Hi All,
I read about Nim for the 1st time last week and decided to make something over
the weekend. I was pleasantly surprised that I was able to cobble something
together that did what I wanted it to do, but am a bit surprised that it's not
faster. Maybe some of you could give some suggestions on how to improve
performance?
The goal was to write a program that takes a word as input from the user, reads
a local file of dictionary words, and returns all of the words in the
dictionary that are anagrams of the user’s original word.
The code I came up with is this.
# anagram_redux.nim
# ---- Imports
----------------------------------------------------------------#
import strutils
import algorithm
import os
import strformat
import tables
import unidecode
# ---- Converts a word into it's lower-case, sorted "signature"
---------------#
proc makeSignature(word: string): string =
let ascii = unidecode(word)
let lower = strutils.toLowerAscii(ascii)
let sorted_word = algorithm.sorted(lower, system.cmp)
let signature = sorted_word.join()
result = signature
# ---- Main body of program
---------------------------------------------------#
proc main =
# -- Parse the lookup word from the argument passed in
if os.paramCount() < 1:
quit("Usage: anagram_redux <word>")
let lookup_word = os.paramStr(1)
let lookup_signature = makeSignature(lookup_word)
echo strformat.fmt("Looking up '{lookup_word}'")
# -- Create empty table to store results in
var anagram_table = tables.initTable[string, seq[string]]()
# -- Open the text file and process it
let file = open("dict.txt")
for word in file.lines():
let signature = makeSignature(word)
if anagram_table.hasKey(signature):
anagram_table[signature].add(word)
else:
anagram_table[signature] = @[word]
# -- Lookup user's word via its signature in the anagram_table
if anagram_table[lookup_signature].len == 1:
echo strformat.fmt("'{lookup_word}' does not have any anagrams")
else:
echo anagram_table[lookup_signature]
main()
Run
The "dict.txt" file with dictionary words is on github at
[https://github.com/lagerratrobe/nim_dev](https://github.com/lagerratrobe/nim_dev).
For comparison, here are the performance numbers I get from this and similar
implementations in Python and R.
* Nim unoptimized
real 0m1.809s
user 0m1.782s
sys 0m0.012s
Run
* Nim --opt:speed
real 0m0.599s
user 0m0.575s
sys 0m0.016s
Run
* Python
real 0m0.348s
user 0m0.325s
sys 0m0.020s
Run
* R
user system elapsed
1.773 0.020 1.803
Run
