code to convert cdec's output to json for derivation editor

3 files changed, 153 insertions, 0 deletions

diff --git a/derivation_to_json/deriv.rb b/derivation_to_json/deriv.rb
new file mode 100755
index 0000000..e6230c3
--- /dev/null
+++ b/derivation_to_json/deriv.rb
@@ -0,0 +1,151 @@
+#!/usr/bin/env ruby
+
+require 'zipf'
+
+def conv_cdec_show_deriv s
+  s.gsub! /\(/, "\n"
+  s.gsub! /[{}]/, ""
+  a = s.split "\n"
+  a.map! { |i|
+    i.strip.gsub /(\s*\))+$/, ""
+  }
+  a.reject! { |i| i.strip=="" }
+  return a
+end
+
+class RuleAndSpan
+  attr_accessor :span, :symbol, :source, :target, :subspans, :done, :id
+
+  def initialize s, id
+    spans, srcs, tgts = splitpipe s.strip
+    _,@symbol = spans.split
+    @span = _.split(",", 2).map { |i| i.gsub(/[<>]/, "").to_i}
+    @source = srcs.strip.split
+    j = 0
+    @source.map! { |i| if i.match(/\[X\]/) then "[#{j+=1}]"  else i end }
+    @target = tgts.strip.split
+    @subspans = []
+    @done = false
+    @id = id
+  end
+
+  def to_s
+    return "#{@id}\t<#{@span.first},#{@span[1]}> [X] ||| #{@source.join " "} ||| #{@target.join " "}"
+  end
+
+  def is_terminal_rule?
+    @source.each { |w|
+      return true if !w.match(/\[S|(\d+)\]/)
+    }
+
+    return false
+  end
+end
+
+def proc_deriv s
+a = conv_cdec_show_deriv s
+
+by_span = {}
+spans = []
+id = 0
+a.each { |line|
+  rs = RuleAndSpan.new line, id
+  id += 1
+  by_span[rs.span] = rs
+  if rs.is_terminal_rule?
+    spans << rs.span
+  end
+}
+
+spans.reverse.each_with_index { |s,k|
+  (spans.size-(k+2)).downto(0) { |l|
+    t = spans[l]
+    if s[0] >= t[0] and s[1] <= t[1]
+      by_span[t].subspans << s
+      break
+    end
+  }
+}
+
+# fix order
+spans.each { |s|
+  by_span[s].subspans.reverse!
+}
+
+def derive span, spans, by_span, o, groups, source
+  if groups.size==0 || groups.last.size>0
+    groups << []
+  end
+  a = nil
+  if source
+    a = span.source
+  else
+    a = span.target
+  end
+  a.each_with_index { |w,k|
+    nt = w.match /\[(\d+)\]/
+    if nt
+      idx = nt.captures.first.to_i-1
+      _ = derive by_span[span.subspans[idx]], spans, by_span, o, groups, source
+      (k+1).upto(a.size-1) { |i|
+        if !a[i].match(/\[(\d+)\]/) && groups.last.size>0
+          groups << []
+        end
+      }
+    else
+      groups.last << ["#{w}", span.id]
+      o << w
+    end
+  }
+  span.done = true
+end
+
+so = []
+source_groups = []
+spans.each { |span|
+  next if by_span[span].done
+  derive by_span[span], spans, by_span, so, source_groups, true
+}
+#puts "SOURCE"
+#puts so.join " "
+#puts source_groups.to_s
+#puts "##{source_groups.size}"
+
+spans.each { |s| by_span[s].done = false }
+
+o = []
+groups = []
+spans.each { |span|
+  next if by_span[span].done
+  derive by_span[span], spans, by_span, o, groups, false
+}
+#puts "TARGET"
+#puts o.join " "
+#puts groups.to_s
+#puts "##{groups.size}"
+
+source_rgroups = []
+rgroups = []
+source_groups.each { |i| source_rgroups << i.first[1] }
+groups.each { |i| rgroups << i.first[1] }
+
+phrase_align = []
+source_rgroups.each { |i|
+  phrase_align << []
+  rgroups.each_with_index { |j,k|
+    if i==j
+      phrase_align.last << k
+    end
+  }
+}
+
+h = {}
+h[:phrase_alignment] =  phrase_align
+h[:source_groups] = source_groups.map { |a| a.map { |i| i.first } }
+h[:target_groups] = groups.map { |a| a.map { |i| i.first } }
+
+return h.to_json
+end
+
+puts proc_deriv STDIN.gets.strip
+
diff --git a/derivation_to_json/example b/derivation_to_json/example
new file mode 100644
index 0000000..5128e09
--- /dev/null
+++ b/derivation_to_json/example
@@ -0,0 +1 @@
+({<0,41> [Goal] ||| [S] ||| [1]}({<0,41> [S] ||| [S] [X] ||| [1] [2]}({<0,37> [S] ||| [S] [X] ||| [1] [2]}({<0,33> [S] ||| [S] [X] ||| [1] [2]}({<0,32> [S] ||| [S] [X] ||| [1] [2]}({<0,30> [S] ||| [S] [X] ||| [1] [2]}({<0,15> [S] ||| [S] [X] ||| [1] [2]}({<0,10> [S] ||| [S] [X] ||| [1] [2]}({<0,9> [S] ||| [S] [X] ||| [1] [2]}({<0,8> [S] ||| [S] [X] ||| [1] [2]}({<0,7> [S] ||| [S] [X] ||| [1] [2]}({<0,5> [S] ||| [X] ||| [1]}({<0,5> [X] ||| hier also [X] , ||| so here [1] ,}({<2,4> [X] ||| ein bescheidener ||| a modest}) ) ) ({<5,7> [X] ||| auf alle ||| to all}) ) ({<7,8> [X] ||| demokratien ||| democracies}) ) ({<8,9> [X] ||| anzuwendender ||| anzuwendender}) ) ({<9,10> [X] ||| vorschlag ||| proposal}) ) ({<10,15> [X] ||| : [X] ideen ||| : [1] ideas}({<11,14> [X] ||| der markt für ||| the market for}) ) ) ({<15,30> [X] ||| funktioniert [X] , [X] der ||| works [1] [2] the}({<16,17> [X] ||| besser ||| better}) ({<18,29> [X] ||| wenn es den [X] ||| if [1]}({<21,29> [X] ||| [X] fällt , die [X] ||| [1] , the [2]}({<21,23> [X] ||| [X] leichter ||| [1] easier}({<21,22> [X] ||| bürgern ||| citizens}) ) ({<26,29> [X] ||| zielkonflikte zwischen [X] ||| trade @-@ offs between [1]}({<28,29> [X] ||| treffsicherheit ||| treffsicherheit}) ) ) ) ) ) ({<30,32> [X] ||| aussagen und ||| statements and}) ) ({<32,33> [X] ||| unterhaltung ||| entertainment}) ) ({<33,37> [X] ||| oder zwischen [X] und ||| or [1] and}({<35,36> [X] ||| treffsicherheit ||| treffsicherheit}) ) ) ({<37,41> [X] ||| [X] zu erkennen . ||| [1] .}({<37,38> [X] ||| parteitreue ||| parteitreue}) ) ) )
diff --git a/derivation_to_json/example.json b/derivation_to_json/example.json
new file mode 100644
index 0000000..b84edad
--- /dev/null
+++ b/derivation_to_json/example.json
@@ -0,0 +1 @@
+{"phrase_alignment":[[0,2],[1],[0,2],[3],[4],[5],[6],[7,9],[8],[7,9],[10,18],[11],[10,18],[12],[13],[14],[15],[16],[17],[10,18],[19],[20],[21,23],[22],[21,23],[24],[25]],"source_groups":[["hier","also"],["ein","bescheidener"],[","],["auf","alle"],["demokratien"],["anzuwendender"],["vorschlag"],[":"],["der","markt","für"],["ideen"],["funktioniert"],["besser"],[","],["wenn","es","den"],["bürgern"],["leichter"],["fällt",",","die"],["zielkonflikte","zwischen"],["treffsicherheit"],["der"],["aussagen","und"],["unterhaltung"],["oder","zwischen"],["treffsicherheit"],["und"],["parteitreue"],["zu","erkennen","."]],"target_groups":[["so","here"],["a","modest"],[","],["to","all"],["democracies"],["anzuwendender"],["proposal"],[":"],["the","market","for"],["ideas"],["works"],["better"],["if"],["citizens"],["easier"],[",","the"],["trade","@-@","offs","between"],["treffsicherheit"],["the"],["statements","and"],["entertainment"],["or"],["treffsicherheit"],["and"],["parteitreue"],["."]]}