lama_byterun/stdlib/Matcher.lama

-- (C) Dmitry Boulytchev, St. Petersburg State University, JetBrains Research, 2020
-- Matcher: simple string matching library.

-- Create a regular expression representation.
-- Arguments:
--    r    --- a string representation for regular expression (as per GNU regexp)
--    name --- a string describing the meaning of the expression in free form
--             (e.g. "identifier", "string constant", etc.), used for error
--             reporting
public fun createRegexp (r, name) {
  var l = [regexp (r), name];
  l
}

-- Create an immutable matcher.
-- Arguments:
--    buf --- a string to match in
--    pos --- an integer beginning position to match from
--    line, col --- line and column numbers
-- This function is internal, do not use it directly.
-- To initially create a matcher use initMatcher function (see below).
fun createMatcher (buf, pos, line, col) {
  -- Shows a matcher in a readable form
  fun show () {
    sprintf ("buf : %-40s\npos : %d\nline: %d\ncol : %d\n", buf, pos, line, col)
  }

  -- Calculates the number of remaining unmatched characters in the buffer
  fun rest () {
    buf.length - pos
  }

  -- Moves the position pointer on given number of characters.
  fun shift (n) {
    var i, l = line, c = col;
    
    for i := pos, i < pos+n, i := i+1 do
      case buf [i] of
        '\n' -> l := l + 1; c := 1
      | '\t' -> c := c + 8
      | _    -> c := c + 1
      esac
    od;
    
    createMatcher (buf, pos + n, l, c)
  }

  fun matchString (s) {
    fun min (x, y) {
      if x < y then x else y fi
    }
    
    -- printf ("Matching string %s against %s...\n", s, substring (buf, pos, min (10, buf.length - pos)));

    if s.length > rest ()
    then Fail (sprintf ("""%s"" expected", s), line, col)
    elif matchSubString (buf, s, pos) then Succ (s, shift (s.length)) 
    else Fail (sprintf ("""%s"" expected at", s), line, col)
    fi
  }

  fun matchRegexp (r) {
    var n;

    fun min (x, y) {
      if x < y then x else y fi
    }
    
    -- printf ("Matching regexp %x against %s...\n", r, substring (buf, pos, min (10, buf.length - pos)));

    if (n := regexpMatch (r[0], buf, pos)) >= 0
    then Succ (substring (buf, pos, n), shift (n))
    else Fail (sprintf ("%s expected", r[1]), line, col)
    fi
  }

  fun eof () {
    if rest () == 0
    then Succ ("", shift (0))
    else Fail ("EOF expected", line, col)
    fi
  }
  
  [show,
   eof,
   matchString,
   matchRegexp,
   fun () {line},
   fun () {col}]  
}

public fun showMatcher (m) {
  m [0] ()
}

public fun endOfMatcher (m) {
  m [1] ()
}

public fun matchString (m, s) {
  m [2] (s)
}

-- Matches against a regexp
public fun matchRegexp (m, r) {
  m [3] (r)
}

-- Gets a line number
public fun getLine (m) {
  m [4] ()
}

-- Gets a column number
public fun getCol (m) {
  m [5] ()
}

-- Creates a fresh matcher from a string buffer
public fun initMatcher (buf) {
  createMatcher (buf, 0, 1, 1)
}