docker/install: build Tesseract from source

.circleci/config.yml

@@ -12,18 +12,18 @@ jobs:
         type: string
     docker:
       - image: cimg/python:<< parameters.python-version >>
+    environment:
+      # cimg/python uses pyenv instead of venv
+      VIRTUAL_ENV: ${PYENV_ROOT}
     steps:
       - checkout
+      - run: git submodule sync && git submodule update --init
       - run: sudo make deps-ubuntu
-      - when:
-          condition:
-            equal: [ '3.6', << parameters.python-version >> ]
-          steps:
-            # speed-up build time for end-of-life Python by holding at latest binary:
-            - run: pip install --prefer-binary -U opencv-python-headless numpy
+      - run: make install-tesseract
+      - run: make install-tesserocr
       - run: make install
-      # PPA tessdata prefix (= ocrd_tesserocr moduledir) is owned by root
-      - run: sudo chmod go+w `dpkg-query -L tesseract-ocr-eng | sed -n s,/eng.traineddata,,p`
+      - run: ocrd resmgr download ocrd-tesserocr-recognize eng.traineddata
+      - run: ocrd resmgr download ocrd-tesserocr-recognize osd.traineddata
       - run: ocrd resmgr download ocrd-tesserocr-recognize deu.traineddata
       - run: ocrd resmgr download ocrd-tesserocr-recognize Fraktur.traineddata
       - run: make test-cli
@@ -36,4 +36,4 @@ workflows:
       - build-python:
           matrix:
             parameters:
-              python-version: ['3.6', '3.7', '3.8', '3.9', '3.10']
+              python-version: ['3.7', '3.8', '3.9', '3.10']

.dockerignore

@@ -5,6 +5,8 @@
 !requirements_test.txt
 !LICENSE
 !README.md
+!repo/tesserocr
+!repo/tesseract
 
 # avoid .git and __pycache__ etc:
 !ocrd_tesserocr/**/*.py

.gitignore

@@ -13,3 +13,4 @@ test-workspace
 /.coverage
 /htmlcov
 /.cache
+build_tesseract

.gitmodules

@@ -0,0 +1,6 @@
+[submodule "repo/tesserocr"]
+	path = repo/tesserocr
+	url = https://github.com/sirfz/tesserocr/
+[submodule "repo/tesseract"]
+	path = repo/tesseract
+	url = https://github.com/tesseract-ocr/tesseract

Dockerfile

@@ -1,4 +1,8 @@
-FROM ocrd/core
+FROM ocrd/core:v2.62.0 AS base
+# set proper locales
+ENV LANG C.UTF-8
+ENV LC_ALL C.UTF-8
+# install ocrd-tesserocr (until here commands for installing tesseract-ocr)
 ARG VCS_REF
 ARG BUILD_DATE
 LABEL \
@@ -7,32 +11,49 @@ LABEL \
     org.label-schema.vcs-url="https://github.com/OCR-D/ocrd_tesserocr" \
     org.label-schema.build-date=$BUILD_DATE
 
-ENV DEBIAN_FRONTEND noninteractive
 ENV PYTHONIOENCODING utf8
 
+# set TESSDATA_PREFIX
+ENV TESSDATA_PREFIX /usr/local/share/tessdata
+
+# set frontend non-interactive to silence interactive tzdata config
+ARG DEBIAN_FRONTEND=noninteractive
+
+# set proper date and timezone in container
+RUN echo "Europe/Berlin" > /etc/timezone
+RUN ln -sf /usr/share/zoneinfo/Europe/Berlin /etc/localtime
+RUN dpkg-reconfigure -f noninteractive tzdata
+
+# diagnostic output - check timezone settings
+# RUN cat /etc/timezone
+
 # avoid HOME/.local/share (hard to predict USER here)
 # so let XDG_DATA_HOME coincide with fixed system location
 # (can still be overridden by derived stages)
 ENV XDG_DATA_HOME /usr/local/share
 
-WORKDIR /build-ocrd
+WORKDIR /build
 COPY setup.py .
 COPY ocrd_tesserocr/ocrd-tool.json .
 COPY README.md .
 COPY requirements.txt .
 COPY requirements_test.txt .
 COPY ocrd_tesserocr ./ocrd_tesserocr
+COPY repo/tesserocr ./repo/tesserocr
+COPY repo/tesseract ./repo/tesseract
 COPY Makefile .
-RUN make deps-ubuntu && \
-    apt-get install -y --no-install-recommends \
-    g++ \
-    && make deps install \
-    && rm -rf /build-ocrd \
+RUN make deps-ubuntu deps install-tesseract install-tesserocr install \
+    && rm -rf /build \
     && apt-get -y remove --auto-remove g++ libtesseract-dev make
+
 # PPA tessdata prefix (= ocrd_tesserocr moduledir) is owned by root
-RUN sudo chmod go+w `dpkg-query -L tesseract-ocr-eng | sed -n s,/eng.traineddata,,p`
+# next line causes failure because tesseract-ocr-eng not existing. Not sure if needed, so skipping
+# RUN sudo chmod go+w `dpkg-query -L tesseract-ocr-eng | sed -n s,/eng.traineddata,,p`
 RUN ocrd resmgr download ocrd-tesserocr-recognize Fraktur.traineddata
 RUN ocrd resmgr download ocrd-tesserocr-recognize deu.traineddata
+RUN ocrd resmgr download ocrd-tesserocr-recognize eng.traineddata
+RUN ocrd resmgr download ocrd-tesserocr-recognize equ.traineddata
+RUN ocrd resmgr download ocrd-tesserocr-recognize osd.traineddata
 
 WORKDIR /data
 VOLUME /data

Makefile

@@ -7,8 +7,20 @@ LOG_LEVEL = INFO
 PYTHONIOENCODING=utf8
 LC_ALL = C.UTF-8
 LANG = C.UTF-8
-export
+ifdef VIRTUAL_ENV
+	TESSERACT_PREFIX = $(VIRTUAL_ENV)
+else
+	TESSERACT_PREFIX = /usr/local
+endif
+
+ifeq ($(PKG_CONFIG_PATH),)
+PKG_CONFIG_PATH := $(TESSERACT_PREFIX)/lib/pkgconfig
+else
+PKG_CONFIG_PATH := $(TESSERACT_PREFIX)/lib/pkgconfig:$(PKG_CONFIG_PATH)
+endif
+export PKG_CONFIG_PATH
 
+export
 
 # pytest args. Set to '-s' to see log output during test execution, '--verbose' to see individual tests. Default: '$(PYTEST_ARGS)'
 PYTEST_ARGS =
@@ -50,24 +62,32 @@ help:
 
 # Dependencies for deployment in an ubuntu/debian linux
 # (lib*-dev merely for building tesserocr with pip)
-# (tesseract-ocr: Ubuntu 18.04 now ships 4.0.0,
-#  which is unsupported. Add the tesseract-ocr PPA
-#  from Alexander Pozdnyakov which provides 4.1.0.
-#  See https://launchpad.net/~alex-p/+archive/ubuntu/tesseract-ocr
-#  for details.)
 deps-ubuntu:
-	apt-get install -y --no-install-recommends software-properties-common
-	-add-apt-repository -u -y ppa:alex-p/tesseract-ocr
-	apt-get install -y \
+	apt-get update && apt-get install -y --no-install-recommends \
+		apt-utils \
+		build-essential \
 		g++ \
 		git \
 		python3 \
 		python3-pip \
-		libtesseract-dev \
+		libjpeg-dev \
+		libgif-dev \
+		libwebp-dev \
+		libopenjp2-7-dev \
+		libpng-dev \
+		libtiff-dev \
+		libtool \
+		pkg-config \
+		tzdata \
+		xzgv \
+		zlib1g-dev \
 		libleptonica-dev \
-		tesseract-ocr-eng \
-		tesseract-ocr-script-frak \
-		tesseract-ocr
+		libpango1.0-dev \
+		libicu-dev \
+		autotools-dev \
+		automake \
+		libcurl4-nss-dev \
+		libarchive-dev
 
 # Install Python deps for install via pip
 deps:
@@ -85,6 +105,24 @@ docker:
 	--build-arg BUILD_DATE=$$(date -u +"%Y-%m-%dT%H:%M:%SZ") \
 	-t $(DOCKER_TAG) .
 
+install-tesserocr: repo/tesserocr
+	$(PIP) install ./$<
+
+install-tesseract: repo/tesseract
+	cd $<; ./autogen.sh
+	mkdir -p build_tesseract
+	cd build_tesseract && $(CURDIR)/repo/tesseract/configure \
+				--prefix=$(TESSERACT_PREFIX) \
+				--disable-openmp \
+				--disable-shared \
+				'CXXFLAGS=-g -O2 -fno-math-errno -Wall -Wextra -Wpedantic -fPIC' && \
+				$(MAKE) install
+	if [[ "$(TESSERACT_PREFIX)" = "/usr"* ]];then ldconfig ;fi
+
+repo/tesserocr repo/tesseract:
+	git submodule sync $@
+	git submodule update --init $@
+
 # Install this package
 install: deps
 	$(PIP) install .
@@ -135,6 +173,12 @@ repo/assets:
 	mkdir -p $(dir $@)
 	git clone https://github.com/OCR-D/assets "$@"
 
+.PHONY: clean
+clean: assets-clean tesseract-clean
+
+tesseract-clean:
+	rm -rf $(CURDIR)/build_tesseract
+	cd repo/tesseract; make distclean
 
 .PHONY: assets-clean
 # Remove symlinks in test/assets