Additional Python and Lua benchmarks for: http://matthewrocklin.com/blog/work/2014/01/13/Text-Benchmarks/

benchmark1.lua

local function groupby(func, seq)
    local t = {}
    for _, val in ipairs(seq) do
        local key = func(val)
        t[key] = t[key] or {}
        t[key][#t[key] + 1] = val
    end
    return t
end

local function dowork(filename)
    local f = io.open(filename, "r")
    local t = {}
    for n1, n2 in string.gmatch(f:read("*all"), "(%S+) (%S+)") do
        t[#t + 1] = {n1, n2}
    end
    f:close()
    return groupby(function(x) return x[1] end, t)
end

return { dowork=dowork }

benchmark1.py

from sys import argv
from toolz import groupby, first
from utils import duration
from re import findall
from operator import itemgetter

filename = argv[1]

with duration():
    with open(filename, 'r') as file:
        result = groupby(itemgetter(0), findall('(\S+) (\S+)', file.read()))

benchmark2.lua

-- instead of calling `groupby`, we store the word pairs as done in the Java example
local function dowork(filename)
    local f = io.open(filename, "r")
    local t = {}
    for k, v in string.gmatch(f:read("*all"), "(%S+) (%S+)") do
        t[k] = t[k] or {}
        t[k][#t[k] + 1] = v
    end
    f:close()
    return t
end

return { dowork=dowork }

install_commands.sh

# these are the commands I used to install `lupa` on 64-bit Ubuntu
git clone https://github.com/scoder/lupa.git
cd lupa
wget http://luajit.org/download/LuaJIT-2.0.2.tar.gz
tar zxvf LuaJIT-2.0.2.tar.gz
cd LuaJIT-2.0.2
make CFLAGS="-fPIC"
cd ../
# add "--user" to below to install only for current user 
# add "--with-cython" if you want to build lupa using cython
python setup.py install

lupa_benchmark1.py

import lupa
from sys import argv
from utils import duration

lua = lupa.LuaRuntime()
t = lua.require('benchmark1')
filename = argv[1]

with duration():
    t.dowork(filename)

lupa_threadedbenchmark1.py

import threading
import lupa
from sys import argv
from utils import duration

thread_count = len(argv) - 1

lua_funcs = [lupa.LuaRuntime().require('benchmark1').dowork
             for _ in range(thread_count)]
results = [None] * thread_count

def runthread(i, filename, lua_func):
    results[i] = lua_func(filename)

threads = [threading.Thread(target=runthread, args=(i, argv[i+1], lua_func))
           for i, lua_func in enumerate(lua_funcs)]

with duration():
    for thread in threads:
        thread.start()
    for thread in threads:
        thread.join()