com!online 2005 May/com_0505_1.iso/opensource/top10/amc_install.exe/{app}/Scripts/IMDB.ifs

Text File | 2005-02-02 | 21.7 KB | 684 lines

// GETINFO SCRIPTING
// IMDB (US) import - click "Editor" tab to change options

{
   To change the options, go to the "GetOption"
   and "CanSetField" functions
}

(***************************************************
 *  Movie importation script for:                  *
 *      IMDB (US), http://us.imdb.com              *
 *                                                 *
 *  (c) 2002-2005 Antoine Potten                   *
 *                          [e-mail]       *
 *                                                 *
 *  For use with Ant Movie Catalog 3.4.3           *
 *  www.antp.be/software/moviecatalog              *
 *                                                 *
 *  Based on the script rewritten for version 3.5  *
 *  Which was based on the script made for         *
 *  version 3.3.x / 3.4.x by                       *
 *  Antoine Potten, Danny Falkov, Kai Blankenhorn, *
 *  lboregard, Ork, Trekkie, Youri Heijnen         *
 *                                                 *
 *  This program is free software; you can         *
 *  redistribute it and/or modify it under the     *
 *  terms of the GNU General Public License as     *
 *  published by the Free Software Foundation;     *
 *  either version 2 of the License, or (at your   *
 *  option) any later version.                     *
 ***************************************************)

program IMDB;

// ***** Here you can change the options of the script *****

function GetOption(OptName: string): Integer;
begin
  case OptName of
    'ImageKind': Result := 2;
        {
          0=No image
          1=IMDB small image, from the main movie page
          2=IMDB large image if found easily, else small image
          3=First search for Amazon large image, then IMDB large one, then IMDB small image if other failed
          4=First search for Amazon large image, then directly take IMDB small image if the first one failed
          5=IMDB large image if found easily, else search for Amazon large image, then take IMDB small image if others failed
        }
    'BatchMode': Result := 0;
        {
          0=Normal working mode, prompts user when needed
          1=Does not display any window, takes the first movie found
        }
    'PopularSearches': Result := 1;
        {
          0=Do not use the popular searches page, directly show full search results
          1=Show popular searches first, I'll click on "Find more" if needed
        }
    'ActorsLayout': Result := 0;
        {
          0=Only actor names, separated by commas
          1=Only actor names, separated by linebreaks
          2=Actors names with character names between parenthesis separated by commas
          3=Actors names with character names between parenthesis separated by linebreaks
          4=Actor names like on IMDB page, with "...." and separated by linebreaks
        }
    'MultipleValuesCountry': Result := 0;
        {
          0=Only take first value for Country
          1=Take full list, separated by commas
          2=Take full list, separated by slashes
        }
    'MultipleValuesCategory': Result := 0;
        {
          0=Only take first value for Category
          1=Take full list, separated by commas
          2=Take full list, separated by slashes
        }
    'MultipleValuesLanguages': Result := 0;
        {
          0=Only take first value for Languages
          1=Take full list, separated by commas
          2=Take full list, separated by slashes
        }
    'DescriptionSelection': Result := 2;
        {
          0=Take the short summary, from main page (faster)
          1=Show a list of available summaries
          2=Take the longest summary
        }
    'GetTagline': Result := 0;
        {
          0=Do not get tagline
          1=Put it in Description field, before the summary
          2=Put it in the Comment field, before the comments
        }
  end;
end;

// Here you can specify which fields can be modified.
// Simply remove a field from the list if it cannot be modified

function CanSetField(fieldID: Integer): Boolean;
begin
  case fieldID of
    fieldNumber,
    fieldMedia,
    fieldMediaType,
    fieldSource,
    fieldDate,
    fieldBorrower,
    fieldRating,
    fieldOriginalTitle,
    fieldTranslatedTitle,
    fieldDirector,
    fieldProducer,
    fieldCountry,
    fieldCategory,
    fieldYear,
    fieldLength,
    fieldActors,
    fieldURL,
    fieldDescription,
    fieldComments,
    fieldVideoFormat,
    fieldVideoBitrate,
    fieldAudioFormat,
    fieldAudioBitrate,
    fieldResolution,
    fieldFrameRate,
    fieldLanguages,
    fieldSubtitles,
    fieldSize,
    fieldDisks:   Result := True;
  else
    Result := False;
  end;
end;

var
  MovieName: string;
  MovieURL: string;
  MovieNumber: string;

var
  RemainingText: string;
  {
    when calling...     this variable contains...
    TextBefore          the text after SearchText
    TextBetween         the text after AfterText
  }

// ***** Like the Pos function, but returns the last occurence instead of the first one *****

function LastPos(ASearch: string; AText: string): Integer;
var
  CurPos, PrevPos: Integer;
begin
  PrevPos := 0;
  CurPos := Pos(ASearch, AText);
  while CurPos > 0 do
  begin
    if PrevPos = 0 then
      PrevPos := CurPos
    else
      PrevPos := PrevPos + CurPos + Length(ASearch) - 1;
    Delete(AText, 1, CurPos + Length(ASearch) - 1);
    CurPos := Pos(ASearch, AText);
  end;
  Result := PrevPos;
end;

// *****
{    Returns the text before SearchText, but not before BeginLimit (if it is not empty),
    It takes the last occurence of BeginLimit found before the position of SearchText  }

function TextBefore(WholeText: string; SearchText: string; BeginLimit: string): string;
var
  FoundPos, PrevPos: Integer;
  WorkText: string;
begin
  Result := '';
  FoundPos := Pos(SearchText, WholeText);
  if FoundPos = 0 then
    Exit;
  WorkText := Copy(WholeText, 1, FoundPos - 1);
  RemainingText := Copy(WholeText, FoundPos + Length(SearchText), Length(WholeText));
  if BeginLimit <> '' then
  begin
    FoundPos := LastPos(BeginLimit, WorkText);
    if FoundPos = 0 then
      Exit
    else
      FoundPos := FoundPos + Length(BeginLimit);
  end
  else
    FoundPos := 1;
  Result := Copy(WorkText, FoundPos, Length(WorkText));
end;

// ***** Returns the text after SearchText *****

function TextAfter(WholeText: string; SearchText: string): string;
var
  FoundPos: Integer;
begin
  Result := '';
  FoundPos := Pos(SearchText, WholeText);
  if FoundPos = 0 then
    Exit;
  Result := Copy(WholeText, FoundPos + Length(SearchText), Length(WholeText));
end;

// *****
{    Returns the text between BeforeText and AfterText (without these two strings),
     It takes the first AfterText occurence found after the position of BeforeText  }

function TextBetween(WholeText: string; BeforeText: string; AfterText: string): string;
var
  FoundPos: Integer;
  WorkText: string;
begin
  Result := '';
  FoundPos := Pos(BeforeText, WholeText);
  if FoundPos = 0 then
    Exit;
  WorkText := Copy(WholeText, FoundPos + Length(BeforeText), Length(WholeText));
  FoundPos := Pos(AfterText, WorkText);
  if FoundPos = 0 then
    Exit;
  Result := Copy(WorkText, 1, FoundPos - 1);
  RemainingText := Copy(WorkText, FoundPos + Length(AfterText), Length(WorkText));
end;

// ***** analyzes the results page that asks to select a movie from a list *****

procedure AnalyzeResultsPage(Address: string);
var
  PageText: string;
  Value: string;
begin
  PageText := GetPage(Address);
  if pos('<title>IMDb', PageText) = 0 then
  begin
    AnalyzeMoviePage(PageText)
  end else
  begin
    if Pos('<b>No Matches.</b>', PageText) > 0 then
    begin
      if GetOption('BatchMode') = 0 then
        ShowMessage('No movie found for this search');
      Exit;
    end;
    PickTreeClear;
    repeat
      Value := TextBefore(PageText, '<ol>', '<b>');
      if Value <> '' then
      begin
        HTMLRemoveTags(Value);
        HTMLDecode(Value);
        PickTreeAdd(Value, '');
      end;
      Value := TextBetween(PageText, '<ol>', '</ol>');
      PageText := RemainingText;
    until not AddMovieTitles(Value);
    Value := TextBefore(PageText, '"><b>more titles</b></a>', '<a href="');
    if Value <> '' then
      PickTreeMoreLink('http://us.imdb.com' + Value);
    if PickTreeExec(Address) then
      AnalyzeResultsPage(Address);
  end;
end;

// ***** adds the titles contained in <ol>'s items *****

function AddMovieTitles(List: string): Boolean;
var
  Value: string;
  Address: string;
begin
  Result := False;
  Value := TextBetween(List, '<li>', '</li>');
  List := RemainingText;
  while Value <> '' do
  begin
    Address := TextBetween(Value, '<a href="', '">');
    HTMLRemoveTags(Value);
    HTMLDecode(Value);
    PickTreeAdd(Value, 'http://us.imdb.com' + Address);
    Result := True;
    Value := TextBetween(List, '<li>', '</li>');
    List := RemainingText;
  end;
end;

// ***** analyzes the page containing movie information *****

procedure AnalyzeMoviePage(PageText: string);
var
  Value, Value2, Value3, FullValue: string;
begin
  MovieNumber := TextBetween(PageText, '<input type="hidden" name="arg" value="', '"><input');
  MovieURL := 'http://us.imdb.com/title/tt' + MovieNumber;
  // URL
  if CanSetField(fieldURL) then
    SetField(fieldURL, MovieURL);
  // Original Title & Year
  if CanSetField(fieldOriginalTitle) or CanSetField(fieldYear) then
  begin
    Value := TextBetween(PageText, '<title>', '</title>');
    Value2 := TextBefore(Value, ' (', '');
    Value := RemainingText;
    HTMLDecode(Value2);
    if CanSetField(fieldOriginalTitle) then
      SetField(fieldOriginalTitle, Value2);
    if Pos('/', Value) > 0 then
      Value2 := TextBefore(Value, '/', '')
    else
      Value2 := TextBefore(Value, ')', '');
    if CanSetField(fieldYear) then
      SetField(fieldYear, Value2);
  end;
  // Rating
  if CanSetField(fieldRating) then
  begin
    Value := TextBetween(PageText, '/rating-stars/', '/rating-vote/');
    Value2 := TextBetween(Value, '<b>', '.');
    if StrToInt(Copy(RemainingText, 1, 1), 0) >= 5 then
      Value2 := IntToStr(StrToInt(Value2, 0) + 1);
    SetField(fieldRating, Value2);
  end;
  // Picture
  case GetOption('ImageKind') of
    1:  ImportSmallPicture(PageText);
    2:  if not ImportLargePicture('http://us.imdb.com/gallery/ss/' + MovieNumber) then
          ImportSmallPicture(PageText);
    3:  if not ImportAmazonPicture(PageText) then
          if not ImportLargePicture('http://us.imdb.com/gallery/ss/' + MovieNumber) then
            ImportSmallPicture(PageText);
    4:  if not ImportAmazonPicture(PageText) then
          ImportSmallPicture(PageText);
    5:  if not ImportLargePicture('http://us.imdb.com/gallery/ss/' + MovieNumber) then
          if not ImportAmazonPicture(PageText) then
            ImportSmallPicture(PageText);
  end;
  // Director
  if CanSetField(fieldDirector) then
  begin
    Value := TextBetween(PageText, '<b class="blackcatheader">Directed by</b>', '</a>');
    Value := StringReplace(TextAfter(Value, '">'), '<br>', ', ');
    HTMLRemoveTags(Value);
    HTMLDecode(Value);
    SetField(fieldDirector, Value);
  end;
  // Actors
  if CanSetField(fieldActors) then
  begin
    Value := TextBetween(PageText, 'ast overview', '</div>');
    if Value = '' then
      Value := TextBetween(PageText, 'redited cast', '</div>');
    if Value <> '' then
    begin
      Value := TextAfter(Value, '</tr> ');
      FullValue := '';
      case GetOption('ActorsLayout') of
        0, 1:
          while Pos('<tr>', Value) > 0 do
          begin
            Value2 := TextBetween(Value, '<tr>', '</tr>');
            Value := RemainingText;
            if Pos('<a href="fullcredits">(more)</a>', Value2) > 0 then
              Break;
            if FullValue <> '' then
              FullValue := FullValue + #13#10;
            FullValue := FullValue + TextBefore(Value2, '</td>', '');
          end;
        2, 3:
          while Pos('<tr>', Value) > 0 do
          begin
            Value2 := TextBetween(Value, '<tr>', '</tr>');
            Value := RemainingText;
            if Pos('<a href="fullcredits">(more)</a>', Value2) > 0 then
              Break;
            if FullValue <> '' then
              FullValue := FullValue + #13#10;
            FullValue := FullValue + TextBefore(Value2, '</td>', '');
            Value2 := TextBetween(RemainingText, '<td valign="top">', '</td>');
            if Value2 <> '' then
              FullValue := FullValue + ' (as ' + Value2 + ')';
          end;
        4:
          begin
            FullValue := TextBefore(Value, '</tr><tr><td colspan="2">', '');
            if FullValue = '' then
              FullValue := Value;
            FullValue := StringReplace(FullValue, '</tr>', #13#10);
          end;
      end;
      HTMLRemoveTags(FullValue);
      HTMLDecode(FullValue);
      case GetOption('ActorsLayout') of
        0, 2:
          FullValue := StringReplace(FullValue, #13#10, ', ');
      end;
      SetField(fieldActors, FullValue);
    end;
  end;
  //Country
  if CanSetField(fieldCountry) then
  begin
    SetField(fieldCountry, ImportList(PageText, GetOption('MultipleValuesCountry'), '/Countries/'));
  end;
  //Category
  if CanSetField(fieldCategory) then
  begin
    SetField(fieldCategory, ImportList(PageText, GetOption('MultipleValuesCategory'), '/Genres/'));
  end;
  // Language
  if CanSetField(fieldLanguages) then
  begin
    SetField(fieldLanguages, ImportList(PageText, GetOption('MultipleValuesLanguages'), '/Languages/'));
  end;
  //Description
  if CanSetField(fieldDescription) then
  begin
    Value := TextBetween(PageText, '<b class="ch">Plot Outline:</b>', '<br><br>');
    if Value = '' then
      Value := TextBetween(PageText, '<b class="ch">Plot Summary:</b>', '<br><br>');
    if Value <> '' then
      SetField(fieldDescription, ImportSummary(Value));
  end;
  // Comments
  if CanSetField(fieldComments) then
  begin
    Value := TextAfter(PageText, '/comments">');
    if Value <> '' then
    begin
      Value := TextBetween(Value, '<p>', '</p>');
      Value := StringReplace(Value, #13#10, ' ');
      Value := StringReplace(Value, '<br>', #13#10);
      HTMLRemoveTags(Value);
      HTMLDecode(Value);
      Value := Trim(Value);
      while Pos('  ', Value) > 0 do
        Value := StringReplace(Value, '  ', ' ');
      while Pos(#13#10, Value) = 1 do
        Delete(Value, 1, 2);
      SetField(fieldComments, Value);
    end;
  end;
  // Length
  if CanSetField(fieldLength) then
  begin
    Value := TextBetween(PageText, '<b class="ch">Runtime:</b>' + #13#10, ' ');
    if Value <> '' then
    begin
      if Pos(':', Value) > 0 then
        SetField(fieldLength, TextAfter(Value, ':'))
      else
        SetField(fieldLength, Value);
    end;
  end;
  // TagLine
  if GetOption('GetTagline') > 0 then
  begin
    Value := TextBetween(PageText, 'Tagline:</b>', #13);
    if Pos('<a', Value) > 0 then
      Value := TextBefore(Value, '<a', '');
    HTMLRemoveTags(Value);
    HTMLDecode(Value);
    Value := Trim(Value);
    if Value <> '' then
    begin
      Value := '"' + Value + '"';
      case GetOption('GetTagline') of
        1:
          if CanSetField(fieldDescription) then
            SetField(fieldDescription, Value + #13#10 + GetField(fieldDescription));
        2:
          if CanSetField(fieldComments) then
            SetField(fieldComments, Value + #13#10 + GetField(fieldComments));
      end;
    end;
  end;
end;

// ***** Imports lists like Genre, Country, etc. depending of the selected option *****

function ImportList(PageText: string; MultipleValues: Integer; StartTag: string): string;
var
  Value, Value2: string;
begin
  if MultipleValues = 0 then
  begin
    Value := TextBetween(PageText, StartTag, '</a>');
    Value2 := TextAfter(Value, '">');
  end
  else
  begin
    Value := TextBetween(PageText, StartTag, #13#10);
    Value2 := TextBefore(Value, ' <a href="/rg', '');
    if Value2 <> '' then
      Value := Value2;
    Value2 := TextAfter(Value, '">');
    HTMLRemoveTags(Value2);
    if MultipleValues = 1 then
      Value2 := StringReplace(Value2, ' / ', ', ');
  end;
  HTMLDecode(Value2);
  Result := Value2;
end;

// ***** functions to import the different pictures kinds, depending of the option selected by user *****

function ImportSmallPicture(PageText: string): Boolean;
var
  Value: string;
begin
  Result := False;
  Value := TextBetween(PageText, '<img border="0" alt="cover" src="', '"');
  if Value <> '' then
  begin
    GetPicture(Value, False);
    Result := True;
  end;
end;

function ImportLargePicture(Address: string): Boolean;
var
  Value, Value2: string;
begin
  Result := True;
  Value := GetPage(Address);
  if SearchForLargePicture(Value, 'Onesheet_text', False) then
    Exit;
  if SearchForLargePicture(Value, 'Onesheet', False) then
    Exit;
  if SearchForLargePicture(Value, 'poster', False) then
    Exit;
  if SearchForLargePicture(Value, 'keyart01', False) then
    Exit;
  if SearchForLargePicture(Value, 'usposter', False) then
    Exit;
  if SearchForLargePicture(Value, 'text', True) then
    Exit;
  if SearchForLargePicture(Value, 'pos01', True) then
    Exit;
  if SearchForLargePicture(Value, 'KeyArt', True) then
    Exit;
  if SearchForLargePicture(Value, 'heet', True) then // Sheet & Onesheet
    Exit;
  if SearchForLargePicture(Value, 'OneSheetv2', True) then
    Exit;
  if SearchForLargePicture(Value, 'artwork', True) then
    Exit;
  Address := TextBetween(Value, 'There are ' + #13#10 + '<a href="', '">');
  if Address <> '' then
    Result := ImportLargePicture('http://us.imdb.com' + Address)
  else
    Result := False;
end;

function SearchForLargePicture(PageText: string; Name: string; PartialName: Boolean): Boolean;
var
  Value: string;
begin
  Result := False;
  if PartialName then
  begin
    Value := TextBefore(PageText, Name + '.jpg', '/');
    if Value = '' then
      Exit
    else
      Name := Value + Name;
  end;
  Value := TextBefore(PageText, 'th-' + Name + '.jpg', 'src="');
  if Value <> '' then
  begin
    GetPicture(Value + Name + '.jpg', False);
    Result := True;
  end;
end;

function ImportAmazonPicture(PageText: string): Boolean;
var
  Value, Value2: string;
begin
  Result := False;
  Value := TextBefore(PageText, '" title="DVD available', '<a href="');
  if Value = '' then
    Exit;
  PageText := GetPage('http://us.imdb.com' + Value);
  if Pos('unable to find exact matches', PageText) > 0 then
    Exit;
  if Pos('You may also be interested in these items...', PageText) > 0 then
    PageText := TextBefore(PageText, 'You may also be interested in these items...', '');
  Value := TextBefore(PageText, 'TZZZZZZZ.jpg', '<img src="');
  if Value = '' then
    Value := TextBefore(PageText, 'THUMBZZZ.jpg', '<img src="');
  if Value <> '' then
  begin
    GetPicture(Value + 'LZZZZZZZ.jpg', False);
    Result := True;
  end;
end;

// ***** Gets summaries for the movie, based on the plot outline given in parameter (that contains the URL to more summaries) *****

function ImportSummary(PlotText: string): string;
var
  Address, Value, Value2, PageText, Longest: string;
begin
  Address := TextBetween(PlotText, '<a href="/rg/title-tease/plotsummary', '">(more)</a>');
  if (Address = '') or (GetOption('DescriptionSelection') = 0) then
  begin
    Result := Trim(TextBefore(PlotText, '<a href="/rg', ''));
    if Result = '' then
      Result := Trim(PlotText);
    HTMLRemoveTags(Result);
    HTMLDecode(Result);
  end
  else
  begin
    PageText := GetPage('http://us.imdb.com/rg/title-tease/plotsummary' + Address);
    PickListClear;
    Longest := '';
    Value := TextBetween(PageText, '<p class="plotpar">', '</p>');
    PageText := RemainingText;
    while Value <> '' do
    begin
      Value := StringReplace(Value, #13#10, ' ');
      Value := StringReplace(Value, '<br>', #13#10);
      HTMLRemoveTags(Value);
      HTMLDecode(Value);
      while Pos('  ', Value) > 0 do
        Value := StringReplace(Value, '  ', ' ');
      if Length(Value) > Length(Longest) then
        Longest := Value;
      PickListAdd(Trim(Value));
      Value := TextBetween(PageText, '<p class="plotpar">', '</p>');
      PageText := RemainingText;
    end;
    if (GetOption('BatchMode') = 1) or (GetOption('DescriptionSelection') = 2) then
      Result := Longest
    else
    begin
      if not PickListExec('Select a description for "' + GetField(fieldOriginalTitle) + '"', Result) then
        Result := '';
    end;
  end;
end;

// ***** beginning of the program *****

begin
  if CheckVersion(3,4,3) then
  begin
    MovieName := GetField(fieldOriginalTitle);
    if MovieName = '' then
      MovieName := GetField(fieldTranslatedTitle);
    if GetOption('BatchMode') = 0 then
    begin
      if not Input('IMDB Import', 'Enter the title or the IMDB URL of the movie:', MovieName) then
        Exit;
    end;
    if MovieName <> '' then
    begin
      if Pos('imdb.com', MovieName) > 0 then
        AnalyzeResultsPage(MovieName)
      else
      begin
        MovieName := StringReplace(MovieName, '&', 'and');
        if (GetOption('BatchMode') = 1) or (GetOption('PopularSearches') = 1) then
          AnalyzeResultsPage('http://us.imdb.com/find?tt=1;q=' + UrlEncode(MovieName))
        else
          AnalyzeResultsPage('http://us.imdb.com/find?more=tt;q=' + UrlEncode(MovieName));
      end;
      DisplayResults;
    end;
  end
  else
    ShowMessage('This script requires a newer version of Ant Movie Catalog (at least the version 3.4.3)');
end.