couchdb - problems grouping collections

I'm having problems trying to get my head around getting a collection of types along with number of times a skill is found in that doc type.

There are a number of document types that have a list of skills.

{
   "skills": "Windows, Network Admin, Linux",
   "type": "Experience"
},
{
   "skills": "Windows, Erlang, Linux",
   "type": "Experience"
},

{
   "skills": "Ruby, Rails, Erlang",
   "type": "Project"
}

I'm trying to get the number of times a skill is found in an document type.

The end result should look something like this:

{
  'type': Experience,
  'skills': [
    {'skill': 'Erlang', 'count': 1},
    {'skill': 'Linux', 'count': 2},
    {'skill': 'Network Admin', 'count': 1},
    {'skill': 'Rails', 'count': 0},
    {'skill': 'Ruby', 'count': 0},
    {'skill': 'Windows', 'count': 2}
  ]
},
{
  'type': Project,
  'skills': [
    {'skill': 'Erlang', 'count': 1},
    开发者_JAVA技巧{'skill': 'Linux', 'count': 0},
    {'skill': 'Network Admin', 'count': 0},
    {'skill': 'Rails', 'count': 1},
    {'skill': 'Ruby', 'count': 1},
    {'skill': 'Windows', 'count': 0}
  ]
}

What would be the best way to do this?

You should store the skill list in your docs as a real list.

{
   "skills": ["Windows", "Network Admin", Linux"],
   "type": "Experience"
}

From there, your map function becomes:

function(doc) {
  for(var skill in doc.skills) {
    emit([doc.type, skill], 1);
  }
}

And the reduce function just sums the result, so use "_sum". You can easily choose to show only Experience or Project with start and end keys.

Here are map and reduce functions (view named skill_split) and a list function that will produce output close to what you asked for using a URL like this http://127.0.0.1:5984/myskills/_design/myskills/_list/transform/skill_split?group=true:

{
   "Experience": {"Erlang":1, "Linux":2, "Network Admin":1, "Windows":2},
   "Project":{"Erlang":1, "Rails":1, "Ruby":1}
}

I can tweak the code to make the output exactly what you asked for, but the list function would be a bit longer.

map:

function(doc) {
  if (doc.type && doc.skills) {
    doc.skills.split(', ').forEach(function(skill) {
      emit([doc.type, skill], 1);
    });
  }
}

reduce:

function(keys, values, rereduce) {
  return sum(values)
}

or use "reduce": "_sum" for short

list (named transform):

function(head, req) {
  var results = {};
  var row;
  while (row = getRow()) {
    var skill_map;
    if (results.hasOwnProperty(row.key[0])) {
      skill_map = results[row.key[0]];
    } else {
      skill_map = {};
      results[row.key[0]] = skill_map;
    }

    if (skill_map.hasOwnProperty(row.key[1])) {
      skill_map[row.key[1]] = skill_map[row.key[1]] + row.value;
    } else {
      skill_map[row.key[1]] = row.value;
    }
  }
  send(JSON.stringify(results));
}

If you don't have access to the JSON object (you do inside a CouchApp), you might have to jump through some hoops to get access to it.