Merge pull request #65775 from Kiwi/ocrmypdf

ocrmypdf: init

pkgs/development/python-modules/hocr-tools/default.nix

@@ -0,0 +1,31 @@
+{ buildPythonPackage
+, fetchFromGitHub
+, lxml
+, pillow
+, reportlab
+, stdenv
+}:
+buildPythonPackage rec {
+  pname = "hocr-tools";
+  version = "1.3.0";
+
+  src = fetchFromGitHub {
+    owner = "tmbdev";
+    repo = "${pname}";
+    rev = "v${version}";
+    sha256 = "14f9hkp7pr677085w8iidwd0la9cjzy3pyj3rdg9b03nz9pc0w6p";
+  };
+
+  # hocr-tools uses a test framework that requires internet access
+  doCheck = false;
+
+  propagatedBuildInputs = [ pillow lxml reportlab ];
+
+  meta = with stdenv.lib; {
+    description = "
+Tools for manipulating and evaluating the hOCR format for representing multi-lingual OCR results by embedding them into HTML";
+    homepage = https://github.com/tmbdev/hocr-tools;
+    license = licenses.asl20;
+    maintainers = [ maintainers.kiwi ];
+  };
+}

pkgs/development/python-modules/pikepdf/default.nix

@@ -0,0 +1,73 @@
+{ attrs
+, buildPythonPackage
+, defusedxml
+, fetchPypi
+, hypothesis
+, isPy3k
+, lxml
+, pillow
+, pybind11
+, pytest
+, pytest-helpers-namespace
+, pytest-timeout
+, pytest_xdist
+, pytestrunner
+, python-xmp-toolkit
+, python3
+, qpdf
+, setuptools-scm-git-archive
+, setuptools_scm
+, stdenv
+}:
+
+buildPythonPackage rec {
+  pname = "pikepdf";
+  version = "1.1.0";
+  disabled = ! isPy3k;
+
+  src = fetchPypi {
+    inherit pname version;
+    sha256 = "14b36r6h3088z2sxp2pqvm171js53hz53mwm1g52iadignjnp0my";
+  };
+
+  buildInputs = [
+    pybind11
+    qpdf
+  ];
+
+  nativeBuildInputs = [
+    setuptools-scm-git-archive
+    setuptools_scm
+  ];
+
+  checkInputs = [
+    attrs
+    hypothesis
+    pillow
+    pytest
+    pytest-helpers-namespace
+    pytest-timeout
+    pytest_xdist
+    pytestrunner
+    python-xmp-toolkit
+  ];
+
+  propagatedBuildInputs = [ defusedxml lxml ];
+
+  postPatch = ''
+    substituteInPlace requirements/test.txt \
+       --replace "pytest >= 3.6.0, < 4.1.0" "pytest >= 4.2.1, < 5"
+  '';
+
+  preBuild = ''
+    HOME=$TMPDIR
+  '';
+
+  meta = with stdenv.lib; {
+    homepage = "https://github.com/pikepdf/pikepdf";
+    description = "Read and write PDFs with Python, powered by qpdf";
+    license = licenses.mpl20;
+    maintainers = [ maintainers.kiwi ];
+  };
+}
+

pkgs/development/python-modules/pytest-helpers-namespace/default.nix

@@ -0,0 +1,35 @@
+{ buildPythonPackage
+, fetchFromGitHub
+, pytest
+, stdenv
+}:
+
+buildPythonPackage rec {
+  pname = "pytest-helpers-namespace";
+  version = "2019.1.8";
+
+  src = fetchFromGitHub {
+    owner = "saltstack";
+    repo = "${pname}";
+    rev = "v${version}";
+    sha256 = "0z9f25d2wpf3lnqzmmnrlvl5b1f7kqwjjf4nzs9x2bpf91s5zny1";
+  };
+
+  buildInputs = [ pytest ];
+
+  checkInputs = [ pytest ];
+
+  checkPhase = ''
+    pytest
+  '';
+
+  # The tests fail with newest pytest. They passed with pytest_3, which no longer exists
+  doCheck = false;
+
+  meta = with stdenv.lib; {
+    homepage = "https://github.com/saltstack/pytest-helpers-namespace";
+    description = "PyTest Helpers Namespace";
+    license = licenses.asl20;
+    maintainers = [ maintainers.kiwi ];
+  };
+}

pkgs/development/python-modules/python-xmp-toolkit/default.nix

@@ -0,0 +1,44 @@
+{ buildPythonPackage
+, exempi
+, fetchFromGitHub
+, mock
+, pythonOlder
+, pytz
+, stdenv
+}:
+
+buildPythonPackage rec {
+  pname = "python-xmp-toolkit";
+  version = "2.0.2";
+
+  # PyPi has version 2.0.1; the tests fail
+  # There are commits for a 2.0.2 release that was never published
+  # Not to github, not to PyPi
+  # This is the latest commit from Jun 29, 2017 (as of Mar 13, 2019)
+  # It includes the commits for the unreleased version 2.0.2 and more
+  # Tests pass with this version
+  src = fetchFromGitHub {
+    owner = "python-xmp-toolkit";
+    repo = "python-xmp-toolkit";
+    rev = "5692bdf8dac3581a0d5fb3c5aeb29be0ab6a54fc";
+    sha256 = "16bylcm183ilzp7mrpdzw0pzp6csv9v5v247914qsv2abg0hgl5y";
+  };
+
+  buildInputs = [ exempi ];
+
+  checkInputs = stdenv.lib.optionals (pythonOlder "3.3") [ mock ];
+
+  propagatedBuildInputs = [ pytz ];
+
+  postPatch = ''
+    substituteInPlace libxmp/exempi.py \
+      --replace "ctypes.util.find_library('exempi')" "'${exempi}/lib/libexempi${stdenv.hostPlatform.extensions.sharedLibrary}'"
+  '';
+
+  meta = with stdenv.lib; {
+    homepage = https://github.com/python-xmp-toolkit/python-xmp-toolkit;
+    description = "Python XMP Toolkit for working with metadata";
+    license = licenses.bsd3;
+    maintainers = [ maintainers.kiwi ];
+  };
+}

pkgs/development/python-modules/ruffus/default.nix

@@ -0,0 +1,53 @@
+{ gevent
+, buildPythonPackage
+, fetchFromGitHub
+, hostname
+, pytest
+, python
+, stdenv
+}:
+
+buildPythonPackage rec {
+  pname = "ruffus";
+  version = "2.8.1";
+
+  src = fetchFromGitHub {
+    owner = "cgat-developers";
+    repo = "${pname}";
+    rev = "v${version}";
+    sha256 = "1gyabqafq4s2sy0prh3k1m8859shzjmfxr7fimx10liflvki96a9";
+  };
+
+  propagatedBuildInputs = [ gevent ];
+
+  postPatch = ''
+    sed -i -e 's|/bin/bash|${stdenv.shell}|'          ruffus/test/Makefile
+    sed -i -e 's|\tpytest|\t${pytest}/bin/pytest|'    ruffus/test/Makefile
+    sed -i -e 's|\tpython|\t${python.interpreter}|'   ruffus/test/Makefile
+    sed -i -e 's|/usr/bin/env bash|${stdenv.shell}|'  ruffus/test/run_all_unit_tests.cmd
+    sed -i -e 's|python3|${python.interpreter}|'      ruffus/test/run_all_unit_tests3.cmd
+    sed -i -e 's|python %s|${python.interpreter} %s|' ruffus/test/test_drmaa_wrapper_run_job_locally.py
+  '';
+
+  makefile = "ruffus/test/Makefile";
+
+  checkInputs = [
+    gevent
+    hostname
+    pytest
+  ];
+
+  checkPhase = ''
+    export HOME=$TMPDIR
+    cd ruffus/test
+    make all PYTEST_OPTIONS="-q --disable-warnings"
+  '';
+
+  meta = with stdenv.lib; {
+    description = "Light-weight Python Computational Pipeline Management";
+    homepage = http://www.ruffus.org.uk;
+    license = licenses.mit;
+    maintainers = [ maintainers.kiwi ];
+  };
+}
+

pkgs/tools/text/ocrmypdf/default.nix

@@ -0,0 +1,103 @@
+{ fetchFromGitHub
+, ghostscript
+, img2pdf
+, jbig2enc
+, leptonica
+, pngquant
+, python3
+, python3Packages
+, qpdf
+, stdenv
+, tesseract4
+, unpaper
+}:
+
+let
+  inherit (python3Packages) buildPythonApplication;
+
+  runtimeDeps = with python3Packages; [
+    ghostscript
+    jbig2enc
+    leptonica
+    pngquant
+    qpdf
+    tesseract4
+    unpaper
+    pillow
+  ];
+
+in buildPythonApplication rec {
+  pname = "ocrmypdf";
+  version = "8.2.3";
+  disabled = ! python3Packages.isPy3k;
+
+  src = fetchFromGitHub {
+    owner = "jbarlow83";
+    repo = "OCRmyPDF";
+    rev = "v${version}";
+    sha256 = "1ldlyhxkav34y9d7g2kx3d4p26c2b82vnwi0ywnfynb16sav36d5";
+  };
+
+  nativeBuildInputs = with python3Packages; [
+    pytestrunner
+    setuptools
+    setuptools-scm-git-archive
+    setuptools_scm
+  ];
+
+  propagatedBuildInputs = with python3Packages; [
+    cffi
+    chardet
+    img2pdf
+    pdfminer
+    pikepdf
+    reportlab
+    ruffus
+  ];
+
+  checkInputs = with python3Packages; [
+    hocr-tools
+    pypdf2
+    pytest
+    pytest-helpers-namespace
+    pytest_xdist
+    pytestcov
+    pytestrunner
+    python-xmp-toolkit
+    setuptools
+  ] ++ runtimeDeps;
+
+
+  postPatch = ''
+    substituteInPlace src/ocrmypdf/leptonica.py \
+      --replace "ffi.dlopen(find_library('lept'))" \
+      'ffi.dlopen("${stdenv.lib.makeLibraryPath [leptonica]}/liblept${stdenv.hostPlatform.extensions.sharedLibrary}")'
+  '';
+
+  # The tests take potentially 20+ minutes, depending on machine
+  doCheck = false;
+
+  # These tests fail and it might be upstream problem... or packaging. :)
+  # development is happening on macos and the pinned test versions are
+  # significantly newer than nixpkgs has. Program still works...
+  # (to the extent I've used it) -- Kiwi
+  checkPhase = ''
+    export HOME=$TMPDIR
+    pytest -k 'not test_force_ocr_on_pdf_with_no_images \
+    and not test_tesseract_crash \
+    and not test_tesseract_crash_autorotate \
+    and not test_ghostscript_pdfa_failure \
+    and not test_gs_render_failure \
+    and not test_gs_raster_failure \
+    and not test_bad_utf8 \
+    and not test_old_unpaper'
+  '';
+
+  meta = with stdenv.lib; {
+    homepage = "https://github.com/jbarlow83/OCRmyPDF";
+    description = "Adds an OCR text layer to scanned PDF files, allowing them to be searched";
+    license = licenses.gpl3;
+    platforms = platforms.linux;
+    maintainers = [ maintainers.kiwi ];
+  };
+}

pkgs/top-level/all-packages.nix

@@ -1704,6 +1704,8 @@ in
 
   hid-listen = callPackage ../tools/misc/hid-listen { };
 
+  hocr-tools = with python3Packages; toPythonApplication hocr-tools;
+
   home-manager = callPackage ../tools/package-management/home-manager {};
 
   hostsblock = callPackage ../tools/misc/hostsblock { };
@@ -1828,6 +1830,8 @@ in
 
   nyx = callPackage ../tools/networking/nyx { };
 
+  ocrmypdf = callPackage ../tools/text/ocrmypdf { };
+
   onboard = callPackage ../applications/misc/onboard { };
 
   xkbd = callPackage ../applications/misc/xkbd { };

pkgs/top-level/python-packages.nix

@@ -643,6 +643,8 @@ in {
 
   hdmedians = callPackage ../development/python-modules/hdmedians { };
 
+  hocr-tools = callPackage ../development/python-modules/hocr-tools { };
+
   holoviews = callPackage ../development/python-modules/holoviews { };
 
   hoomd-blue = toPythonModule (callPackage ../development/python-modules/hoomd-blue {
@@ -1952,6 +1954,8 @@ in {
     hypothesis = self.hypothesis.override { doCheck = false; };
   };
 
+  pytest-helpers-namespace = callPackage ../development/python-modules/pytest-helpers-namespace { };
+
   pytest-httpbin = callPackage ../development/python-modules/pytest-httpbin { };
 
   pytest-asyncio = callPackage ../development/python-modules/pytest-asyncio { };
@@ -4143,6 +4147,8 @@ in {
 
   pika-pool = callPackage ../development/python-modules/pika-pool { };
 
+  pikepdf = callPackage ../development/python-modules/pikepdf { };
+
   kmapper = callPackage ../development/python-modules/kmapper { };
 
   kmsxx = (callPackage ../development/libraries/kmsxx {
@@ -5083,6 +5089,8 @@ in {
 
   ruamel_yaml = callPackage ../development/python-modules/ruamel_yaml { };
 
+  ruffus = callPackage ../development/python-modules/ruffus { };
+
   runsnakerun = callPackage ../development/python-modules/runsnakerun { };
 
   pysendfile = callPackage ../development/python-modules/pysendfile { };
@@ -5669,6 +5677,8 @@ in {
 
   python-u2flib-host = callPackage ../development/python-modules/python-u2flib-host { };
 
+  python-xmp-toolkit = callPackage ../development/python-modules/python-xmp-toolkit { };
+
   pluggy = callPackage ../development/python-modules/pluggy {};
 
   xcffib = callPackage ../development/python-modules/xcffib {};