run mapreduce in bigquery

gistfile1.txt

function mapper(row, emit) {
  if(row.comment) {
  keywords = row.comment.split(' ');
  for(var i=0; i<keywords.length; i++) {
    emit({keyword: keywords[i], count: 1});
  }
  }
}

function reducer(row, emit) {
  var total = 0;
  for(var i=0;i<row.count.length; i++) {
    total += row.count[i];
  }
  emit({keyword: row.keyword, total: total});
}

bigquery.defineFunction(
  'mapper',                           // Name of the function exported to SQL
  ['comment'],                    // Names of input columns
  [{'name': 'keyword', 'type': 'string'},  // Output schema
   {'name': 'count', 'type': 'integer'}],
  mapper                       // Reference to JavaScript UDF
);


bigquery.defineFunction(
  'reducer',                           // Name of the function exported to SQL
  ['keyword', 'count'],                    // Names of input columns
  [{'name': 'keyword', 'type': 'string'},  // Output schema
   {'name': 'total', 'type': 'integer'}],
  reducer                       // Reference to JavaScript UDF
);

select keyword, total 
from reducer(select keyword, nest(count) as count from 
   mapper(select comment from [publicdata:samples.wikipedia] limit 1000)
   group by keyword
   )