

export LANG=en_US.UTF-8

REPOSDIR=/var/working/ruby-trunk-changes-notes
TMPDIR=/tmp/ruby-trunk-changes
URL_PREFIX='https://ruby-trunk-changes.hatenablog.com/archive/'

[ -d $TMPDIR ] || mkdir -p $TMPDIR

if [ -z "$1" ]
then
  YYYYMMDD=$(ruby -rdate -e 'puts (Date.today - 3).to_s')
else
  YYYYMMDD=$1
fi

YYYY=$(echo $YYYYMMDD|cut -d- -f 1)
MM=$(echo $YYYYMMDD|cut -d- -f 2)
DD=$(echo $YYYYMMDD|cut -d- -f 3)

cd $TMPDIR
for i in $(wget -O - "${URL_PREFIX}$YYYY/$MM/$DD"  | grep entry-title-link | grep -o 'href="[^"]*"' | cut -d '"' -f 2)
do
  w3m -T text/html $i|ruby -e '
    HEADER = /: ?r(\d+) \d{4}/
    ARGF.read.slice(/#{HEADER.source}.*?(?=^nagachika \d{4}-)/m).tap {|i| exit unless i }\
        .each_line.slice_before(HEADER).each do |entry|
      open(entry[0].slice(HEADER, 1), "w:utf-8") do |f|
        f.puts entry[1..-1].find_all{|i| ! i.chomp.empty?}.join
      end
    end
  '
done

cd $REPOSDIR
git pull
for i in $(ls $TMPDIR)
do
  echo "## $i"
  cat  $TMPDIR/$i
  echo
  git notes --ref=ruby-trunk-changes add -F $TMPDIR/$i \
            $(git log -1 --grep="^git-svn-id: [^ ]*@$i " --format=%H)
done
git gc
# git push k-tsj refs/notes/ruby-trunk-changes

rm -rf $TMPDIR