🚧 WIP: Migrate to using ocrd:all image + Update tests

migrate-to-ocrd_all
Gerber, Mike 6 months ago
parent fc911f3734
commit 699023c084

@ -1,70 +0,0 @@
FROM ubuntu:22.04
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_VERSION_MINIMUM="2.47.0"
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 \
build-essential \
curl \
git \
xz-utils \
pkg-config \
# For add-apt-repository:
software-properties-common \
# XML utils
libxml2-utils \
xmlstarlet \
# OCR-D uses ImageMagick for pixel density estimation
imagemagick \
# pyenv builds
# TODO: builder container?
libz-dev \
libssl-dev \
libbz2-dev \
liblzma-dev \
libncurses-dev \
libffi-dev \
libreadline-dev \
libsqlite3-dev \
libmagic-dev \
&& \
apt-get clean && \
rm -rf /var/lib/apt/lists/*
# Set up OCR-D logging
RUN echo "setOverrideLogLevel(os.getenv('LOG_LEVEL', 'INFO'))" >/etc/ocrd_logging.py
# Install pyenv
# TODO: do not run as root
# TODO: does just saying "3.7" work as intended?
ENV HOME=/root
ENV PYENV_ROOT=/usr/local/share/pyenv
ENV PATH=$PYENV_ROOT/shims:$PYENV_ROOT/bin:$PATH
RUN \
git clone --depth=1 https://github.com/yyuu/pyenv.git $PYENV_ROOT && \
pyenv install 3.7 && \
pyenv global 3.7 && \
pyenv rehash && \
pip install -U pip wheel && \
pip install setuptools
# Install pip installable-stuff
RUN ${PIP_INSTALL} \
"ocrd >= ${OCRD_VERSION_MINIMUM}"
# Check pip dependencies
RUN pip check
WORKDIR /data
# Default command
CMD ['ocrd']

@ -1,18 +0,0 @@
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG DINGLEHOPPER_VERSION="0.9.2"
# Build pip installable stuff
RUN ${PIP_INSTALL} \
"dinglehopper == $DINGLEHOPPER_VERSION"
# Check pip dependencies
RUN pip check
# Default command
CMD ["ocrd-dinglehopper"]

@ -1,18 +0,0 @@
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda12.1:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG EYNOLLAH_VERSION="0.3.0"
# Build pip installable stuff
RUN ${PIP_INSTALL} \
"eynollah == ${EYNOLLAH_VERSION}"
# Check pip dependencies
RUN pip check
# Default command
CMD ["ocrd-eynollah-segment"]

@ -1,24 +0,0 @@
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_FILEFORMAT_VERSION="0.5.0"
RUN apt-get update && \
apt-get install -y \
git \
openjdk-11-jdk-headless \
wget \
unzip \
&& \
apt-get clean && rm -rf /var/lib/apt/lists/*
RUN git clone --depth 1 --branch v${OCRD_FILEFORMAT_VERSION} https://github.com/OCR-D/ocrd_fileformat.git && \
cd ocrd_fileformat/ && \
git submodule update --init && \
make install-fileformat install PREFIX=/usr/local && \
cd .. && rm -rf ocrd_fileformat/
# Default command
CMD ['ocrd-fileformat-transform']

@ -1,39 +0,0 @@
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_OLENA_VERSION="1.3.0"
# Build ocrd_olena
RUN apt-get update && \
apt-get install -y \
imagemagick \
&& \
apt-get clean && rm -rf /var/lib/apt/lists/*
# Install olena from .deb
RUN curl -sSL --retry 3 -O https://qurator-data.de/~mike.gerber/olena_2.1.0+ocrd-git+2-ubuntu22.04/olena-bin_2.1.0+ocrd-git+2_amd64.deb && \
dpkg -i --force-depends olena-bin_2.1.0+ocrd-git+2_amd64.deb && \
rm -f olena-bin_2.1.0+ocrd-git+2_amd64.deb && \
apt-get update && \
apt-get -f install -y && \
apt-get clean && rm -rf /var/lib/apt/lists/* && \
if ! scribo-cli sauvola --help >/dev/null 2>&1; then echo "Olena/scribo is not installed correctly" >&2; exit 1; fi
RUN curl -sSL --retry 3 -o ocrd_olena.tar.gz https://github.com/OCR-D/ocrd_olena/archive/v${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 && \
${PIP_INSTALL} ocrd && \
apt install xmlstarlet && \
make install PREFIX=/usr/local && \
cd .. && rm -rf ocrd_olena ocrd_olena.tar.gz
# Check pip dependencies
RUN pip check
# Default command
CMD ['ocrd-olena-binarize']

@ -1,19 +0,0 @@
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_SEGMENT_VERSION="0.1.22"
# Build pip installable stuff
RUN ${PIP_INSTALL} \
# Now the real stuff:
"ocrd-segment == ${OCRD_SEGMENT_VERSION}"
# Check pip dependencies
RUN pip check
# Default command
CMD ["ocrd-segment-extract-regions"]

@ -1,31 +0,0 @@
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG TESSDATA_BEST_VERSION="4.0.0"
ARG OCRD_TESSEROCR_VERSION="0.17.0"
ENV TESSDATA_PREFIX /usr/local/share/tessdata
# Install Leptonica and Tesseract.
# TODO: Review if alex-p's repo is still necessary on jammy (jammy has 4.1.1,
# alex-p has 4.1.3, but not for jammy.)
# RUN add-apt-repository ppa:alex-p/tesseract-ocr && \
RUN apt-get update && \
apt-get install -y \
tesseract-ocr \
libtesseract-dev \
&& \
apt-get clean && rm -rf /var/lib/apt/lists/*
# Build pip installable stuff
RUN ${PIP_INSTALL} \
"ocrd_tesserocr == ${OCRD_TESSEROCR_VERSION}"
# Check pip dependencies
RUN pip check
# Default command
CMD ["ocrd-tesserocr-recognize"]

@ -1,18 +0,0 @@
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_WRAP_VERSION="0.1.7"
# Build pip installable stuff
RUN ${PIP_INSTALL} \
"ocrd_wrap == ${OCRD_WRAP_VERSION}"
# Check pip dependencies
RUN pip check
# Default command
CMD ["ocrd-preprocess-image"]

@ -1,20 +0,0 @@
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda12.1:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG SBB_TEXTLINE_DETECTOR_COMMIT="c4df3d6"
# Build pip installable stuff
RUN ${PIP_INSTALL} \
# https://github.com/qurator-spk/sbb_textline_detection/issues/50
"h5py < 3" \
https://github.com/qurator-spk/sbb_textline_detector/archive/$SBB_TEXTLINE_DETECTOR_COMMIT.tar.gz
# Check pip dependencies
RUN pip check
# Default command
CMD ["ocrd-sbb-textline-detector"]

33
build

@ -1,33 +0,0 @@
#!/bin/bash
set -e
self=`realpath $0`
self_dir=`dirname "$self"`
if [ -n "$1" ]; then
sub_images=""
for arg in "$@"; do
arg_sub_image=`echo "$arg" | sed 's/Dockerfile-//'`
NL=$'\n'
sub_images+="$NL$arg_sub_image"
done
else
sub_images=`ls -1 Dockerfile-core* | sed 's/Dockerfile-//'`
sub_images="$sub_images `ls -1 Dockerfile-* | sed 's/Dockerfile-//'`"
fi
echo "Building:"
echo "$sub_images"
echo
# Update base images if we build a core image
if echo "$sub_images" | grep -q core; then
docker pull ubuntu:22.04
docker pull nvidia/cuda:12.1.0-cudnn8-runtime-ubuntu22.04
fi
for sub_image in $sub_images; do
docker build --cache-from=quratorspk/ocrd-galley-$sub_image -t quratorspk/ocrd-galley-$sub_image -f Dockerfile-$sub_image .
done

@ -5,6 +5,8 @@ test_id=`basename $0`
cd `mktemp -d /tmp/$test_id-XXXXX`
# Prepare processors
ocrd resmgr download ocrd-tesserocr-recognize eng.traineddata
ocrd resmgr download ocrd-tesserocr-recognize osd.traineddata
ocrd resmgr download ocrd-tesserocr-recognize Fraktur_GT4HistOCR.traineddata
# Prepare test workspace

@ -5,7 +5,7 @@ test_id=`basename $0`
cd `mktemp -d /tmp/$test_id-XXXXX`
# Prepare processors
ocrd resmgr download ocrd-sbb-binarize default-2021-03-09
ocrd resmgr download ocrd-sbb-binarize default
# Prepare test workspace
wget https://qurator-data.de/examples/actevedef_718448162.first-page+binarization+segmentation.zip
@ -13,4 +13,4 @@ unzip actevedef_718448162.first-page+binarization+segmentation.zip
cd actevedef_718448162.first-page+binarization+segmentation
# Run tests
ocrd-sbb-binarize -P model default-2021-03-09 -I OCR-D-IMG -O TEST-OCRD-SBB-BINARIZE
ocrd-sbb-binarize -P model default -I OCR-D-IMG -O TEST-OCRD-SBB-BINARIZE

@ -1,16 +0,0 @@
#!/bin/sh
set -ex
test_id=`basename $0`
cd `mktemp -d /tmp/$test_id-XXXXX`
# Prepare processors
ocrd resmgr download ocrd-sbb-textline-detector default
# Prepare test workspace
wget https://qurator-data.de/examples/actevedef_718448162.first-page+binarization+segmentation.zip
unzip actevedef_718448162.first-page+binarization+segmentation.zip
cd actevedef_718448162.first-page+binarization+segmentation
# Run tests
ocrd-sbb-textline-detector -P models default -I OCR-D-IMG-BIN -O TEST-EYNOLLAH-SEG

@ -5,10 +5,8 @@ import colorama
from pathlib import Path
from termcolor import colored
from .sub_images import sub_images
DOCKER_IMAGE_PREFIX = os.environ.get("DOCKER_IMAGE_PREFIX", "quratorspk/ocrd-galley")
DOCKER_IMAGE_TAG = os.environ.get("DOCKER_IMAGE_TAG", "latest")
DOCKER_IMAGE_TAG = os.environ.get("DOCKER_IMAGE_TAG", "maximum") # TODO rename
LOG_LEVEL = os.environ.get("LOG_LEVEL", "INFO")
# xdg-user-dirs is only available under Python 3.10+ etc. pp. → it is simpler
@ -26,22 +24,9 @@ def main():
argv = sys.argv.copy()
argv[0] = os.path.basename(argv[0])
# If we're running ocrd resmgr download we need to run the correct subimage.
if argv[:3] == ["ocrd", "resmgr", "download"] or \
argv[:3] == ["ocrd", "resmgr", "list-available"]:
# Default to the base image
sub_image = sub_images[argv[0]]
# But look for a match of the executable
for x in argv[3:]:
if x in sub_images:
sub_image = sub_images[x]
break
else:
sub_image = sub_images[argv[0]]
docker_image = "%s-%s:%s" % (DOCKER_IMAGE_PREFIX, sub_image, DOCKER_IMAGE_TAG)
if DOCKER_IMAGE_TAG != "latest":
docker_image = "ocrd/all:%s" % (DOCKER_IMAGE_TAG, )
if DOCKER_IMAGE_TAG != "maximum":
print(colored(f"Using {docker_image}", 'red'))
docker_run(argv, docker_image)
@ -50,6 +35,7 @@ def docker_run(argv, docker_image):
docker_run_options = []
docker_run_options.extend(["--rm", "-t"])
docker_run_options.extend(["--mount", "type=bind,src=%s,target=/data" % os.getcwd()])
docker_run_options.extend(["--mount", "type=tmpfs,target=/tmp"])
docker_run_options.extend(["--user", "%s:%s" % (os.getuid(), os.getgid())])
docker_run_options.extend(["-e", "LOG_LEVEL=%s" % LOG_LEVEL])
docker_run_options.extend(["-e", "_OCRD_COMPLETE"])

@ -1,3 +1,4 @@
# TODO is a list now, basically (no more sub images)
sub_images = {
"ocrd": "core",
"ocrd-olena-binarize": "ocrd_olena",

Loading…
Cancel
Save