gistfile1.txt

0                       {36271 0 PlusOne}
10                      {25474 39 FieldPathEncodeFinish}
1110                    {10334 1 PlusTwo}
1111                    {10530 11 PushOneLeftDeltaNRightNonZeroPack6Bits}
11000                   {2942 8 PushOneLeftDeltaOneRightNonZero}
11010                   {4128 4 PlusN}
110010                  {1375 2 PlusThree}
110011                  {1837 29 PopAllButOnePlusOne}
11011001                {471 10 PushOneLeftDeltaNRightNonZero}
11011010                {521 7 PushOneLeftDeltaOneRightZero}
11011100                {560 9 PushOneLeftDeltaNRightZero}
11011110                {634 32 PopAllButOnePlusNPack6Bits}
11011111                {646 3 PlusFour}
110110000               {149 30 PopAllButOnePlusN}
110110110               {251 12 PushOneLeftDeltaNRightNonZeroPack8Bits}
110110111               {271 37 NonTopoPenultimatePlusOne}
110111010               {300 31 PopAllButOnePlusNPack3Bits}
110111011               {310 26 PushNAndNonTopological}
1101100010              {99 38 NonTopoComplexPack4Bits}
11011000111             {76 36 NonTopoComplex}
110110001101            {35 5 PushOneLeftDeltaZeroRightZero}
110110001100001         {2 27 PopOnePlusOne}
110110001100100         {3 6 PushOneLeftDeltaZeroRightNonZero}
1101100011000000        {1 23 PushThreeLeftDeltaN}
1101100011000001        {1 24 PushThreePack5LeftDeltaN}
1101100011000100        {1 25 PushN}
1101100011000101        {1 28 PopOnePlusN}
1101100011000110        {1 33 PopNPlusOne}
1101100011000111        {1 34 PopNPlusN}
1101100011001010        {1 35 PopNAndNonTopographical}
11011000110010110       {1 13 PushTwoLeftDeltaZero}
11011000110010111       {1 14 PushTwoPack5LeftDeltaZero}
11011000110011000       {1 15 PushThreeLeftDeltaZero}
11011000110011001       {1 16 PushThreePack5LeftDeltaZero}
11011000110011010       {1 17 PushTwoLeftDeltaOne}
11011000110011011       {1 18 PushTwoPack5LeftDeltaOne}
11011000110011100       {1 19 PushThreeLeftDeltaOne}
11011000110011101       {1 20 PushThreePack5LeftDeltaOne}
11011000110011110       {1 21 PushTwoLeftDeltaN}
11011000110011111       {1 22 PushTwoPack5LeftDeltaN}

the long runs of 0's seem deeply, deeply suspicious to me, but I guess in theory they could be correct.

giving every weight 0 item a weight of 1 (as manta does) yields an entirely different and much more sensible set of codes.

The weights seem to be correct, I checked 3-4 of them randomly. Feel free to use this static lookup function:

/** Static fieldpath lookup */
force_inline fieldop* fieldop_lookup( uint32_t id ) {
    switch ( id ) {
    case 0: // PlusOne
        return &fieldpath_operations[0];
    case 2: // FieldPathEncodeFinish
        return &fieldpath_operations[39];
    case 14: // PlusTwo
        return &fieldpath_operations[1];
    case 15: // PushOneLeftDeltaNRightNonZeroPack6Bits
        return &fieldpath_operations[11];
    case 24: // PushOneLeftDeltaOneRightNonZero
        return &fieldpath_operations[8];
    case 26: // PlusN
        return &fieldpath_operations[4];
    case 50: // PlusThree
        return &fieldpath_operations[2];
    case 51: // PopAllButOnePlusOne
        return &fieldpath_operations[29];
    case 217: // PushOneLeftDeltaNRightNonZero
        return &fieldpath_operations[10];
    case 218: // PushOneLeftDeltaOneRightZero
        return &fieldpath_operations[7];
    case 220: // PushOneLeftDeltaNRightZero
        return &fieldpath_operations[9];
    case 222: // PopAllButOnePlusNPack6Bits
        return &fieldpath_operations[32];
    case 223: // PlusFour
        return &fieldpath_operations[3];
    case 432: // PopAllButOnePlusN
        return &fieldpath_operations[30];
    case 438: // PushOneLeftDeltaNRightNonZeroPack8Bits
        return &fieldpath_operations[12];
    case 439: // NonTopoPenultimatePlusOne
        return &fieldpath_operations[37];
    case 442: // PopAllButOnePlusNPack3Bits
        return &fieldpath_operations[31];
    case 443: // PushNAndNonTopological
        return &fieldpath_operations[26];
    case 866: // NonTopoComplexPack4Bits
        return &fieldpath_operations[38];
    case 1735: // NonTopoComplex
        return &fieldpath_operations[36];
    case 3469: // PushOneLeftDeltaZeroRightZero
        return &fieldpath_operations[5];
    case 27745: // PopOnePlusOne
        return &fieldpath_operations[27];
    case 27749: // PushOneLeftDeltaZeroRightNonZero
        return &fieldpath_operations[6];
    case 55488: // PopNAndNonTopographical
        return &fieldpath_operations[35];
    case 55489: // PopNPlusN
        return &fieldpath_operations[34];
    case 55492: // PushN
        return &fieldpath_operations[25];
    case 55493: // PushThreePack5LeftDeltaN
        return &fieldpath_operations[24];
    case 55494: // PopNPlusOne
        return &fieldpath_operations[33];
    case 55495: // PopOnePlusN
        return &fieldpath_operations[28];
    case 55496: // PushTwoLeftDeltaZero
        return &fieldpath_operations[13];
    case 110994: // PushThreeLeftDeltaZero
        return &fieldpath_operations[15];
    case 110995: // PushTwoPack5LeftDeltaZero
        return &fieldpath_operations[14];
    case 111000: // PushTwoLeftDeltaN
        return &fieldpath_operations[21];
    case 111001: // PushThreePack5LeftDeltaOne
        return &fieldpath_operations[20];
    case 111002: // PushThreeLeftDeltaN
        return &fieldpath_operations[23];
    case 111003: // PushTwoPack5LeftDeltaN
        return &fieldpath_operations[22];
    case 111004: // PushTwoLeftDeltaOne
        return &fieldpath_operations[17];
    case 111005: // PushThreePack5LeftDeltaZero
        return &fieldpath_operations[16];
    case 111006: // PushThreeLeftDeltaOne
        return &fieldpath_operations[19];
    case 111007: // PushTwoPack5LeftDeltaOne
        return &fieldpath_operations[18];
    default:
        return nullptr;
    }
}

if you run this file it'll open a graph of the reconstructed huffman tree, containing all codes. (on linux!)

Update: here it is!

@spheenik thanks! turns out my tree was slightly wrong, but I was able to correct it based on your graph. I wasn't counting the nodes as they were added to increment what you call "num"; for intermediate nodes I was just taking the largest of the leaf node's nums instead of assigning a new num to each intermediate nodes, which put a few things out of order at the bottom of the tree where all the leaf nodes have weights of zero.