TSVReader.clj

(ns net.myexperiments.viciouscycle.util.TSVReader
  (:require [clojure.string :as st])
  (:import (java.util.regex Pattern) (java.util HashMap) (java.io File))
  (:gen-class
    :name    "net.myexperiments.viciouscycle.util.TSVReader"
    :prefix "-"
    :load-impl-ns false
    :methods [^{:static true} [loadTSV [String] java.util.Map]
              ^{:static true} [readMap [String] java.util.HashMap]
              ^{:static true} [convertTSV [String String] String]]))

(defn load-tsv-clj
  "Reads a TSV file in and returns a sorted map with the first column's data
  as keys, and for values, hashmaps of the header row's data as keys to the
  other rows' data as values.  A file with a first row of 'Name\tID\tSalary'
  and a second row of 'John\t1\t100' will return a sorted map of {\"John\"
  {\"Name\" \"John\", \"ID\" 1, \"Salary\" 100}}"
  [filename]
  (let [tbl-rows (st/split-lines  (slurp (File. filename)))
        pat-t (. Pattern compile "\t")
        header (st/split (first tbl-rows) pat-t)
         read-unknown (fn [s] (if (integer? (read-string s)) (read-string s) s))
         make-data (fn [rw] (zipmap header (map read-unknown (st/split rw pat-t))))]
  (apply sorted-map (interleave
                                     (map (fn [rw] (first (st/split rw pat-t))) (rest tbl-rows))
                                     (map make-data (rest tbl-rows))))))
(defn -loadTSV
  "Reads a TSV file in and returns a hash-map with the first column's data
  as keys, and for values, hash-maps of the header row's data as keys to the
  other rows' data as values.  A file with a first row of 'Name\tID\tSalary'
  and a second row of 'John\t1\t100' will return a hash-map of {\"John\"
  {\"Name\" \"John\", \"ID\" \"1\", \"Salary\" \"100\"}}"
  [filename]
  (let [tbl-rows (st/split-lines  (slurp (File. filename)))
        pat-t (. Pattern compile "\t")
        header (st/split (first tbl-rows) pat-t)
        make-data (fn [rw] (zipmap header (st/split rw pat-t)))]
  (apply hash-map (interleave
                                     (map (fn [rw] (first (st/split rw pat-t))) (rest tbl-rows))
                                     (map make-data (rest tbl-rows))))))
(defn read-obj
  "Reads a Clojure object from a file, ignoring exceptions."
  [filename]
  (try
     (read-string (slurp  (File. filename)))
  (catch Exception e nil)))
(defn -readMap
  "Reads a Clojure map from a file, returning a Java HashMap
   equivalent to the data or an empty HashMap on failure."
  [filename]
  (try
    (let [f (read-string (slurp (File. filename)))]
     (if (map? f) (HashMap. f) (HashMap.)))
  (catch Exception e (HashMap. ))))

(defn write-obj
  "Writes a Clojure object to a file, to be loaded with read-obj ."
  [filename obj]
  (binding [*out* (java.io.FileWriter. filename)]
    (prn obj)))

(defn -convertTSV
  "Reads a TSV file (with the name specified by the String infile)
   in with loadTSV, using the same format, and writes it to a
   Clojure file (with the name specified by the String outfile)."
  [infile outfile]
  (binding [*out* (java.io.FileWriter. outfile)]
    (prn (load-tsv-clj infile))))