Add new models

.gitlab-ci.yml

@@ -1,44 +1,71 @@
-image: docker:stable
-
-services:
-  - docker:stable-dind
+image: docker:19.03.13
 
 variables:
-  DOCKER_DRIVER: overlay2
+  DOCKER_TLS_CERTDIR: "/certs"
+
+services:
+  - docker:19.03.13-dind
 
 stages:
   - build
   - push
 
-before_script:
-  - docker login -u gitlab-ci-token -p $CI_JOB_TOKEN $CI_REGISTRY
+.reg_setup:
+  before_script:
+    - apk add --no-cache curl
+    - curl --fail --show-error --location "https://github.com/genuinetools/reg/releases/download/v$REG_VERSION/reg-linux-amd64" --output /usr/local/bin/reg
+    - echo "$REG_SHA256  /usr/local/bin/reg" | sha256sum -c -
+    - chmod a+x /usr/local/bin/reg
+  variables:
+    REG_SHA256: ade837fc5224acd8c34732bf54a94f579b47851cc6a7fd5899a98386b782e228
+    REG_VERSION: 0.16.1
 
-Build:
+build_image:
   script:
-    - docker build --pull -t $CI_REGISTRY_IMAGE:tmp .
-    - docker push $CI_REGISTRY_IMAGE:tmp
+    - docker login -u gitlab-ci-token -p $CI_JOB_TOKEN $CI_REGISTRY
+    - docker build -t $INTERMEDIATE_IMAGE_TAG .
+    - docker push $INTERMEDIATE_IMAGE_TAG
   stage: build
   tags:
-  - docker
+    - docker
+  variables:
+    INTERMEDIATE_IMAGE_TAG: $CI_REGISTRY_IMAGE:$CI_COMMIT_SHA
 
-Push latest:
+push_master:
+  extends:
+    - .reg_setup
   only:
     - master
   script:
-    - docker pull $CI_REGISTRY_IMAGE:tmp
-    - docker tag $CI_REGISTRY_IMAGE:tmp $CI_REGISTRY_IMAGE:latest
-    - docker push $CI_REGISTRY_IMAGE:latest
+    - docker login -u gitlab-ci-token -p $CI_JOB_TOKEN $CI_REGISTRY
+    - docker pull $INTERMEDIATE_IMAGE_TAG
+    - /usr/local/bin/reg rm -d --auth-url $CI_REGISTRY -u $CI_REGISTRY_USER -p $CI_REGISTRY_PASSWORD $INTERMEDIATE_IMAGE_TAG
+    - docker tag $INTERMEDIATE_IMAGE_TAG $IMAGE_TAG
+    - docker push $IMAGE_TAG
   stage: push
   tags:
-  - docker
+    - docker
+  variables:
+    IMAGE_TAG: $CI_REGISTRY_IMAGE:latest
+    INTERMEDIATE_IMAGE_TAG: $CI_REGISTRY_IMAGE:$CI_COMMIT_SHA
 
-Push tag:
+push_other:
+  extends:
+    - .reg_setup
+  except:
+    - master
   only:
+    - branches
     - tags
   script:
-    - docker pull $CI_REGISTRY_IMAGE:tmp
-    - docker tag $CI_REGISTRY_IMAGE:tmp $CI_REGISTRY_IMAGE:$CI_COMMIT_REF_NAME
-    - docker push $CI_REGISTRY_IMAGE:$CI_COMMIT_REF_NAME
+    - docker login -u gitlab-ci-token -p $CI_JOB_TOKEN $CI_REGISTRY
+    - docker pull $INTERMEDIATE_IMAGE_TAG
+    - /usr/local/bin/reg rm -d --auth-url $CI_REGISTRY -u $CI_REGISTRY_USER -p $CI_REGISTRY_PASSWORD $INTERMEDIATE_IMAGE_TAG
+    - docker tag $INTERMEDIATE_IMAGE_TAG $IMAGE_TAG
+    - docker push $IMAGE_TAG
   stage: push
   tags:
-  - docker
+    - docker
+  variables:
+    IMAGE_TAG: $CI_REGISTRY_IMAGE:$CI_COMMIT_REF_NAME
+    INTERMEDIATE_IMAGE_TAG: $CI_REGISTRY_IMAGE:$CI_COMMIT_SHA

Dockerfile

@@ -1,73 +1,101 @@
-FROM debian:9-slim
+FROM debian:buster-slim
 
 
-# Define image metadata
-LABEL maintainer="inf_sfb1288@lists.uni-bielefeld.de"
+LABEL authors="Patrick Jentsch <p.jentsch@uni-bielefeld.de>"
 
 
 ENV LANG=C.UTF-8
 
 
-# Install prerequisites
-RUN apt-get update \
- && apt-get install -y --no-install-recommends \
-      apt-transport-https \
-      ca-certificates \
-      gnupg2 \
-      imagemagick \
-      poppler-utils \
-      python2.7 \
-      python3.5 \
-      wget \
-      zip \
- && rm -rf /var/lib/apt/lists/*
+RUN apt-get update
 
-ENV OCROPY_VERSION 1.3.3
-ADD "https://github.com/tmbdev/ocropy/archive/v${OCROPY_VERSION}.tar.gz" .
-RUN tar -xzf "v${OCROPY_VERSION}.tar.gz" \
- && cd "ocropy-${OCROPY_VERSION}" \
- && apt-get update \
- && apt-get install -y --no-install-recommends \
+
+# Install pipeline dependencies #
+## Install pyFlow ##
+ENV PYFLOW_RELEASE=1.1.20
+ADD "https://github.com/Illumina/pyflow/releases/download/v${PYFLOW_RELEASE}/pyflow-${PYFLOW_RELEASE}.tar.gz" .
+RUN tar -xzf "pyflow-${PYFLOW_RELEASE}.tar.gz" \
+ && cd "pyflow-${PYFLOW_RELEASE}" \
+ && apt-get install --no-install-recommends --yes \
+      python2.7 \
+ && python2.7 setup.py build install \
+ && cd .. \
+ && rm -r "pyflow-${PYFLOW_RELEASE}" "pyflow-${PYFLOW_RELEASE}.tar.gz"
+
+
+## Install ocropy ##
+ENV OCROPY_RELEASE=1.3.3
+ADD "https://github.com/tmbdev/ocropy/archive/v${OCROPY_RELEASE}.tar.gz" .
+RUN tar -xzf "v${OCROPY_RELEASE}.tar.gz" \
+ && cd "ocropy-${OCROPY_RELEASE}" \
+ && apt-get install --no-install-recommends --yes \
       python-pil \
       python-tk \
       $(cat PACKAGES) \
- && rm -rf /var/lib/apt/lists/* \
  && python2.7 setup.py install \
  && cd .. \
- && rm -rf \
-      "ocropy-${OCROPY_VERSION}" \
-      "v${OCROPY_VERSION}.tar.gz"
-
-ENV PYFLOW_VERSION=1.1.20
-ADD "https://github.com/Illumina/pyflow/releases/download/v${PYFLOW_VERSION}/pyflow-${PYFLOW_VERSION}.tar.gz" .
-RUN tar -xzf "pyflow-${PYFLOW_VERSION}.tar.gz" \
- && cd "pyflow-${PYFLOW_VERSION}" \
- && python2.7 setup.py build install \
- && cd .. \
- && rm -rf \
-      "pyflow-${PYFLOW_VERSION}" \
-      "pyflow-${PYFLOW_VERSION}.tar.gz"
-
-RUN echo "deb https://notesalexp.org/tesseract-ocr/stretch/ stretch main" >> /etc/apt/sources.list \
- && wget -O - https://notesalexp.org/debian/alexp_key.asc | apt-key add - \
- && apt-get update \
- && apt-get install -y --no-install-recommends \
-      tesseract-ocr  \
-      tesseract-ocr-deu \
-      tesseract-ocr-eng \
-      tesseract-ocr-enm \
-      tesseract-ocr-fra \
-      tesseract-ocr-frk \
-      tesseract-ocr-frm \
-      tesseract-ocr-ita \
-      tesseract-ocr-por \
-      tesseract-ocr-spa \
- && rm -rf /var/lib/apt/lists/*
+ && rm -r "ocropy-${OCROPY_RELEASE}" "v${OCROPY_RELEASE}.tar.gz"
 
 
-# Install OCR pipeline
-COPY hocrtotei /usr/local/bin
-COPY ocr /usr/local/bin
+## Install Tesseract OCR ##
+ENV TESSERACT_RELEASE=4.1.1
+ADD "https://github.com/tesseract-ocr/tesseract/archive/${TESSERACT_RELEASE}.tar.gz" .
+RUN tar -xzf "${TESSERACT_RELEASE}.tar.gz" \
+ && cd "tesseract-${TESSERACT_RELEASE}" \
+ && apt-get install --no-install-recommends --yes \
+      autoconf \
+      automake \
+      g++ \
+      libjpeg62-turbo-dev \
+      libleptonica-dev \
+      libtiff5-dev \
+      libtool \
+      libpng-dev \
+      make \
+      pkg-config \
+      zlib1g-dev \
+ && ./autogen.sh \
+ && ./configure \
+ && make \
+ && make install \
+ && ldconfig \
+ && cd - > /dev/null \
+ && rm -r "tesseract-${TESSERACT_RELEASE}" "${TESSERACT_RELEASE}.tar.gz"
+
+ENV TESSDATA_BEST_RELEASE=4.1.0
+ADD "https://github.com/tesseract-ocr/tessdata_best/archive/${TESSDATA_BEST_RELEASE}.tar.gz" .
+RUN tar -xzf "${TESSDATA_BEST_RELEASE}.tar.gz" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/ara.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/chi_tra.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/dan.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/deu.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/ell.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/eng.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/enm.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/fra.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/frk.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/frm.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/ita.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/por.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/rus.traineddata" "/usr/local/share/tessdata/" \
+ && mv "tessdata_best-${TESSDATA_BEST_RELEASE}/spa.traineddata" "/usr/local/share/tessdata/" \
+ && rm -r "tessdata_best-${TESSDATA_BEST_RELEASE}" "${TESSDATA_BEST_RELEASE}.tar.gz"
+
+
+## Further dependencies ##
+RUN apt-get install --no-install-recommends --yes \
+      ghostscript \
+      python-pip \
+      python3.7 \
+      zip \
+ && pip install natsort
+
+
+RUN rm -r /var/lib/apt/lists/*
+
+
+## Install Pipeline ##
+COPY hocrtotei ocr /usr/local/bin/
 
 
 ENTRYPOINT ["ocr"]

README.md

@@ -1,66 +1,53 @@
-# OCR
+# OCR - Optical Character Recognition
 
-## Build image
+This software implements a heavily parallelized pipeline to recognize text in PDF files. It is used for nopaque's OCR service but you can also use it standalone, for that purpose a convenient wrapper script is provided.
 
-1. Clone this repository and navigate into it:
-```
-git clone https://gitlab.ub.uni-bielefeld.de/sfb1288inf/ocr.git && cd ocr
+## Software used in this pipeline implementation
+- Official Debian Docker image (buster-slim) and programs from its free repositories: https://hub.docker.com/_/debian
+- ocropy (1.3.3): https://github.com/ocropus/ocropy/releases/tag/v1.3.3
+- pyFlow (1.1.20): https://github.com/Illumina/pyflow/releases/tag/v1.1.20
+- Tesseract OCR (4.1.1): https://github.com/tesseract-ocr/tesseract/releases/tag/4.1.1
+- tessdata_best (4.1.0): https://github.com/tesseract-ocr/tessdata_best/releases/tag/4.1.0
+
+
+## Use this image
+
+1. Create input and output directories for the pipeline.
+``` bash
+mkdir -p /<my_data_location>/input /<my_data_location>/output
 ```
 
-2. Build image:
-```
-docker build -t sfb1288inf/ocr:latest .
-```
+2. Place your PDF files inside `/<my_data_location>/input`. Files should all contain text of the same language.
 
-Alternatively build from the GitLab repository without cloning:
-
-1. Build image:
-```
-docker build -t sfb1288inf/ocr:latest https://gitlab.ub.uni-bielefeld.de/sfb1288inf/ocr.git
+3. Start the pipeline process. Check the [Pipeline arguments](#pipeline-arguments) section for more details.
 ```
+# Option one: Use the wrapper script
+## Install the wrapper script (only on first run). Get it from https://gitlab.ub.uni-bielefeld.de/sfb1288inf/ocr/-/raw/1.0.0/wrapper/ocr, make it executeable and add it to your ${PATH}
+cd /<my_data_location>
+ocr -i input -l <language_code> -o output <optional_pipeline_arguments>
 
-## Download prebuilt image
-
-The GitLab registry provides a prebuilt image. It is automatically created, utilizing the conquaire build servers.
-
-1. Download image:
-```
-docker pull gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:latest
-```
-
-## Run
-
-1. Create input and output directories for the OCR software:
-```
-mkdir -p /<mydatalocation>/files_for_ocr /<mydatalocation>/files_from_ocr
-```
-
-2. Place your files inside the `/<mydatalocation>/files_for_ocr` directory. Files can either be PDF (.pdf) or multipage TIFF (.tiff, .tif) files. Files should all contain text of the same language.
-
-3. Start the OCR process.
-```
+# Option two: Classic Docker style
 docker run \
     --rm \
     -it \
     -u $(id -u $USER):$(id -g $USER) \
-    -v /<mydatalocation>/files_for_ocr:/input \
-    -v /<mydatalocation>/files_from_ocr:/output \
-    sfb1288inf/ocr:latest \
+    -v /<my_data_location>/input:/input \
+    -v /<my_data_location>/output:/output \
+    gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:1.0.0 \
         -i /input \
-        -l <languagecode> \
-        -o /output
+        -l <language_code>
+        -o /output \
+        <optional_pipeline_arguments>
 ```
-The arguments below `sfb1288inf/ocr:latest` are described in the [OCR arguments](#ocr-arguments) part.
 
-If you want to use the prebuilt image, replace `sfb1288inf/ocr:latest` with `gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:latest`.
+4. Check your results in the `/<my_data_location>/output` directory.
+```
 
-4. Check your results in the `/<mydatalocation>/files_from_ocr` directory.
-
-### OCR arguments
+### Pipeline arguments
 
 `-l languagecode`
 * Tells tesseract which language will be used.
-* options = deu (German), eng (English), enm (Middle englisch), fra (French), frk (German Fraktur), frm (Middle french), ita (Italian), por (Portuguese), spa (Spanish)
+* options = ara (Arabic), chi_tra (Chinese - Traditional), dan (Danish), deu (German), ell (Greek, Modern (1453-)), eng (English), enm (Middle englisch), fra (French), frk (German Fraktur), frm (Middle french), ita (Italian), por (Portuguese), rus (Russian), spa (Spanish)
 * required = True
 
 `--keep-intermediates`
@@ -78,15 +65,15 @@ kept.
 * Used to skip binarization with ocropus. If skipped, only the tesseract binarization is used.
 * default = False
 
-Example with all arguments used:
-```
+``` bash
+# Example with all arguments used
 docker run \
     --rm \
     -it \
     -u $(id -u $USER):$(id -g $USER) \
-    -v "$HOME"/ocr/files_for_ocr:/input \
-    -v "$HOME"/ocr/files_from_ocr:/output \
-    sfb1288inf/ocr:latest \
+    -v "$HOME"/ocr/input:/input \
+    -v "$HOME"/ocr/output:/output \
+    gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:1.0.0 \
         -i /input \
         -l eng \
         -o /output \

hocrtotei

@@ -1,26 +1,18 @@
-#!/usr/bin/env python3.5
+#!/usr/bin/env python3.7
 # coding=utf-8
 
+""""Merges hOCR files into a TEI file."""
+
 from xml.sax.saxutils import escape
-import argparse
+from argparse import ArgumentParser
 import xml.etree.ElementTree as ET
 
-parser = argparse.ArgumentParser(
-    description='Merges several hOCR files in order of their occurrence on command line to one TEI result file.'
-)
-parser.add_argument(
-    'i',
-    metavar='hOCR-sourcefile',
-    nargs='+'
-)
-parser.add_argument(
-    'o',
-    metavar='TEI-destfile',
-)
+parser = ArgumentParser(description='Merges hOCR files into a TEI file.')
+parser.add_argument('i', metavar='hOCR-sourcefile', nargs='+')
+parser.add_argument('o', metavar='TEI-destfile',)
 args = parser.parse_args()
 
 output_file = open(args.o, 'w')
-
 output_file.write(
       '<?xml version="1.0" encoding="UTF-8"?>\n'
     + '<TEI xmlns="http://www.tei-c.org/ns/1.0" xml:id="dtabf">\n'
@@ -54,5 +46,4 @@ output_file.write(
       '        </body>\n'
     + '    </text>\n'
     + '</TEI>')
-
 output_file.close()

ocr

@@ -1,556 +1,467 @@
 #!/usr/bin/env python2.7
 # coding=utf-8
 
+"""An OCR pipeline for PDF file processing."""
 
-"""
-ocr
+__author__ = 'Patrick Jentsch <p.jentsch@uni-bielefeld.de>'
+__version__ = '1.0.0'
 
-Usage:  For usage instructions run with option --help
-Author: Patrick Jentsch <p.jentsch@uni-bielefeld.de>
-"""
-
-
-import argparse
+from argparse import ArgumentParser
+from natsort import natsorted
+from pyflow import WorkflowRunner
 import multiprocessing
 import os
-import re
 import sys
-from pyflow import WorkflowRunner
+import tempfile
 
 
-def parse_arguments():
-    parser = argparse.ArgumentParser(
-        description='''Performs OCR of (historical) documents utilizing OCRopus
-                       for preprocessing and Tesseract OCR for OCR. The results
-                       are served as hOCR, PDF, raw text and TEI compliant XML
-                       files.\n
-                       Software requirements: imagemagick, ocropus, pdftoppm,
-                       pdfunite, poppler-utils, pyflow, python2.7, python3.5,
-                       tesseract'''
-    )
-    parser.add_argument(
-        '-i',
-        dest='input_dir',
-        required=True
-    )
-    parser.add_argument(
-        '-l',
-        choices=[
-            'deu', 'eng', 'enm', 'fra', 'frk', 'frm', 'ita', 'por', 'spa'
-        ],
-        dest='lang',
-        required=True
-    )
-    parser.add_argument(
-        '-o',
-        dest='output_dir',
-        required=True
-    )
-    parser.add_argument(
-        '--skip-binarisation',
-        action='store_true',
-        default=False,
-        dest='skip_binarisation',
-        help='skip ocropy binarisation',
-        required=False
-    )
-    parser.add_argument(
-        '--keep-intermediates',
-        action='store_true',
-        default=False,
-        dest='keep_intermediates',
-        help='keep intermediate files',
-        required=False
-    )
-    parser.add_argument(
-        '--nCores',
-        default=min(4, multiprocessing.cpu_count()),
-        dest='n_cores',
-        help='total number of cores available',
-        required=False,
-        type=int
-    )
-    parser.add_argument(
-        '--zip',
-        default='ocr-result-files',
-        dest='zip',
-        type=str,
-        help='package result files in zip bundles and asign an filename prefix',
-        required=False
+TESSERACT_MODELS = ['deu', 'eng', 'enm', 'fra', 'frk', 'frm', 'ita', 'por', 'spa']  # noqa
+
+
+def parse_args():
+    parser = ArgumentParser(
+        description='An OCR pipeline for PDF file processing.',
+        prog='OCR pipeline'
     )
+    parser.add_argument('-i', '--input-directory',
+                        help='Input directory (only PDF files get processed)',
+                        required=True)
+    parser.add_argument('-o', '--output-directory',
+                        help='Output directory',
+                        required=True)
+    parser.add_argument('-l', '--language',
+                        choices=TESSERACT_MODELS,
+                        required=True)
+    parser.add_argument('--binarize',
+                        action='store_true',
+                        help='Use ocropy binarisation as preprocessing step.')
+    parser.add_argument('--log-dir')
+    parser.add_argument('--n-cores',
+                        default=min(4, multiprocessing.cpu_count()),
+                        help='Total number of cores available.',
+                        type=int)
+    parser.add_argument('--intermediate-directory')
+    parser.add_argument('--zip',
+                        help='Zips all results in different archives depending'
+                             ' on result types. Also zips everything into one '
+                             'archive.')
+    parser.add_argument('-v', '--version',
+                        action='version',
+                        version='%(prog)s {}'.format(__version__))
     return parser.parse_args()
 
 
-class OCRWorkflow(WorkflowRunner):
-    def __init__(self, args):
-        self.jobs = analyze_jobs(args.input_dir, args.output_dir)
-        self.skip_binarisation = args.skip_binarisation
-        self.keep_intermediates = args.keep_intermediates
-        self.lang = args.lang
-        self.n_cores = args.n_cores
-        self.output_dir = args.output_dir
-        self.zip = args.zip
+class OCRPipelineJob:
+    """An OCR pipeline job class
+
+    Each input file of the pipeline is represented as an OCR pipeline job,
+    which holds all necessary information for the pipeline to process it.
+
+    Arguments:
+    file -- Path to the file
+    output_dir -- Path to a directory, where job results a stored
+    intermediate_dir -- Path to a directory, where intermediate files are
+                        stored.
+    """
+
+    def __init__(self, file, output_dir, intermediate_dir):
+        self.file = file
+        self.intermediate_dir = intermediate_dir
+        self.name = os.path.basename(file).rsplit('.', 1)[0]
+        self.output_dir = output_dir
+
+
+class OCRPipeline(WorkflowRunner):
+    def __init__(self, input_dir, lang, output_dir, binarize, intermediate_dir,
+                 n_cores, zip):
+        self.input_dir = input_dir
+        self.lang = lang
+        self.output_dir = output_dir
+        self.binarize = binarize
+        if intermediate_dir is None:
+            self.intermediate_dir = os.path.join(output_dir, 'tmp')
+        else:
+            self.intermediate_dir = tempfile.mkdtemp(dir=intermediate_dir)
+        self.n_cores = n_cores
+        if zip is None:
+            self.zip = zip
+        else:
+            if zip.lower().endswith('.zip'):
+                # Remove .zip file extension if provided
+                self.zip = zip[:-4]
+                self.zip = self.zip if self.zip else 'output'
+            else:
+                self.zip = zip
+        self.jobs = collect_jobs(self.input_dir,
+                                 self.output_dir,
+                                 self.intermediate_dir)
 
     def workflow(self):
-        if len(self.jobs) == 0:
+        if not self.jobs:
             return
 
         '''
         ' ##################################################
-        ' # Create output directories                      #
+        ' # setup output directory                         #
         ' ##################################################
         '''
-        create_output_directories_jobs = []
-        for index, job in enumerate(self.jobs):
-            cmd = 'mkdir -p "%s"' % (
-                os.path.join(job['output_dir'], 'tmp')
-            )
-            if self.keep_intermediates:
-                cmd += ' "%s" "%s" "%s" "%s"' % (
-                    os.path.join(job['output_dir'], 'tmp', 'hocr'),
-                    os.path.join(job['output_dir'], 'tmp', 'pdf'),
-                    os.path.join(job['output_dir'], 'tmp', 'tiff'),
-                    os.path.join(job['output_dir'], 'tmp', 'txt')
-                )
-            if not self.skip_binarisation:
-                cmd += ' "%s"' % (
-                    os.path.join(job['output_dir'], 'tmp', 'bin.png')
-                )
-            create_output_directories_jobs.append(
-                self.addTask(
-                    command=cmd,
-                    label='create_output_directories_job_-_%i' % (index)
-                )
-            )
+        setup_output_directory_tasks = []
+        for i, job in enumerate(self.jobs):
+            cmd = 'mkdir'
+            cmd += ' -p'
+            cmd += ' "{}"'.format(job.intermediate_dir)
+            cmd += ' "{}"'.format(os.path.join(job.output_dir, 'poco'))
+            lbl = 'setup_output_directory_-_{}'.format(i)
+            task = self.addTask(command=cmd, label=lbl)
+            setup_output_directory_tasks.append(task)
 
         '''
         ' ##################################################
-        ' # Split                                          #
+        ' # split input                                    #
         ' ##################################################
         '''
-        split_jobs = []
-        split_job_n_cores = min(
-            self.n_cores,
-            max(1, int(self.n_cores / len(self.jobs)))
-        )
-        for index, job in enumerate(self.jobs):
-            if job['filename'].endswith(('.tif', '.tiff')):
-                '''
-                ' This command also works for PDF input but ocropus-nlbin
-                ' is not able to handle the TIFF output of it.
-                '''
-                cmd = 'convert -density 300 "%s" -compress LZW -scene 1 "%s/page-%%d.tif"' % (
-                    job['path'],
-                    os.path.join(job['output_dir'], 'tmp')
-                )
-            else:
-                cmd = 'pdftoppm -r 300 -tiff -tiffcompression lzw "%s" "%s"' % (
-                    job['path'],
-                    os.path.join(job['output_dir'], 'tmp', 'page')
-                )
+        split_input_tasks = []
+        n_cores = min(self.n_cores, max(1, int(self.n_cores / len(self.jobs))))
+        for i, job in enumerate(self.jobs):
+            input_file = job.file
+            output_file = '{}/page-%d.tif'.format(job.intermediate_dir)
+            cmd = 'gs'
+            cmd += ' -dBATCH'
+            cmd += ' -dNOPAUSE'
+            cmd += ' -dNumRenderingThreads={}'.format(n_cores)
+            cmd += ' -dQUIET'
+            cmd += ' -r300'
+            cmd += ' -sDEVICE=tiff24nc'
+            cmd += ' -sCompression=lzw'
+            cmd += ' "-sOutputFile={}"'.format(output_file)
+            cmd += ' "{}"'.format(input_file)
+            deps = 'setup_output_directory_-_{}'.format(i)
+            lbl = 'split_input_-_{}'.format(i)
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl, nCores=n_cores)  # noqa
+            split_input_tasks.append(task)
 
-            split_jobs.append(
-                self.addTask(
-                    command=cmd,
-                    dependencies='create_output_directories_job_-_%i' % (index),
-                    label='split_job_-_%i' % (index),
-                    nCores=split_job_n_cores
-                )
-            )
-
-        if not self.skip_binarisation:
+        if self.binarize:
             '''
-            ' The binarisation_jobs are created based of the output files of
-            ' the split_jobs. So wait until they are finished.
+            ' The binarization_tasks list is created based on the output files
+            ' of the split_tasks. So wait until they are finished.
             '''
             self.waitForTasks()
 
             '''
             ' ##################################################
-            ' # Binarise                                       #
+            ' # binarization                                   #
             ' ##################################################
             '''
-            binarisation_jobs = []
+            binarization_tasks = []
             '''
             ' We run ocropus-nlbin with either four or, if there are less then
             ' four cores available for this workflow, the available core
             ' number.
             '''
-            binarisation_job_n_cores = min(4, self.n_cores)
-            for index, job in enumerate(self.jobs):
-                files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
-                files = filter(lambda x: x.endswith('.tif'), files)
-                files.sort(key=lambda x: int(re.search(r'\d+', x).group(0)))
-                files = map(
-                    lambda x: '"' + os.path.join(job['output_dir'], 'tmp', x) + '"',
-                    files
-                )
-                cmd = 'ocropus-nlbin --output "%s" --parallel "%i" %s' % (
-                    os.path.join(job['output_dir'], 'tmp'),
-                    binarisation_job_n_cores,
-                    ' '.join(files)
-                )
-                binarisation_jobs.append(
-                    self.addTask(
-                        command=cmd,
-                        dependencies='split_job_-_%i' % (index),
-                        label='binarisation_job_-_%i' % (index),
-                        nCores=binarisation_job_n_cores
-                    )
-                )
+            n_cores = min(4, self.n_cores)
+            for i, job in enumerate(self.jobs):
+                input_dir = job.intermediate_dir
+                output_dir = job.intermediate_dir
+                files = filter(lambda x: x.endswith('.tif'), os.listdir(input_dir))  # noqa
+                files = natsorted(files)
+                files = map(lambda x: os.path.join(input_dir, x), files)
+                cmd = 'ocropus-nlbin "{}"'.format('" "'.join(files))
+                cmd += ' --nocheck'
+                cmd += ' --output "{}"'.format(output_dir)
+                cmd += ' --parallel "{}"'.format(n_cores)
+                print(cmd)
+                deps = 'split_input_-_{}'.format(i)
+                lbl = 'binarization_-_{}'.format(i)
+                task = self.addTask(command=cmd, dependencies=deps, label=lbl, nCores=n_cores)  # noqa
+                binarization_tasks.append(task)
 
-            '''
-            ' The post_binarisation_jobs are created based of the output files
-            ' of the binarisation_jobs. So wait until they are finished.
-            '''
             self.waitForTasks()
 
             '''
             ' ##################################################
-            ' # Normalise file names from binarisation         #
+            ' # Renaming of binarization output files          #
             ' ##################################################
             '''
-            post_binarisation_jobs = []
-            for index, job in enumerate(self.jobs):
-                number = 0
-                files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
-                files = filter(lambda x: x.endswith('.bin.png'), files)
-                files.sort()
+            for i, job in enumerate(self.jobs):
+                input_dir = job.intermediate_dir
+                output_dir = job.intermediate_dir
+                files = filter(lambda x: x.endswith('.bin.png'), os.listdir(input_dir))  # noqa
                 for file in files:
-                    cmd = 'mv "%s" "%s"' % (
-                        os.path.join(job['output_dir'], 'tmp', file),
-                        os.path.join(
-                            job['output_dir'],
-                            'tmp',
-                            'page-%i.bin.png' % (int(file.split('.', 1)[0]))
-                        )
-                    )
-                    post_binarisation_jobs.append(
-                        self.addTask(
-                            command=cmd,
-                            dependencies='binarisation_job_-_%i' % (index),
-                            label='post_binarisation_job_-_%i-%i' % (
-                                index,
-                                number
-                            )
-                        )
-                    )
-                    number += 1
+                    # int conversion is done in order to trim leading zeros
+                    page_number = int(file.split('.', 1)[0])
+                    output_file = 'page-{}.bin.png'.format(page_number)
+                    os.rename(os.path.join(output_dir, file),
+                              os.path.join(output_dir, output_file))
 
         '''
-        ' The ocr_jobs are created based of the output files of either the
-        ' split_jobs or post_binarisation_jobs. So wait until they are
+        ' The ocr_tasks are created based of the output files of either the
+        ' split_tasks or binarization_tasks. So wait until they are
         ' finished.
         '''
         self.waitForTasks()
 
         '''
         ' ##################################################
-        ' # Optical Character Recognition                  #
+        ' # ocr                                            #
         ' ##################################################
         '''
-        ocr_jobs = []
-        '''
-        ' Tesseract runs fastest with four cores. So we run it with either four
-        ' or, if there are less then four cores available for this workflow,
-        ' the available core number.
-        '''
-        ocr_job_n_cores = min(4, self.n_cores)
-        for index, job in enumerate(self.jobs):
-            files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
-            if self.skip_binarisation:
-                files = filter(lambda x: x.endswith('.tif'), files)
-            else:
+        ocr_tasks = []
+        for i, job in enumerate(self.jobs):
+            input_dir = job.intermediate_dir
+            output_dir = job.intermediate_dir
+            files = os.listdir(input_dir)
+            if self.binarize:
+                deps = 'binarization_-_{}'.format(i)
                 files = filter(lambda x: x.endswith('.bin.png'), files)
-            files.sort(key=lambda x: int(re.search(r'\d+', x).group(0)))
-            files = map(
-                lambda x: os.path.join(job['output_dir'], 'tmp', x),
-                files
-            )
-            number = 0
-            for file in files:
-                cmd = 'tesseract "%s" "%s" -l "%s" hocr pdf txt' % (
-                    file,
-                    os.path.join(
-                        job['output_dir'],
-                        'tmp',
-                        file.rsplit('.', 1 if self.skip_binarisation else 2)[0]
-                    ),
-                    self.lang
-                )
-                if self.skip_binarisation:
-                    ocr_job_dependencies = 'split_job_-_%i' % (index)
+            else:
+                deps = 'split_input_-_{}'.format(i)
+                files = filter(lambda x: x.endswith('.tif'), files)
+            files = natsorted(files)
+            files = map(lambda x: os.path.join(input_dir, x), files)
+            for j, file in enumerate(files):
+                if self.binarize:
+                    output_file_base = os.path.join(output_dir, file.rsplit('.', 2)[0])  # noqa
                 else:
-                    ocr_job_dependencies = filter(
-                        lambda x: x == 'post_binarisation_job_-_%i-%i' % (
-                            index,
-                            number
-                        ),
-                        post_binarisation_jobs
-                    )
-                ocr_jobs.append(
-                    self.addTask(
-                        command=cmd,
-                        dependencies=ocr_job_dependencies,
-                        label='ocr_job_-_%i-%i' % (index, number),
-                        nCores=ocr_job_n_cores
-                    )
-                )
-                number += 1
+                    output_file_base = os.path.join(output_dir, file.rsplit('.', 1)[0])  # noqa
+                cmd = 'tesseract "{}" "{}"'.format(file, output_file_base)
+                cmd += ' -l "{}"'.format(self.lang)
+                cmd += ' hocr pdf txt'
+                cmd += ' && '
+                cmd += 'sed -i \'s+{}/++g\' "{}".hocr'.format(input_dir, output_file_base)  # noqa
+                lbl = 'ocr_-_{}-{}'.format(i, j)
+                task = self.addTask(command=cmd, dependencies=deps, label=lbl, env={"OMP_THREAD_LIMIT": "1"})  # noqa
+                ocr_tasks.append(task)
 
         '''
         ' The following jobs are created based of the output files of the
-        ' ocr_jobs. So wait until they are finished.
+        ' ocr_tasks. So wait until they are finished.
         '''
         self.waitForTasks()
 
         '''
         ' ##################################################
-        ' # Create TEI P5 files                            #
+        ' # combined pdf creation                          #
         ' ##################################################
         '''
-        hocr_to_tei_jobs = []
-        for index, job in enumerate(self.jobs):
-            files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
-            files = filter(lambda x: x.endswith('.hocr'), files)
-            files.sort(key=lambda x: int(re.search(r'\d+', x).group(0)))
-            files = map(
-                lambda x: '"' + os.path.join(job['output_dir'], 'tmp', x) + '"',
-                files
-            )
-            cmd = 'hocrtotei %s "%s"' % (
-                ' '.join(files),
-                os.path.join(
-                    job['output_dir'],
-                    os.path.join(job['output_dir'], job['name'] + '.xml')
-                )
-            )
-            hocr_to_tei_jobs.append(
-                self.addTask(
-                    command=cmd,
-                    dependencies=filter(
-                        lambda x: x.startswith('ocr_job_-_%i' % (index)),
-                        ocr_jobs
-                    ),
-                    label='hocr_to_tei_job_-_%i' % (index)
-                )
-            )
+        combined_pdf_creation_tasks = []
+        n_cores = min(self.n_cores, max(1, int(self.n_cores / len(self.jobs))))
+        for i, job in enumerate(self.jobs):
+            input_dir = job.intermediate_dir
+            output_file = os.path.join(job.output_dir, '{}.pdf'.format(job.name))  # noqa
+            files = filter(lambda x: x.endswith('.pdf'), os.listdir(input_dir))
+            files = natsorted(files)
+            files = map(lambda x: os.path.join(input_dir, x), files)
+            cmd = 'gs'
+            cmd += ' -dBATCH'
+            cmd += ' -dNOPAUSE'
+            cmd += ' -dNumRenderingThreads={}'.format(n_cores)
+            cmd += ' -dPDFSETTINGS=/ebook'
+            cmd += ' -dQUIET'
+            cmd += ' -sDEVICE=pdfwrite'
+            cmd += ' "-sOutputFile={}"'.format(output_file)
+            cmd += ' "{}"'.format('" "'.join(files))
+            deps = filter(lambda x: x.startswith('ocr_-_{}'.format(i)), ocr_tasks)  # noqa
+            lbl = 'combined_pdf_creation_-_{}'.format(i)
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl, nCores=n_cores)  # noqa
+            combined_pdf_creation_tasks.append(task)
 
         '''
         ' ##################################################
-        ' # Merge PDF files                                #
+        ' # combined txt creation                          #
         ' ##################################################
         '''
-        pdf_merge_jobs = []
-        for index, job in enumerate(self.jobs):
-            files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
-            files = filter(lambda x: x.endswith('.pdf'), files)
-            files.sort(key=lambda x: int(re.search(r'\d+', x).group(0)))
-            files = map(
-                lambda x: '"' + os.path.join(job['output_dir'], 'tmp', x) + '"',
-                files
-            )
-            cmd = 'pdfunite %s "%s"' % (
-                ' '.join(files),
-                os.path.join(
-                    job['output_dir'],
-                    os.path.join(job['output_dir'], job['name'] + '.pdf')
-                )
-            )
-            pdf_merge_jobs.append(
-                self.addTask(
-                    command=cmd,
-                    dependencies=filter(
-                        lambda x: x.startswith('ocr_job_-_%i' % (index)),
-                        ocr_jobs
-                    ),
-                    label='pdf_merge_job_-_%i' % (index)
-                )
-            )
+        combined_txt_creation_tasks = []
+        for i, job in enumerate(self.jobs):
+            input_dir = job.intermediate_dir
+            output_file = os.path.join(job.output_dir, '{}.txt'.format(job.name))  # noqa
+            files = filter(lambda x: x.endswith('.txt'), os.listdir(input_dir))
+            files = natsorted(files)
+            files = map(lambda x: os.path.join(input_dir, x), files)
+            cmd = 'cat "{}" > "{}"'.format('" "'.join(files), output_file)
+            deps = filter(lambda x: x.startswith('ocr_-_{}'.format(i)), ocr_tasks)  # noqa
+            lbl = 'combined_txt_creation_-_{}'.format(i)
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+            combined_txt_creation_tasks.append(task)
 
         '''
         ' ##################################################
-        ' # Merge text files                               #
+        ' # tei p5 creation                                #
         ' ##################################################
         '''
-        txt_merge_jobs = []
-        for index, job in enumerate(self.jobs):
-            files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
-            files = filter(lambda x: x.endswith('.txt'), files)
-            files.sort(key=lambda x: int(re.search(r'\d+', x).group(0)))
-            files = map(
-                lambda x: '"' + os.path.join(job['output_dir'], 'tmp', x) + '"',
-                files
-            )
-            cmd = 'cat %s > "%s"' % (
-                ' '.join(files),
-                os.path.join(
-                    job['output_dir'],
-                    os.path.join(job['output_dir'], job['name'] + '.txt')
-                )
-            )
-            txt_merge_jobs.append(
-                self.addTask(
-                    command=cmd,
-                    dependencies=filter(
-                        lambda x: x.startswith('ocr_job_-_%i' % (index)),
-                        ocr_jobs
-                    ),
-                    label='txt_merge_job_-_%i' % (index)
-                )
-            )
-
-        if self.zip:
-            all_zip_jobs = []
-            all_zip_job_dependencies = (hocr_to_tei_jobs
-                                        + pdf_merge_jobs
-                                        + txt_merge_jobs)
-            cmd = 'cd "%s" && zip "%s"-all-ocr-files.zip */*.{pdf,txt,xml} -x "pyflow.data*" && cd -' % (
-                self.output_dir,
-                self.zip
-            )
-            all_zip_jobs.append(
-                self.addTask(
-                    command=cmd,
-                    dependencies=all_zip_job_dependencies,
-                    label='all_zip_job'
-                )
-            )
-
-            pdf_zip_jobs = []
-            pdf_zip_job_dependencies = all_zip_jobs
-            cmd = 'cd "%s" && zip -m "%s"-ocr-pdf.zip */*.pdf -x "pyflow.data*" && cd -' % (
-                self.output_dir,
-                self.zip
-            )
-            pdf_zip_jobs.append(
-                self.addTask(
-                    command=cmd,
-                    dependencies=pdf_zip_job_dependencies,
-                    label='pdf_zip_job'
-                )
-            )
-
-            txt_zip_jobs = []
-            txt_zip_job_dependencies = all_zip_jobs
-            cmd = 'cd "%s" && zip -m "%s"-ocr-txt.zip */*.txt -x "pyflow.data*" && cd -' % (
-                self.output_dir,
-                self.zip
-            )
-            txt_zip_jobs.append(
-                self.addTask(
-                    command=cmd,
-                    dependencies=txt_zip_job_dependencies,
-                    label='txt_zip_job'
-                )
-            )
-
-            xml_zip_jobs = []
-            xml_zip_job_dependencies = all_zip_jobs
-            cmd = 'cd "%s" && zip -m "%s"-ocr-xml.zip */*.xml -x "pyflow.data*" && cd -' % (
-                self.output_dir,
-                self.zip
-            )
-            xml_zip_jobs.append(
-                self.addTask(
-                    command=cmd,
-                    dependencies=xml_zip_job_dependencies,
-                    label='xml_zip_job'
-                )
-            )
+        tei_p5_creation_tasks = []
+        for i, job in enumerate(self.jobs):
+            input_dir = job.intermediate_dir
+            output_file = os.path.join(job.output_dir, '{}.xml'.format(job.name))  # noqa
+            files = filter(lambda x: x.endswith('.hocr'),
+                           os.listdir(input_dir))
+            files = natsorted(files)
+            files = map(lambda x: os.path.join(input_dir, x), files)
+            cmd = 'hocrtotei "{}" "{}"'.format('" "'.join(files),
+                                               output_file)
+            deps = filter(lambda x: x.startswith('ocr_-_{}'.format(i)), ocr_tasks)  # noqa
+            lbl = 'tei_p5_creation_-_{}'.format(i)
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+            tei_p5_creation_tasks.append(task)
 
         '''
         ' ##################################################
-        ' # Cleanup                                        #
+        ' # poco bundle creation                           #
         ' ##################################################
         '''
-        cleanup_jobs = []
-        if self.keep_intermediates:
-            for index, job in enumerate(self.jobs):
-                cleanup_job_dependencies = [
-                    'hocr_to_tei_job_-_%i' % (index),
-                    'pdf_merge_job_-_%i' % (index),
-                    'txt_merge_job_-_%i' % (index)
-                ]
-                cmd = 'mv "%s"/*.hocr "%s"' % (
-                    os.path.join(job['output_dir'], 'tmp'),
-                    os.path.join(job['output_dir'], 'tmp', 'hocr'),
-                )
-                cmd += ' && mv "%s"/*.pdf "%s"' % (
-                    os.path.join(job['output_dir'], 'tmp'),
-                    os.path.join(job['output_dir'], 'tmp', 'pdf'),
-                )
-                cmd += ' && mv "%s"/*.tif "%s"' % (
-                    os.path.join(job['output_dir'], 'tmp'),
-                    os.path.join(job['output_dir'], 'tmp', 'tiff'),
-                )
-                cmd += ' && mv "%s"/*.txt "%s"' % (
-                    os.path.join(job['output_dir'], 'tmp'),
-                    os.path.join(job['output_dir'], 'tmp', 'txt'),
-                )
-                if not self.skip_binarisation:
-                    cmd += ' && mv "%s"/*.bin.png "%s"' % (
-                        os.path.join(job['output_dir'], 'tmp'),
-                        os.path.join(job['output_dir'], 'tmp', 'bin.png'),
-                    )
-                    cmd += ' && rm "%s"/*.nrm.png' % (
-                        os.path.join(job['output_dir'], 'tmp')
-                    )
-                cleanup_jobs.append(
-                    self.addTask(
-                        command=cmd,
-                        dependencies=cleanup_job_dependencies,
-                        label='cleanup_job_-_%i' % (index)
-                    )
-                )
-        else:
-            for index, job in enumerate(self.jobs):
-                cleanup_job_dependencies = [
-                    'hocr_to_tei_job_-_%i' % (index),
-                    'pdf_merge_job_-_%i' % (index),
-                    'txt_merge_job_-_%i' % (index)
-                ]
-                cmd = 'rm -r "%s"' % (
-                    os.path.join(job['output_dir'], 'tmp')
-                )
-                cleanup_jobs.append(
-                    self.addTask(
-                        command=cmd,
-                        dependencies=cleanup_job_dependencies,
-                        label='cleanup_job_-_%i' % (index)
-                    )
-                )
+        poco_bundle_creation_tasks = []
+        for i, job in enumerate(self.jobs):
+            input_dir = job.intermediate_dir
+            output_dir = os.path.join(job.output_dir, 'poco')
+            files = os.listdir(input_dir)
+            if self.binarize:
+                files = filter(lambda x: x.endswith(('.bin.png', '.hocr')), files)  # noqa
+            else:
+                files = filter(lambda x: x.endswith(('.tif', '.hocr')), files)
+            files = natsorted(files)
+            files = map(lambda x: os.path.join(input_dir, x), files)
+            cmd = 'mv "{}" "{}"'.format('" "'.join(files), output_dir)
+            deps = filter(lambda x: x.startswith('ocr_-_{}'.format(i)), ocr_tasks)  # noqa
+            deps.append('tei_p5_creation_-_{}'.format(i))
+            lbl = 'poco_bundle_creation_-_{}'.format(i)
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+            poco_bundle_creation_tasks.append(task)
+
+        '''
+        ' The following jobs are created based of the output files of the
+        ' combined_pdf_creation_tasks. So wait until they are finished.
+        '''
+        self.waitForTasks()
+
+        '''
+        ' ##################################################
+        ' # cleanup                                        #
+        ' ##################################################
+        '''
+        cleanup_tasks = []
+        for i, job in enumerate(self.jobs):
+            input_dir = job.intermediate_dir
+            cmd = 'rm -r "{}"'.format(input_dir)
+            deps = ['combined_pdf_creation_-_{}'.format(i),
+                    'combined_txt_creation_-_{}'.format(i),
+                    'poco_bundle_creation_-_{}'.format(i),
+                    'tei_p5_creation_-_{}'.format(i)]
+            lbl = 'job_cleanup_-_{}'.format(i)
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+            cleanup_tasks.append(task)
+
+        input_dir = self.intermediate_dir
+        cmd = 'rm -r "{}"'.format(input_dir)
+        deps = cleanup_tasks
+        lbl = 'pipeline_cleanup'
+        task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+        cleanup_tasks.append(task)
+
+        self.waitForTasks()
+
+        '''
+        ' ##################################################
+        ' # zip creation                                   #
+        ' ##################################################
+        '''
+        zip_creation_tasks = []
+        if self.zip is not None:
+            # zip all files
+            cmd = 'cd "{}"'.format(self.output_dir)
+            cmd += ' && '
+            cmd += 'zip'
+            cmd += ' -r'
+            cmd += ' "{}.all.zip" .'.format(self.zip)
+            cmd += ' -x "pyflow.data*" "*tmp*"'
+            cmd += ' -i "*.pdf" "*.txt" "*.xml" "*.hocr" "*.{}"'.format('bin.png' if self.binarize else 'tif')  # noqa
+            cmd += ' && '
+            cmd += 'cd -'
+            deps = combined_pdf_creation_tasks + combined_txt_creation_tasks + poco_bundle_creation_tasks  # noqa
+            lbl = 'zip_creation_-_all'
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+            zip_creation_tasks.append(task)
+            # zip PDF files
+            cmd = 'cd "{}"'.format(self.output_dir)
+            cmd += ' && '
+            cmd += 'zip'
+            cmd += ' -r'
+            cmd += ' "{}.pdf.zip" .'.format(self.zip)
+            cmd += ' -x "pyflow.data*" "*tmp*"'
+            cmd += ' -i "*.pdf"'
+            cmd += ' && '
+            cmd += 'cd -'
+            deps = combined_pdf_creation_tasks
+            lbl = 'zip_creation_-_pdf'
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+            zip_creation_tasks.append(task)
+            # zip TXT files
+            cmd = 'cd "{}"'.format(self.output_dir)
+            cmd += ' && '
+            cmd += 'zip'
+            cmd += ' -r'
+            cmd += ' "{}.txt.zip" .'.format(self.zip)
+            cmd += ' -x "pyflow.data*" "*tmp*"'
+            cmd += ' -i "*.txt"'
+            cmd += ' && '
+            cmd += 'cd -'
+            deps = combined_txt_creation_tasks
+            lbl = 'zip_creation_-_txt'
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+            zip_creation_tasks.append(task)
+            # zip XML files
+            cmd = 'cd "{}"'.format(self.output_dir)
+            cmd += ' && '
+            cmd += 'zip'
+            cmd += ' -r'
+            cmd += ' "{}.xml.zip" .'.format(self.zip)
+            cmd += ' -x "pyflow.data*" "*tmp*"'
+            cmd += ' -i "*.xml"'
+            cmd += ' && '
+            cmd += 'cd -'
+            deps = tei_p5_creation_tasks
+            lbl = 'zip_creation_-_xml'
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+            zip_creation_tasks.append(task)
+            # zip PoCo bundles
+            cmd = 'cd "{}"'.format(self.output_dir)
+            cmd += ' && '
+            cmd += 'zip'
+            cmd += ' -r'
+            cmd += ' "{}.poco.zip" .'.format(self.zip)
+            cmd += ' -x "pyflow.data*" "*tmp*"'
+            cmd += ' -i "*.hocr" "*.{}"'.format('bin.png' if self.binarize else 'tif')  # noqa
+            cmd += ' && '
+            cmd += 'cd -'
+            deps = poco_bundle_creation_tasks
+            lbl = 'zip_creation_-_poco'
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+            zip_creation_tasks.append(task)
 
 
-def analyze_jobs(input_dir, output_dir):
+def collect_jobs(input_dir, output_dir, intermediate_dir):
     jobs = []
-
     for file in os.listdir(input_dir):
         if os.path.isdir(os.path.join(input_dir, file)):
-            jobs += analyze_jobs(
-                os.path.join(input_dir, file),
-                os.path.join(output_dir, file)
-            )
-        elif file.endswith(('.pdf', '.tif', '.tiff')):
-            jobs.append(
-                {
-                    'filename': file,
-                    'name': file.rsplit('.', 1)[0],
-                    'output_dir': os.path.join(output_dir, file),
-                    'path': os.path.join(input_dir, file)
-                }
-            )
-
+            jobs += collect_jobs(os.path.join(input_dir, file),
+                                 os.path.join(output_dir, file),
+                                 os.path.join(intermediate_dir, file))
+        elif file.lower().endswith('.pdf'):
+            job = OCRPipelineJob(os.path.join(input_dir, file),
+                                 os.path.join(output_dir, file),
+                                 os.path.join(intermediate_dir, file))
+            jobs.append(job)
     return jobs
 
 
 def main():
-    args = parse_arguments()
-
-    wflow = OCRWorkflow(args)
-
-    retval = wflow.run(dataDirRoot=args.output_dir, nCores=args.n_cores)
-
+    args = parse_args()
+    ocr_pipeline = OCRPipeline(args.input_directory, args.language,
+                               args.output_directory, args.binarize,
+                               args.intermediate_directory, args.n_cores,
+                               args.zip)
+    retval = ocr_pipeline.run(
+        dataDirRoot=(args.log_dir or args.output_directory),
+        nCores=args.n_cores
+    )
     sys.exit(retval)
 
 

wrapper/ocr

@@ -1,39 +1,43 @@
 #!/usr/bin/env python3
 # coding=utf-8
 
-import argparse
+"""A wrapper to execute the OCR pipeline in a Docker container"""
+
+from argparse import ArgumentParser
 import os
 import subprocess
 
-container_image = 'gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:latest'
-container_input_dir = '/input'
-container_output_dir = '/output'
-uid = str(os.getuid())
-gid = str(os.getgid())
+CONTAINER_IMAGE_TAG = '1.0.0'
+CONTAINER_IMAGE = 'gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:{}'.format(CONTAINER_IMAGE_TAG)  # noqa
+CONTAINER_INPUT_DIR = '/input'
+CONTAINER_INTERMEDIATE_DIR = '/intermediate'
+CONTAINER_OUTPUT_DIR = '/output'
+UID = str(os.getuid())
+GID = str(os.getgid())
 
-parser = argparse.ArgumentParser(add_help=False)
-parser.add_argument(
-    '-i',
-    dest='input_dir',
-    required=False
-)
-parser.add_argument(
-    '-o',
-    dest='output_dir',
-    required=False
-)
+parser = ArgumentParser(add_help=False)
+parser.add_argument('-i', '--input-directory')
+parser.add_argument('-o', '--output-directory')
+parser.add_argument('--intermediate-directory')
 args, remaining_args = parser.parse_known_args()
 
-cmd = ['docker', 'run', '--rm', '-it', '-u', uid + ':' + gid]
-if args.input_dir is not None:
-    host_input_dir = os.path.abspath(args.input_dir)
-    cmd += ['-v', host_input_dir + ':' + container_input_dir]
-    remaining_args += ['-i', container_input_dir]
-if args.output_dir is not None:
-    host_output_dir = os.path.abspath(args.output_dir)
-    cmd += ['-v', host_output_dir + ':' + container_output_dir]
-    remaining_args += ['-o', container_output_dir]
-cmd.append(container_image)
+cmd = ['docker', 'run', '--rm', '-it', '-u', '{}:{}'.format(UID, GID)]
+if args.intermediate_directory is not None:
+    cmd += ['-v', '{}:{}'.format(os.path.abspath(args.intermediate_directory),
+                                 CONTAINER_INTERMEDIATE_DIR)]
+    remaining_args.insert(0, CONTAINER_INTERMEDIATE_DIR)
+    remaining_args.insert(0, '--intermediate-directory')
+if args.output_directory is not None:
+    cmd += ['-v', '{}:{}'.format(os.path.abspath(args.output_directory),
+                                 CONTAINER_OUTPUT_DIR)]
+    remaining_args.insert(0, CONTAINER_OUTPUT_DIR)
+    remaining_args.insert(0, '-o')
+if args.input_directory is not None:
+    cmd += ['-v', '{}:{}'.format(os.path.abspath(args.input_directory),
+                                 CONTAINER_INPUT_DIR)]
+    remaining_args.insert(0, CONTAINER_INPUT_DIR)
+    remaining_args.insert(0, '-i')
+cmd.append(CONTAINER_IMAGE)
 cmd += remaining_args
 
 subprocess.run(cmd)