require 'sqlite3'

$authors = {
"rekkanoryo@cpw.pidgin.im" => "John Bailey",
"rekkanoryo@pidgin.im" => "John Bailey",
"rekkanoryo@rekkanoryo.org" => "John Bailey",
"khc@pidgin.im" => "Ka-Hing CHeung",
"sadrul@pidgin.im" => "Sadrul Chowdhury",
"markdoliner@pidgin.im" => "Mark Doliner",
"seanegan@pidgin.im" => "Sean Egan",
"charkins@pidgin.im" => "Casey Harkins",
"grim@pidgin.im" => "Gary Kramlich",
"rlaager@pidgin.im" => "Richard Laager",
"siege@pidgin.im" => "Christopher O'Brien",
"bartosz@pidgin.im" => "Bartoz Oler",
"deryni@pidgin.im" => "Etan Reisner",
"marv@pidgin.im" => "Tim Ringerbach",
"lschiere@pidgin.im" => "Luke Schierer",
"evands@pidgin.im" => "Evan Schoenberg",
"evands" => "Evan Schoenberg",
"resiak@pidgin.im" => "Will Thompson",
"resiak@soc.pidgin.im" => "Will Thompson",
"nosnilmot@pidgin.im" => "Stu Tomlinson",
"stu@nosnilmot.com" => "Stu Tomlinson",
"datallah@pidgin.im" => "Daniel Atallah",
"wabz@pidgin.im" => "Richard Nelson",
"felipe.contreras@gmail.com" => "Felipe Contreras",
"nix@go-nix.ca" => "Gabriel Schulhof",
"elb@pidgin.im" => "Ethan Blanton",
"kstange@pidgin.im" => "Kevin Stange",
"nwalp@pidgin.im" => "Nathan Walp",
"hbons@pidgin.im" => "Hylke Bons",
"aluink@pidgin.im" => "Eric Polino",
"jeff2@soc.pidgin.im" => "Jeffrey Connelly",
"mshkutkov@soc.pidgin.im" => "Michael Shkutkov",
"wehlhard@soc.pidgin.im" => "William Ehlhardt",
"ecoffey@soc.pidgin.im" => "Eoin Coffey",
"typ0@pidgin.im" => "Carlos Silva",
"wtogami" => "Warren Togami",
"friedel@translate.org.za" => "Friedel Wolff",
"khaledhosny@eglug.org" => "Khaled Hosny",
"ihar.hrachyshka@gmail.com" => "Ihar Hrachyshka",
"josep.puigdemont@gmail.com" => "Josep Puigdemont",
"toniher@softcatala.org" => "Toni Hermoso",
"bjoern@cs.tu-berlin.de" => "Björn Voigt",
"jochenkemnade@web.de" => "Jochen Kemnade",
"panos@echidna-band.com" => "Bouklis Panos",
"luke@lukeross.name" => "Luke Ross",
"jfs@computer.org" => "Javier Fernández-Sanguino Peña",
"okul@linux.ee" => "Ivar Smolin",
"timo.jyrinki@iki.fi" => "Timo Jyrinki",
"zongo_fr@users.sf.net" => "Éric Boumaour",
"mariamarcp@gmail.com" => "Mar Castro",
"fjrial@cesga.es" => "Francisco Rial",
"scraimer@gmail.com" => "Shalom Craimer",
"kelemeng@gnome.hu" => "Kelemen Gábor",
"raireg@yahoo.com" => "Rai S. Regawa",
"satriano@na.infn.it" => "Claudio Satriano",
"takeshi.aihana@gmail.com" => "Takashi Aihana",
"anousak@gmail.com" => "Anousak Souphavah",
"laurynas.biveinis@gmail.com" => "Laurynas Biveinis",
"arangel@linux.net.mk" => "Arangel Angov",
"espenas@gmail.com" => "Espen Stefansen",
"aalam@users.sf.net" => "Amanpreet Singh Alam",
"mauricioc@gmail.com" => "Maurício Collares",
"dumol@gnome.ro" => "Mişu Moldovan",
"miles@filmsi.net" => "Martin Srebotnjak",
"gpopac@gmail.com" => "Miloš Popović",
"xake@telia.com" => "Peter Hjalmarsson",
"rkvsraman@gmail.com" => "RKVS Raman",
"fundawang@gmail.com" => "Funda Wang",
"abelindsay@gmail.com" => "Abel Cheung",
"acli@ada.dhs.org" => "Ambrose C. Li",
"tulliana@gmail.com" => "Serdar Soytetir",
"loptosko@gmail.com" => "loptosko",
"eric.boumaour_pidgin@m4x.org" => "Éric Boumaour",
"mitr@volny.cz" => "Miloslav Trmac",
"hallvard.glad@gmail.com" => "Hallvard Glad",
}

def shell_escape(text)
    [ "<", ">", "!", "?", "|", "(", ")" ].each{|char| text = text.gsub(char, '\\' + char)}
    return text
end

class Commit
  attr_reader :id, :subject
  attr_accessor :body, :date, :author, :parents, :childs
  attr_accessor :counted, :count, :main, :alt, :tail
  attr_accessor :analyzed, :merges, :fork_commits
  attr_accessor :in_master

  def initialize(id)
    @id = id
    @branches = []
    @parents = []
    @childs = []

    @count = 0
    @tail = nil
    @in_master = false
    @merges = []
    @fork_commits = []
  end

  def is_in(branch)
    @branches.member?(branch)
  end

  def git_author()
    real_name = $authors[@author]
    real_name = @author if not real_name
    return "#{real_name} <#{@author}>"
  end

  def git_message()
    tmp = [ @subject, "mtn id: #{@id}" ]
    tmp << @body if @body
    return tmp.join("\n\n");
  end

  def to_s()
    return "commit=#{@id},subject=#{@subject}"
  end

  def []=(name, value)
    case name
    when "branch"
      @branches << value
    when "changelog"
      lines = value.split("\n")
      @subject = lines[0]
      @subject = @subject[0..100] + "..." if @subject.length > 100
      @body = value if lines.length > 0
    when "date"
      @date = value
    when "author"
      @author = value
    end
  end
end

class Branch
  def initialize(name)
    @name = name
    @commits = {}

    @master = nil
    @heads = []
    @tails = []
    @authors = {}
    @lost = []
    @lost_levels = {}
    @rebaseable = []
  end

  def filter_commits(commits)
    commits.each do |id,c|
      next if not c.is_in(@name)
      @commits[id] = c
    end
  end

  def check_commits()
    @commits.each do |id,c|
      tmp_parents = []
      rows = $db.execute("select parent,child from revision_ancestry where child='#{id}'")
      rows.each do |e|
        parent, child = e
        c.parents << parent
        tmp_parents << @commits[parent] if @commits[parent]
      end
      @authors[c.author] = 0 if not @authors[c.author]
      @authors[c.author] += 1
=begin
      tmp = c.body.scan(/patch.* from.([^,\.]*)/i)
      if not tmp.empty?
        p "== found ==", tmp
      end
=end
      rows = $db.execute("select parent,child from revision_ancestry where parent='#{id}'")
      rows.each do |e|
        parent, child = e
        c.childs << child
      end
      if tmp_parents.empty?
        @tails << c
      end
      if c.childs.empty?
        @heads << c
      end
    end
  end

  def find_master()
    check_commits()

    def travel(c)
      return c if c.counted
      c.counted = true
      c.parents.each do |id|
        parent = @commits[id]
        if parent
          travel(parent)
          if parent.count > c.count
            c.alt = c.main
            c.count = parent.count
            c.main = parent
            if parent.tail
              c.tail = parent.tail
            else
              c.tail = parent
            end
          else
            c.alt = parent
          end
        end
      end
      c.count += 1
      return c
    end

    master_count = 0

    @heads.each do |h|
      travel(h)
      if h.count > master_count
        master_count = h.count
        @master = h
      end
    end
  end

  def analyze()
    find_master()

    def analyze_head(h)
      def travel(c)
        return [], [] if not c or c.analyzed == true
        c.analyzed = true
        merges, fork_commits = travel(c.main)
        merges << c if c.parents.length > 1
        fork_commits << c
        return merges, fork_commits
      end

      h.merges, h.fork_commits = travel(h.main)
      # puts "branch=%s,merges=%d" % [h.id, h.merges.length]
      # puts "branch=%s,commits=%d" % [h.id, h.fork_commits.length]

      h.merges.each do |m|
        analyze_head(m.alt) if m.alt
      end
    end

    @heads.each do |h|
      analyze_head(h)
    end

    def travel(c)
      c.in_master = true
      if c.alt
        @rebaseable << c.alt if c.alt.fork_commits.length == 0
      end
      travel(c.main) if c.main
    end

    travel(@master)

    @commits.each do |id,e|
      next if e.in_master
      @lost << e
      if @lost_levels[e.merges.length]
        ll = @lost_levels[e.merges.length]
      else
        ll = []
        @lost_levels[e.merges.length] = ll
      end
      ll << e
    end
  end

  def print_results
      def travel(c)
        puts c
        travel(c.main) if c.main
      end

      puts "== info =="
      puts "name=#{@name}"
      puts "commit_count=#{@commits.length}"
      puts "author_count=#{@authors.length}"
      puts "tail_count=#{@tails.length}"
      puts "lost_count=#{@lost.length}"
      puts "rebaseable_count=#{@rebaseable.length}"
      puts "== lost levels =="
      @lost_levels.each do |l,a|
        puts "level=#{l}:count=#{a.length}"
      end
      puts "== master =="
      travel(@master)
      puts "== rebaseable =="
      @rebaseable.each do |c|
        puts c
      end
      puts "== tails =="
      @tails.each do |c|
        puts c
      end
      puts "== not in master =="
      @lost_levels.each do |l,a|
        puts "=== level=#{l} ==="
        a.each do |c|
          puts c
        end
      end
  end

  def generate_git
    def export_commit(c)
      puts "try #{c.id}"
      commit_dir = "commit-#{c.id}"
      system("./get-commit.sh #{c.id}")
      base_dir = Dir.pwd
      Dir.chdir(commit_dir)
      system("git add .")
      if c.alt
        c = c.alt if c.alt.fork_commits.length == 0
      end
      puts "generating #{c.id}, #{c.git_author}, #{c.git_message}"
      File.open("/tmp/message", "w") { |f| f.write c.git_message }
      system("git commit --quiet --author '#{c.git_author}' -F /tmp/message")
      Dir.chdir(base_dir)
      system("rm -rf #{commit_dir}")
    end

    def travel(c)
      export_commit(c)
      c.childs.each do |e|
        child = @commits[e]
        travel(child) if child and child.in_master
      end
    end

    git_dir = "#{Dir.pwd}/branches/master/.git"
    ENV['GIT_DIR'] = git_dir
    #system("mkdir -p #{}")
    travel(@master.tail)
  end

  def print_all
      @commits.each do |id,c|
        puts "#{id}: forks=#{c.fork_commits.length}, in_master=#{c.in_master}, alt=#{c.alt}, forks=#{c.fork_commits}"
      end
  end

  def print_authors
    puts "== authors =="
    @authors.each do |id,c|
      puts "author=%s,commit_count=%d" % [id, c]
      # puts id
    end
  end
end

@all_commits = {}

$db = SQLite3::Database.new("mtn/pidgin.mtn")
rows = $db.execute("select id,name,value from revision_certs")
rows.each do |e|
  id, name, value = e
  if not @all_commits[id]
    commit = Commit.new(id)
    @all_commits[id] = commit
  else
    commit = @all_commits[id]
  end
  commit[name] = value
end

branch = Branch.new("im.pidgin.pidgin")
branch.filter_commits(@all_commits)
branch.analyze()
# branch.print_results()
# branch.print_authors()
branch.generate_git()