✨ Move processors into their own Docker container
parent
894cbeee32
commit
02eae7b6fa
@ -1,86 +0,0 @@
|
||||
FROM ubuntu:18.04
|
||||
|
||||
ENV LC_ALL=C.UTF-8 LANG=C.UTF-8
|
||||
ENV PIP_DEFAULT_TIMEOUT=120
|
||||
|
||||
ENV OCRD_OLENA_VERSION 1.2.0
|
||||
ENV TESSDATA_BEST_VERSION 4.0.0
|
||||
ENV TESSDATA_PREFIX /usr/local/share/tessdata
|
||||
|
||||
|
||||
RUN echo "APT::Acquire::Retries \"3\";" > /etc/apt/apt.conf.d/80-retries && \
|
||||
apt-get update && \
|
||||
apt-get install -y \
|
||||
curl xz-utils \
|
||||
python3-pip \
|
||||
git \
|
||||
software-properties-common \
|
||||
# For clstm on Ubuntu 19.04:
|
||||
swig libeigen3-dev libpng-dev libprotobuf-dev \
|
||||
# For cv2:
|
||||
libsm6 libxrender1 \
|
||||
# For ocrd_olena:
|
||||
imagemagick \
|
||||
# XML utils
|
||||
libxml2-utils \
|
||||
xmlstarlet \
|
||||
&& \
|
||||
apt-get clean && rm -rf /var/lib/apt/lists/*
|
||||
|
||||
|
||||
# Install Leptonica and Tesseract.
|
||||
RUN add-apt-repository ppa:alex-p/tesseract-ocr && \
|
||||
apt-get update && \
|
||||
apt-get install -y \
|
||||
tesseract-ocr \
|
||||
libtesseract-dev \
|
||||
&& \
|
||||
apt-get clean && rm -rf /var/lib/apt/lists/*
|
||||
|
||||
# Set up OCR-D logging
|
||||
COPY ocrd_logging.py /etc/
|
||||
|
||||
|
||||
# Build ocrd_olena
|
||||
# XXX .deb needs an update
|
||||
RUN curl -sSL --retry 3 -O https://qurator-data.de/~mike.gerber/olena_2.1-0+ocrd-git/olena-bin_2.1-0+ocrd-git_amd64.deb && \
|
||||
dpkg -i --force-depends olena-bin_2.1-0+ocrd-git_amd64.deb && \
|
||||
rm -f olena-bin_2.1-0+ocrd-git_amd64.deb && \
|
||||
apt-get update && \
|
||||
apt-get -f install -y && \
|
||||
apt-get clean && rm -rf /var/lib/apt/lists/*
|
||||
RUN pip3 install --no-cache-dir --upgrade pip && \
|
||||
curl -sSL --retry 3 -o ocrd_olena.tar.gz https://github.com/OCR-D/ocrd_olena/archive/v${OCRD_OLENA_VERSION}.tar.gz && \
|
||||
mkdir ocrd_olena && \
|
||||
tar xvz -C ocrd_olena --strip-components=1 -f ocrd_olena.tar.gz && \
|
||||
cd ocrd_olena && \
|
||||
sed -i 's/^install: deps$/install:/' Makefile && \
|
||||
pip3 install --no-cache-dir --use-feature=2020-resolver ocrd && \
|
||||
make install PREFIX=/usr/local && \
|
||||
cd .. && rm -rf ocrd_olena ocrd_olena.tar.gz
|
||||
|
||||
|
||||
# Copy OCR models
|
||||
RUN mkdir -p /var/lib/calamari-models
|
||||
COPY data/calamari-models/GT4HistOCR /var/lib/calamari-models/GT4HistOCR
|
||||
RUN mkdir -p $TESSDATA_PREFIX
|
||||
ADD data/mirror/github.com/tesseract-ocr/tessdata_best/archive/${TESSDATA_BEST_VERSION}-repacked.tar.gz $TESSDATA_PREFIX/
|
||||
COPY data/tesseract-models/GT4HistOCR/GT4HistOCR_2000000.traineddata $TESSDATA_PREFIX/
|
||||
COPY data/textline_detection /var/lib/textline_detection
|
||||
|
||||
|
||||
# Install requirements
|
||||
# Using pipdeptree here to get more info than from pip3 check
|
||||
COPY requirements.txt /tmp/
|
||||
RUN pip3 install --no-cache-dir --upgrade pip && \
|
||||
pip3 install --no-cache-dir --use-feature=2020-resolver -r /tmp/requirements.txt && \
|
||||
pip3 install --no-cache-dir pipdeptree && \
|
||||
pipdeptree -w fail
|
||||
|
||||
|
||||
COPY my_ocrd_workflow /usr/bin/
|
||||
COPY xsd/* /usr/share/xml/
|
||||
|
||||
|
||||
WORKDIR /data
|
||||
ENTRYPOINT ["/usr/bin/my_ocrd_workflow"]
|
||||
@ -0,0 +1,41 @@
|
||||
FROM ubuntu:18.04
|
||||
|
||||
ENV LC_ALL=C.UTF-8 LANG=C.UTF-8
|
||||
ENV PIP_DEFAULT_TIMEOUT=120
|
||||
|
||||
|
||||
RUN echo "APT::Acquire::Retries \"3\";" > /etc/apt/apt.conf.d/80-retries && \
|
||||
apt-get update && \
|
||||
apt-get install -y \
|
||||
curl xz-utils \
|
||||
python3-pip \
|
||||
# For add-apt-repository:
|
||||
software-properties-common \
|
||||
# XML utils
|
||||
libxml2-utils \
|
||||
xmlstarlet \
|
||||
&& \
|
||||
apt-get clean && \
|
||||
rm -rf /var/lib/apt/lists/*
|
||||
|
||||
|
||||
# Set up OCR-D logging
|
||||
COPY ocrd_logging.py /etc/
|
||||
|
||||
|
||||
# Build pip installable stuff
|
||||
RUN pip3 install --no-cache-dir --upgrade pip && \
|
||||
pip3 install --no-cache-dir \
|
||||
# Resolve conflicts early:
|
||||
'setuptools >= 41.0.0' \
|
||||
'ocrd >= 2.13.1'
|
||||
|
||||
|
||||
# Check pip dependencies
|
||||
RUN pip3 check
|
||||
|
||||
|
||||
WORKDIR /data
|
||||
|
||||
# Default command
|
||||
CMD ['ocrd']
|
||||
@ -0,0 +1,18 @@
|
||||
FROM boxed-base
|
||||
|
||||
|
||||
ENV DINGLEHOPPER_COMMIT 2b98f69
|
||||
|
||||
|
||||
# Build pip installable stuff
|
||||
RUN pip3 install --no-cache-dir \
|
||||
# Now the real stuff:
|
||||
https://github.com/qurator-spk/dinglehopper/archive/$DINGLEHOPPER_COMMIT.tar.gz
|
||||
|
||||
|
||||
# Check pip dependencies
|
||||
RUN pip3 check
|
||||
|
||||
|
||||
# Default command
|
||||
CMD ["ocrd-dinglehopper"]
|
||||
@ -0,0 +1,24 @@
|
||||
FROM boxed-base
|
||||
|
||||
|
||||
# Build pip installable stuff
|
||||
RUN pip3 install --no-cache-dir \
|
||||
# Resolve conflicts early:
|
||||
'tensorflow-gpu == 1.15.*' \
|
||||
'calamari-ocr == 0.3.5' \
|
||||
# Now the real stuff:
|
||||
'ocrd_calamari >= 0.0.7'
|
||||
|
||||
|
||||
# Copy OCR models
|
||||
RUN mkdir -p /var/lib/calamari-models
|
||||
COPY data/calamari-models/GT4HistOCR /var/lib/calamari-models/GT4HistOCR
|
||||
|
||||
|
||||
|
||||
# Check pip dependencies
|
||||
RUN pip3 check
|
||||
|
||||
|
||||
# Default command
|
||||
CMD ["ocrd-calamari-recognize"]
|
||||
@ -0,0 +1,33 @@
|
||||
FROM boxed-base
|
||||
|
||||
ENV OCRD_OLENA_VERSION 1.2.0
|
||||
|
||||
# Build ocrd_olena
|
||||
RUN apt-get update && \
|
||||
apt-get install -y \
|
||||
imagemagick \
|
||||
&& \
|
||||
apt-get clean && rm -rf /var/lib/apt/lists/*
|
||||
RUN curl -sSL --retry 3 -O https://qurator-data.de/~mike.gerber/olena_2.1-0+ocrd-git/olena-bin_2.1-0+ocrd-git_amd64.deb && \
|
||||
dpkg -i --force-depends olena-bin_2.1-0+ocrd-git_amd64.deb && \
|
||||
rm -f olena-bin_2.1-0+ocrd-git_amd64.deb && \
|
||||
apt-get update && \
|
||||
apt-get -f install -y && \
|
||||
apt-get clean && rm -rf /var/lib/apt/lists/*
|
||||
RUN pip3 install --no-cache-dir --upgrade pip && \
|
||||
curl -sSL --retry 3 -o ocrd_olena.tar.gz https://github.com/OCR-D/ocrd_olena/archive/v${OCRD_OLENA_VERSION}.tar.gz && \
|
||||
mkdir ocrd_olena && \
|
||||
tar xvz -C ocrd_olena --strip-components=1 -f ocrd_olena.tar.gz && \
|
||||
cd ocrd_olena && \
|
||||
sed -i 's/^install: deps$/install:/' Makefile && \
|
||||
pip3 install --no-cache-dir --use-feature=2020-resolver ocrd && \
|
||||
make install PREFIX=/usr/local && \
|
||||
cd .. && rm -rf ocrd_olena ocrd_olena.tar.gz
|
||||
|
||||
|
||||
# Check pip dependencies
|
||||
RUN pip3 check
|
||||
|
||||
|
||||
# Default command
|
||||
CMD ['ocrd-olena-binarize']
|
||||
@ -0,0 +1,35 @@
|
||||
FROM boxed-base
|
||||
|
||||
|
||||
ENV TESSDATA_BEST_VERSION 4.0.0
|
||||
ENV TESSDATA_PREFIX /usr/local/share/tessdata
|
||||
|
||||
|
||||
# Install Leptonica and Tesseract.
|
||||
RUN add-apt-repository ppa:alex-p/tesseract-ocr && \
|
||||
apt-get update && \
|
||||
apt-get install -y \
|
||||
tesseract-ocr \
|
||||
libtesseract-dev \
|
||||
&& \
|
||||
apt-get clean && rm -rf /var/lib/apt/lists/*
|
||||
|
||||
|
||||
# Copy OCR models
|
||||
RUN mkdir -p $TESSDATA_PREFIX
|
||||
ADD data/mirror/github.com/tesseract-ocr/tessdata_best/archive/${TESSDATA_BEST_VERSION}-repacked.tar.gz $TESSDATA_PREFIX/
|
||||
COPY data/tesseract-models/GT4HistOCR/GT4HistOCR_2000000.traineddata $TESSDATA_PREFIX/
|
||||
|
||||
|
||||
# Build pip installable stuff
|
||||
RUN pip3 install --no-cache-dir \
|
||||
# Now the real stuff:
|
||||
'ocrd_tesserocr >= 0.9.0'
|
||||
|
||||
|
||||
# Check pip dependencies
|
||||
RUN pip3 check
|
||||
|
||||
|
||||
# Default command
|
||||
CMD ["ocrd-tesserocr-recognize"]
|
||||
@ -0,0 +1,22 @@
|
||||
FROM boxed-base
|
||||
|
||||
|
||||
ENV SBB_TEXTLINE_DETECTOR_COMMIT 8b01d9e
|
||||
|
||||
|
||||
# Build pip installable stuff
|
||||
RUN pip3 install --no-cache-dir \
|
||||
# Now the real stuff:
|
||||
https://github.com/qurator-spk/sbb_textline_detector/archive/$SBB_TEXTLINE_DETECTOR_COMMIT.tar.gz
|
||||
|
||||
|
||||
# Copy OCR models
|
||||
COPY data/textline_detection /var/lib/textline_detection
|
||||
|
||||
|
||||
# Check pip dependencies
|
||||
RUN pip3 check
|
||||
|
||||
|
||||
# Default command
|
||||
CMD ["ocrd-sbb-textline-detector"]
|
||||
@ -1,13 +0,0 @@
|
||||
tensorflow-gpu < 2.0 # Needed for sbb_text_linedetector
|
||||
|
||||
ocrd >= 2.13.1
|
||||
|
||||
# XXX See https://github.com/OCR-D/ocrd_tesserocr/issues/135
|
||||
# ocrd_tesserocr >= 0.8.XXX
|
||||
https://github.com/mikegerber/ocrd_tesserocr/archive/fix/set-pcgtsid.tar.gz
|
||||
|
||||
ocrd_calamari >= 0.0.7
|
||||
|
||||
https://github.com/qurator-spk/sbb_textline_detector/archive/8b01d9e.tar.gz
|
||||
|
||||
https://github.com/qurator-spk/dinglehopper/archive/2b98f69.tar.gz
|
||||
@ -1,31 +1,42 @@
|
||||
#!/bin/sh
|
||||
# Run the my_ocrd_workflow container on the current workspace
|
||||
#!/bin/bash
|
||||
|
||||
set -e # Abort on error
|
||||
|
||||
DOCKER_IMAGE=${DOCKER_IMAGE:-my_ocrd_workflow:latest} # default to locally built
|
||||
|
||||
if echo "$DOCKER_IMAGE" | grep -q "/"; then
|
||||
docker pull "$DOCKER_IMAGE"
|
||||
fi
|
||||
self=`realpath $0`
|
||||
self_dir=`dirname "$self"`
|
||||
|
||||
|
||||
# XXX Work around podman vs docker uid behaviour
|
||||
# Docker run options
|
||||
docker_run_options="--rm -t"
|
||||
docker_run_options="$docker_run_options --mount type=bind,src=\"$(pwd)\",target=/data"
|
||||
# In podman, the container always runs as the real user == uid 0 in container
|
||||
if docker -v 2>&1 | grep -q podman; then
|
||||
user="0:0"
|
||||
else
|
||||
user="`id -u`:`id -g`"
|
||||
fi
|
||||
|
||||
|
||||
# The container currently needs to run privileged to allow it to read from e.g.
|
||||
docker_run_options="$docker_run_options --user $user"
|
||||
docker_run_options="$docker_run_options -e LOG_LEVEL=$LOG_LEVEL"
|
||||
# The containers currently need to run privileged to allow it to read from e.g.
|
||||
# /home on SELinux secured systems such as Fedora. We might want to use udica
|
||||
# instead in the future.
|
||||
docker_run_options="$docker_run_options --privileged=true"
|
||||
|
||||
|
||||
# Build aliases for the containerized ocrd processors
|
||||
build_alias() {
|
||||
local command=$1
|
||||
local docker_image=$2
|
||||
|
||||
alias $command="docker run $docker_run_options $docker_image $command"
|
||||
}
|
||||
shopt -s expand_aliases # Required for non-interactive shells
|
||||
build_alias ocrd boxed-base
|
||||
build_alias ocrd-olena-binarize boxed-ocrd_olena
|
||||
build_alias ocrd-sbb-textline-detector boxed-sbb_textline_detector
|
||||
build_alias ocrd-calamari-recognize boxed-ocrd_calamari
|
||||
build_alias ocrd-tesserocr-recognize boxed-ocrd_tesserocr
|
||||
build_alias ocrd-dinglehopper boxed-dinglehopper
|
||||
|
||||
|
||||
docker run --privileged=true --rm -t \
|
||||
\
|
||||
--user $user \
|
||||
--mount type=bind,src="$(pwd)",target=/data \
|
||||
\
|
||||
-e LOG_LEVEL=$LOG_LEVEL \
|
||||
$DOCKER_IMAGE "$@"
|
||||
. $self_dir/my_ocrd_workflow
|
||||
|
||||
@ -1,4 +0,0 @@
|
||||
#!/bin/sh
|
||||
# Run the my_ocrd_workflow container on the current workspace
|
||||
|
||||
DOCKER_IMAGE=mikegerber/my_ocrd_workflow:stable `dirname $0`/run "$@"
|
||||
Loading…
Reference in New Issue