qurator-spk/ocrd-galley
1
0
Fork 0
mirror of https://github.com/qurator-spk/ocrd-galley.git synced 2025-06-11 23:59:53 +02:00
Code Issues Projects Releases Wiki Activity

Merge branch 'master' of https://github.com/qurator-spk/ocrd-galley

Browse source
This commit is contained in:
Mike Gerber 2023-08-11 18:18:29 +02:00
commit 4c1f198da9
39 changed files with 532 additions and 383 deletions
Download patch file Download diff file Expand all files Collapse all files

74
.drone.star
View file

@ -1,74 +0,0 @@
def main(ctx):
tags = [ctx.build.commit]
if ctx.build.event == "tag":
name = "release"
elif ctx.build.branch == "master":
name = "master"
tags.append("latest")
else:
return
return [
{
"kind": "pipeline",
"name": name,
"steps": [
{
"name": "prepare data",
"image": "alpine",
"commands": [
"apk update && apk add bash curl",
"FORCE_DOWNLOAD=y ./build-tmp-XXX"
]
},
# We can't glob and have to add here manually...
step_for(ctx, "core", tags),
step_for(ctx, "core-cuda10.0", tags),
step_for(ctx, "core-cuda10.1", tags),
step_for(ctx, "dinglehopper", tags),
step_for(ctx, "ocrd_calamari", tags),
step_for(ctx, "ocrd_calamari03", tags),
step_for(ctx, "ocrd_cis", tags),
step_for(ctx, "ocrd_fileformat", tags),
step_for(ctx, "ocrd_olena", tags),
step_for(ctx, "ocrd_segment", tags),
step_for(ctx, "ocrd_tesserocr", tags),
step_for(ctx, "ocrd_wrap", tags),
step_for(ctx, "sbb_binarization", tags),
step_for(ctx, "sbb_textline_detector", tags),
step_for(ctx, "eynollah", tags),
step_for(ctx, "ocrd_anybaseocr", tags),
{
"name": "notify",
"image": "drillster/drone-email",
"settings": {
"host": "172.17.0.1",
"port": "25",
"from": "drone@ci.moegen-wir.net",
},
"when": {
"status": [ "success", "failure" ]
}
}
]
}
]
def step_for(ctx, sub_image, tags):
return {
"name": "build %s" % sub_image,
"image": "plugins/docker",
"settings": {
"build_args": [
"DRONE_COMMIT=%s" % ctx.build.commit,
],
"tags": tags,
"username": { "from_secret": "docker_username" },
"password": { "from_secret": "docker_password" },
"repo": "quratorspk/ocrd-galley-%s" % sub_image,
"dockerfile": "Dockerfile-%s" % sub_image,
}
}

38
.github/list-subimages vendored Executable file
View file

@ -0,0 +1,38 @@
#!/usr/bin/python3
import glob
import re
import sys
import argparse
import json
all_subimages = {re.sub(r"^Dockerfile-", "", dockerfile) for dockerfile in glob.glob("Dockerfile-*")}
core_subimages = {si for si in all_subimages if si.startswith("core")}
rest_subimages = all_subimages - core_subimages
parser = argparse.ArgumentParser(description='List subimages.')
parser.add_argument('--core', action='store_true',
default=False, help='List core subimages')
parser.add_argument('--rest', action='store_true',
default=False, help='List rest subimages')
parser.add_argument('--json', action='store_true',
default=False, help='Return list as JSON')
args = parser.parse_args()
def list_(subimages):
subimages = sorted(subimages)
if args.json:
print(json.dumps(subimages))
else:
print("\n".join(subimages))
if not args.core and not args.rest:
list_(core_subimages | rest_subimages)
if args.core:
list_(core_subimages)
if args.rest:
list_(rest_subimages)

58
.github/workflows/build-subimage.yml vendored Normal file
View file

@ -0,0 +1,58 @@
on:
workflow_call:
inputs:
subimage:
required: true
type: string
tags:
required: true
type: string
secrets:
DOCKERHUB_USERNAME:
required: true
DOCKERHUB_TOKEN:
required: true
jobs:
build-subimage-job:
runs-on: ubuntu-latest
steps:
-
name: Checkout
uses: actions/checkout@v3
# We are checking out explicitly, so build-push-action isn't trying
# to checkout the (unreachable) submodule. (Using "context" there.)
-
name: Set up Docker Buildx
uses: docker/setup-buildx-action@v2
-
name: Docker meta
id: meta
uses: docker/metadata-action@v4
with:
images: |
quratorspk/ocrd-galley-${{ inputs.subimage }}
flavor: |
latest=auto
# latest=auto should generate "latest" for the type=semver tags entry
tags: ${{ inputs.tags }}
-
name: Login to Docker Hub
uses: docker/login-action@v2
with:
username: ${{ secrets.DOCKERHUB_USERNAME }}
password: ${{ secrets.DOCKERHUB_TOKEN }}
-
name: Build ${{ inputs.subimage }}
uses: docker/build-push-action@v4
with:
context: .
file: Dockerfile-${{ inputs.subimage }}
build-args: |
GIT_COMMIT=sha-${{ github.sha }}
BUILDKIT_INLINE_CACHE=1
tags: ${{ steps.meta.outputs.tags }}
push: true
cache-from: quratorspk/ocrd-galley-${{ inputs.subimage }}:sha-${{ github.sha }}

102
.github/workflows/build.yml vendored
View file

@ -1,36 +1,104 @@
name: build
on:
workflow_dispatch:
push:
branches:
- 'master'
- 'test/github-actions'
- 'fix/*'
jobs:
docker:
matrix:
runs-on: ubuntu-latest
outputs:
core: ${{ steps.step1.outputs.core }}
rest: ${{ steps.step1.outputs.rest }}
all: ${{ steps.step1.outputs.all }}
steps:
-
name: Checkout
uses: actions/checkout@v3
# We are checking out explicitly, so build-push-action isn't trying
# to checkout the (unreachable) submodule. (Using "context" there.)
-
name: Set up Docker Buildx
uses: docker/setup-buildx-action@v2
name: Generate outputs
id: step1
run: |
echo "core=$(./.github/list-subimages --core --json)" >>$GITHUB_OUTPUT
echo "rest=$(./.github/list-subimages --rest --json)" >>$GITHUB_OUTPUT
echo "all=$(./.github/list-subimages --json)" >>$GITHUB_OUTPUT
# TODO data
# TODO matrix for all Dockerfiles
echo "GITHUB_OUTPUT:"
cat $GITHUB_OUTPUT
-
name: Build
uses: docker/build-push-action@v4
build-core:
needs: matrix
strategy:
fail-fast: false
matrix:
subimage: ${{ fromJson(needs.matrix.outputs.core) }}
uses: ./.github/workflows/build-subimage.yml
with:
context: .
file: Dockerfile-core
build-args:
DRONE_COMMIT=${{ github.sha }}
push: false
subimage: ${{ matrix.subimage }}
tags: |
type=sha,format=long
# Here: NOT the full tags, just the sha! (they get added below)
secrets: inherit
build-rest:
needs: [matrix, build-core]
strategy:
fail-fast: false
matrix:
subimage: ${{ fromJson(needs.matrix.outputs.rest) }}
uses: ./.github/workflows/build-subimage.yml
with:
subimage: ${{ matrix.subimage }}
tags: |
type=sha,format=long
secrets: inherit
test:
needs: build-rest
runs-on: ubuntu-latest
env:
DOCKER_IMAGE_TAG: sha-${{ github.sha }} # needed to run the correct version through the wrapper
steps:
-
name: Checkout
uses: actions/checkout@v3
-
name: Install wrapper
run: |
sudo apt-get install -y python3-pip
cd wrapper && pip install .
-
name: Test
run: |
ocrd --version
ocrd-dinglehopper --version
# At this point, we have successfully built, uploaded and tested the images. We now just need to add
# tags. We do this by building again, but using the formerly built images to
# cache from.
push-with-tags:
needs: [matrix, test]
strategy:
matrix:
subimage: ${{ fromJson(needs.matrix.outputs.all) }}
uses: ./.github/workflows/build-subimage.yml
with:
subimage: ${{ matrix.subimage }}
tags: |
type=sha,format=long
type=edge,branch=master
type=ref,event=branch
type=semver,pattern={{version}}
# Here: full tags
# Note: Do NOT use event=tag here, unless re-configuring the "latest"
# behavior too as that triggers on event=tag by default. By default,
# "latest" triggers on type=semver here, too (which is wanted).
secrets: inherit

2
.gitignore vendored
View file

@ -1,3 +1,5 @@
build/
# Byte-compiled / optimized / DLL files
__pycache__/
*.py[cod]

3
.gitmodules vendored
View file

@ -1,3 +0,0 @@
[submodule "data"]
path = data
url = git@code.dev.sbb.berlin:qurator/qurator-data.git

45
Dockerfile-core
View file

@ -1,7 +1,7 @@
FROM ubuntu:18.04
FROM ubuntu:22.04
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_VERSION_MINIMUM="2.23.3"
ARG OCRD_VERSION_MINIMUM="2.47.0"
ENV LC_ALL=C.UTF-8 LANG=C.UTF-8
ENV PIP_DEFAULT_TIMEOUT=120
@ -9,10 +9,11 @@ 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 \
build-essential python3-dev \
# For get-pip.py:
python3-distutils \
build-essential \
curl \
git \
xz-utils \
pkg-config \
# For add-apt-repository:
software-properties-common \
# XML utils
@ -20,6 +21,17 @@ RUN echo "APT::Acquire::Retries \"3\";" > /etc/apt/apt.conf.d/80-retries && \
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/*
@ -29,14 +41,19 @@ RUN echo "APT::Acquire::Retries \"3\";" > /etc/apt/apt.conf.d/80-retries && \
RUN echo "setOverrideLogLevel(os.getenv('LOG_LEVEL', 'INFO'))" >/etc/ocrd_logging.py
# Install pip (and setuptools)
# We use get-pip.py here to avoid
# a. having to upgrade from Ubuntu's pip
# b. the dreaded "old script wrapper" error message
RUN curl -sSL https://bootstrap.pypa.io/pip/3.6/get-pip.py -o get-pip.py && \
python3 get-pip.py && \
rm -f get-pip.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} \

53
Dockerfile-core-cuda10.0
View file

@ -1,53 +0,0 @@
FROM nvidia/cuda:10.0-cudnn7-runtime-ubuntu18.04
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_VERSION_MINIMUM="2.23.3"
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 \
build-essential python3-dev \
# For get-pip.py:
python3-distutils \
# For add-apt-repository:
software-properties-common \
# XML utils
libxml2-utils \
xmlstarlet \
# OCR-D uses ImageMagick for pixel density estimation
imagemagick \
&& \
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 pip (and setuptools)
# We use get-pip.py here to avoid
# a. having to upgrade from Ubuntu's pip
# b. the dreaded "old script wrapper" error message
RUN curl -sSL https://bootstrap.pypa.io/pip/3.6/get-pip.py -o get-pip.py && \
python3 get-pip.py && \
rm -f get-pip.py
# Install pip installable-stuff
RUN ${PIP_INSTALL} \
"ocrd >= ${OCRD_VERSION_MINIMUM}"
# Check pip dependencies
RUN pip check
WORKDIR /data
# Default command
CMD ['ocrd']

53
Dockerfile-core-cuda10.1
View file

@ -1,53 +0,0 @@
FROM nvidia/cuda:10.1-cudnn7-runtime-ubuntu18.04
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_VERSION_MINIMUM="2.23.3"
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 \
build-essential python3-dev \
# For get-pip.py:
python3-distutils \
# For add-apt-repository:
software-properties-common \
# XML utils
libxml2-utils \
xmlstarlet \
# OCR-D uses ImageMagick for pixel density estimation
imagemagick \
&& \
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 pip (and setuptools)
# We use get-pip.py here to avoid
# a. having to upgrade from Ubuntu's pip
# b. the dreaded "old script wrapper" error message
RUN curl -sSL https://bootstrap.pypa.io/pip/3.6/get-pip.py -o get-pip.py && \
python3 get-pip.py && \
rm -f get-pip.py
# Install pip installable-stuff
RUN ${PIP_INSTALL} \
"ocrd >= ${OCRD_VERSION_MINIMUM}"
# Check pip dependencies
RUN pip check
WORKDIR /data
# Default command
CMD ['ocrd']

70
Dockerfile-core-cuda12.1 Normal file
View file

@ -0,0 +1,70 @@
FROM nvidia/cuda:12.1.0-cudnn8-runtime-ubuntu22.04
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']

9
Dockerfile-dinglehopper
View file

@ -1,14 +1,13 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG DINGLEHOPPER_COMMIT="dcc10c5"
ARG DINGLEHOPPER_VERSION="0.9.2"
# Build pip installable stuff
RUN ${PIP_INSTALL} \
# Now the real stuff:
https://github.com/qurator-spk/dinglehopper/archive/$DINGLEHOPPER_COMMIT.tar.gz
"dinglehopper == $DINGLEHOPPER_VERSION"
# Check pip dependencies

10
Dockerfile-eynollah
View file

@ -1,8 +1,8 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda10.0:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda12.1:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG EYNOLLAH_VERSION="0.0.10"
ARG EYNOLLAH_VERSION="0.3.0"
# Build pip installable stuff
@ -10,10 +10,6 @@ RUN ${PIP_INSTALL} \
"eynollah == ${EYNOLLAH_VERSION}"
# Copy OCR models
COPY data/eynollah /var/lib/eynollah
# Check pip dependencies
RUN pip check

4
Dockerfile-ocrd_anybaseocr
View file

@ -1,5 +1,5 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda10.1:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda12.1:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_ANYBASEOCR_VERSION="1.8.2"

12
Dockerfile-ocrd_calamari
View file

@ -1,5 +1,5 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda10.1:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda12.1:$GIT_COMMIT
# XXX https://github.com/OCR-D/core/issues/642
@ -12,14 +12,6 @@ RUN ${PIP_INSTALL} \
"ocrd_calamari == $OCRD_CALAMARI_VERSION"
# Copy OCR models
RUN mkdir -p /var/lib/calamari-models/GT4HistOCR
COPY data/calamari-models/GT4HistOCR/2019-12-11T11_10+0100 /var/lib/calamari-models/GT4HistOCR/2019-12-11T11_10+0100
# XXX experimental
#COPY data/calamari-models/GT4HistOCR/2019-12-18T17_24+0100-with-augmentation-UNTESTED /var/lib/calamari-models/GT4HistOCR/2019-12-18T17_24+0100
#COPY data/mirror/github.com/Calamari-OCR/calamari_models/gt4histocr /var/lib/calamari-models/GT4HistOCR-chreul
# Check pip dependencies
RUN pip check

10
Dockerfile-ocrd_calamari03
View file

@ -1,5 +1,5 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda10.0:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda12.1:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
@ -13,12 +13,6 @@ RUN ${PIP_INSTALL} \
'ocrd_calamari == 0.0.7'
# Copy OCR models
RUN mkdir -p /var/lib/calamari-models/GT4HistOCR
COPY data/calamari-models/GT4HistOCR/2019-07-22T15_49+0200 /var/lib/calamari-models/GT4HistOCR/2019-07-22T15_49+0200
# Check pip dependencies
RUN pip check

4
Dockerfile-ocrd_cis
View file

@ -1,5 +1,5 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_CIS_VERSION="0.1.5"

4
Dockerfile-ocrd_fileformat
View file

@ -1,5 +1,5 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_FILEFORMAT_VERSION="0.5.0"

5
Dockerfile-ocrd_olena
View file

@ -1,5 +1,5 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_OLENA_VERSION="1.3.0"
@ -26,6 +26,7 @@ RUN curl -sSL --retry 3 -o ocrd_olena.tar.gz https://github.com/OCR-D/ocrd_olena
cd ocrd_olena && \
sed -i 's/^install: deps$/install:/' Makefile && \
${PIP_INSTALL} ocrd && \
make deps-ubuntu && \
make install PREFIX=/usr/local && \
cd .. && rm -rf ocrd_olena ocrd_olena.tar.gz

4
Dockerfile-ocrd_segment
View file

@ -1,5 +1,5 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_SEGMENT_VERSION="0.1.21"

20
Dockerfile-ocrd_tesserocr
View file

@ -1,15 +1,17 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG TESSDATA_BEST_VERSION="4.0.0"
ARG OCRD_TESSEROCR_VERSION="0.16.0"
ARG OCRD_TESSEROCR_VERSION="0.17.0"
ENV TESSDATA_PREFIX /usr/local/share/tessdata
# Install Leptonica and Tesseract.
RUN add-apt-repository ppa:alex-p/tesseract-ocr && \
apt-get update && \
# 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 \
@ -17,18 +19,10 @@ RUN add-apt-repository ppa:alex-p/tesseract-ocr && \
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/
RUN curl -sSL -O https://ub-backup.bib.uni-mannheim.de/~stweil/ocrd-train/data/Fraktur_5000000/tessdata_fast/Fraktur_50000000.502_198857.traineddata && \
mv *.traineddata $TESSDATA_PREFIX/
# Build pip installable stuff
RUN ${PIP_INSTALL} \
"ocrd_tesserocr == ${OCRD_TESSEROCR_VERSION}"
# Check pip dependencies
RUN pip check

18
Dockerfile-ocrd_trocr Normal file
View file

@ -0,0 +1,18 @@
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_TROCR_COMMIT="250ff1c"
# Build pip installable stuff
RUN ${PIP_INSTALL} \
https://github.com/qurator-spk/ocrd_trocr/archive/$OCRD_TROCR_COMMIT.tar.gz
# Check pip dependencies
RUN pip check
# Default command
CMD ["ocrd-trocr-recognize"]

4
Dockerfile-ocrd_wrap
View file

@ -1,5 +1,5 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG OCRD_WRAP_VERSION="0.1.7"

8
Dockerfile-sbb_binarization
View file

@ -1,5 +1,5 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda10.0:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda12.1:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG SBB_BINARIZATION_VERSION="0.0.10"
@ -11,10 +11,6 @@ RUN ${PIP_INSTALL} \
"sbb_binarization == $SBB_BINARIZATION_VERSION"
# Copy models
COPY data/sbb_binarization/2021-03-09 /var/lib/sbb_binarization
# Check pip dependencies
RUN pip check

8
Dockerfile-sbb_textline_detector
View file

@ -1,5 +1,5 @@
ARG DRONE_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda10.0:$DRONE_COMMIT
ARG GIT_COMMIT="latest"
FROM quratorspk/ocrd-galley-core-cuda12.1:$GIT_COMMIT
ARG PIP_INSTALL="pip install --no-cache-dir"
ARG SBB_TEXTLINE_DETECTOR_COMMIT="c4df3d6"
@ -12,10 +12,6 @@ RUN ${PIP_INSTALL} \
https://github.com/qurator-spk/sbb_textline_detector/archive/$SBB_TEXTLINE_DETECTOR_COMMIT.tar.gz
# Copy OCR models
COPY data/textline_detection /var/lib/textline_detection
# Check pip dependencies
RUN pip check

10
README-DEV.md
View file

@ -3,7 +3,6 @@ How to add a processor
* Add model download to `build` (if necessary)
* Add a Dockerfile
* Add commands to `wrapper/qurator/ocrd_galley/cli.py`
* Add to `.drone.star`
Releasing
---------
@ -27,3 +26,12 @@ issue should be open that reminds us to go back to a versioned release again.
Other than relying on "proper releases", this also has a second purpose: Review
releases of qurator-spk releases.
Test builds
-----------
XXX Review this
```
GIT_COMMIT=test ./build Dockerfile-core Dockerfile-ocrd_tesserocr
DOCKER_IMAGE_TAG=test ./test-ocrd_tesserocr.sh
```

25
README.md
View file

@ -29,14 +29,9 @@ including all dependencies in Docker.
How to use
----------
**Currently, due to problems with the Travis CI, we do not provide pre-built
containers anymore.***
To build the containers yourself using Docker:
~~~
cd ~/devel/ocrd-galley/
./build
~~~
ocrd-galley uses Docker to run the OCR-D images. We provide pre-built container
images that get downloaded automatically when you run the provided wrappers for
the OCR-D processors.
You can then install the wrappers into a Python venv:
~~~
@ -44,9 +39,13 @@ cd ~/devel/ocrd-galley/wrapper
pip install .
~~~
To download models, you need to use the `-a` flag of `ocrd resmgr`:
~~~
ocrd resmgr download -a ocrd-calamari-recognize default
~~~
You may then use the script `my_ocrd_workflow` to use your self-built
containers on an example workspace:
~~~
# Download an example workspace
cd /tmp
@ -110,3 +109,11 @@ cd workspace-xxxxx # output by the last command
~~~
This produces a workspace from the files and then runs the OCR workflow on it.
Build the containers yourself
-----------------------------
To build the containers yourself using Docker:
~~~
cd ~/devel/ocrd-galley/
./build
~~~

32
build
View file

@ -22,38 +22,10 @@ echo "$sub_images"
echo
DATA_SUBDIR=data
get_from_annex() {
annex_get 'calamari-models/GT4HistOCR/2019-07-22T15_49+0200/*.ckpt*'
annex_get 'calamari-models/GT4HistOCR/2019-12-11T11_10+0100/*.ckpt*'
annex_get 'calamari-models/GT4HistOCR/2019-12-18T17_24+0100*/*.ckpt*'
annex_get 'mirror/github.com/Calamari-OCR/calamari_models/gt4histocr/*.ckpt*'
annex_get 'tesseract-models/GT4HistOCR/*.traineddata'
annex_get 'textline_detection/*.h5'
annex_get 'mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz'
annex_get 'sbb_binarization/2021-03-09/*.h5'
annex_get 'eynollah/*.h5'
}
get_from_web() {
download_to 'https://qurator-data.de/calamari-models/GT4HistOCR/model.tar.xz' 'calamari-models/GT4HistOCR/2019-07-22T15_49+0200'
download_to 'https://qurator-data.de/calamari-models/GT4HistOCR/2019-12-11T11_10+0100/model.tar.xz' 'calamari-models/GT4HistOCR/2019-12-11T11_10+0100'
download_to 'https://qurator-data.de/tesseract-models/GT4HistOCR/models.tar' 'tesseract-models/GT4HistOCR'
download_to 'https://qurator-data.de/sbb_textline_detector/models.tar.gz' 'textline_detection'
download_to --strip-components 1 'https://qurator-data.de/sbb_binarization/2021-03-09/models.tar.gz' 'sbb_binarization/2021-03-09'
download_to --no-unpack 'https://qurator-data.de/mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz' 'mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz'
download_to --strip-components 1 'https://qurator-data.de/eynollah/models_eynollah.tar.gz' 'eynollah'
}
. $self_dir/qurator_data_lib.sh
handle_data
# Update base images if we build a core image
if echo "$sub_images" | grep -q core; then
docker pull ubuntu:18.04
docker pull nvidia/cuda:10.0-cudnn7-runtime-ubuntu18.04
docker pull nvidia/cuda:10.1-cudnn7-runtime-ubuntu18.04
docker pull ubuntu:22.04
docker pull nvidia/cuda:12.1.0-cudnn8-runtime-ubuntu22.04
fi
for sub_image in $sub_images; do

18
build-tmp-XXX
View file

@ -1,18 +0,0 @@
#!/bin/bash -x
set -e
self=`realpath $0`
self_dir=`dirname "$self"`
DATA_SUBDIR=data
get_from_web() {
download_to 'https://qurator-data.de/calamari-models/GT4HistOCR/model.tar.xz' 'calamari-models/GT4HistOCR/2019-07-22T15_49+0200'
download_to 'https://qurator-data.de/calamari-models/GT4HistOCR/2019-12-11T11_10+0100/model.tar.xz' 'calamari-models/GT4HistOCR/2019-12-11T11_10+0100'
download_to 'https://qurator-data.de/tesseract-models/GT4HistOCR/models.tar' 'tesseract-models/GT4HistOCR'
download_to 'https://qurator-data.de/sbb_textline_detector/models.tar.gz' 'textline_detection'
download_to --strip-components 1 'https://qurator-data.de/sbb_binarization/2021-03-09/models.tar.gz' 'sbb_binarization/2021-03-09'
download_to --no-unpack 'https://qurator-data.de/mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz' 'mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz'
download_to --strip-components 1 'https://qurator-data.de/eynollah/models_eynollah.tar.gz' 'eynollah'
}
. $self_dir/qurator_data_lib.sh
handle_data

1
data

@ -1 +0,0 @@
Subproject commit 9ab08a3626dde1d38dd622b65e425277cd029722

14
test-core.sh Executable file
View file

@ -0,0 +1,14 @@
#!/bin/sh
set -ex
test_id=`basename $0`
cd `mktemp -d /tmp/$test_id-XXXXX`
# 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 workspace validate \
--page-coordinate-consistency off

13
test-dinglehopper.sh Executable file
View file

@ -0,0 +1,13 @@
#!/bin/sh
set -ex
test_id=`basename $0`
cd `mktemp -d /tmp/$test_id-XXXXX`
# 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-dinglehopper -I OCR-D-GT-PAGE,OCR-D-SEG-LINE-SBB -O DINGLEHOPPER-TEST

12
test-ocrd_olena.sh Executable file
View file

@ -0,0 +1,12 @@
#!/bin/sh
set -ex
cd `mktemp -d /tmp/test-ocrd_olena-XXXXX`
# 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-olena-binarize -I OCR-D-IMG -O TEST-OLENA

17
test-ocrd_tesserocr.sh Executable file
View file

@ -0,0 +1,17 @@
#!/bin/sh
set -ex
cd `mktemp -d /tmp/test-ocrd_tesserocr-XXXXX`
# Prepare processors
ocrd resmgr download ocrd-tesserocr-recognize Fraktur_GT4HistOCR.traineddata
# 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-tesserocr-segment-region -I OCR-D-IMG-BIN -O TEST-TESS-SEG-REG
ocrd-tesserocr-segment-line -I TEST-TESS-SEG-REG -O TEST-TESS-SEG-LINE
ocrd-tesserocr-recognize -I TEST-TESS-SEG-LINE -O TEST-TESS-OCR -P model Fraktur_GT4HistOCR

14
test-ocrd_trocr.sh Executable file
View file

@ -0,0 +1,14 @@
#!/bin/sh
set -ex
cd `mktemp -d /tmp/test-ocrd_trocr-XXXXX`
# Prepare processors
# 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-trocr-recognize -I OCR-D-SEG-LINE-SBB -O TEST-TROCR

1
wrapper/qurator/ocrd_galley/__init__.py
View file

@ -1 +0,0 @@
from .cli import *

90
wrapper/qurator/ocrd_galley/cli.py
View file

@ -1,61 +1,48 @@
import os
import subprocess
import sys
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")
LOG_LEVEL = os.environ.get("LOG_LEVEL", "INFO")
# xdg-user-dirs is only available under Python 3.10+ etc. pp. → it is simpler
# to just roll it on our own.
XDG_CONFIG_HOME = os.environ.get("XDG_CONFIG_HOME", Path.home() / ".config")
XDG_DATA_HOME = os.environ.get("XDG_DATA_HOME", Path.home() / ".local" / "share")
XDG_CACHE_HOME = os.environ.get("XDG_CACHE_HOME", Path.home() / ".cache")
sub_images = {
"ocrd": "core",
"ocrd-olena-binarize": "ocrd_olena",
"ocrd-sbb-binarize": "sbb_binarization",
"ocrd-sbb-textline-detector": "sbb_textline_detector",
"ocrd-calamari-recognize": "ocrd_calamari",
"ocrd-calamari-recognize03": "ocrd_calamari03",
"ocrd-tesserocr-segment-region": "ocrd_tesserocr",
"ocrd-tesserocr-segment-line": "ocrd_tesserocr",
"ocrd-tesserocr-recognize": "ocrd_tesserocr",
"ocrd-dinglehopper": "dinglehopper",
"ocrd-cis-ocropy-clip": "ocrd_cis",
"ocrd-cis-ocropy-resegment": "ocrd_cis",
"ocrd-cis-ocropy-segment": "ocrd_cis",
"ocrd-cis-ocropy-deskew": "ocrd_cis",
"ocrd-cis-ocropy-denoise": "ocrd_cis",
"ocrd-cis-ocropy-binarize": "ocrd_cis",
"ocrd-cis-ocropy-dewarp": "ocrd_cis",
"ocrd-cis-ocropy-recognize": "ocrd_cis",
"ocrd-fileformat-transform": "ocrd_fileformat",
"ocrd-segment-extract-pages": "ocrd_segment",
"ocrd-segment-extract-regions": "ocrd_segment",
"ocrd-segment-extract-lines": "ocrd_segment",
"ocrd-segment-from-masks": "ocrd_segment",
"ocrd-segment-from-coco": "ocrd_segment",
"ocrd-segment-repair": "ocrd_segment",
"ocrd-segment-evaluate": "ocrd_segment",
"ocrd-preprocess-image": "ocrd_wrap",
"ocrd-skimage-normalize": "ocrd_wrap",
"ocrd-skimage-denoise-raw": "ocrd_wrap",
"ocrd-skimage-binarize": "ocrd_wrap",
"ocrd-skimage-denoise": "ocrd_wrap",
"ocrd-eynollah-segment": "eynollah",
"ocrd-anybaseocr-crop": "ocrd_anybaseocr",
"ocrd-anybaseocr-deskew": "ocrd_anybaseocr",
# non OCR-D CLI
"ocr-transform": "ocrd_fileformat",
}
# ocrd_tesserocr
TESSDATA_PREFIX = XDG_DATA_HOME / "ocrd-resources" / "ocrd-tesserocr-recognize"
def main():
colorama.init()
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":
print(colored(f"Using {docker_image}", 'red'))
docker_run(argv, docker_image)
@ -67,6 +54,29 @@ def docker_run(argv, docker_image):
docker_run_options.extend(["-e", "LOG_LEVEL=%s" % LOG_LEVEL])
docker_run_options.extend(["-e", "_OCRD_COMPLETE"])
# home directory
docker_run_options.extend(["-e", "HOME=%s" % Path.home()])
# .config
docker_run_options.extend(["-e", "XDG_CONFIG_HOME=%s" % XDG_CONFIG_HOME])
docker_run_options.extend(["--mount", "type=bind,src=%s,target=%s" %
(XDG_CONFIG_HOME, XDG_CONFIG_HOME)])
# .local/share
docker_run_options.extend(["-e", "XDG_DATA_HOME=%s" % XDG_DATA_HOME])
docker_run_options.extend(["--mount", "type=bind,src=%s,target=%s" %
(XDG_DATA_HOME, XDG_DATA_HOME)])
# .cache
docker_run_options.extend(["-e", "XDG_CACHE_HOME=%s" % XDG_CACHE_HOME])
docker_run_options.extend(["--mount", "type=bind,src=%s,target=%s" %
(XDG_CACHE_HOME, XDG_CACHE_HOME)])
# .huggingface
os.makedirs(Path.home() / ".huggingface", exist_ok=True)
docker_run_options.extend(["--mount", "type=bind,src=%s,target=%s" %
(Path.home() / ".huggingface", Path("/root") / ".huggingface")])
# ocrd_tesserocr
docker_run_options.extend(["-e", "TESSDATA_PREFIX=%s" % TESSDATA_PREFIX])
# JAVA_TOOL_OPTIONS is used for Java proxy settings
if os.environ.get("JAVA_TOOL_OPTIONS"):
docker_run_options.extend(["-e", "JAVA_TOOL_OPTIONS"])

40
wrapper/qurator/ocrd_galley/sub_images.py Normal file
View file

@ -0,0 +1,40 @@
sub_images = {
"ocrd": "core",
"ocrd-olena-binarize": "ocrd_olena",
"ocrd-sbb-binarize": "sbb_binarization",
"ocrd-sbb-textline-detector": "sbb_textline_detector",
"ocrd-calamari-recognize": "ocrd_calamari",
"ocrd-calamari-recognize03": "ocrd_calamari03",
"ocrd-tesserocr-segment-region": "ocrd_tesserocr",
"ocrd-tesserocr-segment-line": "ocrd_tesserocr",
"ocrd-tesserocr-recognize": "ocrd_tesserocr",
"ocrd-dinglehopper": "dinglehopper",
"ocrd-cis-ocropy-clip": "ocrd_cis",
"ocrd-cis-ocropy-resegment": "ocrd_cis",
"ocrd-cis-ocropy-segment": "ocrd_cis",
"ocrd-cis-ocropy-deskew": "ocrd_cis",
"ocrd-cis-ocropy-denoise": "ocrd_cis",
"ocrd-cis-ocropy-binarize": "ocrd_cis",
"ocrd-cis-ocropy-dewarp": "ocrd_cis",
"ocrd-cis-ocropy-recognize": "ocrd_cis",
"ocrd-fileformat-transform": "ocrd_fileformat",
"ocrd-segment-extract-pages": "ocrd_segment",
"ocrd-segment-extract-regions": "ocrd_segment",
"ocrd-segment-extract-lines": "ocrd_segment",
"ocrd-segment-from-masks": "ocrd_segment",
"ocrd-segment-from-coco": "ocrd_segment",
"ocrd-segment-repair": "ocrd_segment",
"ocrd-segment-evaluate": "ocrd_segment",
"ocrd-preprocess-image": "ocrd_wrap",
"ocrd-skimage-normalize": "ocrd_wrap",
"ocrd-skimage-denoise-raw": "ocrd_wrap",
"ocrd-skimage-binarize": "ocrd_wrap",
"ocrd-skimage-denoise": "ocrd_wrap",
"ocrd-eynollah-segment": "eynollah",
"ocrd-anybaseocr-crop": "ocrd_anybaseocr",
"ocrd-anybaseocr-deskew": "ocrd_anybaseocr",
"ocrd-trocr-recognize": "ocrd_trocr",
# non OCR-D CLI
"ocr-transform": "ocrd_fileformat",
}

2
wrapper/requirements.txt Normal file
View file

@ -0,0 +1,2 @@
colorama
termcolor

6
wrapper/setup.py
View file

@ -1,9 +1,12 @@
from io import open
from setuptools import find_packages, setup
from qurator.ocrd_galley.cli import sub_images
from qurator.ocrd_galley.sub_images import sub_images
console_scripts = ["%s=qurator.ocrd_galley.cli:main" % command for command in sub_images.keys()]
with open("requirements.txt") as fp:
install_requires = fp.read()
setup(
name="ocrd-galley",
author="Mike Gerber, The QURATOR SPK Team",
@ -13,6 +16,7 @@ setup(
license="Apache",
packages=find_packages(exclude=["*.tests", "*.tests.*", "tests.*", "tests"]),
namespace_packages=["qurator"],
install_requires=install_requires,
entry_points={
"console_scripts": console_scripts,
},