program GoogleProductsSearch;
// Sample script to search and extract product
// information from Google Products             

const  
  MaxRecords  = 5000; 
  cPrefix     = 'allintitle:';
  cTablePos    = 5;
  
  cInputFile  = 'C:\InputFile.csv';
  cOutputFile = 'C:\OutputFile.csv';  
    
type
  ProductRecord = record    
    sProductName    : string;
    sProductUrl     : string;
    sRetailerName   : string;
    sRetailerUrl    : string;    
    sPrice          : string;
  end;

var 
  I, nProductIndex, nCount : integer;
  ProductArray : array[1..MaxRecords] of ProductRecord;

function ProcessDblQuote(str: string): string;
var
  bFlag : boolean;
begin
  result := str;
  bFlag := False;
  if (Pos('"', str) > 0) then
    begin
      str := StringReplace(str, '"', '""', true, true);
      bFlag := True;
    end;
  if (Pos(#13, str) > 0) or (Pos(',', str) > 0) then
    bflag := True;
  if (bFlag) then               // has double quote or #13
    result := '"' + str + '"';
end;      

function GetProductUrl(str : string): string;
var
  sTmp : string;
  nPos : integer;
begin
  result := '';  
  sTmp := Trim(ExtractSubStr(str, 'href="/product_url?q=', '">'));
  if (sTmp <> '') then
    begin
      sTmp := StringReplace(sTmp, '%3F', '?', True, True); 
      sTmp := StringReplace(sTmp, '%3D', '=', True, True);
      sTmp := StringReplace(sTmp, '%26', '&', True, True);
      nPos := Pos('&amp;', sTmp);
      if (nPos > 0) then
        begin
          sTmp := Trim(Copy(sTmp, 1, nPos-1));
        end;
      result := sTmp; 
    end;  
end;

function GetRetailerName(nIdx : integer): string;
var
  sTmp : string;
  nPos : integer;
begin   
  sTmp := GetTableCell(cTablePos, nIdx, 2);  
  nPos := Pos(#13, sTmp);
  result := sTmp; 
  if (nPos > 0) then
    begin
      result := Copy(sTmp, 1, nPos-1);  
    end;
end;

function GetRetailerUrl(str: string): string;
var
  nPos : integer;
begin
  result := str;
  nPos := Pos('.COM', UpperCase(str));
  if (nPos > 0) then
    begin
      result := Copy(str, 1, nPos+4);
    end; 
end;

procedure GetProductDetails(str : string; nIdx : integer);
var
  sTmp : string;
begin
  nCount := nCount + 1;
  ProductArray[nCount].sProductName := ProcessDblQuote(GetTableCell(cTablePos, nIdx, 1));
  sTmp := GetProductUrl(str);
  ProductArray[nCount].sProductUrl := sTmp;   
  ProductArray[nCount].sRetailerName := GetRetailerName(nIdx);
  ProductArray[nCount].sRetailerUrl := GetRetailerUrl(ProductArray[nCount].sProductUrl);
  ProductArray[nCount].sPrice := ProcessDblQuote(GetTableCell(cTablePos, nIdx, 3));
          
  
  WriteToLog(cOutputFile, '', MatrixGetVal(nProductIndex, 1) + ',' + ProductArray[nCount].sProductName + ',' + 
                              ProductArray[nCount].sProductUrl + ',' + ProductArray[nCount].sRetailerName + ',' + 
                              ProductArray[nCount].sRetailerUrl + ',' + ProductArray[nCount].sPrice);  
end;

procedure OnDocumentComplete(URL : string);
begin 
  if IsPartOf('did not match any products', GetHTMLBody) then
    begin
      WriteToLog(cOutputFile, '', MatrixGetVal(nProductIndex, 1) +',not found');  
      GotoURL('http://froogle.google.com/');
    end
  else if IsPartOf('http://www.google.com/products?q=', URL) then
    begin      
      for I := 1 to 10 do
        begin
          if (Trim(GetTableCell(cTablePos, I, 1)) = '') then
            begin
              break;
            end;        
          GetProductDetails(GetTableCellUrls(cTablePos, I, 1), I);
        end;
      // next page
      if (HyperLinkExists('Next')) then
        ClickHyperLink('Next')
      // goto next record
      else if (nProductIndex < MatrixMaxRow) then
        begin
          GotoURL('http://froogle.google.com/');
        end
      // script finished  
      else
        begin
          ShowMessage('Data extraction completed. '+cOutputFile);
          NewbieScriptEnd;
        end;
    end
  else if IsPartOf('http://www.google.com/products', URL) then 
    begin
      nProductIndex := nProductIndex + 1;
      Fill('q', cPrefix + MatrixGetVal(nProductIndex, 1));
      ClickButton('btnG');
    end;
end;
                         
begin
  nCount := 0;
  nProductIndex := 0;  
  DeleteFile(cOutputFile);
  MatrixReadFile(cInputFile);  
  Navigate('http://www.google.com/products');      
end.