Commit 1ccb3142046f2295874cfe9983db47e601f57f89
1 parent
87f27441
Fix minor docker issues.
Showing
2 changed files
with
17 additions
and
9 deletions
Dockerfile.marcell
... | ... | @@ -84,6 +84,12 @@ WORKDIR $APP_HOME |
84 | 84 | # copy django settings |
85 | 85 | COPY ./collector/collector/docker-settings.py $APP_HOME/collector/collector/settings.py |
86 | 86 | |
87 | +# download and unpack db init file | |
88 | +WORKDIR $APP_HOME/resources | |
89 | +RUN wget https://manage.legis.nlp.ipipan.waw.pl/download/marcell/marcell-init.db.tar.gz && \ | |
90 | + tar -xvf ./marcell-init.db.tar.gz && \ | |
91 | + rm ./marcell-init.db.tar.gz | |
92 | + | |
87 | 93 | # chown all the files to the collector user |
88 | 94 | RUN chown -R collector:collector $HOME |
89 | 95 | |
... | ... | @@ -92,14 +98,15 @@ USER postgres |
92 | 98 | RUN /etc/init.d/postgresql start && \ |
93 | 99 | psql --command "CREATE USER collector WITH SUPERUSER PASSWORD 'collector';" && \ |
94 | 100 | createdb -O collector collector && \ |
95 | - wget https://manage.legis.nlp.ipipan.waw.pl/download/marcell/marcell-init.db.tar.gz && \ | |
96 | - tar -xvf ./marcell-init.db.tar.gz && \ | |
97 | - rm ./marcell-init.db.tar.gz && \ | |
98 | - psql collector < ./marcell-init.db && \ | |
99 | - rm ./marcell-init.db | |
101 | + psql collector < $APP_HOME/resources/marcell-init.db | |
102 | + | |
103 | +# remove db init file | |
104 | +USER collector | |
105 | +RUN rm $APP_HOME/resources/marcell-init.db | |
100 | 106 | |
101 | 107 | # change to the root user |
102 | 108 | USER 0 |
103 | 109 | |
104 | 110 | # run entrypoint-marcell.sh |
105 | -ENTRYPOINT ["/home/collector/app/entrypoint-marcell.sh"] | |
111 | +WORKDIR $APP_HOME | |
112 | +ENTRYPOINT ["/home/collector/app/entrypoint-marcell.sh"] | |
106 | 113 | \ No newline at end of file |
... | ... |
collector/loaders/marcell_rest.py
... | ... | @@ -83,9 +83,10 @@ def _create_document(pipeline, doc_id, metadata, text_file): |
83 | 83 | sequence=metadata_sequence) |
84 | 84 | metadata_sequence += 1 |
85 | 85 | |
86 | - for label in metadata['keywords']: | |
87 | - keyword_obj, _ = Keyword.objects.get_or_create(label=label) | |
88 | - document.keywords.add(keyword_obj) | |
86 | + if 'keywords' in metadata: | |
87 | + for label in metadata['keywords']: | |
88 | + keyword_obj, _ = Keyword.objects.get_or_create(label=label) | |
89 | + document.keywords.add(keyword_obj) | |
89 | 90 | |
90 | 91 | return document |
91 | 92 | |
... | ... |