How to jump to a specific line in a text file and return the line in a string

Mirror

How to jump to a specific line in a text file and return the line in a string (Views: 705)

Problem/Question/Abstract:

I'm trying to write a function that, given a FileName and a line number, returns the entire line in a string.

Answer:

This technique is useful for high-speed processing. Save the sample program file with a .pas or .dpr file name and compile it with dcc32:

{$APPTYPE CONSOLE}
uses
  SysUtils, Classes;

function GrabLine(const AFileName: string; ALine: Integer): string;
var
  fs: TFileStream;
  buf: packed array[0..4095] of Char;
  bufRead: Integer;
  bufPos: PChar;
  lineStart: PChar;
  tmp: string;
begin
  fs := TFileStream.Create(AFileName, fmOpenRead);
  try
    Dec(ALine);
    bufRead := 0;
    bufPos := nil;
    { read the first line specially }
    if ALine = 0 then
    begin
      bufRead := fs.Read(buf, SizeOf(buf));
      if bufRead = 0 then
        raise Exception.Create('Line not found');
      bufPos := buf;
    end
    else
      while ALine > 0 do
      begin
        { read in a buffer }
        bufRead := fs.Read(buf, SizeOf(buf));
        if bufRead = 0 then
          raise Exception.Create('Line not found');
        bufPos := buf;
        while (bufRead > 0) and (ALine > 0) do
        begin
          if bufPos^ = #10 then
            Dec(ALine);
          Inc(bufPos);
          Dec(bufRead);
        end;
      end;
    { Found the beginning of the line at bufPos... scan for end.
      Two cases:
        1) we'll find it before the end of this buffer
        2) it'll go beyond this buffer and into n more buffers }
    lineStart := bufPos;
    while (bufRead > 0) and (bufPos^ <> #10) do
    begin
      Inc(bufPos);
      Dec(bufRead);
    end;
    { if bufRead is positive, we'll have found the end and we can leave. }
    SetString(Result, lineStart, bufPos - lineStart);
    { determine if there are more buffers to process }
    while bufRead = 0 do
    begin
      bufRead := fs.Read(buf, SizeOf(buf));
      lineStart := buf;
      bufPos := buf;
      while (bufRead > 0) and (bufPos^ <> #10) do
      begin
        Inc(bufPos);
        Dec(bufRead);
      end;
      SetString(tmp, lineStart, bufPos - lineStart);
      Result := Result + tmp;
    end;
  finally
    fs.Free;
  end;
end;

function GrabLine2(const s: string; ALine: Integer): string;
var
  sl: TStringList;
begin
  sl := TStringList.Create;
  try
    sl.LoadFromFile(s);
    Result := sl[ALine - 1]; { index off by one }
  finally
    sl.Free;
  end;
end;

begin
  Writeln(GrabLine(ParamStr(1), StrToInt(ParamStr(2))));
  Writeln(GrabLine2(ParamStr(1), StrToInt(ParamStr(2))));
end.

Call it like 'getline testfile.txt 20000', depending on what you call the .pas (or .dpr) file. For large (i.e. tens of megabytes) files, the (rather complex) scanning function easily beats the memory expensive StringList version.

<< Back to main page