Mirror of forum.nim-lang.org

9723 :: TinyRE - Tiny Regex Engine for Nim

• Ward (orginal) [2022-12-14T03:42:34+01:00] view original

  TinyRE is a Nim wrap for a tiny regex engine less than 10K binary size (loc < 1K), and guarantees that input regex will scale O(n) with the size of the string.

  https://github.com/khchen/tinyre

  Features:

  • Support unicode.
  • Support case-insensitive matching.
  • Support global matching.
  • Support most common regex syntax, including:
    • Greedy and non-greedy expressions: *, +, ?, *?, +?, ??
    • Characters sets: [xyz], [^xyz]
    • Meta characters: \s, \S , \ w, \W, \d, \D, \n, \r, \t etc.
    • Ascii or unicode characters: \x00, \u0000, \U00000000
    • Beginning and end assertions: ^, $.
    • Repetition operators: {n}, {n,m}, {n,}.
    • Group and non-capture group: (...), (?:...).
    • Start-of-word and end-of-word assertions: \<, \>.

  Examples:

  import tinyre
  
  doAssert match("abc123", re"\d+") == @["123"]
  doAssert bounds("abc123", re"\d+") == @[3..5]
  doAssert contains("abc123", re"\d+") == true
  doAssert startsWith("abc123", re"[a-z]+") == true
  doAssert endsWith("abc123", re"\d+") == true
  doAssert split("abc123", re"\d+") == @["abc", ""]
  doAssert replacef("abc123", re"([a-z]+)(\d+)", "$2$1") == "123abc"
  
  # reG for global matching
  doAssert match("abc123", reG".") == @["a", "b", "c", "1", "2", "3"]
  
  # reI for case insensitive matching
  doAssert match("abc123", reI"ABC") == @["abc"]
  
  # reU for utf8 matching
  doAssert match("中文", reU"..") == @["中文"]
  Run

---

• slangmgh (orginal) [2022-12-14T07:17:39+01:00] view original

  I like it, thank you.

---

• Araq (orginal) [2022-12-14T08:20:59+01:00] view original

  The standard re/nre modules should link to these awesome alternatives. There is also nim-regex.

---

• Yardanico (orginal) [2022-12-14T08:24:43+01:00] view original

  Nice wrapper! Are you planning to rewrite the underlying C code in Nim in the future? So that it could e.g. work at compile-time or in JS backend.

---

• Ward (orginal) [2022-12-14T11:34:33+01:00] view original

  Add a benchmark result on github.

  # small string: "abc123def".contains("\d+")
  # large string: 6.71 MB text file
  #   email: [\w\.+-]+@[\w\.-]+\.[\w\.-]+
  #   uri: [\w]+://[^/\s?#]+[^\s?#]+(?:\?[^\s#]*)?(?:#[^\s]*)?
  #   ipv4: (?:(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9])\.){3}(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9])
  # compile options: -d:release -d:danger --opt:speed -d:lto
  
  name ............................... min time      avg time    std dv   runs
  tinyre (small string) .............. 0.366 ms      0.379 ms    ±0.018  x1000
  std/re (small string) .............. 5.862 ms      6.218 ms    ±0.171   x797
  nim-regex (small string) .......... 16.132 ms     17.067 ms    ±0.580   x288
  tinyre (large string, email) ..... 140.684 ms    151.663 ms    ±8.625    x33
  std/re (large string, email) ...... 44.793 ms     48.884 ms    ±2.716   x102
  nim-regex (large string, email) .... 3.680 ms      3.921 ms    ±0.132  x1000
  tinyre (large string, uri) ....... 127.465 ms    131.721 ms    ±2.110    x38
  std/re (large string, uri) ........ 40.380 ms     42.812 ms    ±1.175   x117
  nim-regex (large string, uri) ..... 21.400 ms     22.205 ms    ±0.344   x225
  tinyre (large string, ipv4) ...... 182.995 ms    186.441 ms    ±1.057    x27
  std/re (large string, ipv4) ........ 4.854 ms      5.965 ms    ±0.903   x838
  nim-regex (large string, ipv4) ..... 7.569 ms      7.849 ms    ±0.159   x635
  Run

  Port to pure nim is possible, but I don't know is it still "tiny". https://github.com/nitely/nim-regex is already a pure nim regex and works well in compile-time. In JS backend, using https://nim-lang.org/docs/jsre.html seems more reasonable.

---

• DeletedUser (orginal) [2023-03-15T17:19:13+01:00] view original

  I just started with Nim and I'm trying to do the things I did in Python. Can I get groups with tinyRE?

  var ppl = """olle=7, pelle=12, lisa=21, ringhals=42"""

  myMatches = match(ppl, reG"(w+)=(d+)")

  echo myMatches.groups

  @[["olle",7], ["pelle",12], ["lisa",21],["ringhals,42]]

---