Don't process files in subdirectories

.gitlab-ci.yml

@@ -1,44 +1,68 @@
-image: docker:stable
+image: docker:19.03.13
 
 services:
-  - docker:stable-dind
+  - docker:19.03.13-dind
-
-variables:
-  DOCKER_DRIVER: overlay2
 
 stages:
   - build
   - push
 
-before_script:
+variables:
-  - docker login -u gitlab-ci-token -p $CI_JOB_TOKEN $CI_REGISTRY
+  DOCKER_TLS_CERTDIR: "/certs"
+  INTERMEDIATE_IMAGE_TAG: $CI_REGISTRY_IMAGE:$CI_COMMIT_REF_NAME-$CI_COMMIT_SHA
 
-Build:
+.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_image:
   script:
-    - docker build --pull -t $CI_REGISTRY_IMAGE:tmp .
+    - docker login -u gitlab-ci-token -p $CI_JOB_TOKEN $CI_REGISTRY
-    - docker push $CI_REGISTRY_IMAGE:tmp
+    - docker build -t $INTERMEDIATE_IMAGE_TAG .
+    - docker push $INTERMEDIATE_IMAGE_TAG
   stage: build
   tags:
-  - docker
+    - docker
 
-Push latest:
+push_master:
+  extends:
+    - .reg_setup
   only:
     - master
   script:
-    - docker pull $CI_REGISTRY_IMAGE:tmp
+    - docker login -u gitlab-ci-token -p $CI_JOB_TOKEN $CI_REGISTRY
-    - docker tag $CI_REGISTRY_IMAGE:tmp $CI_REGISTRY_IMAGE:latest
+    - docker pull $INTERMEDIATE_IMAGE_TAG
-    - docker push $CI_REGISTRY_IMAGE:latest
+    - /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
 
-Push tag:
+push_other:
+  extends:
+    - .reg_setup
+  except:
+    - master
   only:
+    - branches
     - tags
   script:
-    - docker pull $CI_REGISTRY_IMAGE:tmp
+    - docker login -u gitlab-ci-token -p $CI_JOB_TOKEN $CI_REGISTRY
-    - docker tag $CI_REGISTRY_IMAGE:tmp $CI_REGISTRY_IMAGE:$CI_COMMIT_REF_NAME
+    - docker pull $INTERMEDIATE_IMAGE_TAG
-    - docker push $CI_REGISTRY_IMAGE:$CI_COMMIT_REF_NAME
+    - /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

Dockerfile

@@ -1,73 +1,96 @@
-FROM debian:9-slim
+FROM debian:buster-slim
 
 
-# Define image metadata
+LABEL authors="Patrick Jentsch <p.jentsch@uni-bielefeld.de>, Stephan Porada <porada@posteo.de>"
-LABEL maintainer="inf_sfb1288@lists.uni-bielefeld.de"
 
 
 ENV LANG=C.UTF-8
 
 
-# Install prerequisites
 RUN apt-get update \
- && apt-get install -y --no-install-recommends \
+ && apt-get install --no-install-recommends --yes \
-      apt-transport-https \
+      wget
-      ca-certificates \
-      gnupg2 \
-      imagemagick \
-      poppler-utils \
-      python2.7 \
-      python3.5 \
-      wget \
-      zip \
- && rm -rf /var/lib/apt/lists/*
 
-ENV OCROPY_VERSION 1.3.3
+
-ADD "https://github.com/tmbdev/ocropy/archive/v${OCROPY_VERSION}.tar.gz" .
+# Install the OCR pipeline and it's dependencies #
-RUN tar -xzf "v${OCROPY_VERSION}.tar.gz" \
+## Install pyFlow ##
+ENV PYFLOW_VERSION=1.1.20
+RUN wget --no-check-certificate --quiet \
+      "https://github.com/Illumina/pyflow/releases/download/v${PYFLOW_VERSION}/pyflow-${PYFLOW_VERSION}.tar.gz" \
+ && tar -xzf "pyflow-${PYFLOW_VERSION}.tar.gz" \
+ && cd "pyflow-${PYFLOW_VERSION}" \
+ && apt-get install --no-install-recommends --yes \
+      python2.7 \
+ && python2.7 setup.py build install \
+ && cd - > /dev/null \
+ && rm -r "pyflow-${PYFLOW_VERSION}" "pyflow-${PYFLOW_VERSION}.tar.gz"
+
+
+## Install ocropy ##
+ENV OCROPY_VERSION=1.3.3
+RUN wget --no-check-certificate --quiet \
+      "https://github.com/tmbdev/ocropy/archive/v${OCROPY_VERSION}.tar.gz" \
+ && tar -xzf "v${OCROPY_VERSION}.tar.gz" \
  && cd "ocropy-${OCROPY_VERSION}" \
- && apt-get update \
+ && apt-get install --no-install-recommends --yes \
- && apt-get install -y --no-install-recommends \
+      python2.7 \
       python-pil \
       python-tk \
       $(cat PACKAGES) \
- && rm -rf /var/lib/apt/lists/* \
  && python2.7 setup.py install \
- && cd .. \
+ && cd - > /dev/null \
- && rm -rf \
+ && rm -r "ocropy-${OCROPY_VERSION}" "v${OCROPY_VERSION}.tar.gz"
-      "ocropy-${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/*
 
 
-# Install OCR pipeline
+## Install Tesseract OCR ##
-COPY hocrtotei /usr/local/bin
+ENV TESSERACT_VERSION=4.1.1
-COPY ocr /usr/local/bin
+RUN wget --no-check-certificate --quiet \
+      "https://github.com/tesseract-ocr/tesseract/archive/${TESSERACT_VERSION}.tar.gz" \
+ && tar -xzf "${TESSERACT_VERSION}.tar.gz" \
+ && cd "tesseract-${TESSERACT_VERSION}" \
+ && apt-get install --no-install-recommends --yes \
+      autoconf \
+      automake \
+      g++ \
+      libjpeg62-turbo-dev \
+      libleptonica-dev \
+      libtiff5-dev \
+      libtool \
+      libpng-dev \
+      make \
+      pkg-config \
+      zlib1g-dev \
+ && ./autogen.sh \
+ && ./configure \
+ && make \
+ && make install \
+ && ldconfig \
+ && cd - > /dev/null \
+ && rm -r "tesseract-${TESSERACT_VERSION}" "${TESSERACT_VERSION}.tar.gz"
+
+ENV TESSERACT_MODELS="ara,chi_tra,dan,deu,ell,eng,enm,fra,frk,frm,ita,por,rus,spa"
+ENV TESSDATA_BEST_VERSION=4.1.0
+RUN wget --no-check-certificate --quiet \
+      "https://github.com/tesseract-ocr/tessdata_best/archive/${TESSDATA_BEST_VERSION}.tar.gz" \
+ && tar -xzf "${TESSDATA_BEST_VERSION}.tar.gz" \
+ && for tesseract_model in $(echo ${TESSERACT_MODELS} | tr "," "\n"); do mv "tessdata_best-${TESSDATA_BEST_VERSION}/${tesseract_model}.traineddata" "/usr/local/share/tessdata/"; done \
+ && rm -r "tessdata_best-${TESSDATA_BEST_VERSION}" "${TESSDATA_BEST_VERSION}.tar.gz"
+
+
+## Further dependencies ##
+RUN apt-get install --no-install-recommends --yes \
+      procps \
+      ghostscript \
+      python3.7 \
+      rename \
+      zip
+
+
+## Install Pipeline ##
+COPY hocrtotei ocr /usr/local/bin/
+
+
+RUN rm -r /var/lib/apt/lists/*
 
 
 ENTRYPOINT ["ocr"]

README.md

@@ -1,96 +1,44 @@
-# 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:
+## Software used in this pipeline implementation
-```
+
-git clone https://gitlab.ub.uni-bielefeld.de/sfb1288inf/ocr.git && cd ocr
+- Official Debian Docker image (buster-slim): https://hub.docker.com/_/debian
+  - Software from Debian Buster's free repositories
+- 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:
+2. Place your PDF files inside `/<my_data_location>/input`. Files should all contain text of the same language.
-```
-docker build -t sfb1288inf/ocr:latest .
-```
 
-Alternatively build from the GitLab repository without cloning:
+3. Start the pipeline process. Check the pipeline help (`ocr --help`) for more details.
-
-1. Build image:
-```
-docker build -t sfb1288inf/ocr:latest https://gitlab.ub.uni-bielefeld.de/sfb1288inf/ocr.git
 ```
+# 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/development/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
+# Option two: Classic Docker style
-
-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.
-```
 docker run \
     --rm \
     -it \
     -u $(id -u $USER):$(id -g $USER) \
-    -v /<mydatalocation>/files_for_ocr:/input \
+    -v /<my_data_location>/input:/input \
-    -v /<mydatalocation>/files_from_ocr:/output \
+    -v /<my_data_location>/output:/output \
-    sfb1288inf/ocr:latest \
+    gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:development \
-        -i /input \
+        -i /ocr_pipeline/input \
-        -l <languagecode> \
+        -l <language_code> \
-        -o /output
+        -o /ocr_pipeline/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
-
-`-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)
-* required = True
-
-`--keep-intermediates`
-* If set, all intermediate files created during the OCR process will be
-kept.
-* default = False
-* required = False
-
-`--nCores corenumber`
-* Sets the number of CPU cores being used during the OCR process.
-* default = min(4, multiprocessing.cpu_count())
-* required = False
-
-`--skip-binarisation`
-* Used to skip binarization with ocropus. If skipped, only the tesseract binarization is used.
-* default = False
-
-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 \
-        -i /input \
-        -l eng \
-        -o /output \
-        --keep_intermediates \
-        --nCores 8 \
-        --skip-binarisation
-```

hocrtotei

@@ -1,58 +1,57 @@
-#!/usr/bin/env python3.5
+#!/usr/bin/env python3.7
 # coding=utf-8
 
+""""Convert hOCR to TEI XML."""
+
 from xml.sax.saxutils import escape
-import argparse
+from argparse import ArgumentParser
+import re
 import xml.etree.ElementTree as ET
 
-parser = argparse.ArgumentParser(
+parser = ArgumentParser(description='Convert hOCR to TEI XML.')
-    description='Merges several hOCR files in order of their occurrence on command line to one TEI result file.'
+parser.add_argument('input', metavar='Path to hOCR input file')
-)
+parser.add_argument('output', metavar='Path to TEI output 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')
+tei = ''
-
+tei += '<TEI xmlns="http://www.tei-c.org/ns/1.0">\n'
-output_file.write(
+tei += '  <teiHeader>\n'
-      '<?xml version="1.0" encoding="UTF-8"?>\n'
+tei += '    <fileDesc>\n'
-    + '<TEI xmlns="http://www.tei-c.org/ns/1.0" xml:id="dtabf">\n'
+tei += '      <titleStmt>\n'
-    + '    <teiHeader>\n'
+tei += '        <title></title>\n'
-    + '        <fileDesc>\n'
+tei += '      </titleStmt>\n'
-    + '            <titleStmt/>\n'
+tei += '      <publicationStmt>\n'
-    + '            <publicationStmt/>\n'
+tei += '        <p></p>\n'
-    + '            <sourceDesc/>\n'
+tei += '      </publicationStmt>\n'
-    + '        </fileDesc>\n'
+tei += '      <sourceDesc>\n'
-    + '        <encodingDesc/>\n'
+tei += '        <p></p>\n'
-    + '        <profileDesc/>\n'
+tei += '      </sourceDesc>\n'
-    + '    </teiHeader>\n'
+tei += '    </fileDesc>\n'
-    + '    <text>\n'
+tei += '  </teiHeader>\n'
-    + '        <body>\n'
+tei += '  <text>\n'
-)
+tei += '    <body>\n'
-for index, input_file in enumerate(args.i):
+# Conversion start
-    tree = ET.parse(input_file)
+hocr = ET.parse(args.input)
-    output_file.write('            <pb n="%i"/>\n' % (index + 1))
+for page in hocr.findall('.//*[@class="ocr_page"]'):
-    for para in tree.findall('.//*[@class="ocr_par"]'):
+    page_properties = page.attrib.get('title')
-        output_file.write('            <p>\n')
+    facsimile = re.search(r'image \"(.*?)\"', page_properties).group(1)
+    page_number = re.search(r'ppageno (\d+)', page_properties).group(1)
+    tei += '      <pb facs="{}" n="{}"/>\n'.format(facsimile, page_number)
+    for para in page.findall('.//*[@class="ocr_par"]'):
+        tei += '      <p>\n'
         for line in para.findall('.//*[@class="ocr_line"]'):
-            first_word_in_line = True
+            tei += '        <lb/>'
+            indent = ''
             for word in line.findall('.//*[@class="ocrx_word"]'):
                 if word.text is not None:
-                    output_file.write(('                ' if first_word_in_line else ' ') + escape(word.text.strip()))
+                    tei += indent + escape(word.text.strip())
-                    first_word_in_line = False
+                    indent = ' '
-            if not first_word_in_line:
+            tei += '\n'
-                output_file.write('<lb/>\n')
+        tei += '      </p>\n'
-        output_file.write('            </p>\n')
+# Conversion end
-output_file.write(
+tei += '    </body>\n'
-      '        </body>\n'
+tei += '  </text>\n'
-    + '    </text>\n'
+tei += '</TEI>\n'
-    + '</TEI>')
 
-output_file.close()
+with open(args.output, 'w') as tei_file:
+    tei_file.write(tei)

ocr

@@ -1,556 +1,357 @@
 #!/usr/bin/env python2.7
 # coding=utf-8
 
+"""OCR pipeline for PDF file processing."""
 
-"""
+__author__ = 'Patrick Jentsch <p.jentsch@uni-bielefeld.de>,' \
-ocr
+             'Stephan Porada <porada@posteo.de>'
+__version__ = '1.0.0'
 
-Usage:  For usage instructions run with option --help
+from argparse import ArgumentParser
-Author: Patrick Jentsch <p.jentsch@uni-bielefeld.de>
+from pyflow import WorkflowRunner
-"""
-
-
-import argparse
 import multiprocessing
 import os
-import re
 import sys
-from pyflow import WorkflowRunner
 
 
-def parse_arguments():
+class OCRPipelineJob:
-    parser = argparse.ArgumentParser(
+    """An OCR pipeline job class
-        description='''Performs OCR of (historical) documents utilizing OCRopus
+
-                       for preprocessing and Tesseract OCR for OCR. The results
+    Each input file of the pipeline is represented as an OCR pipeline job,
-                       are served as hOCR, PDF, raw text and TEI compliant XML
+    which holds all necessary information for the pipeline to process it.
-                       files.\n
+
-                       Software requirements: imagemagick, ocropus, pdftoppm,
+    Arguments:
-                       pdfunite, poppler-utils, pyflow, python2.7, python3.5,
+    file -- Path to the file
-                       tesseract'''
+    output_dir -- Path to a directory, where job results a stored
-    )
+    """
-    parser.add_argument(
+
-        '-i',
+    def __init__(self, file, output_dir):
-        dest='input_dir',
+        self.file = file
-        required=True
+        self.name = os.path.basename(file).rsplit('.', 1)[0]
-    )
+        self.output_dir = output_dir
-    parser.add_argument(
+        self.page_dir = os.path.join(output_dir, 'pages')
-        '-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
-    )
-    return parser.parse_args()
 
 
-class OCRWorkflow(WorkflowRunner):
+class OCRPipeline(WorkflowRunner):
-    def __init__(self, args):
+    def __init__(self, input_dir, lang, output_dir, binarize, zip):
-        self.jobs = analyze_jobs(args.input_dir, args.output_dir)
+        self.input_dir = input_dir
-        self.skip_binarisation = args.skip_binarisation
+        self.lang = lang
-        self.keep_intermediates = args.keep_intermediates
+        self.output_dir = output_dir
-        self.lang = args.lang
+        self.binarize = binarize
-        self.n_cores = args.n_cores
+        self.zip = zip
-        self.output_dir = args.output_dir
+        self.jobs = collect_jobs(self.input_dir, self.output_dir)
-        self.zip = args.zip
 
     def workflow(self):
-        if len(self.jobs) == 0:
+        if not self.jobs:
             return
 
         '''
         ' ##################################################
-        ' # Create output directories                      #
+        ' # setup output directory                         #
         ' ##################################################
         '''
-        create_output_directories_jobs = []
+        setup_output_directory_tasks = []
-        for index, job in enumerate(self.jobs):
+        for i, job in enumerate(self.jobs):
-            cmd = 'mkdir -p "%s"' % (
+            cmd = 'mkdir -p "{}"'.format(job.page_dir)
-                os.path.join(job['output_dir'], 'tmp')
+            lbl = 'setup_output_directory_-_{}'.format(i)
-            )
+            task = self.addTask(command=cmd, label=lbl)
-            if self.keep_intermediates:
+            setup_output_directory_tasks.append(task)
-                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)
-                )
-            )
 
         '''
         ' ##################################################
-        ' # Split                                          #
+        ' # split input                                    #
         ' ##################################################
         '''
-        split_jobs = []
+        split_input_tasks = []
-        split_job_n_cores = min(
+        n_cores = max(1, int(self.getNCores() / len(self.jobs)))
-            self.n_cores,
+        for i, job in enumerate(self.jobs):
-            max(1, int(self.n_cores / len(self.jobs)))
+            input_file = job.file
-        )
+            output_file = '{}/page-%d.tif'.format(job.page_dir)
-        for index, job in enumerate(self.jobs):
+            cmd = 'gs'
-            if job['filename'].endswith(('.tif', '.tiff')):
+            cmd += ' -dBATCH'
-                '''
+            cmd += ' -dNOPAUSE'
-                ' This command also works for PDF input but ocropus-nlbin
+            cmd += ' -dNumRenderingThreads={}'.format(n_cores)
-                ' is not able to handle the TIFF output of it.
+            cmd += ' -dQUIET'
-                '''
+            cmd += ' -r300'
-                cmd = 'convert -density 300 "%s" -compress LZW -scene 1 "%s/page-%%d.tif"' % (
+            cmd += ' -sDEVICE=tiff24nc'
-                    job['path'],
+            cmd += ' -sCompression=lzw'
-                    os.path.join(job['output_dir'], 'tmp')
+            cmd += ' "-sOutputFile={}"'.format(output_file)
-                )
+            cmd += ' "{}"'.format(input_file)
-            else:
+            deps = 'setup_output_directory_-_{}'.format(i)
-                cmd = 'pdftoppm -r 300 -tiff -tiffcompression lzw "%s" "%s"' % (
+            lbl = 'split_input_-_{}'.format(i)
-                    job['path'],
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl,
-                    os.path.join(job['output_dir'], 'tmp', 'page')
+                                nCores=n_cores)
-                )
+            split_input_tasks.append(task)
 
-            split_jobs.append(
+        if self.binarize:
-                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:
             '''
-            ' The binarisation_jobs are created based of the output files of
+            ' ##################################################
-            ' the split_jobs. So wait until they are finished.
+            ' # pre binarization                               #
+            ' ##################################################
             '''
-            self.waitForTasks()
+            pre_binarization_tasks = []
+            for i, job in enumerate(self.jobs):
+                input_file = os.path.join(job.output_dir, 'binarization_input_files.txt')  # noqa
+                cmd = 'ls -dv "{}/"* >> "{}"'.format(job.page_dir, input_file)
+                deps = 'split_input_-_{}'.format(i)
+                lbl = 'pre_binarization_-_{}'.format(i)
+                task = self.addTask(command=cmd, dependencies=deps, label=lbl)
+                pre_binarization_tasks.append(task)
 
             '''
             ' ##################################################
-            ' # Binarise                                       #
+            ' # binarization                                   #
             ' ##################################################
             '''
-            binarisation_jobs = []
+            binarization_tasks = []
-            '''
+            n_cores = self.getNCores()
-            ' We run ocropus-nlbin with either four or, if there are less then
+            mem_mb = self.getMemMb()
-            ' four cores available for this workflow, the available core
+            for i, job in enumerate(self.jobs):
-            ' number.
+                input_file = os.path.join(job.output_dir, 'binarization_input_files.txt')  # noqa
-            '''
+                cmd = 'ocropus-nlbin "@{}"'.format(input_file)
-            binarisation_job_n_cores = min(4, self.n_cores)
+                cmd += ' --nocheck'
-            for index, job in enumerate(self.jobs):
+                cmd += ' --output "{}"'.format(job.page_dir)
-                files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
+                cmd += ' --parallel "{}"'.format(n_cores)
-                files = filter(lambda x: x.endswith('.tif'), files)
+                deps = 'pre_binarization_-_{}'.format(i)
-                files.sort(key=lambda x: int(re.search(r'\d+', x).group(0)))
+                lbl = 'binarization_-_{}'.format(i)
-                files = map(
+                task = self.addTask(command=cmd, dependencies=deps, label=lbl,
-                    lambda x: '"' + os.path.join(job['output_dir'], 'tmp', x) + '"',
+                                    memMb=mem_mb, nCores=n_cores)
-                    files
+                binarization_tasks.append(task)
-                )
-                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
-                    )
-                )
-
-            '''
-            ' 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         #
+            ' # post binarization                              #
             ' ##################################################
             '''
-            post_binarisation_jobs = []
+            post_binarization_tasks = []
-            for index, job in enumerate(self.jobs):
+            for i, job in enumerate(self.jobs):
-                number = 0
+                input_file = os.path.join(job.output_dir, 'binarization_input_files.txt')  # noqa
-                files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
+                cmd = 'rm "{}"'.format(input_file)
-                files = filter(lambda x: x.endswith('.bin.png'), files)
+                cmd += ' && '
-                files.sort()
+                cmd += 'cd "{}"'.format(job.page_dir)
-                for file in files:
+                cmd += ' && '
-                    cmd = 'mv "%s" "%s"' % (
+                cmd += 'rm *.{nrm.png,tif}'
-                        os.path.join(job['output_dir'], 'tmp', file),
+                cmd += ' && '
-                        os.path.join(
+                cmd += 'rename \'s/^0*/page-/\' *'
-                            job['output_dir'],
+                cmd += ' && '
-                            'tmp',
+                cmd += 'cd -'
-                            'page-%i.bin.png' % (int(file.split('.', 1)[0]))
+                deps = 'binarization_-_{}'.format(i)
-                        )
+                lbl = 'post_binarization_-_{}'.format(i)
-                    )
+                task = self.addTask(command=cmd, dependencies=deps, label=lbl)
-                    post_binarisation_jobs.append(
+                post_binarization_tasks.append(task)
-                        self.addTask(
-                            command=cmd,
-                            dependencies='binarisation_job_-_%i' % (index),
-                            label='post_binarisation_job_-_%i-%i' % (
-                                index,
-                                number
-                            )
-                        )
-                    )
-                    number += 1
-
-        '''
-        ' The ocr_jobs are created based of the output files of either the
-        ' split_jobs or post_binarisation_jobs. So wait until they are
-        ' finished.
-        '''
-        self.waitForTasks()
 
         '''
         ' ##################################################
-        ' # Optical Character Recognition                  #
+        ' # pre ocr                                        #
         ' ##################################################
         '''
-        ocr_jobs = []
+        pre_ocr_tasks = []
-        '''
+        for i, job in enumerate(self.jobs):
-        ' Tesseract runs fastest with four cores. So we run it with either four
+            input_file = os.path.join(job.output_dir, 'ocr_input_files.txt')
-        ' or, if there are less then four cores available for this workflow,
+            cmd = 'ls -dv "{}/"* >> "{}"'.format(job.page_dir, input_file)
-        ' the available core number.
+            deps = 'post_binarization_-_{}'.format(i) if self.binarize else 'split_input_-_{}'.format(i)  # noqa
-        '''
+            lbl = 'pre_ocr_-_{}'.format(i)
-        ocr_job_n_cores = min(4, self.n_cores)
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
-        for index, job in enumerate(self.jobs):
+            pre_ocr_tasks.append(task)
-            files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
-            if self.skip_binarisation:
-                files = filter(lambda x: x.endswith('.tif'), files)
-            else:
-                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:
-                    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
-
-        '''
-        ' The following jobs are created based of the output files of the
-        ' ocr_jobs. So wait until they are finished.
-        '''
-        self.waitForTasks()
 
         '''
         ' ##################################################
-        ' # Create TEI P5 files                            #
+        ' # ocr                                            #
         ' ##################################################
         '''
-        hocr_to_tei_jobs = []
+        ocr_tasks = []
-        for index, job in enumerate(self.jobs):
+        n_cores = min(4, self.getNCores())
-            files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
+        mem_mb = min(n_cores * 2048, self.getMemMb())
-            files = filter(lambda x: x.endswith('.hocr'), files)
+        for i, job in enumerate(self.jobs):
-            files.sort(key=lambda x: int(re.search(r'\d+', x).group(0)))
+            input_file = os.path.join(job.output_dir, 'ocr_input_files.txt')
-            files = map(
+            output_file_base = os.path.join(job.output_dir, job.name)
-                lambda x: '"' + os.path.join(job['output_dir'], 'tmp', x) + '"',
+            cmd = 'tesseract "{}" "{}"'.format(input_file, output_file_base)
-                files
+            cmd += ' -l "{}"'.format(self.lang)
-            )
+            cmd += ' hocr pdf txt'
-            cmd = 'hocrtotei %s "%s"' % (
+            deps = 'pre_ocr_-_{}'.format(i)
-                ' '.join(files),
+            lbl = 'ocr_-_{}'.format(i)
-                os.path.join(
+            task = self.addTask(command=cmd, dependencies=deps,
-                    job['output_dir'],
+                                env={'OMP_THREAD_LIMIT': '{}'.format(n_cores)},
-                    os.path.join(job['output_dir'], job['name'] + '.xml')
+                                label=lbl, memMb=mem_mb, nCores=n_cores)
-                )
+            ocr_tasks.append(task)
-            )
-            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)
-                )
-            )
 
         '''
         ' ##################################################
-        ' # Merge PDF files                                #
+        ' # post ocr                                       #
         ' ##################################################
         '''
-        pdf_merge_jobs = []
+        post_ocr_tasks = []
-        for index, job in enumerate(self.jobs):
+        for i, job in enumerate(self.jobs):
-            files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
+            input_file = os.path.join(job.output_dir, 'ocr_input_files.txt')
-            files = filter(lambda x: x.endswith('.pdf'), files)
+            output_file_base = os.path.join(job.output_dir, job.name)
-            files.sort(key=lambda x: int(re.search(r'\d+', x).group(0)))
+            cmd = 'rm "{}"'.format(input_file)
-            files = map(
+            cmd += ' && '
-                lambda x: '"' + os.path.join(job['output_dir'], 'tmp', x) + '"',
+            cmd += 'sed -i \'s+{}+pages+g\' "{}.hocr"'.format(job.page_dir, output_file_base)  # noqa
-                files
+            deps = 'ocr_-_{}'.format(i)
-            )
+            lbl = 'post_ocr_-_{}'.format(i)
-            cmd = 'pdfunite %s "%s"' % (
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
-                ' '.join(files),
+            post_ocr_tasks.append(task)
-                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)
-                )
-            )
 
         '''
         ' ##################################################
-        ' # Merge text files                               #
+        ' # hocr to tei                                    #
         ' ##################################################
         '''
-        txt_merge_jobs = []
+        hocr_to_tei_tasks = []
-        for index, job in enumerate(self.jobs):
+        for i, job in enumerate(self.jobs):
-            files = os.listdir(os.path.join(job['output_dir'], 'tmp'))
+            output_file_base = os.path.join(job.output_dir, job.name)
-            files = filter(lambda x: x.endswith('.txt'), files)
+            cmd = 'hocrtotei "{}.hocr" "{}.xml"'.format(output_file_base, output_file_base)  # noqa
-            files.sort(key=lambda x: int(re.search(r'\d+', x).group(0)))
+            deps = 'post_ocr_-_{}'.format(i)
-            files = map(
+            lbl = 'hocr_to_tei_-_{}'.format(i)
-                lambda x: '"' + os.path.join(job['output_dir'], 'tmp', x) + '"',
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
-                files
+            hocr_to_tei_tasks.append(task)
-            )
-            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'
-                )
-            )
 
         '''
         ' ##################################################
-        ' # Cleanup                                        #
+        ' # zip creation                                   #
         ' ##################################################
         '''
-        cleanup_jobs = []
+        zip_creation_tasks = []
-        if self.keep_intermediates:
+        if self.zip is not None:
-            for index, job in enumerate(self.jobs):
+            # zip all files
-                cleanup_job_dependencies = [
+            cmd = 'cd "{}"'.format(self.output_dir)
-                    'hocr_to_tei_job_-_%i' % (index),
+            cmd += ' && '
-                    'pdf_merge_job_-_%i' % (index),
+            cmd += 'zip'
-                    'txt_merge_job_-_%i' % (index)
+            cmd += ' -r'
-                ]
+            cmd += ' "{}.all.zip" .'.format(self.zip)
-                cmd = 'mv "%s"/*.hocr "%s"' % (
+            cmd += ' -x "pyflow.data*" "*tmp*"'
-                    os.path.join(job['output_dir'], 'tmp'),
+            cmd += ' -i "*.pdf" "*.txt" "*.xml" "*.hocr" "*.{}"'.format('bin.png' if self.binarize else 'tif')  # noqa
-                    os.path.join(job['output_dir'], 'tmp', 'hocr'),
+            cmd += ' && '
-                )
+            cmd += 'cd -'
-                cmd += ' && mv "%s"/*.pdf "%s"' % (
+            deps = hocr_to_tei_tasks
-                    os.path.join(job['output_dir'], 'tmp'),
+            lbl = 'zip_creation_-_all'
-                    os.path.join(job['output_dir'], 'tmp', 'pdf'),
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
-                )
+            zip_creation_tasks.append(task)
-                cmd += ' && mv "%s"/*.tif "%s"' % (
+            # zip PDF files
-                    os.path.join(job['output_dir'], 'tmp'),
+            cmd = 'cd "{}"'.format(self.output_dir)
-                    os.path.join(job['output_dir'], 'tmp', 'tiff'),
+            cmd += ' && '
-                )
+            cmd += 'zip'
-                cmd += ' && mv "%s"/*.txt "%s"' % (
+            cmd += ' -r'
-                    os.path.join(job['output_dir'], 'tmp'),
+            cmd += ' "{}.pdf.zip" .'.format(self.zip)
-                    os.path.join(job['output_dir'], 'tmp', 'txt'),
+            cmd += ' -x "pyflow.data*" "*tmp*"'
-                )
+            cmd += ' -i "*.pdf"'
-                if not self.skip_binarisation:
+            cmd += ' && '
-                    cmd += ' && mv "%s"/*.bin.png "%s"' % (
+            cmd += 'cd -'
-                        os.path.join(job['output_dir'], 'tmp'),
+            deps = ocr_tasks
-                        os.path.join(job['output_dir'], 'tmp', 'bin.png'),
+            lbl = 'zip_creation_-_pdf'
-                    )
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
-                    cmd += ' && rm "%s"/*.nrm.png' % (
+            zip_creation_tasks.append(task)
-                        os.path.join(job['output_dir'], 'tmp')
+            # zip TXT files
-                    )
+            cmd = 'cd "{}"'.format(self.output_dir)
-                cleanup_jobs.append(
+            cmd += ' && '
-                    self.addTask(
+            cmd += 'zip'
-                        command=cmd,
+            cmd += ' -r'
-                        dependencies=cleanup_job_dependencies,
+            cmd += ' "{}.txt.zip" .'.format(self.zip)
-                        label='cleanup_job_-_%i' % (index)
+            cmd += ' -x "pyflow.data*" "*tmp*"'
-                    )
+            cmd += ' -i "*.txt"'
-                )
+            cmd += ' && '
-        else:
+            cmd += 'cd -'
-            for index, job in enumerate(self.jobs):
+            deps = ocr_tasks
-                cleanup_job_dependencies = [
+            lbl = 'zip_creation_-_txt'
-                    'hocr_to_tei_job_-_%i' % (index),
+            task = self.addTask(command=cmd, dependencies=deps, label=lbl)
-                    'pdf_merge_job_-_%i' % (index),
+            zip_creation_tasks.append(task)
-                    'txt_merge_job_-_%i' % (index)
+            # zip XML files
-                ]
+            cmd = 'cd "{}"'.format(self.output_dir)
-                cmd = 'rm -r "%s"' % (
+            cmd += ' && '
-                    os.path.join(job['output_dir'], 'tmp')
+            cmd += 'zip'
-                )
+            cmd += ' -r'
-                cleanup_jobs.append(
+            cmd += ' "{}.xml.zip" .'.format(self.zip)
-                    self.addTask(
+            cmd += ' -x "pyflow.data*" "*tmp*"'
-                        command=cmd,
+            cmd += ' -i "*.xml"'
-                        dependencies=cleanup_job_dependencies,
+            cmd += ' && '
-                        label='cleanup_job_-_%i' % (index)
+            cmd += 'cd -'
-                    )
+            deps = hocr_to_tei_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 = post_ocr_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):
     jobs = []
-
     for file in os.listdir(input_dir):
         if os.path.isdir(os.path.join(input_dir, file)):
-            jobs += analyze_jobs(
+            continue
-                os.path.join(input_dir, file),
+        if file.lower().endswith('.pdf'):
-                os.path.join(output_dir, file)
+            job = OCRPipelineJob(os.path.join(input_dir, file),
-            )
+                                 os.path.join(output_dir, file))
-        elif file.endswith(('.pdf', '.tif', '.tiff')):
+            jobs.append(job)
-            jobs.append(
-                {
-                    'filename': file,
-                    'name': file.rsplit('.', 1)[0],
-                    'output_dir': os.path.join(output_dir, file),
-                    'path': os.path.join(input_dir, file)
-                }
-            )
-
     return jobs
 
 
+def parse_args():
+    parser = ArgumentParser(description='OCR pipeline for PDF file processing',
+                            prog='OCR pipeline')
+    parser.add_argument('-i', '--input-dir',
+                        help='Input directory',
+                        required=True)
+    parser.add_argument('-o', '--output-dir',
+                        help='Output directory',
+                        required=True)
+    parser.add_argument('-l', '--language',
+                        choices=list(map(lambda x: x[:-12], filter(lambda x: x.endswith('.traineddata'), os.listdir('/usr/local/share/tessdata')))),  # noqa
+                        help='Language of the input '
+                             '(3-character ISO 639-2 language codes)',
+                        required=True)
+    parser.add_argument('--binarize',
+                        action='store_true',
+                        help='Add binarization as a preprocessing step')
+    parser.add_argument('--log-dir',
+                        help='Logging directory')
+    parser.add_argument('--mem-mb',
+                        help='Amount of system memory to be used (Default: min(--n-cores * 2048, available system memory))',  # noqa
+                        type=int)
+    parser.add_argument('--n-cores',
+                        default=min(4, multiprocessing.cpu_count()),
+                        help='Number of CPU threads to be used (Default: min(4, number of CPUs))',  # noqa
+                        type=int)
+    parser.add_argument('--zip',
+                        help='Create one zip file per filetype')
+    parser.add_argument('-v', '--version',
+                        action='version',
+                        help='Returns the current version of the OCR pipeline',
+                        version='%(prog)s {}'.format(__version__))
+    args = parser.parse_args()
+
+    # Set some tricky default values and check for insufficient input
+    if args.log_dir is None:
+        args.log_dir = args.output_dir
+    if args.n_cores < 1:
+        raise Exception('--n-cores must be greater or equal 1')
+    if args.mem_mb is None:
+        max_mem_mb = int(os.popen('free -t -m').readlines()[-1].split()[1:][0])
+        args.mem_mb = min(args.n_cores * 2048, max_mem_mb)
+    if args.mem_mb < 2048:
+        raise Exception('--mem-mb must be greater or equal 2048')
+    if args.zip is not None and args.zip.lower().endswith('.zip'):
+        # Remove .zip file extension if provided
+        args.zip = args.zip[:-4]
+        args.zip = args.zip if args.zip else 'output'
+    return args
+
+
 def main():
-    args = parse_arguments()
+    args = parse_args()
-
+    ocr_pipeline = OCRPipeline(args.input_dir, args.language, args.output_dir, args.binarize, args.zip)  # noqa
-    wflow = OCRWorkflow(args)
+    retval = ocr_pipeline.run(dataDirRoot=args.log_dir, memMb=args.mem_mb, nCores=args.n_cores)  # noqa
-
-    retval = wflow.run(dataDirRoot=args.output_dir, nCores=args.n_cores)
-
     sys.exit(retval)
 
 

wrapper/ocr

@@ -1,39 +1,38 @@
 #!/usr/bin/env python3
 # coding=utf-8
 
-import argparse
+from argparse import ArgumentParser
 import os
 import subprocess
+import sys
 
-container_image = 'gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:latest'
+CONTAINER_IMAGE = 'gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:1.0.0'
-container_input_dir = '/input'
+CONTAINER_INPUT_DIR = '/input'
-container_output_dir = '/output'
+CONTAINER_OUTPUT_DIR = '/output'
-uid = str(os.getuid())
+CONTAINER_LOG_DIR = '/logs'
-gid = str(os.getgid())
+UID = str(os.getuid())
+GID = str(os.getgid())
 
-parser = argparse.ArgumentParser(add_help=False)
+parser = ArgumentParser(add_help=False)
-parser.add_argument(
+parser.add_argument('-i', '--input-dir')
-    '-i',
+parser.add_argument('-o', '--output-dir')
-    dest='input_dir',
+parser.add_argument('--log-dir')
-    required=False
-)
-parser.add_argument(
-    '-o',
-    dest='output_dir',
-    required=False
-)
 args, remaining_args = parser.parse_known_args()
 
-cmd = ['docker', 'run', '--rm', '-it', '-u', uid + ':' + gid]
+cmd = ['docker', 'run', '--rm', '-it', '-u', '{}:{}'.format(UID, GID)]
 if args.input_dir is not None:
-    host_input_dir = os.path.abspath(args.input_dir)
+    mapping = os.path.abspath(args.input_dir) + ':' + CONTAINER_INPUT_DIR
-    cmd += ['-v', host_input_dir + ':' + container_input_dir]
+    cmd += ['-v', mapping]
-    remaining_args += ['-i', 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)
+    mapping = os.path.abspath(args.output_dir) + ':' + CONTAINER_OUTPUT_DIR
-    cmd += ['-v', host_output_dir + ':' + container_output_dir]
+    cmd += ['-v', mapping]
-    remaining_args += ['-o', container_output_dir]
+    remaining_args += ['-o', CONTAINER_OUTPUT_DIR]
-cmd.append(container_image)
+if args.log_dir is not None:
+    mapping = os.path.abspath(args.log_dir) + ':' + CONTAINER_LOG_DIR
+    cmd += ['-v', mapping]
+    remaining_args += ['--log-dir', CONTAINER_LOG_DIR]
+cmd.append(CONTAINER_IMAGE)
 cmd += remaining_args
 
-subprocess.run(cmd)
+sys.exit(subprocess.run(cmd).returncode)