Apply DA mapping

Signed-off-by: Łukasz Gryglicki <[email protected]>

Author: lukaszgryglicki

company_developers1.txt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:465b09ab7b2e304845d92916bf32ead1e34568b4d9afad5960b3ba60fce69d67
-size 66238518
+oid sha256:32a56c04026b4bae30ab299c26bdbb5cb8e252a1c0fff26f65dd63a0efefff94
+size 66413887

company_developers2.txt

@@ -0,0 +1,58 @@
+def merge(infile1, infile2, outfile)
+  # dbg: set to true to have very verbose output
+  dbg = !ENV['DBG'].nil?
+
+  cmap = {}
+  File.readlines(infile1).each do |line|
+    next if line[0] == '#'
+    index = line.index(' -> ')
+    unless index
+      puts "Broken line: #{line}"
+      binding.pry
+      exit 1
+    end
+    from = line[0..index - 1].strip
+    to = line[index + 4..-1].strip
+    next if from == to
+    if cmap.key?(from) && cmap[from] != to
+      puts "Broken map: already present cmap[#{from}] = #{cmap[from]}, new value different: #{to}"
+      binding.pry
+      exit 1
+    end
+    cmap[from] = to
+  end
+  puts "loaded #{cmap.length} items"
+  cmap.each do |from, to|
+    while cmap.key?(to)
+      newto = cmap[to]
+      cmap[from] = newto
+      puts "key #{from} -> #{to} -> #{newto}: added: #{from} -> #{newto}"
+      to = newto
+    end
+  end
+  File.readlines(infile2).each do |line|
+    next if line[0] == '#'
+    index = line.index(' -> ')
+    unless index
+      puts "Broken line: #{line}"
+      binding.pry
+      exit 1
+    end
+    from = line[0..index - 1].strip
+    to = line[index + 4..-1].strip
+    next if from == to
+    if cmap.key?(from) && cmap[from] != to
+      puts "Broken map: already present cmap[#{from}] = #{cmap[from]}, new value different: #{to}"
+      binding.pry
+      exit 1
+    end
+    cmap[from] = to
+  end
+end
+
+if ARGV.length < 3
+  puts "Arguments required: infile1 infile2 outfile"
+  exit 1
+end
+
+merge(ARGV[0], ARGV[1], ARGV[2])