unit mystringlistunit;

{$mode objfpc}{$H+}

interface

uses
  Classes, SysUtils, RegExpr, Process, Math;

type
  tMyStringlist = class;

  tInputThread = class (tThread)
    fertig: boolean;
    inhalt: tMyStringList;
    proc:   tProcess;
    constructor create(p: tProcess; sl: tMyStringList);
    destructor destroy; override;
    procedure execute; override;
  end;

  tMyStringlist = class (tStringlist)
  private
    line:  longint;
  public
    constructor create;
    procedure loadFromFile(const s: ansiString); override;
    procedure loadFromGz(const s: ansiString);
    procedure saveToGz(const s: ansiString);
    function readln(out s: string): boolean;
    procedure grep(expr: string);
    function eof: boolean;
  end;

implementation

// tInputThread ****************************************************************

constructor tInputThread.create(p: tProcess; sl: tMyStringList);
begin
 inherited create(true);
 fertig:=false;
 inhalt:=sl;
 proc:=p;
 suspended:=false;
end;

destructor tInputThread.destroy;
begin
 inhalt:=nil;
 proc:=nil;
 inherited destroy;
end;

procedure tInputThread.execute;
var
  wb,cwb: longint;
begin
 wb:=0;
 while wb<length(inhalt.text) do begin
   cwb:=proc.input.write(inhalt.text[wb+1],length(inhalt.text)-wb);
   if cwb=0 then
     sleep(1)
   else
     wb:=wb+cwb;
 end;
 proc.CloseInput;
 fertig:=true;
end;

// tMyStringlist ***************************************************************

constructor tMyStringlist.create;
begin
 inherited create;
 line:=0;
end;

procedure tMyStringlist.loadFromFile(const s: ansiString);
var i: longint;
begin
 inherited loadFromFile(s);
 for i:=0 to count-1 do
   self[i]:=trim(self[i]);
 line:=0;
 writeln(inttostr(count)+' Zeilen eingelesen');
end;

procedure tMyStringlist.loadFromGz(const s: ansiString);
var p:     tProcess;
    buf:   ansiString;
    rb,br: longint;
begin
 p:=tProcess.create(nil);
 p.executable:='/usr/bin/zcat';
 p.parameters.add(s);
 p.options:=p.options + [poUsePipes];
 setlength(buf,0);
 br:=0;
 p.execute;
 while p.running do begin
   rb:=p.output.numBytesAvailable;
   if rb>0 then begin
     if length(buf)<br+rb then
       setlength(buf,br+rb+1048576);
     rb:=p.output.read(buf[br+1],rb);
     br:=br+rb;
   end
   else sleep(1);
 end;
 setlength(buf,br);
 rb:=p.output.numBytesAvailable;
 while rb>0 do begin
   setlength(buf,br+rb);
   rb:=p.output.read(buf[br+1],rb);
   br:=br+rb;
   rb:=p.output.numBytesAvailable;
 end;
 text:=buf;
 setlength(buf,0);
 p.free;
 for rb:=0 to count-1 do
   self[rb]:=trim(self[rb]);
 line:=0;
 writeln(inttostr(count)+' Zeilen eingelesen');
end;

procedure tMyStringlist.saveToGz(const s: ansiString);
var
  p:     tProcess;
  buf:   array of byte;
  f:     file;
  rb:    longint;
  it:    tInputThread;
const
  outBufLen = 1024*1024;
begin
 p:=tProcess.create(nil);
 p.executable:='/usr/bin/gzip';
 p.parameters.add('--best');
 p.parameters.add('-c');
 p.options:=p.options + [poUsePipes];
 p.execute;
 setlength(buf,outBufLen);
 fillchar(buf[0],length(buf)*sizeof(buf[0]),$0);
 it:=tInputThread.create(p,self);
 assignfile(f,s);
 rewrite(f,1);
 while p.running or (not it.fertig) or (p.output.numBytesAvailable>0) do begin
   rb:=min(length(buf),p.output.numBytesAvailable);
   if rb>0 then begin
     rb:=p.output.read(buf[0],rb);
     blockwrite(f,buf[0],rb);
   end
   else
     sleep(1); // nix zu Schreiben, nix zu Lesen, also warten wir
 end;
 it.free;
 closefile(f);
end;

function tMyStringlist.readln(out s: string): boolean;
begin
 result:=not eof;
 if not result then begin
   s:='';
   exit;
 end;
 s:=self[line];
 inc(line);
end;

procedure tMyStringlist.grep(expr: string);
var
  re: tRegExpr;
  i:  longint;
begin
  re:=tRegExpr.create;
  re.Expression:=expr;
  for i:=count-1 downto 0 do
    if not re.Exec(self[i]) then
      delete(i);
  re.free;
end;

function tMyStringlist.eof: boolean;
begin
 result:=line>=count;
end;

end.