Todos los datos se extraen de cineol se carga la caratula de alpacine si en alpacine no se encuentra la caratula se carga la pequeña que hay en cineol
Code: Select all
(***************************************************
Ant Movie Catalog importation script
www.antp.be/software/moviecatalog/
[Infos]
Authors=Legrad
Title=Cineol-Alpacine
Description=
Site=www.cineol.net-www.alpacine.com
Language=ES
Version=1.0
Requires=3.5.0
Comments=
License=
GetInfo=1
[Options]
***************************************************)
program CineolAlpacine;
var
MovieName: string;
MovieURL: string;
//---------------------------------------------------------------------
function Acentos(Acento: string) :string;
begin
Acento := StringReplace(Acento, 'á' , 'á');
Acento := StringReplace(Acento, 'é' , 'é');
Acento := StringReplace(Acento, 'á' , 'í');
Acento := StringReplace(Acento, 'Ã' , 'í');
Acento := StringReplace(Acento, 'ó', 'ó');
Acento := StringReplace(Acento, 'ú', 'ú');
Acento := StringReplace(Acento, 'ñ', 'ñ');
result := Acento;
end;
//------------------------------------------------------------------------------------
function BorraComillas(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 = '''' then
c := ' ';
t := t + c;
end
s := t;
result := t;
end;
//------------------------------------------------------------------
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, MovieDate, MovieID, Movie: string;
BeginPos, EndPos: Integer;
BeginPoss, EndPoss: Integer;
begin
strPage := GetPage(Address);
BeginPos := Pos('<h1>Resultados de la búsqueda en CINeol</h1>', strPage);
if(BeginPos > -1)then
begin
PickTreeClear;
Delete(strPage, 1, BeginPos);
BeginPos := Pos('<h3><a href="pelicula/', strPage);
EndPos := 1;
while ((BeginPos > 0) and (EndPos > 0)) do
begin
Delete(strPage, 1, BeginPos);
EndPos := Pos('">', strPage);
MovieId := Copy(strPage,+22, EndPos-22);
MovieAddr := 'http://www.cineol.net/pelicula/' + MovieId;
BeginPoss := Pos('">',strPage);
EndPoss := Pos('</a>', strPage);
MovieTitle := Copy(strPage,BeginPoss, EndPoss);
MovieTitle := TextBetween(MovieTitle ,'">', #13#10);
DeleteTags (MovieTitle);
HTMLDecode(MovieTitle);
MovieTitle := UTF8Decode(MovieTitle);
MovieTitle := Acentos(MovieTitle);
PickTreeAdd(MovieTitle, MovieAddr);
BeginPos := Pos('<h3><a href="pelicula/', 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;
begin
Description := '';
// URL
SetField(fieldURL, Address);
Page := TStringList.Create;
Page.Text := GetPage(Address);
// Titulo Original
LineNr := FindLine('tulo Original</strong>:', Page, 0);
if LineNr <> -1 then
begin
Line := Page.GetString(LineNr);
Item := TextBetween (Line, 'tulo Original</strong>:', '</');
HTMLDecode(Item);
DeleteTags (Item);
Item := Acentos(Item);
SetField(fieldOriginalTitle, Trim (Item));
end;
// año
LineNr := FindLine('<li><strong>A', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('a<li><strong>A',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'o:</strong> ', '</');
Item := Trim(Item );
HTMLDecode(Item);
SetField(fieldYear, Trim (Item));
end;
// Categoria
LineNr := FindLine('nero:</strong> ', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('nero:</strong> ',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'nero:</strong> ', '</');
Item := Trim(Item );
HTMLDecode(Item);
DeleteTags (Item);
Item := Acentos(Item);
SetField(fieldCategory, Trim (Item));
end;
// pais
LineNr := FindLine('<li><strong>Pa', Page, 0);
if LineNr <> -1 then
begin
Line := Page.GetString(LineNr);
Item := TextBetween (Line, 's:</strong>', '</');
Item := StringReplace(Item , '/', ',');
HTMLDecode(Item);
DeleteTags (Item);
Item := Acentos(Item);
SetField(fieldCountry, Trim (Item));
end;
// Duracion
LineNr := FindLine('<li><strong>Duraci', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<li><strong>Duraci',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'n</strong>:', 'minutos');
Item := Trim(Item );
HTMLDecode(Item);
SetField(fieldLength, Trim (Item));
end;
// director
LineNr := FindLine('<li><strong>Direcci', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<li><strong>Direcci',Page.Text), length(Page.Text));
Item := TextBetween (Item, '">', '</a></li>');
Item := Trim(Item );
HTMLDecode(Item);
DeleteTags (Item);
Item := Acentos(Item);
SetField(fieldDirector, Trim (Item));
end;
// productor
LineNr := FindLine('<li><strong>Producci', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<li><strong>Producci',Page.Text), length(Page.Text));
Item := TextBetween (Item, '">', '</a></li>');
Item := Trim(Item );
HTMLDecode(Item);
DeleteTags (Item);
Item := Acentos(Item);
SetField(fieldProducer, Trim (Item));
end;
// sinopsis
LineNr := FindLine('<h2>Sinopsis</h2>', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<h2>Sinopsis</h2>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<h2>Sinopsis</h2>', '</p>');
Item := StringReplace(Item , #13#10, '');
Item := StringReplace(Item , ' ', '');
DeleteTags (Item);
Item := Trim(Item );
HTMLDecode(Item);
Item := Acentos(Item);
SetField(fieldDescription, Trim (Item));
end;
// Reparto
LineNr := FindLine('rpretes</h2>', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('rpretes</h2>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '</tr>', '</table>');
Item := StringReplace(Item , '</a><td>', ' (');
Item := StringReplace(Item , '</td></tr>', '), ');
Item := StringReplace(Item , #13#10, '');
DeleteTags (Item);
Item := Trim(Item );
HTMLDecode(Item);
Item := StringReplace(Item , ' ', ' ');
Item := Acentos(Item);
SetField(fieldActors, Trim (Item));
end;
// Guión
LineNr := FindLine('<li><strong>Gui', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<li><strong>Gui',Page.Text), length(Page.Text));
Item := TextBetween (Item, '">', '</a>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Item := Acentos(Item);
Comments := Comments + 'Guión: ' + Item +#13#10;
end;
// Fotografia
LineNr := FindLine('<li><strong>Fotograf', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<li><strong>Fotograf',Page.Text), length(Page.Text));
Item := TextBetween (Item, '">', '</a>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Item := Acentos(Item);
Comments := Comments + 'Fotografía: ' + Item +#13#10;
end;
// Musica
LineNr := FindLine('sica:</strong>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('sica:</strong>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '">', '</a>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Item := Acentos(Item);
Comments := Comments + 'Música: ' + Item +#13#10;
end;
// formato
LineNr := FindLine('<li><strong>Formato:</strong>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<li><strong>Formato:</strong>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<li><strong>Formato:</strong>', '</');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Item := Acentos(Item);
Comments := Comments + 'Formato: ' + Item +#13#10;
end;
// Estreno
LineNr := FindLine('<li><strong>Estreno en Espa', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<li><strong>Estreno en Espa',Page.Text), length(Page.Text));
Item := TextBetween (Item, '</strong>', '</li>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Item := Acentos(Item);
Comments := Comments + 'Estreno en España: ' + Item +#13#10;
end;
// Recaudacion
LineNr := FindLine('<li><strong>Recaudación Espa', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<li><strong>Recaudación Espa',Page.Text), length(Page.Text));
Item := TextBetween (Item, '</strong>', '</');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Item := Acentos(Item);
Comments := Comments + 'Recaudación España: ' + Item +#13#10;
end;
// Recaudacion mundial
LineNr := FindLine('<li><strong>Estreno Mund', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<li><strong>Estreno Mund',Page.Text), length(Page.Text));
Item := TextBetween (Item, '</strong>', '</');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Item := Acentos(Item);
Comments := Comments + 'Recaudación mundial: ' + Item +#13#10;
end;
SetField(fieldComments, Comments);
// Caratula
LineNr := FindLine('<img src="galeria/carteles/', Page, 0);
if LineNr <> -1 then
begin
Line := Page.GetString(Linenr);
Item := TextBetween (Line, '<img src="galeria/carteles/', '"');
GetPicture ('http://www.cineol.net/galeria/carteles/'+Item);
end;
// Titulo traducido
LineNr := FindLine('<title>', Page, 0);
Line := Page.GetString(LineNr);
Item := TextBetween (Line, '<title>', '::');
HTMLDecode(Item);
DeleteTags (Item);
Item := Acentos(Item);
SetField(fieldTranslatedTitle, Trim (Item));
//--------------------------------------------------------
Item := StringReplace (Item, ' ', '+');
Page.Text := GetPage('http://www.google.es/search?hl=es&as_qdr=all&q=%22'+(Item)+'%22+site%3Aalpacine.com&btnG=Buscar&meta=');
LineNr := FindLine('en el dominio <b>alpacine.com', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('href="http://alpacine.com/pelicula/',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'href="http://alpacine.com/pelicula/', '"');
Page.Text := GetPage('http://alpacine.com/pelicula/'+Item);
LineNr := FindLine('<img src="http://img.alpacine.com/carteles/', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<img src="http://img.alpacine.com/carteles/',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<img src="http://img.alpacine.com/carteles/', '-');
GetPicture ('http://img.alpacine.com/carteles/'+(Item)+'.jpg');
end;
end;
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('Cineol-Alpacine ', 'Buscar:', MovieName);
if(GetOption('Sin resultado') = 0) then Input('Cineol-Alpacine', 'Buscar:', MovieName);
AnalyzePage('http://www.cineol.net/search.php?search=' + UrlEncode(MovieName)+'&select=0&button=Buscar');
end.