Re: string patterns

[Philippe Lhoste <PhiLho@...>]

Sam Roberts wrote:
> Uh, self-documenting if you know lpeg, complete goobledygook if you
> don't!

Of course, that's the case for lot of languages.
The self-documenting part is on the fact that elements get explicit 
names, unlike most regular expressions (I know the named capture stuff..).

At the cost of even more verbosity, one can make it perhaps more 
readable, if knowing the base operators (^0, ^1, +, *):

----- LPegWrapper.lua
#!Lua51.exe

m = require"lpeg"

Pattern = m.P
String = m.P
Range = m.R
Set = m.S
Ref = m.V
Capture = m.C
IndexCapture = m.I
TableCapture = m.T
function Not(p) return 1 - p end

----- LPegTests.lua

#!Lua51.exe

dofile"LPegWrapper.lua"
dofile[[DumpObject.lua]]

space = Set" \t"
sp = space^0
equal = String"="
name = Capture(Not(space + equal)^1, "name")
sval = Capture(Not(space)^1, "value")
singleQuote = String"'"
doubleQuote = String'"'
quotedValue = singleQuote * Capture(Not(singleQuote)^0, "value") * 
singleQuote
doubleQuoteval = doubleQuote * Capture(Not(doubleQuote)^0, "value") * 
doubleQuote
pair = TableCapture(sp * name * sp * equal * sp * (quotedValue + 
doubleQuoteval + sval))
args = TableCapture(pair^0)

function ExtractArgs(s)
    mr = m.match(s, args)
    print(DumpObject(mr))
end
ExtractArgs[[class=q href="http://images.goo gle.se/imghp?hl=sv&tab=wi" 
onClick='return qs(this);' test=foo]]

----- Result

local T =
{
  [1] =
  {
    name = "class",
    value = "q"
  },
  [2] =
  {
    name = "href",
    value = "http://images.goo gle.se/imghp?hl=sv&tab=wi"
  },
  [3] =
  {
    name = "onClick",
    value = "return qs(this);"
  },
  [4] =
  {
    name = "test",
    value = "foo"
  }
}

--
Philippe Lhoste
--  (near) Paris -- France
--  http://Phi.Lho.free.fr
--  --  --  --  --  --  --  --  --  --  --  --  --  --