n-obs: wrapper for Cascalog's "first-n" operation. Returns N observations of all fields for a given source.

gistfile1.clj

(use 'cascalog.api)
(require '[cascalog.ops :as c])

(defn field-parser
  "Convert a string to a fieldname string. If string already 
  starts with `?` or `!`, return the string directly.
  
   Usage:
    (field-parser [\"a\" \"b\"])
    ;=> [\"?a\" \"?\"]"
  [field]
  (if (or (= \? (first field))
          (= \! (first field)))
    field
    (str "?" field)))

(defn field-ify
  "Convert vector of strings into fieldname strings a la `?a`."
  [field-strings]
  (vec (map field-parser field-strings)))

(defn n-obs
  "Returns a tap that will return N observations for all fields
  for a given source."
  [n field-names src]
  (let [fields (field-ify field-names)
        query (<- fields
                  (src :>> fields))]
    (<- fields
      ((c/first-n query n) :>> fields))))