Extract annotations from Marvin 2.0

marvin-annotations.rb

require 'nokogiri'

if ARGV.length != 1
  puts "Usage:\n$ ruby marvin-annotations.rb <path-to-library.mrvi>"
  exit
end

def extract_highlights(book)
  min_ts = 3000000000 # very large
  max_ts = 0

  highlights = book.xpath('highlights/highlight').map do |highlight|
    ts = highlight.xpath('@dateTime').to_s.to_f
    min_ts = [min_ts, ts].min
    max_ts = [max_ts, ts].max
    highlight.xpath('@text').to_s
  end

  { first_ts: min_ts, last_ts: max_ts, highlights: highlights }
end

def extract_book_annotations(book)
  title = book.xpath('@title').to_s
  author = book.xpath('@authorSort').to_s
  subjects = book.xpath('subjects/subject').map do |subject|
    subject.xpath('text()').to_s
  end.join(', ')
  first_ts, last_ts, highlights = extract_highlights(book)
    .values_at(:first_ts, :last_ts, :highlights)

  return '' if highlights.length.zero?

  text_highlights = highlights.join("\n\n")

  %Q{
#{title}
#{'-' * title.length}
_By: #{author}_

**About:** #{subjects}

**Timespan:** From #{Time.at(first_ts).strftime('%b %d, %Y')} to #{Time.at(last_ts).strftime('%b %d, %Y')}

### Annotations
#{text_highlights}
}
end

def extract_annotations(books)
  books.map { |book| extract_book_annotations book }
end

xml = File.open(ARGV[0]) { |f| Nokogiri::XML f }
annotations = extract_annotations xml.xpath "/marvin/book"

puts %Q{
Marvin Annotations
==================

#{annotations.select { |an| an.length > 0 }.join("\n\n")}
}