Initial commit

1 files changed, 122 insertions, 0 deletions

diff --git a/markov.lpr b/markov.lpr
new file mode 100644
index 0000000..a9e7507
--- /dev/null
+++ b/markov.lpr
@@ -0,0 +1,122 @@
+program markov;
+
+{$mode objfpc}{$H+}
+
+uses
+  {$IFDEF UNIX}{$IFDEF UseCThreads}
+  cthreads,
+  {$ENDIF}{$ENDIF}
+  Classes
+  { you can add units after this },
+  sysutils, math;
+
+var
+  wordFileName,s:    string;
+  depth,i,j,current: int64;
+  c:                 char;
+  f:                 textFile;
+  probabilities:     array of extended;
+  total:             extended;
+  c2i:               array[char] of longint;
+  i2c:               string;
+
+begin
+  randomize;
+  if paramCount<>2 then begin
+    writeln(stderr,'usage: markov word-file chain-depth');
+    halt(1);
+  end;
+  wordFileName:=paramStr(1);
+  if not fileExists(wordFileName) then begin
+    writeln(stderr,'file '''+wordFileName+''' does not exist');
+    halt(1);
+  end;
+  try
+    depth:=strToInt(paramstr(2));
+  except
+    writeln(stderr,''''+paramStr(2)+''' is not a valid integer');
+    halt(1);
+  end;
+  if depth<=0 then begin
+    writeln(stderr,intToStr(depth)+' is not positive');
+    halt(1);
+  end;
+
+  i2c:='abcdefghijklmnopqrstuvwxyzäöüßABCDEFGHIJKLMNOPQRSTUVWXYZÄÖÜ';
+  for i:=length(i2c)-1 downto 1 do
+    for j:=length(i2c) downto i+1 do
+      if i2c[i]=i2c[j] then
+        delete(i2c,j,1);
+  for c:=#0 to #255 do
+    c2i[c]:=-1;
+  for i:=1 to length(i2c) do
+    c2i[i2c[i]]:=i;
+
+  setLength(probabilities,round(power(length(i2c)+1,depth)));
+  for i:=0 to length(probabilities)-1 do
+    probabilities[i]:=0;
+
+  assignFile(f,wordFileName);
+  reset(f);
+  while not eof(f) do begin
+    readln(f,s);
+    current:=0; // reset current index
+    for i:=1 to length(s) do begin
+      current:=current*(length(i2c)+1) mod length(probabilities);
+      if c2i[s[i]]<0 then begin
+        if current<>0 then // terminate word
+          probabilities[current]:=probabilities[current]+1;
+        current:=0;
+        continue;
+      end;
+      current:=current + c2i[s[i]];
+      probabilities[current]:=probabilities[current]+1;
+    end;
+  end;
+  closeFile(f);
+
+  for i:=0 to length(probabilities) div (length(i2c)+1)-1 do begin
+    total:=0;
+    for j:=0 to length(i2c) do
+      total:=total + probabilities[j + i * (length(i2c)+1)];
+    if total=0 then
+      continue;
+    total:=1/total;
+    for j:=0 to length(i2c) do
+      probabilities[j + i * (length(i2c)+1)]:= probabilities[j + i * (length(i2c)+1)] * total;
+  end;
+
+(*
+  // debug output
+  for current:=0 to length(probabilities)-1 do begin
+    j:=current;
+    for i:=1 to depth do begin
+      if (j mod (length(i2c)+1)) = 0 then
+        write('_')
+      else
+        write(i2c[j mod (length(i2c)+1)]);
+      j:=j div (length(i2c)+1);
+    end;
+    writeln(' ',probabilities[current]);
+  end; *)
+
+  // generation
+  current:=0;
+  repeat
+    current:=current * (length(i2c)+1) mod length(probabilities);
+    total:=random;
+    while total>0 do begin
+      total:=total - probabilities[current];
+      inc(current);
+    end;
+    dec(current);
+    if current mod (length(i2c)+1) = 0 then
+      break
+    else
+      write(i2c[current mod (length(i2c)+1)]);
+  until false;
+  writeln;
+
+  setLength(probabilities,0);
+end.
+