gnu/packages/machine-learning.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2015, 2016, 2017, 2018, 2019, 2020 Ricardo Wurmus <rekado@elephly.net>
   3 ;;; Copyright © 2016, 2020 Efraim Flashner <efraim@flashner.co.il>
   4 ;;; Copyright © 2016, 2017, 2020 Marius Bakke <mbakke@fastmail.com>
   5 ;;; Copyright © 2016 Hartmut Goebel <h.goebel@crazy-compilers.com>
   6 ;;; Copyright © 2018, 2019, 2020 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 ;;; Copyright © 2019, 2020 Guillaume Le Vaillant <glv@posteo.net>
  15 ;;; Copyright © 2019 Brett Gilio <brettg@gnu.org>
  16 ;;; Copyright © 2020 Konrad Hinsen <konrad.hinsen@fastmail.net>
  17 ;;; Copyright © 2020 Edouard Klein <edk@beaver-labs.com>
  18 ;;;
  19 ;;; This file is part of GNU Guix.
  20 ;;;
  21 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  22 ;;; under the terms of the GNU General Public License as published by
  23 ;;; the Free Software Foundation; either version 3 of the License, or (at
  24 ;;; your option) any later version.
  25 ;;;
  26 ;;; GNU Guix is distributed in the hope that it will be useful, but
  27 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  28 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  29 ;;; GNU General Public License for more details.
  30 ;;;
  31 ;;; You should have received a copy of the GNU General Public License
  32 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  33
  34 (define-module (gnu packages machine-learning)
  35   #:use-module ((guix licenses) #:prefix license:)
  36   #:use-module (guix packages)
  37   #:use-module (guix utils)
  38   #:use-module (guix download)
  39   #:use-module (guix svn-download)
  40   #:use-module (guix build-system cmake)
  41   #:use-module (guix build-system gnu)
  42   #:use-module (guix build-system ocaml)
  43   #:use-module (guix build-system python)
  44   #:use-module (guix build-system r)
  45   #:use-module (guix git-download)
  46   #:use-module (gnu packages)
  47   #:use-module (gnu packages adns)
  48   #:use-module (gnu packages algebra)
  49   #:use-module (gnu packages audio)
  50   #:use-module (gnu packages autotools)
  51   #:use-module (gnu packages base)
  52   #:use-module (gnu packages bash)
  53   #:use-module (gnu packages boost)
  54   #:use-module (gnu packages check)
  55   #:use-module (gnu packages compression)
  56   #:use-module (gnu packages cran)
  57   #:use-module (gnu packages databases)
  58   #:use-module (gnu packages dejagnu)
  59   #:use-module (gnu packages gcc)
  60   #:use-module (gnu packages glib)
  61   #:use-module (gnu packages graphviz)
  62   #:use-module (gnu packages gstreamer)
  63   #:use-module (gnu packages image)
  64   #:use-module (gnu packages linux)
  65   #:use-module (gnu packages maths)
  66   #:use-module (gnu packages mpi)
  67   #:use-module (gnu packages ocaml)
  68   #:use-module (gnu packages onc-rpc)
  69   #:use-module (gnu packages perl)
  70   #:use-module (gnu packages pkg-config)
  71   #:use-module (gnu packages protobuf)
  72   #:use-module (gnu packages python)
  73   #:use-module (gnu packages python-science)
  74   #:use-module (gnu packages python-web)
  75   #:use-module (gnu packages python-xyz)
  76   #:use-module (gnu packages rpc)
  77   #:use-module (gnu packages serialization)
  78   #:use-module (gnu packages sphinx)
  79   #:use-module (gnu packages statistics)
  80   #:use-module (gnu packages sqlite)
  81   #:use-module (gnu packages swig)
  82   #:use-module (gnu packages web)
  83   #:use-module (gnu packages xml)
  84   #:use-module (gnu packages xorg)
  85   #:use-module (ice-9 match))
  86
  87 (define-public fann
  88   ;; The last release is >100 commits behind, so we package from git.
  89   (let ((commit "d71d54788bee56ba4cf7522801270152da5209d7"))
  90     (package
  91       (name "fann")
  92       (version (string-append "2.2.0-1." (string-take commit 8)))
  93       (source (origin
  94                 (method git-fetch)
  95                 (uri (git-reference
  96                       (url "https://github.com/libfann/fann")
  97                       (commit commit)))
  98                 (file-name (string-append name "-" version "-checkout"))
  99                 (sha256
 100                  (base32
 101                   "0ibwpfrjs6q2lijs8slxjgzb2llcl6rk3v2ski4r6215g5jjhg3x"))))
 102       (build-system cmake-build-system)
 103       (arguments
 104        `(#:phases
 105          (modify-phases %standard-phases
 106            (replace 'check
 107              (lambda* (#:key outputs #:allow-other-keys)
 108                (let* ((out (assoc-ref outputs "out")))
 109                  (with-directory-excursion (string-append (getcwd) "/tests")
 110                    (invoke "./fann_tests"))))))))
 111       (home-page "http://leenissen.dk/fann/wp/")
 112       (synopsis "Fast Artificial Neural Network")
 113       (description
 114        "FANN is a neural network library, which implements multilayer
 115 artificial neural networks in C with support for both fully connected and
 116 sparsely connected networks.")
 117       (license license:lgpl2.1))))
 118
 119 (define-public libsvm
 120   (package
 121     (name "libsvm")
 122     (version "3.23")
 123     (source
 124      (origin
 125        (method url-fetch)
 126        (uri (string-append "https://www.csie.ntu.edu.tw/~cjlin/libsvm/"
 127                            name "-" version ".tar.gz"))
 128        (sha256
 129         (base32 "0jpaq0rr92x38p4nk3gjan79ip67m6p80anb28z1d8601miysyi5"))))
 130     (build-system gnu-build-system)
 131     (arguments
 132      `(#:tests? #f                      ; no "check" target
 133        #:phases (modify-phases %standard-phases
 134                   (delete 'configure)
 135                   (replace
 136                    'install             ; no ‘install’ target
 137                    (lambda* (#:key outputs #:allow-other-keys)
 138                      (let* ((out (assoc-ref outputs "out"))
 139                             (bin (string-append out "/bin/")))
 140                        (mkdir-p bin)
 141                        (for-each (lambda (file)
 142                                    (copy-file file (string-append bin file)))
 143                                  '("svm-train"
 144                                    "svm-predict"
 145                                    "svm-scale")))
 146                      #t)))))
 147     (home-page "https://www.csie.ntu.edu.tw/~cjlin/libsvm/")
 148     (synopsis "Library for Support Vector Machines")
 149     (description
 150      "LIBSVM is a machine learning library for support vector
 151 classification, (C-SVC, nu-SVC), regression (epsilon-SVR, nu-SVR) and
 152 distribution estimation (one-class SVM).  It supports multi-class
 153 classification.")
 154     (license license:bsd-3)))
 155
 156 (define-public python-libsvm
 157   (package (inherit libsvm)
 158     (name "python-libsvm")
 159     (build-system gnu-build-system)
 160     (arguments
 161      `(#:tests? #f                      ; no "check" target
 162        #:make-flags '("-C" "python")
 163        #:phases
 164        (modify-phases %standard-phases
 165          (delete 'configure)
 166          (replace
 167           'install                      ; no ‘install’ target
 168           (lambda* (#:key inputs outputs #:allow-other-keys)
 169             (let ((site (string-append (assoc-ref outputs "out")
 170                                        "/lib/python"
 171                                        (string-take
 172                                         (string-take-right
 173                                          (assoc-ref inputs "python") 5) 3)
 174                                        "/site-packages/")))
 175               (substitute* "python/svm.py"
 176                 (("../libsvm.so.2") "libsvm.so.2"))
 177               (mkdir-p site)
 178               (for-each (lambda (file)
 179                           (copy-file file (string-append site (basename file))))
 180                         (find-files "python" "\\.py"))
 181               (copy-file "libsvm.so.2"
 182                          (string-append site "libsvm.so.2")))
 183             #t)))))
 184     (inputs
 185      `(("python" ,python)))
 186     (synopsis "Python bindings of libSVM")))
 187
 188 (define-public ghmm
 189   ;; The latest release candidate is several years and a couple of fixes have
 190   ;; been published since.  This is why we download the sources from the SVN
 191   ;; repository.
 192   (let ((svn-revision 2341))
 193     (package
 194       (name "ghmm")
 195       (version (string-append "0.9-rc3-0." (number->string svn-revision)))
 196       (source (origin
 197                 (method svn-fetch)
 198                 (uri (svn-reference
 199                       (url "http://svn.code.sf.net/p/ghmm/code/trunk")
 200                       (revision svn-revision)))
 201                 (file-name (string-append name "-" version "-checkout"))
 202                 (sha256
 203                  (base32
 204                   "0qbq1rqp94l530f043qzp8aw5lj7dng9wq0miffd7spd1ff638wq"))))
 205       (build-system gnu-build-system)
 206       (arguments
 207        `(#:imported-modules (,@%gnu-build-system-modules
 208                              (guix build python-build-system))
 209          #:modules          ((guix build python-build-system)
 210                              ,@%gnu-build-system-modules)
 211          #:phases
 212          (modify-phases %standard-phases
 213            (add-after 'unpack 'enter-dir
 214              (lambda _ (chdir "ghmm") #t))
 215            (delete 'check)
 216            (add-after 'install 'check
 217              (assoc-ref %standard-phases 'check))
 218            (add-before 'check 'fix-PYTHONPATH
 219              (lambda* (#:key inputs outputs #:allow-other-keys)
 220                (let ((python-version (python-version
 221                                       (assoc-ref inputs "python"))))
 222                  (setenv "PYTHONPATH"
 223                          (string-append (getenv "PYTHONPATH")
 224                                         ":" (assoc-ref outputs "out")
 225                                         "/lib/python" python-version
 226                                         "/site-packages")))
 227                #t))
 228            (add-after 'enter-dir 'fix-runpath
 229              (lambda* (#:key outputs #:allow-other-keys)
 230                (substitute* "ghmmwrapper/setup.py"
 231                  (("^(.*)extra_compile_args = \\[" line indent)
 232                   (string-append indent
 233                                  "extra_link_args = [\"-Wl,-rpath="
 234                                  (assoc-ref outputs "out") "/lib\"],\n"
 235                                  line
 236                                  "\"-Wl,-rpath="
 237                                  (assoc-ref outputs "out")
 238                                  "/lib\", ")))
 239                #t))
 240            (add-after 'enter-dir 'disable-broken-tests
 241              (lambda _
 242                (substitute* "tests/Makefile.am"
 243                  ;; GHMM_SILENT_TESTS is assumed to be a command.
 244                  (("TESTS_ENVIRONMENT.*") "")
 245                  ;; Do not build broken tests.
 246                  (("chmm .*") "")
 247                  (("read_fa .*") "")
 248                  (("mcmc .*") "")
 249                  (("label_higher_order_test.*$")
 250                   "label_higher_order_test\n"))
 251
 252                ;; These Python unittests are broken as there is no gato.
 253                ;; See https://sourceforge.net/p/ghmm/support-requests/3/
 254                (substitute* "ghmmwrapper/ghmmunittests.py"
 255                  (("^(.*)def (testNewXML|testMultipleTransitionClasses|testNewXML)"
 256                    line indent)
 257                   (string-append indent
 258                                  "@unittest.skip(\"Disabled by Guix\")\n"
 259                                  line)))
 260                #t)))))
 261       (inputs
 262        `(("python" ,python-2) ; only Python 2 is supported
 263          ("libxml2" ,libxml2)))
 264       (native-inputs
 265        `(("pkg-config" ,pkg-config)
 266          ("dejagnu" ,dejagnu)
 267          ("swig" ,swig)
 268          ("autoconf" ,autoconf)
 269          ("automake" ,automake)
 270          ("libtool" ,libtool)))
 271       (home-page "http://ghmm.org")
 272       (synopsis "Hidden Markov Model library")
 273       (description
 274        "The General Hidden Markov Model library (GHMM) is a C library with
 275 additional Python bindings implementing a wide range of types of @dfn{Hidden
 276 Markov Models} (HMM) and algorithms: discrete, continuous emissions, basic
 277 training, HMM clustering, HMM mixtures.")
 278       (license license:lgpl2.0+))))
 279
 280 (define-public mcl
 281   (package
 282     (name "mcl")
 283     (version "14.137")
 284     (source (origin
 285               (method url-fetch)
 286               (uri (string-append
 287                     "http://micans.org/mcl/src/mcl-"
 288                     (string-replace-substring version "." "-")
 289                     ".tar.gz"))
 290               (sha256
 291                (base32
 292                 "15xlax3z31lsn62vlg94hkm75nm40q4679amnfg13jm8m2bnhy5m"))))
 293     (build-system gnu-build-system)
 294     (arguments
 295      `(#:configure-flags (list "--enable-blast")))
 296     (inputs
 297      `(("perl" ,perl)))
 298     (home-page "http://micans.org/mcl/")
 299     (synopsis "Clustering algorithm for graphs")
 300     (description
 301      "The MCL algorithm is short for the @dfn{Markov Cluster Algorithm}, a
 302 fast and scalable unsupervised cluster algorithm for graphs (also known as
 303 networks) based on simulation of (stochastic) flow in graphs.")
 304     ;; In the LICENCE file and web page it says "The software is licensed
 305     ;; under the GNU General Public License, version 3.", but in several of
 306     ;; the source code files it suggests GPL3 or later.
 307     ;; http://listserver.ebi.ac.uk/pipermail/mcl-users/2016/000376.html
 308     (license license:gpl3)))
 309
 310 (define-public ocaml-mcl
 311   (package
 312     (name "ocaml-mcl")
 313     (version "12-068oasis4")
 314     (source
 315      (origin
 316        (method git-fetch)
 317        (uri (git-reference
 318              (url "https://github.com/fhcrc/mcl")
 319              (commit version)))
 320        (file-name (git-file-name name version))
 321        (sha256
 322         (base32
 323          "0009dc3h2jp3qg5val452wngpqnbfyhbcxylghq0mrjqxx0jdq5p"))))
 324     (build-system ocaml-build-system)
 325     (arguments
 326      `(#:phases
 327        (modify-phases %standard-phases
 328          (add-before 'configure 'patch-paths
 329            (lambda _
 330              (substitute* "configure"
 331                (("/bin/sh") (which "sh")))
 332              (substitute* "setup.ml"
 333                (("LDFLAGS=-fPIC")
 334                 (string-append "LDFLAGS=-fPIC\"; \"SHELL=" (which "sh")))
 335                (("-std=c89") "-std=gnu99")
 336
 337                ;; This is a mutable string, which is no longer supported.  Use
 338                ;; a byte buffer instead.
 339                (("String.make \\(String.length s\\)")
 340                 "Bytes.make (String.length s)")
 341
 342                ;; These two belong together.
 343                (("OASISString.replace_chars")
 344                 "Bytes.to_string (OASISString.replace_chars")
 345                ((" s;")
 346                 " s);"))
 347              (substitute* "myocamlbuild.ml"
 348                (("std=c89") "std=gnu99"))
 349              ;; Since we build with a more recent OCaml, we have to use C99 or
 350              ;; later.  This causes problems with the old C code.
 351              (substitute* "src/impala/matrix.c"
 352                (("restrict") "restrict_"))
 353              #t)))))
 354     (native-inputs
 355      `(("ocamlbuild" ,ocamlbuild)))
 356     (home-page "https://github.com/fhcrc/mcl")
 357     (synopsis "OCaml wrappers around MCL")
 358     (description
 359      "This package provides OCaml bindings for the MCL graph clustering
 360 algorithm.")
 361     (license license:gpl3)))
 362
 363 (define-public randomjungle
 364   (package
 365     (name "randomjungle")
 366     (version "2.1.0")
 367     (source
 368      (origin
 369        (method url-fetch)
 370        (uri (string-append
 371              "https://www.imbs.uni-luebeck.de/fileadmin/files/Software"
 372              "/randomjungle/randomjungle-" version ".tar_.gz"))
 373        (patches (search-patches "randomjungle-disable-static-build.patch"))
 374        (sha256
 375         (base32
 376          "12c8rf30cla71swx2mf4ww9mfd8jbdw5lnxd7dxhyw1ygrvg6y4w"))))
 377     (build-system gnu-build-system)
 378     (arguments
 379      `(#:configure-flags
 380        (list "--disable-static"
 381              (string-append "--with-boost="
 382                             (assoc-ref %build-inputs "boost")))
 383        #:phases
 384        (modify-phases %standard-phases
 385          (add-before
 386           'configure 'set-CXXFLAGS
 387           (lambda _
 388             (setenv "CXXFLAGS" "-fpermissive ")
 389             #t)))))
 390     (inputs
 391      `(("boost" ,boost)
 392        ("gsl" ,gsl)
 393        ("libxml2" ,libxml2)
 394        ("zlib" ,zlib)))
 395     (native-inputs
 396      `(("gfortran" ,gfortran)
 397        ("gfortran:lib" ,gfortran "lib")))
 398     ;; Non-portable assembly instructions are used so building fails on
 399     ;; platforms other than x86_64 or i686.
 400     (supported-systems '("x86_64-linux" "i686-linux"))
 401     (home-page "https://www.imbs.uni-luebeck.de/forschung/software/details.html#c224")
 402     (synopsis "Implementation of the Random Forests machine learning method")
 403     (description
 404      "Random Jungle is an implementation of Random Forests.  It is supposed to
 405 analyse high dimensional data.  In genetics, it can be used for analysing big
 406 Genome Wide Association (GWA) data.  Random Forests is a powerful machine
 407 learning method.  Most interesting features are variable selection, missing
 408 value imputation, classifier creation, generalization error estimation and
 409 sample proximities between pairs of cases.")
 410     (license license:gpl3+)))
 411
 412 (define-public openfst
 413   (package
 414     (name "openfst")
 415     (version "1.7.2")
 416     (source (origin
 417               (method url-fetch)
 418               (uri (string-append "http://www.openfst.org/twiki/pub/FST/"
 419                                   "FstDownload/openfst-" version ".tar.gz"))
 420               (sha256
 421                (base32
 422                 "0fqgk8195kz21is09gwzwnrg7fr9526bi9mh4apyskapz27pbhr1"))))
 423     (build-system gnu-build-system)
 424     (home-page "http://www.openfst.org")
 425     (synopsis "Library for weighted finite-state transducers")
 426     (description "OpenFst is a library for constructing, combining,
 427 optimizing, and searching weighted finite-state transducers (FSTs).")
 428     (license license:asl2.0)))
 429
 430 (define-public shogun
 431   (package
 432     (name "shogun")
 433     (version "6.1.3")
 434     (source
 435      (origin
 436        (method url-fetch)
 437        (uri (string-append
 438              "ftp://shogun-toolbox.org/shogun/releases/"
 439              (version-major+minor version)
 440              "/sources/shogun-" version ".tar.bz2"))
 441        (sha256
 442         (base32
 443          "1rn9skm3nw6hr7mr3lgp2gfqhi7ii0lyxck7qmqnf8avq349s5jp"))
 444        (modules '((guix build utils)
 445                   (ice-9 rdelim)))
 446        (snippet
 447         '(begin
 448            ;; Remove non-free sources and files referencing them
 449            (for-each delete-file
 450                      (find-files "src/shogun/classifier/svm/"
 451                                  "SVMLight\\.(cpp|h)"))
 452            (for-each delete-file
 453                      (find-files "examples/undocumented/libshogun/"
 454                                  (string-append
 455                                   "(classifier_.*svmlight.*|"
 456                                   "evaluation_cross_validation_locked_comparison).cpp")))
 457            ;; Remove non-free functions.
 458            (define (delete-ifdefs file)
 459              (with-atomic-file-replacement file
 460                (lambda (in out)
 461                  (let loop ((line (read-line in 'concat))
 462                             (skipping? #f))
 463                    (if (eof-object? line)
 464                        #t
 465                        (let ((skip-next?
 466                               (or (and skipping?
 467                                        (not (string-prefix?
 468                                              "#endif //USE_SVMLIGHT" line)))
 469                                   (string-prefix?
 470                                    "#ifdef USE_SVMLIGHT" line))))
 471                          (when (or (not skipping?)
 472                                    (and skipping? (not skip-next?)))
 473                            (display line out))
 474                          (loop (read-line in 'concat) skip-next?)))))))
 475            (for-each delete-ifdefs
 476                      (append
 477                       (find-files "src/shogun/classifier/mkl"
 478                                   "^MKLClassification\\.cpp")
 479                       (find-files "src/shogun/classifier/svm"
 480                                   "^SVMLightOneClass\\.(cpp|h)")
 481                       (find-files "src/shogun/multiclass"
 482                                   "^ScatterSVM\\.(cpp|h)")
 483                       (find-files "src/shogun/kernel/"
 484                                   "^(Kernel|CombinedKernel|ProductKernel)\\.(cpp|h)")
 485                       (find-files "src/shogun/regression/svr"
 486                                   "^(MKLRegression|SVRLight)\\.(cpp|h)")
 487                       (find-files "src/shogun/transfer/domain_adaptation"
 488                                   "^DomainAdaptationSVM\\.(cpp|h)")))
 489            #t))))
 490     (build-system cmake-build-system)
 491     (arguments
 492      '(#:tests? #f ;no check target
 493        #:phases
 494        (modify-phases %standard-phases
 495          (add-after 'unpack 'delete-broken-symlinks
 496            (lambda _
 497              (for-each delete-file '("applications/arts/data"
 498                                      "applications/asp/data"
 499                                      "applications/easysvm/data"
 500                                      "applications/msplicer/data"
 501                                      "applications/ocr/data"
 502                                      "examples/meta/data"
 503                                      "examples/undocumented/data"))
 504              #t))
 505          (add-after 'unpack 'change-R-target-path
 506            (lambda* (#:key outputs #:allow-other-keys)
 507              (substitute* '("src/interfaces/r/CMakeLists.txt"
 508                             "examples/meta/r/CMakeLists.txt")
 509                (("\\$\\{R_COMPONENT_LIB_PATH\\}")
 510                 (string-append (assoc-ref outputs "out")
 511                                "/lib/R/library/")))
 512              #t))
 513          (add-after 'unpack 'fix-octave-modules
 514            (lambda* (#:key outputs #:allow-other-keys)
 515              (substitute* "src/interfaces/octave/CMakeLists.txt"
 516                (("^include_directories\\(\\$\\{OCTAVE_INCLUDE_DIRS\\}")
 517                 "include_directories(${OCTAVE_INCLUDE_DIRS} ${OCTAVE_INCLUDE_DIRS}/octave")
 518                ;; change target directory
 519                (("\\$\\{OCTAVE_OCT_LOCAL_API_FILE_DIR\\}")
 520                 (string-append (assoc-ref outputs "out")
 521                                "/share/octave/packages")))
 522              (substitute* '("src/interfaces/octave/swig_typemaps.i"
 523                             "src/interfaces/octave/sg_print_functions.cpp")
 524                ;; "octave/config.h" and "octave/oct-obj.h" deprecated in Octave.
 525                (("octave/config\\.h") "octave/octave-config.h")
 526                (("octave/oct-obj.h") "octave/ovl.h"))
 527              #t))
 528          (add-after 'unpack 'move-rxcpp
 529            (lambda* (#:key inputs #:allow-other-keys)
 530              (let ((rxcpp-dir "shogun/third-party/rxcpp"))
 531                (mkdir-p rxcpp-dir)
 532                (install-file (assoc-ref inputs "rxcpp") rxcpp-dir)
 533                #t)))
 534          (add-before 'build 'set-HOME
 535            ;; $HOME needs to be set at some point during the build phase
 536            (lambda _ (setenv "HOME" "/tmp") #t)))
 537        #:configure-flags
 538        (list "-DCMAKE_BUILD_WITH_INSTALL_RPATH=TRUE"
 539              "-DUSE_SVMLIGHT=OFF" ;disable proprietary SVMLIGHT
 540              "-DBUILD_META_EXAMPLES=OFF" ;requires unpackaged ctags
 541              ;;"-DINTERFACE_JAVA=ON" ;requires unpackaged jblas
 542              ;;"-DINTERFACE_RUBY=ON" ;requires unpackaged ruby-narray
 543              ;;"-DINTERFACE_PERL=ON" ;"FindPerlLibs" does not exist
 544              ;;"-DINTERFACE_LUA=ON"  ;fails because lua doesn't build pkgconfig file
 545              "-DINTERFACE_OCTAVE=ON"
 546              "-DINTERFACE_PYTHON=ON"
 547              "-DINTERFACE_R=ON")))
 548     (inputs
 549      `(("python" ,python)
 550        ("numpy" ,python-numpy)
 551        ("r-minimal" ,r-minimal)
 552        ("octave" ,octave-cli)
 553        ("swig" ,swig)
 554        ("eigen" ,eigen)
 555        ("hdf5" ,hdf5)
 556        ("atlas" ,atlas)
 557        ("arpack" ,arpack-ng)
 558        ("lapack" ,lapack)
 559        ("glpk" ,glpk)
 560        ("libxml2" ,libxml2)
 561        ("lzo" ,lzo)
 562        ("zlib" ,zlib)))
 563     (native-inputs
 564      `(("pkg-config" ,pkg-config)
 565        ("rxcpp" ,rxcpp)))
 566     ;; Non-portable SSE instructions are used so building fails on platforms
 567     ;; other than x86_64.
 568     (supported-systems '("x86_64-linux"))
 569     (home-page "https://shogun-toolbox.org/")
 570     (synopsis "Machine learning toolbox")
 571     (description
 572      "The Shogun Machine learning toolbox provides a wide range of unified and
 573 efficient Machine Learning (ML) methods.  The toolbox seamlessly
 574 combines multiple data representations, algorithm classes, and general purpose
 575 tools.  This enables both rapid prototyping of data pipelines and extensibility
 576 in terms of new algorithms.")
 577     (license license:gpl3+)))
 578
 579 (define-public rxcpp
 580   (package
 581     (name "rxcpp")
 582     (version "4.1.0")
 583     (source
 584      (origin
 585        (method git-fetch)
 586        (uri (git-reference
 587              (url "https://github.com/ReactiveX/RxCpp")
 588              (commit (string-append "v" version))))
 589        (sha256
 590         (base32 "1rdpa3jlc181jd08nk437aar085h28i45s6nzrv65apb3xyyz0ij"))
 591        (file-name (git-file-name name version))))
 592     (build-system cmake-build-system)
 593     (arguments
 594      `(#:phases
 595        (modify-phases %standard-phases
 596          (add-after 'unpack 'remove-werror
 597            (lambda _
 598              (substitute* (find-files ".")
 599                (("-Werror") ""))
 600              #t))
 601          (replace 'check
 602            (lambda _
 603              (invoke "ctest"))))))
 604     (native-inputs
 605      `(("catch" ,catch-framework)))
 606     (home-page "http://reactivex.io/")
 607     (synopsis "Reactive Extensions for C++")
 608     (description
 609      "The Reactive Extensions for C++ (RxCpp) is a library of algorithms for
 610 values-distributed-in-time.  ReactiveX is a library for composing asynchronous
 611 and event-based programs by using observable sequences.
 612
 613 It extends the observer pattern to support sequences of data and/or events and
 614 adds operators that allow you to compose sequences together declaratively while
 615 abstracting away concerns about things like low-level threading,
 616 synchronization, thread-safety, concurrent data structures, and non-blocking
 617 I/O.")
 618     (license license:asl2.0)))
 619
 620 (define-public r-adaptivesparsity
 621   (package
 622     (name "r-adaptivesparsity")
 623     (version "1.6")
 624     (source (origin
 625               (method url-fetch)
 626               (uri (cran-uri "AdaptiveSparsity" version))
 627               (sha256
 628                (base32
 629                 "0imr5m8mll9j6n4icsv6z9rl5kbnwsp9wvzrg7n90nnmcxq2cz91"))))
 630     (properties
 631      `((upstream-name . "AdaptiveSparsity")))
 632     (build-system r-build-system)
 633     (arguments
 634      `(#:phases
 635        (modify-phases %standard-phases
 636          (add-after 'unpack 'link-against-armadillo
 637            (lambda _
 638              (substitute* "src/Makevars"
 639                (("PKG_LIBS=" prefix)
 640                 (string-append prefix "-larmadillo"))))))))
 641     (propagated-inputs
 642      `(("r-mass" ,r-mass)
 643        ("r-matrix" ,r-matrix)
 644        ("r-rcpp" ,r-rcpp)
 645        ("r-rcpparmadillo" ,r-rcpparmadillo)))
 646     (inputs
 647      `(("armadillo" ,armadillo)))
 648     (home-page "https://cran.r-project.org/web/packages/AdaptiveSparsity")
 649     (synopsis "Adaptive sparsity models")
 650     (description
 651      "This package implements the Figueiredo machine learning algorithm for
 652 adaptive sparsity and the Wong algorithm for adaptively sparse gaussian
 653 geometric models.")
 654     (license license:lgpl3+)))
 655
 656 (define-public gemmlowp-for-tensorflow
 657   ;; The commit hash is taken from "tensorflow/workspace.bzl".
 658   (let ((commit "38ebac7b059e84692f53e5938f97a9943c120d98")
 659         (revision "2"))
 660     (package
 661       (name "gemmlowp")
 662       (version (git-version "0" revision commit))
 663       (source (origin
 664                 (method url-fetch)
 665                 (uri (string-append "https://mirror.bazel.build/"
 666                                     "github.com/google/gemmlowp/archive/"
 667                                     commit ".zip"))
 668                 (file-name (string-append "gemmlowp-" version ".zip"))
 669                 (sha256
 670                  (base32
 671                   "0n56s2g8hrssm4w8qj1v58gfm56a04n9v992ixkmvk6zjiralzxq"))))
 672       (build-system cmake-build-system)
 673       (arguments
 674        `(#:configure-flags
 675          (list ,@(match (%current-system)
 676                    ((or "x86_64-linux" "i686-linux")
 677                     '("-DCMAKE_CXX_FLAGS=-msse2"))
 678                    (_ '())))
 679          #:phases
 680          (modify-phases %standard-phases
 681            ;; This directory contains the CMakeLists.txt.
 682            (add-after 'unpack 'chdir
 683              (lambda _ (chdir "contrib") #t))
 684            ;; There is no install target
 685            (replace 'install
 686              (lambda* (#:key outputs #:allow-other-keys)
 687                (let* ((out (assoc-ref outputs "out"))
 688                       (lib (string-append out "/lib/"))
 689                       (inc (string-append out "/include/")))
 690                  (install-file "../build/libeight_bit_int_gemm.so" lib)
 691                  (for-each (lambda (dir)
 692                              (let ((target (string-append inc "/" dir)))
 693                                (mkdir-p target)
 694                                (for-each (lambda (h)
 695                                            (install-file h target))
 696                                          (find-files (string-append "../" dir)
 697                                                      "\\.h$"))))
 698                            '("meta" "profiling" "public" "fixedpoint"
 699                              "eight_bit_int_gemm" "internal"))
 700                  #t))))))
 701       (native-inputs
 702        `(("unzip" ,unzip)))
 703       (home-page "https://github.com/google/gemmlowp")
 704       (synopsis "Small self-contained low-precision GEMM library")
 705       (description
 706        "This is a small self-contained low-precision @dfn{general matrix
 707 multiplication} (GEMM) library.  It is not a full linear algebra library.
 708 Low-precision means that the input and output matrix entries are integers on
 709 at most 8 bits.  To avoid overflow, results are internally accumulated on more
 710 than 8 bits, and at the end only some significant 8 bits are kept.")
 711       (license license:asl2.0))))
 712
 713 (define-public dlib
 714   (package
 715     (name "dlib")
 716     (version "19.20")
 717     (source (origin
 718               (method url-fetch)
 719               (uri (string-append
 720                     "http://dlib.net/files/dlib-" version ".tar.bz2"))
 721               (sha256
 722                (base32
 723                 "139jyi19qz37wwmmy48gil9d1kkh2r3w3bwdzabha6ayxmba96nz"))
 724               (modules '((guix build utils)))
 725               (snippet
 726                '(begin
 727                   ;; Delete ~13MB of bundled dependencies.
 728                   (delete-file-recursively "dlib/external")
 729                   (delete-file-recursively "docs/dlib/external")
 730                   #t))))
 731     (build-system cmake-build-system)
 732     (arguments
 733      `(#:phases
 734        (modify-phases %standard-phases
 735          (add-after 'unpack 'disable-asserts
 736            (lambda _
 737              ;; config.h recommends explicitly enabling or disabling asserts
 738              ;; when building as a shared library. By default neither is set.
 739              (substitute* "dlib/config.h"
 740                (("^//#define DLIB_DISABLE_ASSERTS") "#define DLIB_DISABLE_ASSERTS"))
 741              #t))
 742          (add-after 'disable-asserts 'disable-failing-tests
 743            (lambda _
 744              ;; One test times out on MIPS, so we need to disable it.
 745              ;; Others are flaky on some platforms.
 746              (let* ((system ,(or (%current-target-system)
 747                                  (%current-system)))
 748                     (disabled-tests (cond
 749                                      ((string-prefix? "mips64" system)
 750                                       '("object_detector" ; timeout
 751                                         "data_io"))
 752                                      ((string-prefix? "armhf" system)
 753                                       '("learning_to_track"))
 754                                      ((string-prefix? "i686" system)
 755                                       '("optimization"))
 756                                      (else '()))))
 757                (for-each
 758                 (lambda (test)
 759                   (substitute* "dlib/test/makefile"
 760                     (((string-append "SRC \\+= " test "\\.cpp")) "")))
 761                 disabled-tests)
 762                #t)))
 763          (replace 'check
 764            (lambda _
 765              ;; No test target, so we build and run the unit tests here.
 766              (let ((test-dir (string-append "../dlib-" ,version "/dlib/test")))
 767                (with-directory-excursion test-dir
 768                  (invoke "make" "-j" (number->string (parallel-job-count)))
 769                  (invoke "./dtest" "--runall"))
 770                #t)))
 771          (add-after 'install 'delete-static-library
 772            (lambda* (#:key outputs #:allow-other-keys)
 773              (delete-file (string-append (assoc-ref outputs "out")
 774                                          "/lib/libdlib.a"))
 775              #t)))))
 776     (native-inputs
 777      `(("pkg-config" ,pkg-config)
 778        ;; For tests.
 779        ("libnsl" ,libnsl)))
 780     (inputs
 781      `(("giflib" ,giflib)
 782        ("lapack" ,lapack)
 783        ("libjpeg" ,libjpeg-turbo)
 784        ("libpng" ,libpng)
 785        ("libx11" ,libx11)
 786        ("openblas" ,openblas)
 787        ("zlib" ,zlib)))
 788     (synopsis
 789      "Toolkit for making machine learning and data analysis applications in C++")
 790     (description
 791      "Dlib is a modern C++ toolkit containing machine learning algorithms and
 792 tools.  It is used in both industry and academia in a wide range of domains
 793 including robotics, embedded devices, mobile phones, and large high performance
 794 computing environments.")
 795     (home-page "http://dlib.net")
 796     (license license:boost1.0)))
 797
 798 (define-public python-scikit-learn
 799   (package
 800     (name "python-scikit-learn")
 801     (version "0.22.1")
 802     (source
 803      (origin
 804        (method git-fetch)
 805        (uri (git-reference
 806              (url "https://github.com/scikit-learn/scikit-learn")
 807              (commit version)))
 808        (file-name (git-file-name name version))
 809        (sha256
 810         (base32
 811          "1xqxv210gsmjw094vc5ghq2y9lmm74qkk22pq6flcjzj51b86jxf"))))
 812     (build-system python-build-system)
 813     (arguments
 814      `(#:phases
 815        (modify-phases %standard-phases
 816          (add-after 'build 'build-ext
 817            (lambda _ (invoke "python" "setup.py" "build_ext" "--inplace") #t))
 818          (replace 'check
 819            (lambda _
 820              ;; Restrict OpenBLAS threads to prevent segfaults while testing!
 821              (setenv "OPENBLAS_NUM_THREADS" "1")
 822
 823              ;; Some tests require write access to $HOME.
 824              (setenv "HOME" "/tmp")
 825
 826              (invoke "pytest" "sklearn" "-m" "not network")))
 827          (add-before 'reset-gzip-timestamps 'make-files-writable
 828            (lambda* (#:key outputs #:allow-other-keys)
 829              ;; Make sure .gz files are writable so that the
 830              ;; 'reset-gzip-timestamps' phase can do its work.
 831              (let ((out (assoc-ref outputs "out")))
 832                (for-each make-file-writable
 833                          (find-files out "\\.gz$"))
 834                #t))))))
 835     (inputs
 836      `(("openblas" ,openblas)))
 837     (native-inputs
 838      `(("python-pytest" ,python-pytest)
 839        ("python-pandas" ,python-pandas) ;for tests
 840        ("python-cython" ,python-cython)))
 841     (propagated-inputs
 842      `(("python-numpy" ,python-numpy)
 843        ("python-scipy" ,python-scipy)
 844        ("python-joblib" ,python-joblib)))
 845     (home-page "https://scikit-learn.org/")
 846     (synopsis "Machine Learning in Python")
 847     (description
 848      "Scikit-learn provides simple and efficient tools for data mining and
 849 data analysis.")
 850     (properties `((python2-variant . ,(delay python2-scikit-learn))))
 851     (license license:bsd-3)))
 852
 853 ;; scikit-learn 0.22 and later only supports Python 3, so we stick with
 854 ;; an older version here.
 855 (define-public python2-scikit-learn
 856   (let ((base (package-with-python2 (strip-python2-variant python-scikit-learn))))
 857     (package
 858       (inherit base)
 859       (version "0.20.4")
 860       (source (origin
 861                 (method git-fetch)
 862                 (uri (git-reference
 863                       (url "https://github.com/scikit-learn/scikit-learn")
 864                       (commit version)))
 865                 (file-name (git-file-name "python-scikit-learn" version))
 866                 (sha256
 867                  (base32
 868                   "08zbzi8yx5wdlxfx9jap61vg1malc9ajf576w7a0liv6jvvrxlpj")))))))
 869
 870 (define-public python-scikit-rebate
 871   (package
 872     (name "python-scikit-rebate")
 873     (version "0.6")
 874     (source (origin
 875               (method url-fetch)
 876               (uri (pypi-uri "skrebate" version))
 877               (sha256
 878                (base32
 879                 "1h7qs9gjxpzqabzhb8rmpv3jpmi5iq41kqdibg48299h94iikiw7"))))
 880     (build-system python-build-system)
 881     ;; Pandas is only needed to run the tests.
 882     (native-inputs
 883      `(("python-pandas" ,python-pandas)))
 884     (propagated-inputs
 885      `(("python-numpy" ,python-numpy)
 886        ("python-scipy" ,python-scipy)
 887        ("python-scikit-learn" ,python-scikit-learn)
 888        ("python-joblib" ,python-joblib)))
 889     (home-page "https://epistasislab.github.io/scikit-rebate/")
 890     (synopsis "Relief-based feature selection algorithms for Python")
 891     (description "Scikit-rebate is a scikit-learn-compatible Python
 892 implementation of ReBATE, a suite of Relief-based feature selection algorithms
 893 for Machine Learning.  These algorithms excel at identifying features that are
 894 predictive of the outcome in supervised learning problems, and are especially
 895 good at identifying feature interactions that are normally overlooked by
 896 standard feature selection algorithms.")
 897     (license license:expat)))
 898
 899 (define-public python-autograd
 900   (let* ((commit "442205dfefe407beffb33550846434baa90c4de7")
 901          (revision "0")
 902          (version (git-version "0.0.0" revision commit)))
 903     (package
 904       (name "python-autograd")
 905       (home-page "https://github.com/HIPS/autograd")
 906       (source (origin
 907                 (method git-fetch)
 908                 (uri (git-reference
 909                       (url home-page)
 910                       (commit commit)))
 911                 (sha256
 912                  (base32
 913                   "189sv2xb0mwnjawa9z7mrgdglc1miaq93pnck26r28fi1jdwg0z4"))
 914                 (file-name (git-file-name name version))))
 915       (version version)
 916       (build-system python-build-system)
 917       (native-inputs
 918        `(("python-nose" ,python-nose)
 919          ("python-pytest" ,python-pytest)))
 920       (propagated-inputs
 921        `(("python-future" ,python-future)
 922          ("python-numpy" ,python-numpy)))
 923       (arguments
 924        `(#:phases (modify-phases %standard-phases
 925                     (replace 'check
 926                       (lambda _
 927                         (invoke "py.test" "-v"))))))
 928       (synopsis "Efficiently computes derivatives of NumPy code")
 929       (description "Autograd can automatically differentiate native Python and
 930 NumPy code.  It can handle a large subset of Python's features, including loops,
 931 ifs, recursion and closures, and it can even take derivatives of derivatives
 932 of derivatives.  It supports reverse-mode differentiation
 933 (a.k.a. backpropagation), which means it can efficiently take gradients of
 934 scalar-valued functions with respect to array-valued arguments, as well as
 935 forward-mode differentiation, and the two can be composed arbitrarily.  The
 936 main intended application of Autograd is gradient-based optimization.")
 937       (license license:expat))))
 938
 939 (define-public python2-autograd
 940   (package-with-python2 python-autograd))
 941
 942 (define-public lightgbm
 943   (package
 944     (name "lightgbm")
 945     (version "2.0.12")
 946     (source (origin
 947               (method git-fetch)
 948               (uri (git-reference
 949                      (url "https://github.com/Microsoft/LightGBM")
 950                      (commit (string-append "v" version))))
 951               (sha256
 952                (base32
 953                 "0jlvyn7k81dzrh9ij3zw576wbgiwmmr26rzpdxjn1dbpc3njpvzi"))
 954               (file-name (git-file-name name version))))
 955     (native-inputs
 956      `(("python-pytest" ,python-pytest)
 957        ("python-nose" ,python-nose)))
 958     (inputs
 959      `(("openmpi" ,openmpi)))
 960     (propagated-inputs
 961      `(("python-numpy" ,python-numpy)
 962        ("python-scipy" ,python-scipy)))
 963     (arguments
 964      `(#:configure-flags
 965        '("-DUSE_MPI=ON")
 966        #:phases
 967        (modify-phases %standard-phases
 968          (replace 'check
 969            (lambda _
 970              (with-directory-excursion "../source"
 971                (invoke "pytest" "tests/c_api_test/test_.py")))))))
 972     (build-system cmake-build-system)
 973     (home-page "https://github.com/Microsoft/LightGBM")
 974     (synopsis "Gradient boosting framework based on decision tree algorithms")
 975     (description "LightGBM is a gradient boosting framework that uses tree
 976 based learning algorithms.  It is designed to be distributed and efficient with
 977 the following advantages:
 978
 979 @itemize
 980 @item Faster training speed and higher efficiency
 981 @item Lower memory usage
 982 @item Better accuracy
 983 @item Parallel and GPU learning supported (not enabled in this package)
 984 @item Capable of handling large-scale data
 985 @end itemize\n")
 986     (license license:expat)))
 987
 988 (define-public vowpal-wabbit
 989   ;; Language bindings not included.
 990   (package
 991     (name "vowpal-wabbit")
 992     (version "8.5.0")
 993     (source (origin
 994               (method git-fetch)
 995               (uri (git-reference
 996                      (url "https://github.com/JohnLangford/vowpal_wabbit")
 997                      (commit version)))
 998               (sha256
 999                (base32
1000                 "04bwzk6ifgnz3fmzid8b7avxf9n5pnx9xcjm61nkjng1vv0bpj8x"))
1001               (file-name (git-file-name name version))))
1002     (inputs
1003      `(("boost" ,boost)
1004        ("zlib" ,zlib)))
1005     (arguments
1006      `(#:configure-flags
1007        (list (string-append "--with-boost="
1008                             (assoc-ref %build-inputs "boost")))
1009        #:phases
1010        (modify-phases %standard-phases
1011          (add-after 'unpack 'make-files-writable
1012            (lambda _
1013              (for-each make-file-writable (find-files "." ".*")) #t)))))
1014     (build-system gnu-build-system)
1015     (home-page "https://github.com/JohnLangford/vowpal_wabbit")
1016     (synopsis "Fast machine learning library for online learning")
1017     (description "Vowpal Wabbit is a machine learning system with techniques
1018 such as online, hashing, allreduce, reductions, learning2search, active, and
1019 interactive learning.")
1020     (license license:bsd-3)))
1021
1022 (define-public python2-fastlmm
1023   (package
1024     (name "python2-fastlmm")
1025     (version "0.2.21")
1026     (source
1027      (origin
1028        (method url-fetch)
1029        (uri (pypi-uri "fastlmm" version ".zip"))
1030        (sha256
1031         (base32
1032          "1q8c34rpmwkfy3r4d5172pzdkpfryj561897z9r3x22gq7813x1m"))))
1033     (build-system python-build-system)
1034     (arguments
1035      `(#:tests? #f ; some test files are missing
1036        #:python ,python-2)) ; only Python 2.7 is supported
1037     (propagated-inputs
1038      `(("python2-numpy" ,python2-numpy)
1039        ("python2-scipy" ,python2-scipy)
1040        ("python2-matplotlib" ,python2-matplotlib)
1041        ("python2-pandas" ,python2-pandas)
1042        ("python2-scikit-learn" ,python2-scikit-learn)
1043        ("python2-pysnptools" ,python2-pysnptools)))
1044     (native-inputs
1045      `(("unzip" ,unzip)
1046        ("python2-cython" ,python2-cython)
1047        ("python2-mock" ,python2-mock)
1048        ("python2-nose" ,python2-nose)))
1049     (home-page "http://research.microsoft.com/en-us/um/redmond/projects/mscompbio/fastlmm/")
1050     (synopsis "Perform genome-wide association studies on large data sets")
1051     (description
1052      "FaST-LMM, which stands for Factored Spectrally Transformed Linear Mixed
1053 Models, is a program for performing both single-SNP and SNP-set genome-wide
1054 association studies (GWAS) on extremely large data sets.")
1055     (license license:asl2.0)))
1056
1057 ;; There have been no proper releases yet.
1058 (define-public kaldi
1059   (let ((commit "d4791c0f3fc1a09c042dac365e120899ee2ad21e")
1060         (revision "2"))
1061     (package
1062       (name "kaldi")
1063       (version (git-version "0" revision commit))
1064       (source (origin
1065                 (method git-fetch)
1066                 (uri (git-reference
1067                       (url "https://github.com/kaldi-asr/kaldi")
1068                       (commit commit)))
1069                 (file-name (git-file-name name version))
1070                 (sha256
1071                  (base32
1072                   "07k80my6f19mhrkwbzhjsnpf9871wmrwkl0ym468i830w67qyjrz"))))
1073       (build-system gnu-build-system)
1074       (arguments
1075        `(#:test-target "test"
1076          #:phases
1077          (modify-phases %standard-phases
1078            (add-after 'unpack 'chdir
1079              (lambda _ (chdir "src") #t))
1080            (replace 'configure
1081              (lambda* (#:key build system inputs outputs #:allow-other-keys)
1082                (when (not (or (string-prefix? "x86_64" system)
1083                               (string-prefix? "i686" system)))
1084                  (substitute* "makefiles/linux_openblas.mk"
1085                    (("-msse -msse2") "")))
1086                (substitute* "makefiles/default_rules.mk"
1087                  (("/bin/bash") (which "bash")))
1088                (substitute* "Makefile"
1089                  (("ext_depend: check_portaudio")
1090                   "ext_depend:"))
1091                (substitute* '("online/Makefile"
1092                               "onlinebin/Makefile"
1093                               "gst-plugin/Makefile")
1094                  (("../../tools/portaudio/install")
1095                   (assoc-ref inputs "portaudio")))
1096
1097                ;; This `configure' script doesn't support variables passed as
1098                ;; arguments, nor does it support "prefix".
1099                (let ((out (assoc-ref outputs "out"))
1100                      (openblas (assoc-ref inputs "openblas"))
1101                      (openfst (assoc-ref inputs "openfst")))
1102                  (substitute* "configure"
1103                    (("check_for_slow_expf;") "")
1104                    ;; This affects the RPATH and also serves as the installation
1105                    ;; directory.
1106                    (("KALDILIBDIR=`pwd`/lib")
1107                     (string-append "KALDILIBDIR=" out "/lib")))
1108                  (mkdir-p out) ; must exist
1109                  (setenv "CONFIG_SHELL" (which "bash"))
1110                  (setenv "OPENFST_VER" ,(package-version openfst))
1111                  (invoke "./configure"
1112                          "--use-cuda=no"
1113                          "--shared"
1114                          (string-append "--openblas-root=" openblas)
1115                          (string-append "--fst-root=" openfst)))))
1116            (add-after 'build 'build-ext-and-gstreamer-plugin
1117              (lambda _
1118                (invoke "make" "-C" "online" "depend")
1119                (invoke "make" "-C" "online")
1120                (invoke "make" "-C" "onlinebin" "depend")
1121                (invoke "make" "-C" "onlinebin")
1122                (invoke "make" "-C" "gst-plugin" "depend")
1123                (invoke "make" "-C" "gst-plugin")
1124                #t))
1125            ;; TODO: also install the executables.
1126            (replace 'install
1127              (lambda* (#:key outputs #:allow-other-keys)
1128                (let* ((out (assoc-ref outputs "out"))
1129                       (inc (string-append out "/include"))
1130                       (lib (string-append out "/lib")))
1131                  (mkdir-p lib)
1132                  ;; The build phase installed symlinks to the actual
1133                  ;; libraries.  Install the actual targets.
1134                  (for-each (lambda (file)
1135                              (let ((target (readlink file)))
1136                                (delete-file file)
1137                                (install-file target lib)))
1138                            (find-files lib "\\.so"))
1139                  ;; Install headers
1140                  (for-each (lambda (file)
1141                              (let ((target-dir (string-append inc "/" (dirname file))))
1142                                (install-file file target-dir)))
1143                            (find-files "." "\\.h"))
1144                  (install-file "gst-plugin/libgstonlinegmmdecodefaster.so"
1145                                (string-append lib "/gstreamer-1.0"))
1146                  #t))))))
1147       (inputs
1148        `(("alsa-lib" ,alsa-lib)
1149          ("gfortran" ,gfortran "lib")
1150          ("glib" ,glib)
1151          ("gstreamer" ,gstreamer)
1152          ("jack" ,jack-1)
1153          ("openblas" ,openblas)
1154          ("openfst" ,openfst)
1155          ("portaudio" ,portaudio)
1156          ("python" ,python)))
1157       (native-inputs
1158        `(("glib" ,glib "bin")             ; glib-genmarshal
1159          ("grep" ,grep)
1160          ("sed" ,sed)
1161          ("pkg-config" ,pkg-config)
1162          ("which" ,which)))
1163       (home-page "https://kaldi-asr.org/")
1164       (synopsis "Speech recognition toolkit")
1165       (description "Kaldi is an extensible toolkit for speech recognition
1166 written in C++.")
1167       (license license:asl2.0))))
1168
1169 (define-public gst-kaldi-nnet2-online
1170   (let ((commit "cb227ef43b66a9835c14eb0ad39e08ee03c210ad")
1171         (revision "2"))
1172     (package
1173       (name "gst-kaldi-nnet2-online")
1174       (version (git-version "0" revision commit))
1175       (source (origin
1176                 (method git-fetch)
1177                 (uri (git-reference
1178                       (url "https://github.com/alumae/gst-kaldi-nnet2-online")
1179                       (commit commit)))
1180                 (file-name (git-file-name name version))
1181                 (sha256
1182                  (base32
1183                   "1i6ffwiavxx07ri0lxix6s8q0r31x7i4xxvhys5jxkixf5q34w8g"))))
1184       (build-system gnu-build-system)
1185       (arguments
1186        `(#:tests? #f                    ; there are none
1187          #:make-flags
1188          (list (string-append "SHELL="
1189                               (assoc-ref %build-inputs "bash") "/bin/bash")
1190                (string-append "KALDI_ROOT="
1191                               (assoc-ref %build-inputs "kaldi-src"))
1192                (string-append "KALDILIBDIR="
1193                               (assoc-ref %build-inputs "kaldi") "/lib")
1194                "KALDI_FLAVOR=dynamic")
1195          #:phases
1196          (modify-phases %standard-phases
1197            (add-after 'unpack 'chdir
1198              (lambda _ (chdir "src") #t))
1199            (replace 'configure
1200              (lambda* (#:key inputs #:allow-other-keys)
1201                (let ((glib (assoc-ref inputs "glib")))
1202                  (setenv "CXXFLAGS" "-fPIC")
1203                  (setenv "CPLUS_INCLUDE_PATH"
1204                          (string-append glib "/include/glib-2.0:"
1205                                         glib "/lib/glib-2.0/include:"
1206                                         (assoc-ref inputs "gstreamer")
1207                                         "/include/gstreamer-1.0")))
1208                (substitute* "Makefile"
1209                  (("include \\$\\(KALDI_ROOT\\)/src/kaldi.mk") "")
1210                  (("\\$\\(error Cannot find") "#"))
1211                #t))
1212            (add-before 'build 'build-depend
1213              (lambda* (#:key make-flags #:allow-other-keys)
1214                (apply invoke "make" "depend" make-flags)))
1215            (replace 'install
1216              (lambda* (#:key outputs #:allow-other-keys)
1217                (let* ((out (assoc-ref outputs "out"))
1218                       (lib (string-append out "/lib/gstreamer-1.0")))
1219                  (install-file "libgstkaldinnet2onlinedecoder.so" lib)
1220                  #t))))))
1221       (inputs
1222        `(("glib" ,glib)
1223          ("gstreamer" ,gstreamer)
1224          ("jansson" ,jansson)
1225          ("openfst" ,openfst)
1226          ("kaldi" ,kaldi)))
1227       (native-inputs
1228        `(("bash" ,bash)
1229          ("glib:bin" ,glib "bin")       ; glib-genmarshal
1230          ("kaldi-src" ,(package-source kaldi))
1231          ("pkg-config" ,pkg-config)))
1232       (home-page "https://kaldi-asr.org/")
1233       (synopsis "Gstreamer plugin for decoding speech")
1234       (description "This package provides a GStreamer plugin that wraps
1235 Kaldi's @code{SingleUtteranceNnet2Decoder}.  It requires iVector-adapted DNN
1236 acoustic models.  The iVectors are adapted to the current audio stream
1237 automatically.")
1238       (license license:asl2.0))))
1239
1240 (define-public kaldi-gstreamer-server
1241   ;; This is the tip of the py3 branch
1242   (let ((commit "f68cab490be7eb0da2af1475fbc16655f50a60cb")
1243         (revision "2"))
1244     (package
1245       (name "kaldi-gstreamer-server")
1246       (version (git-version "0" revision commit))
1247       (source (origin
1248                 (method git-fetch)
1249                 (uri (git-reference
1250                       (url "https://github.com/alumae/kaldi-gstreamer-server")
1251                       (commit commit)))
1252                 (file-name (git-file-name name version))
1253                 (sha256
1254                  (base32
1255                   "17lh1368vkg8ngrcbn2phvigzlmalrqg6djx2gg61qq1a0nj87dm"))))
1256       (build-system gnu-build-system)
1257       (arguments
1258        `(#:tests? #f ; there are no tests that can be run automatically
1259          #:modules ((guix build utils)
1260                     (guix build gnu-build-system)
1261                     (srfi srfi-26))
1262          #:phases
1263          (modify-phases %standard-phases
1264            (delete 'configure)
1265            (replace 'build
1266              (lambda* (#:key outputs #:allow-other-keys)
1267                ;; Disable hash randomization to ensure the generated .pycs
1268                ;; are reproducible.
1269                (setenv "PYTHONHASHSEED" "0")
1270                (with-directory-excursion "kaldigstserver"
1271                  ;; See https://github.com/alumae/kaldi-gstreamer-server/issues/232
1272                  (substitute* "master_server.py"
1273                    (("\\.replace\\('\\\\.*") ")"))
1274
1275                  ;; This is a Python 2 file
1276                  (delete-file "decoder_test.py")
1277                  (delete-file "test-buffer.py")
1278
1279                  (for-each (lambda (file)
1280                              (apply invoke
1281                                     `("python"
1282                                       "-m" "compileall"
1283                                       "-f" ; force rebuild
1284                                       ,file)))
1285                            (find-files "." "\\.py$")))
1286                #t))
1287            (replace 'install
1288              (lambda* (#:key inputs outputs #:allow-other-keys)
1289                (let* ((out (assoc-ref outputs "out"))
1290                       (bin (string-append out "/bin"))
1291                       (share (string-append out "/share/kaldi-gstreamer-server/")))
1292                  ;; Install Python files
1293                  (with-directory-excursion "kaldigstserver"
1294                    (for-each (cut install-file <> share)
1295                              (find-files "." ".*")))
1296
1297                  ;; Install sample configuration files
1298                  (for-each (cut install-file <> share)
1299                            (find-files "." "\\.yaml"))
1300
1301                  ;; Install executables
1302                  (mkdir-p bin)
1303                  (let* ((server (string-append bin "/kaldi-gst-server"))
1304                         (client (string-append bin "/kaldi-gst-client"))
1305                         (worker (string-append bin "/kaldi-gst-worker"))
1306                         (PYTHONPATH (getenv "PYTHONPATH"))
1307                         (GST_PLUGIN_PATH (string-append
1308                                           (assoc-ref inputs "gst-kaldi-nnet2-online")
1309                                           "/lib/gstreamer-1.0:${GST_PLUGIN_PATH}"))
1310                         (wrap (lambda (wrapper what)
1311                                 (with-output-to-file wrapper
1312                                   (lambda _
1313                                     (format #t
1314                                             "#!~a
1315 export PYTHONPATH=~a
1316 export GST_PLUGIN_PATH=~a
1317 exec ~a ~a/~a \"$@\"~%"
1318                                             (which "bash") PYTHONPATH GST_PLUGIN_PATH
1319                                             (which "python") share what)))
1320                                 (chmod wrapper #o555))))
1321                    (for-each wrap
1322                              (list server client worker)
1323                              (list "master_server.py"
1324                                    "client.py"
1325                                    "worker.py")))
1326                  #t))))))
1327       (inputs
1328        `(("gst-kaldi-nnet2-online" ,gst-kaldi-nnet2-online)
1329          ("python" ,python-wrapper)
1330          ("python-pygobject" ,python-pygobject)
1331          ("python-pyyaml" ,python-pyyaml)
1332          ("python-tornado" ,python-tornado-6)))
1333       (home-page "https://github.com/alumae/kaldi-gstreamer-server")
1334       (synopsis "Real-time full-duplex speech recognition server")
1335       (description "This is a real-time full-duplex speech recognition server,
1336 based on the Kaldi toolkit and the GStreamer framework and implemented in
1337 Python.")
1338       (license license:bsd-2))))
1339
1340 ;; Note that Tensorflow includes a "third_party" directory, which seems to not
1341 ;; only contain modified subsets of upstream library source code, but also
1342 ;; adapter headers provided by Google (such as the fft.h header, which is not
1343 ;; part of the upstream project code).  The Tensorflow code includes headers
1344 ;; from the "third_party" directory.  It does not look like we can replace
1345 ;; these headers with unmodified upstream files, so we keep them.
1346 (define-public tensorflow
1347   (package
1348     (name "tensorflow")
1349     (version "1.9.0")
1350     (source
1351      (origin
1352        (method git-fetch)
1353        (uri (git-reference
1354              (url "https://github.com/tensorflow/tensorflow")
1355              (commit (string-append "v" version))))
1356        (file-name (string-append "tensorflow-" version "-checkout"))
1357        (sha256
1358         (base32
1359          "0a9kwha395g3wgxfwln5j8vn9nkspmd75xldrlqdq540w996g8xa"))))
1360     (build-system cmake-build-system)
1361     (arguments
1362      `(#:tests? #f                      ; no "check" target
1363        #:build-type "Release"
1364        #:configure-flags
1365        (let ((protobuf (assoc-ref %build-inputs "protobuf"))
1366              (protobuf:native (assoc-ref %build-inputs "protobuf:native"))
1367              (jsoncpp (assoc-ref %build-inputs "jsoncpp"))
1368              (snappy (assoc-ref %build-inputs "snappy"))
1369              (sqlite (assoc-ref %build-inputs "sqlite")))
1370          (list
1371           ;; Use protobuf from Guix
1372           (string-append "-Dprotobuf_STATIC_LIBRARIES="
1373                          protobuf "/lib/libprotobuf.so")
1374           (string-append "-DPROTOBUF_PROTOC_EXECUTABLE="
1375                          protobuf:native "/bin/protoc")
1376
1377           ;; Use snappy from Guix
1378           (string-append "-Dsnappy_STATIC_LIBRARIES="
1379                          snappy "/lib/libsnappy.so")
1380           ;; Yes, this is not actually the include directory but a prefix...
1381           (string-append "-Dsnappy_INCLUDE_DIR=" snappy)
1382
1383           ;; Use jsoncpp from Guix
1384           (string-append "-Djsoncpp_STATIC_LIBRARIES="
1385                          jsoncpp "/lib/libjsoncpp.so")
1386           ;; Yes, this is not actually the include directory but a prefix...
1387           (string-append "-Djsoncpp_INCLUDE_DIR=" jsoncpp)
1388
1389           ;; Use sqlite from Guix
1390           (string-append "-Dsqlite_STATIC_LIBRARIES="
1391                          sqlite "/lib/libsqlite.a")
1392
1393           ;; Use system libraries wherever possible.  Currently, this
1394           ;; only affects zlib.
1395           "-Dsystemlib_ALL=ON"
1396           "-Dtensorflow_ENABLE_POSITION_INDEPENDENT_CODE=ON"
1397           "-Dtensorflow_BUILD_SHARED_LIB=ON"
1398           "-Dtensorflow_OPTIMIZE_FOR_NATIVE_ARCH=OFF"
1399           "-Dtensorflow_ENABLE_SSL_SUPPORT=OFF"
1400           "-Dtensorflow_BUILD_CONTRIB_KERNELS=OFF"))
1401        #:make-flags
1402        (list "CC=gcc")
1403        #:modules ((ice-9 ftw)
1404                   (guix build utils)
1405                   (guix build cmake-build-system)
1406                   ((guix build python-build-system)
1407                    #:select (python-version)))
1408        #:imported-modules (,@%cmake-build-system-modules
1409                             (guix build python-build-system))
1410        #:phases
1411        (modify-phases %standard-phases
1412          (add-after 'unpack 'set-source-file-times-to-1980
1413            ;; At the end of the tf_python_build_pip_package target, a ZIP
1414            ;; archive should be generated via bdist_wheel, but it fails with
1415            ;; "ZIP does not support timestamps before 1980".  Luckily,
1416            ;; SOURCE_DATE_EPOCH is respected, which we set to some time in
1417            ;; 1980.
1418            (lambda _ (setenv "SOURCE_DATE_EPOCH" "315532800") #t))
1419          ;; See https://github.com/tensorflow/tensorflow/issues/20517#issuecomment-406373913
1420          (add-after 'unpack 'python3.7-compatibility
1421            (lambda _
1422              (substitute* '("tensorflow/python/eager/pywrap_tfe_src.cc"
1423                             "tensorflow/python/lib/core/ndarray_tensor.cc"
1424                             "tensorflow/python/lib/core/py_func.cc")
1425                (("PyUnicode_AsUTF8") "(char *)PyUnicode_AsUTF8"))
1426              (substitute* "tensorflow/c/eager/c_api.h"
1427                (("unsigned char async")
1428                 "unsigned char is_async"))
1429
1430              ;; Remove dependency on tensorboard, a complicated but probably
1431              ;; optional package.
1432              (substitute* "tensorflow/tools/pip_package/setup.py"
1433                ((".*'tensorboard >.*") ""))
1434
1435              ;; Fix the build with python-3.8, taken from rejected upstream patch:
1436              ;; https://github.com/tensorflow/tensorflow/issues/34197
1437              (substitute* (find-files "tensorflow/python" ".*\\.cc$")
1438                (("(nullptr,)(\\ +/. tp_print)" _ _ tp_print)
1439                 (string-append "NULL,   " tp_print)))
1440              #t))
1441          (add-after 'python3.7-compatibility 'chdir
1442            (lambda _ (chdir "tensorflow/contrib/cmake") #t))
1443          (add-after 'chdir 'disable-downloads
1444            (lambda* (#:key inputs #:allow-other-keys)
1445              (substitute* (find-files "external" "\\.cmake$")
1446                (("GIT_REPOSITORY.*") "")
1447                (("GIT_TAG.*") "")
1448                (("PREFIX ")
1449                 "DOWNLOAD_COMMAND \"\"\nPREFIX "))
1450
1451              ;; Use packages from Guix
1452              (let ((grpc (assoc-ref inputs "grpc")))
1453                (substitute* "CMakeLists.txt"
1454                  ;; Sqlite
1455                  (("include\\(sqlite\\)") "")
1456                  (("\\$\\{sqlite_STATIC_LIBRARIES\\}")
1457                   (string-append (assoc-ref inputs "sqlite")
1458                                  "/lib/libsqlite3.so"))
1459                  (("sqlite_copy_headers_to_destination") "")
1460
1461                  ;; PNG
1462                  (("include\\(png\\)") "")
1463                  (("\\$\\{png_STATIC_LIBRARIES\\}")
1464                   (string-append (assoc-ref inputs "libpng")
1465                                  "/lib/libpng16.so"))
1466                  (("png_copy_headers_to_destination") "")
1467
1468                  ;; JPEG
1469                  (("include\\(jpeg\\)") "")
1470                  (("\\$\\{jpeg_STATIC_LIBRARIES\\}")
1471                   (string-append (assoc-ref inputs "libjpeg")
1472                                  "/lib/libjpeg.so"))
1473                  (("jpeg_copy_headers_to_destination") "")
1474
1475                  ;; GIF
1476                  (("include\\(gif\\)") "")
1477                  (("\\$\\{gif_STATIC_LIBRARIES\\}")
1478                   (string-append (assoc-ref inputs "giflib")
1479                                  "/lib/libgif.so"))
1480                  (("gif_copy_headers_to_destination") "")
1481
1482                  ;; lmdb
1483                  (("include\\(lmdb\\)") "")
1484                  (("\\$\\{lmdb_STATIC_LIBRARIES\\}")
1485                   (string-append (assoc-ref inputs "lmdb")
1486                                  "/lib/liblmdb.so"))
1487                  (("lmdb_copy_headers_to_destination") "")
1488
1489                  ;; Protobuf
1490                  (("include\\(protobuf\\)") "")
1491                  (("protobuf_copy_headers_to_destination") "")
1492                  (("^ +protobuf") "")
1493
1494                  ;; gRPC
1495                  (("include\\(grpc\\)")
1496                   "find_package(grpc REQUIRED NAMES gRPC)")
1497                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES grpc\\)") "")
1498
1499                  ;; Eigen
1500                  (("include\\(eigen\\)")
1501                   (string-append "find_package(eigen REQUIRED NAMES Eigen3)
1502 set(eigen_INCLUDE_DIRS ${CMAKE_CURRENT_BINARY_DIR}/external/eigen_archive "
1503                                  (assoc-ref inputs "eigen") "/include/eigen3)"))
1504                  (("^ +eigen") "")
1505
1506                  ;; snappy
1507                  (("include\\(snappy\\)")
1508                   "add_definitions(-DTF_USE_SNAPPY)")
1509                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES snappy\\)") "")
1510
1511                  ;; jsoncpp
1512                  (("include\\(jsoncpp\\)") "")
1513                  (("^ +jsoncpp") ""))
1514
1515                (substitute* "tf_core_framework.cmake"
1516                  ((" grpc") "")
1517                  (("\\$\\{GRPC_BUILD\\}/grpc_cpp_plugin")
1518                   (which "grpc_cpp_plugin"))
1519                  ;; Link with gRPC libraries
1520                  (("add_library\\(tf_protos_cc.*" m)
1521                   (string-append m
1522                                  (format #f "\ntarget_link_libraries(tf_protos_cc PRIVATE \
1523 ~a/lib/libgrpc++_unsecure.a \
1524 ~a/lib/libgrpc_unsecure.a \
1525 ~a/lib/libaddress_sorting.a \
1526 ~a/lib/libgpr.a \
1527 ~a//lib/libcares.so
1528 )\n"
1529                                          grpc grpc grpc grpc
1530                                          (assoc-ref inputs "c-ares"))))))
1531              (substitute* "tf_tools.cmake"
1532                (("add_dependencies\\(\\$\\{proto_text.*") ""))
1533              ;; Remove dependency on bundled grpc
1534              (substitute* "tf_core_distributed_runtime.cmake"
1535                (("tf_core_cpu grpc") "tf_core_cpu"))
1536
1537              ;; This directory is a dependency of many targets.
1538              (mkdir-p "protobuf")
1539              #t))
1540          (add-after 'configure 'unpack-third-party-sources
1541            (lambda* (#:key inputs #:allow-other-keys)
1542              ;; This is needed to configure bundled packages properly.
1543              (setenv "CONFIG_SHELL" (which "bash"))
1544              (for-each
1545               (lambda (name)
1546                 (let* ((what  (assoc-ref inputs (string-append name "-src")))
1547                        (name* (string-map (lambda (c)
1548                                             (if (char=? c #\-)
1549                                                 #\_ c)) name))
1550                        (where (string-append "../build/" name* "/src/" name*)))
1551                   (cond
1552                    ((string-suffix? ".zip" what)
1553                     (mkdir-p where)
1554                     (with-directory-excursion where
1555                       (invoke "unzip" what)))
1556                    ((string-suffix? ".tar.gz" what)
1557                     (mkdir-p where)
1558                     (invoke "tar" "xf" what
1559                             "-C" where "--strip-components=1"))
1560                    (else
1561                     (let ((parent (dirname where)))
1562                       (mkdir-p parent)
1563                       (with-directory-excursion parent
1564                         (when (file-exists? name*)
1565                           (delete-file-recursively name*))
1566                         (copy-recursively what name*)
1567                         (map make-file-writable
1568                              (find-files name* ".*"))))))))
1569               (list "boringssl"
1570                     "cub"
1571                     "double-conversion"
1572                     "farmhash"
1573                     "fft2d"
1574                     "highwayhash"
1575                     "nsync"
1576                     "re2"))
1577
1578              (rename-file "../build/cub/src/cub/cub-1.8.0/"
1579                           "../build/cub/src/cub/cub/")
1580              #t))
1581          (add-after 'unpack 'fix-python-build
1582            (lambda* (#:key inputs outputs #:allow-other-keys)
1583              (mkdir-p "protobuf-src")
1584              (invoke "tar" "xf" (assoc-ref inputs "protobuf:src")
1585                      "-C" "protobuf-src" "--strip-components=1")
1586              (mkdir-p "eigen-src")
1587              (invoke "tar" "xf" (assoc-ref inputs "eigen:src")
1588                      "-C" "eigen-src" "--strip-components=1")
1589
1590              (substitute* "tensorflow/contrib/cmake/tf_python.cmake"
1591                ;; Ensure that all Python dependencies can be found at build time.
1592                (("PYTHONPATH=\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/tf_python" m)
1593                 (string-append m ":" (getenv "PYTHONPATH")))
1594                ;; Take protobuf source files from our source package.
1595                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/protobuf/src/protobuf/src/google")
1596                 (string-append (getcwd) "/protobuf-src/src/google")))
1597
1598              (substitute* '("tensorflow/contrib/cmake/tf_shared_lib.cmake"
1599                             "tensorflow/contrib/cmake/tf_python.cmake")
1600                ;; Take Eigen source files from our source package.
1601                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/eigen/src/eigen/")
1602                 (string-append (getcwd) "/eigen-src/"))
1603                ;; Take Eigen headers from our own package.
1604                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/external/eigen_archive")
1605                 (string-append (assoc-ref inputs "eigen") "/include/eigen3")))
1606
1607              ;; Correct the RUNPATH of ops libraries generated for Python.
1608              ;; TODO: this doesn't work :(
1609              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1610              ;; warning: RUNPATH contains bogus entries: ("/tmp/guix-build-tensorflow-1.9.0.drv-0/source/tensorflow/contrib/build")
1611              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1612              ;; error: depends on 'libpywrap_tensorflow_internal.so', which
1613              ;; cannot be found in RUNPATH ...
1614              (substitute* "tensorflow/contrib/cmake/tf_cc_ops.cmake"
1615                (("set_target_properties.*")
1616                 (string-append "set_target_properties(${_AT_TARGET} PROPERTIES \
1617 COMPILE_FLAGS ${target_compile_flags} \
1618 INSTALL_RPATH_USE_LINK_PATH TRUE \
1619 INSTALL_RPATH " (assoc-ref outputs "out") "/lib)\n")))
1620              #t))
1621          (add-after 'build 'build-pip-package
1622            (lambda* (#:key outputs #:allow-other-keys)
1623              (setenv "LDFLAGS"
1624                      (string-append "-Wl,-rpath="
1625                                     (assoc-ref outputs "out") "/lib"))
1626              (invoke "make" "tf_python_build_pip_package")
1627              #t))
1628          (add-after 'build-pip-package 'install-python
1629            (lambda* (#:key inputs outputs #:allow-other-keys)
1630              (let ((out (assoc-ref outputs "out"))
1631                    (wheel (car (find-files "../build/tf_python/dist/" "\\.whl$")))
1632                    (python-version (python-version
1633                                      (assoc-ref inputs "python"))))
1634                (invoke "python" "-m" "pip" "install" wheel
1635                        (string-append "--prefix=" out))
1636
1637                ;; XXX: broken RUNPATH, see fix-python-build phase.
1638                (delete-file
1639                 (string-append
1640                  out "/lib/python" python-version
1641                  "/site-packages/tensorflow/contrib/"
1642                  "seq2seq/python/ops/lib_beam_search_ops.so"))
1643                #t))))))
1644     (native-inputs
1645      `(("pkg-config" ,pkg-config)
1646        ("protobuf:native" ,protobuf-3.6) ; protoc
1647        ("protobuf:src" ,(package-source protobuf-3.6))
1648        ("eigen:src" ,(package-source eigen-for-tensorflow))
1649        ;; install_pip_packages.sh wants setuptools 39.1.0 specifically.
1650        ("python-setuptools" ,python-setuptools-for-tensorflow)
1651
1652        ;; The commit hashes and URLs for third-party source code are taken
1653        ;; from "tensorflow/workspace.bzl".
1654        ("boringssl-src"
1655         ,(let ((commit "ee7aa02")
1656                (revision "1"))
1657            (origin
1658              (method git-fetch)
1659              (uri (git-reference
1660                    (url "https://boringssl.googlesource.com/boringssl")
1661                    (commit commit)))
1662              (file-name (string-append "boringssl-0-" revision
1663                                        (string-take commit 7)
1664                                        "-checkout"))
1665              (sha256
1666               (base32
1667                "1jf693q0nw0adsic6cgmbdx6g7wr4rj4vxa8j1hpn792fqhd8wgw")))))
1668        ("cub-src"
1669         ,(let ((version "1.8.0"))
1670            (origin
1671              (method url-fetch)
1672              (uri (string-append "https://mirror.bazel.build/github.com/NVlabs/"
1673                                  "cub/archive/" version ".zip"))
1674              (file-name (string-append "cub-" version ".zip"))
1675              (sha256
1676               (base32
1677                "1hsqikqridb90dkxkjr2918dcry6pfh46ccnwrzawl56aamhdykb")))))
1678        ("double-conversion-src"
1679         ,(let ((commit "5664746")
1680                (revision "1"))
1681            (origin
1682              (method git-fetch)
1683              (uri (git-reference
1684                    (url "https://github.com/google/double-conversion")
1685                    (commit commit)))
1686              (file-name
1687               (git-file-name "double-conversion"
1688                              (string-append "0-" revision "."
1689                                             (string-take commit 7))))
1690              (sha256
1691               (base32
1692                "1h5lppqqxcvdg5jq42i5msgwx20ryij3apvmndflngrgdpc04gn1")))))
1693        ("farmhash-src"
1694         ,(let ((commit "816a4ae622e964763ca0862d9dbd19324a1eaf45"))
1695            (origin
1696              (method url-fetch)
1697              (uri (string-append
1698                    "https://mirror.bazel.build/github.com/google/farmhash/archive/"
1699                    commit ".tar.gz"))
1700              (file-name (string-append "farmhash-0-" (string-take commit 7)
1701                                        ".tar.gz"))
1702              (sha256
1703               (base32
1704                "185b2xdxl4d4cnsnv6abg8s22gxvx8673jq2yaq85bz4cdy58q35")))))
1705        ;; The license notice on the home page at
1706        ;; http://www.kurims.kyoto-u.ac.jp/~ooura/fft.html says:
1707        ;;   Copyright Takuya OOURA, 1996-2001
1708        ;;
1709        ;;   You may use, copy, modify and distribute this code for any purpose
1710        ;;   (include commercial use) and without fee. Please refer to this
1711        ;;   package when you modify this code.
1712        ;;
1713        ;; We take the identical tarball from the Bazel mirror, because the URL
1714        ;; at the home page is not versioned and might change.
1715        ("fft2d-src"
1716         ,(origin
1717            (method url-fetch)
1718            (uri "https://mirror.bazel.build/www.kurims.kyoto-u.ac.jp/~ooura/fft.tgz")
1719            (file-name "fft2d.tar.gz")
1720            (sha256
1721             (base32
1722              "15jjkfvhqvl2c0753d2di8hz0pyzn598g74wqy79awdrf1y67fsj"))))
1723        ("highwayhash-src"
1724         ,(let ((commit "be5edafc2e1a455768e260ccd68ae7317b6690ee")
1725                (revision "1"))
1726            (origin
1727              (method git-fetch)
1728              (uri (git-reference
1729                    (url "https://github.com/google/highwayhash")
1730                    (commit commit)))
1731              (file-name (string-append "highwayhash-0-" revision
1732                                        (string-take commit 7)
1733                                        "-checkout"))
1734              (sha256
1735               (base32
1736                "154jwf98cyy54hldr94pgjn85zynly3abpnc1avmb8a18lzwjyb6")))))
1737        ("nsync-src"
1738         ,(let ((version "0559ce013feac8db639ee1bf776aca0325d28777")
1739                (revision "1"))
1740            (origin
1741              (method url-fetch)
1742              (uri (string-append "https://mirror.bazel.build/"
1743                                  "github.com/google/nsync/archive/"
1744                                  version ".tar.gz"))
1745              (file-name (string-append "nsync-0." revision
1746                                        "-" (string-take version 7)
1747                                        ".tar.gz"))
1748              (sha256
1749               (base32
1750                "0qdkyqym34x739mmzv97ah5r7ph462v5xkxqxvidmcfqbi64b132")))))
1751        ("re2-src"
1752         ,(let ((commit "e7efc48")
1753                (revision "1"))
1754            (origin
1755              (method git-fetch)
1756              (uri (git-reference
1757                    (url "https://github.com/google/re2")
1758                    (commit commit)))
1759              (file-name (string-append "re2-0-" revision
1760                                        (string-take commit 7)
1761                                        "-checkout"))
1762              (sha256
1763               (base32
1764                "161g9841rjfsy5pn52fcis0s9hdr7rxvb06pad38j5rppfihvign")))))
1765        ("googletest" ,googletest)
1766        ("swig" ,swig)
1767        ("unzip" ,unzip)))
1768     (propagated-inputs
1769      `(("python-absl-py" ,python-absl-py)
1770        ("python-astor" ,python-astor)
1771        ("python-gast" ,python-gast)
1772        ("python-grpcio" ,python-grpcio)
1773        ("python-numpy" ,python-numpy)
1774        ("python-protobuf" ,python-protobuf-3.6)
1775        ("python-six" ,python-six)
1776        ("python-termcolo" ,python-termcolor)
1777        ("python-wheel" ,python-wheel)))
1778     (inputs
1779      `(("c-ares" ,c-ares)
1780        ("eigen" ,eigen-for-tensorflow)
1781        ("gemmlowp" ,gemmlowp-for-tensorflow)
1782        ("lmdb" ,lmdb)
1783        ("libjpeg" ,libjpeg-turbo)
1784        ("libpng" ,libpng)
1785        ("giflib" ,giflib)
1786        ("grpc" ,grpc-1.16.1 "static")
1787        ("grpc:bin" ,grpc-1.16.1)
1788        ("jsoncpp" ,jsoncpp-for-tensorflow)
1789        ("snappy" ,snappy)
1790        ("sqlite" ,sqlite)
1791        ("protobuf" ,protobuf-3.6)
1792        ("python" ,python-wrapper)
1793        ("zlib" ,zlib)))
1794     (home-page "https://tensorflow.org")
1795     (synopsis "Machine learning framework")
1796     (description
1797      "TensorFlow is a flexible platform for building and training machine
1798 learning models.  It provides a library for high performance numerical
1799 computation and includes high level Python APIs, including both a sequential
1800 API for beginners that allows users to build models quickly by plugging
1801 together building blocks and a subclassing API with an imperative style for
1802 advanced research.")
1803     (license license:asl2.0)))
1804
1805 (define-public python-iml
1806   (package
1807     (name "python-iml")
1808     (version "0.6.2")
1809     (source
1810      (origin
1811        (method url-fetch)
1812        (uri (pypi-uri "iml" version))
1813        (sha256
1814         (base32
1815          "1k8szlpm19rcwcxdny9qdm3gmaqq8akb4xlvrzyz8c2d679aak6l"))))
1816     (build-system python-build-system)
1817     (propagated-inputs
1818      `(("ipython" ,python-ipython)
1819        ("numpy" ,python-numpy)
1820        ("pandas" ,python-pandas)
1821        ("scipy" ,python-scipy)))
1822     (native-inputs
1823      `(("nose" ,python-nose)))
1824     (home-page "https://github.com/interpretable-ml/iml")
1825     (synopsis "Interpretable Machine Learning (iML) package")
1826     (description "Interpretable ML (iML) is a set of data type objects,
1827 visualizations, and interfaces that can be used by any method designed to
1828 explain the predictions of machine learning models (or really the output of
1829 any function).  It currently contains the interface and IO code from the Shap
1830 project, and it will potentially also do the same for the Lime project.")
1831     (license license:expat)))
1832
1833 (define-public python-keras-applications
1834   (package
1835     (name "python-keras-applications")
1836     (version "1.0.8")
1837     (source
1838      (origin
1839        (method url-fetch)
1840        (uri (pypi-uri "Keras_Applications" version))
1841        (sha256
1842         (base32
1843          "1rcz31ca4axa6kzhjx4lwqxbg4wvlljkj8qj9a7p9sfd5fhzjyam"))))
1844     (build-system python-build-system)
1845     ;; The tests require Keras, but this package is needed to build Keras.
1846     (arguments '(#:tests? #f))
1847     (propagated-inputs
1848      `(("python-h5py" ,python-h5py)
1849        ("python-numpy" ,python-numpy)))
1850     (native-inputs
1851      `(("python-pytest" ,python-pytest)
1852        ("python-pytest-cov" ,python-pytest-cov)
1853        ("python-pytest-pep8" ,python-pytest-pep8)
1854        ("python-pytest-xdist" ,python-pytest-xdist)))
1855     (home-page "https://github.com/keras-team/keras-applications")
1856     (synopsis "Reference implementations of popular deep learning models")
1857     (description
1858      "This package provides reference implementations of popular deep learning
1859 models for use with the Keras deep learning framework.")
1860     (license license:expat)))
1861
1862 (define-public python-keras-preprocessing
1863   (package
1864     (name "python-keras-preprocessing")
1865     (version "1.1.0")
1866     (source
1867      (origin
1868        (method url-fetch)
1869        (uri (pypi-uri "Keras_Preprocessing" version))
1870        (sha256
1871         (base32
1872          "1r98nm4k1svsqjyaqkfk23i31bl1kcfcyp7094yyj3c43phfp3as"))))
1873     (build-system python-build-system)
1874     (propagated-inputs
1875      `(("python-numpy" ,python-numpy)
1876        ("python-six" ,python-six)))
1877     (native-inputs
1878      `(("python-pandas" ,python-pandas)
1879        ("python-pillow" ,python-pillow)
1880        ("python-pytest" ,python-pytest)
1881        ("python-pytest-cov" ,python-pytest-cov)
1882        ("python-pytest-xdist" ,python-pytest-xdist)
1883        ("tensorflow" ,tensorflow)))
1884     (home-page "https://github.com/keras-team/keras-preprocessing/")
1885     (synopsis "Data preprocessing and augmentation for deep learning models")
1886     (description
1887      "Keras Preprocessing is the data preprocessing and data augmentation
1888 module of the Keras deep learning library.  It provides utilities for working
1889 with image data, text data, and sequence data.")
1890     (license license:expat)))
1891
1892 (define-public python-keras
1893   (package
1894     (name "python-keras")
1895     (version "2.2.4")
1896     (source
1897      (origin
1898        (method url-fetch)
1899        (uri (pypi-uri "Keras" version))
1900        (patches (search-patches "python-keras-integration-test.patch"))
1901        (sha256
1902         (base32
1903          "1j8bsqzh49vjdxy6l1k4iwax5vpjzniynyd041xjavdzvfii1dlh"))))
1904     (build-system python-build-system)
1905     (arguments
1906      `(#:phases
1907        (modify-phases %standard-phases
1908          (add-after 'unpack 'remove-tests-for-unavailable-features
1909            (lambda _
1910              (delete-file "keras/backend/theano_backend.py")
1911              (delete-file "keras/backend/cntk_backend.py")
1912              (delete-file "tests/keras/backend/backend_test.py")
1913
1914              ;; FIXME: This doesn't work because Tensorflow is missing the
1915              ;; coder ops library.
1916              (delete-file "tests/keras/test_callbacks.py")
1917              #t))
1918          (replace 'check
1919            (lambda _
1920              ;; These tests attempt to download data files from the internet.
1921              (delete-file "tests/integration_tests/test_datasets.py")
1922              (delete-file "tests/integration_tests/imagenet_utils_test.py")
1923
1924              (setenv "PYTHONPATH"
1925                      (string-append (getcwd) "/build/lib:"
1926                                     (getenv "PYTHONPATH")))
1927              (invoke "py.test" "-v"
1928                      "-p" "no:cacheprovider"
1929                      "--ignore" "keras/utils"))))))
1930     (propagated-inputs
1931      `(("python-h5py" ,python-h5py)
1932        ("python-keras-applications" ,python-keras-applications)
1933        ("python-keras-preprocessing" ,python-keras-preprocessing)
1934        ("python-numpy" ,python-numpy)
1935        ("python-pydot" ,python-pydot)
1936        ("python-pyyaml" ,python-pyyaml)
1937        ("python-scipy" ,python-scipy)
1938        ("python-six" ,python-six)
1939        ("tensorflow" ,tensorflow)
1940        ("graphviz" ,graphviz)))
1941     (native-inputs
1942      `(("python-pandas" ,python-pandas)
1943        ("python-pytest" ,python-pytest)
1944        ("python-pytest-cov" ,python-pytest-cov)
1945        ("python-pytest-pep8" ,python-pytest-pep8)
1946        ("python-pytest-timeout" ,python-pytest-timeout)
1947        ("python-pytest-xdist" ,python-pytest-xdist)
1948        ("python-sphinx" ,python-sphinx)
1949        ("python-requests" ,python-requests)))
1950     (home-page "https://github.com/keras-team/keras")
1951     (synopsis "High-level deep learning framework")
1952     (description "Keras is a high-level neural networks API, written in Python
1953 and capable of running on top of TensorFlow.  It was developed with a focus on
1954 enabling fast experimentation.  Use Keras if you need a deep learning library
1955 that:
1956
1957 @itemize
1958 @item Allows for easy and fast prototyping (through user friendliness,
1959   modularity, and extensibility).
1960 @item Supports both convolutional networks and recurrent networks, as well as
1961   combinations of the two.
1962 @item Runs seamlessly on CPU and GPU.
1963 @end itemize\n")
1964     (license license:expat)))
1965
1966 (define-public gloo
1967   (let ((version "0.0.0") ; no proper version tag
1968         (commit "ca528e32fea9ca8f2b16053cff17160290fc84ce")
1969         (revision "0"))
1970     (package
1971       (name "gloo")
1972       (version (git-version version revision commit))
1973       (source
1974        (origin
1975          (method git-fetch)
1976          (uri (git-reference
1977                (url "https://github.com/facebookincubator/gloo")
1978                (commit commit)))
1979          (file-name (git-file-name name version))
1980          (sha256
1981           (base32
1982            "1q9f80zy75f6njrzrqkmhc0g3qxs4gskr7ns2jdqanxa2ww7a99w"))))
1983       (build-system cmake-build-system)
1984       (native-inputs
1985        `(("googletest" ,googletest)))
1986       (arguments
1987        `(#:configure-flags '("-DBUILD_TEST=1")
1988          #:phases
1989          (modify-phases %standard-phases
1990            (replace 'check
1991              (lambda _
1992                (invoke "make" "gloo_test")
1993                #t)))))
1994       (synopsis "Collective communications library")
1995       (description
1996        "Gloo is a collective communications library.  It comes with a
1997 number of collective algorithms useful for machine learning applications.
1998 These include a barrier, broadcast, and allreduce.")
1999       (home-page "https://github.com/facebookincubator/gloo")
2000       (license license:bsd-3))))
2001
2002 (define-public python-umap-learn
2003   (package
2004     (name "python-umap-learn")
2005     (version "0.3.10")
2006     (source
2007      (origin
2008        (method url-fetch)
2009        (uri (pypi-uri "umap-learn" version))
2010        (sha256
2011         (base32
2012          "02ada2yy6km6zgk2836kg1c97yrcpalvan34p8c57446finnpki1"))))
2013     (build-system python-build-system)
2014     (native-inputs
2015      `(("python-joblib" ,python-joblib)
2016        ("python-nose" ,python-nose)))
2017     (propagated-inputs
2018      `(("python-numba" ,python-numba)
2019        ("python-numpy" ,python-numpy)
2020        ("python-scikit-learn" ,python-scikit-learn)
2021        ("python-scipy" ,python-scipy)))
2022     (home-page "https://github.com/lmcinnes/umap")
2023     (synopsis
2024      "Uniform Manifold Approximation and Projection")
2025     (description
2026      "Uniform Manifold Approximation and Projection is a dimension reduction
2027 technique that can be used for visualisation similarly to t-SNE, but also for
2028 general non-linear dimension reduction.")
2029     (license license:bsd-3)))