Merge branch 'develop'

debian and ubuntu both install unpaper 0.4.2 or so. No .deb packages
available at higher version numbers although ArchLinux had something.
Considered making a separate image to handle building and install but
decided that was a premature optimization at this point, so just build
the unpaper that works. All tests pass.

.dockerignore

@@ -0,0 +1,21 @@
+bin/
+build/
+dist/
+include/
+lib/
+ocrmypdf.egg-info/
+staging/
+.git/
+.ruffus_history.sqlite
+MANIFEST.in
+*.sublime*
+*.pdf
+*.rst
+*.pyc
+*/*.pyc
+*/*/*.pyc
+*/*/*/*.pyc
+*/*/*/*/*.pyc
+*/*/*/*/*/*.pyc
+*/*/*/*/*/*/*.pyc
+*/*/*/*/*/*/*/*.pyc

.gitattributes

@@ -0,0 +1,8 @@
+# Always use Unix convention for new lines
+* text eol=lf
+
+# These files are binary and should be left untouched
+# (binary is a macro for -text -diff)
+*.jar	binary
+*.pdf	binary
+*.PDF	binary

.gitignore

@@ -0,0 +1,17 @@
+tmp/
+log/
+*.pyc
+tests/output/
+.ruffus_history.sqlite
+*.sublime-*
+/*.pdf
+build/
+dist/
+*.egg-info/
+venv/
+*/test/output
+bin/
+include/
+lib/
+pip-selfcheck.json
+pyvenv.cfg

.travis.yml

@@ -1,29 +0,0 @@
-
-language: generic
-sudo: required
-dist: trusty
-
-before_install:
-  - sudo add-apt-repository ppa:heyarje/libav-11 -y
-  - sudo add-apt-repository ppa:alex-p/tesseract-ocr -y
-  - sudo add-apt-repository ppa:evl.ms/evil -y  # for a newer version (2.5.0) of pngquant
-  - sudo apt-get update -qq
-  - sudo apt-get install libleptonica-dev -y    # required to build jbig2enc
-  - sudo apt-get install zlib1g-dev -y          # required to build jbig2enc
- # - sudo apt-get install imagemagick -y         # required to convert logo to desktop icon
-
-script:
-  - export OCRMYPDF_VERSION=8.3.2
-  - bash build-appimage.sh
-  # remove previous installed libraries to ensure that the tests use the libraries of the AppImage
-  - sudo apt-get remove libleptonica-dev zlib1g-dev -y
-  - bash test/test-appimage.sh
-  - wget https://github.com/probonopd/uploadtool/raw/master/upload.sh
-
-after_success:
-  - bash upload.sh OCRmyPDF*.AppImage
-
-branches:
-  except:
-    # Do not build tags that we create when we upload to GitHub Releases
-    - /^(?i:continuous)/

Dockerfile

@@ -0,0 +1,87 @@
+# OCRmyPDF
+#
+# VERSION               3.0.0
+FROM      debian:stretch
+MAINTAINER James R. Barlow <jim@purplerock.ca>
+
+# Add unprivileged user
+RUN useradd docker \
+  && mkdir /home/docker \
+  && chown docker:docker /home/docker
+
+# Update system and install our dependencies
+RUN apt-get update && apt-get install -y --no-install-recommends \
+  locales \
+  ghostscript \
+  tesseract-ocr \
+  tesseract-ocr-deu tesseract-ocr-spa tesseract-ocr-eng tesseract-ocr-fra \
+  qpdf \
+  poppler-utils \
+  python3 \
+  python3-pip \
+  python3-venv \
+  python3-reportlab \
+  python3-pil
+
+# Enforce UTF-8
+# Borrowed from https://index.docker.io/u/crosbymichael/python/ 
+RUN dpkg-reconfigure locales && \
+  locale-gen C.UTF-8 && \
+  /usr/sbin/update-locale LANG=C.UTF-8
+ENV LC_ALL C.UTF-8
+
+# Build unpaper 6.1
+RUN apt-get install -y \
+  wget \
+  gcc \
+  libavformat-dev \
+  libavcodec-dev \
+  libavutil-dev \
+  autoconf \
+  automake \
+  make \
+  pkg-config \
+  xsltproc
+
+WORKDIR /root
+RUN wget https://github.com/Flameeyes/unpaper/archive/unpaper-6.1.tar.gz
+RUN tar xf unpaper-6.1.tar.gz
+WORKDIR /root/unpaper-unpaper-6.1
+RUN autoreconf -i
+RUN ./configure CFLAGS="-O2 -march=native -pipe -flto"
+RUN make -j install
+
+RUN apt-get remove -y \
+  gcc \
+  autoconf \
+  automake \
+  pkg-config \
+  xsltproc \
+  make
+RUN apt-get autoremove -y && apt-get clean -y
+RUN rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
+
+# Set up a Python virtualenv and take all of the system packages, so we can
+# rely on the platform packages rather than importing GCC and compiling them
+RUN pyvenv /appenv \
+  && pyvenv --system-site-packages /appenv
+
+COPY . /application/
+
+# Install application and dependencies
+# In this arrangement Pillow and reportlab will be provided by the system
+RUN . /appenv/bin/activate; \
+  pip install --upgrade pip \
+  && pip install --no-cache-dir /application \
+  && pip install --no-cache-dir -r /application/test_requirements.txt
+
+USER docker
+WORKDIR /home/docker
+
+ENV DEFAULT_RUFFUS_HISTORY_FILE=/tmp/.{basename}.ruffus_history.sqlite
+ENV OCRMYPDF_TEST_OUTPUT=/tmp/test-output
+ENV OCRMYPDF_IN_DOCKER=1
+
+# Must use array form of ENTRYPOINT
+# Non-array form does not append other arguments, because that is "intuitive"
+ENTRYPOINT ["/application/docker-wrapper.sh"]

LICENSE.rst

@@ -0,0 +1,20 @@
+Copyright (c) 2013-2015, The OCRmyPDF Authors
+
+Permission is hereby granted, free of charge, to any person obtaining a
+copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice shall be included
+in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

MANIFEST.in

@@ -0,0 +1 @@
+recursive-exclude tests/output *

OCRmyPDF.sh

@@ -0,0 +1,6 @@
+#!/bin/sh
+##############################################################################
+# Copyright (c) 2013-14: fritz-hh from Github (https://github.com/fritz-hh)
+##############################################################################
+
+python3 -m ocrmypdf.main "$@"

README.md

@@ -1,40 +0,0 @@
-# OCRmyPDF-AppImage  [![Build Status](https://travis-ci.com/FPille/OCRmyPDF-AppImage.svg?branch=master)](https://travis-ci.com/FPille/OCRmyPDF-AppImage)
-[AppImage][APPIMAGE] for [OCRmyPDF][OCRMYPDF]
-
-## Usage
-Download OCRmyPDF*.AppImage, make it executable and run it.
-```
-wget https://github.com/FPille/OCRmyPDF-AppImage/releases/download/continuous/OCRmyPDF-8.3.2-x86_64.AppImage
-chmod +x OCRmyPDF*.AppImage
-./OCRmyPDF*.AppImage --help
-```  
-  
-  Beside OCRmyPDF additional command line programs can be run with this AppImage like:
-* ghostscript
-* img2pdf
-* pngquant
-* python3.6
-* qpdf
-* tesseract
-* unpaper  
-
-Just use the program name as first parameter plus options:
-```
-./OCRmyPDF*.AppImage tesseract -v
-tesseract 4.1.0
- leptonica-1.76.0
-  libjpeg 8d (libjpeg-turbo 1.3.0) : libpng 1.2.50 : libtiff 4.0.3 : zlib 1.2.11 : libwebp 0.4.0 : libopenjp2 2.3.0
- Found AVX2
- Found AVX
- Found SSE
-```
-Or create a symlink for the corresponding program:
-```
-ln -s OCRmyPDF*.AppImage tesseract
-./tesseract --list-langs
-```
-
-
-[APPIMAGE]: https://appimage.org
-[OCRMYPDF]: https://github.com/jbarlow83/OCRmyPDF
-

README.rst

@@ -0,0 +1,237 @@
+OCRmyPDF
+========
+
+OCRmyPDF adds an OCR text layer to scanned PDF files, allowing them to
+be searched.
+
+Main features
+-------------
+
+-  Generates a searchable
+   `PDF/A <https://en.wikipedia.org/?title=PDF/A>`__ file from a regular PDF
+   only containing images
+-  Places OCRed text accurately below the image to ease copy / paste
+-  Keeps the exact resolution of the original embedded images
+
+   -  or if requested oversamples the images before OCRing so as to get
+      better results
+
+-  When possible, copies input images directly to output without transcoding them,
+   to preserve image quality
+-  Keeps file size about the same
+-  If requested deskews and/or cleans the image before performing OCR
+-  Validates input and output files
+-  Provides debug mode to enable easy verification of the OCR results
+-  Processes several pages in parallel when more than one CPU core is
+   available
+-  Uses Tesseract OCR engine
+
+For details: please consult the `release notes <RELEASE_NOTES.rst>`__
+
+Motivation
+----------
+
+I searched the web for a free command line tool to OCR PDF files on
+Linux/UNIX: I found many, but none of them were really satisfying.
+
+-  Either they produced PDF files with misplaced text under the image (making copy/paste impossible) 
+-  Or they did not display correctly some escaped HTML characters located in the hOCR file produced by the OCR engine 
+-  Or they changed the resolution of the embedded images
+-  Or they generated PDF files having a ridiculous big size
+-  Or they crashed when trying to OCR some of my PDF files
+-  Or they did not produce valid PDF files (even though they were readable with my current PDF reader)
+-  On top of that none of them produced PDF/A files (format dedicated for long time storage)
+
+... so I decided to develop my own tool (using various existing scripts
+as an inspiration)
+
+Installation
+------------
+
+Download OCRmyPDF here: https://github.com/fritz-hh/OCRmyPDF/releases
+
+You can install it to a Python virtual environment or system-wide. 
+
+Installing the Docker container
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+For many users, installing the Docker container will be easier than installing all of OCRmyPDF's dependencies.
+
+If you have `Docker <https://docs.docker.com/>`__ installed on your system, you can install
+a Docker container of the latest release.
+
+Follow the Docker installation instructions for your platform.  If you can run this command
+successfully, your system is ready to download and execute the image::
+
+   docker run hello-world
+
+Assuming you have a Docker engine running somewhere, you can run these commands to download
+the image::
+
+   docker pull jbarlow83/ocrmypdf
+
+Then give it a more friendly, local name::
+
+   docker tag jbarlow83/ocrmypdf ocrmypdf
+  
+You can then run using the simple command::
+
+   docker run ocrmypdf --help
+  
+To execute the OCRmyPDF on a local file, you must `provide a writable volume to the Docker image <https://docs.docker.com/userguide/dockervolumes/>`__, such as this in this template::
+
+   docker run -v "$(pwd):/home/docker"   ocrmypdf <your arguments>
+
+In this worked example, the current working directory contains an input file called `test.pdf` and the output will go to `output.pdf`:: 
+
+   docker run -v "$(pwd):/home/docker"   ocrmypdf --skip-text test.pdf output.pdf
+
+Note that `ocrmypdf` has its own separate -v argument to control debug verbosity.
+
+Installing on Mac OS X Yosemite
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+If it's not already present, `install Homebrew <http://brew.sh/>`__
+
+Update Homebrew::
+
+   brew update
+   
+Install or upgrade the required Homebrew packages, if any are missing::
+
+   brew install libpng openjpeg jbig2dec     # image libraries
+   brew install qpdf
+   brew install ghostscript
+   brew install python3
+   brew install libxml2
+   brew install leptonica
+   brew install tesseract
+   
+It is also recommended that install Pillow and confirm it can read and write JPEG and PNG files::
+
+   pip3 install --upgrade pip
+   pip3 install --upgrade pillow
+
+To test that your Python imaging library (Pillow) can access JPEG and PNG files, try this command::
+
+   python3 -c "from PIL import Image; im = Image.new('1', (1, 1)); im.save('test.png'); im.save('test.jpg')"
+
+If you have trouble getting Pillow to access JPEG and PNG files, `review the installation instructions <https://pillow.readthedocs.org/installation.html>`__.
+
+You can then install OCRmyPDF from PyPI::
+
+   pip3 install ocrmypdf
+
+The command line program should now be available::
+
+   ocrmypdf --help
+
+Installing on Ubuntu 14.04 LTS
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Installing on Ubuntu 14.04 LTS (trusty) is more difficult than other options.
+
+Update apt-get::
+
+   sudo apt-get update
+   sudo apt-get upgrade
+   
+Install system dependencies::
+
+   sudo apt-get install \
+      zlib1g-dev \
+      libjpeg-dev \
+      ghostscript \
+      tesseract-ocr \
+      qpdf \
+      unpaper \
+      python3-pip \
+      python3-pil \
+      python3-pytest \
+      python3-reportlab
+
+If you wish install OCRmyPDF to the system Python, then install as follows (note this installs new packages
+into your system Python, which could interfere with other programs)::
+
+   sudo pip3 install ocrmypdf
+   
+If you wish to install OCRmyPDF to a virtual environment to isolate system Python from modified, you can
+follow these steps.  This includes a workaround `for a known, unresolved issue in Ubuntu 14.04's ensurepip
+package <http://www.thefourtheye.in/2014/12/Python-venv-problem-with-ensurepip-in-Ubuntu.html>`__::
+
+   sudo apt-get install python3-venv
+   python3 -m venv venv-ocrmypdf --without-pip
+   source venv-ocrmypdf/bin/activate
+   wget -O - -o /dev/null https://bootstrap.pypa.io/get-pip.py | python
+   deactivate
+   pyvenv --system-site-packages venv-ocrmypdf
+   source venv-ocrmypdf/bin/activate
+   pip install ocrmypdf
+
+Ubuntu 14.04 only installs `unpaper` version 0.4.2, which is not supported by OCRmyPDF because it is produces invalid output. This program is an optional dependency, and provides page deskewing and cleaning. See `Dockerfile <Dockerfile>`__ for an example of how to building unpaper 6.1 from source. If you choose to install unpaper later, OCRmyPDF will use the foremost version on the system PATH.
+
+      
+Installing HEAD revision from sources
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+To install the HEAD revision from sources in development mode::
+
+   git clone -b master https://github.com/fritz-hh/OCRmyPDF.git
+   cd OCRmyPDF
+   pip3 install -e .
+   
+On certain Linux/UNIX platforms such as Ubuntu, you may need to use 
+run the install command as superuser::
+
+   sudo pip3 install -e .
+   
+Note that this will alter your system's Python distribution. If you prefer 
+to not install as superuser, you can install the package in a Python virtual environment::
+
+   git clone -b master https://github.com/fritz-hh/OCRmyPDF.git
+   pyvenv venv
+   source venv/bin/activate
+   cd OCRmyPDF
+   pip3 install -e .
+
+If your platform does not have ``pip3``, make sure that Python 3.4+ and the `pip` 
+package are installed.
+
+To run the program::
+   
+   ocrmypdf --help
+
+If not yet installed, the script will notify you about dependencies that
+need to be installed. The script requires specific versions of the
+dependencies. Older version than the ones mentioned in the release notes
+are likely not to be compatible to OCRmyPDF.
+
+Support
+-------
+
+In case you detect an issue, please:
+
+-  Check if your issue is already known
+-  If no problem report exists on github, please create one here:
+   https://github.com/fritz-hh/OCRmyPDF/issues
+-  Describe your problem thoroughly
+-  Append the console output of the script when running the debug mode
+   (-v 1 option)
+-  If possible provide your input PDF file as well as the content of the
+   temporary folder (using a file sharing service like
+   www.file-upload.net)
+
+Press & Media
+-------------
+
+-  `c't 1-2014, page 59 <http://www.heise.de/ct/inhalt/2014/1/58/>`__:
+   Detailed presentation of OCRmyPDF v1.0 in the leading German IT
+   magazine c't
+-  `heise Open Source, 09/2014: Texterkennung mit
+   OCRmyPDF <http://www.heise.de/-2356670>`__
+
+Disclaimer
+----------
+
+The software is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
+CONDITIONS OF ANY KIND, either express or implied.

RELEASE_NOTES.rst

@@ -0,0 +1,494 @@
+RELEASE NOTES
+=============
+
+Please always read this file before installing the package
+
+Download software here: https://github.com/fritz-hh/OCRmyPDF/tags
+
+v3.0:
+=====
+
+New features
+------------
+
+-  Easier installation with a Docker container or Python's ``pip`` package manager 
+-  Eliminated many external dependencies, so it's easier to setup
+-  Now installs ``ocrmypdf`` to ``/usr/local/bin`` or equivalent for system-wide
+   access and easier typing
+-  Improved command line syntax and usage help (``--help``)
+-  Tesseract 3.03+ PDF page rendering can be used instead for better positioning
+   of recognized text (``--pdf-renderer tesseract``)
+-  PDF metadata (title, author, keywords) are now transferred to the 
+   output PDF
+-  PDF metadata can also be set from the command line (``--title``, etc.)
+-  Added test cases to confirm everything is working
+-  Added option to skip extremely large pages that take too long to OCR and are 
+   often not OCRable (e.g. large scanned maps or diagrams); other pages are still
+   processed (``--skip-big``)
+-  Added option to kill Tesseract OCR process if it seems to be taking too long on
+   a page, while still processing other pages (``--tesseract-timeout``)
+
+Changes
+-------
+
+-  New, robust rewrite in Python 3.4+ with ruffus_ pipelines
+-  Now uses Ghostscript 9.14's improved color conversion model to preserve PDF colors
+-  All "tasks" in the pipeline can be executed in parallel on any
+   available CPUs, increasing performance
+-  The ``-o DPI`` argument has been phased out, in favor of ``--oversample DPI``, in
+   case we need ``-o OUTPUTFILE`` in the future
+-  Removed several dependencies, so it's easier to install.  We no 
+   longer use:
+   
+   - GNU parallel_
+   - ImageMagick_
+   - Python 2.7
+   - Poppler
+   - MuPDF_ tools
+   - shell scripts
+   - Java and JHOVE_
+   - libxml2
+
+-  Some new external dependencies are required or optional, compared to v2.x:
+
+   - Ghostscript 9.14+
+   - qpdf_ 5.0.0+
+   - Unpaper_ 6.1 (optional)
+   - some automatically managed Python packages
+  
+.. _ruffus: http://www.ruffus.org.uk/index.html
+.. _parallel: https://www.gnu.org/software/parallel/
+.. _ImageMagick: http://www.imagemagick.org/script/index.php
+.. _MuPDF: http://mupdf.com/docs/
+.. _qpdf: http://qpdf.sourceforge.net/
+.. _Unpaper: https://github.com/Flameeyes/unpaper
+.. _JHOVE: http://jhove.sourceforge.net/
+
+Release candidates
+------------------
+
+-  rc6:
+
+   - dropped libxml2 (Python lxml) since Python 3's internal XML parser is sufficient
+   - set up Docker container
+   - fix Unicode errors if recognized text contains Unicode characters and system locale is not UTF-8
+
+-  rc5:
+
+   - dropped Java and JHOVE in favour of qpdf
+   - improved command line error output
+   - additional tests and bug fixes
+   - tested on Ubuntu 14.04 LTS
+
+-  rc4:
+
+   - dropped MuPDF in favour of qpdf
+   - fixed some installer issues and errors in installation instructions
+   - improve performance: run Ghostscript with multithreaded rendering
+   - improve performance: use multiple cores by default
+   - bug fix: checking for wrong exception on process timeout 
+
+-  rc3: skipping version number intentionally to avoid confusion with Tesseract
+-  rc2: first release for public testing to test-PyPI, Github
+-  rc1: testing release process
+
+Compatibility notes
+-------------------
+
+-  ``./OCRmyPDF.sh`` script is still available for now
+-  Stacking the verbosity option like ``-vvv`` is no longer supported
+
+-  The configuration file ``config.sh`` has been removed.  Instead, you can
+   feed a file to the arguments for common settings:
+
+::
+
+   ocrmypdf input.pdf output.pdf @settings.txt
+
+where ``settings.txt`` contains *one argument per line*, for example:
+
+::
+
+   -l 
+   deu 
+   --author 
+   A. Merkel 
+   --pdf-renderer 
+   tesseract
+
+
+Fixes
+-----
+
+-  Handling of filenames containing spaces: fixed
+
+Notes
+-----
+
+-  Some dependencies may work with lower versions than tested, so try
+   overriding dependencies if they are "in the way" to see if they work.
+
+
+v2.2-stable (2014-09-29):
+=========================
+
+New features
+------------
+
+- None
+
+Changes
+-------
+
+- Update to jhove v1.11
+- Request the python library reportlab v3.0 or newer (So that we could remove a patch to the previous version of reportlab leading to issues for some users)
+
+Fixes
+-----
+
+- Fix bug on Mac OS X (resolution of simlink to OCRmyPDF.sh script) (thanks to jbarlow83)
+- Check if the input pdf file exists before to continue
+
+Tested with
+-----------
+
+- Operating system: FreeBSD 9.2
+- Dependencies:
+
+   - parallel 20140822
+   - poppler-utils 0.24.5
+   - ImageMagick 6.8.9-4 2014-09-17
+   - Unpaper 0.3
+   - tesseract 3.02.02
+   - Python 2.7.8
+   - ghostcript (gs): 9.06
+   - java: openjdk version "1.7.0_65"
+
+
+v2.1-stable (2014-09-20):
+=========================
+
+New features
+------------
+
+-  None
+
+Changes
+-------
+
+-  None
+
+Fixes
+-----
+
+-  Allow execution via simlink
+-  Add support for tesseract 3.03
+-  Add support for newer version of reportlab
+-  Lowered minimum version of gnu parallel
+-  Various typo
+
+Tested with
+-----------
+
+-  Operating system: FreeBSD 9.1
+-  Dependencies:
+-  parallel 20130222
+-  poppler-utils 0.22.2
+-  ImageMagick 6.8.0-7 2013-03-30
+-  Unpaper 0.3
+-  tesseract 3.02.02
+-  Python 2.7.3
+-  ghoscript (gs): 9.06
+-  java: openjdk version "1.7.0\_17"
+
+v2.0-stable (2014-01-25):
+=========================
+
+New features
+------------
+
+-  Check if the language(s) passed using the -l option is supported by
+   tesseract (fixes #60)
+
+Changes
+-------
+
+-  Allow OCRmyPDF to be used with tesseract 3.02.01, even though OCR
+   might fail for few PDF file (see issue #28). Rationale: For some
+   linux distribution, no newer version than tesseract 3.02.01 is
+   available
+
+Fixes
+-----
+
+-  More robust algorithm for checking the version of the installed
+   tesseract package
+
+Tested with
+-----------
+
+-  Operating system: FreeBSD 9.1
+-  Dependencies:
+-  parallel 20130222
+-  poppler-utils 0.22.2
+-  ImageMagick 6.8.0-7 2013-03-30
+-  Unpaper 0.3
+-  tesseract 3.02.02
+-  Python 2.7.3
+-  ghoscript (gs): 9.06
+-  java: openjdk version "1.7.0\_17"
+
+v2.0-rc2 (2014-01-16):
+======================
+
+New features
+------------
+
+-  None
+
+Changes
+-------
+
+-  Size reduction of final PDF file: (fixes #50)
+-  Support for monochrome (Black&White) images (massive size reduction
+   in final PDF: >80%)
+-  Reduced size of grayscale images (by 13% on test PDF file)
+-  Preventing fi, fl ligatures does not require anymore to pass an
+   additional config file to tesseract using the -C option (fixes #58)
+-  Location of temporary folder according to content of environment
+   variable TMPDIR.
+-  Dependency to pdftk removed
+-  Check for compatible versions of dependencies: (fixes #51)
+-  parallel and tesseract
+-  python libraries reportlab and lxml
+
+Fixes
+-----
+
+-  Improved portability with various shells (dash, bash, tcsh) and OS
+   (FreeBSD, MAC OSX, Linux) (fixes #59)
+-  Corrected bug in case the input PDF file contains a space character
+   (fixes #48)
+-  Prevent spurious error message in case there is no image in a PDF
+   page
+-  Prevent collision of temporary folder names (fixes #57)
+
+Tested with
+-----------
+
+-  Operating system: FreeBSD 9.1
+-  Dependencies:
+-  parallel 20130222
+-  poppler-utils 0.22.2
+-  ImageMagick 6.8.0-7 2013-03-30
+-  Unpaper 0.3
+-  tesseract 3.02.02
+-  Python 2.7.3
+-  ghoscript (gs): 9.06
+-  java: openjdk version "1.7.0\_17"
+
+v2.0-rc1 (2014-01-07):
+======================
+
+New features
+------------
+
+-  Huge performance improvement on machines having multiple CPU/cores
+   (processing of several pages concurrently) (fixes #18)
+-  By default prevent from processing a PDF file already containing
+   fonts (i.e. text)(it can be overridden with the -f flag) (fixes #16)
+-  Warn if the resolution is too low to get reasonable OCR results
+   (fixes #37)
+-  New option (-o) to perform automatic oversampling if the image
+   resolution is too low. This can improve OCR results.
+-  Warn if using a tesseract version older than v3.02.02 (as older
+   versions are known to produce invalid output) (fixes #41)
+-  Echo version of the installed dependencies (e.g. tesseract) in debug
+   mode in order to ease support (fixes #35)
+-  Echo the arguments passed to the script in debug mode to ease support
+
+Changes
+-------
+
+-  In debug mode: The debug page is now placed after the respective
+   "normal" page
+-  Reduced disk space usage in temporary folder if -d (deskew) or -c
+   (cleanup) options are not selected
+-  New file src/config.sh containing various configuration parameters
+-  Documentation of the tesseract config file "tess-cfg/no\_ligature"
+   improved
+-  Improved consistency of the temporary file names
+
+Fixes
+-----
+
+-  Improved robustness:
+-  in case vertical resolution differs from horizontal resolution (fixes
+   #38)
+-  in case a PDF page contains more than one image (fixes #36)
+-  Fix a problem occurring if python 3 is the standard interpreter
+   (fixes #33)
+-  Fix a problem occurring if the input PDF file contains special
+   characters like "#" (fixes #34)
+
+Tested with
+-----------
+
+-  Operating system: FreeBSD 9.1
+-  Dependencies:
+-  parallel 20130222
+-  poppler-utils 0.22.2
+-  ImageMagick 6.8.0-7 2013-03-30
+-  Unpaper 0.3
+-  tesseract 3.02.02
+-  Python 2.7.3
+-  pdftk 1.45
+-  ghoscript (gs): 9.06
+-  java: openjdk version "1.7.0\_17"
+
+v1.1-stable (2014-01-06):
+=========================
+
+New features
+------------
+
+-  N/A
+
+Changes
+-------
+
+-  N/A
+
+Fixes
+-----
+
+-  Fixed syntax error (bashism) leading to an error message on certain
+   systems (fixes #42)
+
+Tested with
+-----------
+
+-  Operating system: FreeBSD 9.1
+-  Dependencies:
+-  poppler-utils 0.22.2
+-  ImageMagick 6.8.0-7 2013-03-30
+-  Unpaper 0.3
+-  tesseract 3.02.02
+-  Python 2.7.3
+-  pdftk 1.45
+-  ghoscript (gs): 9.06
+-  java: openjdk version "1.7.0\_17"
+
+v1.0-stable (2013-05-06):
+=========================
+
+New features
+------------
+
+-  In debug mode: compute and echo time required for processing (fixes
+   #26)
+
+Changes
+-------
+
+-  Removed feature to add metadata in final pdf file (because it lead to
+   to final PDF file that does not comply to the PDF/A-1 format)
+-  Removed feature to set same owner & permissions in final PDF file
+   than in input file
+-  Removed many unused jhove files (e.g. documentation, \*.java and
+   \*.class files)
+
+Fixes
+-----
+
+-  Correction to handle correctly path and input PDF files having spaces
+   (fixes #31)
+-  Resolutions (x/y) that are nearly equal are now supported (fixes #25)
+-  Fix compatibility issue with Ubuntu server 12.04 / Ubuntu server
+   10.04 / Linux Mint 13 Maya and probably other Linux distributions
+   (fixes #27)
+-  Commit missing jhove files (\*.jar mainly) due to wrong .gitignore
+
+Tested with
+-----------
+
+-  Operating system: FreeBSD 9.1
+-  Dependencies:
+-  poppler-utils 0.22.2
+-  ImageMagick 6.8.0-7 2013-03-30
+-  Unpaper 0.3
+-  tesseract 3.02.02
+-  Python 2.7.3
+-  pdftk 1.45
+-  ghoscript (gs): 9.06
+-  java: openjdk version "1.7.0\_17"
+
+v1.0-rc2 (2013-04-29):
+======================
+
+New features
+------------
+
+-  Keep temporary files if debug mode is set (fixes #22)
+-  Set same owner & permissions in final PDF file than in input file
+   (fixes #9)
+-  Added metadata in final pdf file (fixes #4)
+
+Changes
+-------
+
+-  N/A
+
+Fixes
+-----
+
+-  Fixed wrong image cropping when deskew option is activated
+-  Exit with error message if page size is not found in hocr file (fixes
+   #21)
+-  Various minor fixes in log messages
+
+Tested with
+-----------
+
+-  Operating system: FreeBSD 9.1
+-  Dependencies:
+-  poppler-utils 0.22.2
+-  ImageMagick 6.8.0-7 2013-03-30
+-  Unpaper 0.3
+-  tesseract 3.02.02
+-  Python 2.7.3
+-  pdftk 1.45
+-  ghoscript (gs): 9.06
+-  java: openjdk version "1.7.0\_17"
+
+v1.0-rc1 (2013-04-26):
+======================
+
+New features
+------------
+
+-  First release candidate
+
+Changes
+-------
+
+-  N/A
+
+Fixes
+-----
+
+-  N/A
+
+Tested with
+-----------
+
+-  Operating system: FreeBSD 9.1
+-  Dependencies:
+-  poppler-utils 0.22.2
+-  ImageMagick 6.8.0-7 2013-03-30
+-  Unpaper 0.3
+-  tesseract 3.02.02
+-  Python 2.7.3
+-  pdftk 1.45
+-  ghoscript (gs): 9.06
+-  java: openjdk version "1.7.0\_17"

appimage/AppRun.sh

@@ -1,105 +0,0 @@
-#! /bin/bash
-
-HERE="$(dirname "$(readlink -f "${0}")")"
-
-export PATH="$HERE/usr/bin:$HERE/usr/local/bin:$HERE/usr/python/bin:$PATH"
-export LD_PRELOAD="$HERE/usr/lib/liblept.so.5"
-export LD_LIBRARY_PATH="$HERE/usr/lib:$HERE/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH"
-export TESSDATA_PREFIX="$HERE/usr/share/tesseract-ocr/4.00/tessdata"
-export GS_LIB="$HERE/usr/share/ghostscript/9.26/lib:$HERE/usr/share/ghostscript/9.26/Resource:$HERE/usr/share/ghostscript/9.26/Resource/Init"
-
-# Allow the AppImage to be symlinked to e.g., /usr/bin/commandname
-# or called with ./Some*.AppImage commandname ...
-# refer to https://github.com/AppImage/AppImageKit/wiki/Bundling-command-line-tools
-
-if [ ! -z "$APPIMAGE" ] ; then
-  BINARY_NAME=$(basename "$ARGV0")
-else
-  BINARY_NAME=$(basename "$0")
-  export APPDIR="$HERE"       # required for the wrapper scripts of linuxdeploy-plugin-python
-fi
-
-usage() {
-echo "
-==============================================================================
-                          AppImage for OCRmyPDF
-==============================================================================
-OCRmyPDF adds an OCR text layer to scanned PDF files, allowing them to be
-searched or copy-pasted.
-
-usage:
-$ARGV0 [ocrmypdf] [--help] [--list-programs]
-                                 [--list-licenses] [--show-license]
-
-ocrmypdf                    execute OCRmyPDF
-
---help                      show this help message
-
---list-programs             list all programs contained in this AppImage
-
---list-licenses             list all licenses contained in this AppImage
-
---show-license [LICENSE]    show content of license file
-"
-}
-
-if [ "$1" == "--help" ] ; then
-    usage
-    exit $?
-fi
-
-if [ "$1" == "--list-programs" ] ; then
-    pushd "$HERE"
-    echo ""
-    echo "Run \"$ARGV0\" with one of the following arguments to run the respective program."
-    echo ""
-    find . -type f -perm /111 ! -path '*/lib/*' -execdir basename {} ";" | sort -u | column
-    echo ""
-    exit $?
-fi
-
-if [ "$1" == "--list-licenses" ] ; then
-    pushd "$HERE"
-    echo ""
-    echo "Run \"$ARGV0\" with one of the following arguments to display the respective license file."
-    echo ""
-    find . -type f \( ! -path '*/tesseract-ocr-*' -o -path '*/tesseract-ocr-eng/*' \) \
-        \( -iname "license*" -o -iname "*copyright*" -o -iname "*copying*" \) -printf "--show-license %P\n" | sort | column
-    echo ""
-    exit $?
-fi
-
-if [ "$1" == "--show-license" ] ; then
-    pushd "$HERE"
-    shift
-    if [ -f "$1" ] ; then
-        less -N "$1"
-        exit $?
-    else
-        echo "\"$1\" is not a valid license file path."
-        exit 1
-    fi
-fi
-
-if [ ! -z "$1" ] && [ -e "$HERE/bin/$1" ] ; then
-    MAIN="$HERE/bin/$1" ; shift
-elif [ ! -z "$1" ] && [ -e "$HERE/usr/bin/$1" ] ; then
-    MAIN="$HERE/usr/bin/$1" ; shift
-elif [ ! -z "$1" ] && [ -e "$HERE/usr/python/bin/$1" ] ; then
-    MAIN="$HERE/usr/python/bin/$1" ; shift
-elif [ ! -z "$1" ] && [ -e "$HERE/usr/local/bin/$1" ] ; then
-    MAIN="$HERE/usr/local/bin/$1" ; shift
-elif [ -e "$HERE/bin/$BINARY_NAME" ] ; then
-    MAIN="$HERE/bin/$BINARY_NAME"
-elif [ -e "$HERE/usr/bin/$BINARY_NAME" ] ; then
-    MAIN="$HERE/usr/bin/$BINARY_NAME"
-elif [ -e "$HERE/usr/python/bin/$BINARY_NAME" ] ; then
-    MAIN="$HERE/usr/python/bin/$BINARY_NAME"
-elif [ -e "$HERE/usr/local/bin/$BINARY_NAME" ] ; then
-    MAIN="$HERE/usr/local/bin/$BINARY_NAME"
-else
-    usage
-    exit $?
-fi
-
-exec "${MAIN}" "$@"

build-appimage.sh

@@ -1,120 +0,0 @@
-#! /bin/bash
-
-set -x
-set -e
-
-# use RAM disk if possible
-if [ "$CI" == "" ] && [ -d /dev/shm ]; then
-    TEMP_BASE=/dev/shm
-else
-    TEMP_BASE=/tmp
-fi
-
-BUILD_DIR=$(mktemp -d -p "$TEMP_BASE" OCRmyPDF-AppImage-build-XXXXXX)
-
-cleanup () {
-    if [ -d "$BUILD_DIR" ]; then
-        rm -rf "$BUILD_DIR"
-    fi
-}
-
-trap cleanup EXIT
-
-# store repo root as variable
-REPO_ROOT=$(readlink -f "$(dirname "$(dirname "$0")")")
-OLD_CWD=$(readlink -f .)
-
-pushd "$BUILD_DIR"
-
-mkdir -p AppDir
-mkdir -p PackageDir
-mkdir -p jbig2
-
-# download linuxdeploy AppImage and linuxdeploy-plugin-python AppImage
-wget https://github.com/TheAssassin/linuxdeploy/releases/download/continuous/linuxdeploy-x86_64.AppImage
-# wget https://github.com/niess/linuxdeploy-plugin-python/releases/download/continuous/linuxdeploy-plugin-python-x86_64.AppImage
-
-# use adapted linuxdeploy-plugin-python instead of the original one (otherwise OCRmyPDF breaks)
-wget https://github.com/FPille/linuxdeploy-plugin-python/releases/download/continuous/linuxdeploy-plugin-python-x86_64.AppImage
-
-chmod +x linuxdeploy*.AppImage
-
-
-ARCH=$(uname -i)
-export ARCH
-
-
-# .desktop file
-cat > ocrmypdf.desktop <<\EOF
-[Desktop Entry]
-Name=ocrmypdf
-Type=Application
-Exec=ocrmypdf
-Icon=ocrmypdf
-Terminal=true
-Comment=OCRmyPDF adds an OCR text layer to scanned PDF files, allowing them to be searched
-Categories=Graphics;Scanning;OCR;
-EOF
-
-
-# download logo and convert it to desktop icon
-# requires Imagemagick (convert)
-wget https://raw.githubusercontent.com/jbarlow83/OCRmyPDF/master/docs/images/logo-social.png
-convert logo-social.png -resize 512x512\> -size 512x512 xc:white +swap -gravity center -composite ocrmypdf.png
-
-
-# download and intsall packages required by OCRmyPDF
-pushd PackageDir
-packages=(tesseract-ocr tesseract-ocr-all libavformat56 ghostscript qpdf pngquant)
-
-for i in "${packages[@]}"
-do
-    apt-get -d -o dir::cache="$PWD" -o Debug::NoLocking=1 --reinstall install "$i" -y
-done
-
-wget -q 'https://www.dropbox.com/s/vaq0kbwi6e6au80/unpaper_6.1-1.deb?raw=1' -O unpaper_6.1-1.deb
-
-find . -type f -name \*.deb -exec dpkg-deb -X {} "$BUILD_DIR"/AppDir \;
-popd
-
-
-# compile and install jbig2
-# requires libleptonica-dev, zlib1g-dev
-wget -q https://github.com/agl/jbig2enc/archive/0.29.tar.gz -O - | \
-     tar xz -C jbig2 --strip-components=1
-pushd jbig2
-./autogen.sh
-./configure --prefix="$BUILD_DIR"/AppDir/usr
-make && make install
-popd
-
-pushd "$BUILD_DIR"/AppDir
-# add some tools to AppDir
-#cp -f   /usr/bin/column     ./usr/bin/
-#cp -f   /bin/less           ./usr/bin/
-
-# remove unnecessary data from AppDir
-[ -d bin ] && rm -rf ./bin
-[ -d etc ] && rm -rf ./etc
-[ -d var ] && rm -rf ./var
-popd
-
-
-# export LD_LIBRARY_PATH so that dependencies of shared libraries can be deployed by linuxdeploy-x86_64.AppImage
-export LD_LIBRARY_PATH="$BUILD_DIR/AppDir/usr/lib:$BUILD_DIR/AppDir/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH"
-
-#OCRMYPDF_VERSION=8.3.2   # exported in .travis.yml file
-export PIP_REQUIREMENTS="ocrmypdf==$OCRMYPDF_VERSION"
-export VERSION="$OCRMYPDF_VERSION"
-export OUTPUT=OCRmyPDF-"$VERSION"-"$ARCH".AppImage
-export PYTHON_SOURCE=https://www.python.org/ftp/python/3.6.8/Python-3.6.8.tgz
-
-./linuxdeploy-x86_64.AppImage --appdir AppDir --plugin python \
-    -d ocrmypdf.desktop -i ocrmypdf.png \
-    --custom-apprun "$REPO_ROOT"/appimage/AppRun.sh --output appimage
-
-
-# move AppImage back to old CWD
-mv "$OUTPUT" "$OLD_CWD"/
-
-popd

docker-wrapper.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+. /appenv/bin/activate
+cd /home/docker
+exec ocrmypdf "$@"

ocrmypdf/__init__.py

@@ -0,0 +1,12 @@
+from enum import IntEnum
+
+
+class ExitCode(IntEnum):
+    ok = 0
+    bad_args = 1
+    input_file = 2
+    missing_dependency = 3
+    invalid_output_pdfa = 4
+    file_access_error = 5
+    already_done_ocr = 6
+    other_error = 15

ocrmypdf/ghostscript.py

@@ -0,0 +1,52 @@
+#!/usr/bin/env python3
+# © 2015 James R. Barlow: github.com/jbarlow83
+
+from tempfile import NamedTemporaryFile
+from subprocess import Popen, PIPE, check_call
+from shutil import copy
+
+
+def rasterize_pdf(input_file, output_file, xres, yres, raster_device, log):
+    with NamedTemporaryFile(delete=True) as tmp:
+        args_gs = [
+            'gs',
+            '-dBATCH', '-dNOPAUSE',
+            '-sDEVICE=%s' % raster_device,
+            '-o', tmp.name,
+            '-r{0}x{1}'.format(str(xres), str(yres)),
+            input_file
+        ]
+
+        p = Popen(args_gs, close_fds=True, stdout=PIPE, stderr=PIPE,
+                  universal_newlines=True)
+        stdout, stderr = p.communicate()
+        if stdout:
+            log.debug(stdout)
+        if stderr:
+            log.error(stderr)
+
+        if p.returncode == 0:
+            copy(tmp.name, output_file)
+        else:
+            log.error('Ghostscript rendering failed')
+
+
+def generate_pdfa(pdf_pages, output_file, threads=1):
+    with NamedTemporaryFile(delete=True) as gs_pdf:
+        args_gs = [
+            "gs",
+            "-dQUIET",
+            "-dBATCH",
+            "-dNOPAUSE",
+            '-dNumRenderingThreads=' + str(threads),
+            "-sDEVICE=pdfwrite",
+            "-sColorConversionStrategy=/RGB",
+            "-sProcessColorModel=DeviceRGB",
+            "-dPDFA",
+            "-sPDFACompatibilityPolicy=2",
+            "-sOutputICCProfile=srgb.icc",
+            "-sOutputFile=" + gs_pdf.name,
+        ]
+        args_gs.extend(pdf_pages)
+        check_call(args_gs)
+        copy(gs_pdf.name, output_file)

ocrmypdf/hocrtransform.py

@@ -0,0 +1,230 @@
+#!/usr/local/bin/python3
+##############################################################################
+# Copyright (c) 2013-14: fritz-hh from Github
+#   (https://github.com/fritz-hh)
+#
+# Copyright (c) 2010: Jonathan Brinley from Github
+#   (https://github.com/jbrinley/HocrConverter)
+#   Initial version by Jonathan Brinley, jonathanbrinley@gmail.com
+##############################################################################
+from reportlab.pdfgen.canvas import Canvas
+from reportlab.lib.units import inch
+from xml.etree import ElementTree
+from PIL import Image
+from collections import namedtuple
+import re
+import argparse
+
+
+Rect = namedtuple('Rect', ['x1', 'y1', 'x2', 'y2'])
+
+
+class HocrTransformError(Exception):
+    pass
+
+
+class HocrTransform():
+
+    """
+    A class for converting documents from the hOCR format.
+    For details of the hOCR format, see:
+    http://docs.google.com/View?docid=dfxcv4vc_67g844kf
+    """
+
+    def __init__(self, hocrFileName, dpi):
+        self.dpi = dpi
+        self.boxPattern = re.compile(r'bbox((\s+\d+){4})')
+
+        self.hocr = ElementTree.parse(hocrFileName)
+
+        # if the hOCR file has a namespace, ElementTree requires its use to
+        # find elements
+        matches = re.match(r'({.*})html', self.hocr.getroot().tag)
+        self.xmlns = ''
+        if matches:
+            self.xmlns = matches.group(1)
+
+        # get dimension in pt (not pixel!!!!) of the OCRed image
+        self.width, self.height = None, None
+        for div in self.hocr.findall(
+                ".//%sdiv[@class='ocr_page']" % (self.xmlns)):
+            coords = self.element_coordinates(div)
+            pt_coords = self.pt_from_pixel(coords)
+            self.width = pt_coords.x2 - pt_coords.x1
+            self.height = pt_coords.y2 - pt_coords.y1
+            # there shouldn't be more than one, and if there is, we don't want
+            # it
+            break
+        if self.width is None or self.height is None:
+            raise HocrTransformError("hocr file is missing page dimensions")
+
+    def __str__(self):
+        """
+        Return the textual content of the HTML body
+        """
+        if self.hocr is None:
+            return ''
+        body = self.hocr.find(".//%sbody" % (self.xmlns))
+        if body:
+            return self._get_element_text(body)
+        else:
+            return ''
+
+    def _get_element_text(self, element):
+        """
+        Return the textual content of the element and its children
+        """
+        text = ''
+        if element.text is not None:
+            text += element.text
+        for child in element.getchildren():
+            text += self._get_element_text(child)
+        if element.tail is not None:
+            text += element.tail
+        return text
+
+    def element_coordinates(self, element):
+        """
+        Returns a tuple containing the coordinates of the bounding box around
+        an element
+        """
+        out = (0, 0, 0, 0)
+        if 'title' in element.attrib:
+            matches = self.boxPattern.search(element.attrib['title'])
+            if matches:
+                coords = matches.group(1).split()
+                out = Rect._make(int(coords[n]) for n in range(4))
+        return out
+
+    def pt_from_pixel(self, pxl):
+        """
+        Returns the quantity in PDF units (pt) given quantity in pixels
+        """
+        return Rect._make(
+            (c / self.dpi * inch) for c in pxl)
+
+    def replace_unsupported_chars(self, s):
+        """
+        Given an input string, returns the corresponding string that:
+        - is available in the helvetica facetype
+        - does not contain any ligature (to allow easy search in the PDF file)
+        """
+        # The 'u' before the character to replace indicates that it is a
+        # unicode character
+        s = s.replace(u"fl", "fl")
+        s = s.replace(u"fi", "fi")
+        return s
+
+    def to_pdf(self, outFileName, imageFileName=None, showBoundingboxes=False,
+               fontname="Helvetica", invisibleText=False):
+        """
+        Creates a PDF file with an image superimposed on top of the text.
+        Text is positioned according to the bounding box of the lines in
+        the hOCR file.
+        The image need not be identical to the image used to create the hOCR
+        file.
+        It can have a lower resolution, different color mode, etc.
+        """
+        # create the PDF file
+        # page size in points (1/72 in.)
+        pdf = Canvas(
+            outFileName, pagesize=(self.width, self.height), pageCompression=1)
+
+        # draw bounding box for each paragraph
+        # light blue for bounding box of paragraph
+        pdf.setStrokeColorRGB(0, 1, 1)
+        # light blue for bounding box of paragraph
+        pdf.setFillColorRGB(0, 1, 1)
+        pdf.setLineWidth(0)		# no line for bounding box
+        for elem in self.hocr.findall(
+                ".//%sp[@class='%s']" % (self.xmlns, "ocr_par")):
+
+            elemtxt = self._get_element_text(elem).rstrip()
+            if len(elemtxt) == 0:
+                continue
+
+            pxl_coords = self.element_coordinates(elem)
+            pt = self.pt_from_pixel(pxl_coords)
+
+            # draw the bbox border
+            if showBoundingboxes:
+                pdf.rect(
+                    pt.x1, self.height - pt.y2, pt.x2 - pt.x1, pt.y2 - pt.y1,
+                    fill=1)
+
+        # check if element with class 'ocrx_word' are available
+        # otherwise use 'ocr_line' as fallback
+        elemclass = "ocr_line"
+        if self.hocr.find(
+                ".//%sspan[@class='ocrx_word']" % (self.xmlns)) is not None:
+            elemclass = "ocrx_word"
+
+        # itterate all text elements
+        # light green for bounding box of word/line
+        pdf.setStrokeColorRGB(1, 0, 0)
+        pdf.setLineWidth(0.5)		# bounding box line width
+        pdf.setDash(6, 3)		# bounding box is dashed
+        pdf.setFillColorRGB(0, 0, 0)  # text in black
+        for elem in self.hocr.findall(
+                ".//%sspan[@class='%s']" % (self.xmlns, elemclass)):
+
+            elemtxt = self._get_element_text(elem).rstrip()
+
+            elemtxt = self.replace_unsupported_chars(elemtxt)
+
+            if len(elemtxt) == 0:
+                continue
+
+            pxl_coords = self.element_coordinates(elem)
+            pt = self.pt_from_pixel(pxl_coords)
+
+            # draw the bbox border
+            if showBoundingboxes:
+                pdf.rect(
+                    pt.x1, self.height - pt.y2, pt.x2 - pt.x1, pt.y2 - pt.y1,
+                    fill=0)
+
+            text = pdf.beginText()
+            fontsize = pt.y2 - pt.y1
+            text.setFont(fontname, fontsize)
+            if invisibleText:
+                text.setTextRenderMode(3)  # Invisible (indicates OCR text)
+
+            # set cursor to bottom left corner of bbox (adjust for dpi)
+            text.setTextOrigin(pt.x1, self.height - pt.y2)
+
+            # scale the width of the text to fill the width of the bbox
+            text.setHorizScale(
+                100 * (pt.x2 - pt.x1) / pdf.stringWidth(
+                    elemtxt, fontname, fontsize))
+
+            # write the text to the page
+            text.textLine(elemtxt)
+            pdf.drawText(text)
+
+        # put the image on the page, scaled to fill the page
+        if imageFileName is not None:
+            pdf.drawImage(imageFileName, 0, 0,
+                          width=self.width, height=self.height)
+
+        # finish up the page and save it
+        pdf.showPage()
+        pdf.save()
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description='Convert hocr file to PDF')
+    parser.add_argument('-b', '--boundingboxes', action="store_true",
+                        default=False, help='Show bounding boxes borders')
+    parser.add_argument('-r', '--resolution', type=int,
+                        default=300,
+                        help='Resolution of the image that was OCRed')
+    parser.add_argument('-i', '--image', default=None,
+                        help='Path to the image to be placed above the text')
+    parser.add_argument('hocrfile', help='Path to the hocr file to be parsed')
+    parser.add_argument(
+        'outputfile', help='Path to the PDF file to be generated')
+    args = parser.parse_args()
+
+    hocr = HocrTransform(args.hocrfile, args.resolution)
+    hocr.to_pdf(args.outputfile, args.image, args.boundingboxes)

ocrmypdf/leptonica.py

@@ -0,0 +1,331 @@
+#!/usr/bin/env python2
+# -*- coding: utf-8 -*-
+#
+# © 2013-15: jbarlow83 from Github (https://github.com/jbarlow83)
+#
+#
+# Use Leptonica to detect find and remove page skew.  Leptonica uses the method
+# of differential square sums, which its author claim is faster and more robust
+# than the Hough transform used by ImageMagick.
+
+from __future__ import print_function, absolute_import, division
+import argparse
+import ctypes as C
+import sys
+import os
+import logging
+from tempfile import TemporaryFile
+
+logger = logging.getLogger(__name__)
+
+
+def stderr(*objs):
+    """Python 2/3 compatible print to stderr.
+    """
+    print("leptonica.py:", *objs, file=sys.stderr)
+
+
+from ctypes.util import find_library
+lept_lib = find_library('lept')
+if not lept_lib:
+    stderr("Could not find the Leptonica library")
+    sys.exit(3)
+try:
+    lept = C.cdll.LoadLibrary(lept_lib)
+except Exception:
+    stderr("Could not load the Leptonica library from %s", lept_lib)
+    sys.exit(3)
+
+
+class _PIXCOLORMAP(C.Structure):
+    """struct PixColormap from Leptonica src/pix.h
+    """
+
+    _fields_ = [
+        ("array", C.c_void_p),
+        ("depth", C.c_int32),
+        ("nalloc", C.c_int32),
+        ("n", C.c_int32)
+    ]
+
+
+class _PIX(C.Structure):
+    """struct Pix from Leptonica src/pix.h
+    """
+
+    _fields_ = [
+        ("w", C.c_uint32),
+        ("h", C.c_uint32),
+        ("d", C.c_uint32),
+        ("wpl", C.c_uint32),
+        ("refcount", C.c_uint32),
+        ("xres", C.c_int32),
+        ("yres", C.c_int32),
+        ("informat", C.c_int32),
+        ("text", C.POINTER(C.c_char)),
+        ("colormap", C.POINTER(_PIXCOLORMAP)),
+        ("data", C.POINTER(C.c_uint32))
+    ]
+
+
+PIX = C.POINTER(_PIX)
+
+lept.pixRead.argtypes = [C.c_char_p]
+lept.pixRead.restype = PIX
+lept.pixScale.argtypes = [PIX, C.c_float, C.c_float]
+lept.pixScale.restype = PIX
+lept.pixDeskew.argtypes = [PIX, C.c_int32]
+lept.pixDeskew.restype = PIX
+lept.pixFindSkew.argtypes = [PIX, C.POINTER(C.c_float), C.POINTER(C.c_float)]
+lept.pixFindSkew.restype = C.c_int32
+lept.pixWriteImpliedFormat.argtypes = [C.c_char_p, PIX, C.c_int32, C.c_int32]
+lept.pixWriteImpliedFormat.restype = C.c_int32
+lept.pixDestroy.argtypes = [C.POINTER(PIX)]
+lept.pixDestroy.restype = None
+lept.getLeptonicaVersion.argtypes = []
+lept.getLeptonicaVersion.restype = C.c_char_p
+
+
+class LeptonicaErrorTrap(object):
+    """Context manager to trap errors reported by Leptonica.
+
+    Leptonica's error return codes are unreliable to the point of being
+    almost useless.  It does, however, write errors to stderr provided that is
+    not disabled at its compile time.  Fortunately this is done using error
+    macros so it is very self-consistent.
+
+    This context manager redirects stderr to a temporary file which is then
+    read and parsed for error messages.  As a side benefit, debug messages
+    from Leptonica are also suppressed.
+
+    """
+    def __enter__(self):
+        self.tmpfile = TemporaryFile()
+
+        # Save the old stderr, and redirect stderr to temporary file
+        self.old_stderr_fileno = os.dup(sys.stderr.fileno())
+        os.dup2(self.tmpfile.fileno(), sys.stderr.fileno())
+        return
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        # Restore old stderr
+        os.dup2(self.old_stderr_fileno, sys.stderr.fileno())
+
+        # Get data from tmpfile (in with block to ensure it is closed)
+        with self.tmpfile as tmpfile:
+            tmpfile.seek(0)  # Cursor will be at end, so move back to beginning
+            leptonica_output = tmpfile.read().decode(errors='replace')
+
+        # If there are Python errors, let them bubble up
+        if exc_type:
+            logger.warning(leptonica_output)
+            return False
+
+        # If there are Leptonica errors, wrap them in Python excpetions
+        if 'Error' in leptonica_output:
+            if 'image file not found' in leptonica_output:
+                raise FileNotFoundError()
+            if 'pixWrite: stream not opened' in leptonica_output:
+                raise LeptonicaIOError()
+            raise LeptonicaError(leptonica_output)
+
+        return False
+
+
+class LeptonicaError(Exception):
+    pass
+
+
+class LeptonicaIOError(LeptonicaError):
+    pass
+
+
+def pixRead(filename):
+    """Load an image file into a PIX object.
+
+    Leptonica can load TIFF, PNM (PBM, PGM, PPM), PNG, and JPEG.  If loading
+    fails then the object will wrap a C null pointer.
+
+    """
+    with LeptonicaErrorTrap():
+        return lept.pixRead(filename.encode(sys.getfilesystemencoding()))
+
+
+def pixScale(pix, scalex, scaley):
+    """Returns the pix object rescaled according to the proportions given."""
+    with LeptonicaErrorTrap():
+        return lept.pixScale(pix, scalex, scaley)
+
+
+def pixDeskew(pix, reduction_factor=0):
+    """Returns the deskewed pix object.
+
+    A clone of the original is returned when the algorithm cannot find a skew
+    angle with sufficient confidence.
+
+    reduction_factor -- amount to downsample (0 for default) when searching
+        for skew angle
+
+    """
+    with LeptonicaErrorTrap():
+        return lept.pixDeskew(pix, reduction_factor)
+
+
+def pixFindSkew(pix):
+    """Returns a tuple (deskew angle in degrees, confidence value).
+
+    Returns (None, None) if no angle is available.
+
+    """
+    with LeptonicaErrorTrap():
+        angle = C.c_float(0.0)
+        confidence = C.c_float(0.0)
+        result = lept.pixFindSkew(pix, C.byref(angle), C.byref(confidence))
+        if result == 0:
+            return (angle.value, confidence.value)
+        else:
+            return (None, None)
+
+
+def pixWriteImpliedFormat(filename, pix, jpeg_quality=0, jpeg_progressive=0):
+    """Write pix to the filename, with the extension indicating format.
+
+    jpeg_quality -- quality (iff JPEG; 1 - 100, 0 for default)
+    jpeg_progressive -- (iff JPEG; 0 for baseline seq., 1 for progressive)
+
+    """
+    fileroot, extension = os.path.splitext(filename)
+    fix_pnm = False
+    if extension.lower() in ('.pbm', '.pgm', '.ppm'):
+        # Leptonica does not process handle these extensions correctly, but
+        # does handle .pnm correctly.  Add another .pnm suffix.
+        filename += '.pnm'
+        fix_pnm = True
+
+    with LeptonicaErrorTrap():
+        lept.pixWriteImpliedFormat(
+            filename.encode(sys.getfilesystemencoding()),
+            pix, jpeg_quality, jpeg_progressive)
+
+    if fix_pnm:
+        from shutil import move
+        move(filename, filename[:-4])   # Remove .pnm suffix
+
+
+def pixDestroy(pix):
+    """Destroy the pix object.
+
+    Function signature is pixDestroy(struct Pix **), hence C.byref() to pass
+    the address of the pointer.
+
+    """
+    with LeptonicaErrorTrap():
+        lept.pixDestroy(C.byref(pix))
+
+
+def getLeptonicaVersion():
+    """Get Leptonica version string.
+
+    Caveat: Leptonica expects the caller to free this memory.  We don't,
+    since that would involve binding to libc to access libc.free(),
+    a pointless effort to reclaim 100 bytes of memory.
+
+    """
+    return lept.getLeptonicaVersion().decode()
+
+
+def deskew(infile, outfile, dpi):
+    try:
+        pix_source = pixRead(infile)
+    except LeptonicaIOError:
+        raise LeptonicaIOError("Failed to open file: %s" % infile)
+
+    if dpi < 150:
+        reduction_factor = 1  # Don't downsample too much if DPI is already low
+    else:
+        reduction_factor = 0  # Use default
+    pix_deskewed = pixDeskew(pix_source, reduction_factor)
+
+    try:
+        pixWriteImpliedFormat(outfile, pix_deskewed)
+    except LeptonicaIOError:
+        raise LeptonicaIOError("Failed to open destination file: %s" % outfile)
+    pixDestroy(pix_source)
+    pixDestroy(pix_deskewed)
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(
+        description="Python wrapper to access Leptonica")
+
+    subparsers = parser.add_subparsers(title='commands',
+                                       description='supported operations')
+
+    parser_deskew = subparsers.add_parser('deskew')
+    parser_deskew.add_argument('-r', '--dpi', dest='dpi', action='store',
+                               type=int, default=300, help='input resolution')
+    parser_deskew.add_argument('infile', help='image to deskew')
+    parser_deskew.add_argument('outfile', help='deskewed output image')
+    parser_deskew.set_defaults(func=deskew)
+
+    args = parser.parse_args()
+
+    if getLeptonicaVersion() != u'leptonica-1.69':
+        print("Unexpected leptonica version: %s" % getLeptonicaVersion())
+
+    args.func(args)
+
+
+def _test_output(mode, extension, im_format):
+    from PIL import Image
+    from tempfile import NamedTemporaryFile
+
+    with NamedTemporaryFile(prefix='test-lept-pnm', suffix=extension, delete=True) as tmpfile:
+        im = Image.new(mode=mode, size=(100, 100))
+        im.save(tmpfile)
+
+        pix = pixRead(tmpfile.name)
+        pixWriteImpliedFormat(tmpfile.name, pix)
+        pixDestroy(pix)
+
+        im_roundtrip = Image.open(tmpfile.name)
+        assert im_roundtrip.mode == im.mode, "leptonica mode differs"
+        assert im_roundtrip.format == im_format, \
+            "{0}: leptonica produced a {1}".format(
+                extension,
+                im_roundtrip.format)
+
+
+def test_pnm_output():
+    params = [['1', '.pbm', 'PPM'], ['L', '.pgm', 'PPM'],
+              ['RGB', '.ppm', 'PPM']]
+    for param in params:
+        _test_output(*param)
+
+
+def test_skew_angle():
+    from PIL import Image, ImageDraw
+    from tempfile import NamedTemporaryFile
+
+    im = Image.new(mode='1', size=(1000, 1000), color=1)
+
+    draw = ImageDraw.Draw(im)
+    for n in range(20):
+        draw.line([(50, 25 + 50*n), (950, 25 + 50*n)], width=1)
+    del draw
+
+    test_angles = [0.1 * ang for ang in range(1, 10)] + \
+                  [float(ang) for ang in range(1, 7)]
+    test_angles += [-ang for ang in test_angles]
+    test_angles = sorted(test_angles)
+
+    for rotate_angle in test_angles:
+        rotated_im = im.rotate(rotate_angle)
+        with NamedTemporaryFile(prefix='lept-skew', suffix='.png', delete=True) as tmpfile:
+            rotated_im.save(tmpfile)
+            pix = pixRead(tmpfile.name)
+            angle, confidence = pixFindSkew(pix)
+            pixDestroy(pix)
+            print('{0} {1}  {2}'.format(rotate_angle, angle, confidence), file=sys.stderr)
+
+

ocrmypdf/main.py

@@ -0,0 +1,904 @@
+#!/usr/bin/env python3
+# © 2015 James R. Barlow: github.com/jbarlow83
+
+from contextlib import suppress
+from tempfile import NamedTemporaryFile, mkdtemp
+import sys
+import os
+import re
+import shutil
+import warnings
+import multiprocessing
+import atexit
+import textwrap
+
+import PyPDF2 as pypdf
+from PIL import Image
+
+from subprocess import Popen, check_call, PIPE, CalledProcessError, \
+    TimeoutExpired, check_output, STDOUT
+try:
+    from subprocess import DEVNULL
+except ImportError:
+    DEVNULL = open(os.devnull, 'wb')
+
+
+from ruffus import transform, suffix, merge, active_if, regex, jobs_limit, \
+    formatter, follows, split, collate, check_if_uptodate
+import ruffus.ruffus_exceptions as ruffus_exceptions
+import ruffus.cmdline as cmdline
+
+from .hocrtransform import HocrTransform
+from .pageinfo import pdf_get_all_pageinfo
+from .pdfa import generate_pdfa_def
+from . import ghostscript
+from . import tesseract
+from . import ExitCode
+
+warnings.simplefilter('ignore', pypdf.utils.PdfReadWarning)
+
+
+BASEDIR = os.path.dirname(os.path.realpath(__file__))
+JHOVE_PATH = os.path.realpath(os.path.join(BASEDIR, 'jhove'))
+JHOVE_JAR = os.path.join(JHOVE_PATH, 'bin', 'JhoveApp.jar')
+JHOVE_CFG = os.path.join(JHOVE_PATH, 'conf', 'jhove.conf')
+
+
+# -------------
+# External dependencies
+
+MINIMUM_TESS_VERSION = '3.02.02'
+
+
+def complain(message):
+    print(*textwrap.wrap(message), file=sys.stderr)
+
+
+if tesseract.version() < MINIMUM_TESS_VERSION:
+    complain(
+        "Please install tesseract {0} or newer "
+        "(currently installed version is {1})".format(
+            MINIMUM_TESS_VERSION, tesseract.version()))
+    sys.exit(ExitCode.missing_dependency)
+
+
+try:
+    import PIL.features
+    check_codec = PIL.features.check_codec
+except (ImportError, AttributeError):
+    def check_codec(codec_name):
+        if codec_name == 'jpg':
+            return 'jpeg_encoder' in dir(Image.core)
+        elif codec_name == 'zlib':
+            return 'zip_encoder' in dir(Image.core)
+        raise NotImplementedError(codec_name)
+
+
+def check_pil_encoder(codec_name, friendly_name):
+    try:
+        if check_codec(codec_name):
+            return
+    except Exception:
+        pass
+    complain(
+        "ERROR: Your version of the Python imaging library (Pillow) was "
+        "compiled without support for " + friendly_name + " encoding/decoding."
+        "\n"
+        "You will need to uninstall Pillow and reinstall it with PNG and JPEG "
+        "support (libjpeg and zlib)."
+        "\n"
+        "See installation instructions for your platform here:\n"
+        "    https://pillow.readthedocs.org/installation.html"
+    )
+    sys.exit(ExitCode.missing_dependency)
+
+
+check_pil_encoder('jpg', 'JPEG')
+check_pil_encoder('zlib', 'PNG')
+
+
+# -------------
+# Parser
+
+parser = cmdline.get_argparse(
+    prog="ocrmypdf",
+    description="Generate searchable PDF file from an image-only PDF file.",
+    version='3.0rc6',
+    fromfile_prefix_chars='@',
+    ignored_args=[
+        'touch_files_only', 'recreate_database', 'checksum_file_name',
+        'key_legend_in_graph', 'draw_graph_horizontally', 'flowchart_format',
+        'forced_tasks', 'target_tasks'])
+
+parser.add_argument(
+    'input_file',
+    help="PDF file containing the images to be OCRed")
+parser.add_argument(
+    'output_file',
+    help="output searchable PDF file")
+parser.add_argument(
+    '-l', '--language', action='append',
+    help="languages of the file to be OCRed")
+
+metadata = parser.add_argument_group(
+    "Metadata options",
+    "Set output PDF/A metadata (default: use input document's title)")
+metadata.add_argument(
+    '--title', type=str,
+    help="set document title (place multiple words in quotes)")
+metadata.add_argument(
+    '--author', type=str,
+    help="set document author")
+metadata.add_argument(
+    '--subject', type=str,
+    help="set document")
+metadata.add_argument(
+    '--keywords', type=str,
+    help="set document keywords")
+
+
+preprocessing = parser.add_argument_group(
+    "Preprocessing options",
+    "Improve OCR quality and final image")
+preprocessing.add_argument(
+    '-d', '--deskew', action='store_true',
+    help="deskew each page before performing OCR")
+preprocessing.add_argument(
+    '-c', '--clean', action='store_true',
+    help="clean pages from scanning artifacts before performing OCR")
+preprocessing.add_argument(
+    '-i', '--clean-final', action='store_true',
+    help="incorporate the cleaned image in the final PDF file")
+preprocessing.add_argument(
+    '--oversample', metavar='DPI', type=int, default=0,
+    help="oversample images to at least the specified DPI, to improve OCR "
+         "results slightly")
+
+parser.add_argument(
+    '-f', '--force-ocr', action='store_true',
+    help="rasterize any fonts or vector images on each page and apply OCR")
+parser.add_argument(
+    '-s', '--skip-text', action='store_true',
+    help="skip OCR on any pages that already contain text, but include the"
+         " page in final output")
+parser.add_argument(
+    '--skip-big', type=float, metavar='MPixels',
+    help="skip OCR on pages larger than the specified amount of megapixels, "
+         "but include skipped pages in final output")
+# parser.add_argument(
+#     '--exact-image', action='store_true',
+#     help="Use original page from PDF without re-rendering")
+
+advanced = parser.add_argument_group(
+    "Advanced",
+    "Advanced options for power users")
+advanced.add_argument(
+    '--tesseract-config', default=[], type=list, action='append',
+    help="additional Tesseract configuration files")
+advanced.add_argument(
+    '--pdf-renderer', choices=['tesseract', 'hocr'], default='hocr',
+    help='choose OCR PDF renderer')
+advanced.add_argument(
+    '--tesseract-timeout', default=180.0, type=float,
+    help='give up on OCR after the timeout, but copy the preprocessed page '
+         'into the final output')
+
+debugging = parser.add_argument_group(
+    "Debugging",
+    "Arguments to help with troubleshooting and debugging")
+debugging.add_argument(
+    '-k', '--keep-temporary-files', action='store_true',
+    help="keep temporary files (helpful for debugging)")
+debugging.add_argument(
+    '-g', '--debug-rendering', action='store_true',
+    help="render each page twice with debug information on second page")
+
+options = parser.parse_args()
+
+
+# ----------
+# Languages
+
+if not options.language:
+    options.language = ['eng']  # Enforce English hegemony
+
+# Support v2.x "eng+deu" language syntax
+if '+' in options.language[0]:
+    options.language = options.language[0].split('+')
+
+if not set(options.language).issubset(tesseract.languages()):
+    complain(
+        "The installed version of tesseract does not have language "
+        "data for the following requested languages: ")
+    for lang in (set(options.language) - tesseract.languages()):
+        complain(lang)
+    sys.exit(ExitCode.bad_args)
+
+
+# ----------
+# Arguments
+
+
+if any((options.deskew, options.clean, options.clean_final)):
+    try:
+        from . import unpaper
+    except ImportError:
+        complain(
+            "Install the 'unpaper' program to use --deskew or --clean.")
+        sys.exit(ExitCode.bad_args)
+else:
+    unpaper = None
+
+if options.debug_rendering and options.pdf_renderer == 'tesseract':
+    complain(
+        "Ignoring --debug-rendering because it is not supported with"
+        "--pdf-renderer=tesseract.")
+
+if options.force_ocr and options.skip_text:
+    complain(
+        "Error: --force-ocr and --skip-text are mutually incompatible.")
+    sys.exit(ExitCode.bad_args)
+
+if options.clean and not options.clean_final \
+        and options.pdf_renderer == 'tesseract':
+    complain(
+        "Tesseract PDF renderer cannot render --clean pages without "
+        "also performing --clean-final, so --clean-final is assumed.")
+
+
+# ----------
+# Logging
+
+
+_logger, _logger_mutex = cmdline.setup_logging(__name__, options.log_file,
+                                               options.verbose)
+
+
+class WrappedLogger:
+
+    def __init__(self, my_logger, my_mutex):
+        self.logger = my_logger
+        self.mutex = my_mutex
+
+    def log(self, *args, **kwargs):
+        with self.mutex:
+            self.logger.log(*args, **kwargs)
+
+    def debug(self, *args, **kwargs):
+        with self.mutex:
+            self.logger.debug(*args, **kwargs)
+
+    def info(self, *args, **kwargs):
+        with self.mutex:
+            self.logger.info(*args, **kwargs)
+
+    def warning(self, *args, **kwargs):
+        with self.mutex:
+            self.logger.warning(*args, **kwargs)
+
+    def error(self, *args, **kwargs):
+        with self.mutex:
+            self.logger.error(*args, **kwargs)
+
+    def critical(self, *args, **kwargs):
+        with self.mutex:
+            self.logger.critical(*args, **kwargs)
+
+_log = WrappedLogger(_logger, _logger_mutex)
+
+
+def re_symlink(input_file, soft_link_name, log=_log):
+    """
+    Helper function: relinks soft symbolic link if necessary
+    """
+    # Guard against soft linking to oneself
+    if input_file == soft_link_name:
+        log.debug("Warning: No symbolic link made. You are using " +
+                  "the original data directory as the working directory.")
+        return
+
+    # Soft link already exists: delete for relink?
+    if os.path.lexists(soft_link_name):
+        # do not delete or overwrite real (non-soft link) file
+        if not os.path.islink(soft_link_name):
+            raise Exception("%s exists and is not a link" % soft_link_name)
+        try:
+            os.unlink(soft_link_name)
+        except:
+            log.debug("Can't unlink %s" % (soft_link_name))
+
+    if not os.path.exists(input_file):
+        raise Exception("trying to create a broken symlink to %s" % input_file)
+
+    log.debug("os.symlink(%s, %s)" % (input_file, soft_link_name))
+
+    # Create symbolic link using absolute path
+    os.symlink(
+        os.path.abspath(input_file),
+        soft_link_name
+    )
+
+
+# -------------
+# The Pipeline
+
+manager = multiprocessing.Manager()
+_pdfinfo = manager.list()
+_pdfinfo_lock = manager.Lock()
+
+work_folder = mkdtemp(prefix="com.github.ocrmypdf.")
+
+
+@atexit.register
+def cleanup_working_files(*args):
+    if options.keep_temporary_files:
+        print("Temporary working files saved at:")
+        print(work_folder)
+    else:
+        with suppress(FileNotFoundError):
+            shutil.rmtree(work_folder)
+
+
+@transform(
+    input=options.input_file,
+    filter=suffix('.pdf'),
+    output='.repaired.pdf',
+    output_dir=work_folder,
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def repair_pdf(
+        input_file,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+    args_qpdf = [
+        'qpdf', input_file, output_file
+    ]
+    try:
+        out = check_output(args_qpdf, stderr=STDOUT, universal_newlines=True)
+    except CalledProcessError as e:
+        if e.returncode == 2:
+            print("{0}: not a valid PDF, and could not repair it.".format(
+                    options.input_file))
+            print("Details:")
+            print(e.output)
+        else:
+            print(e.output)
+        sys.exit(ExitCode.input_file)
+
+    log.debug(out)
+
+    with pdfinfo_lock:
+        pdfinfo.extend(pdf_get_all_pageinfo(output_file))
+        log.info(pdfinfo)
+
+
+def get_pageinfo(input_file, pdfinfo, pdfinfo_lock):
+    pageno = int(os.path.basename(input_file)[0:6]) - 1
+    with pdfinfo_lock:
+        pageinfo = pdfinfo[pageno].copy()
+    return pageinfo
+
+
+def is_ocr_required(pageinfo, log):
+    page = pageinfo['pageno'] + 1
+    ocr_required = True
+    if not pageinfo['images']:
+        # If the page has no images, then it contains vector content or text
+        # or both. It seems quite unlikely that one would find meaningful text
+        # from rasterizing vector content. So skip the page.
+        log.info(
+            "Page {0} has no images - skipping OCR".format(page)
+        )
+        ocr_required = False
+    elif pageinfo['has_text']:
+        s = "Page {0} already has text! – {1}"
+
+        if not options.force_ocr and not options.skip_text:
+            log.error(s.format(page,
+                               "aborting (use --force-ocr to force OCR)"))
+            sys.exit(ExitCode.already_done_ocr)
+        elif options.force_ocr:
+            log.info(s.format(page,
+                              "rasterizing text and running OCR anyway"))
+            ocr_required = True
+        elif options.skip_text:
+            log.info(s.format(page,
+                              "skipping all processing on this page"))
+            ocr_required = False
+
+    if ocr_required and options.skip_big:
+        pixel_count = pageinfo['width_pixels'] * pageinfo['height_pixels']
+        if pixel_count > (options.skip_big * 1000000):
+            ocr_required = False
+            log.info(
+                "Page {0} is very large; skipping due to -b".format(page))
+
+    return ocr_required
+
+
+@split(
+    repair_pdf,
+    os.path.join(work_folder, '*.page.pdf'),
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def split_pages(
+        input_file,
+        output_files,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+
+    for oo in output_files:
+        with suppress(FileNotFoundError):
+            os.unlink(oo)
+
+    pages = check_output(['qpdf', '--show-npages', input_file],
+                         universal_newlines=True, close_fds=True)
+
+    for n in range(int(pages)):
+        args_qpdf = [
+            'qpdf', input_file,
+            '--pages', input_file, '{0}'.format(n + 1), '--',
+            os.path.join(work_folder, '{0:06d}.page.pdf'.format(n + 1))
+        ]
+        check_call(args_qpdf)
+
+    from glob import glob
+    for filename in glob(os.path.join(work_folder, '*.page.pdf')):
+        pageinfo = get_pageinfo(filename, pdfinfo, pdfinfo_lock)
+
+        alt_suffix = '.ocr.page.pdf' if is_ocr_required(pageinfo, log) \
+                     else '.skip.page.pdf'
+        re_symlink(
+            filename,
+            os.path.join(
+                work_folder,
+                os.path.basename(filename)[0:6] + alt_suffix))
+
+
+@transform(
+    input=split_pages,
+    filter=suffix('.ocr.page.pdf'),
+    output='.page.png',
+    output_dir=work_folder,
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def rasterize_with_ghostscript(
+        input_file,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+
+    pageinfo = get_pageinfo(input_file, pdfinfo, pdfinfo_lock)
+
+    device = 'png16m'  # 24-bit
+    if all(image['comp'] == 1 for image in pageinfo['images']):
+        if all(image['bpc'] == 1 for image in pageinfo['images']):
+            device = 'pngmono'
+        elif not any(image['color'] == 'color'
+                     for image in pageinfo['images']):
+            device = 'pnggray'
+
+    xres = max(pageinfo['xres'], options.oversample or 0)
+    yres = max(pageinfo['yres'], options.oversample or 0)
+
+    ghostscript.rasterize_pdf(input_file, output_file, xres, yres, device, log)
+
+
+@transform(
+    input=rasterize_with_ghostscript,
+    filter=suffix(".page.png"),
+    output=".pp-deskew.png",
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def preprocess_deskew(
+        input_file,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+
+    if not options.deskew:
+        re_symlink(input_file, output_file, log)
+        return
+
+    pageinfo = get_pageinfo(input_file, pdfinfo, pdfinfo_lock)
+    dpi = int(pageinfo['xres'])
+
+    unpaper.deskew(input_file, output_file, dpi, log)
+
+
+@transform(
+    input=preprocess_deskew,
+    filter=suffix(".pp-deskew.png"),
+    output=".pp-clean.png",
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def preprocess_clean(
+        input_file,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+
+    if not options.clean:
+        re_symlink(input_file, output_file, log)
+        return
+
+    pageinfo = get_pageinfo(input_file, pdfinfo, pdfinfo_lock)
+    dpi = int(pageinfo['xres'])
+
+    unpaper.clean(input_file, output_file, dpi, log)
+
+
+@active_if(options.pdf_renderer == 'hocr')
+@transform(
+    input=preprocess_clean,
+    filter=suffix(".pp-clean.png"),
+    output=".hocr",
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def ocr_tesseract_hocr(
+        input_file,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+
+    pageinfo = get_pageinfo(input_file, pdfinfo, pdfinfo_lock)
+
+    badxml = os.path.splitext(output_file)[0] + '.badxml'
+
+    args_tesseract = [
+        'tesseract',
+        '-l', '+'.join(options.language),
+        input_file,
+        badxml,
+        'hocr'
+    ] + options.tesseract_config
+    p = Popen(args_tesseract, close_fds=True, stdout=PIPE, stderr=PIPE,
+              universal_newlines=True)
+    try:
+        stdout, stderr = p.communicate(timeout=options.tesseract_timeout)
+    except TimeoutExpired:
+        p.kill()
+        stdout, stderr = p.communicate()
+        # Generate a HOCR file with no recognized text if tesseract times out
+        # Temporary workaround to hocrTransform not being able to function if
+        # it does not have a valid hOCR file.
+        with open(output_file, 'w', encoding="utf-8") as f:
+            f.write(tesseract.HOCR_TEMPLATE.format(
+                pageinfo['width_pixels'],
+                pageinfo['height_pixels']))
+    else:
+        if stdout:
+            log.info(stdout)
+        if stderr:
+            log.error(stderr)
+
+        if p.returncode != 0:
+            raise CalledProcessError(p.returncode, args_tesseract)
+
+        if os.path.exists(badxml + '.html'):
+            # Tesseract 3.02 appends suffix ".html" on its own (.badxml.html)
+            shutil.move(badxml + '.html', badxml)
+        elif os.path.exists(badxml + '.hocr'):
+            # Tesseract 3.03 appends suffix ".hocr" on its own (.badxml.hocr)
+            shutil.move(badxml + '.hocr', badxml)
+
+        # Tesseract 3.03 inserts source filename into hocr file without
+        # escaping it, creating invalid XML and breaking the parser.
+        # As a workaround, rewrite the hocr file, replacing the filename
+        # with a space.  Don't know if Tesseract 3.02 does the same.
+
+        regex_nested_single_quotes = re.compile(
+            r"""title='image "([^"]*)";""")
+        with open(badxml, mode='r', encoding='utf-8') as f_in, \
+                open(output_file, mode='w', encoding='utf-8') as f_out:
+            for line in f_in:
+                line = regex_nested_single_quotes.sub(
+                    r"""title='image " ";""", line)
+                f_out.write(line)
+
+
+@active_if(options.pdf_renderer == 'hocr')
+@collate(
+    input=[rasterize_with_ghostscript, preprocess_deskew, preprocess_clean],
+    filter=regex(r".*/(\d{6})(?:\.page|\.pp-deskew|\.pp-clean)\.png"),
+    output=os.path.join(work_folder, r'\1.image'),
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def select_image_for_pdf(
+        infiles,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+    if options.clean_final:
+        image_suffix = '.pp-clean.png'
+    elif options.deskew:
+        image_suffix = '.pp-deskew.png'
+    else:
+        image_suffix = '.page.png'
+    image = next(ii for ii in infiles if ii.endswith(image_suffix))
+
+    pageinfo = get_pageinfo(image, pdfinfo, pdfinfo_lock)
+    if all(image['enc'] == 'jpeg' for image in pageinfo['images']):
+        # If all images were JPEGs originally, produce a JPEG as output
+        Image.open(image).save(output_file, format='JPEG')
+    else:
+        re_symlink(image, output_file)
+
+
+@active_if(options.pdf_renderer == 'hocr')
+@collate(
+    input=[select_image_for_pdf, ocr_tesseract_hocr],
+    filter=regex(r".*/(\d{6})(?:\.image|\.hocr)"),
+    output=os.path.join(work_folder, r'\1.rendered.pdf'),
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def render_hocr_page(
+        infiles,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+    hocr = next(ii for ii in infiles if ii.endswith('.hocr'))
+    image = next(ii for ii in infiles if ii.endswith('.image'))
+
+    pageinfo = get_pageinfo(image, pdfinfo, pdfinfo_lock)
+    dpi = round(max(pageinfo['xres'], pageinfo['yres'], options.oversample))
+
+    hocrtransform = HocrTransform(hocr, dpi)
+    hocrtransform.to_pdf(output_file, imageFileName=image,
+                         showBoundingboxes=False, invisibleText=True)
+
+
+@active_if(options.pdf_renderer == 'hocr')
+@active_if(options.debug_rendering)
+@collate(
+    input=[select_image_for_pdf, ocr_tesseract_hocr],
+    filter=regex(r".*/(\d{6})(?:\.image|\.hocr)"),
+    output=os.path.join(work_folder, r'\1.debug.pdf'),
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def render_hocr_debug_page(
+        infiles,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+    hocr = next(ii for ii in infiles if ii.endswith('.hocr'))
+    image = next(ii for ii in infiles if ii.endswith('.image'))
+
+    pageinfo = get_pageinfo(image, pdfinfo, pdfinfo_lock)
+    dpi = round(max(pageinfo['xres'], pageinfo['yres'], options.oversample))
+
+    hocrtransform = HocrTransform(hocr, dpi)
+    hocrtransform.to_pdf(output_file, imageFileName=None,
+                         showBoundingboxes=True, invisibleText=False)
+
+
+@active_if(options.pdf_renderer == 'tesseract')
+@collate(
+    input=[preprocess_clean, split_pages],
+    filter=regex(r".*/(\d{6})(?:\.pp-clean\.png|\.page\.pdf)"),
+    output=os.path.join(work_folder, r'\1.rendered.pdf'),
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def tesseract_ocr_and_render_pdf(
+        input_files,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+
+    input_image = next((ii for ii in input_files if ii.endswith('.png')), '')
+    input_pdf = next((ii for ii in input_files if ii.endswith('.pdf')))
+    if not input_image:
+        # Skipping this page
+        re_symlink(input_pdf, output_file)
+        return
+
+    args_tesseract = [
+        'tesseract',
+        '-l', '+'.join(options.language),
+        input_image,
+        os.path.splitext(output_file)[0],  # Tesseract appends suffix
+        'pdf'
+    ] + options.tesseract_config
+    p = Popen(args_tesseract, close_fds=True, stdout=PIPE, stderr=PIPE,
+              universal_newlines=True)
+
+    try:
+        stdout, stderr = p.communicate(timeout=options.tesseract_timeout)
+        if stdout:
+            log.info(stdout)
+        if stderr:
+            log.error(stderr)
+    except TimeoutExpired:
+        p.kill()
+        log.info("Tesseract - page timed out")
+        re_symlink(input_pdf, output_file)
+
+
+@transform(
+    input=repair_pdf,
+    filter=suffix('.repaired.pdf'),
+    output='.pdfa_def.ps',
+    output_dir=work_folder,
+    extras=[_log])
+def generate_postscript_stub(
+        input_file,
+        output_file,
+        log):
+
+    pdf = pypdf.PdfFileReader(input_file)
+
+    def from_document_info(key):
+        # pdf.documentInfo.get() DOES NOT work as expected
+        try:
+            s = pdf.documentInfo[key]
+            return str(s)
+        except KeyError:
+            return ''
+
+    pdfmark = {
+        'title': from_document_info('/Title'),
+        'author': from_document_info('/Author'),
+        'keywords': from_document_info('/Keywords'),
+        'subject': from_document_info('/Subject'),
+    }
+    if options.title:
+        pdfmark['title'] = options.title
+    if options.author:
+        pdfmark['author'] = options.author
+    if options.keywords:
+        pdfmark['keywords'] = options.keywords
+    if options.subject:
+        pdfmark['subject'] = options.subject
+
+    generate_pdfa_def(output_file, pdfmark)
+
+
+@transform(
+    input=split_pages,
+    filter=suffix('.skip.page.pdf'),
+    output='.done.pdf',
+    output_dir=work_folder,
+    extras=[_log])
+def skip_page(
+        input_file,
+        output_file,
+        log):
+    re_symlink(input_file, output_file, log)
+
+
+@merge(
+    input=[render_hocr_page, render_hocr_debug_page, skip_page,
+           tesseract_ocr_and_render_pdf, generate_postscript_stub],
+    output=os.path.join(work_folder, 'merged.pdf'),
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def merge_pages(
+        input_files,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+
+    def input_file_order(s):
+        '''Sort order: All rendered pages followed
+        by their debug page, if any, followed by Postscript stub.
+        Ghostscript documentation has the Postscript stub at the
+        beginning, but it works at the end and also gets document info
+        right that way.'''
+        if s.endswith('.ps'):
+            return 99999999
+        key = int(os.path.basename(s)[0:6]) * 10
+        if 'debug' in os.path.basename(s):
+            key += 1
+        return key
+
+    pdf_pages = sorted(input_files, key=input_file_order)
+    log.info(pdf_pages)
+    ghostscript.generate_pdfa(pdf_pages, output_file, options.jobs or 1)
+
+
+@transform(
+    input=merge_pages,
+    filter=formatter(),
+    output=options.output_file,
+    extras=[_log, _pdfinfo, _pdfinfo_lock])
+def copy_final(
+        input_file,
+        output_file,
+        log,
+        pdfinfo,
+        pdfinfo_lock):
+    shutil.copy(input_file, output_file)
+
+
+def validate_pdfa(
+        input_file,
+        log):
+
+    args_qpdf = [
+        'qpdf',
+        '--check',
+        input_file
+    ]
+
+    try:
+        check_output(args_qpdf, stderr=STDOUT, universal_newlines=True)
+    except CalledProcessError as e:
+        if e.returncode == 2:
+            print("{0}: not a valid PDF, and could not repair it.".format(
+                    options.input_file))
+            print("Details:")
+            print(e.output)
+        elif e.returncode == 3:
+            log.info("qpdf --check returned warnings:")
+            log.info(e.output)
+        else:
+            print(e.output)
+        return False
+
+    return True
+
+
+# @active_if(ocr_required and options.exact_image)
+# @merge([render_hocr_blank_page, extract_single_page],
+#        os.path.join(work_folder, "%04i.merged.pdf") % pageno)
+# def merge_hocr_with_original_page(infiles, output_file):
+#     with open(infiles[0], 'rb') as hocr_input, \
+#             open(infiles[1], 'rb') as page_input, \
+#             open(output_file, 'wb') as output:
+#         hocr_reader = pypdf.PdfFileReader(hocr_input)
+#         page_reader = pypdf.PdfFileReader(page_input)
+#         writer = pypdf.PdfFileWriter()
+
+#         the_page = hocr_reader.getPage(0)
+#         the_page.mergePage(page_reader.getPage(0))
+#         writer.addPage(the_page)
+#         writer.write(output)
+
+
+def available_cpu_count():
+    try:
+        return multiprocessing.cpu_count()
+    except NotImplementedError:
+        pass
+
+    try:
+        import psutil
+        return psutil.cpu_count()
+    except (ImportError, AttributeError):
+        pass
+
+    complain(
+        "Could not get CPU count.  Assuming one (1) CPU."
+        "Use -j N to set manually.")
+    return 1
+
+
+def run_pipeline():
+    if not options.jobs or options.jobs == 1:
+        options.jobs = available_cpu_count()
+
+    try:
+        cmdline.run(options)
+    except ruffus_exceptions.RethrownJobError as e:
+        if options.verbose:
+            print(e)
+
+        # Yuck. Hunt through the ruffus exception to find out what the
+        # return code is supposed to be.
+        for exc in e.args:
+            task_name, job_name, exc_name, exc_value, exc_stack = exc
+            if exc_name == 'builtins.SystemExit':
+                return eval(
+                    exc_value,
+                    {'ExitCode': ExitCode}, {'exc_value': exc_value})
+
+    if not validate_pdfa(options.output_file, _log):
+        _log.warning('Output file: The generated PDF/A file is INVALID')
+        return ExitCode.invalid_output_pdfa
+
+    return ExitCode.ok
+
+
+if __name__ == '__main__':
+    sys.exit(run_pipeline())

ocrmypdf/pageinfo.py

@@ -0,0 +1,144 @@
+#!/usr/bin/env python3
+# © 2015 James R. Barlow: github.com/jbarlow83
+
+from subprocess import Popen, PIPE
+from decimal import Decimal, getcontext
+import re
+import sys
+import PyPDF2 as pypdf
+
+
+FRIENDLY_COLORSPACE = {
+    '/DeviceGray': 'gray',
+    '/CalGray': 'gray',
+    '/DeviceRGB': 'rgb',
+    '/CalRGB': 'rgb',
+    '/DeviceCMYK': 'cmyk',
+    '/Lab': 'lab',
+    '/ICCBased': 'icc',
+    '/Indexed': 'index',
+    '/Separation': 'sep',
+    '/DeviceN': 'devn',
+    '/Pattern': '-'
+}
+
+FRIENDLY_ENCODING = {
+    '/CCITTFaxDecode': 'ccitt',
+    '/DCTDecode': 'jpeg',
+    '/JPXDecode': 'jpx',
+    '/JBIG2Decode': 'jbig2',
+}
+
+FRIENDLY_COMP = {
+    'gray': 1,
+    'rgb': 3,
+    'cmyk': 4,
+    'lab': 3,
+}
+
+
+def _page_has_inline_images(page):
+    # PDF always uses \r\n for separator regardless of platform
+    # Really basic heuristic that might trigger the odd false positive
+    # This is only finds the first image and is not quite spec compliant
+    try:
+        contents = page.getContents()
+        data = contents.getData()
+    except AttributeError:
+        # If we can't access the contents or data (empty page?) then there
+        # are no inline images
+        return False
+
+    begin_image, image_data, end_image = False, False, False
+    for data in re.split(b'\s+', data):
+        if data == b'BI':
+            begin_image = True
+        elif data == b'ID':
+            image_data = True
+        elif data == b'EI':
+            end_image = True
+        if all((begin_image, image_data, end_image)):
+            return True
+    return False
+
+
+def _find_page_images(page, pageinfo):
+    try:
+        page['/Resources']['/XObject']
+    except KeyError:
+        return
+
+    # Look for XObject (out of line images)
+    for xobj in page['/Resources']['/XObject']:
+        # PyPDF2 returns the keys as an iterator
+        pdfimage = page['/Resources']['/XObject'][xobj]
+        if pdfimage['/Subtype'] != '/Image':
+            continue
+        if '/ImageMask' in pdfimage:
+            if pdfimage['/ImageMask']:
+                continue
+        image = {}
+        image['width'] = pdfimage['/Width']
+        image['height'] = pdfimage['/Height']
+        image['bpc'] = pdfimage['/BitsPerComponent']
+        if '/Filter' in pdfimage:
+            filter_ = pdfimage['/Filter']
+            if isinstance(filter_, pypdf.generic.ArrayObject):
+                filter_ = filter_[0]
+            image['enc'] = FRIENDLY_ENCODING.get(filter_, 'image')
+        else:
+            image['enc'] = 'image'
+        if '/ColorSpace' in pdfimage:
+            cs = pdfimage['/ColorSpace']
+            if isinstance(cs, pypdf.generic.ArrayObject):
+                cs = cs[0]
+            image['color'] = FRIENDLY_COLORSPACE.get(cs, '-')
+        else:
+            image['color'] = 'jpx' if image['enc'] == 'jpx' else '?'
+
+        image['comp'] = FRIENDLY_COMP.get(image['color'], '?')
+        image['dpi_w'] = image['width'] / pageinfo['width_inches']
+        image['dpi_h'] = image['height'] / pageinfo['height_inches']
+        image['dpi'] = (image['dpi_w'] * image['dpi_h']) ** Decimal(0.5)
+        yield image
+
+
+def _pdf_get_pageinfo(infile, page: int):
+    pageinfo = {}
+    pageinfo['pageno'] = page
+    pageinfo['images'] = []
+
+    pdf = pypdf.PdfFileReader(infile)
+    page = pdf.pages[page - 1]
+
+    text = page.extractText()
+    pageinfo['has_text'] = (text.strip() != '')
+
+    width_pt = page['/MediaBox'][2] - page['/MediaBox'][0]
+    height_pt = page['/MediaBox'][3] - page['/MediaBox'][1]
+    pageinfo['width_inches'] = width_pt / Decimal(72.0)
+    pageinfo['height_inches'] = height_pt / Decimal(72.0)
+
+    pageinfo['images'] = [im for im in _find_page_images(page, pageinfo)]
+
+    # Look for inline images
+    if _page_has_inline_images(page):
+        raise NotImplementedError(
+            "Warning: input PDF contains inline images - not supported")
+
+    if pageinfo['images']:
+        xres = max(image['dpi_w'] for image in pageinfo['images'])
+        yres = max(image['dpi_h'] for image in pageinfo['images'])
+        pageinfo['xres'], pageinfo['yres'] = xres, yres
+        pageinfo['width_pixels'] = \
+            int(round(xres * pageinfo['width_inches']))
+        pageinfo['height_pixels'] = \
+            int(round(yres * pageinfo['height_inches']))
+
+    return pageinfo
+
+
+def pdf_get_all_pageinfo(infile):
+    pdf = pypdf.PdfFileReader(infile)
+    getcontext().prec = 6
+    return [_pdf_get_pageinfo(infile, n) for n in range(pdf.numPages)]

ocrmypdf/pdfa.py

@@ -0,0 +1,130 @@
+#!/usr/bin/env python3
+# © 2015 James R. Barlow: github.com/jbarlow83
+#
+# Generate a PDFA_def.ps file for Ghostscript >= 9.14
+
+from __future__ import print_function, absolute_import, division
+from string import Template
+from subprocess import Popen, PIPE
+import os
+import codecs
+
+
+# This is a template written in PostScript which is needed to create PDF/A
+# files, from the Ghostscript documentation. Lines beginning with % are
+# comments. Python substitution variables have a '$' prefix.
+pdfa_def_template = u"""%!
+% This is a sample prefix file for creating a PDF/A document.
+% Feel free to modify entries marked with "Customize".
+% This assumes an ICC profile to reside in the file (ISO Coated sb.icc),
+% unless the user modifies the corresponding line below.
+
+% Define entries in the document Info dictionary :
+/ICCProfile ($icc_profile)
+def
+
+[ /Title <$title>
+  /Author <$author>
+  /Subject <$subject>
+  /Keywords <$keywords>
+  /DOCINFO pdfmark
+
+% Define an ICC profile :
+
+[/_objdef {icc_PDFA} /type /stream /OBJ pdfmark
+[{icc_PDFA}
+<<
+  /N currentpagedevice /ProcessColorModel known {
+    currentpagedevice /ProcessColorModel get dup /DeviceGray eq
+    {pop 1} {
+      /DeviceRGB eq
+      {3}{4} ifelse
+    } ifelse
+  } {
+    (ERROR, unable to determine ProcessColorModel) == flush
+  } ifelse
+>> /PUT pdfmark
+[{icc_PDFA} ICCProfile (r) file /PUT pdfmark
+
+% Define the output intent dictionary :
+
+[/_objdef {OutputIntent_PDFA} /type /dict /OBJ pdfmark
+[{OutputIntent_PDFA} <<
+  /Type /OutputIntent             % Must be so (the standard requires).
+  /S /GTS_PDFA1                   % Must be so (the standard requires).
+  /DestOutputProfile {icc_PDFA}            % Must be so (see above).
+  /OutputConditionIdentifier ($icc_identifier)
+>> /PUT pdfmark
+[{Catalog} <</OutputIntents [ {OutputIntent_PDFA} ]>> /PUT pdfmark
+"""
+
+
+def encode_text_string(s: str) -> str:
+    '''Encode text string to hex string for use in a PDF
+
+    From PDF 32000-1:2008 a string object may be included in hexademical form
+    if it is enclosed in angle brackets.  For general Unicode the string should
+    be UTF-16 (big endian) with byte order marks.  A non-hexademical
+    representation is doable but this is preferable since it allows the output
+    Postscript file to be completely ASCII and no escaping of Postscript
+    characters is necessary.
+    '''
+    if s == '':
+        return ''
+    utf16_bytes = s.encode('utf-16be')
+    ascii_hex_bytes = codecs.encode(b'\xfe\xff' + utf16_bytes, 'hex')
+    ascii_hex_str = ascii_hex_bytes.decode('ascii').lower()
+    return ascii_hex_str
+
+
+def _get_pdfa_def(icc_profile, icc_identifier, pdfmark):
+    pdfmark_utf16 = {k: encode_text_string(v) for k, v in pdfmark.items()}
+
+    t = Template(pdfa_def_template)
+    result = t.substitute(icc_profile=icc_profile,
+                          icc_identifier=icc_identifier,
+                          title=pdfmark_utf16.get('title', ''),
+                          author=pdfmark_utf16.get('author', ''),
+                          subject=pdfmark_utf16.get('subject', ''),
+                          keywords=pdfmark_utf16.get('keywords', ''))
+    return result
+
+
+def _get_postscript_icc_path():
+    "Parse Ghostscript's help message to find where iccprofiles are stored"
+
+    p_gs = Popen(['gs', '--help'], close_fds=True, universal_newlines=True,
+                 stdout=PIPE, stderr=PIPE)
+    out, _ = p_gs.communicate()
+    lines = out.splitlines()
+
+    def search_paths(lines):
+        seeking = True
+        for line in lines:
+            if seeking:
+                if line.startswith('Search path'):
+                    seeking = False
+                    continue
+            else:
+                if line.strip().startswith('/'):
+                    yield from (
+                        path.strip() for path in line.split(':')
+                        if path.strip() != '')
+    for root in search_paths(lines):
+        path = os.path.realpath(os.path.join(root, '../iccprofiles'))
+        if os.path.exists(path):
+            return path
+
+
+def generate_pdfa_def(target_filename, pdfmark, icc='sRGB'):
+    if icc == 'sRGB':
+        icc_profile = os.path.join(_get_postscript_icc_path(), 'srgb.icc')
+    else:
+        raise NotImplementedError("Only supporting sRGB")
+
+    ps = _get_pdfa_def(icc_profile, icc, pdfmark)
+
+    # Since PostScript might not handle UTF-8 (it's hard to get a clear
+    # answer), insist on ascii
+    with open(target_filename, 'w', encoding='ascii') as f:
+        f.write(ps)

ocrmypdf/tesseract.py

@@ -0,0 +1,69 @@
+#!/usr/bin/env python3
+# © 2015 James R. Barlow: github.com/jbarlow83
+
+from subprocess import STDOUT, CalledProcessError, check_output
+import sys
+import os
+import re
+from functools import lru_cache
+from . import ExitCode
+
+
+@lru_cache(maxsize=1)
+def version():
+    args_tess = [
+        'tesseract',
+        '--version'
+    ]
+    try:
+        versions = check_output(
+                args_tess, close_fds=True, universal_newlines=True,
+                stderr=STDOUT)
+    except CalledProcessError:
+        print("Could not find Tesseract executable on system PATH.")
+        sys.exit(ExitCode.missing_dependency)
+
+    tesseract_version = re.match(r'tesseract\s(.+)', versions).group(1)
+    return tesseract_version
+
+
+@lru_cache(maxsize=1)
+def languages():
+    args_tess = [
+        'tesseract',
+        '--list-langs'
+    ]
+    try:
+        langs = check_output(
+                args_tess, close_fds=True, universal_newlines=True,
+                stderr=STDOUT)
+    except CalledProcessError as e:
+        print("Tesseract failed to report available languages.")
+        print("Output from Tesseract:")
+        print("-" * 40)
+        print(e.output)
+        sys.exit(ExitCode.missing_dependency)
+    return set(lang.strip() for lang in langs.splitlines()[1:])
+
+
+HOCR_TEMPLATE = '''<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+    "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
+ <head>
+  <title></title>
+  <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+  <meta name='ocr-system' content='tesseract 3.02.02' />
+  <meta name='ocr-capabilities' content='ocr_page ocr_carea ocr_par ocr_line ocrx_word'/>
+ </head>
+ <body>
+  <div class='ocr_page' id='page_1' title='image "x.tif"; bbox 0 0 {0} {1}; ppageno 0'>
+   <div class='ocr_carea' id='block_1_1' title="bbox 0 1 {0} {1}">
+    <p class='ocr_par' dir='ltr' id='par_1' title="bbox 0 1 {0} {1}">
+     <span class='ocr_line' id='line_1' title="bbox 0 1 {0} {1}"><span class='ocrx_word' id='word_1' title="bbox 0 1 {0} {1}"> </span>
+     </span>
+    </p>
+   </div>
+  </div>
+ </body>
+</html>'''

ocrmypdf/test/test_pageinfo.py

@@ -0,0 +1,109 @@
+#!/usr/bin/env python3
+# © 2015 James R. Barlow: github.com/jbarlow83
+
+from ocrmypdf import pageinfo
+from reportlab.pdfgen.canvas import Canvas
+from PIL import Image
+from tempfile import NamedTemporaryFile
+from contextlib import suppress
+import os
+import shutil
+import pytest
+import img2pdf
+from pkg_resources import Requirement, resource_filename
+
+req = Requirement.parse('ocrmypdf')
+
+TEST_OUTPUT = os.environ.get(
+    'OCRMYPDF_TEST_OUTPUT',
+    default=os.path.join(os.path.dirname(__file__), 'output'))
+
+
+def setup_module():
+    with suppress(FileNotFoundError):
+        shutil.rmtree(TEST_OUTPUT)
+    with suppress(FileExistsError):
+        os.mkdir(TEST_OUTPUT)
+
+
+def test_single_page_text():
+    filename = os.path.join(TEST_OUTPUT, 'text.pdf')
+    pdf = Canvas(filename, pagesize=(8*72, 6*72))
+    text = pdf.beginText()
+    text.setFont('Helvetica', 12)
+    text.setTextOrigin(1*72, 3*72)
+    text.textLine("Methink'st thou art a general offence and every"
+                  " man should beat thee.")
+    pdf.drawText(text)
+    pdf.showPage()
+    pdf.save()
+
+    pdfinfo = pageinfo.pdf_get_all_pageinfo(filename)
+
+    assert len(pdfinfo) == 1
+    page = pdfinfo[0]
+
+    assert page['has_text']
+    assert len(page['images']) == 0
+
+
+def test_single_page_image():
+    filename = os.path.join(TEST_OUTPUT, 'image-mono.pdf')
+
+    with NamedTemporaryFile() as im_tmp:
+        im = Image.new('1', (8, 8), 0)
+        for n in range(8):
+            im.putpixel((n, n), 1)
+        im.save(im_tmp.name, format='PNG')
+
+        pdf_bytes = img2pdf.convert([im_tmp.name], dpi=8)
+        with open(filename, 'wb') as pdf:
+            pdf.write(pdf_bytes)
+
+    pdfinfo = pageinfo.pdf_get_all_pageinfo(filename)
+
+    assert len(pdfinfo) == 1
+    page = pdfinfo[0]
+
+    assert not page['has_text']
+    assert len(page['images']) == 1
+
+    pdfimage = page['images'][0]
+    assert pdfimage['width'] == 8
+    assert pdfimage['color'] == 'gray'
+
+    # While unexpected, this is correct
+    # PDF spec says /FlateDecode image must have /BitsPerComponent 8
+    # So mono images get upgraded to 8-bit
+    assert pdfimage['bpc'] == 8
+
+    # DPI in a 1"x1" is the image width
+    assert pdfimage['dpi_w'] == 8
+    assert pdfimage['dpi_h'] == 8
+
+
+def test_single_page_inline_image():
+    filename = os.path.join(TEST_OUTPUT, 'image-mono-inline.pdf')
+    pdf = Canvas(filename, pagesize=(8*72, 6*72))
+    with NamedTemporaryFile() as im_tmp:
+        im = Image.new('1', (8, 8), 0)
+        for n in range(8):
+            im.putpixel((n, n), 1)
+        im.save(im_tmp.name, format='PNG')
+        # Draw image in a 72x72 pt or 1"x1" area
+        pdf.drawInlineImage(im_tmp.name, 0, 0, width=72, height=72)
+        pdf.showPage()
+        pdf.save()
+
+    with pytest.raises(NotImplementedError):
+        pageinfo.pdf_get_all_pageinfo(filename)
+
+
+def test_jpeg():
+    filename = resource_filename(req, 'tests/resources/c02-22.pdf')
+
+    pdfinfo = pageinfo.pdf_get_all_pageinfo(filename)
+
+    pdfimage = pdfinfo[0]['images'][0]
+    assert pdfimage['enc'] == 'jpeg'
+

ocrmypdf/unpaper.py

@@ -0,0 +1,84 @@
+#!/usr/bin/env python3
+# © 2015 James R. Barlow: github.com/jbarlow83
+# unpaper documentation:
+# https://github.com/Flameeyes/unpaper/blob/master/doc/basic-concepts.md
+
+from subprocess import Popen, PIPE
+from tempfile import NamedTemporaryFile
+import sys
+import os
+from functools import lru_cache
+
+
+@lru_cache(maxsize=1)
+def version():
+    args_unpaper = [
+        'unpaper',
+        '--version'
+    ]
+    p_unpaper = Popen(args_unpaper, close_fds=True, universal_newlines=True,
+                      stdout=PIPE, stderr=PIPE)
+    version, _ = p_unpaper.communicate(timeout=5)
+
+    return version.strip()
+
+
+try:
+    from PIL import Image
+except ImportError:
+    print("Could not find Python3 imaging library", file=sys.stderr)
+    raise
+
+
+def run(input_file, output_file, dpi, log, mode_args):
+    args_unpaper = [
+        'unpaper',
+        '-v',
+        '--dpi', str(dpi)
+    ] + mode_args
+
+    SUFFIXES = {'1': '.pbm', 'L': '.pgm', 'RGB': '.ppm'}
+    suffix = ''
+
+    im = Image.open(input_file)
+    suffix = SUFFIXES[im.mode]
+    with NamedTemporaryFile(suffix=suffix) as input_pnm, \
+            NamedTemporaryFile(suffix=suffix, mode="r+b") as output_pnm:
+        im.save(input_pnm, format='PPM')
+        im.close()
+
+        os.unlink(output_pnm.name)
+
+        args_unpaper.extend([input_pnm.name, output_pnm.name])
+        p_unpaper = Popen(
+            args_unpaper, close_fds=True,
+            universal_newlines=True, stdout=PIPE, stderr=PIPE
+            )
+        out, err = p_unpaper.communicate()
+        log.debug(out)
+        log.debug(err)
+
+        Image.open(output_pnm.name).save(output_file)
+
+
+def deskew(input_file, output_file, dpi, log):
+    run(input_file, output_file, dpi, log, [
+        '--mask-scan-size', '100',  # don't blank out narrow columns
+        '--no-border-align',  # don't align visible content to borders
+        '--no-mask-center',   # don't center visible content within page
+        '--no-grayfilter',    # don't remove light gray areas
+        '--no-blackfilter',   # don't remove solid black areas
+        '--no-noisefilter',   # don't remove salt and pepper noise
+        '--no-blurfilter'     # don't remove blurry objects/debris
+    ])
+
+
+def clean(input_file, output_file, dpi, log):
+    run(input_file, output_file, dpi, log, [
+        '--mask-scan-size', '100',  # don't blank out narrow columns
+        '--no-border-align',  # don't align visible content to borders
+        '--no-mask-center',   # don't center visible content within page
+        '--no-grayfilter',    # don't remove light gray areas
+        '--no-blackfilter',   # don't remove solid black areas
+        '--no-deskew',        # don't deskew
+    ])

pipeline.svg

@@ -0,0 +1,235 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"
+ "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<!-- Generated by graphviz version 2.38.0 (20140413.2041)
+ -->
+<!-- Title: Pipeline: Pages: 1 -->
+<svg width="728pt" height="651pt"
+ viewBox="0.00 0.00 728.00 650.53" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
+<g id="graph0" class="graph" transform="scale(1 1) rotate(0) translate(4 646.53)">
+<title>Pipeline:</title>
+<polygon fill="white" stroke="none" points="-4,4 -4,-646.53 724,-646.53 724,4 -4,4"/>
+<g id="clust1" class="cluster"><title>clustertasks</title>
+<polygon fill="none" stroke="black" points="8,-8 8,-634.53 712,-634.53 712,-8 8,-8"/>
+<text text-anchor="middle" x="360" y="-606.53" font-family="Times,serif" font-size="30.00" fill="#ff3232">Pipeline:</text>
+</g>
+<!-- t0 -->
+<g id="node1" class="node"><title>t0</title>
+<polygon fill="#efa03b" stroke="#006000" points="481.791,-588.53 386.209,-588.53 382.209,-584.53 382.209,-552.53 477.791,-552.53 481.791,-556.53 481.791,-588.53"/>
+<polyline fill="none" stroke="#006000" points="477.791,-584.53 382.209,-584.53 "/>
+<polyline fill="none" stroke="#006000" points="477.791,-584.53 477.791,-552.53 "/>
+<polyline fill="none" stroke="#006000" points="477.791,-584.53 481.791,-588.53 "/>
+<text text-anchor="middle" x="432" y="-564.53" font-family="Times,serif" font-size="20.00" fill="#006000">repair_pdf</text>
+</g>
+<!-- t1 -->
+<g id="node2" class="node"><title>t1</title>
+<polygon fill="#efa03b" stroke="black" points="466.782,-509.564 374,-526.497 281.218,-509.564 281.304,-482.165 466.696,-482.165 466.782,-509.564"/>
+<polygon fill="none" stroke="black" points="470.799,-512.902 374,-530.569 277.201,-512.902 277.311,-478.159 470.689,-478.159 470.799,-512.902"/>
+<text text-anchor="middle" x="374" y="-495.991" font-family="Times,serif" font-size="20.00">split_pages</text>
+</g>
+<!-- t0&#45;&gt;t1 -->
+<g id="edge1" class="edge"><title>t0&#45;&gt;t1</title>
+<path fill="none" stroke="gray" d="M417.064,-552.394C412.296,-546.925 406.86,-540.689 401.493,-534.532"/>
+<polygon fill="gray" stroke="gray" points="403.996,-532.077 394.787,-526.838 398.719,-536.676 403.996,-532.077"/>
+</g>
+<!-- t10 -->
+<g id="node12" class="node"><title>t10</title>
+<polygon fill="#efa03b" stroke="#006000" points="704.338,-451.452 493.662,-451.452 489.662,-447.452 489.662,-415.452 700.338,-415.452 704.338,-419.452 704.338,-451.452"/>
+<polyline fill="none" stroke="#006000" points="700.338,-447.452 489.662,-447.452 "/>
+<polyline fill="none" stroke="#006000" points="700.338,-447.452 700.338,-415.452 "/>
+<polyline fill="none" stroke="#006000" points="700.338,-447.452 704.338,-451.452 "/>
+<text text-anchor="middle" x="597" y="-427.452" font-family="Times,serif" font-size="20.00" fill="#006000">generate_postscript_stub</text>
+</g>
+<!-- t0&#45;&gt;t10 -->
+<g id="edge16" class="edge"><title>t0&#45;&gt;t10</title>
+<path fill="none" stroke="gray" d="M453.048,-552.468C461.441,-545.654 471.185,-537.73 480,-530.53 510.227,-505.84 544.753,-477.467 568.421,-457.99"/>
+<polygon fill="gray" stroke="gray" points="570.818,-460.55 576.315,-451.493 566.369,-455.146 570.818,-460.55"/>
+</g>
+<!-- t2 -->
+<g id="node3" class="node"><title>t2</title>
+<polygon fill="#efa03b" stroke="black" points="451.555,-451.452 228.445,-451.452 224.445,-447.452 224.445,-415.452 447.555,-415.452 451.555,-419.452 451.555,-451.452"/>
+<polyline fill="none" stroke="black" points="447.555,-447.452 224.445,-447.452 "/>
+<polyline fill="none" stroke="black" points="447.555,-447.452 447.555,-415.452 "/>
+<polyline fill="none" stroke="black" points="447.555,-447.452 451.555,-451.452 "/>
+<text text-anchor="middle" x="338" y="-427.452" font-family="Times,serif" font-size="20.00">rasterize_with_ghostscript</text>
+</g>
+<!-- t1&#45;&gt;t2 -->
+<g id="edge2" class="edge"><title>t1&#45;&gt;t2</title>
+<path fill="none" stroke="#0044a0" d="M361.611,-478.092C358.514,-472.369 355.171,-466.19 352.003,-460.333"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="355.064,-458.636 347.227,-451.506 348.907,-461.967 355.064,-458.636"/>
+</g>
+<!-- t11 -->
+<g id="node10" class="node"><title>t11</title>
+<polygon fill="#efa03b" stroke="black" points="644.594,-393.452 551.406,-393.452 547.406,-389.452 547.406,-357.452 640.594,-357.452 644.594,-361.452 644.594,-393.452"/>
+<polyline fill="none" stroke="black" points="640.594,-389.452 547.406,-389.452 "/>
+<polyline fill="none" stroke="black" points="640.594,-389.452 640.594,-357.452 "/>
+<polyline fill="none" stroke="black" points="640.594,-389.452 644.594,-393.452 "/>
+<text text-anchor="middle" x="596" y="-369.452" font-family="Times,serif" font-size="20.00">skip_page</text>
+</g>
+<!-- t1&#45;&gt;t11 -->
+<g id="edge13" class="edge"><title>t1&#45;&gt;t11</title>
+<path fill="none" stroke="#0044a0" d="M425.139,-478.007C437.834,-470.731 450.731,-461.83 461,-451.452 473.874,-438.442 466.946,-427.178 481,-415.452 490.212,-407.766 513.965,-399.205 537.508,-392.067"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="538.592,-395.396 547.189,-389.203 536.607,-388.683 538.592,-395.396"/>
+</g>
+<!-- t9 -->
+<g id="node11" class="node"><title>t9</title>
+<polygon fill="#efa03b" stroke="black" points="272.496,-277.452 19.5039,-277.452 15.5039,-273.452 15.5039,-241.452 268.496,-241.452 272.496,-245.452 272.496,-277.452"/>
+<polyline fill="none" stroke="black" points="268.496,-273.452 15.5039,-273.452 "/>
+<polyline fill="none" stroke="black" points="268.496,-273.452 268.496,-241.452 "/>
+<polyline fill="none" stroke="black" points="268.496,-273.452 272.496,-277.452 "/>
+<text text-anchor="middle" x="144" y="-253.452" font-family="Times,serif" font-size="20.00">tesseract_ocr_and_render_pdf</text>
+</g>
+<!-- t1&#45;&gt;t9 -->
+<g id="edge15" class="edge"><title>t1&#45;&gt;t9</title>
+<path fill="none" stroke="#0044a0" d="M277.341,-486.044C254.708,-478.702 232.21,-467.759 215,-451.452 168.207,-407.114 152.056,-329.127 146.633,-287.855"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="150.084,-287.235 145.422,-277.721 143.133,-288.065 150.084,-287.235"/>
+</g>
+<!-- t3 -->
+<g id="node4" class="node"><title>t3</title>
+<polygon fill="#efa03b" stroke="black" points="458.999,-393.452 291.001,-393.452 287.001,-389.452 287.001,-357.452 454.999,-357.452 458.999,-361.452 458.999,-393.452"/>
+<polyline fill="none" stroke="black" points="454.999,-389.452 287.001,-389.452 "/>
+<polyline fill="none" stroke="black" points="454.999,-389.452 454.999,-357.452 "/>
+<polyline fill="none" stroke="black" points="454.999,-389.452 458.999,-393.452 "/>
+<text text-anchor="middle" x="373" y="-369.452" font-family="Times,serif" font-size="20.00">preprocess_deskew</text>
+</g>
+<!-- t2&#45;&gt;t3 -->
+<g id="edge3" class="edge"><title>t2&#45;&gt;t3</title>
+<path fill="none" stroke="#0044a0" d="M348.691,-415.346C351.337,-411.112 354.23,-406.485 357.066,-401.946"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="360.042,-403.788 362.374,-393.453 354.106,-400.078 360.042,-403.788"/>
+</g>
+<!-- t6 -->
+<g id="node6" class="node"><title>t6</title>
+<polygon fill="#efa03b" stroke="black" points="664.375,-277.452 477.625,-277.452 473.625,-273.452 473.625,-241.452 660.375,-241.452 664.375,-245.452 664.375,-277.452"/>
+<polyline fill="none" stroke="black" points="660.375,-273.452 473.625,-273.452 "/>
+<polyline fill="none" stroke="black" points="660.375,-273.452 660.375,-241.452 "/>
+<polyline fill="none" stroke="black" points="660.375,-273.452 664.375,-277.452 "/>
+<text text-anchor="middle" x="569" y="-253.452" font-family="Times,serif" font-size="20.00">select_image_for_pdf</text>
+</g>
+<!-- t2&#45;&gt;t6 -->
+<g id="edge7" class="edge"><title>t2&#45;&gt;t6</title>
+<path fill="none" stroke="#0044a0" d="M423.032,-415.374C438.812,-409.925 454.543,-402.785 468,-393.452 508.201,-365.571 539.287,-316.662 555.808,-286.558"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="559.024,-287.965 560.653,-277.496 552.851,-284.664 559.024,-287.965"/>
+</g>
+<!-- t4 -->
+<g id="node5" class="node"><title>t4</title>
+<polygon fill="#efa03b" stroke="black" points="449.705,-335.452 300.295,-335.452 296.295,-331.452 296.295,-299.452 445.705,-299.452 449.705,-303.452 449.705,-335.452"/>
+<polyline fill="none" stroke="black" points="445.705,-331.452 296.295,-331.452 "/>
+<polyline fill="none" stroke="black" points="445.705,-331.452 445.705,-299.452 "/>
+<polyline fill="none" stroke="black" points="445.705,-331.452 449.705,-335.452 "/>
+<text text-anchor="middle" x="373" y="-311.452" font-family="Times,serif" font-size="20.00">preprocess_clean</text>
+</g>
+<!-- t3&#45;&gt;t4 -->
+<g id="edge4" class="edge"><title>t3&#45;&gt;t4</title>
+<path fill="none" stroke="#0044a0" d="M373,-357.346C373,-353.655 373,-349.665 373,-345.695"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="376.5,-345.453 373,-335.453 369.5,-345.453 376.5,-345.453"/>
+</g>
+<!-- t3&#45;&gt;t6 -->
+<g id="edge6" class="edge"><title>t3&#45;&gt;t6</title>
+<path fill="none" stroke="#0044a0" d="M414.399,-357.412C428.785,-351.039 444.852,-343.405 459,-335.452 466.909,-331.007 506.49,-303.805 535.921,-283.431"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="538.073,-286.198 544.299,-277.625 534.086,-280.444 538.073,-286.198"/>
+</g>
+<!-- t4&#45;&gt;t6 -->
+<g id="edge5" class="edge"><title>t4&#45;&gt;t6</title>
+<path fill="none" stroke="#0044a0" d="M432.605,-299.422C453.701,-293.395 477.617,-286.561 499.465,-280.319"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="500.687,-283.61 509.341,-277.497 498.764,-276.879 500.687,-283.61"/>
+</g>
+<!-- t5 -->
+<g id="node7" class="node"><title>t5</title>
+<polygon fill="#efa03b" stroke="black" points="455.922,-277.452 294.078,-277.452 290.078,-273.452 290.078,-241.452 451.922,-241.452 455.922,-245.452 455.922,-277.452"/>
+<polyline fill="none" stroke="black" points="451.922,-273.452 290.078,-273.452 "/>
+<polyline fill="none" stroke="black" points="451.922,-273.452 451.922,-241.452 "/>
+<polyline fill="none" stroke="black" points="451.922,-273.452 455.922,-277.452 "/>
+<text text-anchor="middle" x="373" y="-253.452" font-family="Times,serif" font-size="20.00">ocr_tesseract_hocr</text>
+</g>
+<!-- t4&#45;&gt;t5 -->
+<g id="edge8" class="edge"><title>t4&#45;&gt;t5</title>
+<path fill="none" stroke="#0044a0" d="M373,-299.346C373,-295.655 373,-291.665 373,-287.695"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="376.5,-287.453 373,-277.453 369.5,-287.453 376.5,-287.453"/>
+</g>
+<!-- t4&#45;&gt;t9 -->
+<g id="edge14" class="edge"><title>t4&#45;&gt;t9</title>
+<path fill="none" stroke="#0044a0" d="M303.36,-299.422C278.157,-293.259 249.509,-286.253 223.521,-279.898"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="224.249,-276.473 213.704,-277.497 222.586,-283.273 224.249,-276.473"/>
+</g>
+<!-- t7 -->
+<g id="node8" class="node"><title>t7</title>
+<polygon fill="#efa03b" stroke="black" points="426.366,-219.452 269.634,-219.452 265.634,-215.452 265.634,-183.452 422.366,-183.452 426.366,-187.452 426.366,-219.452"/>
+<polyline fill="none" stroke="black" points="422.366,-215.452 265.634,-215.452 "/>
+<polyline fill="none" stroke="black" points="422.366,-215.452 422.366,-183.452 "/>
+<polyline fill="none" stroke="black" points="422.366,-215.452 426.366,-219.452 "/>
+<text text-anchor="middle" x="346" y="-195.452" font-family="Times,serif" font-size="20.00">render_hocr_page</text>
+</g>
+<!-- t6&#45;&gt;t7 -->
+<g id="edge9" class="edge"><title>t6&#45;&gt;t7</title>
+<path fill="none" stroke="#0044a0" d="M501.184,-241.422C476.75,-235.286 448.99,-228.315 423.772,-221.982"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="424.429,-218.538 413.877,-219.497 422.724,-225.328 424.429,-218.538"/>
+</g>
+<!-- t8 -->
+<g id="node9" class="node"><title>t8</title>
+<polygon fill="#efa03b" stroke="black" points="663.742,-219.452 448.258,-219.452 444.258,-215.452 444.258,-183.452 659.742,-183.452 663.742,-187.452 663.742,-219.452"/>
+<polyline fill="none" stroke="black" points="659.742,-215.452 444.258,-215.452 "/>
+<polyline fill="none" stroke="black" points="659.742,-215.452 659.742,-183.452 "/>
+<polyline fill="none" stroke="black" points="659.742,-215.452 663.742,-219.452 "/>
+<text text-anchor="middle" x="554" y="-195.452" font-family="Times,serif" font-size="20.00">render_hocr_debug_page</text>
+</g>
+<!-- t6&#45;&gt;t8 -->
+<g id="edge11" class="edge"><title>t6&#45;&gt;t8</title>
+<path fill="none" stroke="#0044a0" d="M564.418,-241.346C563.4,-237.546 562.298,-233.43 561.203,-229.345"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="564.522,-228.207 558.554,-219.453 557.761,-230.018 564.522,-228.207"/>
+</g>
+<!-- t5&#45;&gt;t7 -->
+<g id="edge10" class="edge"><title>t5&#45;&gt;t7</title>
+<path fill="none" stroke="#0044a0" d="M364.752,-241.346C362.816,-237.329 360.708,-232.958 358.629,-228.645"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="361.693,-226.941 354.197,-219.453 355.387,-229.981 361.693,-226.941"/>
+</g>
+<!-- t5&#45;&gt;t8 -->
+<g id="edge12" class="edge"><title>t5&#45;&gt;t8</title>
+<path fill="none" stroke="#0044a0" d="M428.289,-241.346C447.587,-235.375 469.416,-228.622 489.404,-222.437"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="490.53,-225.753 499.049,-219.453 488.461,-219.065 490.53,-225.753"/>
+</g>
+<!-- t12 -->
+<g id="node13" class="node"><title>t12</title>
+<polygon fill="#efa03b" stroke="black" points="473.845,-105.456 554,-78.0208 634.155,-105.456 634.08,-149.848 473.92,-149.848 473.845,-105.456"/>
+<polygon fill="none" stroke="black" points="469.836,-102.581 554,-73.7729 638.164,-102.581 638.078,-153.869 469.922,-153.869 469.836,-102.581"/>
+<text text-anchor="middle" x="554" y="-111.726" font-family="Times,serif" font-size="20.00">merge_pages</text>
+</g>
+<!-- t7&#45;&gt;t12 -->
+<g id="edge20" class="edge"><title>t7&#45;&gt;t12</title>
+<path fill="none" stroke="#0044a0" d="M389.35,-183.419C409.937,-175.33 435.398,-165.326 460.092,-155.624"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="461.479,-158.839 469.507,-151.924 458.919,-152.324 461.479,-158.839"/>
+</g>
+<!-- t8&#45;&gt;t12 -->
+<g id="edge19" class="edge"><title>t8&#45;&gt;t12</title>
+<path fill="none" stroke="#0044a0" d="M554,-183.12C554,-177.585 554,-171.177 554,-164.592"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="557.5,-164.201 554,-154.201 550.5,-164.201 557.5,-164.201"/>
+</g>
+<!-- t11&#45;&gt;t12 -->
+<g id="edge17" class="edge"><title>t11&#45;&gt;t12</title>
+<path fill="none" stroke="#0044a0" d="M615.939,-357.13C634.767,-339.333 661.745,-309.733 673,-277.452 686.754,-238.003 694.347,-219.364 673,-183.452 666.345,-172.256 656.999,-162.875 646.44,-155.047"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="648.378,-152.132 638.152,-149.36 644.417,-157.904 648.378,-152.132"/>
+</g>
+<!-- t9&#45;&gt;t12 -->
+<g id="edge18" class="edge"><title>t9&#45;&gt;t12</title>
+<path fill="none" stroke="#0044a0" d="M164.711,-241.235C186.394,-224.08 222.088,-198.224 257,-183.452 321.895,-155.994 399.893,-139.571 459.67,-130.141"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="460.504,-133.554 469.855,-128.574 459.439,-126.635 460.504,-133.554"/>
+</g>
+<!-- t10&#45;&gt;t12 -->
+<g id="edge22" class="edge"><title>t10&#45;&gt;t12</title>
+<path fill="none" stroke="gray" d="M629.379,-415.337C660.552,-396.255 703,-362.233 703,-318.452 703,-318.452 703,-318.452 703,-258.452 703,-224.066 705.581,-210.223 684,-183.452 673.976,-171.017 660.952,-160.773 647.026,-152.396"/>
+<polygon fill="gray" stroke="gray" points="648.682,-149.312 638.258,-147.419 645.227,-155.4 648.682,-149.312"/>
+</g>
+<!-- t13 -->
+<g id="node14" class="node"><title>t13</title>
+<polygon fill="#efa03b" stroke="black" points="616.338,-52 495.662,-52 491.662,-48 491.662,-16 612.338,-16 616.338,-20 616.338,-52"/>
+<polyline fill="none" stroke="black" points="612.338,-48 491.662,-48 "/>
+<polyline fill="none" stroke="black" points="612.338,-48 612.338,-16 "/>
+<polyline fill="none" stroke="black" points="612.338,-48 616.338,-52 "/>
+<text text-anchor="middle" x="554" y="-28" font-family="Times,serif" font-size="20.00">validate_pdfa</text>
+</g>
+<!-- t12&#45;&gt;t13 -->
+<g id="edge21" class="edge"><title>t12&#45;&gt;t13</title>
+<path fill="none" stroke="#0044a0" d="M554,-73.9482C554,-69.9654 554,-66.007 554,-62.2247"/>
+<polygon fill="#0044a0" stroke="#0044a0" points="557.5,-62.1573 554,-52.1573 550.5,-62.1574 557.5,-62.1573"/>
+</g>
+</g>
+</svg>

requirements.txt

@@ -0,0 +1,4 @@
+ruffus>=2.6.3
+Pillow>=2.4.0
+reportlab>=3.1.44
+PyPDF2>=1.25.1

setup.cfg

@@ -0,0 +1,2 @@
+[bdist_wheel]
+python-tag = py34

setup.py

@@ -0,0 +1,214 @@
+#!/usr/bin/env python3
+# © 2015 James R. Barlow: github.com/jbarlow83
+
+from setuptools import setup
+from subprocess import Popen, STDOUT, check_output, CalledProcessError
+from string import Template
+from collections.abc import Mapping
+import re
+import sys
+
+
+missing_program = '''
+The program '{program}' could not be executed or was not found on your
+system PATH.
+'''
+
+unknown_version = '''
+OCRmyPDF requires '{program}' {need_version} or higher.  Your system has
+'{program}' but we cannot tell what version is installed.  Contact the
+package maintainer.
+'''
+
+old_version = '''
+OCRmyPDF requires '{program}' {need_version} or higher.  Your system appears
+to have {found_version}.  Please update this program.
+'''
+
+okay_its_optional = '''
+This program is OPTIONAL, so installation of OCRmyPDF can proceed, but
+some functionality may be missing.
+'''
+
+not_okay_its_required = '''
+This program is REQUIRED for OCRmyPDF to work.  Installation will abort.
+'''
+
+osx_install_advice = '''
+If you have homebrew installed, try these command to install the missing
+packages:
+    brew update
+    brew upgrade
+    brew install {package}
+'''
+
+linux_install_advice = '''
+On systems with the aptitude package manager (Debian, Ubuntu), try these
+commands:
+    sudo apt-get update
+    sudo apt-get install {package}
+
+On RPM-based systems (Red Hat, Fedora), search for instructions on
+installing the RPM for {program}.
+'''
+
+
+def get_platform():
+    if sys.platform.startswith('freebsd'):
+        return 'freebsd'
+    elif sys.platform.startswith('linux'):
+        return 'linux'
+    return sys.platform
+
+
+def _error_trailer(program, package, optional, **kwargs):
+    if optional:
+        print(okay_its_optional.format(**locals()), file=sys.stderr)
+    else:
+        print(not_okay_its_required.format(**locals()), file=sys.stderr)
+
+    if isinstance(package, Mapping):
+        package = package[get_platform()]
+
+    if get_platform() == 'darwin':
+        print(osx_install_advice.format(**locals()), file=sys.stderr)
+    elif get_platform() == 'linux':
+        print(linux_install_advice.format(**locals()), file=sys.stderr)
+
+
+def error_missing_program(
+        program,
+        package,
+        optional
+        ):
+    print(missing_program.format(**locals()), file=sys.stderr)
+    _error_trailer(**locals())
+
+
+def error_unknown_version(
+        program,
+        package,
+        optional,
+        need_version
+        ):
+    print(unknown_version.format(**locals()), file=sys.stderr)
+    _error_trailer(**locals())
+
+
+def error_old_version(
+        program,
+        package,
+        optional,
+        need_version,
+        found_version
+        ):
+    print(old_version.format(**locals()), file=sys.stderr)
+    _error_trailer(**locals())
+
+
+def check_external_program(
+        program,
+        need_version,
+        package,
+        version_check_args=['--version'],
+        version_scrape_regex=re.compile(r'(\d+\.\d+(?:\.\d+)?)'),
+        optional=False):
+
+    print('Checking for {program} >= {need_version}...'.format(
+            program=program, need_version=need_version))
+    try:
+        result = check_output(
+                [program] + version_check_args,
+                universal_newlines=True, stderr=STDOUT)
+    except (CalledProcessError, FileNotFoundError):
+        error_missing_program(program, package, optional)
+        if not optional:
+            sys.exit(1)
+        print('Continuing install without {program}'.format(program=program))
+        return
+
+    try:
+        found_version = version_scrape_regex.search(result).group(1)
+    except AttributeError:
+        error_unknown_version(program, package, optional, need_version)
+        sys.exit(1)
+
+    if found_version < need_version:
+        error_old_version(program, package, optional, need_version,
+                          found_version)
+
+    print('Found {program} {found_version}'.format(
+            program=program, found_version=found_version))
+
+
+command = next((arg for arg in sys.argv[1:] if not arg.startswith('-')), '')
+
+
+if command.startswith('install') or \
+        command in ['check', 'test', 'nosetests', 'easy_install', 'egg_info']:
+    check_external_program(
+        program='tesseract',
+        need_version='3.02.02',
+        package={'darwin': 'tesseract', 'linux': 'tesseract-ocr'}
+    )
+    check_external_program(
+        program='gs',
+        need_version='9.14',
+        package='ghostscript'
+    )
+    check_external_program(
+        program='unpaper',
+        need_version='6.1',
+        package='unpaper',
+        optional=True
+    )
+    check_external_program(
+        program='qpdf',
+        need_version='5.0.0',
+        package='qpdf',
+        version_check_args=['--version']
+    )
+
+
+if 'upload' in sys.argv[1:]:
+    print('Use twine to upload the package - setup.py upload is insecure')
+    sys.exit(1)
+
+install_requires = open('requirements.txt').read().splitlines()
+tests_require = open('test_requirements.txt').read().splitlines()
+
+setup(
+    name='ocrmypdf',
+    version='3.0rc6',  # also update: release notes, main.py
+    description='OCRmyPDF adds an OCR text layer to scanned PDF files, allowing them to be searched',
+    url='https://github.com/fritz-hh/OCRmyPDF',
+    author='James. R. Barlow',
+    author_email='jim@purplerock.ca',
+    license='Public Domain',
+    packages=['ocrmypdf'],
+    keywords=['PDF', 'OCR', 'optical character recognition', 'PDF/A', 'scanning'],
+    classifiers=[
+        "Programming Language :: Python :: 3",
+        "Development Status :: 4 - Beta",
+        "Environment :: Console",
+        "Intended Audience :: End Users/Desktop",
+        "Intended Audience :: Science/Research",
+        "Intended Audience :: System Administrators",
+        "License :: Public Domain",
+        "Operating System :: MacOS :: MacOS X",
+        "Operating System :: POSIX",
+        "Operating System :: POSIX :: BSD",
+        "Operating System :: POSIX :: Linux",
+        "Topic :: Scientific/Engineering :: Image Recognition",
+        "Topic :: Text Processing :: Indexing",
+        "Topic :: Text Processing :: Linguistic",
+        ],
+    install_requires=install_requires,
+    tests_require=tests_require,
+    entry_points={
+        'console_scripts': [
+            'ocrmypdf = ocrmypdf.main:run_pipeline'
+        ],
+    },
+    include_package_data=True,
+    zip_safe=False)

test/test-appimage.sh

@@ -1,53 +0,0 @@
-#! /bin/bash
-
-set -x
-set -e
-
-chmod +x OCRmyPDF*.AppImage
-
-# run OCRmyPDF to test if the AppImage can ocr a test file
-run_appimage()
-{
-    echo ""
-    ./OCRmyPDF*.AppImage --help
-    echo ""
-    ./OCRmyPDF*.AppImage --list-programs
-    echo ""
-    ./OCRmyPDF*.AppImage --list-licenses
-    echo ""
-    ./OCRmyPDF*.AppImage ocrmypdf -l deu -s -d --jbig2-lossy --optimize 1 "$TRAVIS_BUILD_DIR"/test/test.pdf output.pdf
-    echo ""
-}
-
-
-# check AppImage for common issues
-run_appimagelint()
-{
-    wget https://github.com/TheAssassin/appimagelint/releases/download/continuous/appimagelint-x86_64.AppImage
-    chmod +x appimagelint-x86_64.AppImage
-    ./appimagelint-x86_64.AppImage OCRmyPDF*.AppImage
-}
-
-
-# extract the OCRmyPDF AppImage, install pytest & test requirements and run pytest
-run_pytest()
-{
-    git clone --depth=1 --branch "v$OCRMYPDF_VERSION" https://github.com/jbarlow83/OCRmyPDF.git
-    ./OCRmyPDF*.AppImage --appimage-extract
-
-    pushd squashfs-root
-    ./AppRun python3 -m pip install pytest
-    ./AppRun python3 -m pip install -r ../OCRmyPDF/requirements/test.txt
-    ./AppRun python3 -m pytest ../OCRmyPDF -n auto
-    popd
-}
-
-
-run_appimage
-
-run_appimagelint
-
-# run_pytest
-
-
-

test_requirements.txt

@@ -0,0 +1,2 @@
+img2pdf>=0.1.5
+pytest>=2.7.2

tests/replace_ghostscript_nopdfa.py

@@ -0,0 +1,72 @@
+#!/usr/bin/env python3
+import sys
+import os
+
+'''
+For testing PDF/A generation failures, this wrapper calls ghostscript
+but filters out all PDF/A information so that a regular PDF will be
+created instead.
+
+It assumes that it is called from a staged system PATH where the first
+item on the PATH contains a file named 'gs' which is a symlink to this
+file.  It will strip out the first item on path to invoke the real
+'gs'.  That is, it expects this when called
+
+1.   tests/output/[...test name...]/bin/gs is a symlink to this file
+2.   tests/output/bin is the first item on PATH
+3.   The real executable is on the path
+
+OCRmyPDF also calls "gs --version" and "gs --help".  gs answers both
+on stdout so, this wrapper prints to stderr.
+
+'''
+
+
+def pdfa_param(arg):
+    if arg.startswith('-sPDFA'):
+        return True
+    if arg.startswith('-dPDFA'):
+        return True
+    if arg.endswith('.ps'):
+        return True
+    if 'ColorConversionStrategy' in arg:
+        return True
+    if 'ProcessColorModel' in arg:
+        return True
+    if 'OutputICCProfile' in arg:
+        return True
+    return False
+
+
+if __name__ == '__main__':
+    sys_args = sys.argv[1:]
+
+    print("Fake Ghostscript wrapper", file=sys.stderr)
+
+    if any(pdfa_param(arg) for arg in sys_args):
+        # We were asked to produce a PDF/A
+        # Filter out PDF/A arguments
+        args = [arg for arg in sys.argv[1:]
+                if not pdfa_param(arg)]
+
+        # Tell Ghostscript to create a PDF 1.3 instead so JHOVE won't
+        # think it's a PDF/A
+        indexof_pdfwrite = next(n for n, item in enumerate(args)
+                                if 'pdfwrite' in item)
+        args.insert(indexof_pdfwrite + 1, '-dCompatibilityLevel=1.3')
+        print("Rewrote arguments", file=sys.stderr)
+    else:
+        args = sys_args
+        print("Keeping arguments", file=sys.stderr)
+
+    exec_path = os.environ['PATH'].split(os.pathsep)
+    env = os.environ.copy()
+    env['PATH'] = os.pathsep.join(exec_path[1:])
+
+    print("Calling real Ghostscript with modified args and env",
+          file=sys.stderr)
+    print(args, file=sys.stderr)
+    print(env['PATH'], file=sys.stderr)
+
+    sys.stderr.flush()
+    os.execvpe('gs', ['gs'] + args, env)

tests/resources/NOTE.rst

@@ -0,0 +1,32 @@
+All test resources must come from free public domain sources for
+copyright reasons.
+
+Test files do not necessarily produce perfect (or even good) OCR
+results.
+
++-------------------+--------------------------------------------------------------------------------+
+| File              | Source                                                                         |
++===================+================================================================================+
+| graph.pdf         | Wikimedia                                                                      |
++-------------------+--------------------------------------------------------------------------------+
+| c02-22.pdf        | Project Gutenberg: https://www.gutenberg.org/files/76/76-h/images/c02-22.jpg   |
++-------------------+--------------------------------------------------------------------------------+
+| LinnSequencer.jpg | Wikimedia_                                                                     |
++-------------------+--------------------------------------------------------------------------------+
+| congress.jpg      | http://www.baxleystamps.com/litho/meiji/courts_1871.jpg                        |
++-------------------+--------------------------------------------------------------------------------+
+| blank.pdf         | Blank page from Adobe Illustrator CC 2015                                      |
++-------------------+--------------------------------------------------------------------------------+
+| enormous.pdf      | PNG file saved to PDF using img2pdf                                            |
++-------------------+--------------------------------------------------------------------------------+
+| invalid.pdf       | PDF file header followed by EOF marker; not valid                              |
++-------------------+--------------------------------------------------------------------------------+
+| multipage.pdf     | several other files concatenated                                               |
++-------------------+--------------------------------------------------------------------------------+
+| skew.pdf          | skewed version of c02-22.PDF                                                   |
++-------------------+--------------------------------------------------------------------------------+
+| Test_Issue_28.pdf | file with some syntax errors                                                   |
++-------------------+--------------------------------------------------------------------------------+
+
+
+.. _Wikimedia: https://upload.wikimedia.org/wikipedia/en/b/b7/LinnSequencer_hardware_MIDI_sequencer_brochure_page_2_300dpi.jpg

tests/resources/invalid.pdf

@@ -0,0 +1,3 @@
+%PDF-1.3
+This is not a valid PDF file
+%%EOF

tests/test_main.py

@@ -0,0 +1,310 @@
+#!/usr/bin/env python3
+# © 2015 James R. Barlow: github.com/jbarlow83
+
+from __future__ import print_function
+from subprocess import Popen, PIPE, check_output
+import os
+import shutil
+from contextlib import suppress
+import sys
+import pytest
+from ocrmypdf.pageinfo import pdf_get_all_pageinfo
+import PyPDF2 as pypdf
+from ocrmypdf import ExitCode
+
+
+if sys.version_info.major < 3:
+    print("Requires Python 3.4+")
+    sys.exit(1)
+
+TESTS_ROOT = os.path.abspath(os.path.dirname(__file__))
+PROJECT_ROOT = os.path.dirname(TESTS_ROOT)
+OCRMYPDF = os.path.join(PROJECT_ROOT, 'OCRmyPDF.sh')
+TEST_RESOURCES = os.path.join(PROJECT_ROOT, 'tests', 'resources')
+TEST_OUTPUT = os.environ.get(
+    'OCRMYPDF_TEST_OUTPUT',
+    default=os.path.join(PROJECT_ROOT, 'tests', 'output'))
+TEST_BINARY_PATH = os.path.join(TEST_OUTPUT, 'fakebin')
+
+
+def setup_module():
+    with suppress(FileNotFoundError):
+        shutil.rmtree(TEST_OUTPUT)
+    with suppress(FileExistsError):
+        os.mkdir(TEST_OUTPUT)
+
+
+def run_ocrmypdf_sh(input_file, output_file, *args):
+    sh_args = ['sh', OCRMYPDF] + list(args) + [input_file, output_file]
+    sh = Popen(
+        sh_args, close_fds=True, stdout=PIPE, stderr=PIPE,
+        universal_newlines=True)
+    out, err = sh.communicate()
+    return sh, out, err
+
+
+def _make_input(input_basename):
+    return os.path.join(TEST_RESOURCES, input_basename)
+
+
+def _make_output(output_basename):
+    return os.path.join(TEST_OUTPUT, output_basename)
+
+
+def check_ocrmypdf(input_basename, output_basename, *args):
+    input_file = _make_input(input_basename)
+    output_file = _make_output(output_basename)
+
+    sh, _, err = run_ocrmypdf_sh(input_file, output_file, *args)
+    assert sh.returncode == 0, err
+    assert os.path.exists(output_file), "Output file not created"
+    assert os.stat(output_file).st_size > 100, "PDF too small or empty"
+    return output_file
+
+
+def run_ocrmypdf_env(input_basename, output_basename, *args, env=None):
+    input_file = _make_input(input_basename)
+    output_file = _make_output(output_basename)
+
+    if env is None:
+        env = os.environ
+
+    p_args = ['ocrmypdf'] + list(args) + [input_file, output_file]
+    p = Popen(
+        p_args, close_fds=True, stdout=PIPE, stderr=PIPE,
+        universal_newlines=True, env=env)
+    out, err = p.communicate()
+    return p, out, err
+
+
+def test_quick():
+    check_ocrmypdf('c02-22.pdf', 'test_quick.pdf')
+
+
+def test_deskew():
+    # Run with deskew
+    deskewed_pdf = check_ocrmypdf('skew.pdf', 'test_deskew.pdf', '-d')
+
+    # Now render as an image again and use Leptonica to find the skew angle
+    # to confirm that it was deskewed
+    from ocrmypdf.ghostscript import rasterize_pdf
+    import logging
+    log = logging.getLogger()
+
+    deskewed_png = _make_output('deskewed.png')
+
+    rasterize_pdf(
+        deskewed_pdf,
+        deskewed_png,
+        xres=150,
+        yres=150,
+        raster_device='pngmono',
+        log=log)
+
+    from ocrmypdf.leptonica import pixRead, pixDestroy, pixFindSkew
+    pix = pixRead(deskewed_png)
+    skew_angle, skew_confidence = pixFindSkew(pix)
+    pix = pixDestroy(pix)
+
+    print(skew_angle)
+    assert -0.5 < skew_angle < 0.5, "Deskewing failed"
+
+
+def test_clean():
+    check_ocrmypdf('skew.pdf', 'test_clean.pdf', '-c')
+
+
+def test_preserve_metadata():
+    pdf_before = pypdf.PdfFileReader(_make_input('graph.pdf'))
+
+    output = check_ocrmypdf('graph.pdf', 'test_metadata_preserve.pdf')
+
+    pdf_after = pypdf.PdfFileReader(output)
+
+    for key in ('/Title', '/Author'):
+        assert pdf_before.documentInfo[key] == pdf_after.documentInfo[key]
+
+
+def test_override_metadata():
+    input_file = _make_input('c02-22.pdf')
+    output_file = _make_output('test_override_metadata.pdf')
+
+    german = 'Du siehst den Wald vor lauter Bäumen nicht.'
+    chinese = '孔子'
+    high_unicode = 'U+1030C is: 𐌌'
+
+    p, out, err = run_ocrmypdf_env(
+        input_file, output_file,
+        '--title', german,
+        '--author', chinese,
+        '--subject', high_unicode)
+
+    if p.returncode == ExitCode.invalid_output_pdfa:
+        print("Got invalid PDF return code, as expected - JHOVE bug")
+    assert p.returncode in (ExitCode.ok, ExitCode.invalid_output_pdfa)
+
+    pdf = output_file
+
+    out_pdfinfo = check_output(['pdfinfo', pdf], universal_newlines=True)
+    lines_pdfinfo = out_pdfinfo.splitlines()
+    pdfinfo = {}
+    for line in lines_pdfinfo:
+        k, v = line.strip().split(':', maxsplit=1)
+        pdfinfo[k.strip()] = v.strip()
+
+    assert pdfinfo['Title'] == german
+    assert pdfinfo['Author'] == chinese
+    assert pdfinfo['Subject'] == high_unicode
+    assert pdfinfo.get('Keywords', '') == ''
+
+
+def check_oversample(renderer):
+    oversampled_pdf = check_ocrmypdf(
+        'skew.pdf', 'test_oversample_%s.pdf' % renderer, '--oversample', '300',
+        '--pdf-renderer', renderer)
+
+    pdfinfo = pdf_get_all_pageinfo(oversampled_pdf)
+
+    print(pdfinfo[0]['xres'])
+    assert abs(pdfinfo[0]['xres'] - 300) < 1
+
+
+def test_oversample():
+    yield check_oversample, 'hocr'
+    yield check_oversample, 'tesseract'
+
+
+def test_repeat_ocr():
+    sh, _, _ = run_ocrmypdf_sh('graph_ocred.pdf', 'wontwork.pdf')
+    assert sh.returncode != 0
+
+
+def test_force_ocr():
+    out = check_ocrmypdf('graph_ocred.pdf', 'test_force.pdf', '-f')
+    pdfinfo = pdf_get_all_pageinfo(out)
+    assert pdfinfo[0]['has_text']
+
+
+def test_skip_ocr():
+    check_ocrmypdf('graph_ocred.pdf', 'test_skip.pdf', '-s')
+
+
+def test_argsfile():
+    with open(_make_output('test_argsfile.txt'), 'w') as argsfile:
+        print('--title', 'ArgsFile Test', '--author', 'Test Cases',
+              sep='\n', end='\n', file=argsfile)
+    check_ocrmypdf('graph.pdf', 'test_argsfile.pdf',
+                   '@' + _make_output('test_argsfile.txt'))
+
+
+def check_ocr_timeout(renderer):
+    out = check_ocrmypdf('skew.pdf', 'test_timeout_%s.pdf' % renderer,
+                         '--tesseract-timeout', '1.0')
+    pdfinfo = pdf_get_all_pageinfo(out)
+    assert pdfinfo[0]['has_text'] == False
+
+
+def test_ocr_timeout():
+    yield check_ocr_timeout, 'hocr'
+    yield check_ocr_timeout, 'tesseract'
+
+
+def test_skip_big():
+    out = check_ocrmypdf('enormous.pdf', 'test_enormous.pdf',
+                         '--skip-big', '10')
+    pdfinfo = pdf_get_all_pageinfo(out)
+    assert pdfinfo[0]['has_text'] == False
+
+
+def check_maximum_options(renderer):
+    check_ocrmypdf(
+        'multipage.pdf', 'test_multipage%s.pdf' % renderer,
+        '-d', '-c', '-i', '-g', '-f', '-k', '--oversample', '300',
+        '--skip-big', '10', '--title', 'Too Many Weird Files',
+        '--author', 'py.test', '--pdf-renderer', renderer)
+
+
+def test_maximum_options():
+    yield check_maximum_options, 'hocr'
+    yield check_maximum_options, 'tesseract'
+
+
+def override_binary(binary, replacement):
+    '''Create a directory that contains a symlink named 'binary' that
+    points to replacement, another program to use in place of the
+    regular binary for testing.
+
+    override_binary('gs', 'replace_gs.py') will create an environment
+    in which "gs" will invoke replace_gs.py.
+
+    Not thread-safe with other test at the moment.
+
+    Returns the os.environ["PATH"] string under which this binary will
+    be invoked.'''
+
+    replacement_path = os.path.abspath(os.path.join(TESTS_ROOT,
+                                                    replacement))
+    subdir = os.path.splitext(os.path.basename(replacement))[0]
+    binary_path = os.path.abspath(os.path.join(TEST_BINARY_PATH,
+                                               subdir,
+                                               binary))
+    with suppress(FileExistsError):
+        os.makedirs(os.path.dirname(binary_path))
+    assert os.path.isdir(os.path.dirname(binary_path))
+    assert not os.path.lexists(binary_path)
+    print("symlink %s -> %s" % (replacement_path, binary_path))
+    os.symlink(replacement_path, binary_path)
+
+    os.chmod(replacement_path, int('755', base=8))
+
+    return os.path.dirname(binary_path) + os.pathsep + os.environ["PATH"]
+
+
+@pytest.fixture
+def break_ghostscript_pdfa():
+    return override_binary('gs', 'replace_ghostscript_nopdfa.py')
+
+
+@pytest.mark.skipif(os.environ.get('OCRMYPDF_IN_DOCKER', False),
+                    reason="Requires writable filesystem")
+def test_ghostscript_pdfa_fails(break_ghostscript_pdfa):
+    env = os.environ.copy()
+    env['PATH'] = break_ghostscript_pdfa
+
+    p, out, err = run_ocrmypdf_env(
+        'graph_ocred.pdf', 'not_a_pdfa.pdf', '-v', '1', '--skip-text', env=env)
+    assert p.returncode == ExitCode.ok, err  # no longer using JHOVE PDFA check
+
+
+def test_tesseract_missing_tessdata():
+    env = os.environ.copy()
+    env['TESSDATA_PREFIX'] = '/tmp'
+
+    p, _, err = run_ocrmypdf_env(
+        'graph_ocred.pdf', 'not_a_pdfa.pdf', '-v', '1', '--skip-text', env=env)
+    assert p.returncode == ExitCode.missing_dependency, err
+
+
+def test_invalid_input_pdf():
+    p, out, err = run_ocrmypdf_env(
+        'invalid.pdf', 'wont_be_created.pdf')
+    assert p.returncode == ExitCode.input_file, err
+
+
+def test_blank_input_pdf():
+    p, out, err = run_ocrmypdf_env(
+        'blank.pdf', 'still_blank.pdf')
+    assert p.returncode == ExitCode.ok
+
+
+def test_french():
+    p, out, err = run_ocrmypdf_env(
+        'francais.pdf', 'francais.pdf', '-l', 'fra')
+    assert p.returncode == ExitCode.ok, \
+        "This test may fail if Tesseract language packs are missing"
+
+
+def test_klingon():
+    p, out, err = run_ocrmypdf_env(
+        'francais.pdf', 'francais.pdf', '-l', 'klz')
+    assert p.returncode == ExitCode.bad_args