Dockerfile.base

FROM python:3.9-bookworm
ENV PYTHONUNBUFFERED 1
ENV DEBIAN_FRONTEND=noninteractive

RUN set -e \
 && echo 'deb http://deb.debian.org/debian bookworm non-free' >> /etc/apt/sources.list \
 && ( apt-get update -y || sleep 15 && apt-get update -y) \
 && apt-get install -y --no-install-recommends \
     python3 \
     python3-dev \
     python3-pip \
     python3-numpy \
     python3-icu \
     libicu-dev \
     pkg-config \
     build-essential \
     p7zip-full p7zip-rar \
     cpanminus \
     libgsf-1-dev \
     postgresql-client \
     imagemagick \
     tesseract-ocr \
     tesseract-ocr-all \
     libtiff-tools \
     ghostscript qpdf \
     wget curl git file \
     gpg gpg-agent \
     gosu tini \
     s3fs pdftk poppler-utils ghostscript nodejs npm qpdf \
     davfs2 \
     cuneiform \
 && cpanm --notest Email::Outlook::Message \
 && apt-get clean && rm -rf /var/lib/apt/lists/* \
 && pip3 install --upgrade pip \
 && pip3 install pipenv

# disable tesseract multithreading
ENV OMP_THREAD_LIMIT 1

# download libpst
ENV LIBPST_VERSION libpst-0.6.76
ENV LIBPST_URL https://www.five-ten-sg.com/libpst/packages/$LIBPST_VERSION.tar.gz
# ENV LIBPST_URL https://github.com/liquidinvestigations/snoop-deps/raw/master/$LIBPST_VERSION.tar.gz
RUN wget $LIBPST_URL --progress=dot:giga \
  && tar zxvf $LIBPST_VERSION.tar.gz \
  && rm -f $LIBPST_VERSION.tar.gz \
  && mv $LIBPST_VERSION /opt/libpst \
  && cd /opt/libpst \
  && ./configure --disable-python --prefix="`pwd`" \
  && make \
  && make install
ENV PATH="/opt/libpst/bin:${PATH}"

# download recent version of 7z
ENV SEVEN_Z_VERSION 7z2107-linux-x64
RUN wget https://github.com/liquidinvestigations/snoop-deps/raw/master/$SEVEN_Z_VERSION.tar.xz --progress=dot:giga \
  && tar xvf $SEVEN_Z_VERSION.tar.xz --one-top-level=7z \
  && rm -f $SEVEN_Z_VERSION.tar.xz \
  && mkdir /opt/7z \
  && mv 7z/7zz /opt/7z \
  && rm -r 7z
ENV PATH="/opt/7z:${PATH}"

# install pdf2pdfocr
RUN pip3 install --upgrade Pillow reportlab packaging psutil Pillow reportlab \
 && pip3 install --upgrade lxml beautifulsoup4 \
 && pip3 install --upgrade wheel \
 && pip3 install PyPDF2==1.27.12
RUN rm /etc/ImageMagick-6/policy.xml
RUN git clone https://github.com/liquidinvestigations/pdf2pdfocr --branch master2 /opt/pdf2pdfocr \
 && cd /opt/pdf2pdfocr \
 && ./install_command

# download others
ADD https://github.com/liquidinvestigations/snoop-deps/raw/master/wait_v2.3.0 /wait
RUN chmod +x /wait
# Download & Install TINI
ADD https://github.com/liquidinvestigations/snoop-deps/raw/master/tini_v0.19.0 /tini
RUN chmod +x /tini

# extra python libs
RUN pip install -e 'git+https://github.com/liquidinvestigations/pycld2.git#egg=pycld2'

# install PDF tools
RUN mkdir -p /opt/hoover/snoop/pdf-tools
ADD ./pdf-tools /opt/hoover/snoop/pdf-tools
RUN bash /opt/hoover/snoop/pdf-tools/install.sh

# install snoop libs
RUN mkdir -p /opt/hoover/snoop
WORKDIR /opt/hoover/snoop
ADD Pipfile Pipfile.lock ./
RUN pipenv install --system --deploy --ignore-pipfile