zile asked me to improve and correct the current AMG script for 3.5.0 Final version.
I made some modifications and enchancements, also some bugs were corrected.
I could not test it much but seem to OK. If you find a bug please let me know.
KG
Code: Select all
(***************************************************
Ant Movie Catalog importation script
www.antp.be/software/moviecatalog/
[Infos]
Authors=Hubert Kosior, KaraGarga
Title=All Movie Guide
Description=All Movie Guide detailed info import with small picture
Site=http://allmovie.com
Language=EN
Version=0.3 / 04.2005
Requires=3.5.0
Comments=send bugs and reports to: hubert@tm1.net|a bug corrected by Antoine Potten|to do:| - producer's name instad of producing company| - display movie categories when movie list hit (after searching)
License=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. |
GetInfo=1
[Options]
CategoryOptions=3|3|1=Only import firts category|2=Import all categories and divide by "/"|3=Import all categories and divide by ","
ProducerOptions=0|0|0=Import Production Companies into Producer Field|1=Import Theme into Producer Field|2=Import Tones into Producer Field|3=Import Moods into Producer Field|4=Import Releaser Company into Producer Field
SynopsisOptions=2|1|1=Import into Description Field|2=Import into Comments Field|0=DO NOT import Synopsis
ReviewOptions=2|2|1=Import into Description Field|2=Import into Comments Field|0=DO NOT import Review
AwardsOptions=2|2|1=Import into Description Field|2=Import into Comments Field|0=DO NOT import Awards List
CastOptions=3|3|1=Import Cast divided by ";"|2=Import Cast as a list (AMG Default)|3=Import Cast as a list (like IMDB)|4=Import Cast as a list within paranthesis|5=Import Cast within paranthesis
FieldforCredits=2|2|0=DO NOT import Production Credits|1=Import Production Credits into Description Field|2=Import Production Credits into Comments Field
CreditsOptions=2|2|1=Import Credits as a list (like AMG)|2=Import Credits as a list (like IMDB)|3=Import Credits as a list (within paranthesis)
***************************************************)
program AllMovie;
uses
StringUtils1;
var
MovieName: string;
// simple string procedures
function StringReplaceAll(S, Old, New: string): string;
begin
while Pos(Old, S) > 0 do
S := StringReplace(S, Old, New);
Result := S;
end;
procedure CutAfter(var Str: string; Pattern: string);
begin
Str := Copy(str, Pos(Pattern, Str) + Length(Pattern), Length(Str));
end;
procedure CutBefore(var Str: string; Pattern: string);
begin
Str := Copy(Str, Pos(Pattern, Str), Length(Str));
end;
// Loads and analyses page from internet (list of movies or direct hit)
procedure AnalyzePage(Address: string);
var
Page: TStringList;
begin
Page := TStringList.Create;
Page.Text := GetPage(Address);
// movie list
Sleep(500);
if Pos('movie titles like: ', Page.Text) > 0 then
begin
PickTreeClear;
PickTreeAdd('Search results', '');
AddMoviesTitles(Page);
if PickTreeExec(Address) then
AnalyzePage(Address);
// refine search
end
else
if Pos('Sorry, there is too many possible matches, please adjust your search.', Page.Text) > 0 then
begin
ShowMessage('Sorry, there is too many possible matches, please adjust your search.');
if Input('All Movie Import', 'Enter the title of the movie:', MovieName) then
AnalyzePage('http://allmovie.com/cg/avg.dll?p=avg&type=2&srch=' + URLEncode(MovieName));
// direct hit
end
else
begin
if CanSetField(fieldURL) then SetField(FieldURL, Address);
AnalyzeMoviePage(Page)
end;
end;
// Extracts movie details from page
procedure AnalyzeMoviePage(MoviePage: TStringList);
var
Page: string;
Value: string;
begin
Page := MoviePage.Text;
// Original title
if CanSetField(fieldOriginalTitle) then
begin
Value := TextBetween(Page, '<FONT SIZE="+2"><B>', '</B>');
SetField(fieldOriginalTitle, Value);
end;
// Year
if CanSetField(fieldYear) then
begin
SetField(fieldYear, GetStringFromHTML(Page, '<B>'+GetField(fieldOriginalTitle)+'</B>', '</TR>', '</B>'));
end;
// Country
if CanSetField(fieldCountry) then
begin
SetField(fieldCountry, GetStringFromHTML(Page, '<B>'+GetField(fieldOriginalTitle)+'</B>', '<I>', '</I>'));
end;
// Length
if CanSetField(fieldLength) then
begin
SetField(fieldLength, GetStringFromHTML(Page, '<B>'+GetField(fieldOriginalTitle)+'</B>', '</I> - ', ' min'));
end;
// AKA -> translated title
if CanSetField(fieldTranslatedTitle) then
begin
SetField(fieldTranslatedTitle, GetStringFromHTML(Page, '>AKA', '</TD>', '</td>'));
end;
// Rating (multiplied by 2, because 0 <= AMG rating <= 5)
if CanSetField(fieldRating) then
begin
Value := GetStringFromHTML(Page, '>AMG Rating', 'alt="', ' Stars');
if Length(Value) > 0 then
begin
SetField(fieldRating, FloatToStr(StrToFloat(Value)*2));
end;
end;
// Director
if CanSetField(fieldDirector) then
begin
SetField(fieldDirector, GetStringFromHTML(Page, '>Director', '</TD>', '</td>'));
end;
// Genre -> category
if CanSetField(fieldCategory) then
begin
if GetOption('CategoryOptions') = 1 then
Value := TextBetween(Page, 'Genre/Type </TD>', '</A>');
if Value <> '' then
begin
HTMLRemoveTags(Value);
SetField(fieldCategory, Value);
end;
if GetOption('CategoryOptions') = 2 then
Value := TextBetween(Page, 'Genre/Type </TD>', '</td>');
if Value <> '' then
begin
Value := StringReplace(Value, ',', ' /');
HTMLRemoveTags(Value);
SetField(fieldCategory, Value);
end;
if GetOption('CategoryOptions') = 3 then
SetField(fieldCategory, GetStringFromHTML(Page, '>Genre/Type', '</TD>', '</td>'));
end;
// Producing company -> producer
if CanSetField(fieldProducer) then
begin
if GetOption('ProducerOptions') = 0 then
//SetField(fieldProducer, GetStringFromHTML(Page, 'Produced by ', '<TD>', '</TD></TR>'));
Value := TextBetween(Page, 'Produced by ', '</A></TD></TR>');
if Value <> '' then
begin
HTMLRemoveTags(Value);
SetField(fieldProducer, Value);
end;
if GetOption('ProducerOptions') = 1 then
Value := TextBetween(Page, 'Themes ', '</A></td></tr>');
if Value <> '' then
begin
Value := StringReplace(Value, ',', ' /');
HTMLRemoveTags(Value);
SetField(fieldProducer, Value);
end;
if GetOption('ProducerOptions') = 2 then
Value := TextBetween(Page, 'Tones ', '</A></td></tr>');
if Value <> '' then
begin
Value := StringReplace(Value, ',', ' /');
HTMLRemoveTags(Value);
SetField(fieldProducer, Value);
end;
if GetOption('ProducerOptions') = 3 then
Value := TextBetween(Page, 'Moods ', '</A></td></tr>');
if Value <> '' then
begin
Value := StringReplace(Value, ',', ' /');
HTMLRemoveTags(Value);
SetField(fieldProducer, Value);
end;
if GetOption('ProducerOptions') = 4 then
SetField(fieldProducer, GetStringFromHTML(Page, '>Released by', '</TD>', '</TD>'));
end;
// Image
if CanSetPicture then
begin
Value := GetStringFromHTML(Page, 'http://image.allmusic.com', '', '"');
if Length(Value) > 0 then GetPicture(Value);
end;
// Cast -> actors
// adjust semicolons
if CanSetField(fieldActors) then
begin
Value := StringReplaceAll(Page, '</I></TD></TR>', '; ');
Value := GetStringFromHTML(Value, '<A Name="CAST">', '</td></tr>', '</TABLE>');
if Length(Value) > 0 then
begin
// remove double spaces if only actor name given
while Pos(' ', Value) > 0 do
Delete(Value, Pos(' ', Value), 2);
// remove trailing "; "
if Copy(Value, Length(Value) - 1, 2) = '; ' then
Value := Copy(Value, 0, Length(Value) - 2);
if GetOption('CastOptions') = 1 then
begin
SetField(fieldActors, Value);
end;
if GetOption('CastOptions') = 2 then
begin
Value := StringReplace(Value, '; ', #13#10);
HTMLRemoveTags(Value);
SetField(fieldActors, Value);
end;
if GetOption('CastOptions') = 3 then
begin
Value := StringReplace(Value, '; ', #13#10);
Value := StringReplace(Value, '-', '...');
SetField(fieldActors, Value);
end;
if GetOption('CastOptions') = 4 then
begin
Value := StringReplace(Value, '; ', ')'+#13#10);
Value := StringReplace(Value, '-', '(');
SetField(fieldActors, Value);
end;
if GetOption('CastOptions') = 5 then
begin
Value := StringReplace(Value, '; ', '), ');
Value := StringReplace(Value, '-', '(');
SetField(fieldActors, Value);
end;
end;
end;
// Plot synopsis
if CanSetField(fieldComments) or CanSetField(fieldDescription) then
begin
Value := GetStringFromHTML(Page, '<A Name="PLOT">', '</table>', '</table>');
if Length(Value) > 0 then
begin
if GetOption('SynopsisOptions') = 0 then
begin
end;
if GetOption('SynopsisOptions') = 1 then
begin
SetField(fieldDescription, 'AMG SYNOPSIS: '+Value+#13#10+#13#10);
end;
if GetOption('SynopsisOptions') = 2 then
begin
SetField(fieldComments, 'AMG SYNOPSIS: '+Value+#13#10+#13#10);
end;
end;
end;
// Review -> description
if CanSetField(fieldComments) or CanSetField(fieldDescription) then
begin
Value := GetStringFromHTML(Page, '<A Name="REVIEW">', '</table>', '</table>');
if Length(Value) > 0 then
begin
if GetOption('ReviewOptions') = 0 then
begin
end;
if GetOption('ReviewOptions') = 1 then
begin
SetField(fieldDescription, GetField(fieldDescription)+'AMG REVIEW: '+Value+#13#10+#13#10);
end;
if GetOption('ReviewOptions') = 2 then
begin
SetField(fieldComments, GetField(fieldComments)+'AMG REVIEW: '+Value+#13#10+#13#10);
end;
end;
end;
// Awards -> description
// adjust spaces and line feeds
if CanSetField(fieldComments) or CanSetField(fieldDescription) then
begin
Value := StringReplaceAll(Page, '> <FONT', ''); // space before title
Value := StringReplaceAll(Value, '</FONT> </td><td WIDTH=209>', ' - '); // minus before name
Value := StringReplaceAll(Value, ' </A></FONT></td>', ' - '); // minus after name (1)
Value := StringReplaceAll(Value, ' </FONT></td>', ' - '); // minus after name (2)
Value := StringReplaceAll(Value, '</FONT> </td></tr>', + #13#10); // newline after academy name
Value := GetStringFromHTML(Value, '<A Name="AWRD">', '</td></tr>', '</TABLE>');
Value := StringReplaceAll(Value, ' ', ' ');
Value := StringReplaceAll(Value, ' - - ', ' - ');
if Length(Value) > 0 then
begin
if GetOption('AwardsOptions') = 0 then
begin
end;
if GetOption('AwardsOptions') = 1 then
begin
SetField(fieldDescription, GetField(fieldDescription)+'AWARDS:'+#13#10+Value+#13#10);
end;
if GetOption('AwardsOptions') = 2 then
begin
SetField(fieldComments, GetField(fieldComments)+'AWARDS:'+#13#10+Value+#13#10);
end;
end;
end;
// ProductionCredits
// adjust semicolons
if CanSetField(fieldComments) or CanSetField(fieldDescription) then
begin
Value := StringReplaceAll(Page, '</I></TD></TR>', '; ');
Value := GetStringFromHTML(Value, '<A Name="CRED">', '</td></tr>', '</TABLE>');
if Length(Value) > 0 then
begin
// remove double spaces if only actor name given
while Pos(' ', Value) > 0 do
Delete(Value, Pos(' ', Value), 2);
// remove trailing "; "
if Copy(Value, Length(Value) - 1, 2) = '; ' then
Value := Copy(Value, 0, Length(Value) - 2);
if GetOption('FieldforCredits') = 1 then
begin
if GetOption('CreditsOptions') = 1 then
begin
Value := StringReplace(Value, '; ', #13#10);
HTMLRemoveTags(Value);
SetField(fieldDescription, GetField(fieldDescription)+'PRODUCTION CREDITS:'+#13#10+Value);
end;
if GetOption('CreditsOptions') = 2 then
begin
Value := StringReplace(Value, '; ', #13#10);
Value := StringReplace(Value, '-', '...');
SetField(fieldDescription, GetField(fieldDescription)+'PRODUCTION CREDITS:'+#13#10+Value);
end;
if GetOption('CreditsOptions') = 3 then
begin
Value := StringReplace(Value, '; ', ')'+#13#10);
Value := StringReplace(Value, '-', '(');
SetField(fieldDescription, GetField(fieldDescription)+'PRODUCTION CREDITS:'+#13#10+Value+')');
end;
end;
if GetOption('FieldforCredits') = 2 then
begin
if GetOption('CreditsOptions') = 1 then
begin
Value := StringReplace(Value, '; ', #13#10);
HTMLRemoveTags(Value);
SetField(fieldComments, GetField(fieldComments)+'PRODUCTION CREDITS:'+#13#10+Value);
end;
if GetOption('CreditsOptions') = 2 then
begin
Value := StringReplace(Value, '; ', #13#10);
Value := StringReplace(Value, '-', '...');
SetField(fieldComments, GetField(fieldComments)+'PRODUCTION CREDITS:'+#13#10+Value);
end;
if GetOption('CreditsOptions') = 3 then
begin
Value := StringReplace(Value, '; ', ')'+#13#10);
Value := StringReplace(Value, '-', '(');
SetField(fieldComments, GetField(fieldComments)+'PRODUCTION CREDITS:'+#13#10+Value+')');
end;
end;
end;
end;
// remove trailing newline from description or comments
Value := GetField(fieldDescription);
if Copy(Value, Length(Value) - 1, 2) = #13#10 then
begin
Value := Copy(Value, 0, Length(Value) - 2);
SetField(fieldDescription, Value);
end;
Value := GetField(fieldComments);
if Copy(Value, Length(Value) - 1, 2) = #13#10 then
begin
Value := Copy(Value, 0, Length(Value) - 2);
SetField(fieldComments, Value);
end;
end;
// Adds movie titles from search results to tree
procedure AddMoviesTitles(ResultsPage: TStringList);
var
Page: string;
MovieTitle, MovieAddress: string;
begin
Page := ResultsPage.Text;
// Every movie entry begins with string "<A HREF='/cg/avg.dll?"
while Pos('<A HREF="/cg/avg.dll?', Page) > 0 do
begin
CutBefore(Page, '<A HREF="/cg/avg.dll?');
MovieAddress := 'http://allmovie.com' + GetStringFromHTML(Page, '<A', '"', '">');
MovieTitle := GetStringFromHTML(Page, '<A', '', '</tr>');
MovieTitle := StringReplace(MovieTitle, ')', '), ');
CutAfter(Page, '</tr>');
// add movie to list
PickTreeAdd(MovieTitle, MovieAddress);
end;
end;
// Extracts single movie detail (like director, genre) from page
function GetStringFromHTML(Page, StartTag, CutTag, EndTag: string): string;
begin
Result := '';
// recognition tag - if present, extract detail from page, otherwise assume detail is not present
if Pos(StartTag, Page) > 0 then begin
CutBefore(Page, StartTag);
// optional cut tag helps finding right string in html page
if Length(CutTag) > 0 then
CutAfter(Page, CutTag);
// movie detail copied with html tags up to end string
Result := Copy(Page, 0, Pos(EndTag, Page) - 1);
// remove html tags and decode html string
HTMLRemoveTags(Result);
HTMLDecode(Result);
// ShowMessage('DEBUG: GetStringFromHTML - StartTag "'+StartTag+'", CutTag "'+CutTag+'", EndTag "'+EndTag+'", Result "'+Result+'" ___ '+Page);
end;
end;
procedure RemovePronoun(var Str: string);
var
i: Integer;
s: string;
c: char;
begin
// remove pronouns
s := UpperCase(Copy(Str, 0, 4));
if (s = 'LES ') or (s = 'UNE ') or (s = 'THE ') then
Str := Copy(Str, 5, Length(Str) - 4)
else
begin
s := Copy(s, 0, 3);
if (s = 'LE ') or (s = 'UN ') then
Str := Copy(Str, 4, Length(Str) - 3)
else
begin
s := Copy(s, 0, 2);
if (s = 'L''') or (s = 'L ') or (s = 'A ') then
Str := Copy(Str, 3, Length(Str) - 2)
end;
end;
// remove non-letters, non-digits and non-spaces
s := '';
for i := 1 to Length(Str) do begin
c := StrGet(Str, i);
if ((c<'a') or (c>'z')) and
((c<'A') or (c>'Z')) and
((c<'0') or (c>'9')) and
(c<>' ') then
else
s := s + Copy(Str, i, 1);
end;
Str := s;
end;
begin
if CheckVersion(3,5,0) then begin
MovieName := GetField(fieldOriginalTitle);
if MovieName = '' then MovieName := GetField(fieldTranslatedTitle);
if Input('All Movie Import', 'Enter title (only letters, digits and spaces):', MovieName) then
begin
if Pos('allmovie.com', MovieName) > 0 then
AnalyzePage(MovieName)
else
begin
RemovePronoun(MovieName);
AnalyzePage('http://allmovie.com/cg/avg.dll?p=avg&type=2&srch=' + StringReplace(URLEncode(MovieName), '%20', '+'));
end;
end;
end else
ShowMessage('This script requires a newer version of Ant Movie Catalog (at least the version 3.5.0)');
end.
