main_testing5/bfc.scala
changeset 404 bf20a9fa5c29
parent 384 6e1237691307
child 463 0315d9983cd0
--- a/main_testing5/bfc.scala	Mon Nov 08 01:39:00 2021 +0000
+++ b/main_testing5/bfc.scala	Mon Nov 08 01:49:28 2021 +0000
@@ -1,9 +1,7 @@
-// Core Part about a "Compiler" for the Brainf*** language
+// Part 2 about a "Compiler" for the Brainf*** language
 //======================================================
 
-
-object CW10b {
-
+object M5b {
 
 // !!! Copy any function you need from file bf.scala !!!
 //
@@ -11,61 +9,7 @@
 // implement it, but do not make any changes to the
 // templates below.
 
-type Mem = Map[Int, Int]
 
-import io.Source
-import scala.util._
-
-def load_bff(name: String) : String = 
-  Try(scala.io.Source.fromFile(name)("ISO-8859-1").mkString).getOrElse("")
-
-def sread(mem: Mem, mp: Int) : Int = mem.getOrElse(mp, 0)
-
-def write(mem: Mem, mp: Int, v: Int) : Mem = mem + (mp -> v)
-
-def jumpRight(prog: String, pc: Int, level: Int) : Int = {
-    pc match {
-        case pc: Int if (pc >= 0 && pc < prog.length) => {
-            prog(pc) match {
-                case '[' => jumpRight(prog, pc + 1, level + 1)
-                case ']' => if (level == 0) pc + 1 else jumpRight(prog, pc + 1, level - 1)
-                case _ => jumpRight(prog, pc + 1, level)
-            }
-        }
-        case _ => pc
-    }
-}
-
-def jumpLeft(prog: String, pc: Int, level: Int) : Int = {
-    pc match {
-        case pc: Int if (pc >= 0 && pc < prog.length) => {
-            prog(pc) match {
-                case '[' => if (level == 0) pc + 1 else jumpLeft(prog, pc - 1, level - 1)
-                case ']' => jumpLeft(prog, pc - 1, level + 1)
-                case _ => jumpLeft(prog, pc - 1, level)
-            }
-        }
-        case _ => pc
-    }
-}
-
-def get_position(prog: String, pc: Int, level: Int) : Int = {
-  prog(pc) match {
-    case '[' => jumpRight(prog, pc + 1, level)
-    case ']' => jumpLeft(prog, pc - 1, level)
-    case _ => println("Something went horrible wrong, I am sorry"); 0
-  }
-}
-
-// DEBUGGING INFORMATION FOR COMPILERS!!!
-//
-// Compiler, even real ones, are fiendishly difficult to get
-// to produce correct code. One way to debug them is to run
-// example programs ``unoptimised''; and then optimised. Does
-// the optimised version still produce the same result?
-
-
-// for timing purposes
 def time_needed[T](n: Int, code: => T) = {
   val start = System.nanoTime()
   for (i <- 0 until n) code
@@ -73,17 +17,88 @@
   (end - start)/(n * 1.0e9)
 }
 
+type Mem = Map[Int, Int]
 
 
-// TASKS
-//=======
+import io.Source
+import scala.util._
+
+def load_bff(name: String) : String = 
+  Try(Source.fromFile(name)("ISO-8859-1").mkString).getOrElse("")
+
+def sread(mem: Mem, mp: Int) : Int = 
+  mem.getOrElse(mp, 0)
+
+def write(mem: Mem, mp: Int, v: Int) : Mem =
+  mem.updated(mp, v)
+
+def jumpRight(prog: String, pc: Int, level: Int) : Int = {
+  if (prog.length <= pc) pc 
+  else (prog(pc), level) match {
+    case (']', 0) => pc + 1
+    case (']', l) => jumpRight(prog, pc + 1, l - 1)
+    case ('[', l) => jumpRight(prog, pc + 1, l + 1)
+    case (_, l) => jumpRight(prog, pc + 1, l)
+  }
+}
+
+def jumpLeft(prog: String, pc: Int, level: Int) : Int = {
+  if (pc < 0) pc 
+  else (prog(pc), level) match {
+    case ('[', 0) => pc + 1
+    case ('[', l) => jumpLeft(prog, pc - 1, l - 1)
+    case (']', l) => jumpLeft(prog, pc - 1, l + 1)
+    case (_, l) => jumpLeft(prog, pc - 1, l)
+  }
+}
+
+def compute(prog: String, pc: Int, mp: Int, mem: Mem) : Mem = {
+  if (0 <= pc && pc < prog.length) { 
+    val (new_pc, new_mp, new_mem) = prog(pc) match {
+      case '>' => (pc + 1, mp + 1, mem)
+      case '<' => (pc + 1, mp - 1, mem)
+      case '+' => (pc + 1, mp, write(mem, mp, sread(mem, mp) + 1))
+      case '-' => (pc + 1, mp, write(mem, mp, sread(mem, mp) - 1))
+      case '.' => { print(sread(mem, mp).toChar); (pc + 1, mp, mem) }
+      case '['  => 
+	if (sread(mem, mp) == 0) (jumpRight(prog, pc + 1, 0), mp, mem) else (pc + 1, mp, mem) 
+      case ']'  => 
+	if (sread(mem, mp) != 0) (jumpLeft(prog, pc - 1, 0), mp, mem) else (pc + 1, mp, mem) 
+      case _ => (pc + 1, mp, mem)
+    }		     
+    compute(prog, new_pc, new_mp, new_mem)	
+  }
+  else mem
+}
+
+def run(prog: String, m: Mem = Map()) = compute(prog, 0, 0, m)
+
+
+// The baseline to what we can compare our "compiler"
+// implemented below. It should require something like 
+// 60 seconds for the calculation on my laptop
+//
+//time_needed(1, run(load_bff("benchmark.bf")))
+
+
+
+// DEBUGGING INFORMATION!!!
+//
+// Compiler, even real ones, are fiedishly difficult to get
+// to prduce correct code. The point is that for example for
+// the sierpinski program, they need to still generate code
+// that displays such a triangle. If yes, then one usually
+// can take comfort that all is well. If not, then something
+// went wrong during the optimisations.
+
+
 
 // (5) Write a function jtable that precomputes the "jump
 //     table" for a bf-program. This function takes a bf-program 
 //     as an argument and Returns a Map[Int, Int]. The 
-//     purpose of this map is to record the information about
-//     pc positions where '[' or a ']' are stored. The information
-//     is to which pc-position do we need to jump next?
+//     purpose of this map is to record the information
+//     that given on the position pc is a '[' or a ']',
+//     then to which pc-position do we need to jump next?
 // 
 //     For example for the program
 //    
@@ -102,54 +117,48 @@
 //     jtable. You can use the jumpLeft and jumpRight functions
 //     from Part 1 for calculating the jtable.
 //
-//     Then adapt the compute and run functions from Part 1 
-//     in order to take advantage of the information stored in the jtable. 
+//     Then adapt the compute and run functions from Part 1 in order 
+//     to take advantage of the information stored in the jtable. 
 //     This means whenever jumpLeft and jumpRight was called previously,
-//     you should immediately look up the jump address in the jtable.
-//  for ((char, index) <- str.zipWithIndex if (List('[', ']').contains(char))) yield (index, get_position(str, index, 0))
+//     you should look up the jump address in the jtable.
  
 
-def jtable(pg: String) : Map[Int, Int] = {
-  val table = for ((char, index) <- pg.zipWithIndex if (List('[', ']').contains(char))) yield (index, get_position(pg, index, 0))
-  table.toMap
-}
+def jtable(pg: String) : Map[Int, Int] = 
+    (0 until pg.length).collect { pc => pg(pc) match {
+      case '[' => (pc -> jumpRight(pg, pc + 1, 0))
+      case ']' => (pc -> jumpLeft(pg, pc - 1, 0))
+    }}.toMap
 
 
 // testcase
-//
 // jtable("""+++++[->++++++++++<]>--<+++[->>++++++++++<<]>>++<<----------[+>.>.<+<]""")
 // =>  Map(69 -> 61, 5 -> 20, 60 -> 70, 27 -> 44, 43 -> 28, 19 -> 6)
 
 
 def compute2(pg: String, tb: Map[Int, Int], pc: Int, mp: Int, mem: Mem) : Mem = {
-  pc match {
-    case pc: Int if (pc >= 0 && pc < pg.length) => {
-      pg(pc) match {
-        case '>' => compute2(pg, tb, pc + 1, mp + 1, mem)
-        case '<' => compute2(pg, tb, pc + 1, mp - 1, mem)
-        case '+' => compute2(pg, tb, pc + 1, mp, write(mem, mp, sread(mem, mp) + 1))
-        case '-' => compute2(pg, tb, pc + 1, mp, write(mem, mp, sread(mem, mp) - 1))
-        case '.' => print(sread(mem, mp).toChar); compute2(pg, tb, pc + 1, mp, mem)
-        case '[' => if (sread(mem, mp) == 0) compute2(pg, tb, tb(pc), mp, mem) else compute2(pg, tb, pc + 1, mp, mem)
-        case ']' => if (sread(mem, mp) != 0) compute2(pg, tb, tb(pc), mp, mem) else compute2(pg, tb, pc + 1, mp, mem)
-        case '*' => compute2(pg, tb, pc + 1, mp, write(mem, mp, sread(mem, mp) * sread(mem, mp - 1)))
-        case '@' => compute2(pg, tb, pc + 1, mp, write(mem, mem(mp), sread(mem, mp - 1)))
-        case '#' => print(sread(mem, mp)); compute2(pg, tb, pc + 1, mp, mem)
-        case _ => compute2(pg, tb, pc + 1, mp, mem)
-      }
-    }
-    case _ => mem
+  if (0 <= pc && pc < pg.length) { 
+    val (new_pc, new_mp, new_mem) = pg(pc) match {
+      case '>' => (pc + 1, mp + 1, mem)
+      case '<' => (pc + 1, mp - 1, mem)
+      case '+' => (pc + 1, mp, write(mem, mp, sread(mem, mp) + 1))
+      case '-' => (pc + 1, mp, write(mem, mp, sread(mem, mp) - 1))
+      case '.' => { print(sread(mem, mp).toChar); (pc + 1, mp, mem) }
+       case '['  => 
+	if (sread(mem, mp) == 0) (tb(pc), mp, mem) else (pc + 1, mp, mem) 
+      case ']'  => 
+	if (sread(mem, mp) != 0) (tb(pc), mp, mem) else (pc + 1, mp, mem) 
+      case _ => (pc + 1, mp, mem)
+    }		     
+    compute2(pg, tb, new_pc, new_mp, new_mem)	
   }
-}
-
-def run2(pg: String, m: Mem = Map()) = {
-  compute2(pg, jtable(pg), 0, 0, m)
+  else mem
 }
 
 
-// testcases
-// time_needed(1, run2(load_bff("./main5/benchmark.bf")))
-// time_needed(1, run2(load_bff("./main5/sierpinski.bf")))
+def run2(pg: String, m: Mem = Map()) = 
+  compute2(pg, jtable(pg), 0, 0, m)
+
+//time_needed(1, run2(load_bff("benchmark.bf")))
 
 
 
@@ -161,63 +170,56 @@
 // that is write(mem, mp, 0). 
 //
 // The easiest way to modify a string in this way is to use the regular
-// expression """[^<>+-.,\[\]]""", which recognises everything that is 
+// expression """[^<>+-.\[\]""", which recognises everything that is 
 // not a bf-command and replace it by the empty string. Similarly the
-// regular expression """\[-\]""" finds all occurrences of [-] and 
-// by using the Scala method .replaceAll you can replace it with the 
+// regular expression """\[-\]""" finds all occurences of [-] and 
+// by using the Scala method .replaceAll you can repplace it with the 
 // string "0" standing for the new bf-command.
-// load_bff("./main5/mandelbrot.bf").replaceAll("""[^<>+‐.\[\]@#*]""", "").replaceAll("""\[-\]""", "0")
-
-// "Correct" regex
-// s.replaceAll("""[^<>+‐.\[\]@#*]""", "").replaceAll("""\[-\]""", "0")
-// s.replaceAll("""[^<>+-.,\[\]]""", "").replaceAll("""\[-\]""", "0")
 
 def optimise(s: String) : String = {
-  //s.replaceAll("""[^<>+-.\[\]@#*]""","")
-  // .replaceAll("""\[-\]""", "0")
-  s.replaceAll("""[^<>+-.\[\]]""", "").replaceAll("""\[-\]""", "0")
+  s.replaceAll("""[^<>+-.\[\]]""","")
+   .replaceAll("""\[-\]""", "0")
 }
 
+
 def compute3(pg: String, tb: Map[Int, Int], pc: Int, mp: Int, mem: Mem) : Mem = {
-  pc match {
-    case pc: Int if (pc >= 0 && pc < pg.length) => {
-      pg(pc) match {
-        case '>' => compute3(pg, tb, pc + 1, mp + 1, mem)
-        case '<' => compute3(pg, tb, pc + 1, mp - 1, mem)
-        case '+' => compute3(pg, tb, pc + 1, mp, write(mem, mp, sread(mem, mp) + 1))
-        case '-' => compute3(pg, tb, pc + 1, mp, write(mem, mp, sread(mem, mp) - 1))
-        case '.' => print(sread(mem, mp).toChar); compute3(pg, tb, pc + 1, mp, mem)
-        case '[' => if (sread(mem, mp) == 0) compute3(pg, tb, tb(pc), mp, mem) else compute3(pg, tb, pc + 1, mp, mem)
-        case ']' => if (sread(mem, mp) != 0) compute3(pg, tb, tb(pc), mp, mem) else compute3(pg, tb, pc + 1, mp, mem)
-        case '*' => compute3(pg, tb, pc + 1, mp, write(mem, mp, sread(mem, mp) * sread(mem, mp - 1)))
-        case '@' => compute3(pg, tb, pc + 1, mp, write(mem, mem(mp), sread(mem, mp - 1)))
-        case '#' => print(sread(mem, mp)); compute3(pg, tb, pc + 1, mp, mem)
-        case '0' => compute3(pg, tb, pc + 1, mp, write(mem, mp, 0))
-        case _ => compute3(pg, tb, pc + 1, mp, mem)
-      }
-    }
-    case _ => mem
+  if (0 <= pc && pc < pg.length) { 
+    val (new_pc, new_mp, new_mem) = pg(pc) match {
+      case '0' => (pc + 1, mp, write(mem, mp, 0))
+      case '>' => (pc + 1, mp + 1, mem)
+      case '<' => (pc + 1, mp - 1, mem)
+      case '+' => (pc + 1, mp, write(mem, mp, sread(mem, mp) + 1))
+      case '-' => (pc + 1, mp, write(mem, mp, sread(mem, mp) - 1))
+      case '.' => { print(sread(mem, mp).toChar); (pc + 1, mp, mem) }
+      case '['  => 
+	if (sread(mem, mp) == 0) (tb(pc), mp, mem) else (pc + 1, mp, mem) 
+      case ']'  => 
+	if (sread(mem, mp) != 0) (tb(pc), mp, mem) else (pc + 1, mp, mem) 
+      case _ => (pc + 1, mp, mem)
+    }		     
+    compute3(pg, tb, new_pc, new_mp, new_mem)	
   }
+  else mem
 }
 
-def run3(pg: String, m: Mem = Map()) = {
-  val optimised = optimise(pg)
-  compute3(optimised, jtable(optimised), 0, 0, m)
+def run3(pg: String, m: Mem = Map()) = { 
+  val pg_opt = optimise(pg)
+  compute3(pg_opt, jtable(pg_opt), 0, 0, m)
 }
 
 
 // testcases
-//
-// optimise(load_bff("./main5/benchmark.bf"))          // should have inserted 0's
-// optimise(load_bff("./main5/mandelbrot.bf")).length  // => 11205
-// 
-// time_needed(1, run3(load_bff("./main5/benchmark.bf")))
-// time_needed(1, run3(load_bff("./main5/mandelbrot.bf")))
+
+//println(optimise(load_bff("collatz.bf")))
+//optimise(load_bff("benchmark.bf"))          // should have inserted 0's
+//optimise(load_bff("mandelbrot.bf")).length  // => 11203
+ 
+//time_needed(1, run3(load_bff("benchmark.bf")))
 
 
 
 // (7)  Write a function combine which replaces sequences
-// of repeated increment and decrement commands by appropriate
+// of repated increment and decrement commands by appropriate
 // two-character commands. For example for sequences of +
 //
 //              orig bf-cmds  | replacement
@@ -237,82 +239,82 @@
 //  Adapt the compute4 and run4 functions such that they can deal
 //  appropriately with such two-character commands.
 
-// val alphabet = "АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯ"
-val alphabet = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
-
-// Try any alphabet, it will work as long as the character is recognised and the characters are unique
-
-def get_number_from_character(char: Char) : Int = {
-  alphabet.indexOf(char) + 1
+def splice(cs: List[Char], acc: List[(Char, Int)]) : List[(Char, Int)] = (cs, acc) match {
+  case (Nil, acc) => acc  
+  case ('[' :: cs, acc) => splice(cs, ('[', 1) :: acc)
+  case (']' :: cs, acc) => splice(cs, (']', 1) :: acc)
+  case ('.' :: cs, acc) => splice(cs, ('.', 1) :: acc)
+  case ('0' :: cs, acc) => splice(cs, ('0', 1) :: acc)
+  case (c :: cs, Nil) => splice(cs, List((c, 1)))
+  case (c :: cs, (d, n) :: acc) => 
+    if (c == d && n < 26) splice(cs, (c, n + 1) :: acc)
+    else splice(cs, (c, 1) :: (d, n) :: acc)
 }
 
-def get_character_from_number(int: Int) : Char = {
-  alphabet(int - 1)
-}
+def spl(s: String) = splice(s.toList, Nil).reverse
 
-@annotation.tailrec 
-def split_by_repetition(string : String, list : List[String] = Nil) : List[String] = {
-    if(string.size == 0) list.reverse 
-    else {
-        val (left_substring, right_substring) = string.span(_ == string(0))
-        split_by_repetition(right_substring, left_substring :: list)
-    }
-}
+//spl(load_bff("benchmark.bf"))
 
 def combine(s: String) : String = {
-  val split_strings = split_by_repetition(s)
-  val lists = for (string <- split_strings) yield {
-    if (List("+"(0), "-"(0), "<"(0), ">"(0)).contains(string.head)) {
-      val long_repeat = s"${string.head}${alphabet.last}" * (string.size / alphabet.length)
-      val short_repeat = if ((string.size % alphabet.length) != 0) s"${string.head}${get_character_from_number(string.size % alphabet.length)}" else ""
-      long_repeat + short_repeat
-    } else string
-  }
-  lists.mkString("")
-}
-
-// testcase
-// combine(load_bff("./main5/benchmark.bf"))
-
-
-def compute4(pg: String, tb: Map[Int, Int], pc: Int, mp: Int, mem: Mem) : Mem = {
-  pc match {
-    case pc: Int if (pc >= 0 && pc < pg.length) => {
-      pg(pc) match {
-        case '>' => val number = get_number_from_character(pg(pc + 1)); compute4(pg, tb, pc + 2, mp + number, mem)
-        case '<' => val number = get_number_from_character(pg(pc + 1)); compute4(pg, tb, pc + 2, mp - number, mem)
-        case '+' => val number = get_number_from_character(pg(pc + 1)); compute4(pg, tb, pc + 2, mp, write(mem, mp, sread(mem, mp) + number))
-        case '-' => val number = get_number_from_character(pg(pc + 1)); compute4(pg, tb, pc + 2, mp, write(mem, mp, sread(mem, mp) - number))
-        case '.' => print(sread(mem, mp).toChar); compute4(pg, tb, pc + 1, mp, mem)
-        case '[' => if (sread(mem, mp) == 0) compute4(pg, tb, tb(pc), mp, mem) else compute4(pg, tb, pc + 1, mp, mem)
-        case ']' => if (sread(mem, mp) != 0) compute4(pg, tb, tb(pc), mp, mem) else compute4(pg, tb, pc + 1, mp, mem)
-        case '*' => compute4(pg, tb, pc + 1, mp, write(mem, mp, sread(mem, mp) * sread(mem, mp - 1)))
-        case '@' => compute4(pg, tb, pc + 1, mp, write(mem, mem(mp), sread(mem, mp - 1)))
-        case '#' => print(sread(mem, mp)); compute4(pg, tb, pc + 1, mp, mem)
-        case '0' => compute4(pg, tb, pc + 1, mp, write(mem, mp, 0))
-        case _ => compute4(pg, tb, pc + 1, mp, mem)
-      }
-    }
-    case _ => mem
-  }
+  (for ((c, n) <- spl(s)) yield c match {
+    case '>' => List('>', (n + '@').toChar)
+    case '<' => List('<', (n + '@').toChar)
+    case '+' => List('+', (n + '@').toChar)
+    case '-' => List('-', (n + '@').toChar)
+    case _ => List(c)
+  }).flatten.mkString
 }
 
 
-// should call first optimise and then combine on the input string
-//
-def run4(pg: String, m: Mem = Map()) = {
-  val processed_prog = combine(optimise(pg))
-  compute4(processed_prog, jtable(processed_prog), 0, 0, m)
+//combine(load_bff("benchmark.bf"))
+
+def compute4(pg: String, tb: Map[Int, Int], pc: Int, mp: Int, mem: Mem) : Mem = {
+  if (0 <= pc && pc < pg.length) { 
+    val (new_pc, new_mp, new_mem) = pg(pc) match {
+      case '0' => (pc + 1, mp, write(mem, mp, 0))
+      case '>' => (pc + 2, mp + (pg(pc + 1) - '@'), mem)
+      case '<' => (pc + 2, mp - (pg(pc + 1) - '@'), mem)
+      case '+' => (pc + 2, mp, write(mem, mp, sread(mem, mp) + (pg(pc + 1) - '@')))
+      case '-' => (pc + 2, mp, write(mem, mp, sread(mem, mp) - (pg(pc + 1) - '@')))
+      case '.' => { print(sread(mem, mp).toChar); (pc + 1, mp, mem) }
+       case '['  => 
+	if (sread(mem, mp) == 0) (tb(pc), mp, mem) else (pc + 1, mp, mem) 
+      case ']'  => 
+	if (sread(mem, mp) != 0) (tb(pc), mp, mem) else (pc + 1, mp, mem) 
+      case _ => (pc + 1, mp, mem)
+    }		     
+    compute4(pg, tb, new_pc, new_mp, new_mem)	
+  }
+  else mem
 }
 
+def run4(pg: String, m: Mem = Map()) = { 
+  val pg_opt = combine(optimise(pg))
+  compute4(pg_opt, jtable(pg_opt), 0, 0, m)
+}
+
+// testcases
+//println(combine(optimise(load_bff("mandelbrot.bf").drop(123))))
+
+//combine(optimise(load_bff("benchmark.bf"))) // => """>A+B[<A+M>A-A]<A[[....."""
+
+//time_needed(1, run4(load_bff("benchmark.bf")))
+
+//time_needed(1, run(load_bff("sierpinski.bf"))) 
+//time_needed(1, run4(load_bff("sierpinski.bf"))) 
+
+//println(time_needed(1, run4(load_bff("mandelbrot.bf"))))
+
 
-// testcases
-// combine(optimise(load_bff("./main5/benchmark.bf"))) // => """>A+B[<A+M>A-A]<A[[....."""
 
-// testcases (they should now run much faster)
-// time_needed(1, run4(load_bff("./main5/benchmark.bf")))
-// time_needed(1, run4(load_bff("./main5/sierpinski.bf"))) 
-// time_needed(1, run4(load_bff("./main5/mandelbrot.bf")))
 
 
 }
+
+/*
+import CW10b._
+println(time_needed(1, run(load_bff("collatz.bf"))))
+println(time_needed(1, run2(load_bff("collatz.bf"))))
+println(time_needed(1, run3(load_bff("collatz.bf"))))
+println(time_needed(1, run4(load_bff("collatz.bf"))))
+*/