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-science)
  73   #:use-module (gnu packages python-web)
  74   #:use-module (gnu packages python-xyz)
  75   #:use-module (gnu packages serialization)
  76   #:use-module (gnu packages sphinx)
  77   #:use-module (gnu packages statistics)
  78   #:use-module (gnu packages sqlite)
  79   #:use-module (gnu packages swig)
  80   #:use-module (gnu packages tls)
  81   #:use-module (gnu packages web)
  82   #:use-module (gnu packages xml)
  83   #:use-module (gnu packages xorg)
  84   #:use-module (ice-9 match))
  85
  86 (define-public fann
  87   ;; The last release is >100 commits behind, so we package from git.
  88   (let ((commit "d71d54788bee56ba4cf7522801270152da5209d7"))
  89     (package
  90       (name "fann")
  91       (version (string-append "2.2.0-1." (string-take commit 8)))
  92       (source (origin
  93                 (method git-fetch)
  94                 (uri (git-reference
  95                       (url "https://github.com/libfann/fann.git")
  96                       (commit commit)))
  97                 (file-name (string-append name "-" version "-checkout"))
  98                 (sha256
  99                  (base32
 100                   "0ibwpfrjs6q2lijs8slxjgzb2llcl6rk3v2ski4r6215g5jjhg3x"))))
 101       (build-system cmake-build-system)
 102       (arguments
 103        `(#:phases
 104          (modify-phases %standard-phases
 105            (replace 'check
 106              (lambda* (#:key outputs #:allow-other-keys)
 107                (let* ((out (assoc-ref outputs "out")))
 108                  (with-directory-excursion (string-append (getcwd) "/tests")
 109                    (invoke "./fann_tests"))))))))
 110       (home-page "http://leenissen.dk/fann/wp/")
 111       (synopsis "Fast Artificial Neural Network")
 112       (description
 113        "FANN is a neural network library, which implements multilayer
 114 artificial neural networks in C with support for both fully connected and
 115 sparsely connected networks.")
 116       (license license:lgpl2.1))))
 117
 118 (define-public libsvm
 119   (package
 120     (name "libsvm")
 121     (version "3.23")
 122     (source
 123      (origin
 124        (method url-fetch)
 125        (uri (string-append "https://www.csie.ntu.edu.tw/~cjlin/libsvm/"
 126                            name "-" version ".tar.gz"))
 127        (sha256
 128         (base32 "0jpaq0rr92x38p4nk3gjan79ip67m6p80anb28z1d8601miysyi5"))))
 129     (build-system gnu-build-system)
 130     (arguments
 131      `(#:tests? #f                      ; no "check" target
 132        #:phases (modify-phases %standard-phases
 133                   (delete 'configure)
 134                   (replace
 135                    'install             ; no ‘install’ target
 136                    (lambda* (#:key outputs #:allow-other-keys)
 137                      (let* ((out (assoc-ref outputs "out"))
 138                             (bin (string-append out "/bin/")))
 139                        (mkdir-p bin)
 140                        (for-each (lambda (file)
 141                                    (copy-file file (string-append bin file)))
 142                                  '("svm-train"
 143                                    "svm-predict"
 144                                    "svm-scale")))
 145                      #t)))))
 146     (home-page "http://www.csie.ntu.edu.tw/~cjlin/libsvm/")
 147     (synopsis "Library for Support Vector Machines")
 148     (description
 149      "LIBSVM is a machine learning library for support vector
 150 classification, (C-SVC, nu-SVC), regression (epsilon-SVR, nu-SVR) and
 151 distribution estimation (one-class SVM).  It supports multi-class
 152 classification.")
 153     (license license:bsd-3)))
 154
 155 (define-public python-libsvm
 156   (package (inherit libsvm)
 157     (name "python-libsvm")
 158     (build-system gnu-build-system)
 159     (arguments
 160      `(#:tests? #f                      ; no "check" target
 161        #:make-flags '("-C" "python")
 162        #:phases
 163        (modify-phases %standard-phases
 164          (delete 'configure)
 165          (replace
 166           'install                      ; no ‘install’ target
 167           (lambda* (#:key inputs outputs #:allow-other-keys)
 168             (let ((site (string-append (assoc-ref outputs "out")
 169                                        "/lib/python"
 170                                        (string-take
 171                                         (string-take-right
 172                                          (assoc-ref inputs "python") 5) 3)
 173                                        "/site-packages/")))
 174               (substitute* "python/svm.py"
 175                 (("../libsvm.so.2") "libsvm.so.2"))
 176               (mkdir-p site)
 177               (for-each (lambda (file)
 178                           (copy-file file (string-append site (basename file))))
 179                         (find-files "python" "\\.py"))
 180               (copy-file "libsvm.so.2"
 181                          (string-append site "libsvm.so.2")))
 182             #t)))))
 183     (inputs
 184      `(("python" ,python)))
 185     (synopsis "Python bindings of libSVM")))
 186
 187 (define-public ghmm
 188   ;; The latest release candidate is several years and a couple of fixes have
 189   ;; been published since.  This is why we download the sources from the SVN
 190   ;; repository.
 191   (let ((svn-revision 2341))
 192     (package
 193       (name "ghmm")
 194       (version (string-append "0.9-rc3-0." (number->string svn-revision)))
 195       (source (origin
 196                 (method svn-fetch)
 197                 (uri (svn-reference
 198                       (url "http://svn.code.sf.net/p/ghmm/code/trunk")
 199                       (revision svn-revision)))
 200                 (file-name (string-append name "-" version "-checkout"))
 201                 (sha256
 202                  (base32
 203                   "0qbq1rqp94l530f043qzp8aw5lj7dng9wq0miffd7spd1ff638wq"))))
 204       (build-system gnu-build-system)
 205       (arguments
 206        `(#:imported-modules (,@%gnu-build-system-modules
 207                              (guix build python-build-system))
 208          #:phases
 209          (modify-phases %standard-phases
 210            (add-after 'unpack 'enter-dir
 211              (lambda _ (chdir "ghmm") #t))
 212            (delete 'check)
 213            (add-after 'install 'check
 214              (assoc-ref %standard-phases 'check))
 215            (add-before 'check 'fix-PYTHONPATH
 216              (lambda* (#:key inputs outputs #:allow-other-keys)
 217                (let ((python-version (python-version
 218                                       (assoc-ref inputs "python"))))
 219                  (setenv "PYTHONPATH"
 220                          (string-append (getenv "PYTHONPATH")
 221                                         ":" (assoc-ref outputs "out")
 222                                         "/lib/python" python-version
 223                                         "/site-packages")))
 224                #t))
 225            (add-after 'enter-dir 'fix-runpath
 226              (lambda* (#:key outputs #:allow-other-keys)
 227                (substitute* "ghmmwrapper/setup.py"
 228                  (("^(.*)extra_compile_args = \\[" line indent)
 229                   (string-append indent
 230                                  "extra_link_args = [\"-Wl,-rpath="
 231                                  (assoc-ref outputs "out") "/lib\"],\n"
 232                                  line
 233                                  "\"-Wl,-rpath="
 234                                  (assoc-ref outputs "out")
 235                                  "/lib\", ")))
 236                #t))
 237            (add-after 'enter-dir 'disable-broken-tests
 238              (lambda _
 239                (substitute* "tests/Makefile.am"
 240                  ;; GHMM_SILENT_TESTS is assumed to be a command.
 241                  (("TESTS_ENVIRONMENT.*") "")
 242                  ;; Do not build broken tests.
 243                  (("chmm .*") "")
 244                  (("read_fa .*") "")
 245                  (("mcmc .*") "")
 246                  (("label_higher_order_test.*$")
 247                   "label_higher_order_test\n"))
 248
 249                ;; These Python unittests are broken as there is no gato.
 250                ;; See https://sourceforge.net/p/ghmm/support-requests/3/
 251                (substitute* "ghmmwrapper/ghmmunittests.py"
 252                  (("^(.*)def (testNewXML|testMultipleTransitionClasses|testNewXML)"
 253                    line indent)
 254                   (string-append indent
 255                                  "@unittest.skip(\"Disabled by Guix\")\n"
 256                                  line)))
 257                #t)))))
 258       (inputs
 259        `(("python" ,python-2) ; only Python 2 is supported
 260          ("libxml2" ,libxml2)))
 261       (native-inputs
 262        `(("pkg-config" ,pkg-config)
 263          ("dejagnu" ,dejagnu)
 264          ("swig" ,swig)
 265          ("autoconf" ,autoconf)
 266          ("automake" ,automake)
 267          ("libtool" ,libtool)))
 268       (home-page "http://ghmm.org")
 269       (synopsis "Hidden Markov Model library")
 270       (description
 271        "The General Hidden Markov Model library (GHMM) is a C library with
 272 additional Python bindings implementing a wide range of types of @dfn{Hidden
 273 Markov Models} (HMM) and algorithms: discrete, continuous emissions, basic
 274 training, HMM clustering, HMM mixtures.")
 275       (license license:lgpl2.0+))))
 276
 277 (define-public mcl
 278   (package
 279     (name "mcl")
 280     (version "14.137")
 281     (source (origin
 282               (method url-fetch)
 283               (uri (string-append
 284                     "http://micans.org/mcl/src/mcl-"
 285                     (string-replace-substring version "." "-")
 286                     ".tar.gz"))
 287               (sha256
 288                (base32
 289                 "15xlax3z31lsn62vlg94hkm75nm40q4679amnfg13jm8m2bnhy5m"))))
 290     (build-system gnu-build-system)
 291     (arguments
 292      `(#:configure-flags (list "--enable-blast")))
 293     (inputs
 294      `(("perl" ,perl)))
 295     (home-page "http://micans.org/mcl/")
 296     (synopsis "Clustering algorithm for graphs")
 297     (description
 298      "The MCL algorithm is short for the @dfn{Markov Cluster Algorithm}, a
 299 fast and scalable unsupervised cluster algorithm for graphs (also known as
 300 networks) based on simulation of (stochastic) flow in graphs.")
 301     ;; In the LICENCE file and web page it says "The software is licensed
 302     ;; under the GNU General Public License, version 3.", but in several of
 303     ;; the source code files it suggests GPL3 or later.
 304     ;; http://listserver.ebi.ac.uk/pipermail/mcl-users/2016/000376.html
 305     (license license:gpl3)))
 306
 307 (define-public ocaml-mcl
 308   (package
 309     (name "ocaml-mcl")
 310     (version "12-068oasis4")
 311     (source
 312      (origin
 313        (method git-fetch)
 314        (uri (git-reference
 315              (url "https://github.com/fhcrc/mcl.git")
 316              (commit version)))
 317        (file-name (git-file-name name version))
 318        (sha256
 319         (base32
 320          "0009dc3h2jp3qg5val452wngpqnbfyhbcxylghq0mrjqxx0jdq5p"))))
 321     (build-system ocaml-build-system)
 322     (arguments
 323      `(#:phases
 324        (modify-phases %standard-phases
 325          (add-before 'configure 'patch-paths
 326            (lambda _
 327              (substitute* "configure"
 328                (("/bin/sh") (which "sh")))
 329              (substitute* "setup.ml"
 330                (("LDFLAGS=-fPIC")
 331                 (string-append "LDFLAGS=-fPIC\"; \"SHELL=" (which "sh")))
 332                (("-std=c89") "-std=gnu99")
 333
 334                ;; This is a mutable string, which is no longer supported.  Use
 335                ;; a byte buffer instead.
 336                (("String.make \\(String.length s\\)")
 337                 "Bytes.make (String.length s)")
 338
 339                ;; These two belong together.
 340                (("OASISString.replace_chars")
 341                 "Bytes.to_string (OASISString.replace_chars")
 342                ((" s;")
 343                 " s);"))
 344              (substitute* "myocamlbuild.ml"
 345                (("std=c89") "std=gnu99"))
 346              ;; Since we build with a more recent OCaml, we have to use C99 or
 347              ;; later.  This causes problems with the old C code.
 348              (substitute* "src/impala/matrix.c"
 349                (("restrict") "restrict_"))
 350              #t)))))
 351     (native-inputs
 352      `(("ocamlbuild" ,ocamlbuild)))
 353     (home-page "https://github.com/fhcrc/mcl")
 354     (synopsis "OCaml wrappers around MCL")
 355     (description
 356      "This package provides OCaml bindings for the MCL graph clustering
 357 algorithm.")
 358     (license license:gpl3)))
 359
 360 (define-public randomjungle
 361   (package
 362     (name "randomjungle")
 363     (version "2.1.0")
 364     (source
 365      (origin
 366        (method url-fetch)
 367        (uri (string-append
 368              "https://www.imbs.uni-luebeck.de/fileadmin/files/Software"
 369              "/randomjungle/randomjungle-" version ".tar_.gz"))
 370        (patches (search-patches "randomjungle-disable-static-build.patch"))
 371        (sha256
 372         (base32
 373          "12c8rf30cla71swx2mf4ww9mfd8jbdw5lnxd7dxhyw1ygrvg6y4w"))))
 374     (build-system gnu-build-system)
 375     (arguments
 376      `(#:configure-flags
 377        (list "--disable-static"
 378              (string-append "--with-boost="
 379                             (assoc-ref %build-inputs "boost")))
 380        #:phases
 381        (modify-phases %standard-phases
 382          (add-before
 383           'configure 'set-CXXFLAGS
 384           (lambda _
 385             (setenv "CXXFLAGS" "-fpermissive ")
 386             #t)))))
 387     (inputs
 388      `(("boost" ,boost)
 389        ("gsl" ,gsl)
 390        ("libxml2" ,libxml2)
 391        ("zlib" ,zlib)))
 392     (native-inputs
 393      `(("gfortran" ,gfortran)
 394        ("gfortran:lib" ,gfortran "lib")))
 395     ;; Non-portable assembly instructions are used so building fails on
 396     ;; platforms other than x86_64 or i686.
 397     (supported-systems '("x86_64-linux" "i686-linux"))
 398     (home-page "https://www.imbs.uni-luebeck.de/forschung/software/details.html#c224")
 399     (synopsis "Implementation of the Random Forests machine learning method")
 400     (description
 401      "Random Jungle is an implementation of Random Forests.  It is supposed to
 402 analyse high dimensional data.  In genetics, it can be used for analysing big
 403 Genome Wide Association (GWA) data.  Random Forests is a powerful machine
 404 learning method.  Most interesting features are variable selection, missing
 405 value imputation, classifier creation, generalization error estimation and
 406 sample proximities between pairs of cases.")
 407     (license license:gpl3+)))
 408
 409 (define-public openfst
 410   (package
 411     (name "openfst")
 412     (version "1.7.2")
 413     (source (origin
 414               (method url-fetch)
 415               (uri (string-append "http://www.openfst.org/twiki/pub/FST/"
 416                                   "FstDownload/openfst-" version ".tar.gz"))
 417               (sha256
 418                (base32
 419                 "0fqgk8195kz21is09gwzwnrg7fr9526bi9mh4apyskapz27pbhr1"))))
 420     (build-system gnu-build-system)
 421     (home-page "http://www.openfst.org")
 422     (synopsis "Library for weighted finite-state transducers")
 423     (description "OpenFst is a library for constructing, combining,
 424 optimizing, and searching weighted finite-state transducers (FSTs).")
 425     (license license:asl2.0)))
 426
 427 (define-public shogun
 428   (package
 429     (name "shogun")
 430     (version "6.1.3")
 431     (source
 432      (origin
 433        (method url-fetch)
 434        (uri (string-append
 435              "ftp://shogun-toolbox.org/shogun/releases/"
 436              (version-major+minor version)
 437              "/sources/shogun-" version ".tar.bz2"))
 438        (sha256
 439         (base32
 440          "1rn9skm3nw6hr7mr3lgp2gfqhi7ii0lyxck7qmqnf8avq349s5jp"))
 441        (modules '((guix build utils)
 442                   (ice-9 rdelim)))
 443        (snippet
 444         '(begin
 445            ;; Remove non-free sources and files referencing them
 446            (for-each delete-file
 447                      (find-files "src/shogun/classifier/svm/"
 448                                  "SVMLight\\.(cpp|h)"))
 449            (for-each delete-file
 450                      (find-files "examples/undocumented/libshogun/"
 451                                  (string-append
 452                                   "(classifier_.*svmlight.*|"
 453                                   "evaluation_cross_validation_locked_comparison).cpp")))
 454            ;; Remove non-free functions.
 455            (define (delete-ifdefs file)
 456              (with-atomic-file-replacement file
 457                (lambda (in out)
 458                  (let loop ((line (read-line in 'concat))
 459                             (skipping? #f))
 460                    (if (eof-object? line)
 461                        #t
 462                        (let ((skip-next?
 463                               (or (and skipping?
 464                                        (not (string-prefix?
 465                                              "#endif //USE_SVMLIGHT" line)))
 466                                   (string-prefix?
 467                                    "#ifdef USE_SVMLIGHT" line))))
 468                          (when (or (not skipping?)
 469                                    (and skipping? (not skip-next?)))
 470                            (display line out))
 471                          (loop (read-line in 'concat) skip-next?)))))))
 472            (for-each delete-ifdefs
 473                      (append
 474                       (find-files "src/shogun/classifier/mkl"
 475                                   "^MKLClassification\\.cpp")
 476                       (find-files "src/shogun/classifier/svm"
 477                                   "^SVMLightOneClass\\.(cpp|h)")
 478                       (find-files "src/shogun/multiclass"
 479                                   "^ScatterSVM\\.(cpp|h)")
 480                       (find-files "src/shogun/kernel/"
 481                                   "^(Kernel|CombinedKernel|ProductKernel)\\.(cpp|h)")
 482                       (find-files "src/shogun/regression/svr"
 483                                   "^(MKLRegression|SVRLight)\\.(cpp|h)")
 484                       (find-files "src/shogun/transfer/domain_adaptation"
 485                                   "^DomainAdaptationSVM\\.(cpp|h)")))
 486            #t))))
 487     (build-system cmake-build-system)
 488     (arguments
 489      '(#:tests? #f ;no check target
 490        #:phases
 491        (modify-phases %standard-phases
 492          (add-after 'unpack 'delete-broken-symlinks
 493            (lambda _
 494              (for-each delete-file '("applications/arts/data"
 495                                      "applications/asp/data"
 496                                      "applications/easysvm/data"
 497                                      "applications/msplicer/data"
 498                                      "applications/ocr/data"
 499                                      "examples/meta/data"
 500                                      "examples/undocumented/data"))
 501              #t))
 502          (add-after 'unpack 'change-R-target-path
 503            (lambda* (#:key outputs #:allow-other-keys)
 504              (substitute* '("src/interfaces/r/CMakeLists.txt"
 505                             "examples/meta/r/CMakeLists.txt")
 506                (("\\$\\{R_COMPONENT_LIB_PATH\\}")
 507                 (string-append (assoc-ref outputs "out")
 508                                "/lib/R/library/")))
 509              #t))
 510          (add-after 'unpack 'fix-octave-modules
 511            (lambda* (#:key outputs #:allow-other-keys)
 512              (substitute* "src/interfaces/octave/CMakeLists.txt"
 513                (("^include_directories\\(\\$\\{OCTAVE_INCLUDE_DIRS\\}")
 514                 "include_directories(${OCTAVE_INCLUDE_DIRS} ${OCTAVE_INCLUDE_DIRS}/octave")
 515                ;; change target directory
 516                (("\\$\\{OCTAVE_OCT_LOCAL_API_FILE_DIR\\}")
 517                 (string-append (assoc-ref outputs "out")
 518                                "/share/octave/packages")))
 519              (substitute* '("src/interfaces/octave/swig_typemaps.i"
 520                             "src/interfaces/octave/sg_print_functions.cpp")
 521                ;; "octave/config.h" and "octave/oct-obj.h" deprecated in Octave.
 522                (("octave/config\\.h") "octave/octave-config.h")
 523                (("octave/oct-obj.h") "octave/ovl.h"))
 524              #t))
 525          (add-after 'unpack 'move-rxcpp
 526            (lambda* (#:key inputs #:allow-other-keys)
 527              (let ((rxcpp-dir "shogun/third-party/rxcpp"))
 528                (mkdir-p rxcpp-dir)
 529                (install-file (assoc-ref inputs "rxcpp") rxcpp-dir)
 530                #t)))
 531          (add-before 'build 'set-HOME
 532            ;; $HOME needs to be set at some point during the build phase
 533            (lambda _ (setenv "HOME" "/tmp") #t)))
 534        #:configure-flags
 535        (list "-DCMAKE_BUILD_WITH_INSTALL_RPATH=TRUE"
 536              "-DUSE_SVMLIGHT=OFF" ;disable proprietary SVMLIGHT
 537              "-DBUILD_META_EXAMPLES=OFF" ;requires unpackaged ctags
 538              ;;"-DINTERFACE_JAVA=ON" ;requires unpackaged jblas
 539              ;;"-DINTERFACE_RUBY=ON" ;requires unpackaged ruby-narray
 540              ;;"-DINTERFACE_PERL=ON" ;"FindPerlLibs" does not exist
 541              ;;"-DINTERFACE_LUA=ON"  ;fails because lua doesn't build pkgconfig file
 542              "-DINTERFACE_OCTAVE=ON"
 543              "-DINTERFACE_PYTHON=ON"
 544              "-DINTERFACE_R=ON")))
 545     (inputs
 546      `(("python" ,python)
 547        ("numpy" ,python-numpy)
 548        ("r-minimal" ,r-minimal)
 549        ("octave" ,octave-cli)
 550        ("swig" ,swig)
 551        ("eigen" ,eigen)
 552        ("hdf5" ,hdf5)
 553        ("atlas" ,atlas)
 554        ("arpack" ,arpack-ng)
 555        ("lapack" ,lapack)
 556        ("glpk" ,glpk)
 557        ("libxml2" ,libxml2)
 558        ("lzo" ,lzo)
 559        ("zlib" ,zlib)))
 560     (native-inputs
 561      `(("pkg-config" ,pkg-config)
 562        ("rxcpp" ,rxcpp)))
 563     ;; Non-portable SSE instructions are used so building fails on platforms
 564     ;; other than x86_64.
 565     (supported-systems '("x86_64-linux"))
 566     (home-page "http://shogun-toolbox.org/")
 567     (synopsis "Machine learning toolbox")
 568     (description
 569      "The Shogun Machine learning toolbox provides a wide range of unified and
 570 efficient Machine Learning (ML) methods.  The toolbox seamlessly allows to
 571 combine multiple data representations, algorithm classes, and general purpose
 572 tools.  This enables both rapid prototyping of data pipelines and extensibility
 573 in terms of new algorithms.")
 574     (license license:gpl3+)))
 575
 576 (define-public rxcpp
 577   (package
 578     (name "rxcpp")
 579     (version "4.1.0")
 580     (source
 581      (origin
 582        (method git-fetch)
 583        (uri (git-reference
 584              (url "https://github.com/ReactiveX/RxCpp.git")
 585              (commit (string-append "v" version))))
 586        (sha256
 587         (base32 "1rdpa3jlc181jd08nk437aar085h28i45s6nzrv65apb3xyyz0ij"))
 588        (file-name (git-file-name name version))))
 589     (build-system cmake-build-system)
 590     (arguments
 591      `(#:phases
 592        (modify-phases %standard-phases
 593          (add-after 'unpack 'remove-werror
 594            (lambda _
 595              (substitute* (find-files ".")
 596                (("-Werror") ""))
 597              #t))
 598          (replace 'check
 599            (lambda _
 600              (invoke "ctest"))))))
 601     (native-inputs
 602      `(("catch" ,catch-framework)))
 603     (home-page "http://reactivex.io/")
 604     (synopsis "Reactive Extensions for C++")
 605     (description
 606      "The Reactive Extensions for C++ (RxCpp) is a library of algorithms for
 607 values-distributed-in-time.  ReactiveX is a library for composing asynchronous
 608 and event-based programs by using observable sequences.
 609
 610 It extends the observer pattern to support sequences of data and/or events and
 611 adds operators that allow you to compose sequences together declaratively while
 612 abstracting away concerns about things like low-level threading,
 613 synchronization, thread-safety, concurrent data structures, and non-blocking
 614 I/O.")
 615     (license license:asl2.0)))
 616
 617 (define-public r-adaptivesparsity
 618   (package
 619     (name "r-adaptivesparsity")
 620     (version "1.6")
 621     (source (origin
 622               (method url-fetch)
 623               (uri (cran-uri "AdaptiveSparsity" version))
 624               (sha256
 625                (base32
 626                 "0imr5m8mll9j6n4icsv6z9rl5kbnwsp9wvzrg7n90nnmcxq2cz91"))))
 627     (properties
 628      `((upstream-name . "AdaptiveSparsity")))
 629     (build-system r-build-system)
 630     (arguments
 631      `(#:phases
 632        (modify-phases %standard-phases
 633          (add-after 'unpack 'link-against-armadillo
 634            (lambda _
 635              (substitute* "src/Makevars"
 636                (("PKG_LIBS=" prefix)
 637                 (string-append prefix "-larmadillo"))))))))
 638     (propagated-inputs
 639      `(("r-mass" ,r-mass)
 640        ("r-matrix" ,r-matrix)
 641        ("r-rcpp" ,r-rcpp)
 642        ("r-rcpparmadillo" ,r-rcpparmadillo)))
 643     (inputs
 644      `(("armadillo" ,armadillo)))
 645     (home-page "https://cran.r-project.org/web/packages/AdaptiveSparsity")
 646     (synopsis "Adaptive sparsity models")
 647     (description
 648      "This package implements the Figueiredo machine learning algorithm for
 649 adaptive sparsity and the Wong algorithm for adaptively sparse gaussian
 650 geometric models.")
 651     (license license:lgpl3+)))
 652
 653 (define-public gemmlowp-for-tensorflow
 654   ;; The commit hash is taken from "tensorflow/workspace.bzl".
 655   (let ((commit "38ebac7b059e84692f53e5938f97a9943c120d98")
 656         (revision "2"))
 657     (package
 658       (name "gemmlowp")
 659       (version (git-version "0" revision commit))
 660       (source (origin
 661                 (method url-fetch)
 662                 (uri (string-append "https://mirror.bazel.build/"
 663                                     "github.com/google/gemmlowp/archive/"
 664                                     commit ".zip"))
 665                 (file-name (string-append "gemmlowp-" version ".zip"))
 666                 (sha256
 667                  (base32
 668                   "0n56s2g8hrssm4w8qj1v58gfm56a04n9v992ixkmvk6zjiralzxq"))))
 669       (build-system cmake-build-system)
 670       (arguments
 671        `(#:configure-flags
 672          (list ,@(match (%current-system)
 673                    ((or "x86_64-linux" "i686-linux")
 674                     '("-DCMAKE_CXX_FLAGS=-msse2"))
 675                    (_ '())))
 676          #:phases
 677          (modify-phases %standard-phases
 678            ;; This directory contains the CMakeLists.txt.
 679            (add-after 'unpack 'chdir
 680              (lambda _ (chdir "contrib") #t))
 681            ;; There is no install target
 682            (replace 'install
 683              (lambda* (#:key outputs #:allow-other-keys)
 684                (let* ((out (assoc-ref outputs "out"))
 685                       (lib (string-append out "/lib/"))
 686                       (inc (string-append out "/include/")))
 687                  (install-file "../build/libeight_bit_int_gemm.so" lib)
 688                  (for-each (lambda (dir)
 689                              (let ((target (string-append inc "/" dir)))
 690                                (mkdir-p target)
 691                                (for-each (lambda (h)
 692                                            (install-file h target))
 693                                          (find-files (string-append "../" dir)
 694                                                      "\\.h$"))))
 695                            '("meta" "profiling" "public" "fixedpoint"
 696                              "eight_bit_int_gemm" "internal"))
 697                  #t))))))
 698       (native-inputs
 699        `(("unzip" ,unzip)))
 700       (home-page "https://github.com/google/gemmlowp")
 701       (synopsis "Small self-contained low-precision GEMM library")
 702       (description
 703        "This is a small self-contained low-precision @dfn{general matrix
 704 multiplication} (GEMM) library.  It is not a full linear algebra library.
 705 Low-precision means that the input and output matrix entries are integers on
 706 at most 8 bits.  To avoid overflow, results are internally accumulated on more
 707 than 8 bits, and at the end only some significant 8 bits are kept.")
 708       (license license:asl2.0))))
 709
 710 (define-public dlib
 711   (package
 712     (name "dlib")
 713     (version "19.7")
 714     (source (origin
 715               (method url-fetch)
 716               (uri (string-append
 717                     "http://dlib.net/files/dlib-" version ".tar.bz2"))
 718               (sha256
 719                (base32
 720                 "1mljz02kwkrbggyncxv5fpnyjdybw2qihaacb3js8yfkw12vwpc2"))
 721               (modules '((guix build utils)))
 722               (snippet
 723                '(begin
 724                   ;; Delete ~13MB of bundled dependencies.
 725                   (delete-file-recursively "dlib/external")
 726                   (delete-file-recursively "docs/dlib/external")
 727                   #t))))
 728     (build-system cmake-build-system)
 729     (arguments
 730      `(#:phases
 731        (modify-phases %standard-phases
 732          (add-after 'unpack 'disable-asserts
 733            (lambda _
 734              ;; config.h recommends explicitly enabling or disabling asserts
 735              ;; when building as a shared library. By default neither is set.
 736              (substitute* "dlib/config.h"
 737                (("^//#define DLIB_DISABLE_ASSERTS") "#define DLIB_DISABLE_ASSERTS"))
 738              #t))
 739          (add-after 'disable-asserts 'disable-failing-tests
 740            (lambda _
 741              ;; One test times out on MIPS, so we need to disable it.
 742              ;; Others are flaky on some platforms.
 743              (let* ((system ,(or (%current-target-system)
 744                                  (%current-system)))
 745                     (disabled-tests (cond
 746                                      ((string-prefix? "mips64" system)
 747                                       '("object_detector" ; timeout
 748                                         "data_io"))
 749                                      ((string-prefix? "armhf" system)
 750                                       '("learning_to_track"))
 751                                      ((string-prefix? "i686" system)
 752                                       '("optimization"))
 753                                      (else '()))))
 754                (for-each
 755                 (lambda (test)
 756                   (substitute* "dlib/test/makefile"
 757                     (((string-append "SRC \\+= " test "\\.cpp")) "")))
 758                 disabled-tests)
 759                #t)))
 760          (replace 'check
 761            (lambda _
 762              ;; No test target, so we build and run the unit tests here.
 763              (let ((test-dir (string-append "../dlib-" ,version "/dlib/test")))
 764                (with-directory-excursion test-dir
 765                  (invoke "make" "-j" (number->string (parallel-job-count)))
 766                  (invoke "./dtest" "--runall"))
 767                #t)))
 768          (add-after 'install 'delete-static-library
 769            (lambda* (#:key outputs #:allow-other-keys)
 770              (delete-file (string-append (assoc-ref outputs "out")
 771                                          "/lib/libdlib.a"))
 772              #t)))))
 773     (native-inputs
 774      `(("pkg-config" ,pkg-config)
 775        ;; For tests.
 776        ("libnsl" ,libnsl)))
 777     (inputs
 778      `(("giflib" ,giflib)
 779        ("lapack" ,lapack)
 780        ("libjpeg" ,libjpeg)
 781        ("libpng" ,libpng)
 782        ("libx11" ,libx11)
 783        ("openblas" ,openblas)
 784        ("zlib" ,zlib)))
 785     (synopsis
 786      "Toolkit for making machine learning and data analysis applications in C++")
 787     (description
 788      "Dlib is a modern C++ toolkit containing machine learning algorithms and
 789 tools.  It is used in both industry and academia in a wide range of domains
 790 including robotics, embedded devices, mobile phones, and large high performance
 791 computing environments.")
 792     (home-page "http://dlib.net")
 793     (license license:boost1.0)))
 794
 795 (define-public python-scikit-learn
 796   (package
 797     (name "python-scikit-learn")
 798     (version "0.20.4")
 799     (source
 800      (origin
 801        (method git-fetch)
 802        (uri (git-reference
 803              (url "https://github.com/scikit-learn/scikit-learn.git")
 804              (commit version)))
 805        (file-name (git-file-name name version))
 806        (sha256
 807         (base32
 808          "08zbzi8yx5wdlxfx9jap61vg1malc9ajf576w7a0liv6jvvrxlpj"))))
 809     (build-system python-build-system)
 810     (arguments
 811      `(#:phases
 812        (modify-phases %standard-phases
 813          (add-after 'build 'build-ext
 814            (lambda _ (invoke "python" "setup.py" "build_ext" "--inplace") #t))
 815          (replace 'check
 816            (lambda _
 817              ;; Restrict OpenBLAS threads to prevent segfaults while testing!
 818              (setenv "OPENBLAS_NUM_THREADS" "1")
 819
 820              ;; Some tests require write access to $HOME.
 821              (setenv "HOME" "/tmp")
 822
 823              (invoke "pytest" "sklearn" "-m" "not network")))
 824          (add-before 'reset-gzip-timestamps 'make-files-writable
 825            (lambda* (#:key outputs #:allow-other-keys)
 826              ;; Make sure .gz files are writable so that the
 827              ;; 'reset-gzip-timestamps' phase can do its work.
 828              (let ((out (assoc-ref outputs "out")))
 829                (for-each make-file-writable
 830                          (find-files out "\\.gz$"))
 831                #t))))))
 832     (inputs
 833      `(("openblas" ,openblas)))
 834     (native-inputs
 835      `(("python-pytest" ,python-pytest)
 836        ("python-pandas" ,python-pandas) ;for tests
 837        ("python-cython" ,python-cython)))
 838     (propagated-inputs
 839      `(("python-numpy" ,python-numpy)
 840        ("python-scipy" ,python-scipy)))
 841     (home-page "http://scikit-learn.org/")
 842     (synopsis "Machine Learning in Python")
 843     (description
 844      "Scikit-learn provides simple and efficient tools for data mining and
 845 data analysis.")
 846     (license license:bsd-3)))
 847
 848 (define-public python2-scikit-learn
 849   (package-with-python2 python-scikit-learn))
 850
 851 (define-public python-autograd
 852   (let* ((commit "442205dfefe407beffb33550846434baa90c4de7")
 853          (revision "0")
 854          (version (git-version "0.0.0" revision commit)))
 855     (package
 856       (name "python-autograd")
 857       (home-page "https://github.com/HIPS/autograd")
 858       (source (origin
 859                 (method git-fetch)
 860                 (uri (git-reference
 861                       (url home-page)
 862                       (commit commit)))
 863                 (sha256
 864                  (base32
 865                   "189sv2xb0mwnjawa9z7mrgdglc1miaq93pnck26r28fi1jdwg0z4"))
 866                 (file-name (git-file-name name version))))
 867       (version version)
 868       (build-system python-build-system)
 869       (native-inputs
 870        `(("python-nose" ,python-nose)
 871          ("python-pytest" ,python-pytest)))
 872       (propagated-inputs
 873        `(("python-future" ,python-future)
 874          ("python-numpy" ,python-numpy)))
 875       (arguments
 876        `(#:phases (modify-phases %standard-phases
 877                     (replace 'check
 878                       (lambda _
 879                         (invoke "py.test" "-v"))))))
 880       (synopsis "Efficiently computes derivatives of NumPy code")
 881       (description "Autograd can automatically differentiate native Python and
 882 NumPy code.  It can handle a large subset of Python's features, including loops,
 883 ifs, recursion and closures, and it can even take derivatives of derivatives
 884 of derivatives.  It supports reverse-mode differentiation
 885 (a.k.a. backpropagation), which means it can efficiently take gradients of
 886 scalar-valued functions with respect to array-valued arguments, as well as
 887 forward-mode differentiation, and the two can be composed arbitrarily.  The
 888 main intended application of Autograd is gradient-based optimization.")
 889       (license license:expat))))
 890
 891 (define-public python2-autograd
 892   (package-with-python2 python-autograd))
 893
 894 (define-public lightgbm
 895   (package
 896     (name "lightgbm")
 897     (version "2.0.12")
 898     (source (origin
 899               (method url-fetch)
 900               (uri (string-append
 901                     "https://github.com/Microsoft/LightGBM/archive/v"
 902                     version ".tar.gz"))
 903               (sha256
 904                (base32
 905                 "132zf0yk0545mg72hyzxm102g3hpb6ixx9hnf8zd2k55gas6cjj1"))
 906               (file-name (string-append name "-" version ".tar.gz"))))
 907     (native-inputs
 908      `(("python-pytest" ,python-pytest)
 909        ("python-nose" ,python-nose)))
 910     (inputs
 911      `(("openmpi" ,openmpi)))
 912     (propagated-inputs
 913      `(("python-numpy" ,python-numpy)
 914        ("python-scipy" ,python-scipy)))
 915     (arguments
 916      `(#:configure-flags
 917        '("-DUSE_MPI=ON")
 918        #:phases
 919        (modify-phases %standard-phases
 920          (replace 'check
 921            (lambda* (#:key outputs #:allow-other-keys)
 922              (with-directory-excursion ,(string-append "../LightGBM-" version)
 923                (invoke "pytest" "tests/c_api_test/test_.py")))))))
 924     (build-system cmake-build-system)
 925     (home-page "https://github.com/Microsoft/LightGBM")
 926     (synopsis "Gradient boosting framework based on decision tree algorithms")
 927     (description "LightGBM is a gradient boosting framework that uses tree
 928 based learning algorithms.  It is designed to be distributed and efficient with
 929 the following advantages:
 930
 931 @itemize
 932 @item Faster training speed and higher efficiency
 933 @item Lower memory usage
 934 @item Better accuracy
 935 @item Parallel and GPU learning supported (not enabled in this package)
 936 @item Capable of handling large-scale data
 937 @end itemize\n")
 938     (license license:expat)))
 939
 940 (define-public vowpal-wabbit
 941   ;; Language bindings not included.
 942   (package
 943     (name "vowpal-wabbit")
 944     (version "8.5.0")
 945     (source (origin
 946               (method url-fetch)
 947               (uri (string-append
 948                     "https://github.com/JohnLangford/vowpal_wabbit/archive/"
 949                     version ".tar.gz"))
 950               (sha256
 951                (base32
 952                 "0clp2kb7rk5sckhllxjr5a651awf4s8dgzg4659yh4hf5cqnf0gr"))
 953               (file-name (string-append name "-" version ".tar.gz"))))
 954     (inputs
 955      `(("boost" ,boost)
 956        ("zlib" ,zlib)))
 957     (arguments
 958      `(#:configure-flags
 959        (list (string-append "--with-boost="
 960                             (assoc-ref %build-inputs "boost")))))
 961     (build-system gnu-build-system)
 962     (home-page "https://github.com/JohnLangford/vowpal_wabbit")
 963     (synopsis "Fast machine learning library for online learning")
 964     (description "Vowpal Wabbit is a machine learning system with techniques
 965 such as online, hashing, allreduce, reductions, learning2search, active, and
 966 interactive learning.")
 967     (license license:bsd-3)))
 968
 969 (define-public python2-fastlmm
 970   (package
 971     (name "python2-fastlmm")
 972     (version "0.2.21")
 973     (source
 974      (origin
 975        (method url-fetch)
 976        (uri (pypi-uri "fastlmm" version ".zip"))
 977        (sha256
 978         (base32
 979          "1q8c34rpmwkfy3r4d5172pzdkpfryj561897z9r3x22gq7813x1m"))))
 980     (build-system python-build-system)
 981     (arguments
 982      `(#:tests? #f ; some test files are missing
 983        #:python ,python-2)) ; only Python 2.7 is supported
 984     (propagated-inputs
 985      `(("python2-numpy" ,python2-numpy)
 986        ("python2-scipy" ,python2-scipy)
 987        ("python2-matplotlib" ,python2-matplotlib)
 988        ("python2-pandas" ,python2-pandas)
 989        ("python2-scikit-learn" ,python2-scikit-learn)
 990        ("python2-pysnptools" ,python2-pysnptools)))
 991     (native-inputs
 992      `(("unzip" ,unzip)
 993        ("python2-cython" ,python2-cython)
 994        ("python2-mock" ,python2-mock)
 995        ("python2-nose" ,python2-nose)))
 996     (home-page "http://research.microsoft.com/en-us/um/redmond/projects/mscompbio/fastlmm/")
 997     (synopsis "Perform genome-wide association studies on large data sets")
 998     (description
 999      "FaST-LMM, which stands for Factored Spectrally Transformed Linear Mixed
1000 Models, is a program for performing both single-SNP and SNP-set genome-wide
1001 association studies (GWAS) on extremely large data sets.")
1002     (license license:asl2.0)))
1003
1004 ;; There have been no proper releases yet.
1005 (define-public kaldi
1006   (let ((commit "2f95609f0bb085bd3a1dc5eb0a39f3edea59e606")
1007         (revision "1"))
1008     (package
1009       (name "kaldi")
1010       (version (git-version "0" revision commit))
1011       (source (origin
1012                 (method git-fetch)
1013                 (uri (git-reference
1014                       (url "https://github.com/kaldi-asr/kaldi.git")
1015                       (commit commit)))
1016                 (file-name (git-file-name name version))
1017                 (sha256
1018                  (base32
1019                   "082qh3pfi7hvncylp4xsmkfahbd7gb0whdfa4rwrx7fxk9rdh3kz"))))
1020       (build-system gnu-build-system)
1021       (arguments
1022        `(#:test-target "test"
1023          #:phases
1024          (modify-phases %standard-phases
1025            (add-after 'unpack 'chdir
1026              (lambda _ (chdir "src") #t))
1027            (replace 'configure
1028              (lambda* (#:key build system inputs outputs #:allow-other-keys)
1029                (when (not (or (string-prefix? "x86_64" system)
1030                               (string-prefix? "i686" system)))
1031                  (substitute* "makefiles/linux_openblas.mk"
1032                    (("-msse -msse2") "")))
1033                (substitute* "makefiles/default_rules.mk"
1034                  (("/bin/bash") (which "bash")))
1035                (substitute* "Makefile"
1036                  (("ext_depend: check_portaudio")
1037                   "ext_depend:"))
1038                (substitute* '("online/Makefile"
1039                               "onlinebin/Makefile"
1040                               "gst-plugin/Makefile")
1041                  (("../../tools/portaudio/install")
1042                   (assoc-ref inputs "portaudio")))
1043
1044                ;; This `configure' script doesn't support variables passed as
1045                ;; arguments, nor does it support "prefix".
1046                (let ((out (assoc-ref outputs "out"))
1047                      (openblas (assoc-ref inputs "openblas"))
1048                      (openfst (assoc-ref inputs "openfst")))
1049                  (substitute* "configure"
1050                    (("check_for_slow_expf;") "")
1051                    ;; This affects the RPATH and also serves as the installation
1052                    ;; directory.
1053                    (("KALDILIBDIR=`pwd`/lib")
1054                     (string-append "KALDILIBDIR=" out "/lib")))
1055                  (mkdir-p out) ; must exist
1056                  (setenv "CONFIG_SHELL" (which "bash"))
1057                  (setenv "OPENFST_VER" ,(package-version openfst))
1058                  (invoke "./configure"
1059                          "--use-cuda=no"
1060                          "--shared"
1061                          (string-append "--openblas-root=" openblas)
1062                          (string-append "--fst-root=" openfst)))))
1063            (add-after 'build 'build-ext-and-gstreamer-plugin
1064              (lambda _
1065                (invoke "make" "-C" "online" "depend")
1066                (invoke "make" "-C" "online")
1067                (invoke "make" "-C" "onlinebin" "depend")
1068                (invoke "make" "-C" "onlinebin")
1069                (invoke "make" "-C" "gst-plugin" "depend")
1070                (invoke "make" "-C" "gst-plugin")
1071                #t))
1072            ;; TODO: also install the executables.
1073            (replace 'install
1074              (lambda* (#:key outputs #:allow-other-keys)
1075                (let* ((out (assoc-ref outputs "out"))
1076                       (inc (string-append out "/include"))
1077                       (lib (string-append out "/lib")))
1078                  (mkdir-p lib)
1079                  ;; The build phase installed symlinks to the actual
1080                  ;; libraries.  Install the actual targets.
1081                  (for-each (lambda (file)
1082                              (let ((target (readlink file)))
1083                                (delete-file file)
1084                                (install-file target lib)))
1085                            (find-files lib "\\.so"))
1086                  ;; Install headers
1087                  (for-each (lambda (file)
1088                              (let ((target-dir (string-append inc "/" (dirname file))))
1089                                (install-file file target-dir)))
1090                            (find-files "." "\\.h"))
1091                  (install-file "gst-plugin/libgstonlinegmmdecodefaster.so"
1092                                (string-append lib "/gstreamer-1.0"))
1093                  #t))))))
1094       (inputs
1095        `(("alsa-lib" ,alsa-lib)
1096          ("gfortran" ,gfortran "lib")
1097          ("glib" ,glib)
1098          ("gstreamer" ,gstreamer)
1099          ("jack" ,jack-1)
1100          ("openblas" ,openblas)
1101          ("openfst" ,openfst)
1102          ("portaudio" ,portaudio)
1103          ("python" ,python)))
1104       (native-inputs
1105        `(("glib" ,glib "bin")             ; glib-genmarshal
1106          ("grep" ,grep)
1107          ("sed" ,sed)
1108          ("pkg-config" ,pkg-config)
1109          ("which" ,which)))
1110       (home-page "https://kaldi-asr.org/")
1111       (synopsis "Speech recognition toolkit")
1112       (description "Kaldi is an extensible toolkit for speech recognition
1113 written in C++.")
1114       (license license:asl2.0))))
1115
1116 (define-public gst-kaldi-nnet2-online
1117   (let ((commit "617e43e73c7cc45eb9119028c02bd4178f738c4a")
1118         (revision "1"))
1119     (package
1120       (name "gst-kaldi-nnet2-online")
1121       (version (git-version "0" revision commit))
1122       (source (origin
1123                 (method git-fetch)
1124                 (uri (git-reference
1125                       (url "https://github.com/alumae/gst-kaldi-nnet2-online.git")
1126                       (commit commit)))
1127                 (file-name (git-file-name name version))
1128                 (sha256
1129                  (base32
1130                   "0xh3w67b69818s6ib02ara4lw7wamjdmh4jznvkpzrs4skbs9jx9"))))
1131       (build-system gnu-build-system)
1132       (arguments
1133        `(#:tests? #f                    ; there are none
1134          #:make-flags
1135          (list (string-append "SHELL="
1136                               (assoc-ref %build-inputs "bash") "/bin/bash")
1137                (string-append "KALDI_ROOT="
1138                               (assoc-ref %build-inputs "kaldi-src"))
1139                (string-append "KALDILIBDIR="
1140                               (assoc-ref %build-inputs "kaldi") "/lib")
1141                "KALDI_FLAVOR=dynamic")
1142          #:phases
1143          (modify-phases %standard-phases
1144            (add-after 'unpack 'chdir
1145              (lambda _ (chdir "src") #t))
1146            (replace 'configure
1147              (lambda* (#:key inputs #:allow-other-keys)
1148                (let ((glib (assoc-ref inputs "glib")))
1149                  (setenv "CXXFLAGS" "-fPIC")
1150                  (setenv "CPLUS_INCLUDE_PATH"
1151                          (string-append glib "/include/glib-2.0:"
1152                                         glib "/lib/glib-2.0/include:"
1153                                         (assoc-ref inputs "gstreamer")
1154                                         "/include/gstreamer-1.0")))
1155                (substitute* "Makefile"
1156                  (("include \\$\\(KALDI_ROOT\\)/src/kaldi.mk") "")
1157                  (("\\$\\(error Cannot find") "#"))
1158                #t))
1159            (add-before 'build 'build-depend
1160              (lambda* (#:key make-flags #:allow-other-keys)
1161                (apply invoke "make" "depend" make-flags)))
1162            (replace 'install
1163              (lambda* (#:key outputs #:allow-other-keys)
1164                (let* ((out (assoc-ref outputs "out"))
1165                       (lib (string-append out "/lib/gstreamer-1.0")))
1166                  (install-file "libgstkaldinnet2onlinedecoder.so" lib)
1167                  #t))))))
1168       (inputs
1169        `(("glib" ,glib)
1170          ("gstreamer" ,gstreamer)
1171          ("jansson" ,jansson)
1172          ("openfst" ,openfst)
1173          ("kaldi" ,kaldi)))
1174       (native-inputs
1175        `(("bash" ,bash)
1176          ("glib:bin" ,glib "bin")       ; glib-genmarshal
1177          ("kaldi-src" ,(package-source kaldi))
1178          ("pkg-config" ,pkg-config)))
1179       (home-page "https://kaldi-asr.org/")
1180       (synopsis "Gstreamer plugin for decoding speech")
1181       (description "This package provides a GStreamer plugin that wraps
1182 Kaldi's @code{SingleUtteranceNnet2Decoder}.  It requires iVector-adapted DNN
1183 acoustic models.  The iVectors are adapted to the current audio stream
1184 automatically.")
1185       (license license:asl2.0))))
1186
1187 (define-public kaldi-gstreamer-server
1188   (let ((commit "1735ba49c5dc0ebfc184e45105fc600cd9f1f508")
1189         (revision "1"))
1190     (package
1191       (name "kaldi-gstreamer-server")
1192       (version (git-version "0" revision commit))
1193       (source (origin
1194                 (method git-fetch)
1195                 (uri (git-reference
1196                       (url "https://github.com/alumae/kaldi-gstreamer-server.git")
1197                       (commit commit)))
1198                 (file-name (git-file-name name version))
1199                 (sha256
1200                  (base32
1201                   "0j701m7lbwmzqxsfanj882v7881hrbmpqybbczbxqpcbg8q34w0k"))))
1202       (build-system gnu-build-system)
1203       (arguments
1204        `(#:tests? #f ; there are no tests that can be run automatically
1205          #:modules ((guix build utils)
1206                     (guix build gnu-build-system)
1207                     (srfi srfi-26))
1208          #:phases
1209          (modify-phases %standard-phases
1210            (delete 'configure)
1211            (replace 'build
1212              (lambda* (#:key outputs #:allow-other-keys)
1213                ;; Disable hash randomization to ensure the generated .pycs
1214                ;; are reproducible.
1215                (setenv "PYTHONHASHSEED" "0")
1216                (with-directory-excursion "kaldigstserver"
1217                  (for-each (lambda (file)
1218                              (apply invoke
1219                                     `("python"
1220                                       "-m" "compileall"
1221                                       "-f" ; force rebuild
1222                                       ,file)))
1223                            (find-files "." "\\.py$")))
1224                #t))
1225            (replace 'install
1226              (lambda* (#:key inputs outputs #:allow-other-keys)
1227                (let* ((out (assoc-ref outputs "out"))
1228                       (bin (string-append out "/bin"))
1229                       (share (string-append out "/share/kaldi-gstreamer-server/")))
1230                  ;; Install Python files
1231                  (with-directory-excursion "kaldigstserver"
1232                    (for-each (cut install-file <> share)
1233                              (find-files "." ".*")))
1234
1235                  ;; Install sample configuration files
1236                  (for-each (cut install-file <> share)
1237                            (find-files "." "\\.yaml"))
1238
1239                  ;; Install executables
1240                  (mkdir-p bin)
1241                  (let* ((server (string-append bin "/kaldi-gst-server"))
1242                         (client (string-append bin "/kaldi-gst-client"))
1243                         (worker (string-append bin "/kaldi-gst-worker"))
1244                         (PYTHONPATH (getenv "PYTHONPATH"))
1245                         (GST_PLUGIN_PATH (string-append
1246                                           (assoc-ref inputs "gst-kaldi-nnet2-online")
1247                                           "/lib/gstreamer-1.0:${GST_PLUGIN_PATH}"))
1248                         (wrap (lambda (wrapper what)
1249                                 (with-output-to-file wrapper
1250                                   (lambda _
1251                                     (format #t
1252                                             "#!~a
1253 export PYTHONPATH=~a
1254 export GST_PLUGIN_PATH=~a
1255 exec ~a ~a/~a \"$@\"~%"
1256                                             (which "bash") PYTHONPATH GST_PLUGIN_PATH
1257                                             (which "python") share what)))
1258                                 (chmod wrapper #o555))))
1259                    (for-each wrap
1260                              (list server client worker)
1261                              (list "master_server.py"
1262                                    "client.py"
1263                                    "worker.py")))
1264                  #t))))))
1265       (inputs
1266        `(("gst-kaldi-nnet2-online" ,gst-kaldi-nnet2-online)
1267          ("python2" ,python-2)
1268          ("python2-futures" ,python2-futures)
1269          ("python2-pygobject" ,python2-pygobject)
1270          ("python2-pyyaml" ,python2-pyyaml)
1271          ("python2-tornado" ,python2-tornado)
1272          ("python2-ws4py" ,python2-ws4py-for-kaldi-gstreamer-server)))
1273       (home-page "https://github.com/alumae/kaldi-gstreamer-server")
1274       (synopsis "Real-time full-duplex speech recognition server")
1275       (description "This is a real-time full-duplex speech recognition server,
1276 based on the Kaldi toolkit and the GStreamer framework and implemented in
1277 Python.")
1278       (license license:bsd-2))))
1279
1280 (define-public grpc
1281   (package
1282     (name "grpc")
1283     (version "1.16.1")
1284     (source (origin
1285               (method git-fetch)
1286               (uri (git-reference
1287                     (url "https://github.com/grpc/grpc.git")
1288                     (commit (string-append "v" version))))
1289               (file-name (git-file-name name version))
1290               (sha256
1291                (base32
1292                 "1jimqz3115f9pli5w6ik9wi7mjc7ix6y7yrq4a1ab9fc3dalj7p2"))))
1293     (build-system cmake-build-system)
1294     (arguments
1295      `(#:tests? #f ; no test target
1296        #:configure-flags
1297        (list "-DgRPC_ZLIB_PROVIDER=package"
1298              "-DgRPC_CARES_PROVIDER=package"
1299              "-DgRPC_SSL_PROVIDER=package"
1300              "-DgRPC_PROTOBUF_PROVIDER=package")))
1301     (inputs
1302      `(("c-ares" ,c-ares/cmake)
1303        ("openssl" ,openssl)
1304        ("zlib" ,zlib)))
1305     (native-inputs
1306      `(("protobuf" ,protobuf)
1307        ("python" ,python-wrapper)))
1308     (home-page "https://grpc.io")
1309     (synopsis "High performance universal RPC framework")
1310     (description "gRPC is a modern high performance @dfn{Remote Procedure Call}
1311 (RPC) framework that can run in any environment.  It can efficiently connect
1312 services in and across data centers with pluggable support for load balancing,
1313 tracing, health checking and authentication.  It is also applicable in last
1314 mile of distributed computing to connect devices, mobile applications and
1315 browsers to backend services.")
1316     (license license:asl2.0)))
1317
1318 ;; Note that Tensorflow includes a "third_party" directory, which seems to not
1319 ;; only contain modified subsets of upstream library source code, but also
1320 ;; adapter headers provided by Google (such as the fft.h header, which is not
1321 ;; part of the upstream project code).  The Tensorflow code includes headers
1322 ;; from the "third_party" directory.  It does not look like we can replace
1323 ;; these headers with unmodified upstream files, so we keep them.
1324 (define-public tensorflow
1325   (package
1326     (name "tensorflow")
1327     (version "1.9.0")
1328     (source
1329      (origin
1330        (method git-fetch)
1331        (uri (git-reference
1332              (url "https://github.com/tensorflow/tensorflow.git")
1333              (commit (string-append "v" version))))
1334        (file-name (string-append "tensorflow-" version "-checkout"))
1335        (sha256
1336         (base32
1337          "0a9kwha395g3wgxfwln5j8vn9nkspmd75xldrlqdq540w996g8xa"))))
1338     (build-system cmake-build-system)
1339     (arguments
1340      `(#:tests? #f                      ; no "check" target
1341        #:build-type "Release"
1342        #:configure-flags
1343        (let ((protobuf (assoc-ref %build-inputs "protobuf"))
1344              (protobuf:native (assoc-ref %build-inputs "protobuf:native"))
1345              (jsoncpp (assoc-ref %build-inputs "jsoncpp"))
1346              (snappy (assoc-ref %build-inputs "snappy"))
1347              (sqlite (assoc-ref %build-inputs "sqlite")))
1348          (list
1349           ;; Use protobuf from Guix
1350           (string-append "-Dprotobuf_STATIC_LIBRARIES="
1351                          protobuf "/lib/libprotobuf.so")
1352           (string-append "-DPROTOBUF_PROTOC_EXECUTABLE="
1353                          protobuf:native "/bin/protoc")
1354
1355           ;; Use snappy from Guix
1356           (string-append "-Dsnappy_STATIC_LIBRARIES="
1357                          snappy "/lib/libsnappy.so")
1358           ;; Yes, this is not actually the include directory but a prefix...
1359           (string-append "-Dsnappy_INCLUDE_DIR=" snappy)
1360
1361           ;; Use jsoncpp from Guix
1362           (string-append "-Djsoncpp_STATIC_LIBRARIES="
1363                          jsoncpp "/lib/libjsoncpp.so")
1364           ;; Yes, this is not actually the include directory but a prefix...
1365           (string-append "-Djsoncpp_INCLUDE_DIR=" jsoncpp)
1366
1367           ;; Use sqlite from Guix
1368           (string-append "-Dsqlite_STATIC_LIBRARIES="
1369                          sqlite "/lib/libsqlite.a")
1370
1371           ;; Use system libraries wherever possible.  Currently, this
1372           ;; only affects zlib.
1373           "-Dsystemlib_ALL=ON"
1374           "-Dtensorflow_ENABLE_POSITION_INDEPENDENT_CODE=ON"
1375           "-Dtensorflow_BUILD_SHARED_LIB=ON"
1376           "-Dtensorflow_OPTIMIZE_FOR_NATIVE_ARCH=OFF"
1377           "-Dtensorflow_ENABLE_SSL_SUPPORT=OFF"
1378           "-Dtensorflow_BUILD_CONTRIB_KERNELS=OFF"))
1379        #:make-flags
1380        (list "CC=gcc")
1381        #:modules ((ice-9 ftw)
1382                   (guix build utils)
1383                   (guix build cmake-build-system))
1384        #:phases
1385        (modify-phases %standard-phases
1386          (add-after 'unpack 'set-source-file-times-to-1980
1387            ;; At the end of the tf_python_build_pip_package target, a ZIP
1388            ;; archive should be generated via bdist_wheel, but it fails with
1389            ;; "ZIP does not support timestamps before 1980".  Luckily,
1390            ;; SOURCE_DATE_EPOCH is respected, which we set to some time in
1391            ;; 1980.
1392            (lambda _ (setenv "SOURCE_DATE_EPOCH" "315532800") #t))
1393          ;; See https://github.com/tensorflow/tensorflow/issues/20517#issuecomment-406373913
1394          (add-after 'unpack 'python3.7-compatibility
1395            (lambda _
1396              (substitute* '("tensorflow/python/eager/pywrap_tfe_src.cc"
1397                             "tensorflow/python/lib/core/ndarray_tensor.cc"
1398                             "tensorflow/python/lib/core/py_func.cc")
1399                (("PyUnicode_AsUTF8") "(char *)PyUnicode_AsUTF8"))
1400              (substitute* "tensorflow/c/eager/c_api.h"
1401                (("unsigned char async")
1402                 "unsigned char is_async"))
1403
1404              ;; Remove dependency on tensorboard, a complicated but probably
1405              ;; optional package.
1406              (substitute* "tensorflow/tools/pip_package/setup.py"
1407                ((".*'tensorboard >.*") ""))
1408              #t))
1409          (add-after 'python3.7-compatibility 'chdir
1410            (lambda _ (chdir "tensorflow/contrib/cmake") #t))
1411          (add-after 'chdir 'disable-downloads
1412            (lambda* (#:key inputs #:allow-other-keys)
1413              (substitute* (find-files "external" "\\.cmake$")
1414                (("GIT_REPOSITORY.*") "")
1415                (("GIT_TAG.*") "")
1416                (("PREFIX ")
1417                 "DOWNLOAD_COMMAND \"\"\nPREFIX "))
1418
1419              ;; Use packages from Guix
1420              (let ((grpc (assoc-ref inputs "grpc")))
1421                (substitute* "CMakeLists.txt"
1422                  ;; Sqlite
1423                  (("include\\(sqlite\\)") "")
1424                  (("\\$\\{sqlite_STATIC_LIBRARIES\\}")
1425                   (string-append (assoc-ref inputs "sqlite")
1426                                  "/lib/libsqlite3.so"))
1427                  (("sqlite_copy_headers_to_destination") "")
1428
1429                  ;; PNG
1430                  (("include\\(png\\)") "")
1431                  (("\\$\\{png_STATIC_LIBRARIES\\}")
1432                   (string-append (assoc-ref inputs "libpng")
1433                                  "/lib/libpng16.so"))
1434                  (("png_copy_headers_to_destination") "")
1435
1436                  ;; JPEG
1437                  (("include\\(jpeg\\)") "")
1438                  (("\\$\\{jpeg_STATIC_LIBRARIES\\}")
1439                   (string-append (assoc-ref inputs "libjpeg")
1440                                  "/lib/libjpeg.so"))
1441                  (("jpeg_copy_headers_to_destination") "")
1442
1443                  ;; GIF
1444                  (("include\\(gif\\)") "")
1445                  (("\\$\\{gif_STATIC_LIBRARIES\\}")
1446                   (string-append (assoc-ref inputs "giflib")
1447                                  "/lib/libgif.so"))
1448                  (("gif_copy_headers_to_destination") "")
1449
1450                  ;; lmdb
1451                  (("include\\(lmdb\\)") "")
1452                  (("\\$\\{lmdb_STATIC_LIBRARIES\\}")
1453                   (string-append (assoc-ref inputs "lmdb")
1454                                  "/lib/liblmdb.so"))
1455                  (("lmdb_copy_headers_to_destination") "")
1456
1457                  ;; Protobuf
1458                  (("include\\(protobuf\\)") "")
1459                  (("protobuf_copy_headers_to_destination") "")
1460                  (("^ +protobuf") "")
1461
1462                  ;; gRPC
1463                  (("include\\(grpc\\)")
1464                   "find_package(grpc REQUIRED NAMES gRPC)")
1465                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES grpc\\)") "")
1466
1467                  ;; Eigen
1468                  (("include\\(eigen\\)")
1469                   (string-append "find_package(eigen REQUIRED NAMES Eigen3)
1470 set(eigen_INCLUDE_DIRS ${CMAKE_CURRENT_BINARY_DIR}/external/eigen_archive "
1471                                  (assoc-ref inputs "eigen") "/include/eigen3)"))
1472                  (("^ +eigen") "")
1473
1474                  ;; snappy
1475                  (("include\\(snappy\\)")
1476                   "add_definitions(-DTF_USE_SNAPPY)")
1477                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES snappy\\)") "")
1478
1479                  ;; jsoncpp
1480                  (("include\\(jsoncpp\\)") "")
1481                  (("^ +jsoncpp") ""))
1482
1483                (substitute* "tf_core_framework.cmake"
1484                  ((" grpc") "")
1485                  (("\\$\\{GRPC_BUILD\\}/grpc_cpp_plugin")
1486                   (which "grpc_cpp_plugin"))
1487                  ;; Link with gRPC libraries
1488                  (("add_library\\(tf_protos_cc.*" m)
1489                   (string-append m
1490                                  (format #f "\ntarget_link_libraries(tf_protos_cc PRIVATE \
1491 ~a/lib/libgrpc++_unsecure.a \
1492 ~a/lib/libgrpc_unsecure.a \
1493 ~a/lib/libaddress_sorting.a \
1494 ~a/lib/libgpr.a \
1495 ~a//lib/libcares.so
1496 )\n"
1497                                          grpc grpc grpc grpc
1498                                          (assoc-ref inputs "c-ares"))))))
1499              (substitute* "tf_tools.cmake"
1500                (("add_dependencies\\(\\$\\{proto_text.*") ""))
1501              ;; Remove dependency on bundled grpc
1502              (substitute* "tf_core_distributed_runtime.cmake"
1503                (("tf_core_cpu grpc") "tf_core_cpu"))
1504
1505              ;; This directory is a dependency of many targets.
1506              (mkdir-p "protobuf")
1507              #t))
1508          (add-after 'configure 'unpack-third-party-sources
1509            (lambda* (#:key inputs #:allow-other-keys)
1510              ;; This is needed to configure bundled packages properly.
1511              (setenv "CONFIG_SHELL" (which "bash"))
1512              (for-each
1513               (lambda (name)
1514                 (let* ((what  (assoc-ref inputs (string-append name "-src")))
1515                        (name* (string-map (lambda (c)
1516                                             (if (char=? c #\-)
1517                                                 #\_ c)) name))
1518                        (where (string-append "../build/" name* "/src/" name*)))
1519                   (cond
1520                    ((string-suffix? ".zip" what)
1521                     (mkdir-p where)
1522                     (with-directory-excursion where
1523                       (invoke "unzip" what)))
1524                    ((string-suffix? ".tar.gz" what)
1525                     (mkdir-p where)
1526                     (invoke "tar" "xf" what
1527                             "-C" where "--strip-components=1"))
1528                    (else
1529                     (let ((parent (dirname where)))
1530                       (mkdir-p parent)
1531                       (with-directory-excursion parent
1532                         (when (file-exists? name*)
1533                           (delete-file-recursively name*))
1534                         (copy-recursively what name*)
1535                         (map make-file-writable
1536                              (find-files name* ".*"))))))))
1537               (list "boringssl"
1538                     "cub"
1539                     "double-conversion"
1540                     "farmhash"
1541                     "fft2d"
1542                     "highwayhash"
1543                     "nsync"
1544                     "re2"))
1545
1546              (rename-file "../build/cub/src/cub/cub-1.8.0/"
1547                           "../build/cub/src/cub/cub/")
1548              #t))
1549          (add-after 'unpack 'fix-python-build
1550            (lambda* (#:key inputs outputs #:allow-other-keys)
1551              (mkdir-p "protobuf-src")
1552              (invoke "tar" "xf" (assoc-ref inputs "protobuf:src")
1553                      "-C" "protobuf-src" "--strip-components=1")
1554              (mkdir-p "eigen-src")
1555              (invoke "tar" "xf" (assoc-ref inputs "eigen:src")
1556                      "-C" "eigen-src" "--strip-components=1")
1557
1558              (substitute* "tensorflow/contrib/cmake/tf_python.cmake"
1559                ;; Ensure that all Python dependencies can be found at build time.
1560                (("PYTHONPATH=\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/tf_python" m)
1561                 (string-append m ":" (getenv "PYTHONPATH")))
1562                ;; Take protobuf source files from our source package.
1563                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/protobuf/src/protobuf/src/google")
1564                 (string-append (getcwd) "/protobuf-src/src/google")))
1565
1566              (substitute* '("tensorflow/contrib/cmake/tf_shared_lib.cmake"
1567                             "tensorflow/contrib/cmake/tf_python.cmake")
1568                ;; Take Eigen source files from our source package.
1569                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/eigen/src/eigen/")
1570                 (string-append (getcwd) "/eigen-src/"))
1571                ;; Take Eigen headers from our own package.
1572                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/external/eigen_archive")
1573                 (string-append (assoc-ref inputs "eigen") "/include/eigen3")))
1574
1575              ;; Correct the RUNPATH of ops libraries generated for Python.
1576              ;; TODO: this doesn't work :(
1577              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1578              ;; warning: RUNPATH contains bogus entries: ("/tmp/guix-build-tensorflow-1.9.0.drv-0/source/tensorflow/contrib/build")
1579              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1580              ;; error: depends on 'libpywrap_tensorflow_internal.so', which
1581              ;; cannot be found in RUNPATH ...
1582              (substitute* "tensorflow/contrib/cmake/tf_cc_ops.cmake"
1583                (("set_target_properties.*")
1584                 (string-append "set_target_properties(${_AT_TARGET} PROPERTIES \
1585 COMPILE_FLAGS ${target_compile_flags} \
1586 INSTALL_RPATH_USE_LINK_PATH TRUE \
1587 INSTALL_RPATH " (assoc-ref outputs "out") "/lib)\n")))
1588              #t))
1589          (add-after 'build 'build-pip-package
1590            (lambda* (#:key outputs #:allow-other-keys)
1591              (setenv "LDFLAGS"
1592                      (string-append "-Wl,-rpath="
1593                                     (assoc-ref outputs "out") "/lib"))
1594              (invoke "make" "tf_python_build_pip_package")
1595              #t))
1596          (add-after 'build-pip-package 'install-python
1597            (lambda* (#:key outputs #:allow-other-keys)
1598              (let ((out (assoc-ref outputs "out"))
1599                    (wheel (car (find-files "../build/tf_python/dist/" "\\.whl$"))))
1600                (invoke "python" "-m" "pip" "install" wheel
1601                        (string-append "--prefix=" out))
1602
1603                ;; XXX: broken RUNPATH, see fix-python-build phase.
1604                (delete-file
1605                 (string-append
1606                  out "/lib/python3.7/site-packages/tensorflow/contrib/"
1607                  "seq2seq/python/ops/lib_beam_search_ops.so"))
1608                #t))))))
1609     (native-inputs
1610      `(("pkg-config" ,pkg-config)
1611        ("protobuf:native" ,protobuf-3.6) ; protoc
1612        ("protobuf:src" ,(package-source protobuf-3.6))
1613        ("eigen:src" ,(package-source eigen-for-tensorflow))
1614        ;; install_pip_packages.sh wants setuptools 39.1.0 specifically.
1615        ("python-setuptools" ,python-setuptools-for-tensorflow)
1616
1617        ;; The commit hashes and URLs for third-party source code are taken
1618        ;; from "tensorflow/workspace.bzl".
1619        ("boringssl-src"
1620         ,(let ((commit "ee7aa02")
1621                (revision "1"))
1622            (origin
1623              (method git-fetch)
1624              (uri (git-reference
1625                    (url "https://boringssl.googlesource.com/boringssl")
1626                    (commit commit)))
1627              (file-name (string-append "boringssl-0-" revision
1628                                        (string-take commit 7)
1629                                        "-checkout"))
1630              (sha256
1631               (base32
1632                "1jf693q0nw0adsic6cgmbdx6g7wr4rj4vxa8j1hpn792fqhd8wgw")))))
1633        ("cub-src"
1634         ,(let ((version "1.8.0"))
1635            (origin
1636              (method url-fetch)
1637              (uri (string-append "https://mirror.bazel.build/github.com/NVlabs/"
1638                                  "cub/archive/" version ".zip"))
1639              (file-name (string-append "cub-" version ".zip"))
1640              (sha256
1641               (base32
1642                "1hsqikqridb90dkxkjr2918dcry6pfh46ccnwrzawl56aamhdykb")))))
1643        ("double-conversion-src"
1644         ,(let ((commit "5664746")
1645                (revision "1"))
1646            (origin
1647              (method git-fetch)
1648              (uri (git-reference
1649                    (url "https://github.com/google/double-conversion.git")
1650                    (commit commit)))
1651              (file-name
1652               (git-file-name "double-conversion"
1653                              (string-append "0-" revision "."
1654                                             (string-take commit 7))))
1655              (sha256
1656               (base32
1657                "1h5lppqqxcvdg5jq42i5msgwx20ryij3apvmndflngrgdpc04gn1")))))
1658        ("farmhash-src"
1659         ,(let ((commit "816a4ae622e964763ca0862d9dbd19324a1eaf45"))
1660            (origin
1661              (method url-fetch)
1662              (uri (string-append
1663                    "https://mirror.bazel.build/github.com/google/farmhash/archive/"
1664                    commit ".tar.gz"))
1665              (file-name (string-append "farmhash-0-" (string-take commit 7)
1666                                        ".tar.gz"))
1667              (sha256
1668               (base32
1669                "185b2xdxl4d4cnsnv6abg8s22gxvx8673jq2yaq85bz4cdy58q35")))))
1670        ;; The license notice on the home page at
1671        ;; http://www.kurims.kyoto-u.ac.jp/~ooura/fft.html says:
1672        ;;   Copyright Takuya OOURA, 1996-2001
1673        ;;
1674        ;;   You may use, copy, modify and distribute this code for any purpose
1675        ;;   (include commercial use) and without fee. Please refer to this
1676        ;;   package when you modify this code.
1677        ;;
1678        ;; We take the identical tarball from the Bazel mirror, because the URL
1679        ;; at the home page is not versioned and might change.
1680        ("fft2d-src"
1681         ,(origin
1682            (method url-fetch)
1683            (uri "https://mirror.bazel.build/www.kurims.kyoto-u.ac.jp/~ooura/fft.tgz")
1684            (file-name "fft2d.tar.gz")
1685            (sha256
1686             (base32
1687              "15jjkfvhqvl2c0753d2di8hz0pyzn598g74wqy79awdrf1y67fsj"))))
1688        ("highwayhash-src"
1689         ,(let ((commit "be5edafc2e1a455768e260ccd68ae7317b6690ee")
1690                (revision "1"))
1691            (origin
1692              (method git-fetch)
1693              (uri (git-reference
1694                    (url "https://github.com/google/highwayhash.git")
1695                    (commit commit)))
1696              (file-name (string-append "highwayhash-0-" revision
1697                                        (string-take commit 7)
1698                                        "-checkout"))
1699              (sha256
1700               (base32
1701                "154jwf98cyy54hldr94pgjn85zynly3abpnc1avmb8a18lzwjyb6")))))
1702        ("nsync-src"
1703         ,(let ((version "0559ce013feac8db639ee1bf776aca0325d28777")
1704                (revision "1"))
1705            (origin
1706              (method url-fetch)
1707              (uri (string-append "https://mirror.bazel.build/"
1708                                  "github.com/google/nsync/archive/"
1709                                  version ".tar.gz"))
1710              (file-name (string-append "nsync-0." revision
1711                                        "-" (string-take version 7)
1712                                        ".tar.gz"))
1713              (sha256
1714               (base32
1715                "0qdkyqym34x739mmzv97ah5r7ph462v5xkxqxvidmcfqbi64b132")))))
1716        ("re2-src"
1717         ,(let ((commit "e7efc48")
1718                (revision "1"))
1719            (origin
1720              (method git-fetch)
1721              (uri (git-reference
1722                    (url "https://github.com/google/re2")
1723                    (commit commit)))
1724              (file-name (string-append "re2-0-" revision
1725                                        (string-take commit 7)
1726                                        "-checkout"))
1727              (sha256
1728               (base32
1729                "161g9841rjfsy5pn52fcis0s9hdr7rxvb06pad38j5rppfihvign")))))
1730        ("googletest" ,googletest)
1731        ("swig" ,swig)
1732        ("unzip" ,unzip)))
1733     (propagated-inputs
1734      `(("python-absl-py" ,python-absl-py)
1735        ("python-astor" ,python-astor)
1736        ("python-gast" ,python-gast)
1737        ("python-grpcio" ,python-grpcio)
1738        ("python-numpy" ,python-numpy)
1739        ("python-protobuf" ,python-protobuf-3.6)
1740        ("python-six" ,python-six)
1741        ("python-termcolo" ,python-termcolor)
1742        ("python-wheel" ,python-wheel)))
1743     (inputs
1744      `(("c-ares" ,c-ares)
1745        ("eigen" ,eigen-for-tensorflow)
1746        ("gemmlowp" ,gemmlowp-for-tensorflow)
1747        ("lmdb" ,lmdb)
1748        ("libjpeg" ,libjpeg)
1749        ("libpng" ,libpng)
1750        ("giflib" ,giflib)
1751        ("grpc" ,grpc)
1752        ("jsoncpp" ,jsoncpp-for-tensorflow)
1753        ("snappy" ,snappy)
1754        ("sqlite" ,sqlite)
1755        ("protobuf" ,protobuf-3.6)
1756        ("python" ,python-wrapper)
1757        ("zlib" ,zlib)))
1758     (home-page "https://tensorflow.org")
1759     (synopsis "Machine learning framework")
1760     (description
1761      "TensorFlow is a flexible platform for building and training machine
1762 learning models.  It provides a library for high performance numerical
1763 computation and includes high level Python APIs, including both a sequential
1764 API for beginners that allows users to build models quickly by plugging
1765 together building blocks and a subclassing API with an imperative style for
1766 advanced research.")
1767     (license license:asl2.0)))
1768
1769 (define-public python-iml
1770   (package
1771     (name "python-iml")
1772     (version "0.6.2")
1773     (source
1774      (origin
1775        (method url-fetch)
1776        (uri (pypi-uri "iml" version))
1777        (sha256
1778         (base32
1779          "1k8szlpm19rcwcxdny9qdm3gmaqq8akb4xlvrzyz8c2d679aak6l"))))
1780     (build-system python-build-system)
1781     (propagated-inputs
1782      `(("ipython" ,python-ipython)
1783        ("nose" ,python-nose)
1784        ("numpy" ,python-numpy)
1785        ("pandas" ,python-pandas)
1786        ("scipy" ,python-scipy)))
1787     (home-page "http://github.com/interpretable-ml/iml")
1788     (synopsis "Interpretable Machine Learning (iML) package")
1789     (description "Interpretable ML (iML) is a set of data type objects,
1790 visualizations, and interfaces that can be used by any method designed to
1791 explain the predictions of machine learning models (or really the output of
1792 any function).  It currently contains the interface and IO code from the Shap
1793 project, and it will potentially also do the same for the Lime project.")
1794     (license license:expat)))
1795
1796 (define-public python-keras-applications
1797   (package
1798     (name "python-keras-applications")
1799     (version "1.0.8")
1800     (source
1801      (origin
1802        (method url-fetch)
1803        (uri (pypi-uri "Keras_Applications" version))
1804        (sha256
1805         (base32
1806          "1rcz31ca4axa6kzhjx4lwqxbg4wvlljkj8qj9a7p9sfd5fhzjyam"))))
1807     (build-system python-build-system)
1808     ;; The tests require Keras, but this package is needed to build Keras.
1809     (arguments '(#:tests? #f))
1810     (propagated-inputs
1811      `(("python-h5py" ,python-h5py)
1812        ("python-numpy" ,python-numpy)))
1813     (native-inputs
1814      `(("python-pytest" ,python-pytest)
1815        ("python-pytest-cov" ,python-pytest-cov)
1816        ("python-pytest-pep8" ,python-pytest-pep8)
1817        ("python-pytest-xdist" ,python-pytest-xdist)))
1818     (home-page "https://github.com/keras-team/keras-applications")
1819     (synopsis "Reference implementations of popular deep learning models")
1820     (description
1821      "This package provides reference implementations of popular deep learning
1822 models for use with the Keras deep learning framework.")
1823     (license license:expat)))
1824
1825 (define-public python-keras-preprocessing
1826   (package
1827     (name "python-keras-preprocessing")
1828     (version "1.1.0")
1829     (source
1830      (origin
1831        (method url-fetch)
1832        (uri (pypi-uri "Keras_Preprocessing" version))
1833        (sha256
1834         (base32
1835          "1r98nm4k1svsqjyaqkfk23i31bl1kcfcyp7094yyj3c43phfp3as"))))
1836     (build-system python-build-system)
1837     (propagated-inputs
1838      `(("python-numpy" ,python-numpy)
1839        ("python-six" ,python-six)))
1840     (native-inputs
1841      `(("python-pandas" ,python-pandas)
1842        ("python-pillow" ,python-pillow)
1843        ("python-pytest" ,python-pytest)
1844        ("python-pytest-cov" ,python-pytest-cov)
1845        ("python-pytest-xdist" ,python-pytest-xdist)
1846        ("tensorflow" ,tensorflow)))
1847     (home-page "https://github.com/keras-team/keras-preprocessing/")
1848     (synopsis "Data preprocessing and augmentation for deep learning models")
1849     (description
1850      "Keras Preprocessing is the data preprocessing and data augmentation
1851 module of the Keras deep learning library.  It provides utilities for working
1852 with image data, text data, and sequence data.")
1853     (license license:expat)))
1854
1855 (define-public python-keras
1856   (package
1857     (name "python-keras")
1858     (version "2.2.4")
1859     (source
1860      (origin
1861        (method url-fetch)
1862        (uri (pypi-uri "Keras" version))
1863        (patches (search-patches "python-keras-integration-test.patch"))
1864        (sha256
1865         (base32
1866          "1j8bsqzh49vjdxy6l1k4iwax5vpjzniynyd041xjavdzvfii1dlh"))))
1867     (build-system python-build-system)
1868     (arguments
1869      `(#:phases
1870        (modify-phases %standard-phases
1871          (add-after 'unpack 'remove-tests-for-unavailable-features
1872            (lambda _
1873              (delete-file "keras/backend/theano_backend.py")
1874              (delete-file "keras/backend/cntk_backend.py")
1875              (delete-file "tests/keras/backend/backend_test.py")
1876
1877              ;; FIXME: This doesn't work because Tensorflow is missing the
1878              ;; coder ops library.
1879              (delete-file "tests/keras/test_callbacks.py")
1880              #t))
1881          (replace 'check
1882            (lambda _
1883              ;; These tests attempt to download data files from the internet.
1884              (delete-file "tests/integration_tests/test_datasets.py")
1885              (delete-file "tests/integration_tests/imagenet_utils_test.py")
1886
1887              (setenv "PYTHONPATH"
1888                      (string-append (getcwd) "/build/lib:"
1889                                     (getenv "PYTHONPATH")))
1890              (invoke "py.test" "-v"
1891                      "-p" "no:cacheprovider"
1892                      "--ignore" "keras/utils"))))))
1893     (propagated-inputs
1894      `(("python-h5py" ,python-h5py)
1895        ("python-keras-applications" ,python-keras-applications)
1896        ("python-keras-preprocessing" ,python-keras-preprocessing)
1897        ("python-numpy" ,python-numpy)
1898        ("python-pydot" ,python-pydot)
1899        ("python-pyyaml" ,python-pyyaml)
1900        ("python-scipy" ,python-scipy)
1901        ("python-six" ,python-six)
1902        ("tensorflow" ,tensorflow)
1903        ("graphviz" ,graphviz)))
1904     (native-inputs
1905      `(("python-pandas" ,python-pandas)
1906        ("python-pytest" ,python-pytest)
1907        ("python-pytest-cov" ,python-pytest-cov)
1908        ("python-pytest-pep8" ,python-pytest-pep8)
1909        ("python-pytest-timeout" ,python-pytest-timeout)
1910        ("python-pytest-xdist" ,python-pytest-xdist)
1911        ("python-sphinx" ,python-sphinx)
1912        ("python-requests" ,python-requests)))
1913     (home-page "https://github.com/keras-team/keras")
1914     (synopsis "High-level deep learning framework")
1915     (description "Keras is a high-level neural networks API, written in Python
1916 and capable of running on top of TensorFlow.  It was developed with a focus on
1917 enabling fast experimentation.  Use Keras if you need a deep learning library
1918 that:
1919
1920 @itemize
1921 @item Allows for easy and fast prototyping (through user friendliness,
1922   modularity, and extensibility).
1923 @item Supports both convolutional networks and recurrent networks, as well as
1924   combinations of the two.
1925 @item Runs seamlessly on CPU and GPU.
1926 @end itemize\n")
1927     (license license:expat)))
1928
1929 (define-public sbcl-cl-libsvm-format
1930   (let ((commit "3300f84fd8d9f5beafc114f543f9d83417c742fb")
1931         (revision "0"))
1932     (package
1933       (name "sbcl-cl-libsvm-format")
1934       (version (git-version "0.1.0" revision commit))
1935       (source
1936        (origin
1937          (method git-fetch)
1938          (uri (git-reference
1939                (url "https://github.com/masatoi/cl-libsvm-format.git")
1940                (commit commit)))
1941          (file-name (git-file-name name version))
1942          (sha256
1943           (base32
1944            "0284aj84xszhkhlivaigf9qj855fxad3mzmv3zfr0qzb5k0nzwrg"))))
1945       (build-system asdf-build-system/sbcl)
1946       (native-inputs
1947        `(("prove" ,sbcl-prove)
1948          ("prove-asdf" ,sbcl-prove-asdf)))
1949       (inputs
1950        `(("alexandria" ,sbcl-alexandria)))
1951       (synopsis "LibSVM data format reader for Common Lisp")
1952       (description
1953        "This Common Lisp library provides a fast reader for data in LibSVM
1954 format.")
1955       (home-page "https://github.com/masatoi/cl-libsvm-format")
1956       (license license:expat))))
1957
1958 (define-public cl-libsvm-format
1959   (sbcl-package->cl-source-package sbcl-cl-libsvm-format))
1960
1961 (define-public ecl-cl-libsvm-format
1962   (sbcl-package->ecl-package sbcl-cl-libsvm-format))
1963
1964 (define-public sbcl-cl-online-learning
1965   (let ((commit "fc7a34f4f161cd1c7dd747d2ed8f698947781423")
1966         (revision "0"))
1967     (package
1968       (name "sbcl-cl-online-learning")
1969       (version (git-version "0.5" revision commit))
1970       (source
1971        (origin
1972          (method git-fetch)
1973          (uri (git-reference
1974                (url "https://github.com/masatoi/cl-online-learning.git")
1975                (commit commit)))
1976          (file-name (git-file-name name version))
1977          (sha256
1978           (base32
1979            "14x95rlg80ay5hv645ki57pqvy12v28hz4k1w0f6bsfi2rmpxchq"))))
1980       (build-system asdf-build-system/sbcl)
1981       (native-inputs
1982        `(("prove" ,sbcl-prove)
1983          ("prove-asdf" ,sbcl-prove-asdf)))
1984       (inputs
1985        `(("cl-libsvm-format" ,sbcl-cl-libsvm-format)
1986          ("cl-store" ,sbcl-cl-store)))
1987       (arguments
1988        `(;; FIXME: Tests pass but then the check phase crashes
1989          #:tests? #f))
1990       (synopsis "Online Machine Learning for Common Lisp")
1991       (description
1992        "This library contains a collection of machine learning algorithms for
1993 online linear classification written in Common Lisp.")
1994       (home-page "https://github.com/masatoi/cl-online-learning")
1995       (license license:expat))))
1996
1997 (define-public cl-online-learning
1998   (sbcl-package->cl-source-package sbcl-cl-online-learning))
1999
2000 (define-public ecl-cl-online-learning
2001   (sbcl-package->ecl-package sbcl-cl-online-learning))
2002
2003 (define-public sbcl-cl-random-forest
2004   (let ((commit "85fbdd4596d40e824f70f1b7cf239cf544e49d51")
2005         (revision "0"))
2006     (package
2007       (name "sbcl-cl-random-forest")
2008       (version (git-version "0.1" revision commit))
2009       (source
2010        (origin
2011          (method git-fetch)
2012          (uri (git-reference
2013                (url "https://github.com/masatoi/cl-random-forest.git")
2014                (commit commit)))
2015          (file-name (git-file-name name version))
2016          (sha256
2017           (base32
2018            "097xv60i1ndz68sg9p4pc7c5gvyp9i1xgw966b4wwfq3x6hbz421"))))
2019       (build-system asdf-build-system/sbcl)
2020       (native-inputs
2021        `(("prove" ,sbcl-prove)
2022          ("prove-asdf" ,sbcl-prove-asdf)
2023          ("trivial-garbage" ,sbcl-trivial-garbage)))
2024       (inputs
2025        `(("alexandria" ,sbcl-alexandria)
2026          ("cl-libsvm-format" ,sbcl-cl-libsvm-format)
2027          ("cl-online-learning" ,sbcl-cl-online-learning)
2028          ("lparallel" ,sbcl-lparallel)))
2029       (arguments
2030        `(;; The tests download data from the Internet
2031          #:tests? #f
2032          #:phases
2033          (modify-phases %standard-phases
2034            (add-after 'unpack 'add-sb-cltl2-dependency
2035              (lambda _
2036                ;; sb-cltl2 is required by lparallel when using sbcl, but it is
2037                ;; not loaded automatically.
2038                (substitute* "cl-random-forest.asd"
2039                  (("\\(in-package :cl-user\\)")
2040                   "(in-package :cl-user) #+sbcl (require :sb-cltl2)"))
2041                #t)))))
2042       (synopsis "Random Forest and Global Refinement for Common Lisp")
2043       (description
2044        "CL-random-forest is an implementation of Random Forest for multiclass
2045 classification and univariate regression written in Common Lisp.  It also
2046 includes an implementation of Global Refinement of Random Forest.")
2047       (home-page "https://github.com/masatoi/cl-random-forest")
2048       (license license:expat))))
2049
2050 (define-public cl-random-forest
2051   (sbcl-package->cl-source-package sbcl-cl-random-forest))
2052
2053 (define-public ecl-cl-random-forest
2054   (sbcl-package->ecl-package sbcl-cl-random-forest))
2055
2056 (define-public gloo
2057   (let ((version "0.0.0") ; no proper version tag
2058         (commit "ca528e32fea9ca8f2b16053cff17160290fc84ce")
2059         (revision "0"))
2060     (package
2061       (name "gloo")
2062       (version (git-version version revision commit))
2063       (source
2064        (origin
2065          (method git-fetch)
2066          (uri (git-reference
2067                (url "https://github.com/facebookincubator/gloo.git")
2068                (commit commit)))
2069          (file-name (git-file-name name version))
2070          (sha256
2071           (base32
2072            "1q9f80zy75f6njrzrqkmhc0g3qxs4gskr7ns2jdqanxa2ww7a99w"))))
2073       (build-system cmake-build-system)
2074       (native-inputs
2075        `(("googletest" ,googletest)))
2076       (arguments
2077        `(#:configure-flags '("-DBUILD_TEST=1")
2078          #:phases
2079          (modify-phases %standard-phases
2080            (replace 'check
2081              (lambda _
2082                (invoke "make" "gloo_test")
2083                #t)))))
2084       (synopsis "Collective communications library")
2085       (description
2086        "Gloo is a collective communications library.  It comes with a
2087 number of collective algorithms useful for machine learning applications.
2088 These include a barrier, broadcast, and allreduce.")
2089       (home-page "https://github.com/facebookincubator/gloo")
2090       (license license:bsd-3))))