gnu/packages/machine-learning.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2015, 2016, 2017, 2018, 2019 Ricardo Wurmus <rekado@elephly.net>
   3 ;;; Copyright © 2016 Efraim Flashner <efraim@flashner.co.il>
   4 ;;; Copyright © 2016, 2017 Marius Bakke <mbakke@fastmail.com>
   5 ;;; Copyright © 2016 Hartmut Goebel <h.goebel@crazy-compilers.com>
   6 ;;; Copyright © 2018, 2019 Tobias Geerinckx-Rice <me@tobias.gr>
   7 ;;; Copyright © 2018 Kei Kebreau <kkebreau@posteo.net>
   8 ;;; Copyright © 2018 Mark Meyer <mark@ofosos.org>
   9 ;;; Copyright © 2018 Ben Woodcroft <donttrustben@gmail.com>
  10 ;;; Copyright © 2018 Fis Trivial <ybbs.daans@hotmail.com>
  11 ;;; Copyright © 2018 Julien Lepiller <julien@lepiller.eu>
  12 ;;; Copyright © 2018 Björn Höfling <bjoern.hoefling@bjoernhoefling.de>
  13 ;;; Copyright © 2019 Nicolas Goaziou <mail@nicolasgoaziou.fr>
  14 ;;; Copyright © 2019 Guillaume Le Vaillant <glv@posteo.net>
  15 ;;;
  16 ;;; This file is part of GNU Guix.
  17 ;;;
  18 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  19 ;;; under the terms of the GNU General Public License as published by
  20 ;;; the Free Software Foundation; either version 3 of the License, or (at
  21 ;;; your option) any later version.
  22 ;;;
  23 ;;; GNU Guix is distributed in the hope that it will be useful, but
  24 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  25 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  26 ;;; GNU General Public License for more details.
  27 ;;;
  28 ;;; You should have received a copy of the GNU General Public License
  29 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  30
  31 (define-module (gnu packages machine-learning)
  32   #:use-module ((guix licenses) #:prefix license:)
  33   #:use-module (guix packages)
  34   #:use-module (guix utils)
  35   #:use-module (guix download)
  36   #:use-module (guix svn-download)
  37   #:use-module (guix build-system asdf)
  38   #:use-module (guix build-system cmake)
  39   #:use-module (guix build-system gnu)
  40   #:use-module (guix build-system ocaml)
  41   #:use-module (guix build-system python)
  42   #:use-module (guix build-system r)
  43   #:use-module (guix git-download)
  44   #:use-module (gnu packages)
  45   #:use-module (gnu packages adns)
  46   #:use-module (gnu packages algebra)
  47   #:use-module (gnu packages audio)
  48   #:use-module (gnu packages autotools)
  49   #:use-module (gnu packages base)
  50   #:use-module (gnu packages bash)
  51   #:use-module (gnu packages boost)
  52   #:use-module (gnu packages check)
  53   #:use-module (gnu packages compression)
  54   #:use-module (gnu packages cran)
  55   #:use-module (gnu packages databases)
  56   #:use-module (gnu packages dejagnu)
  57   #:use-module (gnu packages gcc)
  58   #:use-module (gnu packages glib)
  59   #:use-module (gnu packages graphviz)
  60   #:use-module (gnu packages gstreamer)
  61   #:use-module (gnu packages image)
  62   #:use-module (gnu packages linux)
  63   #:use-module (gnu packages lisp)
  64   #:use-module (gnu packages maths)
  65   #:use-module (gnu packages mpi)
  66   #:use-module (gnu packages ocaml)
  67   #:use-module (gnu packages onc-rpc)
  68   #:use-module (gnu packages perl)
  69   #:use-module (gnu packages pkg-config)
  70   #:use-module (gnu packages protobuf)
  71   #:use-module (gnu packages python)
  72   #:use-module (gnu packages python-web)
  73   #:use-module (gnu packages python-xyz)
  74   #:use-module (gnu packages serialization)
  75   #:use-module (gnu packages sphinx)
  76   #:use-module (gnu packages statistics)
  77   #:use-module (gnu packages sqlite)
  78   #:use-module (gnu packages swig)
  79   #:use-module (gnu packages tls)
  80   #:use-module (gnu packages web)
  81   #:use-module (gnu packages xml)
  82   #:use-module (gnu packages xorg)
  83   #:use-module (ice-9 match))
  84
  85 (define-public fann
  86   ;; The last release is >100 commits behind, so we package from git.
  87   (let ((commit "d71d54788bee56ba4cf7522801270152da5209d7"))
  88     (package
  89       (name "fann")
  90       (version (string-append "2.2.0-1." (string-take commit 8)))
  91       (source (origin
  92                 (method git-fetch)
  93                 (uri (git-reference
  94                       (url "https://github.com/libfann/fann.git")
  95                       (commit commit)))
  96                 (file-name (string-append name "-" version "-checkout"))
  97                 (sha256
  98                  (base32
  99                   "0ibwpfrjs6q2lijs8slxjgzb2llcl6rk3v2ski4r6215g5jjhg3x"))))
 100       (build-system cmake-build-system)
 101       (arguments
 102        `(#:phases
 103          (modify-phases %standard-phases
 104            (replace 'check
 105              (lambda* (#:key outputs #:allow-other-keys)
 106                (let* ((out (assoc-ref outputs "out")))
 107                  (with-directory-excursion (string-append (getcwd) "/tests")
 108                    (invoke "./fann_tests"))))))))
 109       (home-page "http://leenissen.dk/fann/wp/")
 110       (synopsis "Fast Artificial Neural Network")
 111       (description
 112        "FANN is a free open source neural network library, which implements
 113 multilayer artificial neural networks in C with support for both fully
 114 connected and sparsely connected networks.")
 115       (license license:lgpl2.1))))
 116
 117 (define-public libsvm
 118   (package
 119     (name "libsvm")
 120     (version "3.23")
 121     (source
 122      (origin
 123        (method url-fetch)
 124        (uri (string-append "https://www.csie.ntu.edu.tw/~cjlin/libsvm/"
 125                            name "-" version ".tar.gz"))
 126        (sha256
 127         (base32 "0jpaq0rr92x38p4nk3gjan79ip67m6p80anb28z1d8601miysyi5"))))
 128     (build-system gnu-build-system)
 129     (arguments
 130      `(#:tests? #f                      ; no "check" target
 131        #:phases (modify-phases %standard-phases
 132                   (delete 'configure)
 133                   (replace
 134                    'install             ; no ‘install’ target
 135                    (lambda* (#:key outputs #:allow-other-keys)
 136                      (let* ((out (assoc-ref outputs "out"))
 137                             (bin (string-append out "/bin/")))
 138                        (mkdir-p bin)
 139                        (for-each (lambda (file)
 140                                    (copy-file file (string-append bin file)))
 141                                  '("svm-train"
 142                                    "svm-predict"
 143                                    "svm-scale")))
 144                      #t)))))
 145     (home-page "http://www.csie.ntu.edu.tw/~cjlin/libsvm/")
 146     (synopsis "Library for Support Vector Machines")
 147     (description
 148      "LIBSVM is a machine learning library for support vector
 149 classification, (C-SVC, nu-SVC), regression (epsilon-SVR, nu-SVR) and
 150 distribution estimation (one-class SVM).  It supports multi-class
 151 classification.")
 152     (license license:bsd-3)))
 153
 154 (define-public python-libsvm
 155   (package (inherit libsvm)
 156     (name "python-libsvm")
 157     (build-system gnu-build-system)
 158     (arguments
 159      `(#:tests? #f                      ; no "check" target
 160        #:make-flags '("-C" "python")
 161        #:phases
 162        (modify-phases %standard-phases
 163          (delete 'configure)
 164          (replace
 165           'install                      ; no ‘install’ target
 166           (lambda* (#:key inputs outputs #:allow-other-keys)
 167             (let ((site (string-append (assoc-ref outputs "out")
 168                                        "/lib/python"
 169                                        (string-take
 170                                         (string-take-right
 171                                          (assoc-ref inputs "python") 5) 3)
 172                                        "/site-packages/")))
 173               (substitute* "python/svm.py"
 174                 (("../libsvm.so.2") "libsvm.so.2"))
 175               (mkdir-p site)
 176               (for-each (lambda (file)
 177                           (copy-file file (string-append site (basename file))))
 178                         (find-files "python" "\\.py"))
 179               (copy-file "libsvm.so.2"
 180                          (string-append site "libsvm.so.2")))
 181             #t)))))
 182     (inputs
 183      `(("python" ,python)))
 184     (synopsis "Python bindings of libSVM")))
 185
 186 (define-public ghmm
 187   ;; The latest release candidate is several years and a couple of fixes have
 188   ;; been published since.  This is why we download the sources from the SVN
 189   ;; repository.
 190   (let ((svn-revision 2341))
 191     (package
 192       (name "ghmm")
 193       (version (string-append "0.9-rc3-0." (number->string svn-revision)))
 194       (source (origin
 195                 (method svn-fetch)
 196                 (uri (svn-reference
 197                       (url "http://svn.code.sf.net/p/ghmm/code/trunk")
 198                       (revision svn-revision)))
 199                 (file-name (string-append name "-" version "-checkout"))
 200                 (sha256
 201                  (base32
 202                   "0qbq1rqp94l530f043qzp8aw5lj7dng9wq0miffd7spd1ff638wq"))))
 203       (build-system gnu-build-system)
 204       (arguments
 205        `(#:imported-modules (,@%gnu-build-system-modules
 206                              (guix build python-build-system))
 207          #:phases
 208          (modify-phases %standard-phases
 209            (add-after 'unpack 'enter-dir
 210              (lambda _ (chdir "ghmm") #t))
 211            (delete 'check)
 212            (add-after 'install 'check
 213              (assoc-ref %standard-phases 'check))
 214            (add-before 'check 'fix-PYTHONPATH
 215              (lambda* (#:key inputs outputs #:allow-other-keys)
 216                (let ((python-version (python-version
 217                                       (assoc-ref inputs "python"))))
 218                  (setenv "PYTHONPATH"
 219                          (string-append (getenv "PYTHONPATH")
 220                                         ":" (assoc-ref outputs "out")
 221                                         "/lib/python" python-version
 222                                         "/site-packages")))
 223                #t))
 224            (add-after 'enter-dir 'fix-runpath
 225              (lambda* (#:key outputs #:allow-other-keys)
 226                (substitute* "ghmmwrapper/setup.py"
 227                  (("^(.*)extra_compile_args = \\[" line indent)
 228                   (string-append indent
 229                                  "extra_link_args = [\"-Wl,-rpath="
 230                                  (assoc-ref outputs "out") "/lib\"],\n"
 231                                  line
 232                                  "\"-Wl,-rpath="
 233                                  (assoc-ref outputs "out")
 234                                  "/lib\", ")))
 235                #t))
 236            (add-after 'enter-dir 'disable-broken-tests
 237              (lambda _
 238                (substitute* "tests/Makefile.am"
 239                  ;; GHMM_SILENT_TESTS is assumed to be a command.
 240                  (("TESTS_ENVIRONMENT.*") "")
 241                  ;; Do not build broken tests.
 242                  (("chmm .*") "")
 243                  (("read_fa .*") "")
 244                  (("mcmc .*") "")
 245                  (("label_higher_order_test.*$")
 246                   "label_higher_order_test\n"))
 247
 248                ;; These Python unittests are broken as there is no gato.
 249                ;; See https://sourceforge.net/p/ghmm/support-requests/3/
 250                (substitute* "ghmmwrapper/ghmmunittests.py"
 251                  (("^(.*)def (testNewXML|testMultipleTransitionClasses|testNewXML)"
 252                    line indent)
 253                   (string-append indent
 254                                  "@unittest.skip(\"Disabled by Guix\")\n"
 255                                  line)))
 256                #t)))))
 257       (inputs
 258        `(("python" ,python-2) ; only Python 2 is supported
 259          ("libxml2" ,libxml2)))
 260       (native-inputs
 261        `(("pkg-config" ,pkg-config)
 262          ("dejagnu" ,dejagnu)
 263          ("swig" ,swig)
 264          ("autoconf" ,autoconf)
 265          ("automake" ,automake)
 266          ("libtool" ,libtool)))
 267       (home-page "http://ghmm.org")
 268       (synopsis "Hidden Markov Model library")
 269       (description
 270        "The General Hidden Markov Model library (GHMM) is a C library with
 271 additional Python bindings implementing a wide range of types of @dfn{Hidden
 272 Markov Models} (HMM) and algorithms: discrete, continuous emissions, basic
 273 training, HMM clustering, HMM mixtures.")
 274       (license license:lgpl2.0+))))
 275
 276 (define-public mcl
 277   (package
 278     (name "mcl")
 279     (version "14.137")
 280     (source (origin
 281               (method url-fetch)
 282               (uri (string-append
 283                     "http://micans.org/mcl/src/mcl-"
 284                     (string-replace-substring version "." "-")
 285                     ".tar.gz"))
 286               (sha256
 287                (base32
 288                 "15xlax3z31lsn62vlg94hkm75nm40q4679amnfg13jm8m2bnhy5m"))))
 289     (build-system gnu-build-system)
 290     (arguments
 291      `(#:configure-flags (list "--enable-blast")))
 292     (inputs
 293      `(("perl" ,perl)))
 294     (home-page "http://micans.org/mcl/")
 295     (synopsis "Clustering algorithm for graphs")
 296     (description
 297      "The MCL algorithm is short for the @dfn{Markov Cluster Algorithm}, a
 298 fast and scalable unsupervised cluster algorithm for graphs (also known as
 299 networks) based on simulation of (stochastic) flow in graphs.")
 300     ;; In the LICENCE file and web page it says "The software is licensed
 301     ;; under the GNU General Public License, version 3.", but in several of
 302     ;; the source code files it suggests GPL3 or later.
 303     ;; http://listserver.ebi.ac.uk/pipermail/mcl-users/2016/000376.html
 304     (license license:gpl3)))
 305
 306 (define-public ocaml-mcl
 307   (package
 308     (name "ocaml-mcl")
 309     (version "12-068oasis4")
 310     (source
 311      (origin
 312        (method git-fetch)
 313        (uri (git-reference
 314              (url "https://github.com/fhcrc/mcl.git")
 315              (commit version)))
 316        (file-name (git-file-name name version))
 317        (sha256
 318         (base32
 319          "0009dc3h2jp3qg5val452wngpqnbfyhbcxylghq0mrjqxx0jdq5p"))))
 320     (build-system ocaml-build-system)
 321     (arguments
 322      `(#:phases
 323        (modify-phases %standard-phases
 324          (add-before 'configure 'patch-paths
 325            (lambda _
 326              (substitute* "configure"
 327                (("/bin/sh") (which "sh")))
 328              (substitute* "setup.ml"
 329                (("LDFLAGS=-fPIC")
 330                 (string-append "LDFLAGS=-fPIC\"; \"SHELL=" (which "sh")))
 331                (("-std=c89") "-std=gnu99")
 332
 333                ;; This is a mutable string, which is no longer supported.  Use
 334                ;; a byte buffer instead.
 335                (("String.make \\(String.length s\\)")
 336                 "Bytes.make (String.length s)")
 337
 338                ;; These two belong together.
 339                (("OASISString.replace_chars")
 340                 "Bytes.to_string (OASISString.replace_chars")
 341                ((" s;")
 342                 " s);"))
 343              (substitute* "myocamlbuild.ml"
 344                (("std=c89") "std=gnu99"))
 345              ;; Since we build with a more recent OCaml, we have to use C99 or
 346              ;; later.  This causes problems with the old C code.
 347              (substitute* "src/impala/matrix.c"
 348                (("restrict") "restrict_"))
 349              #t)))))
 350     (native-inputs
 351      `(("ocamlbuild" ,ocamlbuild)))
 352     (home-page "https://github.com/fhcrc/mcl")
 353     (synopsis "OCaml wrappers around MCL")
 354     (description
 355      "This package provides OCaml bindings for the MCL graph clustering
 356 algorithm.")
 357     (license license:gpl3)))
 358
 359 (define-public randomjungle
 360   (package
 361     (name "randomjungle")
 362     (version "2.1.0")
 363     (source
 364      (origin
 365        (method url-fetch)
 366        (uri (string-append
 367              "https://www.imbs.uni-luebeck.de/fileadmin/files/Software"
 368              "/randomjungle/randomjungle-" version ".tar_.gz"))
 369        (patches (search-patches "randomjungle-disable-static-build.patch"))
 370        (sha256
 371         (base32
 372          "12c8rf30cla71swx2mf4ww9mfd8jbdw5lnxd7dxhyw1ygrvg6y4w"))))
 373     (build-system gnu-build-system)
 374     (arguments
 375      `(#:configure-flags
 376        (list "--disable-static"
 377              (string-append "--with-boost="
 378                             (assoc-ref %build-inputs "boost")))
 379        #:phases
 380        (modify-phases %standard-phases
 381          (add-before
 382           'configure 'set-CXXFLAGS
 383           (lambda _
 384             (setenv "CXXFLAGS" "-fpermissive ")
 385             #t)))))
 386     (inputs
 387      `(("boost" ,boost)
 388        ("gsl" ,gsl)
 389        ("libxml2" ,libxml2)
 390        ("zlib" ,zlib)))
 391     (native-inputs
 392      `(("gfortran" ,gfortran)
 393        ("gfortran:lib" ,gfortran "lib")))
 394     ;; Non-portable assembly instructions are used so building fails on
 395     ;; platforms other than x86_64 or i686.
 396     (supported-systems '("x86_64-linux" "i686-linux"))
 397     (home-page "https://www.imbs.uni-luebeck.de/forschung/software/details.html#c224")
 398     (synopsis "Implementation of the Random Forests machine learning method")
 399     (description
 400      "Random Jungle is an implementation of Random Forests.  It is supposed to
 401 analyse high dimensional data.  In genetics, it can be used for analysing big
 402 Genome Wide Association (GWA) data.  Random Forests is a powerful machine
 403 learning method.  Most interesting features are variable selection, missing
 404 value imputation, classifier creation, generalization error estimation and
 405 sample proximities between pairs of cases.")
 406     (license license:gpl3+)))
 407
 408 (define-public openfst
 409   (package
 410     (name "openfst")
 411     (version "1.7.2")
 412     (source (origin
 413               (method url-fetch)
 414               (uri (string-append "http://www.openfst.org/twiki/pub/FST/"
 415                                   "FstDownload/openfst-" version ".tar.gz"))
 416               (sha256
 417                (base32
 418                 "0fqgk8195kz21is09gwzwnrg7fr9526bi9mh4apyskapz27pbhr1"))))
 419     (build-system gnu-build-system)
 420     (home-page "http://www.openfst.org")
 421     (synopsis "Library for weighted finite-state transducers")
 422     (description "OpenFst is a library for constructing, combining,
 423 optimizing, and searching weighted finite-state transducers (FSTs).")
 424     (license license:asl2.0)))
 425
 426 (define-public shogun
 427   (package
 428     (name "shogun")
 429     (version "6.1.3")
 430     (source
 431      (origin
 432        (method url-fetch)
 433        (uri (string-append
 434              "ftp://shogun-toolbox.org/shogun/releases/"
 435              (version-major+minor version)
 436              "/sources/shogun-" version ".tar.bz2"))
 437        (sha256
 438         (base32
 439          "1rn9skm3nw6hr7mr3lgp2gfqhi7ii0lyxck7qmqnf8avq349s5jp"))
 440        (modules '((guix build utils)
 441                   (ice-9 rdelim)))
 442        (snippet
 443         '(begin
 444            ;; Remove non-free sources and files referencing them
 445            (for-each delete-file
 446                      (find-files "src/shogun/classifier/svm/"
 447                                  "SVMLight\\.(cpp|h)"))
 448            (for-each delete-file
 449                      (find-files "examples/undocumented/libshogun/"
 450                                  (string-append
 451                                   "(classifier_.*svmlight.*|"
 452                                   "evaluation_cross_validation_locked_comparison).cpp")))
 453            ;; Remove non-free functions.
 454            (define (delete-ifdefs file)
 455              (with-atomic-file-replacement file
 456                (lambda (in out)
 457                  (let loop ((line (read-line in 'concat))
 458                             (skipping? #f))
 459                    (if (eof-object? line)
 460                        #t
 461                        (let ((skip-next?
 462                               (or (and skipping?
 463                                        (not (string-prefix?
 464                                              "#endif //USE_SVMLIGHT" line)))
 465                                   (string-prefix?
 466                                    "#ifdef USE_SVMLIGHT" line))))
 467                          (when (or (not skipping?)
 468                                    (and skipping? (not skip-next?)))
 469                            (display line out))
 470                          (loop (read-line in 'concat) skip-next?)))))))
 471            (for-each delete-ifdefs
 472                      (append
 473                       (find-files "src/shogun/classifier/mkl"
 474                                   "^MKLClassification\\.cpp")
 475                       (find-files "src/shogun/classifier/svm"
 476                                   "^SVMLightOneClass\\.(cpp|h)")
 477                       (find-files "src/shogun/multiclass"
 478                                   "^ScatterSVM\\.(cpp|h)")
 479                       (find-files "src/shogun/kernel/"
 480                                   "^(Kernel|CombinedKernel|ProductKernel)\\.(cpp|h)")
 481                       (find-files "src/shogun/regression/svr"
 482                                   "^(MKLRegression|SVRLight)\\.(cpp|h)")
 483                       (find-files "src/shogun/transfer/domain_adaptation"
 484                                   "^DomainAdaptationSVM\\.(cpp|h)")))
 485            #t))))
 486     (build-system cmake-build-system)
 487     (arguments
 488      '(#:tests? #f ;no check target
 489        #:phases
 490        (modify-phases %standard-phases
 491          (add-after 'unpack 'delete-broken-symlinks
 492            (lambda _
 493              (for-each delete-file '("applications/arts/data"
 494                                      "applications/asp/data"
 495                                      "applications/easysvm/data"
 496                                      "applications/msplicer/data"
 497                                      "applications/ocr/data"
 498                                      "examples/meta/data"
 499                                      "examples/undocumented/data"))
 500              #t))
 501          (add-after 'unpack 'change-R-target-path
 502            (lambda* (#:key outputs #:allow-other-keys)
 503              (substitute* '("src/interfaces/r/CMakeLists.txt"
 504                             "examples/meta/r/CMakeLists.txt")
 505                (("\\$\\{R_COMPONENT_LIB_PATH\\}")
 506                 (string-append (assoc-ref outputs "out")
 507                                "/lib/R/library/")))
 508              #t))
 509          (add-after 'unpack 'fix-octave-modules
 510            (lambda* (#:key outputs #:allow-other-keys)
 511              (substitute* "src/interfaces/octave/CMakeLists.txt"
 512                (("^include_directories\\(\\$\\{OCTAVE_INCLUDE_DIRS\\}")
 513                 "include_directories(${OCTAVE_INCLUDE_DIRS} ${OCTAVE_INCLUDE_DIRS}/octave")
 514                ;; change target directory
 515                (("\\$\\{OCTAVE_OCT_LOCAL_API_FILE_DIR\\}")
 516                 (string-append (assoc-ref outputs "out")
 517                                "/share/octave/packages")))
 518              (substitute* '("src/interfaces/octave/swig_typemaps.i"
 519                             "src/interfaces/octave/sg_print_functions.cpp")
 520                ;; "octave/config.h" and "octave/oct-obj.h" deprecated in Octave.
 521                (("octave/config\\.h") "octave/octave-config.h")
 522                (("octave/oct-obj.h") "octave/ovl.h"))
 523              #t))
 524          (add-after 'unpack 'move-rxcpp
 525            (lambda* (#:key inputs #:allow-other-keys)
 526              (let ((rxcpp-dir "shogun/third-party/rxcpp"))
 527                (mkdir-p rxcpp-dir)
 528                (install-file (assoc-ref inputs "rxcpp") rxcpp-dir)
 529                #t)))
 530          (add-before 'build 'set-HOME
 531            ;; $HOME needs to be set at some point during the build phase
 532            (lambda _ (setenv "HOME" "/tmp") #t)))
 533        #:configure-flags
 534        (list "-DCMAKE_BUILD_WITH_INSTALL_RPATH=TRUE"
 535              "-DUSE_SVMLIGHT=OFF" ;disable proprietary SVMLIGHT
 536              "-DBUILD_META_EXAMPLES=OFF" ;requires unpackaged ctags
 537              ;;"-DINTERFACE_JAVA=ON" ;requires unpackaged jblas
 538              ;;"-DINTERFACE_RUBY=ON" ;requires unpackaged ruby-narray
 539              ;;"-DINTERFACE_PERL=ON" ;"FindPerlLibs" does not exist
 540              ;;"-DINTERFACE_LUA=ON"  ;fails because lua doesn't build pkgconfig file
 541              "-DINTERFACE_OCTAVE=ON"
 542              "-DINTERFACE_PYTHON=ON"
 543              "-DINTERFACE_R=ON")))
 544     (inputs
 545      `(("python" ,python)
 546        ("numpy" ,python-numpy)
 547        ("r-minimal" ,r-minimal)
 548        ("octave" ,octave-cli)
 549        ("swig" ,swig)
 550        ("eigen" ,eigen)
 551        ("hdf5" ,hdf5)
 552        ("atlas" ,atlas)
 553        ("arpack" ,arpack-ng)
 554        ("lapack" ,lapack)
 555        ("glpk" ,glpk)
 556        ("libxml2" ,libxml2)
 557        ("lzo" ,lzo)
 558        ("zlib" ,zlib)))
 559     (native-inputs
 560      `(("pkg-config" ,pkg-config)
 561        ("rxcpp" ,rxcpp)))
 562     ;; Non-portable SSE instructions are used so building fails on platforms
 563     ;; other than x86_64.
 564     (supported-systems '("x86_64-linux"))
 565     (home-page "http://shogun-toolbox.org/")
 566     (synopsis "Machine learning toolbox")
 567     (description
 568      "The Shogun Machine learning toolbox provides a wide range of unified and
 569 efficient Machine Learning (ML) methods.  The toolbox seamlessly allows to
 570 combine multiple data representations, algorithm classes, and general purpose
 571 tools.  This enables both rapid prototyping of data pipelines and extensibility
 572 in terms of new algorithms.")
 573     (license license:gpl3+)))
 574
 575 (define-public rxcpp
 576   (package
 577     (name "rxcpp")
 578     (version "4.1.0")
 579     (source
 580      (origin
 581        (method git-fetch)
 582        (uri (git-reference
 583              (url "https://github.com/ReactiveX/RxCpp.git")
 584              (commit (string-append "v" version))))
 585        (sha256
 586         (base32 "1rdpa3jlc181jd08nk437aar085h28i45s6nzrv65apb3xyyz0ij"))
 587        (file-name (git-file-name name version))))
 588     (build-system cmake-build-system)
 589     (arguments
 590      `(#:phases
 591        (modify-phases %standard-phases
 592          (add-after 'unpack 'remove-werror
 593            (lambda _
 594              (substitute* (find-files ".")
 595                (("-Werror") ""))
 596              #t))
 597          (replace 'check
 598            (lambda _
 599              (invoke "ctest"))))))
 600     (native-inputs
 601      `(("catch" ,catch-framework)))
 602     (home-page "http://reactivex.io/")
 603     (synopsis "Reactive Extensions for C++")
 604     (description
 605      "The Reactive Extensions for C++ (RxCpp) is a library of algorithms for
 606 values-distributed-in-time.  ReactiveX is a library for composing asynchronous
 607 and event-based programs by using observable sequences.
 608
 609 It extends the observer pattern to support sequences of data and/or events and
 610 adds operators that allow you to compose sequences together declaratively while
 611 abstracting away concerns about things like low-level threading,
 612 synchronization, thread-safety, concurrent data structures, and non-blocking
 613 I/O.")
 614     (license license:asl2.0)))
 615
 616 (define-public r-adaptivesparsity
 617   (package
 618     (name "r-adaptivesparsity")
 619     (version "1.6")
 620     (source (origin
 621               (method url-fetch)
 622               (uri (cran-uri "AdaptiveSparsity" version))
 623               (sha256
 624                (base32
 625                 "0imr5m8mll9j6n4icsv6z9rl5kbnwsp9wvzrg7n90nnmcxq2cz91"))))
 626     (properties
 627      `((upstream-name . "AdaptiveSparsity")))
 628     (build-system r-build-system)
 629     (arguments
 630      `(#:phases
 631        (modify-phases %standard-phases
 632          (add-after 'unpack 'link-against-armadillo
 633            (lambda _
 634              (substitute* "src/Makevars"
 635                (("PKG_LIBS=" prefix)
 636                 (string-append prefix "-larmadillo"))))))))
 637     (propagated-inputs
 638      `(("r-mass" ,r-mass)
 639        ("r-matrix" ,r-matrix)
 640        ("r-rcpp" ,r-rcpp)
 641        ("r-rcpparmadillo" ,r-rcpparmadillo)))
 642     (inputs
 643      `(("armadillo" ,armadillo)))
 644     (home-page "https://cran.r-project.org/web/packages/AdaptiveSparsity")
 645     (synopsis "Adaptive sparsity models")
 646     (description
 647      "This package implements the Figueiredo machine learning algorithm for
 648 adaptive sparsity and the Wong algorithm for adaptively sparse gaussian
 649 geometric models.")
 650     (license license:lgpl3+)))
 651
 652 (define-public gemmlowp-for-tensorflow
 653   ;; The commit hash is taken from "tensorflow/workspace.bzl".
 654   (let ((commit "38ebac7b059e84692f53e5938f97a9943c120d98")
 655         (revision "2"))
 656     (package
 657       (name "gemmlowp")
 658       (version (git-version "0" revision commit))
 659       (source (origin
 660                 (method url-fetch)
 661                 (uri (string-append "https://mirror.bazel.build/"
 662                                     "github.com/google/gemmlowp/archive/"
 663                                     commit ".zip"))
 664                 (file-name (string-append "gemmlowp-" version ".zip"))
 665                 (sha256
 666                  (base32
 667                   "0n56s2g8hrssm4w8qj1v58gfm56a04n9v992ixkmvk6zjiralzxq"))))
 668       (build-system cmake-build-system)
 669       (arguments
 670        `(#:configure-flags
 671          (list ,@(match (%current-system)
 672                    ((or "x86_64-linux" "i686-linux")
 673                     '("-DCMAKE_CXX_FLAGS=-msse2"))
 674                    (_ '())))
 675          #:phases
 676          (modify-phases %standard-phases
 677            ;; This directory contains the CMakeLists.txt.
 678            (add-after 'unpack 'chdir
 679              (lambda _ (chdir "contrib") #t))
 680            ;; There is no install target
 681            (replace 'install
 682              (lambda* (#:key outputs #:allow-other-keys)
 683                (let* ((out (assoc-ref outputs "out"))
 684                       (lib (string-append out "/lib/"))
 685                       (inc (string-append out "/include/")))
 686                  (install-file "../build/libeight_bit_int_gemm.so" lib)
 687                  (for-each (lambda (dir)
 688                              (let ((target (string-append inc "/" dir)))
 689                                (mkdir-p target)
 690                                (for-each (lambda (h)
 691                                            (install-file h target))
 692                                          (find-files (string-append "../" dir)
 693                                                      "\\.h$"))))
 694                            '("meta" "profiling" "public" "fixedpoint"
 695                              "eight_bit_int_gemm" "internal"))
 696                  #t))))))
 697       (native-inputs
 698        `(("unzip" ,unzip)))
 699       (home-page "https://github.com/google/gemmlowp")
 700       (synopsis "Small self-contained low-precision GEMM library")
 701       (description
 702        "This is a small self-contained low-precision @dfn{general matrix
 703 multiplication} (GEMM) library.  It is not a full linear algebra library.
 704 Low-precision means that the input and output matrix entries are integers on
 705 at most 8 bits.  To avoid overflow, results are internally accumulated on more
 706 than 8 bits, and at the end only some significant 8 bits are kept.")
 707       (license license:asl2.0))))
 708
 709 (define-public dlib
 710   (package
 711     (name "dlib")
 712     (version "19.7")
 713     (source (origin
 714               (method url-fetch)
 715               (uri (string-append
 716                     "http://dlib.net/files/dlib-" version ".tar.bz2"))
 717               (sha256
 718                (base32
 719                 "1mljz02kwkrbggyncxv5fpnyjdybw2qihaacb3js8yfkw12vwpc2"))
 720               (modules '((guix build utils)))
 721               (snippet
 722                '(begin
 723                   ;; Delete ~13MB of bundled dependencies.
 724                   (delete-file-recursively "dlib/external")
 725                   (delete-file-recursively "docs/dlib/external")
 726                   #t))))
 727     (build-system cmake-build-system)
 728     (arguments
 729      `(#:phases
 730        (modify-phases %standard-phases
 731          (add-after 'unpack 'disable-asserts
 732            (lambda _
 733              ;; config.h recommends explicitly enabling or disabling asserts
 734              ;; when building as a shared library. By default neither is set.
 735              (substitute* "dlib/config.h"
 736                (("^//#define DLIB_DISABLE_ASSERTS") "#define DLIB_DISABLE_ASSERTS"))
 737              #t))
 738          (add-after 'disable-asserts 'disable-failing-tests
 739            (lambda _
 740              ;; One test times out on MIPS, so we need to disable it.
 741              ;; Others are flaky on some platforms.
 742              (let* ((system ,(or (%current-target-system)
 743                                  (%current-system)))
 744                     (disabled-tests (cond
 745                                      ((string-prefix? "mips64" system)
 746                                       '("object_detector" ; timeout
 747                                         "data_io"))
 748                                      ((string-prefix? "armhf" system)
 749                                       '("learning_to_track"))
 750                                      ((string-prefix? "i686" system)
 751                                       '("optimization"))
 752                                      (else '()))))
 753                (for-each
 754                 (lambda (test)
 755                   (substitute* "dlib/test/makefile"
 756                     (((string-append "SRC \\+= " test "\\.cpp")) "")))
 757                 disabled-tests)
 758                #t)))
 759          (replace 'check
 760            (lambda _
 761              ;; No test target, so we build and run the unit tests here.
 762              (let ((test-dir (string-append "../dlib-" ,version "/dlib/test")))
 763                (with-directory-excursion test-dir
 764                  (invoke "make" "-j" (number->string (parallel-job-count)))
 765                  (invoke "./dtest" "--runall"))
 766                #t)))
 767          (add-after 'install 'delete-static-library
 768            (lambda* (#:key outputs #:allow-other-keys)
 769              (delete-file (string-append (assoc-ref outputs "out")
 770                                          "/lib/libdlib.a"))
 771              #t)))))
 772     (native-inputs
 773      `(("pkg-config" ,pkg-config)
 774        ;; For tests.
 775        ("libnsl" ,libnsl)))
 776     (inputs
 777      `(("giflib" ,giflib)
 778        ("lapack" ,lapack)
 779        ("libjpeg" ,libjpeg)
 780        ("libpng" ,libpng)
 781        ("libx11" ,libx11)
 782        ("openblas" ,openblas)
 783        ("zlib" ,zlib)))
 784     (synopsis
 785      "Toolkit for making machine learning and data analysis applications in C++")
 786     (description
 787      "Dlib is a modern C++ toolkit containing machine learning algorithms and
 788 tools.  It is used in both industry and academia in a wide range of domains
 789 including robotics, embedded devices, mobile phones, and large high performance
 790 computing environments.")
 791     (home-page "http://dlib.net")
 792     (license license:boost1.0)))
 793
 794 (define-public python-scikit-learn
 795   (package
 796     (name "python-scikit-learn")
 797     (version "0.20.4")
 798     (source
 799      (origin
 800        (method git-fetch)
 801        (uri (git-reference
 802              (url "https://github.com/scikit-learn/scikit-learn.git")
 803              (commit version)))
 804        (file-name (git-file-name name version))
 805        (sha256
 806         (base32
 807          "08zbzi8yx5wdlxfx9jap61vg1malc9ajf576w7a0liv6jvvrxlpj"))))
 808     (build-system python-build-system)
 809     (arguments
 810      `(#:phases
 811        (modify-phases %standard-phases
 812          (add-after 'build 'build-ext
 813            (lambda _ (invoke "python" "setup.py" "build_ext" "--inplace") #t))
 814          (replace 'check
 815            (lambda _
 816              ;; Restrict OpenBLAS threads to prevent segfaults while testing!
 817              (setenv "OPENBLAS_NUM_THREADS" "1")
 818
 819              ;; Some tests require write access to $HOME.
 820              (setenv "HOME" "/tmp")
 821
 822              (invoke "pytest" "sklearn" "-m" "not network")))
 823          (add-before 'reset-gzip-timestamps 'make-files-writable
 824            (lambda* (#:key outputs #:allow-other-keys)
 825              ;; Make sure .gz files are writable so that the
 826              ;; 'reset-gzip-timestamps' phase can do its work.
 827              (let ((out (assoc-ref outputs "out")))
 828                (for-each make-file-writable
 829                          (find-files out "\\.gz$"))
 830                #t))))))
 831     (inputs
 832      `(("openblas" ,openblas)))
 833     (native-inputs
 834      `(("python-pytest" ,python-pytest)
 835        ("python-pandas" ,python-pandas) ;for tests
 836        ("python-cython" ,python-cython)))
 837     (propagated-inputs
 838      `(("python-numpy" ,python-numpy)
 839        ("python-scipy" ,python-scipy)))
 840     (home-page "http://scikit-learn.org/")
 841     (synopsis "Machine Learning in Python")
 842     (description
 843      "Scikit-learn provides simple and efficient tools for data mining and
 844 data analysis.")
 845     (license license:bsd-3)))
 846
 847 (define-public python2-scikit-learn
 848   (package-with-python2 python-scikit-learn))
 849
 850 (define-public python-autograd
 851   (let* ((commit "442205dfefe407beffb33550846434baa90c4de7")
 852          (revision "0")
 853          (version (git-version "0.0.0" revision commit)))
 854     (package
 855       (name "python-autograd")
 856       (home-page "https://github.com/HIPS/autograd")
 857       (source (origin
 858                 (method git-fetch)
 859                 (uri (git-reference
 860                       (url home-page)
 861                       (commit commit)))
 862                 (sha256
 863                  (base32
 864                   "189sv2xb0mwnjawa9z7mrgdglc1miaq93pnck26r28fi1jdwg0z4"))
 865                 (file-name (git-file-name name version))))
 866       (version version)
 867       (build-system python-build-system)
 868       (native-inputs
 869        `(("python-nose" ,python-nose)
 870          ("python-pytest" ,python-pytest)))
 871       (propagated-inputs
 872        `(("python-future" ,python-future)
 873          ("python-numpy" ,python-numpy)))
 874       (arguments
 875        `(#:phases (modify-phases %standard-phases
 876                     (replace 'check
 877                       (lambda _
 878                         (invoke "py.test" "-v"))))))
 879       (synopsis "Efficiently computes derivatives of NumPy code")
 880       (description "Autograd can automatically differentiate native Python and
 881 NumPy code.  It can handle a large subset of Python's features, including loops,
 882 ifs, recursion and closures, and it can even take derivatives of derivatives
 883 of derivatives.  It supports reverse-mode differentiation
 884 (a.k.a. backpropagation), which means it can efficiently take gradients of
 885 scalar-valued functions with respect to array-valued arguments, as well as
 886 forward-mode differentiation, and the two can be composed arbitrarily.  The
 887 main intended application of Autograd is gradient-based optimization.")
 888       (license license:expat))))
 889
 890 (define-public python2-autograd
 891   (package-with-python2 python-autograd))
 892
 893 (define-public lightgbm
 894   (package
 895     (name "lightgbm")
 896     (version "2.0.12")
 897     (source (origin
 898               (method url-fetch)
 899               (uri (string-append
 900                     "https://github.com/Microsoft/LightGBM/archive/v"
 901                     version ".tar.gz"))
 902               (sha256
 903                (base32
 904                 "132zf0yk0545mg72hyzxm102g3hpb6ixx9hnf8zd2k55gas6cjj1"))
 905               (file-name (string-append name "-" version ".tar.gz"))))
 906     (native-inputs
 907      `(("python-pytest" ,python-pytest)
 908        ("python-nose" ,python-nose)))
 909     (inputs
 910      `(("openmpi" ,openmpi)))
 911     (propagated-inputs
 912      `(("python-numpy" ,python-numpy)
 913        ("python-scipy" ,python-scipy)))
 914     (arguments
 915      `(#:configure-flags
 916        '("-DUSE_MPI=ON")
 917        #:phases
 918        (modify-phases %standard-phases
 919          (replace 'check
 920            (lambda* (#:key outputs #:allow-other-keys)
 921              (with-directory-excursion ,(string-append "../LightGBM-" version)
 922                (invoke "pytest" "tests/c_api_test/test_.py")))))))
 923     (build-system cmake-build-system)
 924     (home-page "https://github.com/Microsoft/LightGBM")
 925     (synopsis "Gradient boosting framework based on decision tree algorithms")
 926     (description "LightGBM is a gradient boosting framework that uses tree
 927 based learning algorithms.  It is designed to be distributed and efficient with
 928 the following advantages:
 929
 930 @itemize
 931 @item Faster training speed and higher efficiency
 932 @item Lower memory usage
 933 @item Better accuracy
 934 @item Parallel and GPU learning supported (not enabled in this package)
 935 @item Capable of handling large-scale data
 936 @end itemize\n")
 937     (license license:expat)))
 938
 939 (define-public vowpal-wabbit
 940   ;; Language bindings not included.
 941   (package
 942     (name "vowpal-wabbit")
 943     (version "8.5.0")
 944     (source (origin
 945               (method url-fetch)
 946               (uri (string-append
 947                     "https://github.com/JohnLangford/vowpal_wabbit/archive/"
 948                     version ".tar.gz"))
 949               (sha256
 950                (base32
 951                 "0clp2kb7rk5sckhllxjr5a651awf4s8dgzg4659yh4hf5cqnf0gr"))
 952               (file-name (string-append name "-" version ".tar.gz"))))
 953     (inputs
 954      `(("boost" ,boost)
 955        ("zlib" ,zlib)))
 956     (arguments
 957      `(#:configure-flags
 958        (list (string-append "--with-boost="
 959                             (assoc-ref %build-inputs "boost")))))
 960     (build-system gnu-build-system)
 961     (home-page "https://github.com/JohnLangford/vowpal_wabbit")
 962     (synopsis "Fast machine learning library for online learning")
 963     (description "Vowpal Wabbit is a machine learning system with techniques
 964 such as online, hashing, allreduce, reductions, learning2search, active, and
 965 interactive learning.")
 966     (license license:bsd-3)))
 967
 968 (define-public python2-fastlmm
 969   (package
 970     (name "python2-fastlmm")
 971     (version "0.2.21")
 972     (source
 973      (origin
 974        (method url-fetch)
 975        (uri (pypi-uri "fastlmm" version ".zip"))
 976        (sha256
 977         (base32
 978          "1q8c34rpmwkfy3r4d5172pzdkpfryj561897z9r3x22gq7813x1m"))))
 979     (build-system python-build-system)
 980     (arguments
 981      `(#:tests? #f ; some test files are missing
 982        #:python ,python-2)) ; only Python 2.7 is supported
 983     (propagated-inputs
 984      `(("python2-numpy" ,python2-numpy)
 985        ("python2-scipy" ,python2-scipy)
 986        ("python2-matplotlib" ,python2-matplotlib)
 987        ("python2-pandas" ,python2-pandas)
 988        ("python2-scikit-learn" ,python2-scikit-learn)
 989        ("python2-pysnptools" ,python2-pysnptools)))
 990     (native-inputs
 991      `(("unzip" ,unzip)
 992        ("python2-cython" ,python2-cython)
 993        ("python2-mock" ,python2-mock)
 994        ("python2-nose" ,python2-nose)))
 995     (home-page "http://research.microsoft.com/en-us/um/redmond/projects/mscompbio/fastlmm/")
 996     (synopsis "Perform genome-wide association studies on large data sets")
 997     (description
 998      "FaST-LMM, which stands for Factored Spectrally Transformed Linear Mixed
 999 Models, is a program for performing both single-SNP and SNP-set genome-wide
1000 association studies (GWAS) on extremely large data sets.")
1001     (license license:asl2.0)))
1002
1003 ;; There have been no proper releases yet.
1004 (define-public kaldi
1005   (let ((commit "2f95609f0bb085bd3a1dc5eb0a39f3edea59e606")
1006         (revision "1"))
1007     (package
1008       (name "kaldi")
1009       (version (git-version "0" revision commit))
1010       (source (origin
1011                 (method git-fetch)
1012                 (uri (git-reference
1013                       (url "https://github.com/kaldi-asr/kaldi.git")
1014                       (commit commit)))
1015                 (file-name (git-file-name name version))
1016                 (sha256
1017                  (base32
1018                   "082qh3pfi7hvncylp4xsmkfahbd7gb0whdfa4rwrx7fxk9rdh3kz"))))
1019       (build-system gnu-build-system)
1020       (arguments
1021        `(#:test-target "test"
1022          #:phases
1023          (modify-phases %standard-phases
1024            (add-after 'unpack 'chdir
1025              (lambda _ (chdir "src") #t))
1026            (replace 'configure
1027              (lambda* (#:key build system inputs outputs #:allow-other-keys)
1028                (when (not (or (string-prefix? "x86_64" system)
1029                               (string-prefix? "i686" system)))
1030                  (substitute* "makefiles/linux_openblas.mk"
1031                    (("-msse -msse2") "")))
1032                (substitute* "makefiles/default_rules.mk"
1033                  (("/bin/bash") (which "bash")))
1034                (substitute* "Makefile"
1035                  (("ext_depend: check_portaudio")
1036                   "ext_depend:"))
1037                (substitute* '("online/Makefile"
1038                               "onlinebin/Makefile"
1039                               "gst-plugin/Makefile")
1040                  (("../../tools/portaudio/install")
1041                   (assoc-ref inputs "portaudio")))
1042
1043                ;; This `configure' script doesn't support variables passed as
1044                ;; arguments, nor does it support "prefix".
1045                (let ((out (assoc-ref outputs "out"))
1046                      (openblas (assoc-ref inputs "openblas"))
1047                      (openfst (assoc-ref inputs "openfst")))
1048                  (substitute* "configure"
1049                    (("check_for_slow_expf;") "")
1050                    ;; This affects the RPATH and also serves as the installation
1051                    ;; directory.
1052                    (("KALDILIBDIR=`pwd`/lib")
1053                     (string-append "KALDILIBDIR=" out "/lib")))
1054                  (mkdir-p out) ; must exist
1055                  (setenv "CONFIG_SHELL" (which "bash"))
1056                  (setenv "OPENFST_VER" ,(package-version openfst))
1057                  (invoke "./configure"
1058                          "--use-cuda=no"
1059                          "--shared"
1060                          (string-append "--openblas-root=" openblas)
1061                          (string-append "--fst-root=" openfst)))))
1062            (add-after 'build 'build-ext-and-gstreamer-plugin
1063              (lambda _
1064                (invoke "make" "-C" "online" "depend")
1065                (invoke "make" "-C" "online")
1066                (invoke "make" "-C" "onlinebin" "depend")
1067                (invoke "make" "-C" "onlinebin")
1068                (invoke "make" "-C" "gst-plugin" "depend")
1069                (invoke "make" "-C" "gst-plugin")
1070                #t))
1071            ;; TODO: also install the executables.
1072            (replace 'install
1073              (lambda* (#:key outputs #:allow-other-keys)
1074                (let* ((out (assoc-ref outputs "out"))
1075                       (inc (string-append out "/include"))
1076                       (lib (string-append out "/lib")))
1077                  (mkdir-p lib)
1078                  ;; The build phase installed symlinks to the actual
1079                  ;; libraries.  Install the actual targets.
1080                  (for-each (lambda (file)
1081                              (let ((target (readlink file)))
1082                                (delete-file file)
1083                                (install-file target lib)))
1084                            (find-files lib "\\.so"))
1085                  ;; Install headers
1086                  (for-each (lambda (file)
1087                              (let ((target-dir (string-append inc "/" (dirname file))))
1088                                (install-file file target-dir)))
1089                            (find-files "." "\\.h"))
1090                  (install-file "gst-plugin/libgstonlinegmmdecodefaster.so"
1091                                (string-append lib "/gstreamer-1.0"))
1092                  #t))))))
1093       (inputs
1094        `(("alsa-lib" ,alsa-lib)
1095          ("gfortran" ,gfortran "lib")
1096          ("glib" ,glib)
1097          ("gstreamer" ,gstreamer)
1098          ("jack" ,jack-1)
1099          ("openblas" ,openblas)
1100          ("openfst" ,openfst)
1101          ("portaudio" ,portaudio)
1102          ("python" ,python)))
1103       (native-inputs
1104        `(("glib" ,glib "bin")             ; glib-genmarshal
1105          ("grep" ,grep)
1106          ("sed" ,sed)
1107          ("pkg-config" ,pkg-config)
1108          ("which" ,which)))
1109       (home-page "https://kaldi-asr.org/")
1110       (synopsis "Speech recognition toolkit")
1111       (description "Kaldi is an extensible toolkit for speech recognition
1112 written in C++.")
1113       (license license:asl2.0))))
1114
1115 (define-public gst-kaldi-nnet2-online
1116   (let ((commit "617e43e73c7cc45eb9119028c02bd4178f738c4a")
1117         (revision "1"))
1118     (package
1119       (name "gst-kaldi-nnet2-online")
1120       (version (git-version "0" revision commit))
1121       (source (origin
1122                 (method git-fetch)
1123                 (uri (git-reference
1124                       (url "https://github.com/alumae/gst-kaldi-nnet2-online.git")
1125                       (commit commit)))
1126                 (file-name (git-file-name name version))
1127                 (sha256
1128                  (base32
1129                   "0xh3w67b69818s6ib02ara4lw7wamjdmh4jznvkpzrs4skbs9jx9"))))
1130       (build-system gnu-build-system)
1131       (arguments
1132        `(#:tests? #f                    ; there are none
1133          #:make-flags
1134          (list (string-append "SHELL="
1135                               (assoc-ref %build-inputs "bash") "/bin/bash")
1136                (string-append "KALDI_ROOT="
1137                               (assoc-ref %build-inputs "kaldi-src"))
1138                (string-append "KALDILIBDIR="
1139                               (assoc-ref %build-inputs "kaldi") "/lib")
1140                "KALDI_FLAVOR=dynamic")
1141          #:phases
1142          (modify-phases %standard-phases
1143            (add-after 'unpack 'chdir
1144              (lambda _ (chdir "src") #t))
1145            (replace 'configure
1146              (lambda* (#:key inputs #:allow-other-keys)
1147                (let ((glib (assoc-ref inputs "glib")))
1148                  (setenv "CXXFLAGS" "-fPIC")
1149                  (setenv "CPLUS_INCLUDE_PATH"
1150                          (string-append glib "/include/glib-2.0:"
1151                                         glib "/lib/glib-2.0/include:"
1152                                         (assoc-ref inputs "gstreamer")
1153                                         "/include/gstreamer-1.0")))
1154                (substitute* "Makefile"
1155                  (("include \\$\\(KALDI_ROOT\\)/src/kaldi.mk") "")
1156                  (("\\$\\(error Cannot find") "#"))
1157                #t))
1158            (add-before 'build 'build-depend
1159              (lambda* (#:key make-flags #:allow-other-keys)
1160                (apply invoke "make" "depend" make-flags)))
1161            (replace 'install
1162              (lambda* (#:key outputs #:allow-other-keys)
1163                (let* ((out (assoc-ref outputs "out"))
1164                       (lib (string-append out "/lib/gstreamer-1.0")))
1165                  (install-file "libgstkaldinnet2onlinedecoder.so" lib)
1166                  #t))))))
1167       (inputs
1168        `(("glib" ,glib)
1169          ("gstreamer" ,gstreamer)
1170          ("jansson" ,jansson)
1171          ("openfst" ,openfst)
1172          ("kaldi" ,kaldi)))
1173       (native-inputs
1174        `(("bash" ,bash)
1175          ("glib:bin" ,glib "bin")       ; glib-genmarshal
1176          ("kaldi-src" ,(package-source kaldi))
1177          ("pkg-config" ,pkg-config)))
1178       (home-page "https://kaldi-asr.org/")
1179       (synopsis "Gstreamer plugin for decoding speech")
1180       (description "This package provides a GStreamer plugin that wraps
1181 Kaldi's @code{SingleUtteranceNnet2Decoder}.  It requires iVector-adapted DNN
1182 acoustic models.  The iVectors are adapted to the current audio stream
1183 automatically.")
1184       (license license:asl2.0))))
1185
1186 (define-public kaldi-gstreamer-server
1187   (let ((commit "1735ba49c5dc0ebfc184e45105fc600cd9f1f508")
1188         (revision "1"))
1189     (package
1190       (name "kaldi-gstreamer-server")
1191       (version (git-version "0" revision commit))
1192       (source (origin
1193                 (method git-fetch)
1194                 (uri (git-reference
1195                       (url "https://github.com/alumae/kaldi-gstreamer-server.git")
1196                       (commit commit)))
1197                 (file-name (git-file-name name version))
1198                 (sha256
1199                  (base32
1200                   "0j701m7lbwmzqxsfanj882v7881hrbmpqybbczbxqpcbg8q34w0k"))))
1201       (build-system gnu-build-system)
1202       (arguments
1203        `(#:tests? #f ; there are no tests that can be run automatically
1204          #:modules ((guix build utils)
1205                     (guix build gnu-build-system)
1206                     (srfi srfi-26))
1207          #:phases
1208          (modify-phases %standard-phases
1209            (delete 'configure)
1210            (replace 'build
1211              (lambda* (#:key outputs #:allow-other-keys)
1212                ;; Disable hash randomization to ensure the generated .pycs
1213                ;; are reproducible.
1214                (setenv "PYTHONHASHSEED" "0")
1215                (with-directory-excursion "kaldigstserver"
1216                  (for-each (lambda (file)
1217                              (apply invoke
1218                                     `("python"
1219                                       "-m" "compileall"
1220                                       "-f" ; force rebuild
1221                                       ,file)))
1222                            (find-files "." "\\.py$")))
1223                #t))
1224            (replace 'install
1225              (lambda* (#:key inputs outputs #:allow-other-keys)
1226                (let* ((out (assoc-ref outputs "out"))
1227                       (bin (string-append out "/bin"))
1228                       (share (string-append out "/share/kaldi-gstreamer-server/")))
1229                  ;; Install Python files
1230                  (with-directory-excursion "kaldigstserver"
1231                    (for-each (cut install-file <> share)
1232                              (find-files "." ".*")))
1233
1234                  ;; Install sample configuration files
1235                  (for-each (cut install-file <> share)
1236                            (find-files "." "\\.yaml"))
1237
1238                  ;; Install executables
1239                  (mkdir-p bin)
1240                  (let* ((server (string-append bin "/kaldi-gst-server"))
1241                         (client (string-append bin "/kaldi-gst-client"))
1242                         (worker (string-append bin "/kaldi-gst-worker"))
1243                         (PYTHONPATH (getenv "PYTHONPATH"))
1244                         (GST_PLUGIN_PATH (string-append
1245                                           (assoc-ref inputs "gst-kaldi-nnet2-online")
1246                                           "/lib/gstreamer-1.0:${GST_PLUGIN_PATH}"))
1247                         (wrap (lambda (wrapper what)
1248                                 (with-output-to-file wrapper
1249                                   (lambda _
1250                                     (format #t
1251                                             "#!~a
1252 export PYTHONPATH=~a
1253 export GST_PLUGIN_PATH=~a
1254 exec ~a ~a/~a \"$@\"~%"
1255                                             (which "bash") PYTHONPATH GST_PLUGIN_PATH
1256                                             (which "python") share what)))
1257                                 (chmod wrapper #o555))))
1258                    (for-each wrap
1259                              (list server client worker)
1260                              (list "master_server.py"
1261                                    "client.py"
1262                                    "worker.py")))
1263                  #t))))))
1264       (inputs
1265        `(("gst-kaldi-nnet2-online" ,gst-kaldi-nnet2-online)
1266          ("python2" ,python-2)
1267          ("python2-futures" ,python2-futures)
1268          ("python2-pygobject" ,python2-pygobject)
1269          ("python2-pyyaml" ,python2-pyyaml)
1270          ("python2-tornado" ,python2-tornado)
1271          ("python2-ws4py" ,python2-ws4py-for-kaldi-gstreamer-server)))
1272       (home-page "https://github.com/alumae/kaldi-gstreamer-server")
1273       (synopsis "Real-time full-duplex speech recognition server")
1274       (description "This is a real-time full-duplex speech recognition server,
1275 based on the Kaldi toolkit and the GStreamer framework and implemented in
1276 Python.")
1277       (license license:bsd-2))))
1278
1279 (define-public grpc
1280   (package
1281     (name "grpc")
1282     (version "1.16.1")
1283     (source (origin
1284               (method git-fetch)
1285               (uri (git-reference
1286                     (url "https://github.com/grpc/grpc.git")
1287                     (commit (string-append "v" version))))
1288               (file-name (git-file-name name version))
1289               (sha256
1290                (base32
1291                 "1jimqz3115f9pli5w6ik9wi7mjc7ix6y7yrq4a1ab9fc3dalj7p2"))))
1292     (build-system cmake-build-system)
1293     (arguments
1294      `(#:tests? #f ; no test target
1295        #:configure-flags
1296        (list "-DgRPC_ZLIB_PROVIDER=package"
1297              "-DgRPC_CARES_PROVIDER=package"
1298              "-DgRPC_SSL_PROVIDER=package"
1299              "-DgRPC_PROTOBUF_PROVIDER=package")))
1300     (inputs
1301      `(("c-ares" ,c-ares/cmake)
1302        ("openssl" ,openssl)
1303        ("zlib" ,zlib)))
1304     (native-inputs
1305      `(("protobuf" ,protobuf-next)
1306        ("python" ,python-wrapper)))
1307     (home-page "https://grpc.io")
1308     (synopsis "High performance universal RPC framework")
1309     (description "gRPC is a modern open source high performance @dfn{Remote
1310 Procedure Call} (RPC) framework that can run in any environment.  It can
1311 efficiently connect services in and across data centers with pluggable support
1312 for load balancing, tracing, health checking and authentication.  It is also
1313 applicable in last mile of distributed computing to connect devices, mobile
1314 applications and browsers to backend services.")
1315     (license license:asl2.0)))
1316
1317 ;; Note that Tensorflow includes a "third_party" directory, which seems to not
1318 ;; only contain modified subsets of upstream library source code, but also
1319 ;; adapter headers provided by Google (such as the fft.h header, which is not
1320 ;; part of the upstream project code).  The Tensorflow code includes headers
1321 ;; from the "third_party" directory.  It does not look like we can replace
1322 ;; these headers with unmodified upstream files, so we keep them.
1323 (define-public tensorflow
1324   (package
1325     (name "tensorflow")
1326     (version "1.9.0")
1327     (source
1328      (origin
1329        (method git-fetch)
1330        (uri (git-reference
1331              (url "https://github.com/tensorflow/tensorflow.git")
1332              (commit (string-append "v" version))))
1333        (file-name (string-append "tensorflow-" version "-checkout"))
1334        (sha256
1335         (base32
1336          "0a9kwha395g3wgxfwln5j8vn9nkspmd75xldrlqdq540w996g8xa"))))
1337     (build-system cmake-build-system)
1338     (arguments
1339      `(#:tests? #f                      ; no "check" target
1340        #:build-type "Release"
1341        #:configure-flags
1342        (let ((protobuf (assoc-ref %build-inputs "protobuf"))
1343              (protobuf:native (assoc-ref %build-inputs "protobuf:native"))
1344              (jsoncpp (assoc-ref %build-inputs "jsoncpp"))
1345              (snappy (assoc-ref %build-inputs "snappy"))
1346              (sqlite (assoc-ref %build-inputs "sqlite")))
1347          (list
1348           ;; Use protobuf from Guix
1349           (string-append "-Dprotobuf_STATIC_LIBRARIES="
1350                          protobuf "/lib/libprotobuf.so")
1351           (string-append "-DPROTOBUF_PROTOC_EXECUTABLE="
1352                          protobuf:native "/bin/protoc")
1353
1354           ;; Use snappy from Guix
1355           (string-append "-Dsnappy_STATIC_LIBRARIES="
1356                          snappy "/lib/libsnappy.so")
1357           ;; Yes, this is not actually the include directory but a prefix...
1358           (string-append "-Dsnappy_INCLUDE_DIR=" snappy)
1359
1360           ;; Use jsoncpp from Guix
1361           (string-append "-Djsoncpp_STATIC_LIBRARIES="
1362                          jsoncpp "/lib/libjsoncpp.so")
1363           ;; Yes, this is not actually the include directory but a prefix...
1364           (string-append "-Djsoncpp_INCLUDE_DIR=" jsoncpp)
1365
1366           ;; Use sqlite from Guix
1367           (string-append "-Dsqlite_STATIC_LIBRARIES="
1368                          sqlite "/lib/libsqlite.a")
1369
1370           ;; Use system libraries wherever possible.  Currently, this
1371           ;; only affects zlib.
1372           "-Dsystemlib_ALL=ON"
1373           "-Dtensorflow_ENABLE_POSITION_INDEPENDENT_CODE=ON"
1374           "-Dtensorflow_BUILD_SHARED_LIB=ON"
1375           "-Dtensorflow_OPTIMIZE_FOR_NATIVE_ARCH=OFF"
1376           "-Dtensorflow_ENABLE_SSL_SUPPORT=OFF"
1377           "-Dtensorflow_BUILD_CONTRIB_KERNELS=OFF"))
1378        #:make-flags
1379        (list "CC=gcc")
1380        #:modules ((ice-9 ftw)
1381                   (guix build utils)
1382                   (guix build cmake-build-system))
1383        #:phases
1384        (modify-phases %standard-phases
1385          (add-after 'unpack 'set-source-file-times-to-1980
1386            ;; At the end of the tf_python_build_pip_package target, a ZIP
1387            ;; archive should be generated via bdist_wheel, but it fails with
1388            ;; "ZIP does not support timestamps before 1980".  Luckily,
1389            ;; SOURCE_DATE_EPOCH is respected, which we set to some time in
1390            ;; 1980.
1391            (lambda _ (setenv "SOURCE_DATE_EPOCH" "315532800") #t))
1392          ;; See https://github.com/tensorflow/tensorflow/issues/20517#issuecomment-406373913
1393          (add-after 'unpack 'python3.7-compatibility
1394            (lambda _
1395              (substitute* '("tensorflow/python/eager/pywrap_tfe_src.cc"
1396                             "tensorflow/python/lib/core/ndarray_tensor.cc"
1397                             "tensorflow/python/lib/core/py_func.cc")
1398                (("PyUnicode_AsUTF8") "(char *)PyUnicode_AsUTF8"))
1399              (substitute* "tensorflow/c/eager/c_api.h"
1400                (("unsigned char async")
1401                 "unsigned char is_async"))
1402
1403              ;; Remove dependency on tensorboard, a complicated but probably
1404              ;; optional package.
1405              (substitute* "tensorflow/tools/pip_package/setup.py"
1406                ((".*'tensorboard >.*") ""))
1407              #t))
1408          (add-after 'python3.7-compatibility 'chdir
1409            (lambda _ (chdir "tensorflow/contrib/cmake") #t))
1410          (add-after 'chdir 'disable-downloads
1411            (lambda* (#:key inputs #:allow-other-keys)
1412              (substitute* (find-files "external" "\\.cmake$")
1413                (("GIT_REPOSITORY.*") "")
1414                (("GIT_TAG.*") "")
1415                (("PREFIX ")
1416                 "DOWNLOAD_COMMAND \"\"\nPREFIX "))
1417
1418              ;; Use packages from Guix
1419              (let ((grpc (assoc-ref inputs "grpc")))
1420                (substitute* "CMakeLists.txt"
1421                  ;; Sqlite
1422                  (("include\\(sqlite\\)") "")
1423                  (("\\$\\{sqlite_STATIC_LIBRARIES\\}")
1424                   (string-append (assoc-ref inputs "sqlite")
1425                                  "/lib/libsqlite3.so"))
1426                  (("sqlite_copy_headers_to_destination") "")
1427
1428                  ;; PNG
1429                  (("include\\(png\\)") "")
1430                  (("\\$\\{png_STATIC_LIBRARIES\\}")
1431                   (string-append (assoc-ref inputs "libpng")
1432                                  "/lib/libpng16.so"))
1433                  (("png_copy_headers_to_destination") "")
1434
1435                  ;; JPEG
1436                  (("include\\(jpeg\\)") "")
1437                  (("\\$\\{jpeg_STATIC_LIBRARIES\\}")
1438                   (string-append (assoc-ref inputs "libjpeg")
1439                                  "/lib/libjpeg.so"))
1440                  (("jpeg_copy_headers_to_destination") "")
1441
1442                  ;; GIF
1443                  (("include\\(gif\\)") "")
1444                  (("\\$\\{gif_STATIC_LIBRARIES\\}")
1445                   (string-append (assoc-ref inputs "giflib")
1446                                  "/lib/libgif.so"))
1447                  (("gif_copy_headers_to_destination") "")
1448
1449                  ;; lmdb
1450                  (("include\\(lmdb\\)") "")
1451                  (("\\$\\{lmdb_STATIC_LIBRARIES\\}")
1452                   (string-append (assoc-ref inputs "lmdb")
1453                                  "/lib/liblmdb.so"))
1454                  (("lmdb_copy_headers_to_destination") "")
1455
1456                  ;; Protobuf
1457                  (("include\\(protobuf\\)") "")
1458                  (("protobuf_copy_headers_to_destination") "")
1459                  (("^ +protobuf") "")
1460
1461                  ;; gRPC
1462                  (("include\\(grpc\\)")
1463                   "find_package(grpc REQUIRED NAMES gRPC)")
1464                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES grpc\\)") "")
1465
1466                  ;; Eigen
1467                  (("include\\(eigen\\)")
1468                   (string-append "find_package(eigen REQUIRED NAMES Eigen3)
1469 set(eigen_INCLUDE_DIRS ${CMAKE_CURRENT_BINARY_DIR}/external/eigen_archive "
1470                                  (assoc-ref inputs "eigen") "/include/eigen3)"))
1471                  (("^ +eigen") "")
1472
1473                  ;; snappy
1474                  (("include\\(snappy\\)")
1475                   "add_definitions(-DTF_USE_SNAPPY)")
1476                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES snappy\\)") "")
1477
1478                  ;; jsoncpp
1479                  (("include\\(jsoncpp\\)") "")
1480                  (("^ +jsoncpp") ""))
1481
1482                (substitute* "tf_core_framework.cmake"
1483                  ((" grpc") "")
1484                  (("\\$\\{GRPC_BUILD\\}/grpc_cpp_plugin")
1485                   (which "grpc_cpp_plugin"))
1486                  ;; Link with gRPC libraries
1487                  (("add_library\\(tf_protos_cc.*" m)
1488                   (string-append m
1489                                  (format #f "\ntarget_link_libraries(tf_protos_cc PRIVATE \
1490 ~a/lib/libgrpc++_unsecure.a \
1491 ~a/lib/libgrpc_unsecure.a \
1492 ~a/lib/libaddress_sorting.a \
1493 ~a/lib/libgpr.a \
1494 ~a//lib/libcares.so
1495 )\n"
1496                                          grpc grpc grpc grpc
1497                                          (assoc-ref inputs "c-ares"))))))
1498              (substitute* "tf_tools.cmake"
1499                (("add_dependencies\\(\\$\\{proto_text.*") ""))
1500              ;; Remove dependency on bundled grpc
1501              (substitute* "tf_core_distributed_runtime.cmake"
1502                (("tf_core_cpu grpc") "tf_core_cpu"))
1503
1504              ;; This directory is a dependency of many targets.
1505              (mkdir-p "protobuf")
1506              #t))
1507          (add-after 'configure 'unpack-third-party-sources
1508            (lambda* (#:key inputs #:allow-other-keys)
1509              ;; This is needed to configure bundled packages properly.
1510              (setenv "CONFIG_SHELL" (which "bash"))
1511              (for-each
1512               (lambda (name)
1513                 (let* ((what  (assoc-ref inputs (string-append name "-src")))
1514                        (name* (string-map (lambda (c)
1515                                             (if (char=? c #\-)
1516                                                 #\_ c)) name))
1517                        (where (string-append "../build/" name* "/src/" name*)))
1518                   (cond
1519                    ((string-suffix? ".zip" what)
1520                     (mkdir-p where)
1521                     (with-directory-excursion where
1522                       (invoke "unzip" what)))
1523                    ((string-suffix? ".tar.gz" what)
1524                     (mkdir-p where)
1525                     (invoke "tar" "xf" what
1526                             "-C" where "--strip-components=1"))
1527                    (else
1528                     (let ((parent (dirname where)))
1529                       (mkdir-p parent)
1530                       (with-directory-excursion parent
1531                         (when (file-exists? name*)
1532                           (delete-file-recursively name*))
1533                         (copy-recursively what name*)
1534                         (map make-file-writable
1535                              (find-files name* ".*"))))))))
1536               (list "boringssl"
1537                     "cub"
1538                     "double-conversion"
1539                     "farmhash"
1540                     "fft2d"
1541                     "highwayhash"
1542                     "nsync"
1543                     "re2"))
1544
1545              (rename-file "../build/cub/src/cub/cub-1.8.0/"
1546                           "../build/cub/src/cub/cub/")
1547              #t))
1548          (add-after 'unpack 'fix-python-build
1549            (lambda* (#:key inputs outputs #:allow-other-keys)
1550              (mkdir-p "protobuf-src")
1551              (invoke "tar" "xf" (assoc-ref inputs "protobuf:src")
1552                      "-C" "protobuf-src" "--strip-components=1")
1553              (mkdir-p "eigen-src")
1554              (invoke "tar" "xf" (assoc-ref inputs "eigen:src")
1555                      "-C" "eigen-src" "--strip-components=1")
1556
1557              (substitute* "tensorflow/contrib/cmake/tf_python.cmake"
1558                ;; Ensure that all Python dependencies can be found at build time.
1559                (("PYTHONPATH=\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/tf_python" m)
1560                 (string-append m ":" (getenv "PYTHONPATH")))
1561                ;; Take protobuf source files from our source package.
1562                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/protobuf/src/protobuf/src/google")
1563                 (string-append (getcwd) "/protobuf-src/src/google")))
1564
1565              (substitute* '("tensorflow/contrib/cmake/tf_shared_lib.cmake"
1566                             "tensorflow/contrib/cmake/tf_python.cmake")
1567                ;; Take Eigen source files from our source package.
1568                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/eigen/src/eigen/")
1569                 (string-append (getcwd) "/eigen-src/"))
1570                ;; Take Eigen headers from our own package.
1571                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/external/eigen_archive")
1572                 (string-append (assoc-ref inputs "eigen") "/include/eigen3")))
1573
1574              ;; Correct the RUNPATH of ops libraries generated for Python.
1575              ;; TODO: this doesn't work :(
1576              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1577              ;; warning: RUNPATH contains bogus entries: ("/tmp/guix-build-tensorflow-1.9.0.drv-0/source/tensorflow/contrib/build")
1578              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1579              ;; error: depends on 'libpywrap_tensorflow_internal.so', which
1580              ;; cannot be found in RUNPATH ...
1581              (substitute* "tensorflow/contrib/cmake/tf_cc_ops.cmake"
1582                (("set_target_properties.*")
1583                 (string-append "set_target_properties(${_AT_TARGET} PROPERTIES \
1584 COMPILE_FLAGS ${target_compile_flags} \
1585 INSTALL_RPATH_USE_LINK_PATH TRUE \
1586 INSTALL_RPATH " (assoc-ref outputs "out") "/lib)\n")))
1587              #t))
1588          (add-after 'build 'build-pip-package
1589            (lambda* (#:key outputs #:allow-other-keys)
1590              (setenv "LDFLAGS"
1591                      (string-append "-Wl,-rpath="
1592                                     (assoc-ref outputs "out") "/lib"))
1593              (invoke "make" "tf_python_build_pip_package")
1594              #t))
1595          (add-after 'build-pip-package 'install-python
1596            (lambda* (#:key outputs #:allow-other-keys)
1597              (let ((out (assoc-ref outputs "out"))
1598                    (wheel (car (find-files "../build/tf_python/dist/" "\\.whl$"))))
1599                (invoke "python" "-m" "pip" "install" wheel
1600                        (string-append "--prefix=" out))
1601
1602                ;; XXX: broken RUNPATH, see fix-python-build phase.
1603                (delete-file
1604                 (string-append
1605                  out "/lib/python3.7/site-packages/tensorflow/contrib/"
1606                  "seq2seq/python/ops/lib_beam_search_ops.so"))
1607                #t))))))
1608     (native-inputs
1609      `(("pkg-config" ,pkg-config)
1610        ("protobuf:native" ,protobuf-next) ; protoc
1611        ("protobuf:src" ,(package-source protobuf-next))
1612        ("eigen:src" ,(package-source eigen-for-tensorflow))
1613        ;; install_pip_packages.sh wants setuptools 39.1.0 specifically.
1614        ("python-setuptools" ,python-setuptools-for-tensorflow)
1615
1616        ;; The commit hashes and URLs for third-party source code are taken
1617        ;; from "tensorflow/workspace.bzl".
1618        ("boringssl-src"
1619         ,(let ((commit "ee7aa02")
1620                (revision "1"))
1621            (origin
1622              (method git-fetch)
1623              (uri (git-reference
1624                    (url "https://boringssl.googlesource.com/boringssl")
1625                    (commit commit)))
1626              (file-name (string-append "boringssl-0-" revision
1627                                        (string-take commit 7)
1628                                        "-checkout"))
1629              (sha256
1630               (base32
1631                "1jf693q0nw0adsic6cgmbdx6g7wr4rj4vxa8j1hpn792fqhd8wgw")))))
1632        ("cub-src"
1633         ,(let ((version "1.8.0"))
1634            (origin
1635              (method url-fetch)
1636              (uri (string-append "https://mirror.bazel.build/github.com/NVlabs/"
1637                                  "cub/archive/" version ".zip"))
1638              (file-name (string-append "cub-" version ".zip"))
1639              (sha256
1640               (base32
1641                "1hsqikqridb90dkxkjr2918dcry6pfh46ccnwrzawl56aamhdykb")))))
1642        ("double-conversion-src"
1643         ,(let ((commit "5664746")
1644                (revision "1"))
1645            (origin
1646              (method git-fetch)
1647              (uri (git-reference
1648                    (url "https://github.com/google/double-conversion.git")
1649                    (commit commit)))
1650              (file-name
1651               (git-file-name "double-conversion"
1652                              (string-append "0-" revision "."
1653                                             (string-take commit 7))))
1654              (sha256
1655               (base32
1656                "1h5lppqqxcvdg5jq42i5msgwx20ryij3apvmndflngrgdpc04gn1")))))
1657        ("farmhash-src"
1658         ,(let ((commit "816a4ae622e964763ca0862d9dbd19324a1eaf45"))
1659            (origin
1660              (method url-fetch)
1661              (uri (string-append
1662                    "https://mirror.bazel.build/github.com/google/farmhash/archive/"
1663                    commit ".tar.gz"))
1664              (file-name (string-append "farmhash-0-" (string-take commit 7)
1665                                        ".tar.gz"))
1666              (sha256
1667               (base32
1668                "185b2xdxl4d4cnsnv6abg8s22gxvx8673jq2yaq85bz4cdy58q35")))))
1669        ;; The license notice on the home page at
1670        ;; http://www.kurims.kyoto-u.ac.jp/~ooura/fft.html says:
1671        ;;   Copyright Takuya OOURA, 1996-2001
1672        ;;
1673        ;;   You may use, copy, modify and distribute this code for any purpose
1674        ;;   (include commercial use) and without fee. Please refer to this
1675        ;;   package when you modify this code.
1676        ;;
1677        ;; We take the identical tarball from the Bazel mirror, because the URL
1678        ;; at the home page is not versioned and might change.
1679        ("fft2d-src"
1680         ,(origin
1681            (method url-fetch)
1682            (uri "https://mirror.bazel.build/www.kurims.kyoto-u.ac.jp/~ooura/fft.tgz")
1683            (file-name "fft2d.tar.gz")
1684            (sha256
1685             (base32
1686              "15jjkfvhqvl2c0753d2di8hz0pyzn598g74wqy79awdrf1y67fsj"))))
1687        ("highwayhash-src"
1688         ,(let ((commit "be5edafc2e1a455768e260ccd68ae7317b6690ee")
1689                (revision "1"))
1690            (origin
1691              (method git-fetch)
1692              (uri (git-reference
1693                    (url "https://github.com/google/highwayhash.git")
1694                    (commit commit)))
1695              (file-name (string-append "highwayhash-0-" revision
1696                                        (string-take commit 7)
1697                                        "-checkout"))
1698              (sha256
1699               (base32
1700                "154jwf98cyy54hldr94pgjn85zynly3abpnc1avmb8a18lzwjyb6")))))
1701        ("nsync-src"
1702         ,(let ((version "0559ce013feac8db639ee1bf776aca0325d28777")
1703                (revision "1"))
1704            (origin
1705              (method url-fetch)
1706              (uri (string-append "https://mirror.bazel.build/"
1707                                  "github.com/google/nsync/archive/"
1708                                  version ".tar.gz"))
1709              (file-name (string-append "nsync-0." revision
1710                                        "-" (string-take version 7)
1711                                        ".tar.gz"))
1712              (sha256
1713               (base32
1714                "0qdkyqym34x739mmzv97ah5r7ph462v5xkxqxvidmcfqbi64b132")))))
1715        ("re2-src"
1716         ,(let ((commit "e7efc48")
1717                (revision "1"))
1718            (origin
1719              (method git-fetch)
1720              (uri (git-reference
1721                    (url "https://github.com/google/re2")
1722                    (commit commit)))
1723              (file-name (string-append "re2-0-" revision
1724                                        (string-take commit 7)
1725                                        "-checkout"))
1726              (sha256
1727               (base32
1728                "161g9841rjfsy5pn52fcis0s9hdr7rxvb06pad38j5rppfihvign")))))
1729        ("googletest" ,googletest)
1730        ("swig" ,swig)
1731        ("unzip" ,unzip)))
1732     (propagated-inputs
1733      `(("python-absl-py" ,python-absl-py)
1734        ("python-astor" ,python-astor)
1735        ("python-gast" ,python-gast)
1736        ("python-grpcio" ,python-grpcio)
1737        ("python-numpy" ,python-numpy)
1738        ("python-protobuf" ,python-protobuf-next)
1739        ("python-six" ,python-six)
1740        ("python-termcolo" ,python-termcolor)
1741        ("python-wheel" ,python-wheel)))
1742     (inputs
1743      `(("c-ares" ,c-ares)
1744        ("eigen" ,eigen-for-tensorflow)
1745        ("gemmlowp" ,gemmlowp-for-tensorflow)
1746        ("lmdb" ,lmdb)
1747        ("libjpeg" ,libjpeg)
1748        ("libpng" ,libpng)
1749        ("giflib" ,giflib)
1750        ("grpc" ,grpc)
1751        ("jsoncpp" ,jsoncpp-for-tensorflow)
1752        ("snappy" ,snappy)
1753        ("sqlite" ,sqlite)
1754        ("protobuf" ,protobuf-next)
1755        ("python" ,python-wrapper)
1756        ("zlib" ,zlib)))
1757     (home-page "https://tensorflow.org")
1758     (synopsis "Machine learning framework")
1759     (description
1760      "TensorFlow is a flexible platform for building and training machine
1761 learning models.  It provides a library for high performance numerical
1762 computation and includes high level Python APIs, including both a sequential
1763 API for beginners that allows users to build models quickly by plugging
1764 together building blocks and a subclassing API with an imperative style for
1765 advanced research.")
1766     (license license:asl2.0)))
1767
1768 (define-public python-iml
1769   (package
1770     (name "python-iml")
1771     (version "0.6.2")
1772     (source
1773      (origin
1774        (method url-fetch)
1775        (uri (pypi-uri "iml" version))
1776        (sha256
1777         (base32
1778          "1k8szlpm19rcwcxdny9qdm3gmaqq8akb4xlvrzyz8c2d679aak6l"))))
1779     (build-system python-build-system)
1780     (propagated-inputs
1781      `(("ipython" ,python-ipython)
1782        ("nose" ,python-nose)
1783        ("numpy" ,python-numpy)
1784        ("pandas" ,python-pandas)
1785        ("scipy" ,python-scipy)))
1786     (home-page "http://github.com/interpretable-ml/iml")
1787     (synopsis "Interpretable Machine Learning (iML) package")
1788     (description "Interpretable ML (iML) is a set of data type objects,
1789 visualizations, and interfaces that can be used by any method designed to
1790 explain the predictions of machine learning models (or really the output of
1791 any function).  It currently contains the interface and IO code from the Shap
1792 project, and it will potentially also do the same for the Lime project.")
1793     (license license:expat)))
1794
1795 (define-public python-keras-applications
1796   (package
1797     (name "python-keras-applications")
1798     (version "1.0.8")
1799     (source
1800      (origin
1801        (method url-fetch)
1802        (uri (pypi-uri "Keras_Applications" version))
1803        (sha256
1804         (base32
1805          "1rcz31ca4axa6kzhjx4lwqxbg4wvlljkj8qj9a7p9sfd5fhzjyam"))))
1806     (build-system python-build-system)
1807     ;; The tests require Keras, but this package is needed to build Keras.
1808     (arguments '(#:tests? #f))
1809     (propagated-inputs
1810      `(("python-h5py" ,python-h5py)
1811        ("python-numpy" ,python-numpy)))
1812     (native-inputs
1813      `(("python-pytest" ,python-pytest)
1814        ("python-pytest-cov" ,python-pytest-cov)
1815        ("python-pytest-pep8" ,python-pytest-pep8)
1816        ("python-pytest-xdist" ,python-pytest-xdist)))
1817     (home-page "https://github.com/keras-team/keras-applications")
1818     (synopsis "Reference implementations of popular deep learning models")
1819     (description
1820      "This package provides reference implementations of popular deep learning
1821 models for use with the Keras deep learning framework.")
1822     (license license:expat)))
1823
1824 (define-public python-keras-preprocessing
1825   (package
1826     (name "python-keras-preprocessing")
1827     (version "1.1.0")
1828     (source
1829      (origin
1830        (method url-fetch)
1831        (uri (pypi-uri "Keras_Preprocessing" version))
1832        (sha256
1833         (base32
1834          "1r98nm4k1svsqjyaqkfk23i31bl1kcfcyp7094yyj3c43phfp3as"))))
1835     (build-system python-build-system)
1836     (propagated-inputs
1837      `(("python-numpy" ,python-numpy)
1838        ("python-six" ,python-six)))
1839     (native-inputs
1840      `(("python-pandas" ,python-pandas)
1841        ("python-pillow" ,python-pillow)
1842        ("python-pytest" ,python-pytest)
1843        ("python-pytest-cov" ,python-pytest-cov)
1844        ("python-pytest-xdist" ,python-pytest-xdist)
1845        ("tensorflow" ,tensorflow)))
1846     (home-page "https://github.com/keras-team/keras-preprocessing/")
1847     (synopsis "Data preprocessing and augmentation for deep learning models")
1848     (description
1849      "Keras Preprocessing is the data preprocessing and data augmentation
1850 module of the Keras deep learning library.  It provides utilities for working
1851 with image data, text data, and sequence data.")
1852     (license license:expat)))
1853
1854 (define-public python-keras
1855   (package
1856     (name "python-keras")
1857     (version "2.2.4")
1858     (source
1859      (origin
1860        (method url-fetch)
1861        (uri (pypi-uri "Keras" version))
1862        (patches (search-patches "python-keras-integration-test.patch"))
1863        (sha256
1864         (base32
1865          "1j8bsqzh49vjdxy6l1k4iwax5vpjzniynyd041xjavdzvfii1dlh"))))
1866     (build-system python-build-system)
1867     (arguments
1868      `(#:phases
1869        (modify-phases %standard-phases
1870          (add-after 'unpack 'remove-tests-for-unavailable-features
1871            (lambda _
1872              (delete-file "keras/backend/theano_backend.py")
1873              (delete-file "keras/backend/cntk_backend.py")
1874              (delete-file "tests/keras/backend/backend_test.py")
1875
1876              ;; FIXME: This doesn't work because Tensorflow is missing the
1877              ;; coder ops library.
1878              (delete-file "tests/keras/test_callbacks.py")
1879              #t))
1880          (replace 'check
1881            (lambda _
1882              ;; These tests attempt to download data files from the internet.
1883              (delete-file "tests/integration_tests/test_datasets.py")
1884              (delete-file "tests/integration_tests/imagenet_utils_test.py")
1885
1886              (setenv "PYTHONPATH"
1887                      (string-append (getcwd) "/build/lib:"
1888                                     (getenv "PYTHONPATH")))
1889              (invoke "py.test" "-v"
1890                      "-p" "no:cacheprovider"
1891                      "--ignore" "keras/utils"))))))
1892     (propagated-inputs
1893      `(("python-h5py" ,python-h5py)
1894        ("python-keras-applications" ,python-keras-applications)
1895        ("python-keras-preprocessing" ,python-keras-preprocessing)
1896        ("python-numpy" ,python-numpy)
1897        ("python-pydot" ,python-pydot)
1898        ("python-pyyaml" ,python-pyyaml)
1899        ("python-scipy" ,python-scipy)
1900        ("python-six" ,python-six)
1901        ("tensorflow" ,tensorflow)
1902        ("graphviz" ,graphviz)))
1903     (native-inputs
1904      `(("python-pandas" ,python-pandas)
1905        ("python-pytest" ,python-pytest)
1906        ("python-pytest-cov" ,python-pytest-cov)
1907        ("python-pytest-pep8" ,python-pytest-pep8)
1908        ("python-pytest-timeout" ,python-pytest-timeout)
1909        ("python-pytest-xdist" ,python-pytest-xdist)
1910        ("python-sphinx" ,python-sphinx)
1911        ("python-requests" ,python-requests)))
1912     (home-page "https://github.com/keras-team/keras")
1913     (synopsis "High-level deep learning framework")
1914     (description "Keras is a high-level neural networks API, written in Python
1915 and capable of running on top of TensorFlow.  It was developed with a focus on
1916 enabling fast experimentation.  Use Keras if you need a deep learning library
1917 that:
1918
1919 @itemize
1920 @item Allows for easy and fast prototyping (through user friendliness,
1921   modularity, and extensibility).
1922 @item Supports both convolutional networks and recurrent networks, as well as
1923   combinations of the two.
1924 @item Runs seamlessly on CPU and GPU.
1925 @end itemize\n")
1926     (license license:expat)))
1927
1928 (define-public sbcl-cl-libsvm-format
1929   (let ((commit "3300f84fd8d9f5beafc114f543f9d83417c742fb")
1930         (revision "0"))
1931     (package
1932       (name "sbcl-cl-libsvm-format")
1933       (version (git-version "0.1.0" revision commit))
1934       (source
1935        (origin
1936          (method git-fetch)
1937          (uri (git-reference
1938                (url "https://github.com/masatoi/cl-libsvm-format.git")
1939                (commit commit)))
1940          (file-name (git-file-name name version))
1941          (sha256
1942           (base32
1943            "0284aj84xszhkhlivaigf9qj855fxad3mzmv3zfr0qzb5k0nzwrg"))))
1944       (build-system asdf-build-system/sbcl)
1945       (native-inputs
1946        `(("prove" ,sbcl-prove)
1947          ("prove-asdf" ,sbcl-prove-asdf)))
1948       (inputs
1949        `(("alexandria" ,sbcl-alexandria)))
1950       (synopsis "LibSVM data format reader for Common Lisp")
1951       (description
1952        "This Common Lisp library provides a fast reader for data in LibSVM
1953 format.")
1954       (home-page "https://github.com/masatoi/cl-libsvm-format")
1955       (license license:expat))))
1956
1957 (define-public cl-libsvm-format
1958   (sbcl-package->cl-source-package sbcl-cl-libsvm-format))
1959
1960 (define-public ecl-cl-libsvm-format
1961   (sbcl-package->ecl-package sbcl-cl-libsvm-format))
1962
1963 (define-public sbcl-cl-online-learning
1964   (let ((commit "fc7a34f4f161cd1c7dd747d2ed8f698947781423")
1965         (revision "0"))
1966     (package
1967       (name "sbcl-cl-online-learning")
1968       (version (git-version "0.5" revision commit))
1969       (source
1970        (origin
1971          (method git-fetch)
1972          (uri (git-reference
1973                (url "https://github.com/masatoi/cl-online-learning.git")
1974                (commit commit)))
1975          (file-name (git-file-name name version))
1976          (sha256
1977           (base32
1978            "14x95rlg80ay5hv645ki57pqvy12v28hz4k1w0f6bsfi2rmpxchq"))))
1979       (build-system asdf-build-system/sbcl)
1980       (native-inputs
1981        `(("prove" ,sbcl-prove)
1982          ("prove-asdf" ,sbcl-prove-asdf)))
1983       (inputs
1984        `(("cl-libsvm-format" ,sbcl-cl-libsvm-format)
1985          ("cl-store" ,sbcl-cl-store)))
1986       (arguments
1987        `(;; FIXME: Tests pass but then the check phase crashes
1988          #:tests? #f))
1989       (synopsis "Online Machine Learning for Common Lisp")
1990       (description
1991        "This library contains a collection of machine learning algorithms for
1992 online linear classification written in Common Lisp.")
1993       (home-page "https://github.com/masatoi/cl-online-learning")
1994       (license license:expat))))
1995
1996 (define-public cl-online-learning
1997   (sbcl-package->cl-source-package sbcl-cl-online-learning))
1998
1999 (define-public ecl-cl-online-learning
2000   (sbcl-package->ecl-package sbcl-cl-online-learning))
2001
2002 (define-public sbcl-cl-random-forest
2003   (let ((commit "85fbdd4596d40e824f70f1b7cf239cf544e49d51")
2004         (revision "0"))
2005     (package
2006       (name "sbcl-cl-random-forest")
2007       (version (git-version "0.1" revision commit))
2008       (source
2009        (origin
2010          (method git-fetch)
2011          (uri (git-reference
2012                (url "https://github.com/masatoi/cl-random-forest.git")
2013                (commit commit)))
2014          (file-name (git-file-name name version))
2015          (sha256
2016           (base32
2017            "097xv60i1ndz68sg9p4pc7c5gvyp9i1xgw966b4wwfq3x6hbz421"))))
2018       (build-system asdf-build-system/sbcl)
2019       (native-inputs
2020        `(("prove" ,sbcl-prove)
2021          ("prove-asdf" ,sbcl-prove-asdf)
2022          ("trivial-garbage" ,sbcl-trivial-garbage)))
2023       (inputs
2024        `(("alexandria" ,sbcl-alexandria)
2025          ("cl-libsvm-format" ,sbcl-cl-libsvm-format)
2026          ("cl-online-learning" ,sbcl-cl-online-learning)
2027          ("lparallel" ,sbcl-lparallel)))
2028       (arguments
2029        `(;; The tests download data from the Internet
2030          #:tests? #f
2031          #:phases
2032          (modify-phases %standard-phases
2033            (add-after 'unpack 'add-sb-cltl2-dependency
2034              (lambda _
2035                ;; sb-cltl2 is required by lparallel when using sbcl, but it is
2036                ;; not loaded automatically.
2037                (substitute* "cl-random-forest.asd"
2038                  (("\\(in-package :cl-user\\)")
2039                   "(in-package :cl-user) #+sbcl (require :sb-cltl2)"))
2040                #t)))))
2041       (synopsis "Random Forest and Global Refinement for Common Lisp")
2042       (description
2043        "CL-random-forest is an implementation of Random Forest for multiclass
2044 classification and univariate regression written in Common Lisp.  It also
2045 includes an implementation of Global Refinement of Random Forest.")
2046       (home-page "https://github.com/masatoi/cl-random-forest")
2047       (license license:expat))))
2048
2049 (define-public cl-random-forest
2050   (sbcl-package->cl-source-package sbcl-cl-random-forest))
2051
2052 (define-public ecl-cl-random-forest
2053   (sbcl-package->ecl-package sbcl-cl-random-forest))