Mirror of forum.nim-lang.org

81 :: Map-Reduce functionality?

• bb (orginal) [2012-10-06T08:25:15+02:00] view original

  Hi. It's been a pleasure to met with this great language.

  So I'm trying to convert my old school Python code into Nimrod like as the followings:

  
  import unicode
  import strutils
  
  # print s[2]
  const s = "ニムロッド" # "Nimrod" in Japanese, 5 unicode characters length
  var list: seq[string] = @[]
  for x in runes(s):
      list.add(toUTF8(x))
  echo list[2]
  
  # print s[1:4]
  echo list[1..4].join
  
  # print s[2:]
  echo list[2..(-1)].join
  Run

  However, it seems the first part is not clear as it should be. My intention is:

  
  var list: seq[string] = reduce(toUTF8, runes(s))
  Run

  Is there any way to do something like this? Thanks in advance.

---

• Araq (orginal) [2012-10-06T22:09:53+02:00] view original

  Since closures have only been added recently, the stdlib still lacks foldleft/right. It's easy to implement them though.

  You don't need a seq[string] and then join all the strings together. Nimrod's strings are mutable, so just use add for strings directly.

---

• bb (orginal) [2012-10-07T02:58:16+02:00] view original

  1. I see. I'm going to wait for it. Nimrod is so promising anyway :)

  2. I think, I don't have to do it for syntax, but I have to, for semantics. I would like to write an Unicode library which supports all basic functions (replace, toupper, etc.):

  
  import unicode
  import strutils
  
  proc u(s: string): seq[string] =
    var list: seq[string] = @[]
    for x in runes(s):
      list.add(toUTF8(x))
    return list
  
  const s = "ニムロッド"
  const v = u(s)
  echo v[2]
  echo v[1..4].join
  echo v[2..(-1)].join
  Run

  This one really works. But the following one didn't work, because adding every Unicode characters into one string variable makes one binary string (and result in binary outputs; not Unicode character sequences).

  
  import unicode
  
  proc u(s: string): string =
    var r: string = ""
    for x in runes(s):
      r.add(toUTF8(x))
    return r
  
  const s = "ニムロッド"
  const v = u(s)
  echo v[2]
  echo v[1..4]
  echo v[2..(-1)]
  Run

  It would be great if we can handle both Unicode string and character at once.

---

• Araq (orginal) [2012-10-07T10:57:59+02:00] view original

  1. Learn to use 'result'. Typical newbie mistake. ;-)
  2. You could overload {} for unicode aware access to strings:

     
     
     proc `{}`*(s: string, i: int): TRune = runeAt(s, i)
     Run

  But I see runeAt doesn't implement what you want here since i is a byte index then. Well, feel free to add a proper implementation to unicode.nim and do a pull request. ;-)

  3. Beware I didn't test:

  
  proc foldl[U, V](fn: proc (a: V, b: U): V, s: openArray[U], z: V): V =
    result = z
    for i in 0..s.high:
      result = fn(result, s[i])
  Run

---

• bb (orginal) [2012-10-07T13:21:05+02:00] view original

  Thank you so much for your help!

  Wow, I didn't even imagine the handy 'result' exists. I'll check the manual first, and think how can I make a good contribution to unicode.nim (hopefully compatible way).

---