Merge pull request #5 from dlcs/feature/update_deps

Update aws pip packages + multi stage build

Dockerfile

@@ -1,20 +1,30 @@
+FROM debian:bullseye as build
+
+# avoid issue with packages requiring interaction (e.g. tzdata)
+ARG DEBIAN_FRONTEND=noninteractive
+RUN apt-get update && apt-get install -y wget cmake clang git autoconf pkg-config
+
+# Change submodule to https as we're cloning only. Avoids issues with ssh
+# 8bb209c0c21476ee904a is 0.4 with some bugfixes
+RUN mkdir /home/pdfalto && cd /home/pdfalto \
+    && git clone https://github.com/kermitt2/pdfalto.git && cd pdfalto && git checkout 8bb209c0c21476ee904a && ./install_deps.sh \
+    && git submodule set-url xpdf-4.03 https://github.com/kermitt2/xpdf-4.03.git && git submodule update --init --recursive \
+    && cmake ./ && make
+
 FROM python:3.11-slim
 
 LABEL maintainer="Donald Gray <[email protected]>"
 LABEL org.opencontainers.image.source=https://github.com/dlcs/pdf-to-alto
 LABEL org.opencontainers.image.description="Extract ALTO from PDF"
 
-COPY /deps/pdfalto /usr/bin/pdfalto
+COPY --from=build  /home/pdfalto/pdfalto/pdfalto /usr/bin/pdfalto
 
 COPY requirements.txt /opt/app/requirements.txt
 
 WORKDIR /opt/app
 RUN pip install --no-cache-dir -r requirements.txt
-
-COPY app /opt/app/app
-COPY monitor.py /opt/app/monitor.py
-COPY wait-for-localstack.sh /opt/app/wait-for-localstack.sh
+COPY . /opt/app
 
 RUN chmod +x wait-for-localstack.sh
 
-CMD ["python3", "/opt/app/monitor.py"]
+CMD ["python3", "/opt/app/monitor.py"]

requirements.txt

@@ -1,5 +1,5 @@
-boto3==1.20.33
-botocore==1.23.33
+boto3==1.29.2
+botocore==1.32.2
 certifi==2021.10.8
 charset-normalizer==2.0.10
 colorama==0.4.4
@@ -11,6 +11,6 @@ pycryptodome==3.12.0
 PyMuPDF==1.22.5
 python-dateutil==2.8.2
 requests==2.27.1
-s3transfer==0.5.0
+s3transfer==0.7.0
 six==1.16.0
 urllib3==1.26.8