ithayer

(ns ignacio.tfidf (:require [clojure.contrib.string  :as string])) ;; Simple tfidf in clojure, for fun.

(def stopwords (set (string/split #"\n" (slurp "./stopwords.txt"))))

(defn tokenize [raw-text] ;; Lowercases and splits on non-letters, non-numbers.
  (remove stopwords (string/split #"[^a-z0-9äöüáéíóúãâêîôûàèìòùçñ]+" (string/lower-case raw-text))))

(defn idf2 [n-docs match] (Math/pow (Math/log (/ n-docs (count (keys match)))) 2))

(defn index-one [fname] ;; Index for one file. Given an fname, returns a map of token -> map of (fname, count)

ithayer

[org.clojars.ithayer/plaid-penguin "1.0.0"]

ithayer

var sys = require('sys'),
    Path  = require('path'),
    Mu = require('../mu'),
    Script = process.binding('evals').Script;

exports.compile = compile;
exports.compilePartial = compilePartial;

function RenderEventEmitter(options) {
  this.chunkSize = options.chunkSize || 1024;