!7

cprint CompressorSpike {
  switchable S snippetID = #1020763;
  LinkedHashMap<S> versions;
  
  start-thread {
    print("Snippet ID: " + snippetID);
    dm_onFieldChange snippetID(r { setField(versions := null); dm_reload() });
    if (versions == null) {
      versions = mapToLinkedHashMap(textChangesOfSnippet(snippetID), sv -> pair(str(sv.versionID), sv.previousValue));
      versions.put("latest", loadSnippet(snippetID));
      versions = mapValuesToLinkedHashMap toLinesAndBack(versions); // canonicalize line breaks
      change();
    }
    print("Have " + nVersions(versions) + " with a total of " + nLines(totalLineCount(values(versions))));
    
    new Compressor().run();
  }
  
  abstract sclass Chunk {
    abstract S text(L<Chunk> chunks);
  }
  
  srecord CPair(int i1, int i2) > Chunk {
    S text(L<Chunk> chunks) {
      ret linesLL(chunks.get(i1).text(chunks), chunks.get(i2).text(chunks));
    }
  }
  
  srecord CPrim(S s) > Chunk {
    S text(L<Chunk> chunks) { ret s; }
  }
  
  class Compressor {
    Map<S, LS> textIDToLines = mapValuesToLinkedHashMap lines(versions);
    LS allUniqueLines;
    new L<Chunk> chunks;
    Map<S, Int> lineIndex;
    new Map<PairS, Int> linePairIndex;
    
    run {
      allUniqueLines = uniquify(concatLists(values(textIDToLines)));
      for (S line : allUniqueLines)
        chunks.add(new CPrim(line));
      lineIndex = listIndex(collect s(chunks));
      for (S id, LS lines : textIDToLines) {
        new L<Int> encoded;
        encoded = map(lines, line -> lineIndex.get(line));
        print(id + ": " + joinWithSpace(encoded));
        assertEquals(lines(lines), decode(encoded));
      }
    }
    
    S decode(L<Int> encoded) {
      ret lines(lambdaMap chunkText(encoded));
    }
    
    S chunkText(int idx) {
      ret chunks.get(idx).text(chunks);
    }
  }
}