path: root/server.rb

#!/usr/bin/env ruby

require 'sinatra'
require 'sinatra/cross_origin'
require 'sinatra/reloader'
require 'nanomsg'
require 'zipf'
require 'json'
require 'haml'
require 'uri'
require_relative './derivation_to_json/derivation_to_json'
require_relative './phrase_alignment/phrase_alignment'

# #############################################################################
# Load configuration file and setup global variables
# #############################################################################
require_relative "#{ARGV[0]}" # load configuration for this session
$lock             = false     # lock if currently learning/translating
$last_reply       = nil       # cache last reply
$confirmed        = true      # client received translation?
$additional_rules = []        # corrected OOVs and newly extracted rules
$rejected_rules   = []        # known rules
if !FileTest.exist? LOCK_FILE # locked?
  $db  = {}                   # data file (JSON format)
  $env = {}                   # environment variables (socket connections to daemons)
end

# #############################################################################
# Daemons
# #############################################################################
DIR="/fast_scratch/simianer/lfpe"
$daemons = {
  :tokenizer        => "#{DIR}/lfpe/util/nanomsg_wrapper.rb -a tokenize   -S '__ADDR__' -e #{EXTERNAL} -l #{TARGET_LANG}",
  :tokenizer_src    => "#{DIR}/lfpe/util/nanomsg_wrapper.rb -a tokenize   -S '__ADDR__' -e #{EXTERNAL} -l #{SOURCE_LANG}",
  :detokenizer      => "#{DIR}/lfpe/util/nanomsg_wrapper.rb -a detokenize -S '__ADDR__' -e #{EXTERNAL} -l #{TARGET_LANG}",
  :detokenizer_src  => "#{DIR}/lfpe/util/nanomsg_wrapper.rb -a detokenize -S '__ADDR__' -e #{EXTERNAL} -l #{SOURCE_LANG}",
  :truecaser        => "#{DIR}/lfpe/util/nanomsg_wrapper.rb -a truecase   -S '__ADDR__' -e #{EXTERNAL} -t #{SESSION_DIR}/truecase.model",
  #:lowercaser   => "#{DIR}/lfpe/util/nanomsg_wrapper.rb -a lowercase  -S '__ADDR__' -e #{EXTERNAL}",
  :dtrain           => "#{CDEC}/training/dtrain/dtrain_net_interface -c #{SESSION_DIR}/dtrain.ini -d #{WORK_DIR}/dtrain.debug.json -o #{WORK_DIR}/weights -a '__ADDR__' -E -R",
  :extractor        => "python -m cdec.sa.extract -c #{SESSION_DIR}/sa.ini --online -u -S '__ADDR__'",
  :aligner_fwd      => "#{CDEC}/word-aligner/net_fa -f #{SESSION_DIR}/forward.params  -m #{FWD_MEAN_SRCLEN_MULT}  -T #{FWD_TENSION}  --sock_url '__ADDR__'",
  :aligner_back     => "#{CDEC}/word-aligner/net_fa -f #{SESSION_DIR}/backward.params -m #{BACK_MEAN_SRCLEN_MULT} -T #{BACK_TENSION} --sock_url '__ADDR__'",
  :atools           => "#{CDEC}/utils/atools_net -c grow-diag-final-and -S '__ADDR__'"
}

# #############################################################################
# Set-up Sinatra
# #############################################################################
set :bind,              SERVER_IP
set :port,              WEB_PORT
set :allow_origin,      :any
set :allow_methods,     [:get, :post, :options]
set :allow_credentials, true
set :max_age,           "1728000"
set :expose_headers,    ['Content-Type']
set :public_folder,     File.dirname(__FILE__) + '/static'

# #############################################################################
# Helper functions
# #############################################################################
def logmsg name, msg
  STDERR.write "[#{name}] #{msg}\n"
end

def start_daemon cmd, name, addr
  logmsg :server, "starting #{name} daemon"
  cmd.gsub! '__ADDR__', addr
  pid = fork do
    exec cmd
  end
  sock = NanoMsg::PairSocket.new
  sock.connect addr
  logmsg :server, "< got #{sock.recv} from #{name}"

  return sock, pid
end

def stop_all_daemons
  logmsg :server, "shutting down all daemons"
  $env.each { |name,p|
    p[:socket].send "shutdown"                   # every daemon shuts down
                                                 # after receiving this keyword
    logmsg :server, "< #{name} is #{p[:socket].recv}"
  }
end

def update_database reset=false
  if !reset
    $db['progress'] += 1
  else
    $db['progress'] = 0
  end
  j = JSON.generate $db
  f = WriteFile.new DB_FILE
  f.write j.to_s
  f.close
end

def init
                                                          # data from JSON file
  $db = JSON.parse ReadFile.read DB_FILE
                                                            # working directory
  `mkdir -p #{WORK_DIR}/`
  `mkdir #{WORK_DIR}/g`
                                             # setup environment, start daemons
  port = BEGIN_PORT_RANGE
  $daemons.each { |name,cmd|
    sock, pid = start_daemon cmd, name, "tcp://127.0.0.1:#{port}"
    $env[name] = { :socket => sock, :pid => pid }
    port += 1
  }
                                                                   #  lock file
  `touch #{LOCK_FILE}`
end

def send_recv daemon, msg                            # simple pair communcation
  socket = $env[daemon][:socket]                     # query -> answer
  logmsg daemon, "> sending message: '#{msg}'"
  socket.send msg
  logmsg daemon, "waiting ..."
  ans = socket.recv.force_encoding("UTF-8").strip
  logmsg daemon, "< received answer: '#{ans}'"

  return ans
end

def clean_str s                   # FIXME replace chars w/ something reasonable
  s.gsub! "[", " "
  s.gsub! "]", " "
  s.gsub! "|", " "

  return s
end

# #############################################################################
# Run init() [just once]
# #############################################################################
init if !FileTest.exist?(LOCK_FILE)

# #############################################################################
# Routes
# #############################################################################
get '/' do
  cross_origin

  return ""                                                            # return
end

post '/next' do      # (receive post-edit, update models), send next translation
  cross_origin                           # enable Cross-Origin Resource Sharing
  reply = request.body.read
  logmsg :server, "raw JSON client reply: #{reply}"
  data = JSON.parse(URI.decode(reply))
  logmsg :server, "parsed reply: #{data.to_s}"
                                                  # already processing request?
  return "locked" if $lock                                    # return (locked)
  $lock = true                                                           # lock
  key = data['key']              # TODO do something with it, e.g. simple auth?
  if data["OOV"]                                              # OOV corrections
    logmsg :server, "received OOV corrections"
    grammar = "#{WORK_DIR}/g/#{$db['progress']}.grammar"
    src, tgt = splitpipe(data["correct"]) # format:src1\tsrc2\tsrc..|||tgt1\t..
    tgt = clean_str tgt
    src = src.split("\t").map { |i| i.strip }
    tgt = tgt.split("\t").map { |i| i.strip }
    src.each_with_index { |s,i|
      next if s==''||tgt[i]==''
      as = ""
      tgt[i].split.each_index { |k| as += " 0-#{k}" }
      r = "[X] ||| #{s} ||| #{tgt[i]} ||| NewRule=1 OOVFix=1 ||| #{as}"
      $additional_rules << r
    }
    $confirmed = true
  end
# received post-edit -> update models
# 0. save raw post-edit
# 1. tokenize [for each phrase]
# 2. truecase [for each phrase]
# 2.5 extract new rules
# 3. save processed post-edit
# 4. update weights
# 5. update grammar extractor
# 5a. forward alignment
# 5b. backward alignment
# 5c. symmetrize alignment
# 5d. actual update
# 6. update database
  if data["EDIT"]
    logmsg :server, "received post-edit"
                                                        # 0. save raw post-edit
    source = data["source_value"]
    post_edit = ''
    if data["type"] == 'g'                                # graphical interface
      post_edit = data["target"].join(" ")
      e = []
      logmsg :server, "post-edit before processing: '#{post_edit}'"
      data["target"].each_with_index { |i,j|
                                                                # [1.] tokenize
        _ = clean_str send_recv(:tokenizer, URI.decode(i))
        prev = _[0]                                       # use received casing
                                                                 #[2.] truecase
        _ = send_recv :truecaser, _
        _[0] = prev if j>0
        e << _
      }
      logmsg :server, "post-edit after processing: '#{e.join " "}'"
      f = []
      data["source_raw"].each { |i| f << URI.decode(i) }
                                                      # 2.5 new rule extraction
      new_rules = PhrasePhraseExtraction.extract_rules f, e, data["align"], true
      grammar = "#{WORK_DIR}/g/#{$db['progress']}.grammar"
      sts = {}
      ReadFile.readlines_strip(grammar).each { |r|
        s = splitpipe(r.to_s)[1..2].map{|i|i.strip.lstrip}.join(" ||| ")
        sts[s] = true
      }

      f = WriteFile.new "#{WORK_DIR}/#{$db['progress']}.new_rules"
      new_rules = new_rules.map { |r| r.as_trule_string }
      logmsg :server, "# rules before filtering #{new_rules.size}"
      _ = new_rules.dup
      new_rules.reject! { |rs|
        s = splitpipe(rs)[1..2].map{|i|i.strip.lstrip}.join(" ||| ")
        sts.has_key? s
      }
      logmsg :server, "# rules after filtering #{new_rules.size}"
      (_-new_rules).each { |r|
        logmsg :server, "rejected rule [already known]: '#{r}'"
      }
      $additional_rules += new_rules
      $rejected_rules   += _-new_rules
      f.write new_rules.join "\n"
      f.close
    else                                                       # text interface
      post_edit = data["post_edit"]
    end
    post_edit.strip!
    post_edit.lstrip!
    post_edit = clean_str post_edit                      # FIXME escape [ and ]
                                                                      # fill db
    $db['feedback']           << reply
    $db['post_edits_raw']     << post_edit
    $db['svg']                << data['svg']
    $db['durations']          << data['duration'].to_f
    $db['post_edits_display'] << send_recv(:detokenizer, post_edit)
    # 1. tokenize
      logmsg :server, "tokenizing post-edit"
      post_edit = send_recv :tokenizer, post_edit
    # 2. truecase
      logmsg :server, "truecasing post-edit"
      post_edit = send_recv :truecaser, post_edit
    # 3. save processed post-edits
      logmsg :db, "saving processed post-edit"
      $db['post_edits'] << post_edit.strip
    nochange = false
    if data['nochange']
      logmsg :server, "no change -> no update!"
      nochange = true
    end
    if !NOLEARN && !NOMT && !nochange
      logmsg :server, "updating ..."
                               # 4. update weights
                               # nb: this uses unaltered grammar [no new rules]
      grammar = "#{WORK_DIR}/g/#{$db['progress']}.grammar"
      annotated_source = "<seg grammar=\"#{grammar}\"> #{source} </seg>"
      send_recv :dtrain, "#{annotated_source} ||| #{post_edit}"
                                                  # 5. update grammar extractor
                                                  # 5a. get forward alignment
      source_lc = source.downcase
      post_edit_lc = post_edit.downcase
      a_fwd = send_recv :aligner_fwd, "#{source_lc} ||| #{post_edit_lc}"
                                                   # 5b. get backward alignment
      a_back = send_recv :aligner_back, "#{source_lc} ||| #{post_edit_lc}"
                                                     # 5c. symmetrize alignment
      a = send_recv :atools, "#{a_fwd} ||| #{a_back}"
                                                          # 5d actual extractor
      send_recv :extractor, "default_context ||| #{source} ||| #{post_edit} ||| #{a}"
                                                           # 6. update database
      logmsg :db, "updating database"
      $db['updated'] << true
    else
      $db['updated'] << false
    end
    update_database
  end
  source     = $db['source_segments'][$db['progress']]
  source = source.strip.lstrip
  raw_source = $db['raw_source_segments'][$db['progress']]
  raw_source = raw_source.strip.lstrip
  if !source                                                    # input is done
    logmsg :server, "end of input, sending 'fin'"
    $lock = false
    return {'fin'=>true}.to_json                                       # return
  elsif !$confirmed \
    || ($confirmed && $last_reply && $last_reply!="" \
        && !data["EDIT"] && !$last_reply.to_json["oovs"])     # send last reply
    logmsg :server, "locked, re-sending last reply"
    logmsg :server, "last_reply: '#{$last_reply}'"
    $lock = false
    return $last_reply                                                 # return
  else
# translate next sentence
# 0. no mt?
# 1. generate grammar
# - known rules
# - additional rules
# 2. check for OOV
# 3. translate
# 4. detokenize
# 5. reply
                                                                    # 0. no mt?
    if NOMT
      $lock = false
      logmsg :server, "no mt"
      obj = Hash.new
      obj["progress"]   = $db["progress"]
      obj["source"]     = source
      obj["raw_source"] = raw_source
      return obj.to_json                                               # return
    end
                                     # 1. generate grammar for current sentence
    grammar = "#{WORK_DIR}/g/#{$db['progress']}.grammar"
    send_recv :extractor, "default_context ||| #{source} ||| #{grammar}"
                                                                # - known rules
    logmsg :server, "annotating known rules"
    match = {}
    $rejected_rules.each { |r|
      _,src,tgt,_,_ = splitpipe r
      match["#{src.strip.lstrip} ||| #{tgt.strip.lstrip}".hash] = true
    }
    all_rules = ReadFile.readlines_strip grammar
    all_rules.each_with_index { |r,j|
      nt,src,tgt,f,a = splitpipe(r).map { |i| i.strip.lstrip }
      if match["#{src} ||| #{tgt}".hash]
        ar = "#{nt} ||| #{src} ||| #{tgt} ||| #{f} KnownRule=1 ||| #{a}"
        logmsg :server, "replacing rule '#{r}' with '#{ar}'"
        all_rules[j] = ar
      end
    }
    WriteFile.new(grammar).write all_rules.join("\n")+"\n"
                                                           # - additional rules
    $additional_rules.each { |rule|
     logmsg :server, "adding rule '#{rule}' to grammar '#{grammar}'"
      s = splitpipe(rule)[1..2].map{|i|i.strip.lstrip}.join(" ||| ")
      `echo "#{rule}" >> #{grammar}`
    }
    # 2. check for OOVs
    src_r = ReadFile.readlines(grammar).map {
      |l| splitpipe(l)[1].strip.split
    }.flatten.uniq
    oovs = []
    source.split.each { |token|
      if !src_r.include? token
        oovs << token
        logmsg :server, "OOV token: '#{token}'"
      end
    }
    oovs.uniq!
    logmsg :server, "have OOVs: '#{oovs.to_s}'"
    if oovs.size > 0                                                     # OOVs
      obj = Hash.new
      obj["oovs"]      = oovs
      obj["progress"]  = $db['progress']
      raw_source_annot = "#{raw_source}"
      oovs.each { |o|
        raw_source_annot.gsub! "#{o}", "***#{o}###"
      }
      obj["raw_source"] = raw_source_annot
      $last_reply = obj.to_json
      logmsg :server, "OOV reply: '#{$last_reply}'"
      $lock = false
      $confirmed = false
      return $last_reply                                               # return
    end
    # 3. translation
    msg = "act:translate ||| <seg grammar=\"#{grammar}\"> #{source} </seg>"
    derivation_str = send_recv :dtrain, msg
    obj_str = proc_deriv derivation_str
    obj = JSON.parse obj_str
    obj["transl"] = obj["target_groups"].join " "
    # 4. detokenizer
    obj["transl_detok"] = send_recv(:detokenizer, obj["transl"]).strip
    obj["target_groups"].each_index { |j|
      prev = obj["target_groups"][j][0]
      obj["target_groups"][j] = send_recv(:detokenizer, obj["target_groups"][j]).strip
      obj["target_groups"][j][0]=prev if j > 0
    }
    obj["source"]     = source
    obj["progress"]   = $db['progress']
    obj["raw_source"] = raw_source
    w_idx = 0
    obj["source_groups_raw"] = []
    obj["source_groups"].each { |i|
      obj["source_groups_raw"] << String.new(i)
    }
    obj["source_groups_raw"][0][0] = source[0]
    obj["source_groups"][0][0] = obj["source_groups"][0][0].upcase
    obj["source_groups"].each_with_index { |i,j|
      prev = obj["source_groups"][j][0]
      obj["source_groups"][j] = send_recv(:detokenizer_src, obj["source_groups"][j]).strip
      obj["source_groups"][j][0]=prev if j > 0
    }
                                                                         # save
    $db["derivations"]      << derivation_str
    $db["derivations_proc"] << obj_str
    $db["mt_raw"]           << obj["transl"]
    $db["mt"]               << obj["transl_detok"]
                                                                     # 5. reply
    $last_reply = obj.to_json
    $lock = false
    $confirmed = false
    logmsg :server, "response: '#{$last_reply}'"
    return $last_reply                                                 # return
  end

  return "{}"                                                  # return [ERROR]
end

get '/debug' do                                                    # debug view
  data = {}
  data = JSON.parse ReadFile.read(DB_FILE).force_encoding("UTF-8")
  if data["durations"].size == 0
    data["durations"] << -1
  end

  fn = "#{WORK_DIR}/dtrain.debug.json"
  pairwise_ranking_data                     = {}
  pairwise_ranking_data["kbest"]            = []
  pairwise_ranking_data["weights_before"]   = {}
  pairwise_ranking_data["weights_after"]    = {}
  pairwise_ranking_data["best_match_score"] = 0
  if File.exist? fn
    pairwise_ranking_data = JSON.parse ReadFile.read(fn).force_encoding("UTF-8")
  end

  haml :debug, :locals => { :data => data,
                            :pairwise_ranking_data => pairwise_ranking_data, \
                            :progress => $db["progress"]-1,
                            :additional_rules => $additional_rules, \
                            :rejected_rules => $rejected_rules, \
                            :session_key => SESSION_KEY }
end

get '/confirm' do                        # client confirms received translation
  cross_origin
  $confirmed = true
  logmsg :server, "confirmed = #{$confirmed}"

  return "#{$confirmed}"
end

get '/set_learning_rate/:rate' do
  rate = params[:rate].to_f
  logmsg :server, "set learning rate: #{rate}"
  return "locked" if $lock
  send_recv :dtrain, "set_learning_rate #{rate}"

  return "set learning rate to: #{rate}"
end

get '/set_learning_rate/sparse/:rate' do
  rate = params[:rate]
  logmsg :server, "set sparse learning rate: #{rate}"
  return "locked" if $lock
  send_recv :dtrain, "set_sparse_learning_rate #{rate}"

  return "set sparse learning rate to: #{rate}"
end

get '/reset_progress' do                                # reset current session
  return "locked" if $lock
  $db = JSON.parse ReadFile.read DB_FILE
  $db['post_edits'].clear
  $db['post_edits_raw'].clear
  $db['post_edits_display'].clear
  $db['mt'].clear
  $db['mt_raw'].clear
  $db['updated'].clear
  $db['durations'].clear
  $db['derivations'].clear
  $db['derivations_proc'].clear
  $db['svg'].clear
  $db['feedback'].clear
  $db['progress'] = -1
  update_database true
  $confirmed = true
  $last_reply = nil

  return "progress reset: done"
end

get '/reset_weights' do                                         # reset weights
  logmsg :server, "reset weights"
  return "locked" if $lock
  send_recv :dtrain, "reset_weights"

  return "reset weights: done"
end

get '/reset_extractor' do                             # reset grammar extractor
  logmsg :server, "reset extractor"
  return "locked" if $lock
  send_recv :extractor, "default_context ||| drop"

  return "reset extractor: done"
end

get '/reset_new_rules' do                               # removed learned rules
  $additional_rules.clear
  $rejected_rules.clear

  return "reset new rules: done"
end

get '/shutdown' do                          # stop daemons and shut down server
  logmsg :server, "shutting down daemons"
  stop_all_daemons

  return "stopped all daemons, ready to shutdown"
end