[UPDATE DVDMovies(AR)] 1.9 fix field producer

gnbusetto · Post by **gnbusetto** » 2007-11-05 11:24:32

(***************************************************

Ant Movie Catalog importation script
www.antp.be/software/moviecatalog/

[Infos]
Authors=Legrad , GNBusetto & Bad4u
Title=DVDmovies (AR)
Description=
Site=www.dvdmovies.com.ar
Language=ES
Version=1.7
Requires=3.5.0
Comments=
License=
GetInfo=1

[Options]

***************************************************)

program DVDmovies;
var
MovieName: string;
MovieURL: string;
//------------------------------------------------------------------
function FindLine(Pattern: string; List: TStringList; StartAt: Integer): Integer;
var
i: Integer;
begin
Result := -1;
if StartAt < 0 then
StartAt := 0;
for i := StartAt to List.Count-1 do
if Pos(Pattern, List.GetString(i)) <> 0 then
begin
Result := i;
Break;
end;
end;
//------------------------------------------------------------------------------------
function TextBetween(var S: string; StartTag: string; EndTag: string): string;
var
InitialPos: Integer;
begin
InitialPos := Pos(StartTag, S);
if InitialPos = 0 then
result := ''
else
begin
Delete(S, 1, InitialPos + Length(StartTag) - 1);
InitialPos := Pos(EndTag, S);
if InitialPos = 0 then
result := S
else
begin
result := copy(S, 1, InitialPos - 1);
Delete(S, 1, InitialPos + 1);
end;
end;
end;

//------------------------------------------------------------------------------------
function DeleteTags(var S: string): string;
var
n,len, tag: Integer;
c: char;
t: String;
begin

tag := 0;
t := '';
len := length(s);

for n :=1 to len do
begin
c := Copy(s,n,1);


if c = #9 then
c := ' ';

if(tag=1) then
begin
if(c='>') then tag := 0;
continue;
end
else
begin
if(c='<') then
begin
tag := 1;
continue;
end;
t := t + c;
end;
end
s := t;
result := t;
end;


//------------------------------------------------------------------------------------
procedure AnalyzePage(Address: string);
var
strPage, MovieAddr, MovieTitle,MovieTitle1, MovieDate, MovieID, Movie: string;
BeginPos, EndPos: Integer;
BeginPoss, EndPoss: Integer;
begin
strPage := GetPage('http://www.dvdmovies.com.ar');
strPage := GetPage(Address);
BeginPos := Pos('Su búsqueda...', strPage);
if(BeginPos > -1)then
begin
PickTreeClear;
Delete(strPage, 1, BeginPos);
BeginPos := Pos('<a href="peliculaFromDB.php?ProdID=', strPage);
EndPos := 1;
while ((BeginPos > 0) and (EndPos > 0)) do

begin
Delete(strPage, 1, BeginPos);
EndPos := Pos('" title=', strPage);
MovieId := Copy(strPage,+35, EndPos-35);
MovieAddr := 'http://www.dvdmovies.com.ar/ventas/peliculaFromDB.php?ProdID=' + MovieId;
BeginPoss := Pos('"_self">',strPage);
EndPoss := Pos('</a> </td>', strPage);
MovieTitle := Copy(strPage,BeginPoss, EndPoss);
HTMLDecode(MovieTitle);
MovieTitle := TextBetween(MovieTitle ,'"_self">', '</a> </td>');
MovieTitle := AnsiLowerCase(MovieTitle);
MovieTitle := AnsiUpFirstLetter(MovieTitle);
///
PickTreeAdd(MovieTitle, MovieAddr);
BeginPos := Pos('<a href="peliculaFromDB.php?ProdID=', strPage);
if(Pos('</body>', strPage) < BeginPos) then
BeginPos := -1;
end;

end;
PickTreeSort;
PickTreeExec(Address)
AnalyzeMoviePage(Address);
end;
//------------------------------------------------------------------------------------
procedure AnalyzeMoviePage(Address: string);
var
Page: TStringList;
LineNr: Integer;
Line: string;
Item: string;
Comments: string;
Actors: string;
Directors: string;
Description: string;
Busca: integer;
Commments: string;
Copiapage: string;
Posinic: integer;

begin

// inicio

Description := '';                                    // inicializo description con cadena vacia
Page := TStringList.Create;                           // creo string page
Page.Text := GetPage(Address);                        // pongo pag. web en string page
Copiapage := GetPage(Address);                        // pongo pag. web en string Copiapage
Posinic := Pos('<u><b>', Copiapage);                  // pongo puntero en posicion inicial en copiapage
Delete(Copiapage, 1, Posinic);                        // borro lo que hay antes de la posic. inic. en string copiapage

Page.Text := StringReplace(Page.Text , '<BR>', '<br>');
Page.Text := StringReplace(Page.Text , '<TR>', '<tr>');
Page.Text := StringReplace(Page.Text , '<A>', '<a>');
Page.Text := StringReplace(Page.Text , '<U>', '<u>');
Page.Text := StringReplace(Page.Text , '<B>', '<b>');
Page.Text := StringReplace(Page.Text , '</A>', '</a>');
Page.Text := StringReplace(Page.Text , 'TD', 'td');
Page.Text := StringReplace(Page.Text , '</A>', '</a>');
Page.Text := StringReplace(Page.Text , '</FONT>', '</font>');
Page.Text := StringReplace(Page.Text , '<A HREF', '<a href');
// URL
SetField(fieldURL, Address);                          // pongo direccion web en campo fieldURL


// Titulo traducido
LineNr := FindLine('<u><b>', Page, 0);                // busco linea en string page
Line := Page.GetString(LineNr);                       // pongo nro. de linea en line
Item := TextBetween (Line, '<u><b>', '</b></u>');     // busco texto entre parametros y lo pongo en item
HTMLDecode(Item);                                     // decodifico
SetField(fieldTranslatedTitle, Trim (Item));          // pongo valor del campo


// año

begin
Item := TextBetween(Copiapage,'(', ')');              // busco y pongo valor entre () en copiapage
HTMLDecode(Item);                                     // decodifico
SetField(fieldYear, Trim (Item));                     // pongo valor del campo
end;


// Titulo original

begin                                                 // busco y pongo valor entre...
Item := TextBetween (Copiapage, '<br>(Título original: <b>', '</b>');
HTMLDecode(Item);                                     // decodifico
SetField(fieldOriginalTitle, Trim (Item));            // pongo valor del campo
end;


// Genero

begin                                                 // busco y pongo valor entre...
Item := copy(Page.Text, pos('<br>Género:',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<br>Género:', '<br>');
Item := StringReplace(Item , ' ', '');                // borro caracteres vacios
Item := StringReplace(Item , '/', ', ');              // reemplazo barras por comas
Item := StringReplace(Item , #13#10, '');             // borro retornos carro
Item := StringReplace(Item , #9, '');                 // borro tabulaciones
HTMLDecode(Item);                                     // decodifico
SetField(fieldCategory, Trim (Item));                 // pongo valor del campo
end;


// sinopsis

LineNr := FindLine('<br>Género:', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<br></font><font color=black face="Verdana',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<font color=black face="Verdana, Arial, Helvetica" size="1">', '<a href="#COMENTARIOS"');
Item := StringReplace(Item , #13#10, '');
Item := StringReplace(Item , #9, '');
DeleteTags (Item);
Item := Trim(Item );
HTMLDecode(Item);
SetField(fieldDescription, Trim (Item));
end;


// Reparto
LineNr := FindLine('<u>Protagonistas:</u> <br>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<u>Protagonistas:</u> <br>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '"Medio">', '<br>');
Item := StringReplace(Item , '</a>', ', ');
Item := Trim(Item );
DeleteTags(Item);
HTMLDecode(Item);
Posinic := length(Item);
Delete(Item, Posinic -7 , Posinic);
SetField(fieldActors, Trim (Item));
end;

// director
LineNr := FindLine('<u>Director:</u>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<u>Director:</u>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '"Medio">', '<br>');
Item := StringReplace(Item , '</a>', ', ');
Item := Trim(Item );
DeleteTags(Item);
HTMLDecode(Item);
Posinic := length(Item);
Delete(Item, Posinic -7 , Posinic);
SetField(fieldDirector, Trim (Item));
end;


// Productor
LineNr := FindLine('Producida por ', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Producida por ',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Producida por ', '.');
Item := StringReplace(Item , '/', ', ');
Item := StringReplace(Item , #13#10, '');
Item := Trim(Item );
DeleteTags(Item);
HTMLDecode(Item);
SetField(fieldProducer, Trim (Item));
end;


// Formato de video
LineNr := FindLine('<u>Video:</u>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<u>Video:</u>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<u>Video:</u>', '<u>Audio:</u>');
Item := StringReplace(Item , #13#10, '');
DeleteTags(Item);
HTMLDecode(Item);
SetField(fieldVideoFormat, Trim (Item));
end;


// Subtitulos
LineNr := FindLine('<u>Subt', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<u>Subt',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<u>Subt', '<br><br>');
Item := StringReplace(Item , #13#10, '');
Item := StringReplace(Item , ' ', ', ');
DeleteTags(Item);
HTMLDecode(Item);
Item := StringReplace(Item , 'ítulos:', '');
Posinic := length(Item);
Delete(Item, Posinic -10 , Posinic);
SetField(fieldSubtitles, Trim (Item));
end;


// Lenguaje
LineNr := FindLine('<u>Audio:</u> ', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<u>Audio:</u>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<u>Audio:</u>', '<br><br>');
Delete(Item, 1, 1);
Item := StringReplace(Item , '<br>', ', ');
Item := StringReplace(Item , #13#10, '');
DeleteTags(Item);
Delete( Item, 1, 1);
HTMLDecode(Item);
SetField(fieldLanguages, Trim (Item));
end;


// duracion
LineNr := FindLine('<u>Duración:</u> ', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<u>Duración:</u> ',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<u>Duración:</u> ', 'minutos');
SetField(fieldLength, Trim (Item));
end;
//**********************************************************************************
// Extras:
Comments := '';

LineNr := FindLine('<br><br><u>Extras:</u><br>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<br><br><u>Extras:</u><br>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<br><br><u>Extras:</u><br>', '</font>');
Item := StringReplace(Item , #13#10, '');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Extras: ' + Item +#13#10;
end;

// Capitulos:
LineNr := FindLine('<u>Capítulos:</u> ', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<u>Capítulos:</u> ',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<u>Capítulos:</u> ', '<br>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Capítulos: ' + Item +#13#10;
end;


// Origen del producto
LineNr := FindLine('Origen Producto:</u>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Origen Producto:</u>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Origen Producto:</u>', '</font>');
Item := StringReplace(Item , #13#10,'');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Origen del producto: '+Item + #13#10 ;
end;


// Lanzamiento
LineNr := FindLine('Lanzamiento', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Lanzamiento',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'size="1">', '</font></td>');
Item := StringReplace(Item , #13#10,'');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Lanzamiento: '+Item + #13#10 ;
end;

// Puntuación
LineNr := FindLine('size="1">Puntaje', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('size="1">Puntaje',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'size="1">Puntaje', '</font><hr');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Puntuación: '+Item + #13#10 ;
end;

// Contenido del DVD
LineNr := FindLine('Contenido DVD:</b><br>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Contenido DVD:</b><br>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Contenido DVD:</b><br>', '<br><br>');
Item := StringReplace(Item , #13#10,'');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Contenido del DVD: '+Item + #13#10 ;
end;

// Características
LineNr := FindLine('<b>Características:</b>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<b>Características:</b>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<b>Características:</b>', '<br><br>');
Item := StringReplace(Item , #13#10,'');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Características: '+Item + #13#10 ;
end;

SetField(fieldComments, Comments);

// Caratula
LineNr := FindLine('<IMG SRC="http://www.dvdmovies.com.ar', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<IMG SRC="http://www.dvdmovies.com.ar',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<IMG SRC="http://www.dvdmovies.com.ar', 'jpg"');
Item := Trim(Item );
HTMLDecode(Item);
GetPicture ('http://www.dvdmovies.com.ar'+Item+'jpg');

end;

end;
//-------------------------------------------------------------------------
begin
if (CheckVersion(3,5,0)=FALSe) then
begin
ShowMessage('Se requiere Ant Movie Catalog versión 3.5 o superior');
exit;
end;

MovieName := GetField(fieldTranslatedTitle);
if MovieName = '' then
MovieName := GetField(fieldOriginalTitle);
Input('DVDmovies', 'Buscar:', MovieName);
AnalyzePage('http://www.dvdmovies.com.ar/ventas/peliculas_avanzado.php?BuscarTituloEsp=' + UrlEncode(MovieName));
end.

Post by **antp** » 2007-11-05 15:16:11

What is the difference between this one and the other topic?
viewtopic.php?p=24458
Both scripts have version 1.7 in their properties

It is maybe better to delete older versions to prevent users from getting the old one

gnbusetto · Post by **gnbusetto** » 2007-11-05 21:24:57

The difference is a space.

That`s make the field are filled with trash until the script found ". "(maybe many lines later).

And I'm sorry, you'r right , i have to increment version in properties and i'm not do it.

Post by **antp** » 2007-11-05 22:14:20

Thanks, so I took the script above and changed version to 1.9 in the version uploaded on my server

antp.be forum

[UPDATE DVDMovies(AR)] 1.9 fix field producer

[UPDATE DVDMovies(AR)] 1.9 fix field producer

IN the field producer , fix ". " now it's ".&