PythonでSGMファイルを利用したい

前提・実現したいこと

Pythonで文書分類をやろうとしているのですが、ダウンロードしてきたreuters21578のデータをPython上で扱えるようにするのに苦戦しています。BeautifulSoupでやってみてますが、他に良い方法があれば教えていただきたいです。

発生している問題・エラーメッセージ

BeautifulSoupのメソッドでトピックが存在する記事のtopicsタグをリストにしようとすると、以下のようにトピックが存在しないものまでリストに追加されてしまいます。

該当のソースコード

Python
1from bs4 import BeautifulSoup
2
3with open ("/content/drive/My Drive/antiword/lewis.dtd") as f:
4    dtd = f.read()
5with open ("/content/drive/My Drive/antiword/reut2-000.sgm") as f:
6    soup = BeautifulSoup(dtd + f.read(), "lxml")
7
8topics = []
9for article in soup.find_all(topics="YES"):
10    topics.append(article.find("topics"))
11print(topics)

SGML
1<!DOCTYPE lewis SYSTEM "lewis.dtd">
2<REUTERS TOPICS="YES" LEWISSPLIT="TRAIN" CGISPLIT="TRAINING-SET" OLDID="5544" NEWID="1">
3<DATE>26-FEB-1987 15:01:01.79</DATE>
4<TOPICS><D>cocoa</D></TOPICS>
5<PLACES><D>el-salvador</D><D>usa</D><D>uruguay</D></PLACES>
6<PEOPLE></PEOPLE>
7<ORGS></ORGS>
8<EXCHANGES></EXCHANGES>
9<COMPANIES></COMPANIES>
10<UNKNOWN> 
11&#5;&#5;&#5;C T
12&#22;&#22;&#1;f0704&#31;reute
13u f BC-BAHIA-COCOA-REVIEW   02-26 0105</UNKNOWN>
14<TEXT>&#2;
15<TITLE>BAHIA COCOA REVIEW</TITLE>
16<DATELINE>    SALVADOR, Feb 26 - </DATELINE><BODY>Showers continued throughout the week in
17the Bahia cocoa zone, alleviating the drought since early
18January and improving prospects for the coming temporao,
19although normal humidity levels have not been restored,
20Comissaria Smith said in its weekly review.
21    The dry period means the temporao will be late this year.
22    Arrivals for the week ended February 22 were 155,221 bags
23of 60 kilos making a cumulative total for the season of 5.93
24mln against 5.81 at the same stage last year. Again it seems
25that cocoa delivered earlier on consignment was included in the
26arrivals figures.
27    Comissaria Smith said there is still some doubt as to how
28much old crop cocoa is still available as harvesting has
29practically come to an end. With total Bahia crop estimates
30around 6.4 mln bags and sales standing at almost 6.2 mln there
31are a few hundred thousand bags still in the hands of farmers,
32middlemen, exporters and processors.
33    There are doubts as to how much of this cocoa would be fit
34for export as shippers are now experiencing dificulties in
35obtaining +Bahia superior+ certificates.
36    In view of the lower quality over recent weeks farmers have
37sold a good part of their cocoa held on consignment.
38    Comissaria Smith said spot bean prices rose to 340 to 350
39cruzados per arroba of 15 kilos.
40    Bean shippers were reluctant to offer nearby shipment and
41only limited sales were booked for March shipment at 1,750 to
421,780 dlrs per tonne to ports to be named.
43    New crop sales were also light and all to open ports with
44June/July going at 1,850 and 1,880 dlrs and at 35 and 45 dlrs
45under New York july, Aug/Sept at 1,870, 1,875 and 1,880 dlrs
46per tonne FOB.
47    Routine sales of butter were made. March/April sold at
484,340, 4,345 and 4,350 dlrs.
49    April/May butter went at 2.27 times New York May, June/July
50at 4,400 and 4,415 dlrs, Aug/Sept at 4,351 to 4,450 dlrs and at
512.27 and 2.28 times New York Sept and Oct/Dec at 4,480 dlrs and
522.27 times New York Dec, Comissaria Smith said.
53    Destinations were the U.S., Covertible currency areas,
54Uruguay and open ports.
55    Cake sales were registered at 785 to 995 dlrs for
56March/April, 785 dlrs for May, 753 dlrs for Aug and 0.39 times
57New York Dec for Oct/Dec.
58    Buyers were the U.S., Argentina, Uruguay and convertible
59currency areas.
60    Liquor sales were limited with March/April selling at 2,325
61and 2,380 dlrs, June/July at 2,375 dlrs and at 1.25 times New
62York July, Aug/Sept at 2,400 dlrs and at 1.25 times New York
63Sept and Oct/Dec at 1.25 times New York Dec, Comissaria Smith
64said.
65    Total Bahia sales are currently estimated at 6.13 mln bags
66against the 1986/87 crop and 1.06 mln bags against the 1987/88
67crop.
68    Final figures for the period to February 28 are expected to
69be published by the Brazilian Cocoa Trade Commission after
70carnival which ends midday on February 27.
71 Reuter
72&#3;</BODY></TEXT>
73</REUTERS>