How to unpack a table with a column of structs while preserving other columns?

I have a table with 2 fields, ['A', 'B'] where 'B' has structs of 1xn in each cell corresponding to A. So the table looks something like this:

A B

1 {1xn}

2 {1xm}

. .

. .

What I want is to flatten the structs while retaining the corresponding relations with the other column (The values in the other column can be repeated):

A B

1 1x1

1 1x2

1 1x3

2 1x1

2 1x2

. .

How can I acheive this?

Additional information:

The structs have all the same fields, just different number of rows

UPDATE:

I have attached a sample test file that can represent this better. What I currently have in the test_01.mat file is this:

What I am trying to get is a flattened structure after unpacking the contents of the structs in the second column like this:

Please let me know if I can provide any additional information, thanks!

[Adam Danz]

You probably want a combination of varfun and struct2array.

[Parth Uday Sanghvi]

Okay thanks I'll try researching more about varfun. However, could you elaborate the logic you have in mind?

[Adam Danz]

Actually, you won't need varfun because it looks like you're only operating on one column of the table. Look up Struct2array in the documentation to get a sense of what it does. I think that's what you're looking for but I don't have a clear picture of what your structures look like. You would apply it to every row of that column.

[Adam Danz]

Sure. If demo data is provided I'd be happy to.

[Parth Uday Sanghvi]

Updated with a small subset of the data I am working with. As you can see I need to create a super table with the contents of the struct in the cell unpacked into indiviual rows while maintaining the relation with the first column. Thanks!

[Adam Danz]

Your structures are not purely numeric so you'd have to use sruct2table (or struct2cell) rather than struct2array.

Answer 1

[Adam Danz]

Tnew is the input table (attached).

gtTbl = cellfun(@struct2table, Tnew.groundtruth2DBB, 'Unif', false); 
T = [repelem(Tnew(:,1), cellfun(@numel, Tnew.groundtruth2DBB),1), vertcat(gtTbl{:})];

Result

T =
  16×5 table
                                        sequenceName                                         objid                     gtBb2D                         classname         has3dbox
    _____________________________________________________________________________________    _____    ________________________________________    __________________    ________
    {'SUNRGBD/kv2/kinect2data/000002_2014-05-26_14-23-37_260595134347_rgbf000103-resize'}      0          328        152        346        320    {'bed'           }     true   
    {'SUNRGBD/kv2/kinect2data/000002_2014-05-26_14-23-37_260595134347_rgbf000103-resize'}      1       572.54     275.64     98.378     53.177    {'night_stand'   }     true   
    {'SUNRGBD/kv2/kinect2data/000002_2014-05-26_14-23-37_260595134347_rgbf000103-resize'}      2       332.36     318.18      316.4     209.16    {'ottoman'       }     true   
    {'SUNRGBD/kv2/kinect2data/000002_2014-05-26_14-23-37_260595134347_rgbf000103-resize'}      3       157.76     10.635     156.87     362.49    {'dresser_mirror'}     true   
    {'SUNRGBD/kv2/kinect2data/000002_2014-05-26_14-23-37_260595134347_rgbf000103-resize'}      4            1          1        221        529    {'dresser'       }     true   
    {'SUNRGBD/kv2/kinect2data/000002_2014-05-26_14-23-37_260595134347_rgbf000103-resize'}      5        614.2      200.3      54.95     93.946    {'lamp'          }     false  
    {'SUNRGBD/kv2/kinect2data/000002_2014-05-26_14-23-37_260595134347_rgbf000103-resize'}      6       414.49     230.43     153.26     48.155    {'pillow'        }     false  
    {'SUNRGBD/kv2/kinect2data/000003_2014-05-26_14-24-42_260595134347_rgbf000040-resize'}      0            1        128        498        402    {'bed'           }     true   
    {'SUNRGBD/kv2/kinect2data/000003_2014-05-26_14-24-42_260595134347_rgbf000040-resize'}      1          462        303        197        195    {'night_stand'   }     true   
    {'SUNRGBD/kv2/kinect2data/000003_2014-05-26_14-24-42_260595134347_rgbf000040-resize'}      2       338.93     254.42     130.43     72.563    {'pillow'        }     false  
    {'SUNRGBD/kv2/kinect2data/000003_2014-05-26_14-24-42_260595134347_rgbf000040-resize'}      3       546.23     181.08     74.738     150.76    {'lamp'          }     false  
    {'SUNRGBD/kv2/kinect2data/000003_2014-05-26_14-24-42_260595134347_rgbf000040-resize'}      4       241.29     236.64     100.59     58.495    {'pillow'        }     false  
    {'SUNRGBD/kv2/kinect2data/000003_2014-05-26_14-24-42_260595134347_rgbf000040-resize'}      5       140.03     212.71     110.79     51.405    {'pillow'        }     false  
    {'SUNRGBD/kv2/kinect2data/000003_2014-05-26_14-24-42_260595134347_rgbf000040-resize'}      6       7.9766      38.11     75.334     130.28    {'lamp'          }     false  
    {'SUNRGBD/kv2/kinect2data/000003_2014-05-26_14-24-42_260595134347_rgbf000040-resize'}      7       46.973     195.87     106.35     48.746    {'night_stand'   }     false  
    {'SUNRGBD/kv2/kinect2data/000003_2014-05-26_14-24-42_260595134347_rgbf000040-resize'}      8      -1.7726     1.7726     68.244     238.41    {'dresser_mirror'}     false 

[Adam Danz]

Note, if the sequenceName column position is variable or unpredictable, do not use the hard-coded index of 1. Instead, use

seqIdx = strcmp(Tnew.Properties.VariableNames,'sequenceName');
. . . Tnew(:,seqIdx) . . .

[Parth Uday Sanghvi]

So this works great in the test filel, unfortunately when I try doing this on the entire dataset, I get the following error, and I am not sure why this might be? The input is a single column as you displayed in the test file. I am also attaching the complete mat file for reference. Really appreciate the quick responses!

Error using struct2table (line 30)
Input structure must be a scalar structure, or a structure array with one column or one row.

[Adam Danz]

That's because you've got at least 1 empty structure. The empty structures are in rows,

find(cellfun(@isempty, x.groundtruth2DBB))

The solution is easy, remove the rows with empty structures before implementing the 2 lines in my answer.

% Remove rows with empty structs
x(cellfun(@isempty, x.groundtruth2DBB),:) = []; 

[Parth Uday Sanghvi]

Worked great, thanks so much. Closing now

[Adam Danz]

Glad I could help. Alternatively you could replace the empty structures with default values if you want to keep those rows in the data.