Code: Select all
(***************************************************
Ant Movie Catalog importation script
www.antp.be/software/moviecatalog/
[Infos]
Authors=Legrad
Title=Comohacercine(ES)
Description=
Site=www.Comohacercine.com
Language=ES
Version=1.0
Requires=3.5.0
Comments=
License=
GetInfo=1
[Options]
***************************************************)
program Comohacercine;
var
MovieName: string;
MovieURL: string;
//---------------------------------------------------------------------
function Acentos(Acento: string) :string;
begin
Acento := StringReplace(Acento, 'Ã-' , 'á');
//Acento := StringReplace(Acento, 'Ã' , 'á');
Acento := StringReplace(Acento, 'ã¡' , 'á');
Acento := StringReplace(Acento, 'iã³', 'á');
Acento := StringReplace(Acento, 'Ã' , 'á');
Acento := StringReplace(Acento, 'ã', 'á');
Acento := StringReplace(Acento, 'é', 'é');
Acento := StringReplace(Acento, 'ã©' , 'é');
Acento := StringReplace(Acento, 'ãª', 'é');
Acento := StringReplace(Acento, 'Ã', 'í');
Acento := StringReplace(Acento, 'ã', 'í');
Acento := StringReplace(Acento, 'ó', 'ó');
Acento := StringReplace(Acento, 'ã³', 'ó');
Acento := StringReplace(Acento, 'á³', 'ó');
Acento := StringReplace(Acento, 'ú', 'ù');
Acento := StringReplace(Acento, 'ñ', 'ñ');
Acento := StringReplace(Acento, 'ã±', 'ñ');
Acento := StringReplace(Acento, 'º', 'º');
Acento := StringReplace(Acento, '&#', '');
Acento := StringReplace(Acento, '39;', '');
Acento := StringReplace(Acento, 'Â', '');
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('</b> en el dominio <b>www.comohacercine.com</b>', strPage);
if(BeginPos > -1)then
begin
PickTreeClear;
Delete(strPage, 1, BeginPos);
BeginPos := Pos('http://www.comohacercine.com/cine_ficha.php?id=', strPage);
EndPos := 1;
while ((BeginPos > 0) and (EndPos > 0)) do
begin
Delete(strPage, 1, BeginPos);
EndPos := Pos('" class', strPage);
MovieId := Copy(strPage,+47, EndPos-47);
MovieAddr := 'http://www.comohacercine.com/cine_ficha.php?id=' + MovieId;
BeginPoss := Pos(')"',strPage);
EndPoss := Pos('</a>', strPage);
MovieTitle := Copy(strPage,BeginPoss, EndPoss);
MovieTitle := TextBetween(MovieTitle ,'>', '-');
MovieTitle :=Acentos(MovieTitle);
DeleteTags(MovieTitle);
PickTreeAdd(MovieTitle, MovieAddr);
BeginPos := Pos('http://www.comohacercine.com/cine_ficha.php?id=', strPage);
if(Pos('</body>', strPage) < BeginPos) then
BeginPos := -1;
end;
end;
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 := '';
Comments := '';
// URL
SetField(fieldURL, Address);
Page := TStringList.Create;
Page.Text := GetPage(Address);
// Titulo traducido
LineNr := FindLine('<title>', Page, 0);
Line := Page.GetString(LineNr);
Item := TextBetween (Line, '<title>', '-');
HTMLDecode(Item);
SetField(fieldTranslatedTitle, Trim (Item));
// Titulo original
LineNr := FindLine('<span class="link-articulos2">', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<span class="link-articulos2">',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<br><i>', '</i> ');
HTMLDecode(Item);
SetField(fieldOriginalTitle, Trim (Item));
end;
// sinopsis
LineNr := FindLine('<span class="link-articulos2', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<span class="link-articulos2',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'class="v10">', '</td>');
Item := StringReplace (Item, #13#10, '');
DeleteTags (Item);
Item := Trim(Item );
HTMLDecode(Item);
SetField(fieldDescription, Trim (Item));
end;
// año
LineNr := FindLine('FICHA TÉCNICA</span>', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('FICHA TÉCNICA</span>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'FICHA TÉCNICA</span>', ', ');
Item := StringReplace(Item ,#13#10, '');
DeleteTags(Item);
HTMLDecode(Item);
Item := Trim(Item );
HTMLDecode(Item);
SetField(fieldYear, Trim (Item));
end;
// pais
LineNr := FindLine('FICHA TÉCNICA</span>', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('class="v10">',Page.Text), length(Page.Text));
Item := TextBetween (Item, ', ', '<BR><b>');
HTMLDecode(Item);
Item := StringReplace(Item , 'EE.UU.', 'USA');
SetField(fieldCountry, Trim (Item));
end;
// director
LineNr := FindLine('<b>Director:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<b>Director:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<b>Director:</b> <font class=colorcitodenuri>', '</font><BR>');
Item := Trim(Item );
HTMLDecode(Item);
SetField(fieldDirector, Trim (Item));
end;
// Duracion
LineNr := FindLine('Duración:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('Duración:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Duración:</b> <font class=colorcitodenuri>', 'minutos');
Item := Trim(Item );
HTMLDecode(Item);
SetField(fieldLength, Trim (Item));
end;
// Reparto
LineNr := FindLine('Actores principales: </b><font class=colorcitodenuri>', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('Actores principales: </b><font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Actores principales: </b><font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item );
HTMLDecode(Item);
SetField(fieldActors, Trim (Item));
end;
// Productor
LineNr := FindLine('>Producción:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('>Producción:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '>Producción:</b> <font class=colorcitodenuri>', '</font><BR>');
Item := Trim(Item );
HTMLDecode(Item);
SetField(fieldProducer, Trim (Item));
end;
// Guión
LineNr := FindLine('Guión:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Guión:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Guión:</b> <font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Guión: ' + Item +#13#10;
end;
// Fotografía
LineNr := FindLine('Dirección de fotografía:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Dirección de fotografía:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Dirección de fotografía:</b> <font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Fotografía: ' + Item +#13#10;
end;
// Vestuario
LineNr := FindLine('Vestuario:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Vestuario:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Vestuario:</b> <font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Vestuario: ' + Item +#13#10;
end;
// Montaje
LineNr := FindLine('Montaje:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Montaje:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Montaje:</b> <font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Montaje: ' + Item +#13#10;
end;
// Banda Sonora
LineNr := FindLine('Banda sonora:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Banda sonora:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Banda sonora:</b> <font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Banda Sonora: ' + Item +#13#10;
end;
// Ficha Artistica
LineNr := FindLine('FICHA ARTÍSTICA</span> ', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('FICHA ARTÍSTICA</span> ',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Ficha Artistica: ' + Item +#13#10;
end;
// Distribuidora
LineNr := FindLine('<b>Distribuidora:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<b>Distribuidora:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<b>Distribuidora:</b> <font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Distribuidora: ' + Item +#13#10;
end;
// Calificación
LineNr := FindLine('<b>Calificación:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('<b>Calificación:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<b>Calificación:</b> <font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Calificación: ' + Item +#13#10;
end;
// Fecha de estreno
LineNr := FindLine('Fecha de estreno:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Fecha de estreno:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Fecha de estreno:</b> <font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Fecha de estreno: ' + Item +#13#10;
end;
// Fecha de lanzamiento
LineNr := FindLine('Fecha de lanzamiento:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Fecha de lanzamiento:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Fecha de lanzamiento:</b> <font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Fecha de Lanzamiento: '+Item+#13#10;
end;
// Precio
LineNr := FindLine('Precio:</b> <font class=colorcitodenuri>', Page, 0);
if LineNr <> -1 then
begin
Item := copy(Page.Text, pos('Precio:</b> <font class=colorcitodenuri>',Page.Text), length(Page.Text));
Item := TextBetween (Item, 'Precio:</b> <font class=colorcitodenuri>', '</font><BR>');
DeleteTags (Item);
Item := Trim(Item);
HTMLDecode(Item);
Comments := Comments + 'Precio: '+Item+#13#10;
end;
// Categoria
LineNr := FindLine('<b>Género: </b>', Page, 0);
if LineNr > 0 then
begin
Item := copy(Page.Text, pos('<b>Género: </b>',Page.Text), length(Page.Text));
Item := TextBetween (Item, '<b>Género: </b>', '.<br');
Item := Trim(Item );
HTMLDecode(Item);
SetField(fieldCategory, Trim (Item));
end;
//caratula pequeña
LineNr := FindLine('<img src="img_dvd/', Page, 0);
if LineNr <> -1 then
begin
Line := Page.GetString(Linenr);
Item := TextBetween (Line, '<img src="img_dvd/', '" width');
GetPicture ('http://www.comohacercine.com/img_dvd/'+Item);
end;
SetField(fieldComments, Comments);
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('Comohacercine', 'Buscar:', MovieName);
if(GetOption('Sin resultado') = 0) then Input('Comohacercine', 'Buscar:', MovieName);
AnalyzePage('http://www.google.es/search?num=100&hl=es&as_qdr=all&q=allinanchor%3A++%22' + UrlEncode(MovieName)+'%22+site%3Awww.comohacercine.com&btnG=B%C3%BAsqueda&meta=');
end.