sankey diagram d3.js problem with large data

The D3's Sankey plugin uses the numeric (index) position of the element in the nodes array for the source and target properties of the links array. In your example you are using names. You can preprocess the file before use to make the replacement, or make the replacement in real time using:

d3.json("sankey-data.json", function(error, data) {
   if(error) console.log(error)

   function getIndex(name) { // descobre o índice do nome
       return data.nodes.findIndex(function(d) {
           return d.name == name;
       });
   }

   for(var i = 0; i < data.links.length; i++) { // troca nomes por índices
       data.links[i].target = getIndex(data.links[i].target);
       data.links[i].source = getIndex(data.links[i].source);
   }

   // agora o array *data* pode ser usado pelo sankey

...

A second problem you may encounter is an infinite loop, as it seems to me that your data contains circular paths, which is not supported by Sankey. Test your file with less data (removing circular paths) and then take a look at link that cites some alternatives to address this issue.