Added support for Shift + Click selection of rows

[patorjk]

An implementation of the feature suggested in #663

The UX for shift+click is modeled off of how GMail does it. For example, if you select/unselect row A and then hold shift and select/unselect row B:

• If row B is being selected, then the rows between A and B (including A) will be selected.
• If row B is being unselected, then the rows between A and B (including A) will be unselected.

If any filters are removed or columns resorted or selection headers toggled or data fetched (in the case of Server Side tables), between the selection of rows A and B, then there will be no special effect.

[coveralls]

Coverage increased (+1.03%) to 76.248% when pulling 7581dab on patorjk:shift-click into d1914e4 on gregnb:master.

[patorjk]

This PR is most likely not compatible with #825, however, I could adjust it so that in the MUIDataTable.js file, when it was in the process of adding the adjacent rows, it would stop if it hit the maxSelectedRows threshold.

[gabrielliwerant]

This is a cool feature, thank you for working on this!

I just reviewed and suggested a number of changes to #825, so how about we wait for that to be merged and integrated with yours before review?

[patorjk]

Sounds good, I’ll be unavailable next week but could make any adjustments tomorrow or the week after that.

[gabrielliwerant]

@patorjk The isRowSelectable changes are in latest master now, if you want to pull those.

[patorjk]

Sounds good, I’ll pull them and update this PR early next week!

[patorjk]

For trying this out, I added (but didn't commit) an isRowSelectable function that looked like this:

  isRowSelectable: function(dataIndex, selectedRows) {
    //return true;
    if (selectedRows.data.length > 4 && selectedRows.data.filter(d => d.dataIndex === dataIndex).length === 0) return false;
    return data[dataIndex][1] != "Attorney";
  }

[gabrielliwerant]

Worked really well for me! I tested in multiple scenarios, including the isSelectable you pointed out. I only have a few nitpicks, but otherwise looks great.

[gabrielliwerant]

It would be self-documenting and probably a bit less expensive to use lodash.clonedeep, which is in the library, here instead.

[gabrielliwerant]

I believe we can use let here and below on line 129 since the variables don't need to leave their respective blocks.