gnu/packages/machine-learning.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2015, 2016, 2017, 2018, 2019 Ricardo Wurmus <rekado@elephly.net>
   3 ;;; Copyright © 2016 Efraim Flashner <efraim@flashner.co.il>
   4 ;;; Copyright © 2016, 2017 Marius Bakke <mbakke@fastmail.com>
   5 ;;; Copyright © 2016 Hartmut Goebel <h.goebel@crazy-compilers.com>
   6 ;;; Copyright © 2018, 2019 Tobias Geerinckx-Rice <me@tobias.gr>
   7 ;;; Copyright © 2018 Kei Kebreau <kkebreau@posteo.net>
   8 ;;; Copyright © 2018 Mark Meyer <mark@ofosos.org>
   9 ;;; Copyright © 2018 Ben Woodcroft <donttrustben@gmail.com>
  10 ;;; Copyright © 2018 Fis Trivial <ybbs.daans@hotmail.com>
  11 ;;; Copyright © 2018 Julien Lepiller <julien@lepiller.eu>
  12 ;;; Copyright © 2018 Björn Höfling <bjoern.hoefling@bjoernhoefling.de>
  13 ;;; Copyright © 2019 Nicolas Goaziou <mail@nicolasgoaziou.fr>
  14 ;;;
  15 ;;; This file is part of GNU Guix.
  16 ;;;
  17 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  18 ;;; under the terms of the GNU General Public License as published by
  19 ;;; the Free Software Foundation; either version 3 of the License, or (at
  20 ;;; your option) any later version.
  21 ;;;
  22 ;;; GNU Guix is distributed in the hope that it will be useful, but
  23 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  24 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  25 ;;; GNU General Public License for more details.
  26 ;;;
  27 ;;; You should have received a copy of the GNU General Public License
  28 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  29
  30 (define-module (gnu packages machine-learning)
  31   #:use-module ((guix licenses) #:prefix license:)
  32   #:use-module (guix packages)
  33   #:use-module (guix utils)
  34   #:use-module (guix download)
  35   #:use-module (guix svn-download)
  36   #:use-module (guix build-system cmake)
  37   #:use-module (guix build-system gnu)
  38   #:use-module (guix build-system ocaml)
  39   #:use-module (guix build-system python)
  40   #:use-module (guix build-system r)
  41   #:use-module (guix git-download)
  42   #:use-module (gnu packages)
  43   #:use-module (gnu packages adns)
  44   #:use-module (gnu packages algebra)
  45   #:use-module (gnu packages audio)
  46   #:use-module (gnu packages autotools)
  47   #:use-module (gnu packages base)
  48   #:use-module (gnu packages bash)
  49   #:use-module (gnu packages boost)
  50   #:use-module (gnu packages check)
  51   #:use-module (gnu packages compression)
  52   #:use-module (gnu packages cran)
  53   #:use-module (gnu packages databases)
  54   #:use-module (gnu packages dejagnu)
  55   #:use-module (gnu packages gcc)
  56   #:use-module (gnu packages glib)
  57   #:use-module (gnu packages graphviz)
  58   #:use-module (gnu packages gstreamer)
  59   #:use-module (gnu packages image)
  60   #:use-module (gnu packages linux)
  61   #:use-module (gnu packages maths)
  62   #:use-module (gnu packages mpi)
  63   #:use-module (gnu packages ocaml)
  64   #:use-module (gnu packages onc-rpc)
  65   #:use-module (gnu packages perl)
  66   #:use-module (gnu packages pkg-config)
  67   #:use-module (gnu packages protobuf)
  68   #:use-module (gnu packages python)
  69   #:use-module (gnu packages python-web)
  70   #:use-module (gnu packages python-xyz)
  71   #:use-module (gnu packages serialization)
  72   #:use-module (gnu packages sphinx)
  73   #:use-module (gnu packages statistics)
  74   #:use-module (gnu packages sqlite)
  75   #:use-module (gnu packages swig)
  76   #:use-module (gnu packages tls)
  77   #:use-module (gnu packages web)
  78   #:use-module (gnu packages xml)
  79   #:use-module (gnu packages xorg)
  80   #:use-module (ice-9 match))
  81
  82 (define-public fann
  83   ;; The last release is >100 commits behind, so we package from git.
  84   (let ((commit "d71d54788bee56ba4cf7522801270152da5209d7"))
  85     (package
  86       (name "fann")
  87       (version (string-append "2.2.0-1." (string-take commit 8)))
  88       (source (origin
  89                 (method git-fetch)
  90                 (uri (git-reference
  91                       (url "https://github.com/libfann/fann.git")
  92                       (commit commit)))
  93                 (file-name (string-append name "-" version "-checkout"))
  94                 (sha256
  95                  (base32
  96                   "0ibwpfrjs6q2lijs8slxjgzb2llcl6rk3v2ski4r6215g5jjhg3x"))))
  97       (build-system cmake-build-system)
  98       (arguments
  99        `(#:phases
 100          (modify-phases %standard-phases
 101            (replace 'check
 102              (lambda* (#:key outputs #:allow-other-keys)
 103                (let* ((out (assoc-ref outputs "out")))
 104                  (with-directory-excursion (string-append (getcwd) "/tests")
 105                    (invoke "./fann_tests"))))))))
 106       (home-page "http://leenissen.dk/fann/wp/")
 107       (synopsis "Fast Artificial Neural Network")
 108       (description
 109        "FANN is a free open source neural network library, which implements
 110 multilayer artificial neural networks in C with support for both fully
 111 connected and sparsely connected networks.")
 112       (license license:lgpl2.1))))
 113
 114 (define-public libsvm
 115   (package
 116     (name "libsvm")
 117     (version "3.23")
 118     (source
 119      (origin
 120        (method url-fetch)
 121        (uri (string-append "https://www.csie.ntu.edu.tw/~cjlin/libsvm/"
 122                            name "-" version ".tar.gz"))
 123        (sha256
 124         (base32 "0jpaq0rr92x38p4nk3gjan79ip67m6p80anb28z1d8601miysyi5"))))
 125     (build-system gnu-build-system)
 126     (arguments
 127      `(#:tests? #f                      ; no "check" target
 128        #:phases (modify-phases %standard-phases
 129                   (delete 'configure)
 130                   (replace
 131                    'install             ; no ‘install’ target
 132                    (lambda* (#:key outputs #:allow-other-keys)
 133                      (let* ((out (assoc-ref outputs "out"))
 134                             (bin (string-append out "/bin/")))
 135                        (mkdir-p bin)
 136                        (for-each (lambda (file)
 137                                    (copy-file file (string-append bin file)))
 138                                  '("svm-train"
 139                                    "svm-predict"
 140                                    "svm-scale")))
 141                      #t)))))
 142     (home-page "http://www.csie.ntu.edu.tw/~cjlin/libsvm/")
 143     (synopsis "Library for Support Vector Machines")
 144     (description
 145      "LIBSVM is a machine learning library for support vector
 146 classification, (C-SVC, nu-SVC), regression (epsilon-SVR, nu-SVR) and
 147 distribution estimation (one-class SVM).  It supports multi-class
 148 classification.")
 149     (license license:bsd-3)))
 150
 151 (define-public python-libsvm
 152   (package (inherit libsvm)
 153     (name "python-libsvm")
 154     (build-system gnu-build-system)
 155     (arguments
 156      `(#:tests? #f                      ; no "check" target
 157        #:make-flags '("-C" "python")
 158        #:phases
 159        (modify-phases %standard-phases
 160          (delete 'configure)
 161          (replace
 162           'install                      ; no ‘install’ target
 163           (lambda* (#:key inputs outputs #:allow-other-keys)
 164             (let ((site (string-append (assoc-ref outputs "out")
 165                                        "/lib/python"
 166                                        (string-take
 167                                         (string-take-right
 168                                          (assoc-ref inputs "python") 5) 3)
 169                                        "/site-packages/")))
 170               (substitute* "python/svm.py"
 171                 (("../libsvm.so.2") "libsvm.so.2"))
 172               (mkdir-p site)
 173               (for-each (lambda (file)
 174                           (copy-file file (string-append site (basename file))))
 175                         (find-files "python" "\\.py"))
 176               (copy-file "libsvm.so.2"
 177                          (string-append site "libsvm.so.2")))
 178             #t)))))
 179     (inputs
 180      `(("python" ,python)))
 181     (synopsis "Python bindings of libSVM")))
 182
 183 (define-public ghmm
 184   ;; The latest release candidate is several years and a couple of fixes have
 185   ;; been published since.  This is why we download the sources from the SVN
 186   ;; repository.
 187   (let ((svn-revision 2341))
 188     (package
 189       (name "ghmm")
 190       (version (string-append "0.9-rc3-0." (number->string svn-revision)))
 191       (source (origin
 192                 (method svn-fetch)
 193                 (uri (svn-reference
 194                       (url "http://svn.code.sf.net/p/ghmm/code/trunk")
 195                       (revision svn-revision)))
 196                 (file-name (string-append name "-" version "-checkout"))
 197                 (sha256
 198                  (base32
 199                   "0qbq1rqp94l530f043qzp8aw5lj7dng9wq0miffd7spd1ff638wq"))))
 200       (build-system gnu-build-system)
 201       (arguments
 202        `(#:imported-modules (,@%gnu-build-system-modules
 203                              (guix build python-build-system))
 204          #:phases
 205          (modify-phases %standard-phases
 206            (add-after 'unpack 'enter-dir
 207              (lambda _ (chdir "ghmm") #t))
 208            (delete 'check)
 209            (add-after 'install 'check
 210              (assoc-ref %standard-phases 'check))
 211            (add-before 'check 'fix-PYTHONPATH
 212              (lambda* (#:key inputs outputs #:allow-other-keys)
 213                (let ((python-version (python-version
 214                                       (assoc-ref inputs "python"))))
 215                  (setenv "PYTHONPATH"
 216                          (string-append (getenv "PYTHONPATH")
 217                                         ":" (assoc-ref outputs "out")
 218                                         "/lib/python" python-version
 219                                         "/site-packages")))
 220                #t))
 221            (add-after 'enter-dir 'fix-runpath
 222              (lambda* (#:key outputs #:allow-other-keys)
 223                (substitute* "ghmmwrapper/setup.py"
 224                  (("^(.*)extra_compile_args = \\[" line indent)
 225                   (string-append indent
 226                                  "extra_link_args = [\"-Wl,-rpath="
 227                                  (assoc-ref outputs "out") "/lib\"],\n"
 228                                  line
 229                                  "\"-Wl,-rpath="
 230                                  (assoc-ref outputs "out")
 231                                  "/lib\", ")))
 232                #t))
 233            (add-after 'enter-dir 'disable-broken-tests
 234              (lambda _
 235                (substitute* "tests/Makefile.am"
 236                  ;; GHMM_SILENT_TESTS is assumed to be a command.
 237                  (("TESTS_ENVIRONMENT.*") "")
 238                  ;; Do not build broken tests.
 239                  (("chmm .*") "")
 240                  (("read_fa .*") "")
 241                  (("mcmc .*") "")
 242                  (("label_higher_order_test.*$")
 243                   "label_higher_order_test\n"))
 244
 245                ;; These Python unittests are broken as there is no gato.
 246                ;; See https://sourceforge.net/p/ghmm/support-requests/3/
 247                (substitute* "ghmmwrapper/ghmmunittests.py"
 248                  (("^(.*)def (testNewXML|testMultipleTransitionClasses|testNewXML)"
 249                    line indent)
 250                   (string-append indent
 251                                  "@unittest.skip(\"Disabled by Guix\")\n"
 252                                  line)))
 253                #t)))))
 254       (inputs
 255        `(("python" ,python-2) ; only Python 2 is supported
 256          ("libxml2" ,libxml2)))
 257       (native-inputs
 258        `(("pkg-config" ,pkg-config)
 259          ("dejagnu" ,dejagnu)
 260          ("swig" ,swig)
 261          ("autoconf" ,autoconf)
 262          ("automake" ,automake)
 263          ("libtool" ,libtool)))
 264       (home-page "http://ghmm.org")
 265       (synopsis "Hidden Markov Model library")
 266       (description
 267        "The General Hidden Markov Model library (GHMM) is a C library with
 268 additional Python bindings implementing a wide range of types of @dfn{Hidden
 269 Markov Models} (HMM) and algorithms: discrete, continuous emissions, basic
 270 training, HMM clustering, HMM mixtures.")
 271       (license license:lgpl2.0+))))
 272
 273 (define-public mcl
 274   (package
 275     (name "mcl")
 276     (version "14.137")
 277     (source (origin
 278               (method url-fetch)
 279               (uri (string-append
 280                     "http://micans.org/mcl/src/mcl-"
 281                     (string-replace-substring version "." "-")
 282                     ".tar.gz"))
 283               (sha256
 284                (base32
 285                 "15xlax3z31lsn62vlg94hkm75nm40q4679amnfg13jm8m2bnhy5m"))))
 286     (build-system gnu-build-system)
 287     (arguments
 288      `(#:configure-flags (list "--enable-blast")))
 289     (inputs
 290      `(("perl" ,perl)))
 291     (home-page "http://micans.org/mcl/")
 292     (synopsis "Clustering algorithm for graphs")
 293     (description
 294      "The MCL algorithm is short for the @dfn{Markov Cluster Algorithm}, a
 295 fast and scalable unsupervised cluster algorithm for graphs (also known as
 296 networks) based on simulation of (stochastic) flow in graphs.")
 297     ;; In the LICENCE file and web page it says "The software is licensed
 298     ;; under the GNU General Public License, version 3.", but in several of
 299     ;; the source code files it suggests GPL3 or later.
 300     ;; http://listserver.ebi.ac.uk/pipermail/mcl-users/2016/000376.html
 301     (license license:gpl3)))
 302
 303 (define-public ocaml-mcl
 304   (package
 305     (name "ocaml-mcl")
 306     (version "12-068oasis4")
 307     (source
 308      (origin
 309        (method git-fetch)
 310        (uri (git-reference
 311              (url "https://github.com/fhcrc/mcl.git")
 312              (commit version)))
 313        (file-name (git-file-name name version))
 314        (sha256
 315         (base32
 316          "0009dc3h2jp3qg5val452wngpqnbfyhbcxylghq0mrjqxx0jdq5p"))))
 317     (build-system ocaml-build-system)
 318     (arguments
 319      `(#:phases
 320        (modify-phases %standard-phases
 321          (add-before 'configure 'patch-paths
 322            (lambda _
 323              (substitute* "configure"
 324                (("/bin/sh") (which "sh")))
 325              (substitute* "setup.ml"
 326                (("LDFLAGS=-fPIC")
 327                 (string-append "LDFLAGS=-fPIC\"; \"SHELL=" (which "sh")))
 328                (("-std=c89") "-std=gnu99")
 329
 330                ;; This is a mutable string, which is no longer supported.  Use
 331                ;; a byte buffer instead.
 332                (("String.make \\(String.length s\\)")
 333                 "Bytes.make (String.length s)")
 334
 335                ;; These two belong together.
 336                (("OASISString.replace_chars")
 337                 "Bytes.to_string (OASISString.replace_chars")
 338                ((" s;")
 339                 " s);"))
 340              (substitute* "myocamlbuild.ml"
 341                (("std=c89") "std=gnu99"))
 342              ;; Since we build with a more recent OCaml, we have to use C99 or
 343              ;; later.  This causes problems with the old C code.
 344              (substitute* "src/impala/matrix.c"
 345                (("restrict") "restrict_"))
 346              #t)))))
 347     (native-inputs
 348      `(("ocamlbuild" ,ocamlbuild)))
 349     (home-page "https://github.com/fhcrc/mcl")
 350     (synopsis "OCaml wrappers around MCL")
 351     (description
 352      "This package provides OCaml bindings for the MCL graph clustering
 353 algorithm.")
 354     (license license:gpl3)))
 355
 356 (define-public randomjungle
 357   (package
 358     (name "randomjungle")
 359     (version "2.1.0")
 360     (source
 361      (origin
 362        (method url-fetch)
 363        (uri (string-append
 364              "https://www.imbs.uni-luebeck.de/fileadmin/files/Software"
 365              "/randomjungle/randomjungle-" version ".tar_.gz"))
 366        (patches (search-patches "randomjungle-disable-static-build.patch"))
 367        (sha256
 368         (base32
 369          "12c8rf30cla71swx2mf4ww9mfd8jbdw5lnxd7dxhyw1ygrvg6y4w"))))
 370     (build-system gnu-build-system)
 371     (arguments
 372      `(#:configure-flags
 373        (list "--disable-static"
 374              (string-append "--with-boost="
 375                             (assoc-ref %build-inputs "boost")))
 376        #:phases
 377        (modify-phases %standard-phases
 378          (add-before
 379           'configure 'set-CXXFLAGS
 380           (lambda _
 381             (setenv "CXXFLAGS" "-fpermissive ")
 382             #t)))))
 383     (inputs
 384      `(("boost" ,boost)
 385        ("gsl" ,gsl)
 386        ("libxml2" ,libxml2)
 387        ("zlib" ,zlib)))
 388     (native-inputs
 389      `(("gfortran" ,gfortran)
 390        ("gfortran:lib" ,gfortran "lib")))
 391     ;; Non-portable assembly instructions are used so building fails on
 392     ;; platforms other than x86_64 or i686.
 393     (supported-systems '("x86_64-linux" "i686-linux"))
 394     (home-page "https://www.imbs.uni-luebeck.de/forschung/software/details.html#c224")
 395     (synopsis "Implementation of the Random Forests machine learning method")
 396     (description
 397      "Random Jungle is an implementation of Random Forests.  It is supposed to
 398 analyse high dimensional data.  In genetics, it can be used for analysing big
 399 Genome Wide Association (GWA) data.  Random Forests is a powerful machine
 400 learning method.  Most interesting features are variable selection, missing
 401 value imputation, classifier creation, generalization error estimation and
 402 sample proximities between pairs of cases.")
 403     (license license:gpl3+)))
 404
 405 (define-public openfst
 406   (package
 407     (name "openfst")
 408     (version "1.7.2")
 409     (source (origin
 410               (method url-fetch)
 411               (uri (string-append "http://www.openfst.org/twiki/pub/FST/"
 412                                   "FstDownload/openfst-" version ".tar.gz"))
 413               (sha256
 414                (base32
 415                 "0fqgk8195kz21is09gwzwnrg7fr9526bi9mh4apyskapz27pbhr1"))))
 416     (build-system gnu-build-system)
 417     (home-page "http://www.openfst.org")
 418     (synopsis "Library for weighted finite-state transducers")
 419     (description "OpenFst is a library for constructing, combining,
 420 optimizing, and searching weighted finite-state transducers (FSTs).")
 421     (license license:asl2.0)))
 422
 423 (define-public shogun
 424   (package
 425     (name "shogun")
 426     (version "6.1.3")
 427     (source
 428      (origin
 429        (method url-fetch)
 430        (uri (string-append
 431              "ftp://shogun-toolbox.org/shogun/releases/"
 432              (version-major+minor version)
 433              "/sources/shogun-" version ".tar.bz2"))
 434        (sha256
 435         (base32
 436          "1rn9skm3nw6hr7mr3lgp2gfqhi7ii0lyxck7qmqnf8avq349s5jp"))
 437        (modules '((guix build utils)
 438                   (ice-9 rdelim)))
 439        (snippet
 440         '(begin
 441            ;; Remove non-free sources and files referencing them
 442            (for-each delete-file
 443                      (find-files "src/shogun/classifier/svm/"
 444                                  "SVMLight\\.(cpp|h)"))
 445            (for-each delete-file
 446                      (find-files "examples/undocumented/libshogun/"
 447                                  (string-append
 448                                   "(classifier_.*svmlight.*|"
 449                                   "evaluation_cross_validation_locked_comparison).cpp")))
 450            ;; Remove non-free functions.
 451            (define (delete-ifdefs file)
 452              (with-atomic-file-replacement file
 453                (lambda (in out)
 454                  (let loop ((line (read-line in 'concat))
 455                             (skipping? #f))
 456                    (if (eof-object? line)
 457                        #t
 458                        (let ((skip-next?
 459                               (or (and skipping?
 460                                        (not (string-prefix?
 461                                              "#endif //USE_SVMLIGHT" line)))
 462                                   (string-prefix?
 463                                    "#ifdef USE_SVMLIGHT" line))))
 464                          (when (or (not skipping?)
 465                                    (and skipping? (not skip-next?)))
 466                            (display line out))
 467                          (loop (read-line in 'concat) skip-next?)))))))
 468            (for-each delete-ifdefs
 469                      (append
 470                       (find-files "src/shogun/classifier/mkl"
 471                                   "^MKLClassification\\.cpp")
 472                       (find-files "src/shogun/classifier/svm"
 473                                   "^SVMLightOneClass\\.(cpp|h)")
 474                       (find-files "src/shogun/multiclass"
 475                                   "^ScatterSVM\\.(cpp|h)")
 476                       (find-files "src/shogun/kernel/"
 477                                   "^(Kernel|CombinedKernel|ProductKernel)\\.(cpp|h)")
 478                       (find-files "src/shogun/regression/svr"
 479                                   "^(MKLRegression|SVRLight)\\.(cpp|h)")
 480                       (find-files "src/shogun/transfer/domain_adaptation"
 481                                   "^DomainAdaptationSVM\\.(cpp|h)")))
 482            #t))))
 483     (build-system cmake-build-system)
 484     (arguments
 485      '(#:tests? #f ;no check target
 486        #:phases
 487        (modify-phases %standard-phases
 488          (add-after 'unpack 'delete-broken-symlinks
 489            (lambda _
 490              (for-each delete-file '("applications/arts/data"
 491                                      "applications/asp/data"
 492                                      "applications/easysvm/data"
 493                                      "applications/msplicer/data"
 494                                      "applications/ocr/data"
 495                                      "examples/meta/data"
 496                                      "examples/undocumented/data"))
 497              #t))
 498          (add-after 'unpack 'change-R-target-path
 499            (lambda* (#:key outputs #:allow-other-keys)
 500              (substitute* '("src/interfaces/r/CMakeLists.txt"
 501                             "examples/meta/r/CMakeLists.txt")
 502                (("\\$\\{R_COMPONENT_LIB_PATH\\}")
 503                 (string-append (assoc-ref outputs "out")
 504                                "/lib/R/library/")))
 505              #t))
 506          (add-after 'unpack 'fix-octave-modules
 507            (lambda* (#:key outputs #:allow-other-keys)
 508              (substitute* "src/interfaces/octave/CMakeLists.txt"
 509                (("^include_directories\\(\\$\\{OCTAVE_INCLUDE_DIRS\\}")
 510                 "include_directories(${OCTAVE_INCLUDE_DIRS} ${OCTAVE_INCLUDE_DIRS}/octave")
 511                ;; change target directory
 512                (("\\$\\{OCTAVE_OCT_LOCAL_API_FILE_DIR\\}")
 513                 (string-append (assoc-ref outputs "out")
 514                                "/share/octave/packages")))
 515              (substitute* '("src/interfaces/octave/swig_typemaps.i"
 516                             "src/interfaces/octave/sg_print_functions.cpp")
 517                ;; "octave/config.h" and "octave/oct-obj.h" deprecated in Octave.
 518                (("octave/config\\.h") "octave/octave-config.h")
 519                (("octave/oct-obj.h") "octave/ovl.h"))
 520              #t))
 521          (add-after 'unpack 'move-rxcpp
 522            (lambda* (#:key inputs #:allow-other-keys)
 523              (let ((rxcpp-dir "shogun/third-party/rxcpp"))
 524                (mkdir-p rxcpp-dir)
 525                (install-file (assoc-ref inputs "rxcpp") rxcpp-dir)
 526                #t)))
 527          (add-before 'build 'set-HOME
 528            ;; $HOME needs to be set at some point during the build phase
 529            (lambda _ (setenv "HOME" "/tmp") #t)))
 530        #:configure-flags
 531        (list "-DCMAKE_BUILD_WITH_INSTALL_RPATH=TRUE"
 532              "-DUSE_SVMLIGHT=OFF" ;disable proprietary SVMLIGHT
 533              "-DBUILD_META_EXAMPLES=OFF" ;requires unpackaged ctags
 534              ;;"-DINTERFACE_JAVA=ON" ;requires unpackaged jblas
 535              ;;"-DINTERFACE_RUBY=ON" ;requires unpackaged ruby-narray
 536              ;;"-DINTERFACE_PERL=ON" ;"FindPerlLibs" does not exist
 537              ;;"-DINTERFACE_LUA=ON"  ;fails because lua doesn't build pkgconfig file
 538              "-DINTERFACE_OCTAVE=ON"
 539              "-DINTERFACE_PYTHON=ON"
 540              "-DINTERFACE_R=ON")))
 541     (inputs
 542      `(("python" ,python)
 543        ("numpy" ,python-numpy)
 544        ("r-minimal" ,r-minimal)
 545        ("octave" ,octave-cli)
 546        ("swig" ,swig)
 547        ("eigen" ,eigen)
 548        ("hdf5" ,hdf5)
 549        ("atlas" ,atlas)
 550        ("arpack" ,arpack-ng)
 551        ("lapack" ,lapack)
 552        ("glpk" ,glpk)
 553        ("libxml2" ,libxml2)
 554        ("lzo" ,lzo)
 555        ("zlib" ,zlib)))
 556     (native-inputs
 557      `(("pkg-config" ,pkg-config)
 558        ("rxcpp" ,rxcpp)))
 559     ;; Non-portable SSE instructions are used so building fails on platforms
 560     ;; other than x86_64.
 561     (supported-systems '("x86_64-linux"))
 562     (home-page "http://shogun-toolbox.org/")
 563     (synopsis "Machine learning toolbox")
 564     (description
 565      "The Shogun Machine learning toolbox provides a wide range of unified and
 566 efficient Machine Learning (ML) methods.  The toolbox seamlessly allows to
 567 combine multiple data representations, algorithm classes, and general purpose
 568 tools.  This enables both rapid prototyping of data pipelines and extensibility
 569 in terms of new algorithms.")
 570     (license license:gpl3+)))
 571
 572 (define-public rxcpp
 573   (package
 574     (name "rxcpp")
 575     (version "4.1.0")
 576     (source
 577      (origin
 578        (method git-fetch)
 579        (uri (git-reference
 580              (url "https://github.com/ReactiveX/RxCpp.git")
 581              (commit (string-append "v" version))))
 582        (sha256
 583         (base32 "1rdpa3jlc181jd08nk437aar085h28i45s6nzrv65apb3xyyz0ij"))
 584        (file-name (git-file-name name version))))
 585     (build-system cmake-build-system)
 586     (arguments
 587      `(#:phases
 588        (modify-phases %standard-phases
 589          (add-after 'unpack 'remove-werror
 590            (lambda _
 591              (substitute* (find-files ".")
 592                (("-Werror") ""))
 593              #t))
 594          (replace 'check
 595            (lambda _
 596              (invoke "ctest"))))))
 597     (native-inputs
 598      `(("catch" ,catch-framework)))
 599     (home-page "http://reactivex.io/")
 600     (synopsis "Reactive Extensions for C++")
 601     (description
 602      "The Reactive Extensions for C++ (RxCpp) is a library of algorithms for
 603 values-distributed-in-time.  ReactiveX is a library for composing asynchronous
 604 and event-based programs by using observable sequences.
 605
 606 It extends the observer pattern to support sequences of data and/or events and
 607 adds operators that allow you to compose sequences together declaratively while
 608 abstracting away concerns about things like low-level threading,
 609 synchronization, thread-safety, concurrent data structures, and non-blocking
 610 I/O.")
 611     (license license:asl2.0)))
 612
 613 (define-public r-adaptivesparsity
 614   (package
 615     (name "r-adaptivesparsity")
 616     (version "1.6")
 617     (source (origin
 618               (method url-fetch)
 619               (uri (cran-uri "AdaptiveSparsity" version))
 620               (sha256
 621                (base32
 622                 "0imr5m8mll9j6n4icsv6z9rl5kbnwsp9wvzrg7n90nnmcxq2cz91"))))
 623     (properties
 624      `((upstream-name . "AdaptiveSparsity")))
 625     (build-system r-build-system)
 626     (arguments
 627      `(#:phases
 628        (modify-phases %standard-phases
 629          (add-after 'unpack 'link-against-armadillo
 630            (lambda _
 631              (substitute* "src/Makevars"
 632                (("PKG_LIBS=" prefix)
 633                 (string-append prefix "-larmadillo"))))))))
 634     (propagated-inputs
 635      `(("r-mass" ,r-mass)
 636        ("r-matrix" ,r-matrix)
 637        ("r-rcpp" ,r-rcpp)
 638        ("r-rcpparmadillo" ,r-rcpparmadillo)))
 639     (inputs
 640      `(("armadillo" ,armadillo)))
 641     (home-page "https://cran.r-project.org/web/packages/AdaptiveSparsity")
 642     (synopsis "Adaptive sparsity models")
 643     (description
 644      "This package implements the Figueiredo machine learning algorithm for
 645 adaptive sparsity and the Wong algorithm for adaptively sparse gaussian
 646 geometric models.")
 647     (license license:lgpl3+)))
 648
 649 (define-public gemmlowp-for-tensorflow
 650   ;; The commit hash is taken from "tensorflow/workspace.bzl".
 651   (let ((commit "38ebac7b059e84692f53e5938f97a9943c120d98")
 652         (revision "2"))
 653     (package
 654       (name "gemmlowp")
 655       (version (git-version "0" revision commit))
 656       (source (origin
 657                 (method url-fetch)
 658                 (uri (string-append "https://mirror.bazel.build/"
 659                                     "github.com/google/gemmlowp/archive/"
 660                                     commit ".zip"))
 661                 (file-name (string-append "gemmlowp-" version ".zip"))
 662                 (sha256
 663                  (base32
 664                   "0n56s2g8hrssm4w8qj1v58gfm56a04n9v992ixkmvk6zjiralzxq"))))
 665       (build-system cmake-build-system)
 666       (arguments
 667        `(#:configure-flags
 668          (list ,@(match (%current-system)
 669                    ((or "x86_64-linux" "i686-linux")
 670                     '("-DCMAKE_CXX_FLAGS=-msse2"))
 671                    (_ '())))
 672          #:phases
 673          (modify-phases %standard-phases
 674            ;; This directory contains the CMakeLists.txt.
 675            (add-after 'unpack 'chdir
 676              (lambda _ (chdir "contrib") #t))
 677            ;; There is no install target
 678            (replace 'install
 679              (lambda* (#:key outputs #:allow-other-keys)
 680                (let* ((out (assoc-ref outputs "out"))
 681                       (lib (string-append out "/lib/"))
 682                       (inc (string-append out "/include/")))
 683                  (install-file "../build/libeight_bit_int_gemm.so" lib)
 684                  (for-each (lambda (dir)
 685                              (let ((target (string-append inc "/" dir)))
 686                                (mkdir-p target)
 687                                (for-each (lambda (h)
 688                                            (install-file h target))
 689                                          (find-files (string-append "../" dir)
 690                                                      "\\.h$"))))
 691                            '("meta" "profiling" "public" "fixedpoint"
 692                              "eight_bit_int_gemm" "internal"))
 693                  #t))))))
 694       (native-inputs
 695        `(("unzip" ,unzip)))
 696       (home-page "https://github.com/google/gemmlowp")
 697       (synopsis "Small self-contained low-precision GEMM library")
 698       (description
 699        "This is a small self-contained low-precision @dfn{general matrix
 700 multiplication} (GEMM) library.  It is not a full linear algebra library.
 701 Low-precision means that the input and output matrix entries are integers on
 702 at most 8 bits.  To avoid overflow, results are internally accumulated on more
 703 than 8 bits, and at the end only some significant 8 bits are kept.")
 704       (license license:asl2.0))))
 705
 706 (define-public dlib
 707   (package
 708     (name "dlib")
 709     (version "19.7")
 710     (source (origin
 711               (method url-fetch)
 712               (uri (string-append
 713                     "http://dlib.net/files/dlib-" version ".tar.bz2"))
 714               (sha256
 715                (base32
 716                 "1mljz02kwkrbggyncxv5fpnyjdybw2qihaacb3js8yfkw12vwpc2"))
 717               (modules '((guix build utils)))
 718               (snippet
 719                '(begin
 720                   ;; Delete ~13MB of bundled dependencies.
 721                   (delete-file-recursively "dlib/external")
 722                   (delete-file-recursively "docs/dlib/external")
 723                   #t))))
 724     (build-system cmake-build-system)
 725     (arguments
 726      `(#:phases
 727        (modify-phases %standard-phases
 728          (add-after 'unpack 'disable-asserts
 729            (lambda _
 730              ;; config.h recommends explicitly enabling or disabling asserts
 731              ;; when building as a shared library. By default neither is set.
 732              (substitute* "dlib/config.h"
 733                (("^//#define DLIB_DISABLE_ASSERTS") "#define DLIB_DISABLE_ASSERTS"))
 734              #t))
 735          (add-after 'disable-asserts 'disable-failing-tests
 736            (lambda _
 737              ;; One test times out on MIPS, so we need to disable it.
 738              ;; Others are flaky on some platforms.
 739              (let* ((system ,(or (%current-target-system)
 740                                  (%current-system)))
 741                     (disabled-tests (cond
 742                                      ((string-prefix? "mips64" system)
 743                                       '("object_detector" ; timeout
 744                                         "data_io"))
 745                                      ((string-prefix? "armhf" system)
 746                                       '("learning_to_track"))
 747                                      ((string-prefix? "i686" system)
 748                                       '("optimization"))
 749                                      (else '()))))
 750                (for-each
 751                 (lambda (test)
 752                   (substitute* "dlib/test/makefile"
 753                     (((string-append "SRC \\+= " test "\\.cpp")) "")))
 754                 disabled-tests)
 755                #t)))
 756          (replace 'check
 757            (lambda _
 758              ;; No test target, so we build and run the unit tests here.
 759              (let ((test-dir (string-append "../dlib-" ,version "/dlib/test")))
 760                (with-directory-excursion test-dir
 761                  (invoke "make" "-j" (number->string (parallel-job-count)))
 762                  (invoke "./dtest" "--runall"))
 763                #t)))
 764          (add-after 'install 'delete-static-library
 765            (lambda* (#:key outputs #:allow-other-keys)
 766              (delete-file (string-append (assoc-ref outputs "out")
 767                                          "/lib/libdlib.a"))
 768              #t)))))
 769     (native-inputs
 770      `(("pkg-config" ,pkg-config)
 771        ;; For tests.
 772        ("libnsl" ,libnsl)))
 773     (inputs
 774      `(("giflib" ,giflib)
 775        ("lapack" ,lapack)
 776        ("libjpeg" ,libjpeg)
 777        ("libpng" ,libpng)
 778        ("libx11" ,libx11)
 779        ("openblas" ,openblas)
 780        ("zlib" ,zlib)))
 781     (synopsis
 782      "Toolkit for making machine learning and data analysis applications in C++")
 783     (description
 784      "Dlib is a modern C++ toolkit containing machine learning algorithms and
 785 tools.  It is used in both industry and academia in a wide range of domains
 786 including robotics, embedded devices, mobile phones, and large high performance
 787 computing environments.")
 788     (home-page "http://dlib.net")
 789     (license license:boost1.0)))
 790
 791 (define-public python-scikit-learn
 792   (package
 793     (name "python-scikit-learn")
 794     (version "0.20.3")
 795     (source
 796      (origin
 797        (method git-fetch)
 798        (uri (git-reference
 799              (url "https://github.com/scikit-learn/scikit-learn.git")
 800              (commit version)))
 801        (file-name (git-file-name name version))
 802        (sha256
 803         (base32
 804          "08aaby5zphfxy83mggg35bwyka7wk91l2qijh8kk0bl08dikq8dl"))))
 805     (build-system python-build-system)
 806     (arguments
 807      `(#:phases
 808        (modify-phases %standard-phases
 809          (add-after 'build 'build-ext
 810            (lambda _ (invoke "python" "setup.py" "build_ext" "--inplace") #t))
 811          (replace 'check
 812            (lambda _
 813              ;; Restrict OpenBLAS threads to prevent segfaults while testing!
 814              (setenv "OPENBLAS_NUM_THREADS" "1")
 815
 816              ;; Some tests require write access to $HOME.
 817              (setenv "HOME" "/tmp")
 818
 819              (invoke "pytest" "sklearn" "-m" "not network")))
 820          (add-before 'reset-gzip-timestamps 'make-files-writable
 821            (lambda* (#:key outputs #:allow-other-keys)
 822              ;; Make sure .gz files are writable so that the
 823              ;; 'reset-gzip-timestamps' phase can do its work.
 824              (let ((out (assoc-ref outputs "out")))
 825                (for-each make-file-writable
 826                          (find-files out "\\.gz$"))
 827                #t))))))
 828     (inputs
 829      `(("openblas" ,openblas)))
 830     (native-inputs
 831      `(("python-pytest" ,python-pytest)
 832        ("python-pandas" ,python-pandas) ;for tests
 833        ("python-cython" ,python-cython)))
 834     (propagated-inputs
 835      `(("python-numpy" ,python-numpy)
 836        ("python-scipy" ,python-scipy)))
 837     (home-page "http://scikit-learn.org/")
 838     (synopsis "Machine Learning in Python")
 839     (description
 840      "Scikit-learn provides simple and efficient tools for data mining and
 841 data analysis.")
 842     (license license:bsd-3)))
 843
 844 (define-public python2-scikit-learn
 845   (package-with-python2 python-scikit-learn))
 846
 847 (define-public python-autograd
 848   (let* ((commit "442205dfefe407beffb33550846434baa90c4de7")
 849          (revision "0")
 850          (version (git-version "0.0.0" revision commit)))
 851     (package
 852       (name "python-autograd")
 853       (home-page "https://github.com/HIPS/autograd")
 854       (source (origin
 855                 (method git-fetch)
 856                 (uri (git-reference
 857                       (url home-page)
 858                       (commit commit)))
 859                 (sha256
 860                  (base32
 861                   "189sv2xb0mwnjawa9z7mrgdglc1miaq93pnck26r28fi1jdwg0z4"))
 862                 (file-name (git-file-name name version))))
 863       (version version)
 864       (build-system python-build-system)
 865       (native-inputs
 866        `(("python-nose" ,python-nose)
 867          ("python-pytest" ,python-pytest)))
 868       (propagated-inputs
 869        `(("python-future" ,python-future)
 870          ("python-numpy" ,python-numpy)))
 871       (arguments
 872        `(#:phases (modify-phases %standard-phases
 873                     (replace 'check
 874                       (lambda _
 875                         (invoke "py.test" "-v"))))))
 876       (synopsis "Efficiently computes derivatives of NumPy code")
 877       (description "Autograd can automatically differentiate native Python and
 878 NumPy code.  It can handle a large subset of Python's features, including loops,
 879 ifs, recursion and closures, and it can even take derivatives of derivatives
 880 of derivatives.  It supports reverse-mode differentiation
 881 (a.k.a. backpropagation), which means it can efficiently take gradients of
 882 scalar-valued functions with respect to array-valued arguments, as well as
 883 forward-mode differentiation, and the two can be composed arbitrarily.  The
 884 main intended application of Autograd is gradient-based optimization.")
 885       (license license:expat))))
 886
 887 (define-public python2-autograd
 888   (package-with-python2 python-autograd))
 889
 890 (define-public lightgbm
 891   (package
 892     (name "lightgbm")
 893     (version "2.0.12")
 894     (source (origin
 895               (method url-fetch)
 896               (uri (string-append
 897                     "https://github.com/Microsoft/LightGBM/archive/v"
 898                     version ".tar.gz"))
 899               (sha256
 900                (base32
 901                 "132zf0yk0545mg72hyzxm102g3hpb6ixx9hnf8zd2k55gas6cjj1"))
 902               (file-name (string-append name "-" version ".tar.gz"))))
 903     (native-inputs
 904      `(("python-pytest" ,python-pytest)
 905        ("python-nose" ,python-nose)))
 906     (inputs
 907      `(("openmpi" ,openmpi)))
 908     (propagated-inputs
 909      `(("python-numpy" ,python-numpy)
 910        ("python-scipy" ,python-scipy)))
 911     (arguments
 912      `(#:configure-flags
 913        '("-DUSE_MPI=ON")
 914        #:phases
 915        (modify-phases %standard-phases
 916          (replace 'check
 917            (lambda* (#:key outputs #:allow-other-keys)
 918              (with-directory-excursion ,(string-append "../LightGBM-" version)
 919                (invoke "pytest" "tests/c_api_test/test_.py")))))))
 920     (build-system cmake-build-system)
 921     (home-page "https://github.com/Microsoft/LightGBM")
 922     (synopsis "Gradient boosting framework based on decision tree algorithms")
 923     (description "LightGBM is a gradient boosting framework that uses tree
 924 based learning algorithms.  It is designed to be distributed and efficient with
 925 the following advantages:
 926
 927 @itemize
 928 @item Faster training speed and higher efficiency
 929 @item Lower memory usage
 930 @item Better accuracy
 931 @item Parallel and GPU learning supported (not enabled in this package)
 932 @item Capable of handling large-scale data
 933 @end itemize\n")
 934     (license license:expat)))
 935
 936 (define-public vowpal-wabbit
 937   ;; Language bindings not included.
 938   (package
 939     (name "vowpal-wabbit")
 940     (version "8.5.0")
 941     (source (origin
 942               (method url-fetch)
 943               (uri (string-append
 944                     "https://github.com/JohnLangford/vowpal_wabbit/archive/"
 945                     version ".tar.gz"))
 946               (sha256
 947                (base32
 948                 "0clp2kb7rk5sckhllxjr5a651awf4s8dgzg4659yh4hf5cqnf0gr"))
 949               (file-name (string-append name "-" version ".tar.gz"))))
 950     (inputs
 951      `(("boost" ,boost)
 952        ("zlib" ,zlib)))
 953     (arguments
 954      `(#:configure-flags
 955        (list (string-append "--with-boost="
 956                             (assoc-ref %build-inputs "boost")))))
 957     (build-system gnu-build-system)
 958     (home-page "https://github.com/JohnLangford/vowpal_wabbit")
 959     (synopsis "Fast machine learning library for online learning")
 960     (description "Vowpal Wabbit is a machine learning system with techniques
 961 such as online, hashing, allreduce, reductions, learning2search, active, and
 962 interactive learning.")
 963     (license license:bsd-3)))
 964
 965 (define-public python2-fastlmm
 966   (package
 967     (name "python2-fastlmm")
 968     (version "0.2.21")
 969     (source
 970      (origin
 971        (method url-fetch)
 972        (uri (pypi-uri "fastlmm" version ".zip"))
 973        (sha256
 974         (base32
 975          "1q8c34rpmwkfy3r4d5172pzdkpfryj561897z9r3x22gq7813x1m"))))
 976     (build-system python-build-system)
 977     (arguments
 978      `(#:tests? #f ; some test files are missing
 979        #:python ,python-2)) ; only Python 2.7 is supported
 980     (propagated-inputs
 981      `(("python2-numpy" ,python2-numpy)
 982        ("python2-scipy" ,python2-scipy)
 983        ("python2-matplotlib" ,python2-matplotlib)
 984        ("python2-pandas" ,python2-pandas)
 985        ("python2-scikit-learn" ,python2-scikit-learn)
 986        ("python2-pysnptools" ,python2-pysnptools)))
 987     (native-inputs
 988      `(("unzip" ,unzip)
 989        ("python2-cython" ,python2-cython)
 990        ("python2-mock" ,python2-mock)
 991        ("python2-nose" ,python2-nose)))
 992     (home-page "http://research.microsoft.com/en-us/um/redmond/projects/mscompbio/fastlmm/")
 993     (synopsis "Perform genome-wide association studies on large data sets")
 994     (description
 995      "FaST-LMM, which stands for Factored Spectrally Transformed Linear Mixed
 996 Models, is a program for performing both single-SNP and SNP-set genome-wide
 997 association studies (GWAS) on extremely large data sets.")
 998     (license license:asl2.0)))
 999
1000 ;; There have been no proper releases yet.
1001 (define-public kaldi
1002   (let ((commit "2f95609f0bb085bd3a1dc5eb0a39f3edea59e606")
1003         (revision "1"))
1004     (package
1005       (name "kaldi")
1006       (version (git-version "0" revision commit))
1007       (source (origin
1008                 (method git-fetch)
1009                 (uri (git-reference
1010                       (url "https://github.com/kaldi-asr/kaldi.git")
1011                       (commit commit)))
1012                 (file-name (git-file-name name version))
1013                 (sha256
1014                  (base32
1015                   "082qh3pfi7hvncylp4xsmkfahbd7gb0whdfa4rwrx7fxk9rdh3kz"))))
1016       (build-system gnu-build-system)
1017       (arguments
1018        `(#:test-target "test"
1019          #:phases
1020          (modify-phases %standard-phases
1021            (add-after 'unpack 'chdir
1022              (lambda _ (chdir "src") #t))
1023            (replace 'configure
1024              (lambda* (#:key build system inputs outputs #:allow-other-keys)
1025                (when (not (or (string-prefix? "x86_64" system)
1026                               (string-prefix? "i686" system)))
1027                  (substitute* "makefiles/linux_openblas.mk"
1028                    (("-msse -msse2") "")))
1029                (substitute* "makefiles/default_rules.mk"
1030                  (("/bin/bash") (which "bash")))
1031                (substitute* "Makefile"
1032                  (("ext_depend: check_portaudio")
1033                   "ext_depend:"))
1034                (substitute* '("online/Makefile"
1035                               "onlinebin/Makefile"
1036                               "gst-plugin/Makefile")
1037                  (("../../tools/portaudio/install")
1038                   (assoc-ref inputs "portaudio")))
1039
1040                ;; This `configure' script doesn't support variables passed as
1041                ;; arguments, nor does it support "prefix".
1042                (let ((out (assoc-ref outputs "out"))
1043                      (openblas (assoc-ref inputs "openblas"))
1044                      (openfst (assoc-ref inputs "openfst")))
1045                  (substitute* "configure"
1046                    (("check_for_slow_expf;") "")
1047                    ;; This affects the RPATH and also serves as the installation
1048                    ;; directory.
1049                    (("KALDILIBDIR=`pwd`/lib")
1050                     (string-append "KALDILIBDIR=" out "/lib")))
1051                  (mkdir-p out) ; must exist
1052                  (setenv "CONFIG_SHELL" (which "bash"))
1053                  (setenv "OPENFST_VER" ,(package-version openfst))
1054                  (invoke "./configure"
1055                          "--use-cuda=no"
1056                          "--shared"
1057                          (string-append "--openblas-root=" openblas)
1058                          (string-append "--fst-root=" openfst)))))
1059            (add-after 'build 'build-ext-and-gstreamer-plugin
1060              (lambda _
1061                (invoke "make" "-C" "online" "depend")
1062                (invoke "make" "-C" "online")
1063                (invoke "make" "-C" "onlinebin" "depend")
1064                (invoke "make" "-C" "onlinebin")
1065                (invoke "make" "-C" "gst-plugin" "depend")
1066                (invoke "make" "-C" "gst-plugin")
1067                #t))
1068            ;; TODO: also install the executables.
1069            (replace 'install
1070              (lambda* (#:key outputs #:allow-other-keys)
1071                (let* ((out (assoc-ref outputs "out"))
1072                       (inc (string-append out "/include"))
1073                       (lib (string-append out "/lib")))
1074                  (mkdir-p lib)
1075                  ;; The build phase installed symlinks to the actual
1076                  ;; libraries.  Install the actual targets.
1077                  (for-each (lambda (file)
1078                              (let ((target (readlink file)))
1079                                (delete-file file)
1080                                (install-file target lib)))
1081                            (find-files lib "\\.so"))
1082                  ;; Install headers
1083                  (for-each (lambda (file)
1084                              (let ((target-dir (string-append inc "/" (dirname file))))
1085                                (install-file file target-dir)))
1086                            (find-files "." "\\.h"))
1087                  (install-file "gst-plugin/libgstonlinegmmdecodefaster.so"
1088                                (string-append lib "/gstreamer-1.0"))
1089                  #t))))))
1090       (inputs
1091        `(("alsa-lib" ,alsa-lib)
1092          ("gfortran" ,gfortran "lib")
1093          ("glib" ,glib)
1094          ("gstreamer" ,gstreamer)
1095          ("jack" ,jack-1)
1096          ("openblas" ,openblas)
1097          ("openfst" ,openfst)
1098          ("portaudio" ,portaudio)
1099          ("python" ,python)))
1100       (native-inputs
1101        `(("glib" ,glib "bin")             ; glib-genmarshal
1102          ("grep" ,grep)
1103          ("sed" ,sed)
1104          ("pkg-config" ,pkg-config)
1105          ("which" ,which)))
1106       (home-page "https://kaldi-asr.org/")
1107       (synopsis "Speech recognition toolkit")
1108       (description "Kaldi is an extensible toolkit for speech recognition
1109 written in C++.")
1110       (license license:asl2.0))))
1111
1112 (define-public gst-kaldi-nnet2-online
1113   (let ((commit "617e43e73c7cc45eb9119028c02bd4178f738c4a")
1114         (revision "1"))
1115     (package
1116       (name "gst-kaldi-nnet2-online")
1117       (version (git-version "0" revision commit))
1118       (source (origin
1119                 (method git-fetch)
1120                 (uri (git-reference
1121                       (url "https://github.com/alumae/gst-kaldi-nnet2-online.git")
1122                       (commit commit)))
1123                 (file-name (git-file-name name version))
1124                 (sha256
1125                  (base32
1126                   "0xh3w67b69818s6ib02ara4lw7wamjdmh4jznvkpzrs4skbs9jx9"))))
1127       (build-system gnu-build-system)
1128       (arguments
1129        `(#:tests? #f                    ; there are none
1130          #:make-flags
1131          (list (string-append "SHELL="
1132                               (assoc-ref %build-inputs "bash") "/bin/bash")
1133                (string-append "KALDI_ROOT="
1134                               (assoc-ref %build-inputs "kaldi-src"))
1135                (string-append "KALDILIBDIR="
1136                               (assoc-ref %build-inputs "kaldi") "/lib")
1137                "KALDI_FLAVOR=dynamic")
1138          #:phases
1139          (modify-phases %standard-phases
1140            (add-after 'unpack 'chdir
1141              (lambda _ (chdir "src") #t))
1142            (replace 'configure
1143              (lambda* (#:key inputs #:allow-other-keys)
1144                (let ((glib (assoc-ref inputs "glib")))
1145                  (setenv "CXXFLAGS" "-fPIC")
1146                  (setenv "CPLUS_INCLUDE_PATH"
1147                          (string-append glib "/include/glib-2.0:"
1148                                         glib "/lib/glib-2.0/include:"
1149                                         (assoc-ref inputs "gstreamer")
1150                                         "/include/gstreamer-1.0")))
1151                (substitute* "Makefile"
1152                  (("include \\$\\(KALDI_ROOT\\)/src/kaldi.mk") "")
1153                  (("\\$\\(error Cannot find") "#"))
1154                #t))
1155            (add-before 'build 'build-depend
1156              (lambda* (#:key make-flags #:allow-other-keys)
1157                (apply invoke "make" "depend" make-flags)))
1158            (replace 'install
1159              (lambda* (#:key outputs #:allow-other-keys)
1160                (let* ((out (assoc-ref outputs "out"))
1161                       (lib (string-append out "/lib/gstreamer-1.0")))
1162                  (install-file "libgstkaldinnet2onlinedecoder.so" lib)
1163                  #t))))))
1164       (inputs
1165        `(("glib" ,glib)
1166          ("gstreamer" ,gstreamer)
1167          ("jansson" ,jansson)
1168          ("openfst" ,openfst)
1169          ("kaldi" ,kaldi)))
1170       (native-inputs
1171        `(("bash" ,bash)
1172          ("glib:bin" ,glib "bin")       ; glib-genmarshal
1173          ("kaldi-src" ,(package-source kaldi))
1174          ("pkg-config" ,pkg-config)))
1175       (home-page "https://kaldi-asr.org/")
1176       (synopsis "Gstreamer plugin for decoding speech")
1177       (description "This package provides a GStreamer plugin that wraps
1178 Kaldi's @code{SingleUtteranceNnet2Decoder}.  It requires iVector-adapted DNN
1179 acoustic models.  The iVectors are adapted to the current audio stream
1180 automatically.")
1181       (license license:asl2.0))))
1182
1183 (define-public kaldi-gstreamer-server
1184   (let ((commit "1735ba49c5dc0ebfc184e45105fc600cd9f1f508")
1185         (revision "1"))
1186     (package
1187       (name "kaldi-gstreamer-server")
1188       (version (git-version "0" revision commit))
1189       (source (origin
1190                 (method git-fetch)
1191                 (uri (git-reference
1192                       (url "https://github.com/alumae/kaldi-gstreamer-server.git")
1193                       (commit commit)))
1194                 (file-name (git-file-name name version))
1195                 (sha256
1196                  (base32
1197                   "0j701m7lbwmzqxsfanj882v7881hrbmpqybbczbxqpcbg8q34w0k"))))
1198       (build-system gnu-build-system)
1199       (arguments
1200        `(#:tests? #f ; there are no tests that can be run automatically
1201          #:modules ((guix build utils)
1202                     (guix build gnu-build-system)
1203                     (srfi srfi-26))
1204          #:phases
1205          (modify-phases %standard-phases
1206            (delete 'configure)
1207            (replace 'build
1208              (lambda* (#:key outputs #:allow-other-keys)
1209                ;; Disable hash randomization to ensure the generated .pycs
1210                ;; are reproducible.
1211                (setenv "PYTHONHASHSEED" "0")
1212                (with-directory-excursion "kaldigstserver"
1213                  (for-each (lambda (file)
1214                              (apply invoke
1215                                     `("python"
1216                                       "-m" "compileall"
1217                                       "-f" ; force rebuild
1218                                       ,file)))
1219                            (find-files "." "\\.py$")))
1220                #t))
1221            (replace 'install
1222              (lambda* (#:key inputs outputs #:allow-other-keys)
1223                (let* ((out (assoc-ref outputs "out"))
1224                       (bin (string-append out "/bin"))
1225                       (share (string-append out "/share/kaldi-gstreamer-server/")))
1226                  ;; Install Python files
1227                  (with-directory-excursion "kaldigstserver"
1228                    (for-each (cut install-file <> share)
1229                              (find-files "." ".*")))
1230
1231                  ;; Install sample configuration files
1232                  (for-each (cut install-file <> share)
1233                            (find-files "." "\\.yaml"))
1234
1235                  ;; Install executables
1236                  (mkdir-p bin)
1237                  (let* ((server (string-append bin "/kaldi-gst-server"))
1238                         (client (string-append bin "/kaldi-gst-client"))
1239                         (worker (string-append bin "/kaldi-gst-worker"))
1240                         (PYTHONPATH (getenv "PYTHONPATH"))
1241                         (GST_PLUGIN_PATH (string-append
1242                                           (assoc-ref inputs "gst-kaldi-nnet2-online")
1243                                           "/lib/gstreamer-1.0:${GST_PLUGIN_PATH}"))
1244                         (wrap (lambda (wrapper what)
1245                                 (with-output-to-file wrapper
1246                                   (lambda _
1247                                     (format #t
1248                                             "#!~a
1249 export PYTHONPATH=~a
1250 export GST_PLUGIN_PATH=~a
1251 exec ~a ~a/~a \"$@\"~%"
1252                                             (which "bash") PYTHONPATH GST_PLUGIN_PATH
1253                                             (which "python") share what)))
1254                                 (chmod wrapper #o555))))
1255                    (for-each wrap
1256                              (list server client worker)
1257                              (list "master_server.py"
1258                                    "client.py"
1259                                    "worker.py")))
1260                  #t))))))
1261       (inputs
1262        `(("gst-kaldi-nnet2-online" ,gst-kaldi-nnet2-online)
1263          ("python2" ,python-2)
1264          ("python2-futures" ,python2-futures)
1265          ("python2-pygobject" ,python2-pygobject)
1266          ("python2-pyyaml" ,python2-pyyaml)
1267          ("python2-tornado" ,python2-tornado)
1268          ("python2-ws4py" ,python2-ws4py-for-kaldi-gstreamer-server)))
1269       (home-page "https://github.com/alumae/kaldi-gstreamer-server")
1270       (synopsis "Real-time full-duplex speech recognition server")
1271       (description "This is a real-time full-duplex speech recognition server,
1272 based on the Kaldi toolkit and the GStreamer framework and implemented in
1273 Python.")
1274       (license license:bsd-2))))
1275
1276 (define-public grpc
1277   (package
1278     (name "grpc")
1279     (version "1.16.1")
1280     (source (origin
1281               (method git-fetch)
1282               (uri (git-reference
1283                     (url "https://github.com/grpc/grpc.git")
1284                     (commit (string-append "v" version))))
1285               (file-name (git-file-name name version))
1286               (sha256
1287                (base32
1288                 "1jimqz3115f9pli5w6ik9wi7mjc7ix6y7yrq4a1ab9fc3dalj7p2"))))
1289     (build-system cmake-build-system)
1290     (arguments
1291      `(#:tests? #f ; no test target
1292        #:configure-flags
1293        (list "-DgRPC_ZLIB_PROVIDER=package"
1294              "-DgRPC_CARES_PROVIDER=package"
1295              "-DgRPC_SSL_PROVIDER=package"
1296              "-DgRPC_PROTOBUF_PROVIDER=package")))
1297     (inputs
1298      `(("c-ares" ,c-ares/cmake)
1299        ("openssl" ,openssl)
1300        ("zlib" ,zlib)))
1301     (native-inputs
1302      `(("protobuf" ,protobuf-next)
1303        ("python" ,python-wrapper)))
1304     (home-page "https://grpc.io")
1305     (synopsis "High performance universal RPC framework")
1306     (description "gRPC is a modern open source high performance @dfn{Remote
1307 Procedure Call} (RPC) framework that can run in any environment.  It can
1308 efficiently connect services in and across data centers with pluggable support
1309 for load balancing, tracing, health checking and authentication.  It is also
1310 applicable in last mile of distributed computing to connect devices, mobile
1311 applications and browsers to backend services.")
1312     (license license:asl2.0)))
1313
1314 ;; Note that Tensorflow includes a "third_party" directory, which seems to not
1315 ;; only contain modified subsets of upstream library source code, but also
1316 ;; adapter headers provided by Google (such as the fft.h header, which is not
1317 ;; part of the upstream project code).  The Tensorflow code includes headers
1318 ;; from the "third_party" directory.  It does not look like we can replace
1319 ;; these headers with unmodified upstream files, so we keep them.
1320 (define-public tensorflow
1321   (package
1322     (name "tensorflow")
1323     (version "1.9.0")
1324     (source
1325      (origin
1326        (method git-fetch)
1327        (uri (git-reference
1328              (url "https://github.com/tensorflow/tensorflow.git")
1329              (commit (string-append "v" version))))
1330        (file-name (string-append "tensorflow-" version "-checkout"))
1331        (sha256
1332         (base32
1333          "0a9kwha395g3wgxfwln5j8vn9nkspmd75xldrlqdq540w996g8xa"))))
1334     (build-system cmake-build-system)
1335     (arguments
1336      `(#:tests? #f                      ; no "check" target
1337        #:build-type "Release"
1338        #:configure-flags
1339        (let ((protobuf (assoc-ref %build-inputs "protobuf"))
1340              (protobuf:native (assoc-ref %build-inputs "protobuf:native"))
1341              (jsoncpp (assoc-ref %build-inputs "jsoncpp"))
1342              (snappy (assoc-ref %build-inputs "snappy"))
1343              (sqlite (assoc-ref %build-inputs "sqlite")))
1344          (list
1345           ;; Use protobuf from Guix
1346           (string-append "-Dprotobuf_STATIC_LIBRARIES="
1347                          protobuf "/lib/libprotobuf.so")
1348           (string-append "-DPROTOBUF_PROTOC_EXECUTABLE="
1349                          protobuf:native "/bin/protoc")
1350
1351           ;; Use snappy from Guix
1352           (string-append "-Dsnappy_STATIC_LIBRARIES="
1353                          snappy "/lib/libsnappy.so")
1354           ;; Yes, this is not actually the include directory but a prefix...
1355           (string-append "-Dsnappy_INCLUDE_DIR=" snappy)
1356
1357           ;; Use jsoncpp from Guix
1358           (string-append "-Djsoncpp_STATIC_LIBRARIES="
1359                          jsoncpp "/lib/libjsoncpp.so")
1360           ;; Yes, this is not actually the include directory but a prefix...
1361           (string-append "-Djsoncpp_INCLUDE_DIR=" jsoncpp)
1362
1363           ;; Use sqlite from Guix
1364           (string-append "-Dsqlite_STATIC_LIBRARIES="
1365                          sqlite "/lib/libsqlite.a")
1366
1367           ;; Use system libraries wherever possible.  Currently, this
1368           ;; only affects zlib.
1369           "-Dsystemlib_ALL=ON"
1370           "-Dtensorflow_ENABLE_POSITION_INDEPENDENT_CODE=ON"
1371           "-Dtensorflow_BUILD_SHARED_LIB=ON"
1372           "-Dtensorflow_OPTIMIZE_FOR_NATIVE_ARCH=OFF"
1373           "-Dtensorflow_ENABLE_SSL_SUPPORT=OFF"
1374           "-Dtensorflow_BUILD_CONTRIB_KERNELS=OFF"))
1375        #:make-flags
1376        (list "CC=gcc")
1377        #:modules ((ice-9 ftw)
1378                   (guix build utils)
1379                   (guix build cmake-build-system))
1380        #:phases
1381        (modify-phases %standard-phases
1382          (add-after 'unpack 'set-source-file-times-to-1980
1383            ;; At the end of the tf_python_build_pip_package target, a ZIP
1384            ;; archive should be generated via bdist_wheel, but it fails with
1385            ;; "ZIP does not support timestamps before 1980".  Luckily,
1386            ;; SOURCE_DATE_EPOCH is respected, which we set to some time in
1387            ;; 1980.
1388            (lambda _ (setenv "SOURCE_DATE_EPOCH" "315532800") #t))
1389          ;; See https://github.com/tensorflow/tensorflow/issues/20517#issuecomment-406373913
1390          (add-after 'unpack 'python3.7-compatibility
1391            (lambda _
1392              (substitute* '("tensorflow/python/eager/pywrap_tfe_src.cc"
1393                             "tensorflow/python/lib/core/ndarray_tensor.cc"
1394                             "tensorflow/python/lib/core/py_func.cc")
1395                (("PyUnicode_AsUTF8") "(char *)PyUnicode_AsUTF8"))
1396              (substitute* "tensorflow/c/eager/c_api.h"
1397                (("unsigned char async")
1398                 "unsigned char is_async"))
1399
1400              ;; Remove dependency on tensorboard, a complicated but probably
1401              ;; optional package.
1402              (substitute* "tensorflow/tools/pip_package/setup.py"
1403                ((".*'tensorboard >.*") ""))
1404              #t))
1405          (add-after 'python3.7-compatibility 'chdir
1406            (lambda _ (chdir "tensorflow/contrib/cmake") #t))
1407          (add-after 'chdir 'disable-downloads
1408            (lambda* (#:key inputs #:allow-other-keys)
1409              (substitute* (find-files "external" "\\.cmake$")
1410                (("GIT_REPOSITORY.*") "")
1411                (("GIT_TAG.*") "")
1412                (("PREFIX ")
1413                 "DOWNLOAD_COMMAND \"\"\nPREFIX "))
1414
1415              ;; Use packages from Guix
1416              (let ((grpc (assoc-ref inputs "grpc")))
1417                (substitute* "CMakeLists.txt"
1418                  ;; Sqlite
1419                  (("include\\(sqlite\\)") "")
1420                  (("\\$\\{sqlite_STATIC_LIBRARIES\\}")
1421                   (string-append (assoc-ref inputs "sqlite")
1422                                  "/lib/libsqlite3.so"))
1423                  (("sqlite_copy_headers_to_destination") "")
1424
1425                  ;; PNG
1426                  (("include\\(png\\)") "")
1427                  (("\\$\\{png_STATIC_LIBRARIES\\}")
1428                   (string-append (assoc-ref inputs "libpng")
1429                                  "/lib/libpng16.so"))
1430                  (("png_copy_headers_to_destination") "")
1431
1432                  ;; JPEG
1433                  (("include\\(jpeg\\)") "")
1434                  (("\\$\\{jpeg_STATIC_LIBRARIES\\}")
1435                   (string-append (assoc-ref inputs "libjpeg")
1436                                  "/lib/libjpeg.so"))
1437                  (("jpeg_copy_headers_to_destination") "")
1438
1439                  ;; GIF
1440                  (("include\\(gif\\)") "")
1441                  (("\\$\\{gif_STATIC_LIBRARIES\\}")
1442                   (string-append (assoc-ref inputs "giflib")
1443                                  "/lib/libgif.so"))
1444                  (("gif_copy_headers_to_destination") "")
1445
1446                  ;; lmdb
1447                  (("include\\(lmdb\\)") "")
1448                  (("\\$\\{lmdb_STATIC_LIBRARIES\\}")
1449                   (string-append (assoc-ref inputs "lmdb")
1450                                  "/lib/liblmdb.so"))
1451                  (("lmdb_copy_headers_to_destination") "")
1452
1453                  ;; Protobuf
1454                  (("include\\(protobuf\\)") "")
1455                  (("protobuf_copy_headers_to_destination") "")
1456                  (("^ +protobuf") "")
1457
1458                  ;; gRPC
1459                  (("include\\(grpc\\)")
1460                   "find_package(grpc REQUIRED NAMES gRPC)")
1461                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES grpc\\)") "")
1462
1463                  ;; Eigen
1464                  (("include\\(eigen\\)")
1465                   (string-append "find_package(eigen REQUIRED NAMES Eigen3)
1466 set(eigen_INCLUDE_DIRS ${CMAKE_CURRENT_BINARY_DIR}/external/eigen_archive "
1467                                  (assoc-ref inputs "eigen") "/include/eigen3)"))
1468                  (("^ +eigen") "")
1469
1470                  ;; snappy
1471                  (("include\\(snappy\\)")
1472                   "add_definitions(-DTF_USE_SNAPPY)")
1473                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES snappy\\)") "")
1474
1475                  ;; jsoncpp
1476                  (("include\\(jsoncpp\\)") "")
1477                  (("^ +jsoncpp") ""))
1478
1479                (substitute* "tf_core_framework.cmake"
1480                  ((" grpc") "")
1481                  (("\\$\\{GRPC_BUILD\\}/grpc_cpp_plugin")
1482                   (which "grpc_cpp_plugin"))
1483                  ;; Link with gRPC libraries
1484                  (("add_library\\(tf_protos_cc.*" m)
1485                   (string-append m
1486                                  (format #f "\ntarget_link_libraries(tf_protos_cc PRIVATE \
1487 ~a/lib/libgrpc++_unsecure.a \
1488 ~a/lib/libgrpc_unsecure.a \
1489 ~a/lib/libaddress_sorting.a \
1490 ~a/lib/libgpr.a \
1491 ~a//lib/libcares.so
1492 )\n"
1493                                          grpc grpc grpc grpc
1494                                          (assoc-ref inputs "c-ares"))))))
1495              (substitute* "tf_tools.cmake"
1496                (("add_dependencies\\(\\$\\{proto_text.*") ""))
1497              ;; Remove dependency on bundled grpc
1498              (substitute* "tf_core_distributed_runtime.cmake"
1499                (("tf_core_cpu grpc") "tf_core_cpu"))
1500
1501              ;; This directory is a dependency of many targets.
1502              (mkdir-p "protobuf")
1503              #t))
1504          (add-after 'configure 'unpack-third-party-sources
1505            (lambda* (#:key inputs #:allow-other-keys)
1506              ;; This is needed to configure bundled packages properly.
1507              (setenv "CONFIG_SHELL" (which "bash"))
1508              (for-each
1509               (lambda (name)
1510                 (let* ((what  (assoc-ref inputs (string-append name "-src")))
1511                        (name* (string-map (lambda (c)
1512                                             (if (char=? c #\-)
1513                                                 #\_ c)) name))
1514                        (where (string-append "../build/" name* "/src/" name*)))
1515                   (cond
1516                    ((string-suffix? ".zip" what)
1517                     (mkdir-p where)
1518                     (with-directory-excursion where
1519                       (invoke "unzip" what)))
1520                    ((string-suffix? ".tar.gz" what)
1521                     (mkdir-p where)
1522                     (invoke "tar" "xf" what
1523                             "-C" where "--strip-components=1"))
1524                    (else
1525                     (let ((parent (dirname where)))
1526                       (mkdir-p parent)
1527                       (with-directory-excursion parent
1528                         (when (file-exists? name*)
1529                           (delete-file-recursively name*))
1530                         (copy-recursively what name*)
1531                         (map make-file-writable
1532                              (find-files name* ".*"))))))))
1533               (list "boringssl"
1534                     "cub"
1535                     "double-conversion"
1536                     "farmhash"
1537                     "fft2d"
1538                     "highwayhash"
1539                     "nsync"
1540                     "re2"))
1541
1542              (rename-file "../build/cub/src/cub/cub-1.8.0/"
1543                           "../build/cub/src/cub/cub/")
1544              #t))
1545          (add-after 'unpack 'fix-python-build
1546            (lambda* (#:key inputs outputs #:allow-other-keys)
1547              (mkdir-p "protobuf-src")
1548              (invoke "tar" "xf" (assoc-ref inputs "protobuf:src")
1549                      "-C" "protobuf-src" "--strip-components=1")
1550              (mkdir-p "eigen-src")
1551              (invoke "tar" "xf" (assoc-ref inputs "eigen:src")
1552                      "-C" "eigen-src" "--strip-components=1")
1553
1554              (substitute* "tensorflow/contrib/cmake/tf_python.cmake"
1555                ;; Ensure that all Python dependencies can be found at build time.
1556                (("PYTHONPATH=\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/tf_python" m)
1557                 (string-append m ":" (getenv "PYTHONPATH")))
1558                ;; Take protobuf source files from our source package.
1559                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/protobuf/src/protobuf/src/google")
1560                 (string-append (getcwd) "/protobuf-src/src/google")))
1561
1562              (substitute* '("tensorflow/contrib/cmake/tf_shared_lib.cmake"
1563                             "tensorflow/contrib/cmake/tf_python.cmake")
1564                ;; Take Eigen source files from our source package.
1565                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/eigen/src/eigen/")
1566                 (string-append (getcwd) "/eigen-src/"))
1567                ;; Take Eigen headers from our own package.
1568                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/external/eigen_archive")
1569                 (string-append (assoc-ref inputs "eigen") "/include/eigen3")))
1570
1571              ;; Correct the RUNPATH of ops libraries generated for Python.
1572              ;; TODO: this doesn't work :(
1573              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1574              ;; warning: RUNPATH contains bogus entries: ("/tmp/guix-build-tensorflow-1.9.0.drv-0/source/tensorflow/contrib/build")
1575              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1576              ;; error: depends on 'libpywrap_tensorflow_internal.so', which
1577              ;; cannot be found in RUNPATH ...
1578              (substitute* "tensorflow/contrib/cmake/tf_cc_ops.cmake"
1579                (("set_target_properties.*")
1580                 (string-append "set_target_properties(${_AT_TARGET} PROPERTIES \
1581 COMPILE_FLAGS ${target_compile_flags} \
1582 INSTALL_RPATH_USE_LINK_PATH TRUE \
1583 INSTALL_RPATH " (assoc-ref outputs "out") "/lib)\n")))
1584              #t))
1585          (add-after 'build 'build-pip-package
1586            (lambda* (#:key outputs #:allow-other-keys)
1587              (setenv "LDFLAGS"
1588                      (string-append "-Wl,-rpath="
1589                                     (assoc-ref outputs "out") "/lib"))
1590              (invoke "make" "tf_python_build_pip_package")
1591              #t))
1592          (add-after 'build-pip-package 'install-python
1593            (lambda* (#:key outputs #:allow-other-keys)
1594              (let ((out (assoc-ref outputs "out"))
1595                    (wheel (car (find-files "../build/tf_python/dist/" "\\.whl$"))))
1596                (invoke "python" "-m" "pip" "install" wheel
1597                        (string-append "--prefix=" out))
1598
1599                ;; XXX: broken RUNPATH, see fix-python-build phase.
1600                (delete-file
1601                 (string-append
1602                  out "/lib/python3.7/site-packages/tensorflow/contrib/"
1603                  "seq2seq/python/ops/lib_beam_search_ops.so"))
1604                #t))))))
1605     (native-inputs
1606      `(("pkg-config" ,pkg-config)
1607        ("protobuf:native" ,protobuf-next) ; protoc
1608        ("protobuf:src" ,(package-source protobuf-next))
1609        ("eigen:src" ,(package-source eigen-for-tensorflow))
1610        ;; install_pip_packages.sh wants setuptools 39.1.0 specifically.
1611        ("python-setuptools" ,python-setuptools-for-tensorflow)
1612
1613        ;; The commit hashes and URLs for third-party source code are taken
1614        ;; from "tensorflow/workspace.bzl".
1615        ("boringssl-src"
1616         ,(let ((commit "ee7aa02")
1617                (revision "1"))
1618            (origin
1619              (method git-fetch)
1620              (uri (git-reference
1621                    (url "https://boringssl.googlesource.com/boringssl")
1622                    (commit commit)))
1623              (file-name (string-append "boringssl-0-" revision
1624                                        (string-take commit 7)
1625                                        "-checkout"))
1626              (sha256
1627               (base32
1628                "1jf693q0nw0adsic6cgmbdx6g7wr4rj4vxa8j1hpn792fqhd8wgw")))))
1629        ("cub-src"
1630         ,(let ((version "1.8.0"))
1631            (origin
1632              (method url-fetch)
1633              (uri (string-append "https://mirror.bazel.build/github.com/NVlabs/"
1634                                  "cub/archive/" version ".zip"))
1635              (file-name (string-append "cub-" version ".zip"))
1636              (sha256
1637               (base32
1638                "1hsqikqridb90dkxkjr2918dcry6pfh46ccnwrzawl56aamhdykb")))))
1639        ("double-conversion-src"
1640         ,(let ((commit "5664746")
1641                (revision "1"))
1642            (origin
1643              (method git-fetch)
1644              (uri (git-reference
1645                    (url "https://github.com/google/double-conversion.git")
1646                    (commit commit)))
1647              (file-name
1648               (git-file-name "double-conversion"
1649                              (string-append "0-" revision "."
1650                                             (string-take commit 7))))
1651              (sha256
1652               (base32
1653                "1h5lppqqxcvdg5jq42i5msgwx20ryij3apvmndflngrgdpc04gn1")))))
1654        ("farmhash-src"
1655         ,(let ((commit "816a4ae622e964763ca0862d9dbd19324a1eaf45"))
1656            (origin
1657              (method url-fetch)
1658              (uri (string-append
1659                    "https://mirror.bazel.build/github.com/google/farmhash/archive/"
1660                    commit ".tar.gz"))
1661              (file-name (string-append "farmhash-0-" (string-take commit 7)
1662                                        ".tar.gz"))
1663              (sha256
1664               (base32
1665                "185b2xdxl4d4cnsnv6abg8s22gxvx8673jq2yaq85bz4cdy58q35")))))
1666        ;; The license notice on the home page at
1667        ;; http://www.kurims.kyoto-u.ac.jp/~ooura/fft.html says:
1668        ;;   Copyright Takuya OOURA, 1996-2001
1669        ;;
1670        ;;   You may use, copy, modify and distribute this code for any purpose
1671        ;;   (include commercial use) and without fee. Please refer to this
1672        ;;   package when you modify this code.
1673        ;;
1674        ;; We take the identical tarball from the Bazel mirror, because the URL
1675        ;; at the home page is not versioned and might change.
1676        ("fft2d-src"
1677         ,(origin
1678            (method url-fetch)
1679            (uri "https://mirror.bazel.build/www.kurims.kyoto-u.ac.jp/~ooura/fft.tgz")
1680            (file-name "fft2d.tar.gz")
1681            (sha256
1682             (base32
1683              "15jjkfvhqvl2c0753d2di8hz0pyzn598g74wqy79awdrf1y67fsj"))))
1684        ("highwayhash-src"
1685         ,(let ((commit "be5edafc2e1a455768e260ccd68ae7317b6690ee")
1686                (revision "1"))
1687            (origin
1688              (method git-fetch)
1689              (uri (git-reference
1690                    (url "https://github.com/google/highwayhash.git")
1691                    (commit commit)))
1692              (file-name (string-append "highwayhash-0-" revision
1693                                        (string-take commit 7)
1694                                        "-checkout"))
1695              (sha256
1696               (base32
1697                "154jwf98cyy54hldr94pgjn85zynly3abpnc1avmb8a18lzwjyb6")))))
1698        ("nsync-src"
1699         ,(let ((version "0559ce013feac8db639ee1bf776aca0325d28777")
1700                (revision "1"))
1701            (origin
1702              (method url-fetch)
1703              (uri (string-append "https://mirror.bazel.build/"
1704                                  "github.com/google/nsync/archive/"
1705                                  version ".tar.gz"))
1706              (file-name (string-append "nsync-0." revision
1707                                        "-" (string-take version 7)
1708                                        ".tar.gz"))
1709              (sha256
1710               (base32
1711                "0qdkyqym34x739mmzv97ah5r7ph462v5xkxqxvidmcfqbi64b132")))))
1712        ("re2-src"
1713         ,(let ((commit "e7efc48")
1714                (revision "1"))
1715            (origin
1716              (method git-fetch)
1717              (uri (git-reference
1718                    (url "https://github.com/google/re2")
1719                    (commit commit)))
1720              (file-name (string-append "re2-0-" revision
1721                                        (string-take commit 7)
1722                                        "-checkout"))
1723              (sha256
1724               (base32
1725                "161g9841rjfsy5pn52fcis0s9hdr7rxvb06pad38j5rppfihvign")))))
1726        ("googletest" ,googletest)
1727        ("swig" ,swig)
1728        ("unzip" ,unzip)))
1729     (propagated-inputs
1730      `(("python-absl-py" ,python-absl-py)
1731        ("python-astor" ,python-astor)
1732        ("python-gast" ,python-gast)
1733        ("python-grpcio" ,python-grpcio)
1734        ("python-numpy" ,python-numpy)
1735        ("python-protobuf" ,python-protobuf-next)
1736        ("python-six" ,python-six)
1737        ("python-termcolo" ,python-termcolor)
1738        ("python-wheel" ,python-wheel)))
1739     (inputs
1740      `(("c-ares" ,c-ares)
1741        ("eigen" ,eigen-for-tensorflow)
1742        ("gemmlowp" ,gemmlowp-for-tensorflow)
1743        ("lmdb" ,lmdb)
1744        ("libjpeg" ,libjpeg)
1745        ("libpng" ,libpng)
1746        ("giflib" ,giflib)
1747        ("grpc" ,grpc)
1748        ("jsoncpp" ,jsoncpp-for-tensorflow)
1749        ("snappy" ,snappy)
1750        ("sqlite" ,sqlite)
1751        ("protobuf" ,protobuf-next)
1752        ("python" ,python-wrapper)
1753        ("zlib" ,zlib)))
1754     (home-page "https://tensorflow.org")
1755     (synopsis "Machine learning framework")
1756     (description
1757      "TensorFlow is a flexible platform for building and training machine
1758 learning models.  It provides a library for high performance numerical
1759 computation and includes high level Python APIs, including both a sequential
1760 API for beginners that allows users to build models quickly by plugging
1761 together building blocks and a subclassing API with an imperative style for
1762 advanced research.")
1763     (license license:asl2.0)))
1764
1765 (define-public python-iml
1766   (package
1767     (name "python-iml")
1768     (version "0.6.2")
1769     (source
1770      (origin
1771        (method url-fetch)
1772        (uri (pypi-uri "iml" version))
1773        (sha256
1774         (base32
1775          "1k8szlpm19rcwcxdny9qdm3gmaqq8akb4xlvrzyz8c2d679aak6l"))))
1776     (build-system python-build-system)
1777     (propagated-inputs
1778      `(("ipython" ,python-ipython)
1779        ("nose" ,python-nose)
1780        ("numpy" ,python-numpy)
1781        ("pandas" ,python-pandas)
1782        ("scipy" ,python-scipy)))
1783     (home-page "http://github.com/interpretable-ml/iml")
1784     (synopsis "Interpretable Machine Learning (iML) package")
1785     (description "Interpretable ML (iML) is a set of data type objects,
1786 visualizations, and interfaces that can be used by any method designed to
1787 explain the predictions of machine learning models (or really the output of
1788 any function).  It currently contains the interface and IO code from the Shap
1789 project, and it will potentially also do the same for the Lime project.")
1790     (license license:expat)))
1791
1792 (define-public python-keras-applications
1793   (package
1794     (name "python-keras-applications")
1795     (version "1.0.8")
1796     (source
1797      (origin
1798        (method url-fetch)
1799        (uri (pypi-uri "Keras_Applications" version))
1800        (sha256
1801         (base32
1802          "1rcz31ca4axa6kzhjx4lwqxbg4wvlljkj8qj9a7p9sfd5fhzjyam"))))
1803     (build-system python-build-system)
1804     ;; The tests require Keras, but this package is needed to build Keras.
1805     (arguments '(#:tests? #f))
1806     (propagated-inputs
1807      `(("python-h5py" ,python-h5py)
1808        ("python-numpy" ,python-numpy)))
1809     (native-inputs
1810      `(("python-pytest" ,python-pytest)
1811        ("python-pytest-cov" ,python-pytest-cov)
1812        ("python-pytest-pep8" ,python-pytest-pep8)
1813        ("python-pytest-xdist" ,python-pytest-xdist)))
1814     (home-page "https://github.com/keras-team/keras-applications")
1815     (synopsis "Reference implementations of popular deep learning models")
1816     (description
1817      "This package provides reference implementations of popular deep learning
1818 models for use with the Keras deep learning framework.")
1819     (license license:expat)))
1820
1821 (define-public python-keras-preprocessing
1822   (package
1823     (name "python-keras-preprocessing")
1824     (version "1.1.0")
1825     (source
1826      (origin
1827        (method url-fetch)
1828        (uri (pypi-uri "Keras_Preprocessing" version))
1829        (sha256
1830         (base32
1831          "1r98nm4k1svsqjyaqkfk23i31bl1kcfcyp7094yyj3c43phfp3as"))))
1832     (build-system python-build-system)
1833     (propagated-inputs
1834      `(("python-numpy" ,python-numpy)
1835        ("python-six" ,python-six)))
1836     (native-inputs
1837      `(("python-pandas" ,python-pandas)
1838        ("python-pillow" ,python-pillow)
1839        ("python-pytest" ,python-pytest)
1840        ("python-pytest-cov" ,python-pytest-cov)
1841        ("python-pytest-xdist" ,python-pytest-xdist)
1842        ("tensorflow" ,tensorflow)))
1843     (home-page "https://github.com/keras-team/keras-preprocessing/")
1844     (synopsis "Data preprocessing and augmentation for deep learning models")
1845     (description
1846      "Keras Preprocessing is the data preprocessing and data augmentation
1847 module of the Keras deep learning library.  It provides utilities for working
1848 with image data, text data, and sequence data.")
1849     (license license:expat)))
1850
1851 (define-public python-keras
1852   (package
1853     (name "python-keras")
1854     (version "2.2.4")
1855     (source
1856      (origin
1857        (method url-fetch)
1858        (uri (pypi-uri "Keras" version))
1859        (patches (search-patches "python-keras-integration-test.patch"))
1860        (sha256
1861         (base32
1862          "1j8bsqzh49vjdxy6l1k4iwax5vpjzniynyd041xjavdzvfii1dlh"))))
1863     (build-system python-build-system)
1864     (arguments
1865      `(#:phases
1866        (modify-phases %standard-phases
1867          (add-after 'unpack 'remove-tests-for-unavailable-features
1868            (lambda _
1869              (delete-file "keras/backend/theano_backend.py")
1870              (delete-file "keras/backend/cntk_backend.py")
1871              (delete-file "tests/keras/backend/backend_test.py")
1872
1873              ;; FIXME: This doesn't work because Tensorflow is missing the
1874              ;; coder ops library.
1875              (delete-file "tests/keras/test_callbacks.py")
1876              #t))
1877          (replace 'check
1878            (lambda _
1879              ;; These tests attempt to download data files from the internet.
1880              (delete-file "tests/integration_tests/test_datasets.py")
1881              (delete-file "tests/integration_tests/imagenet_utils_test.py")
1882
1883              (setenv "PYTHONPATH"
1884                      (string-append (getcwd) "/build/lib:"
1885                                     (getenv "PYTHONPATH")))
1886              (invoke "py.test" "-v"
1887                      "-p" "no:cacheprovider"
1888                      "--ignore" "keras/utils"))))))
1889     (propagated-inputs
1890      `(("python-h5py" ,python-h5py)
1891        ("python-keras-applications" ,python-keras-applications)
1892        ("python-keras-preprocessing" ,python-keras-preprocessing)
1893        ("python-numpy" ,python-numpy)
1894        ("python-pydot" ,python-pydot)
1895        ("python-pyyaml" ,python-pyyaml)
1896        ("python-scipy" ,python-scipy)
1897        ("python-six" ,python-six)
1898        ("tensorflow" ,tensorflow)
1899        ("graphviz" ,graphviz)))
1900     (native-inputs
1901      `(("python-pandas" ,python-pandas)
1902        ("python-pytest" ,python-pytest)
1903        ("python-pytest-cov" ,python-pytest-cov)
1904        ("python-pytest-pep8" ,python-pytest-pep8)
1905        ("python-pytest-timeout" ,python-pytest-timeout)
1906        ("python-pytest-xdist" ,python-pytest-xdist)
1907        ("python-sphinx" ,python-sphinx)
1908        ("python-requests" ,python-requests)))
1909     (home-page "https://github.com/keras-team/keras")
1910     (synopsis "High-level deep learning framework")
1911     (description "Keras is a high-level neural networks API, written in Python
1912 and capable of running on top of TensorFlow.  It was developed with a focus on
1913 enabling fast experimentation.  Use Keras if you need a deep learning library
1914 that:
1915
1916 @itemize
1917 @item Allows for easy and fast prototyping (through user friendliness,
1918   modularity, and extensibility).
1919 @item Supports both convolutional networks and recurrent networks, as well as
1920   combinations of the two.
1921 @item Runs seamlessly on CPU and GPU.
1922 @end itemize\n")
1923     (license license:expat)))