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 ;;; Copyright © 2019 Brett Gilio <brettg@gnu.org>
  16 ;;;
  17 ;;; This file is part of GNU Guix.
  18 ;;;
  19 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  20 ;;; under the terms of the GNU General Public License as published by
  21 ;;; the Free Software Foundation; either version 3 of the License, or (at
  22 ;;; your option) any later version.
  23 ;;;
  24 ;;; GNU Guix is distributed in the hope that it will be useful, but
  25 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  26 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  27 ;;; GNU General Public License for more details.
  28 ;;;
  29 ;;; You should have received a copy of the GNU General Public License
  30 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  31
  32 (define-module (gnu packages machine-learning)
  33   #:use-module ((guix licenses) #:prefix license:)
  34   #:use-module (guix packages)
  35   #:use-module (guix utils)
  36   #:use-module (guix download)
  37   #:use-module (guix svn-download)
  38   #:use-module (guix build-system asdf)
  39   #:use-module (guix build-system cmake)
  40   #:use-module (guix build-system gnu)
  41   #:use-module (guix build-system ocaml)
  42   #:use-module (guix build-system python)
  43   #:use-module (guix build-system r)
  44   #:use-module (guix git-download)
  45   #:use-module (gnu packages)
  46   #:use-module (gnu packages adns)
  47   #:use-module (gnu packages algebra)
  48   #:use-module (gnu packages audio)
  49   #:use-module (gnu packages autotools)
  50   #:use-module (gnu packages base)
  51   #:use-module (gnu packages bash)
  52   #:use-module (gnu packages boost)
  53   #:use-module (gnu packages check)
  54   #:use-module (gnu packages compression)
  55   #:use-module (gnu packages cran)
  56   #:use-module (gnu packages databases)
  57   #:use-module (gnu packages dejagnu)
  58   #:use-module (gnu packages gcc)
  59   #:use-module (gnu packages glib)
  60   #:use-module (gnu packages graphviz)
  61   #:use-module (gnu packages gstreamer)
  62   #:use-module (gnu packages image)
  63   #:use-module (gnu packages linux)
  64   #:use-module (gnu packages lisp-xyz)
  65   #:use-module (gnu packages maths)
  66   #:use-module (gnu packages mpi)
  67   #:use-module (gnu packages ocaml)
  68   #:use-module (gnu packages onc-rpc)
  69   #:use-module (gnu packages perl)
  70   #:use-module (gnu packages pkg-config)
  71   #:use-module (gnu packages protobuf)
  72   #:use-module (gnu packages python)
  73   #:use-module (gnu packages python-science)
  74   #:use-module (gnu packages python-web)
  75   #:use-module (gnu packages python-xyz)
  76   #:use-module (gnu packages serialization)
  77   #:use-module (gnu packages sphinx)
  78   #:use-module (gnu packages statistics)
  79   #:use-module (gnu packages sqlite)
  80   #:use-module (gnu packages swig)
  81   #:use-module (gnu packages tls)
  82   #:use-module (gnu packages web)
  83   #:use-module (gnu packages xml)
  84   #:use-module (gnu packages xorg)
  85   #:use-module (ice-9 match))
  86
  87 (define-public fann
  88   ;; The last release is >100 commits behind, so we package from git.
  89   (let ((commit "d71d54788bee56ba4cf7522801270152da5209d7"))
  90     (package
  91       (name "fann")
  92       (version (string-append "2.2.0-1." (string-take commit 8)))
  93       (source (origin
  94                 (method git-fetch)
  95                 (uri (git-reference
  96                       (url "https://github.com/libfann/fann.git")
  97                       (commit commit)))
  98                 (file-name (string-append name "-" version "-checkout"))
  99                 (sha256
 100                  (base32
 101                   "0ibwpfrjs6q2lijs8slxjgzb2llcl6rk3v2ski4r6215g5jjhg3x"))))
 102       (build-system cmake-build-system)
 103       (arguments
 104        `(#:phases
 105          (modify-phases %standard-phases
 106            (replace 'check
 107              (lambda* (#:key outputs #:allow-other-keys)
 108                (let* ((out (assoc-ref outputs "out")))
 109                  (with-directory-excursion (string-append (getcwd) "/tests")
 110                    (invoke "./fann_tests"))))))))
 111       (home-page "http://leenissen.dk/fann/wp/")
 112       (synopsis "Fast Artificial Neural Network")
 113       (description
 114        "FANN is a neural network library, which implements multilayer
 115 artificial neural networks in C with support for both fully connected and
 116 sparsely connected networks.")
 117       (license license:lgpl2.1))))
 118
 119 (define-public libsvm
 120   (package
 121     (name "libsvm")
 122     (version "3.23")
 123     (source
 124      (origin
 125        (method url-fetch)
 126        (uri (string-append "https://www.csie.ntu.edu.tw/~cjlin/libsvm/"
 127                            name "-" version ".tar.gz"))
 128        (sha256
 129         (base32 "0jpaq0rr92x38p4nk3gjan79ip67m6p80anb28z1d8601miysyi5"))))
 130     (build-system gnu-build-system)
 131     (arguments
 132      `(#:tests? #f                      ; no "check" target
 133        #:phases (modify-phases %standard-phases
 134                   (delete 'configure)
 135                   (replace
 136                    'install             ; no ‘install’ target
 137                    (lambda* (#:key outputs #:allow-other-keys)
 138                      (let* ((out (assoc-ref outputs "out"))
 139                             (bin (string-append out "/bin/")))
 140                        (mkdir-p bin)
 141                        (for-each (lambda (file)
 142                                    (copy-file file (string-append bin file)))
 143                                  '("svm-train"
 144                                    "svm-predict"
 145                                    "svm-scale")))
 146                      #t)))))
 147     (home-page "http://www.csie.ntu.edu.tw/~cjlin/libsvm/")
 148     (synopsis "Library for Support Vector Machines")
 149     (description
 150      "LIBSVM is a machine learning library for support vector
 151 classification, (C-SVC, nu-SVC), regression (epsilon-SVR, nu-SVR) and
 152 distribution estimation (one-class SVM).  It supports multi-class
 153 classification.")
 154     (license license:bsd-3)))
 155
 156 (define-public python-libsvm
 157   (package (inherit libsvm)
 158     (name "python-libsvm")
 159     (build-system gnu-build-system)
 160     (arguments
 161      `(#:tests? #f                      ; no "check" target
 162        #:make-flags '("-C" "python")
 163        #:phases
 164        (modify-phases %standard-phases
 165          (delete 'configure)
 166          (replace
 167           'install                      ; no ‘install’ target
 168           (lambda* (#:key inputs outputs #:allow-other-keys)
 169             (let ((site (string-append (assoc-ref outputs "out")
 170                                        "/lib/python"
 171                                        (string-take
 172                                         (string-take-right
 173                                          (assoc-ref inputs "python") 5) 3)
 174                                        "/site-packages/")))
 175               (substitute* "python/svm.py"
 176                 (("../libsvm.so.2") "libsvm.so.2"))
 177               (mkdir-p site)
 178               (for-each (lambda (file)
 179                           (copy-file file (string-append site (basename file))))
 180                         (find-files "python" "\\.py"))
 181               (copy-file "libsvm.so.2"
 182                          (string-append site "libsvm.so.2")))
 183             #t)))))
 184     (inputs
 185      `(("python" ,python)))
 186     (synopsis "Python bindings of libSVM")))
 187
 188 (define-public ghmm
 189   ;; The latest release candidate is several years and a couple of fixes have
 190   ;; been published since.  This is why we download the sources from the SVN
 191   ;; repository.
 192   (let ((svn-revision 2341))
 193     (package
 194       (name "ghmm")
 195       (version (string-append "0.9-rc3-0." (number->string svn-revision)))
 196       (source (origin
 197                 (method svn-fetch)
 198                 (uri (svn-reference
 199                       (url "http://svn.code.sf.net/p/ghmm/code/trunk")
 200                       (revision svn-revision)))
 201                 (file-name (string-append name "-" version "-checkout"))
 202                 (sha256
 203                  (base32
 204                   "0qbq1rqp94l530f043qzp8aw5lj7dng9wq0miffd7spd1ff638wq"))))
 205       (build-system gnu-build-system)
 206       (arguments
 207        `(#:imported-modules (,@%gnu-build-system-modules
 208                              (guix build python-build-system))
 209          #:modules          ((guix build python-build-system)
 210                              ,@%gnu-build-system-modules)
 211          #:phases
 212          (modify-phases %standard-phases
 213            (add-after 'unpack 'enter-dir
 214              (lambda _ (chdir "ghmm") #t))
 215            (delete 'check)
 216            (add-after 'install 'check
 217              (assoc-ref %standard-phases 'check))
 218            (add-before 'check 'fix-PYTHONPATH
 219              (lambda* (#:key inputs outputs #:allow-other-keys)
 220                (let ((python-version (python-version
 221                                       (assoc-ref inputs "python"))))
 222                  (setenv "PYTHONPATH"
 223                          (string-append (getenv "PYTHONPATH")
 224                                         ":" (assoc-ref outputs "out")
 225                                         "/lib/python" python-version
 226                                         "/site-packages")))
 227                #t))
 228            (add-after 'enter-dir 'fix-runpath
 229              (lambda* (#:key outputs #:allow-other-keys)
 230                (substitute* "ghmmwrapper/setup.py"
 231                  (("^(.*)extra_compile_args = \\[" line indent)
 232                   (string-append indent
 233                                  "extra_link_args = [\"-Wl,-rpath="
 234                                  (assoc-ref outputs "out") "/lib\"],\n"
 235                                  line
 236                                  "\"-Wl,-rpath="
 237                                  (assoc-ref outputs "out")
 238                                  "/lib\", ")))
 239                #t))
 240            (add-after 'enter-dir 'disable-broken-tests
 241              (lambda _
 242                (substitute* "tests/Makefile.am"
 243                  ;; GHMM_SILENT_TESTS is assumed to be a command.
 244                  (("TESTS_ENVIRONMENT.*") "")
 245                  ;; Do not build broken tests.
 246                  (("chmm .*") "")
 247                  (("read_fa .*") "")
 248                  (("mcmc .*") "")
 249                  (("label_higher_order_test.*$")
 250                   "label_higher_order_test\n"))
 251
 252                ;; These Python unittests are broken as there is no gato.
 253                ;; See https://sourceforge.net/p/ghmm/support-requests/3/
 254                (substitute* "ghmmwrapper/ghmmunittests.py"
 255                  (("^(.*)def (testNewXML|testMultipleTransitionClasses|testNewXML)"
 256                    line indent)
 257                   (string-append indent
 258                                  "@unittest.skip(\"Disabled by Guix\")\n"
 259                                  line)))
 260                #t)))))
 261       (inputs
 262        `(("python" ,python-2) ; only Python 2 is supported
 263          ("libxml2" ,libxml2)))
 264       (native-inputs
 265        `(("pkg-config" ,pkg-config)
 266          ("dejagnu" ,dejagnu)
 267          ("swig" ,swig)
 268          ("autoconf" ,autoconf)
 269          ("automake" ,automake)
 270          ("libtool" ,libtool)))
 271       (home-page "http://ghmm.org")
 272       (synopsis "Hidden Markov Model library")
 273       (description
 274        "The General Hidden Markov Model library (GHMM) is a C library with
 275 additional Python bindings implementing a wide range of types of @dfn{Hidden
 276 Markov Models} (HMM) and algorithms: discrete, continuous emissions, basic
 277 training, HMM clustering, HMM mixtures.")
 278       (license license:lgpl2.0+))))
 279
 280 (define-public mcl
 281   (package
 282     (name "mcl")
 283     (version "14.137")
 284     (source (origin
 285               (method url-fetch)
 286               (uri (string-append
 287                     "http://micans.org/mcl/src/mcl-"
 288                     (string-replace-substring version "." "-")
 289                     ".tar.gz"))
 290               (sha256
 291                (base32
 292                 "15xlax3z31lsn62vlg94hkm75nm40q4679amnfg13jm8m2bnhy5m"))))
 293     (build-system gnu-build-system)
 294     (arguments
 295      `(#:configure-flags (list "--enable-blast")))
 296     (inputs
 297      `(("perl" ,perl)))
 298     (home-page "http://micans.org/mcl/")
 299     (synopsis "Clustering algorithm for graphs")
 300     (description
 301      "The MCL algorithm is short for the @dfn{Markov Cluster Algorithm}, a
 302 fast and scalable unsupervised cluster algorithm for graphs (also known as
 303 networks) based on simulation of (stochastic) flow in graphs.")
 304     ;; In the LICENCE file and web page it says "The software is licensed
 305     ;; under the GNU General Public License, version 3.", but in several of
 306     ;; the source code files it suggests GPL3 or later.
 307     ;; http://listserver.ebi.ac.uk/pipermail/mcl-users/2016/000376.html
 308     (license license:gpl3)))
 309
 310 (define-public ocaml-mcl
 311   (package
 312     (name "ocaml-mcl")
 313     (version "12-068oasis4")
 314     (source
 315      (origin
 316        (method git-fetch)
 317        (uri (git-reference
 318              (url "https://github.com/fhcrc/mcl.git")
 319              (commit version)))
 320        (file-name (git-file-name name version))
 321        (sha256
 322         (base32
 323          "0009dc3h2jp3qg5val452wngpqnbfyhbcxylghq0mrjqxx0jdq5p"))))
 324     (build-system ocaml-build-system)
 325     (arguments
 326      `(#:phases
 327        (modify-phases %standard-phases
 328          (add-before 'configure 'patch-paths
 329            (lambda _
 330              (substitute* "configure"
 331                (("/bin/sh") (which "sh")))
 332              (substitute* "setup.ml"
 333                (("LDFLAGS=-fPIC")
 334                 (string-append "LDFLAGS=-fPIC\"; \"SHELL=" (which "sh")))
 335                (("-std=c89") "-std=gnu99")
 336
 337                ;; This is a mutable string, which is no longer supported.  Use
 338                ;; a byte buffer instead.
 339                (("String.make \\(String.length s\\)")
 340                 "Bytes.make (String.length s)")
 341
 342                ;; These two belong together.
 343                (("OASISString.replace_chars")
 344                 "Bytes.to_string (OASISString.replace_chars")
 345                ((" s;")
 346                 " s);"))
 347              (substitute* "myocamlbuild.ml"
 348                (("std=c89") "std=gnu99"))
 349              ;; Since we build with a more recent OCaml, we have to use C99 or
 350              ;; later.  This causes problems with the old C code.
 351              (substitute* "src/impala/matrix.c"
 352                (("restrict") "restrict_"))
 353              #t)))))
 354     (native-inputs
 355      `(("ocamlbuild" ,ocamlbuild)))
 356     (home-page "https://github.com/fhcrc/mcl")
 357     (synopsis "OCaml wrappers around MCL")
 358     (description
 359      "This package provides OCaml bindings for the MCL graph clustering
 360 algorithm.")
 361     (license license:gpl3)))
 362
 363 (define-public randomjungle
 364   (package
 365     (name "randomjungle")
 366     (version "2.1.0")
 367     (source
 368      (origin
 369        (method url-fetch)
 370        (uri (string-append
 371              "https://www.imbs.uni-luebeck.de/fileadmin/files/Software"
 372              "/randomjungle/randomjungle-" version ".tar_.gz"))
 373        (patches (search-patches "randomjungle-disable-static-build.patch"))
 374        (sha256
 375         (base32
 376          "12c8rf30cla71swx2mf4ww9mfd8jbdw5lnxd7dxhyw1ygrvg6y4w"))))
 377     (build-system gnu-build-system)
 378     (arguments
 379      `(#:configure-flags
 380        (list "--disable-static"
 381              (string-append "--with-boost="
 382                             (assoc-ref %build-inputs "boost")))
 383        #:phases
 384        (modify-phases %standard-phases
 385          (add-before
 386           'configure 'set-CXXFLAGS
 387           (lambda _
 388             (setenv "CXXFLAGS" "-fpermissive ")
 389             #t)))))
 390     (inputs
 391      `(("boost" ,boost)
 392        ("gsl" ,gsl)
 393        ("libxml2" ,libxml2)
 394        ("zlib" ,zlib)))
 395     (native-inputs
 396      `(("gfortran" ,gfortran)
 397        ("gfortran:lib" ,gfortran "lib")))
 398     ;; Non-portable assembly instructions are used so building fails on
 399     ;; platforms other than x86_64 or i686.
 400     (supported-systems '("x86_64-linux" "i686-linux"))
 401     (home-page "https://www.imbs.uni-luebeck.de/forschung/software/details.html#c224")
 402     (synopsis "Implementation of the Random Forests machine learning method")
 403     (description
 404      "Random Jungle is an implementation of Random Forests.  It is supposed to
 405 analyse high dimensional data.  In genetics, it can be used for analysing big
 406 Genome Wide Association (GWA) data.  Random Forests is a powerful machine
 407 learning method.  Most interesting features are variable selection, missing
 408 value imputation, classifier creation, generalization error estimation and
 409 sample proximities between pairs of cases.")
 410     (license license:gpl3+)))
 411
 412 (define-public openfst
 413   (package
 414     (name "openfst")
 415     (version "1.7.2")
 416     (source (origin
 417               (method url-fetch)
 418               (uri (string-append "http://www.openfst.org/twiki/pub/FST/"
 419                                   "FstDownload/openfst-" version ".tar.gz"))
 420               (sha256
 421                (base32
 422                 "0fqgk8195kz21is09gwzwnrg7fr9526bi9mh4apyskapz27pbhr1"))))
 423     (build-system gnu-build-system)
 424     (home-page "http://www.openfst.org")
 425     (synopsis "Library for weighted finite-state transducers")
 426     (description "OpenFst is a library for constructing, combining,
 427 optimizing, and searching weighted finite-state transducers (FSTs).")
 428     (license license:asl2.0)))
 429
 430 (define-public shogun
 431   (package
 432     (name "shogun")
 433     (version "6.1.3")
 434     (source
 435      (origin
 436        (method url-fetch)
 437        (uri (string-append
 438              "ftp://shogun-toolbox.org/shogun/releases/"
 439              (version-major+minor version)
 440              "/sources/shogun-" version ".tar.bz2"))
 441        (sha256
 442         (base32
 443          "1rn9skm3nw6hr7mr3lgp2gfqhi7ii0lyxck7qmqnf8avq349s5jp"))
 444        (modules '((guix build utils)
 445                   (ice-9 rdelim)))
 446        (snippet
 447         '(begin
 448            ;; Remove non-free sources and files referencing them
 449            (for-each delete-file
 450                      (find-files "src/shogun/classifier/svm/"
 451                                  "SVMLight\\.(cpp|h)"))
 452            (for-each delete-file
 453                      (find-files "examples/undocumented/libshogun/"
 454                                  (string-append
 455                                   "(classifier_.*svmlight.*|"
 456                                   "evaluation_cross_validation_locked_comparison).cpp")))
 457            ;; Remove non-free functions.
 458            (define (delete-ifdefs file)
 459              (with-atomic-file-replacement file
 460                (lambda (in out)
 461                  (let loop ((line (read-line in 'concat))
 462                             (skipping? #f))
 463                    (if (eof-object? line)
 464                        #t
 465                        (let ((skip-next?
 466                               (or (and skipping?
 467                                        (not (string-prefix?
 468                                              "#endif //USE_SVMLIGHT" line)))
 469                                   (string-prefix?
 470                                    "#ifdef USE_SVMLIGHT" line))))
 471                          (when (or (not skipping?)
 472                                    (and skipping? (not skip-next?)))
 473                            (display line out))
 474                          (loop (read-line in 'concat) skip-next?)))))))
 475            (for-each delete-ifdefs
 476                      (append
 477                       (find-files "src/shogun/classifier/mkl"
 478                                   "^MKLClassification\\.cpp")
 479                       (find-files "src/shogun/classifier/svm"
 480                                   "^SVMLightOneClass\\.(cpp|h)")
 481                       (find-files "src/shogun/multiclass"
 482                                   "^ScatterSVM\\.(cpp|h)")
 483                       (find-files "src/shogun/kernel/"
 484                                   "^(Kernel|CombinedKernel|ProductKernel)\\.(cpp|h)")
 485                       (find-files "src/shogun/regression/svr"
 486                                   "^(MKLRegression|SVRLight)\\.(cpp|h)")
 487                       (find-files "src/shogun/transfer/domain_adaptation"
 488                                   "^DomainAdaptationSVM\\.(cpp|h)")))
 489            #t))))
 490     (build-system cmake-build-system)
 491     (arguments
 492      '(#:tests? #f ;no check target
 493        #:phases
 494        (modify-phases %standard-phases
 495          (add-after 'unpack 'delete-broken-symlinks
 496            (lambda _
 497              (for-each delete-file '("applications/arts/data"
 498                                      "applications/asp/data"
 499                                      "applications/easysvm/data"
 500                                      "applications/msplicer/data"
 501                                      "applications/ocr/data"
 502                                      "examples/meta/data"
 503                                      "examples/undocumented/data"))
 504              #t))
 505          (add-after 'unpack 'change-R-target-path
 506            (lambda* (#:key outputs #:allow-other-keys)
 507              (substitute* '("src/interfaces/r/CMakeLists.txt"
 508                             "examples/meta/r/CMakeLists.txt")
 509                (("\\$\\{R_COMPONENT_LIB_PATH\\}")
 510                 (string-append (assoc-ref outputs "out")
 511                                "/lib/R/library/")))
 512              #t))
 513          (add-after 'unpack 'fix-octave-modules
 514            (lambda* (#:key outputs #:allow-other-keys)
 515              (substitute* "src/interfaces/octave/CMakeLists.txt"
 516                (("^include_directories\\(\\$\\{OCTAVE_INCLUDE_DIRS\\}")
 517                 "include_directories(${OCTAVE_INCLUDE_DIRS} ${OCTAVE_INCLUDE_DIRS}/octave")
 518                ;; change target directory
 519                (("\\$\\{OCTAVE_OCT_LOCAL_API_FILE_DIR\\}")
 520                 (string-append (assoc-ref outputs "out")
 521                                "/share/octave/packages")))
 522              (substitute* '("src/interfaces/octave/swig_typemaps.i"
 523                             "src/interfaces/octave/sg_print_functions.cpp")
 524                ;; "octave/config.h" and "octave/oct-obj.h" deprecated in Octave.
 525                (("octave/config\\.h") "octave/octave-config.h")
 526                (("octave/oct-obj.h") "octave/ovl.h"))
 527              #t))
 528          (add-after 'unpack 'move-rxcpp
 529            (lambda* (#:key inputs #:allow-other-keys)
 530              (let ((rxcpp-dir "shogun/third-party/rxcpp"))
 531                (mkdir-p rxcpp-dir)
 532                (install-file (assoc-ref inputs "rxcpp") rxcpp-dir)
 533                #t)))
 534          (add-before 'build 'set-HOME
 535            ;; $HOME needs to be set at some point during the build phase
 536            (lambda _ (setenv "HOME" "/tmp") #t)))
 537        #:configure-flags
 538        (list "-DCMAKE_BUILD_WITH_INSTALL_RPATH=TRUE"
 539              "-DUSE_SVMLIGHT=OFF" ;disable proprietary SVMLIGHT
 540              "-DBUILD_META_EXAMPLES=OFF" ;requires unpackaged ctags
 541              ;;"-DINTERFACE_JAVA=ON" ;requires unpackaged jblas
 542              ;;"-DINTERFACE_RUBY=ON" ;requires unpackaged ruby-narray
 543              ;;"-DINTERFACE_PERL=ON" ;"FindPerlLibs" does not exist
 544              ;;"-DINTERFACE_LUA=ON"  ;fails because lua doesn't build pkgconfig file
 545              "-DINTERFACE_OCTAVE=ON"
 546              "-DINTERFACE_PYTHON=ON"
 547              "-DINTERFACE_R=ON")))
 548     (inputs
 549      `(("python" ,python)
 550        ("numpy" ,python-numpy)
 551        ("r-minimal" ,r-minimal)
 552        ("octave" ,octave-cli)
 553        ("swig" ,swig)
 554        ("eigen" ,eigen)
 555        ("hdf5" ,hdf5)
 556        ("atlas" ,atlas)
 557        ("arpack" ,arpack-ng)
 558        ("lapack" ,lapack)
 559        ("glpk" ,glpk)
 560        ("libxml2" ,libxml2)
 561        ("lzo" ,lzo)
 562        ("zlib" ,zlib)))
 563     (native-inputs
 564      `(("pkg-config" ,pkg-config)
 565        ("rxcpp" ,rxcpp)))
 566     ;; Non-portable SSE instructions are used so building fails on platforms
 567     ;; other than x86_64.
 568     (supported-systems '("x86_64-linux"))
 569     (home-page "http://shogun-toolbox.org/")
 570     (synopsis "Machine learning toolbox")
 571     (description
 572      "The Shogun Machine learning toolbox provides a wide range of unified and
 573 efficient Machine Learning (ML) methods.  The toolbox seamlessly allows to
 574 combine multiple data representations, algorithm classes, and general purpose
 575 tools.  This enables both rapid prototyping of data pipelines and extensibility
 576 in terms of new algorithms.")
 577     (license license:gpl3+)))
 578
 579 (define-public rxcpp
 580   (package
 581     (name "rxcpp")
 582     (version "4.1.0")
 583     (source
 584      (origin
 585        (method git-fetch)
 586        (uri (git-reference
 587              (url "https://github.com/ReactiveX/RxCpp.git")
 588              (commit (string-append "v" version))))
 589        (sha256
 590         (base32 "1rdpa3jlc181jd08nk437aar085h28i45s6nzrv65apb3xyyz0ij"))
 591        (file-name (git-file-name name version))))
 592     (build-system cmake-build-system)
 593     (arguments
 594      `(#:phases
 595        (modify-phases %standard-phases
 596          (add-after 'unpack 'remove-werror
 597            (lambda _
 598              (substitute* (find-files ".")
 599                (("-Werror") ""))
 600              #t))
 601          (replace 'check
 602            (lambda _
 603              (invoke "ctest"))))))
 604     (native-inputs
 605      `(("catch" ,catch-framework)))
 606     (home-page "http://reactivex.io/")
 607     (synopsis "Reactive Extensions for C++")
 608     (description
 609      "The Reactive Extensions for C++ (RxCpp) is a library of algorithms for
 610 values-distributed-in-time.  ReactiveX is a library for composing asynchronous
 611 and event-based programs by using observable sequences.
 612
 613 It extends the observer pattern to support sequences of data and/or events and
 614 adds operators that allow you to compose sequences together declaratively while
 615 abstracting away concerns about things like low-level threading,
 616 synchronization, thread-safety, concurrent data structures, and non-blocking
 617 I/O.")
 618     (license license:asl2.0)))
 619
 620 (define-public r-adaptivesparsity
 621   (package
 622     (name "r-adaptivesparsity")
 623     (version "1.6")
 624     (source (origin
 625               (method url-fetch)
 626               (uri (cran-uri "AdaptiveSparsity" version))
 627               (sha256
 628                (base32
 629                 "0imr5m8mll9j6n4icsv6z9rl5kbnwsp9wvzrg7n90nnmcxq2cz91"))))
 630     (properties
 631      `((upstream-name . "AdaptiveSparsity")))
 632     (build-system r-build-system)
 633     (arguments
 634      `(#:phases
 635        (modify-phases %standard-phases
 636          (add-after 'unpack 'link-against-armadillo
 637            (lambda _
 638              (substitute* "src/Makevars"
 639                (("PKG_LIBS=" prefix)
 640                 (string-append prefix "-larmadillo"))))))))
 641     (propagated-inputs
 642      `(("r-mass" ,r-mass)
 643        ("r-matrix" ,r-matrix)
 644        ("r-rcpp" ,r-rcpp)
 645        ("r-rcpparmadillo" ,r-rcpparmadillo)))
 646     (inputs
 647      `(("armadillo" ,armadillo)))
 648     (home-page "https://cran.r-project.org/web/packages/AdaptiveSparsity")
 649     (synopsis "Adaptive sparsity models")
 650     (description
 651      "This package implements the Figueiredo machine learning algorithm for
 652 adaptive sparsity and the Wong algorithm for adaptively sparse gaussian
 653 geometric models.")
 654     (license license:lgpl3+)))
 655
 656 (define-public gemmlowp-for-tensorflow
 657   ;; The commit hash is taken from "tensorflow/workspace.bzl".
 658   (let ((commit "38ebac7b059e84692f53e5938f97a9943c120d98")
 659         (revision "2"))
 660     (package
 661       (name "gemmlowp")
 662       (version (git-version "0" revision commit))
 663       (source (origin
 664                 (method url-fetch)
 665                 (uri (string-append "https://mirror.bazel.build/"
 666                                     "github.com/google/gemmlowp/archive/"
 667                                     commit ".zip"))
 668                 (file-name (string-append "gemmlowp-" version ".zip"))
 669                 (sha256
 670                  (base32
 671                   "0n56s2g8hrssm4w8qj1v58gfm56a04n9v992ixkmvk6zjiralzxq"))))
 672       (build-system cmake-build-system)
 673       (arguments
 674        `(#:configure-flags
 675          (list ,@(match (%current-system)
 676                    ((or "x86_64-linux" "i686-linux")
 677                     '("-DCMAKE_CXX_FLAGS=-msse2"))
 678                    (_ '())))
 679          #:phases
 680          (modify-phases %standard-phases
 681            ;; This directory contains the CMakeLists.txt.
 682            (add-after 'unpack 'chdir
 683              (lambda _ (chdir "contrib") #t))
 684            ;; There is no install target
 685            (replace 'install
 686              (lambda* (#:key outputs #:allow-other-keys)
 687                (let* ((out (assoc-ref outputs "out"))
 688                       (lib (string-append out "/lib/"))
 689                       (inc (string-append out "/include/")))
 690                  (install-file "../build/libeight_bit_int_gemm.so" lib)
 691                  (for-each (lambda (dir)
 692                              (let ((target (string-append inc "/" dir)))
 693                                (mkdir-p target)
 694                                (for-each (lambda (h)
 695                                            (install-file h target))
 696                                          (find-files (string-append "../" dir)
 697                                                      "\\.h$"))))
 698                            '("meta" "profiling" "public" "fixedpoint"
 699                              "eight_bit_int_gemm" "internal"))
 700                  #t))))))
 701       (native-inputs
 702        `(("unzip" ,unzip)))
 703       (home-page "https://github.com/google/gemmlowp")
 704       (synopsis "Small self-contained low-precision GEMM library")
 705       (description
 706        "This is a small self-contained low-precision @dfn{general matrix
 707 multiplication} (GEMM) library.  It is not a full linear algebra library.
 708 Low-precision means that the input and output matrix entries are integers on
 709 at most 8 bits.  To avoid overflow, results are internally accumulated on more
 710 than 8 bits, and at the end only some significant 8 bits are kept.")
 711       (license license:asl2.0))))
 712
 713 (define-public dlib
 714   (package
 715     (name "dlib")
 716     (version "19.7")
 717     (source (origin
 718               (method url-fetch)
 719               (uri (string-append
 720                     "http://dlib.net/files/dlib-" version ".tar.bz2"))
 721               (sha256
 722                (base32
 723                 "1mljz02kwkrbggyncxv5fpnyjdybw2qihaacb3js8yfkw12vwpc2"))
 724               (modules '((guix build utils)))
 725               (snippet
 726                '(begin
 727                   ;; Delete ~13MB of bundled dependencies.
 728                   (delete-file-recursively "dlib/external")
 729                   (delete-file-recursively "docs/dlib/external")
 730                   #t))))
 731     (build-system cmake-build-system)
 732     (arguments
 733      `(#:phases
 734        (modify-phases %standard-phases
 735          (add-after 'unpack 'disable-asserts
 736            (lambda _
 737              ;; config.h recommends explicitly enabling or disabling asserts
 738              ;; when building as a shared library. By default neither is set.
 739              (substitute* "dlib/config.h"
 740                (("^//#define DLIB_DISABLE_ASSERTS") "#define DLIB_DISABLE_ASSERTS"))
 741              #t))
 742          (add-after 'disable-asserts 'disable-failing-tests
 743            (lambda _
 744              ;; One test times out on MIPS, so we need to disable it.
 745              ;; Others are flaky on some platforms.
 746              (let* ((system ,(or (%current-target-system)
 747                                  (%current-system)))
 748                     (disabled-tests (cond
 749                                      ((string-prefix? "mips64" system)
 750                                       '("object_detector" ; timeout
 751                                         "data_io"))
 752                                      ((string-prefix? "armhf" system)
 753                                       '("learning_to_track"))
 754                                      ((string-prefix? "i686" system)
 755                                       '("optimization"))
 756                                      (else '()))))
 757                (for-each
 758                 (lambda (test)
 759                   (substitute* "dlib/test/makefile"
 760                     (((string-append "SRC \\+= " test "\\.cpp")) "")))
 761                 disabled-tests)
 762                #t)))
 763          (replace 'check
 764            (lambda _
 765              ;; No test target, so we build and run the unit tests here.
 766              (let ((test-dir (string-append "../dlib-" ,version "/dlib/test")))
 767                (with-directory-excursion test-dir
 768                  (invoke "make" "-j" (number->string (parallel-job-count)))
 769                  (invoke "./dtest" "--runall"))
 770                #t)))
 771          (add-after 'install 'delete-static-library
 772            (lambda* (#:key outputs #:allow-other-keys)
 773              (delete-file (string-append (assoc-ref outputs "out")
 774                                          "/lib/libdlib.a"))
 775              #t)))))
 776     (native-inputs
 777      `(("pkg-config" ,pkg-config)
 778        ;; For tests.
 779        ("libnsl" ,libnsl)))
 780     (inputs
 781      `(("giflib" ,giflib)
 782        ("lapack" ,lapack)
 783        ("libjpeg" ,libjpeg-turbo)
 784        ("libpng" ,libpng)
 785        ("libx11" ,libx11)
 786        ("openblas" ,openblas)
 787        ("zlib" ,zlib)))
 788     (synopsis
 789      "Toolkit for making machine learning and data analysis applications in C++")
 790     (description
 791      "Dlib is a modern C++ toolkit containing machine learning algorithms and
 792 tools.  It is used in both industry and academia in a wide range of domains
 793 including robotics, embedded devices, mobile phones, and large high performance
 794 computing environments.")
 795     (home-page "http://dlib.net")
 796     (license license:boost1.0)))
 797
 798 (define-public python-scikit-learn
 799   (package
 800     (name "python-scikit-learn")
 801     (version "0.20.4")
 802     (source
 803      (origin
 804        (method git-fetch)
 805        (uri (git-reference
 806              (url "https://github.com/scikit-learn/scikit-learn.git")
 807              (commit version)))
 808        (file-name (git-file-name name version))
 809        (sha256
 810         (base32
 811          "08zbzi8yx5wdlxfx9jap61vg1malc9ajf576w7a0liv6jvvrxlpj"))))
 812     (build-system python-build-system)
 813     (arguments
 814      `(#:phases
 815        (modify-phases %standard-phases
 816          (add-after 'build 'build-ext
 817            (lambda _ (invoke "python" "setup.py" "build_ext" "--inplace") #t))
 818          (replace 'check
 819            (lambda _
 820              ;; Restrict OpenBLAS threads to prevent segfaults while testing!
 821              (setenv "OPENBLAS_NUM_THREADS" "1")
 822
 823              ;; Some tests require write access to $HOME.
 824              (setenv "HOME" "/tmp")
 825
 826              (invoke "pytest" "sklearn" "-m" "not network")))
 827          (add-before 'reset-gzip-timestamps 'make-files-writable
 828            (lambda* (#:key outputs #:allow-other-keys)
 829              ;; Make sure .gz files are writable so that the
 830              ;; 'reset-gzip-timestamps' phase can do its work.
 831              (let ((out (assoc-ref outputs "out")))
 832                (for-each make-file-writable
 833                          (find-files out "\\.gz$"))
 834                #t))))))
 835     (inputs
 836      `(("openblas" ,openblas)))
 837     (native-inputs
 838      `(("python-pytest" ,python-pytest)
 839        ("python-pandas" ,python-pandas) ;for tests
 840        ("python-cython" ,python-cython)))
 841     (propagated-inputs
 842      `(("python-numpy" ,python-numpy)
 843        ("python-scipy" ,python-scipy)))
 844     (home-page "http://scikit-learn.org/")
 845     (synopsis "Machine Learning in Python")
 846     (description
 847      "Scikit-learn provides simple and efficient tools for data mining and
 848 data analysis.")
 849     (license license:bsd-3)))
 850
 851 (define-public python2-scikit-learn
 852   (package-with-python2 python-scikit-learn))
 853
 854 (define-public python-autograd
 855   (let* ((commit "442205dfefe407beffb33550846434baa90c4de7")
 856          (revision "0")
 857          (version (git-version "0.0.0" revision commit)))
 858     (package
 859       (name "python-autograd")
 860       (home-page "https://github.com/HIPS/autograd")
 861       (source (origin
 862                 (method git-fetch)
 863                 (uri (git-reference
 864                       (url home-page)
 865                       (commit commit)))
 866                 (sha256
 867                  (base32
 868                   "189sv2xb0mwnjawa9z7mrgdglc1miaq93pnck26r28fi1jdwg0z4"))
 869                 (file-name (git-file-name name version))))
 870       (version version)
 871       (build-system python-build-system)
 872       (native-inputs
 873        `(("python-nose" ,python-nose)
 874          ("python-pytest" ,python-pytest)))
 875       (propagated-inputs
 876        `(("python-future" ,python-future)
 877          ("python-numpy" ,python-numpy)))
 878       (arguments
 879        `(#:phases (modify-phases %standard-phases
 880                     (replace 'check
 881                       (lambda _
 882                         (invoke "py.test" "-v"))))))
 883       (synopsis "Efficiently computes derivatives of NumPy code")
 884       (description "Autograd can automatically differentiate native Python and
 885 NumPy code.  It can handle a large subset of Python's features, including loops,
 886 ifs, recursion and closures, and it can even take derivatives of derivatives
 887 of derivatives.  It supports reverse-mode differentiation
 888 (a.k.a. backpropagation), which means it can efficiently take gradients of
 889 scalar-valued functions with respect to array-valued arguments, as well as
 890 forward-mode differentiation, and the two can be composed arbitrarily.  The
 891 main intended application of Autograd is gradient-based optimization.")
 892       (license license:expat))))
 893
 894 (define-public python2-autograd
 895   (package-with-python2 python-autograd))
 896
 897 (define-public lightgbm
 898   (package
 899     (name "lightgbm")
 900     (version "2.0.12")
 901     (source (origin
 902               (method url-fetch)
 903               (uri (string-append
 904                     "https://github.com/Microsoft/LightGBM/archive/v"
 905                     version ".tar.gz"))
 906               (sha256
 907                (base32
 908                 "132zf0yk0545mg72hyzxm102g3hpb6ixx9hnf8zd2k55gas6cjj1"))
 909               (file-name (string-append name "-" version ".tar.gz"))))
 910     (native-inputs
 911      `(("python-pytest" ,python-pytest)
 912        ("python-nose" ,python-nose)))
 913     (inputs
 914      `(("openmpi" ,openmpi)))
 915     (propagated-inputs
 916      `(("python-numpy" ,python-numpy)
 917        ("python-scipy" ,python-scipy)))
 918     (arguments
 919      `(#:configure-flags
 920        '("-DUSE_MPI=ON")
 921        #:phases
 922        (modify-phases %standard-phases
 923          (replace 'check
 924            (lambda* (#:key outputs #:allow-other-keys)
 925              (with-directory-excursion ,(string-append "../LightGBM-" version)
 926                (invoke "pytest" "tests/c_api_test/test_.py")))))))
 927     (build-system cmake-build-system)
 928     (home-page "https://github.com/Microsoft/LightGBM")
 929     (synopsis "Gradient boosting framework based on decision tree algorithms")
 930     (description "LightGBM is a gradient boosting framework that uses tree
 931 based learning algorithms.  It is designed to be distributed and efficient with
 932 the following advantages:
 933
 934 @itemize
 935 @item Faster training speed and higher efficiency
 936 @item Lower memory usage
 937 @item Better accuracy
 938 @item Parallel and GPU learning supported (not enabled in this package)
 939 @item Capable of handling large-scale data
 940 @end itemize\n")
 941     (license license:expat)))
 942
 943 (define-public vowpal-wabbit
 944   ;; Language bindings not included.
 945   (package
 946     (name "vowpal-wabbit")
 947     (version "8.5.0")
 948     (source (origin
 949               (method url-fetch)
 950               (uri (string-append
 951                     "https://github.com/JohnLangford/vowpal_wabbit/archive/"
 952                     version ".tar.gz"))
 953               (sha256
 954                (base32
 955                 "0clp2kb7rk5sckhllxjr5a651awf4s8dgzg4659yh4hf5cqnf0gr"))
 956               (file-name (string-append name "-" version ".tar.gz"))))
 957     (inputs
 958      `(("boost" ,boost)
 959        ("zlib" ,zlib)))
 960     (arguments
 961      `(#:configure-flags
 962        (list (string-append "--with-boost="
 963                             (assoc-ref %build-inputs "boost")))))
 964     (build-system gnu-build-system)
 965     (home-page "https://github.com/JohnLangford/vowpal_wabbit")
 966     (synopsis "Fast machine learning library for online learning")
 967     (description "Vowpal Wabbit is a machine learning system with techniques
 968 such as online, hashing, allreduce, reductions, learning2search, active, and
 969 interactive learning.")
 970     (license license:bsd-3)))
 971
 972 (define-public python2-fastlmm
 973   (package
 974     (name "python2-fastlmm")
 975     (version "0.2.21")
 976     (source
 977      (origin
 978        (method url-fetch)
 979        (uri (pypi-uri "fastlmm" version ".zip"))
 980        (sha256
 981         (base32
 982          "1q8c34rpmwkfy3r4d5172pzdkpfryj561897z9r3x22gq7813x1m"))))
 983     (build-system python-build-system)
 984     (arguments
 985      `(#:tests? #f ; some test files are missing
 986        #:python ,python-2)) ; only Python 2.7 is supported
 987     (propagated-inputs
 988      `(("python2-numpy" ,python2-numpy)
 989        ("python2-scipy" ,python2-scipy)
 990        ("python2-matplotlib" ,python2-matplotlib)
 991        ("python2-pandas" ,python2-pandas)
 992        ("python2-scikit-learn" ,python2-scikit-learn)
 993        ("python2-pysnptools" ,python2-pysnptools)))
 994     (native-inputs
 995      `(("unzip" ,unzip)
 996        ("python2-cython" ,python2-cython)
 997        ("python2-mock" ,python2-mock)
 998        ("python2-nose" ,python2-nose)))
 999     (home-page "http://research.microsoft.com/en-us/um/redmond/projects/mscompbio/fastlmm/")
1000     (synopsis "Perform genome-wide association studies on large data sets")
1001     (description
1002      "FaST-LMM, which stands for Factored Spectrally Transformed Linear Mixed
1003 Models, is a program for performing both single-SNP and SNP-set genome-wide
1004 association studies (GWAS) on extremely large data sets.")
1005     (license license:asl2.0)))
1006
1007 ;; There have been no proper releases yet.
1008 (define-public kaldi
1009   (let ((commit "2f95609f0bb085bd3a1dc5eb0a39f3edea59e606")
1010         (revision "1"))
1011     (package
1012       (name "kaldi")
1013       (version (git-version "0" revision commit))
1014       (source (origin
1015                 (method git-fetch)
1016                 (uri (git-reference
1017                       (url "https://github.com/kaldi-asr/kaldi.git")
1018                       (commit commit)))
1019                 (file-name (git-file-name name version))
1020                 (sha256
1021                  (base32
1022                   "082qh3pfi7hvncylp4xsmkfahbd7gb0whdfa4rwrx7fxk9rdh3kz"))))
1023       (build-system gnu-build-system)
1024       (arguments
1025        `(#:test-target "test"
1026          #:phases
1027          (modify-phases %standard-phases
1028            (add-after 'unpack 'chdir
1029              (lambda _ (chdir "src") #t))
1030            (replace 'configure
1031              (lambda* (#:key build system inputs outputs #:allow-other-keys)
1032                (when (not (or (string-prefix? "x86_64" system)
1033                               (string-prefix? "i686" system)))
1034                  (substitute* "makefiles/linux_openblas.mk"
1035                    (("-msse -msse2") "")))
1036                (substitute* "makefiles/default_rules.mk"
1037                  (("/bin/bash") (which "bash")))
1038                (substitute* "Makefile"
1039                  (("ext_depend: check_portaudio")
1040                   "ext_depend:"))
1041                (substitute* '("online/Makefile"
1042                               "onlinebin/Makefile"
1043                               "gst-plugin/Makefile")
1044                  (("../../tools/portaudio/install")
1045                   (assoc-ref inputs "portaudio")))
1046
1047                ;; This `configure' script doesn't support variables passed as
1048                ;; arguments, nor does it support "prefix".
1049                (let ((out (assoc-ref outputs "out"))
1050                      (openblas (assoc-ref inputs "openblas"))
1051                      (openfst (assoc-ref inputs "openfst")))
1052                  (substitute* "configure"
1053                    (("check_for_slow_expf;") "")
1054                    ;; This affects the RPATH and also serves as the installation
1055                    ;; directory.
1056                    (("KALDILIBDIR=`pwd`/lib")
1057                     (string-append "KALDILIBDIR=" out "/lib")))
1058                  (mkdir-p out) ; must exist
1059                  (setenv "CONFIG_SHELL" (which "bash"))
1060                  (setenv "OPENFST_VER" ,(package-version openfst))
1061                  (invoke "./configure"
1062                          "--use-cuda=no"
1063                          "--shared"
1064                          (string-append "--openblas-root=" openblas)
1065                          (string-append "--fst-root=" openfst)))))
1066            (add-after 'build 'build-ext-and-gstreamer-plugin
1067              (lambda _
1068                (invoke "make" "-C" "online" "depend")
1069                (invoke "make" "-C" "online")
1070                (invoke "make" "-C" "onlinebin" "depend")
1071                (invoke "make" "-C" "onlinebin")
1072                (invoke "make" "-C" "gst-plugin" "depend")
1073                (invoke "make" "-C" "gst-plugin")
1074                #t))
1075            ;; TODO: also install the executables.
1076            (replace 'install
1077              (lambda* (#:key outputs #:allow-other-keys)
1078                (let* ((out (assoc-ref outputs "out"))
1079                       (inc (string-append out "/include"))
1080                       (lib (string-append out "/lib")))
1081                  (mkdir-p lib)
1082                  ;; The build phase installed symlinks to the actual
1083                  ;; libraries.  Install the actual targets.
1084                  (for-each (lambda (file)
1085                              (let ((target (readlink file)))
1086                                (delete-file file)
1087                                (install-file target lib)))
1088                            (find-files lib "\\.so"))
1089                  ;; Install headers
1090                  (for-each (lambda (file)
1091                              (let ((target-dir (string-append inc "/" (dirname file))))
1092                                (install-file file target-dir)))
1093                            (find-files "." "\\.h"))
1094                  (install-file "gst-plugin/libgstonlinegmmdecodefaster.so"
1095                                (string-append lib "/gstreamer-1.0"))
1096                  #t))))))
1097       (inputs
1098        `(("alsa-lib" ,alsa-lib)
1099          ("gfortran" ,gfortran "lib")
1100          ("glib" ,glib)
1101          ("gstreamer" ,gstreamer)
1102          ("jack" ,jack-1)
1103          ("openblas" ,openblas)
1104          ("openfst" ,openfst)
1105          ("portaudio" ,portaudio)
1106          ("python" ,python)))
1107       (native-inputs
1108        `(("glib" ,glib "bin")             ; glib-genmarshal
1109          ("grep" ,grep)
1110          ("sed" ,sed)
1111          ("pkg-config" ,pkg-config)
1112          ("which" ,which)))
1113       (home-page "https://kaldi-asr.org/")
1114       (synopsis "Speech recognition toolkit")
1115       (description "Kaldi is an extensible toolkit for speech recognition
1116 written in C++.")
1117       (license license:asl2.0))))
1118
1119 (define-public gst-kaldi-nnet2-online
1120   (let ((commit "617e43e73c7cc45eb9119028c02bd4178f738c4a")
1121         (revision "1"))
1122     (package
1123       (name "gst-kaldi-nnet2-online")
1124       (version (git-version "0" revision commit))
1125       (source (origin
1126                 (method git-fetch)
1127                 (uri (git-reference
1128                       (url "https://github.com/alumae/gst-kaldi-nnet2-online.git")
1129                       (commit commit)))
1130                 (file-name (git-file-name name version))
1131                 (sha256
1132                  (base32
1133                   "0xh3w67b69818s6ib02ara4lw7wamjdmh4jznvkpzrs4skbs9jx9"))))
1134       (build-system gnu-build-system)
1135       (arguments
1136        `(#:tests? #f                    ; there are none
1137          #:make-flags
1138          (list (string-append "SHELL="
1139                               (assoc-ref %build-inputs "bash") "/bin/bash")
1140                (string-append "KALDI_ROOT="
1141                               (assoc-ref %build-inputs "kaldi-src"))
1142                (string-append "KALDILIBDIR="
1143                               (assoc-ref %build-inputs "kaldi") "/lib")
1144                "KALDI_FLAVOR=dynamic")
1145          #:phases
1146          (modify-phases %standard-phases
1147            (add-after 'unpack 'chdir
1148              (lambda _ (chdir "src") #t))
1149            (replace 'configure
1150              (lambda* (#:key inputs #:allow-other-keys)
1151                (let ((glib (assoc-ref inputs "glib")))
1152                  (setenv "CXXFLAGS" "-fPIC")
1153                  (setenv "CPLUS_INCLUDE_PATH"
1154                          (string-append glib "/include/glib-2.0:"
1155                                         glib "/lib/glib-2.0/include:"
1156                                         (assoc-ref inputs "gstreamer")
1157                                         "/include/gstreamer-1.0")))
1158                (substitute* "Makefile"
1159                  (("include \\$\\(KALDI_ROOT\\)/src/kaldi.mk") "")
1160                  (("\\$\\(error Cannot find") "#"))
1161                #t))
1162            (add-before 'build 'build-depend
1163              (lambda* (#:key make-flags #:allow-other-keys)
1164                (apply invoke "make" "depend" make-flags)))
1165            (replace 'install
1166              (lambda* (#:key outputs #:allow-other-keys)
1167                (let* ((out (assoc-ref outputs "out"))
1168                       (lib (string-append out "/lib/gstreamer-1.0")))
1169                  (install-file "libgstkaldinnet2onlinedecoder.so" lib)
1170                  #t))))))
1171       (inputs
1172        `(("glib" ,glib)
1173          ("gstreamer" ,gstreamer)
1174          ("jansson" ,jansson)
1175          ("openfst" ,openfst)
1176          ("kaldi" ,kaldi)))
1177       (native-inputs
1178        `(("bash" ,bash)
1179          ("glib:bin" ,glib "bin")       ; glib-genmarshal
1180          ("kaldi-src" ,(package-source kaldi))
1181          ("pkg-config" ,pkg-config)))
1182       (home-page "https://kaldi-asr.org/")
1183       (synopsis "Gstreamer plugin for decoding speech")
1184       (description "This package provides a GStreamer plugin that wraps
1185 Kaldi's @code{SingleUtteranceNnet2Decoder}.  It requires iVector-adapted DNN
1186 acoustic models.  The iVectors are adapted to the current audio stream
1187 automatically.")
1188       (license license:asl2.0))))
1189
1190 (define-public kaldi-gstreamer-server
1191   (let ((commit "1735ba49c5dc0ebfc184e45105fc600cd9f1f508")
1192         (revision "1"))
1193     (package
1194       (name "kaldi-gstreamer-server")
1195       (version (git-version "0" revision commit))
1196       (source (origin
1197                 (method git-fetch)
1198                 (uri (git-reference
1199                       (url "https://github.com/alumae/kaldi-gstreamer-server.git")
1200                       (commit commit)))
1201                 (file-name (git-file-name name version))
1202                 (sha256
1203                  (base32
1204                   "0j701m7lbwmzqxsfanj882v7881hrbmpqybbczbxqpcbg8q34w0k"))))
1205       (build-system gnu-build-system)
1206       (arguments
1207        `(#:tests? #f ; there are no tests that can be run automatically
1208          #:modules ((guix build utils)
1209                     (guix build gnu-build-system)
1210                     (srfi srfi-26))
1211          #:phases
1212          (modify-phases %standard-phases
1213            (delete 'configure)
1214            (replace 'build
1215              (lambda* (#:key outputs #:allow-other-keys)
1216                ;; Disable hash randomization to ensure the generated .pycs
1217                ;; are reproducible.
1218                (setenv "PYTHONHASHSEED" "0")
1219                (with-directory-excursion "kaldigstserver"
1220                  (for-each (lambda (file)
1221                              (apply invoke
1222                                     `("python"
1223                                       "-m" "compileall"
1224                                       "-f" ; force rebuild
1225                                       ,file)))
1226                            (find-files "." "\\.py$")))
1227                #t))
1228            (replace 'install
1229              (lambda* (#:key inputs outputs #:allow-other-keys)
1230                (let* ((out (assoc-ref outputs "out"))
1231                       (bin (string-append out "/bin"))
1232                       (share (string-append out "/share/kaldi-gstreamer-server/")))
1233                  ;; Install Python files
1234                  (with-directory-excursion "kaldigstserver"
1235                    (for-each (cut install-file <> share)
1236                              (find-files "." ".*")))
1237
1238                  ;; Install sample configuration files
1239                  (for-each (cut install-file <> share)
1240                            (find-files "." "\\.yaml"))
1241
1242                  ;; Install executables
1243                  (mkdir-p bin)
1244                  (let* ((server (string-append bin "/kaldi-gst-server"))
1245                         (client (string-append bin "/kaldi-gst-client"))
1246                         (worker (string-append bin "/kaldi-gst-worker"))
1247                         (PYTHONPATH (getenv "PYTHONPATH"))
1248                         (GST_PLUGIN_PATH (string-append
1249                                           (assoc-ref inputs "gst-kaldi-nnet2-online")
1250                                           "/lib/gstreamer-1.0:${GST_PLUGIN_PATH}"))
1251                         (wrap (lambda (wrapper what)
1252                                 (with-output-to-file wrapper
1253                                   (lambda _
1254                                     (format #t
1255                                             "#!~a
1256 export PYTHONPATH=~a
1257 export GST_PLUGIN_PATH=~a
1258 exec ~a ~a/~a \"$@\"~%"
1259                                             (which "bash") PYTHONPATH GST_PLUGIN_PATH
1260                                             (which "python") share what)))
1261                                 (chmod wrapper #o555))))
1262                    (for-each wrap
1263                              (list server client worker)
1264                              (list "master_server.py"
1265                                    "client.py"
1266                                    "worker.py")))
1267                  #t))))))
1268       (inputs
1269        `(("gst-kaldi-nnet2-online" ,gst-kaldi-nnet2-online)
1270          ("python2" ,python-2)
1271          ("python2-futures" ,python2-futures)
1272          ("python2-pygobject" ,python2-pygobject)
1273          ("python2-pyyaml" ,python2-pyyaml)
1274          ("python2-tornado" ,python2-tornado)
1275          ("python2-ws4py" ,python2-ws4py-for-kaldi-gstreamer-server)))
1276       (home-page "https://github.com/alumae/kaldi-gstreamer-server")
1277       (synopsis "Real-time full-duplex speech recognition server")
1278       (description "This is a real-time full-duplex speech recognition server,
1279 based on the Kaldi toolkit and the GStreamer framework and implemented in
1280 Python.")
1281       (license license:bsd-2))))
1282
1283 (define-public grpc
1284   (package
1285     (name "grpc")
1286     (version "1.16.1")
1287     (source (origin
1288               (method git-fetch)
1289               (uri (git-reference
1290                     (url "https://github.com/grpc/grpc.git")
1291                     (commit (string-append "v" version))))
1292               (file-name (git-file-name name version))
1293               (sha256
1294                (base32
1295                 "1jimqz3115f9pli5w6ik9wi7mjc7ix6y7yrq4a1ab9fc3dalj7p2"))))
1296     (build-system cmake-build-system)
1297     (arguments
1298      `(#:tests? #f ; no test target
1299        #:configure-flags
1300        (list "-DgRPC_ZLIB_PROVIDER=package"
1301              "-DgRPC_CARES_PROVIDER=package"
1302              "-DgRPC_SSL_PROVIDER=package"
1303              "-DgRPC_PROTOBUF_PROVIDER=package")))
1304     (inputs
1305      `(("c-ares" ,c-ares/cmake)
1306        ("openssl" ,openssl)
1307        ("zlib" ,zlib)))
1308     (native-inputs
1309      `(("protobuf" ,protobuf)
1310        ("python" ,python-wrapper)))
1311     (home-page "https://grpc.io")
1312     (synopsis "High performance universal RPC framework")
1313     (description "gRPC is a modern high performance @dfn{Remote Procedure Call}
1314 (RPC) framework that can run in any environment.  It can efficiently connect
1315 services in and across data centers with pluggable support for load balancing,
1316 tracing, health checking and authentication.  It is also applicable in last
1317 mile of distributed computing to connect devices, mobile applications and
1318 browsers to backend services.")
1319     (license license:asl2.0)))
1320
1321 ;; Note that Tensorflow includes a "third_party" directory, which seems to not
1322 ;; only contain modified subsets of upstream library source code, but also
1323 ;; adapter headers provided by Google (such as the fft.h header, which is not
1324 ;; part of the upstream project code).  The Tensorflow code includes headers
1325 ;; from the "third_party" directory.  It does not look like we can replace
1326 ;; these headers with unmodified upstream files, so we keep them.
1327 (define-public tensorflow
1328   (package
1329     (name "tensorflow")
1330     (version "1.9.0")
1331     (source
1332      (origin
1333        (method git-fetch)
1334        (uri (git-reference
1335              (url "https://github.com/tensorflow/tensorflow.git")
1336              (commit (string-append "v" version))))
1337        (file-name (string-append "tensorflow-" version "-checkout"))
1338        (sha256
1339         (base32
1340          "0a9kwha395g3wgxfwln5j8vn9nkspmd75xldrlqdq540w996g8xa"))))
1341     (build-system cmake-build-system)
1342     (arguments
1343      `(#:tests? #f                      ; no "check" target
1344        #:build-type "Release"
1345        #:configure-flags
1346        (let ((protobuf (assoc-ref %build-inputs "protobuf"))
1347              (protobuf:native (assoc-ref %build-inputs "protobuf:native"))
1348              (jsoncpp (assoc-ref %build-inputs "jsoncpp"))
1349              (snappy (assoc-ref %build-inputs "snappy"))
1350              (sqlite (assoc-ref %build-inputs "sqlite")))
1351          (list
1352           ;; Use protobuf from Guix
1353           (string-append "-Dprotobuf_STATIC_LIBRARIES="
1354                          protobuf "/lib/libprotobuf.so")
1355           (string-append "-DPROTOBUF_PROTOC_EXECUTABLE="
1356                          protobuf:native "/bin/protoc")
1357
1358           ;; Use snappy from Guix
1359           (string-append "-Dsnappy_STATIC_LIBRARIES="
1360                          snappy "/lib/libsnappy.so")
1361           ;; Yes, this is not actually the include directory but a prefix...
1362           (string-append "-Dsnappy_INCLUDE_DIR=" snappy)
1363
1364           ;; Use jsoncpp from Guix
1365           (string-append "-Djsoncpp_STATIC_LIBRARIES="
1366                          jsoncpp "/lib/libjsoncpp.so")
1367           ;; Yes, this is not actually the include directory but a prefix...
1368           (string-append "-Djsoncpp_INCLUDE_DIR=" jsoncpp)
1369
1370           ;; Use sqlite from Guix
1371           (string-append "-Dsqlite_STATIC_LIBRARIES="
1372                          sqlite "/lib/libsqlite.a")
1373
1374           ;; Use system libraries wherever possible.  Currently, this
1375           ;; only affects zlib.
1376           "-Dsystemlib_ALL=ON"
1377           "-Dtensorflow_ENABLE_POSITION_INDEPENDENT_CODE=ON"
1378           "-Dtensorflow_BUILD_SHARED_LIB=ON"
1379           "-Dtensorflow_OPTIMIZE_FOR_NATIVE_ARCH=OFF"
1380           "-Dtensorflow_ENABLE_SSL_SUPPORT=OFF"
1381           "-Dtensorflow_BUILD_CONTRIB_KERNELS=OFF"))
1382        #:make-flags
1383        (list "CC=gcc")
1384        #:modules ((ice-9 ftw)
1385                   (guix build utils)
1386                   (guix build cmake-build-system))
1387        #:phases
1388        (modify-phases %standard-phases
1389          (add-after 'unpack 'set-source-file-times-to-1980
1390            ;; At the end of the tf_python_build_pip_package target, a ZIP
1391            ;; archive should be generated via bdist_wheel, but it fails with
1392            ;; "ZIP does not support timestamps before 1980".  Luckily,
1393            ;; SOURCE_DATE_EPOCH is respected, which we set to some time in
1394            ;; 1980.
1395            (lambda _ (setenv "SOURCE_DATE_EPOCH" "315532800") #t))
1396          ;; See https://github.com/tensorflow/tensorflow/issues/20517#issuecomment-406373913
1397          (add-after 'unpack 'python3.7-compatibility
1398            (lambda _
1399              (substitute* '("tensorflow/python/eager/pywrap_tfe_src.cc"
1400                             "tensorflow/python/lib/core/ndarray_tensor.cc"
1401                             "tensorflow/python/lib/core/py_func.cc")
1402                (("PyUnicode_AsUTF8") "(char *)PyUnicode_AsUTF8"))
1403              (substitute* "tensorflow/c/eager/c_api.h"
1404                (("unsigned char async")
1405                 "unsigned char is_async"))
1406
1407              ;; Remove dependency on tensorboard, a complicated but probably
1408              ;; optional package.
1409              (substitute* "tensorflow/tools/pip_package/setup.py"
1410                ((".*'tensorboard >.*") ""))
1411              #t))
1412          (add-after 'python3.7-compatibility 'chdir
1413            (lambda _ (chdir "tensorflow/contrib/cmake") #t))
1414          (add-after 'chdir 'disable-downloads
1415            (lambda* (#:key inputs #:allow-other-keys)
1416              (substitute* (find-files "external" "\\.cmake$")
1417                (("GIT_REPOSITORY.*") "")
1418                (("GIT_TAG.*") "")
1419                (("PREFIX ")
1420                 "DOWNLOAD_COMMAND \"\"\nPREFIX "))
1421
1422              ;; Use packages from Guix
1423              (let ((grpc (assoc-ref inputs "grpc")))
1424                (substitute* "CMakeLists.txt"
1425                  ;; Sqlite
1426                  (("include\\(sqlite\\)") "")
1427                  (("\\$\\{sqlite_STATIC_LIBRARIES\\}")
1428                   (string-append (assoc-ref inputs "sqlite")
1429                                  "/lib/libsqlite3.so"))
1430                  (("sqlite_copy_headers_to_destination") "")
1431
1432                  ;; PNG
1433                  (("include\\(png\\)") "")
1434                  (("\\$\\{png_STATIC_LIBRARIES\\}")
1435                   (string-append (assoc-ref inputs "libpng")
1436                                  "/lib/libpng16.so"))
1437                  (("png_copy_headers_to_destination") "")
1438
1439                  ;; JPEG
1440                  (("include\\(jpeg\\)") "")
1441                  (("\\$\\{jpeg_STATIC_LIBRARIES\\}")
1442                   (string-append (assoc-ref inputs "libjpeg")
1443                                  "/lib/libjpeg.so"))
1444                  (("jpeg_copy_headers_to_destination") "")
1445
1446                  ;; GIF
1447                  (("include\\(gif\\)") "")
1448                  (("\\$\\{gif_STATIC_LIBRARIES\\}")
1449                   (string-append (assoc-ref inputs "giflib")
1450                                  "/lib/libgif.so"))
1451                  (("gif_copy_headers_to_destination") "")
1452
1453                  ;; lmdb
1454                  (("include\\(lmdb\\)") "")
1455                  (("\\$\\{lmdb_STATIC_LIBRARIES\\}")
1456                   (string-append (assoc-ref inputs "lmdb")
1457                                  "/lib/liblmdb.so"))
1458                  (("lmdb_copy_headers_to_destination") "")
1459
1460                  ;; Protobuf
1461                  (("include\\(protobuf\\)") "")
1462                  (("protobuf_copy_headers_to_destination") "")
1463                  (("^ +protobuf") "")
1464
1465                  ;; gRPC
1466                  (("include\\(grpc\\)")
1467                   "find_package(grpc REQUIRED NAMES gRPC)")
1468                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES grpc\\)") "")
1469
1470                  ;; Eigen
1471                  (("include\\(eigen\\)")
1472                   (string-append "find_package(eigen REQUIRED NAMES Eigen3)
1473 set(eigen_INCLUDE_DIRS ${CMAKE_CURRENT_BINARY_DIR}/external/eigen_archive "
1474                                  (assoc-ref inputs "eigen") "/include/eigen3)"))
1475                  (("^ +eigen") "")
1476
1477                  ;; snappy
1478                  (("include\\(snappy\\)")
1479                   "add_definitions(-DTF_USE_SNAPPY)")
1480                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES snappy\\)") "")
1481
1482                  ;; jsoncpp
1483                  (("include\\(jsoncpp\\)") "")
1484                  (("^ +jsoncpp") ""))
1485
1486                (substitute* "tf_core_framework.cmake"
1487                  ((" grpc") "")
1488                  (("\\$\\{GRPC_BUILD\\}/grpc_cpp_plugin")
1489                   (which "grpc_cpp_plugin"))
1490                  ;; Link with gRPC libraries
1491                  (("add_library\\(tf_protos_cc.*" m)
1492                   (string-append m
1493                                  (format #f "\ntarget_link_libraries(tf_protos_cc PRIVATE \
1494 ~a/lib/libgrpc++_unsecure.a \
1495 ~a/lib/libgrpc_unsecure.a \
1496 ~a/lib/libaddress_sorting.a \
1497 ~a/lib/libgpr.a \
1498 ~a//lib/libcares.so
1499 )\n"
1500                                          grpc grpc grpc grpc
1501                                          (assoc-ref inputs "c-ares"))))))
1502              (substitute* "tf_tools.cmake"
1503                (("add_dependencies\\(\\$\\{proto_text.*") ""))
1504              ;; Remove dependency on bundled grpc
1505              (substitute* "tf_core_distributed_runtime.cmake"
1506                (("tf_core_cpu grpc") "tf_core_cpu"))
1507
1508              ;; This directory is a dependency of many targets.
1509              (mkdir-p "protobuf")
1510              #t))
1511          (add-after 'configure 'unpack-third-party-sources
1512            (lambda* (#:key inputs #:allow-other-keys)
1513              ;; This is needed to configure bundled packages properly.
1514              (setenv "CONFIG_SHELL" (which "bash"))
1515              (for-each
1516               (lambda (name)
1517                 (let* ((what  (assoc-ref inputs (string-append name "-src")))
1518                        (name* (string-map (lambda (c)
1519                                             (if (char=? c #\-)
1520                                                 #\_ c)) name))
1521                        (where (string-append "../build/" name* "/src/" name*)))
1522                   (cond
1523                    ((string-suffix? ".zip" what)
1524                     (mkdir-p where)
1525                     (with-directory-excursion where
1526                       (invoke "unzip" what)))
1527                    ((string-suffix? ".tar.gz" what)
1528                     (mkdir-p where)
1529                     (invoke "tar" "xf" what
1530                             "-C" where "--strip-components=1"))
1531                    (else
1532                     (let ((parent (dirname where)))
1533                       (mkdir-p parent)
1534                       (with-directory-excursion parent
1535                         (when (file-exists? name*)
1536                           (delete-file-recursively name*))
1537                         (copy-recursively what name*)
1538                         (map make-file-writable
1539                              (find-files name* ".*"))))))))
1540               (list "boringssl"
1541                     "cub"
1542                     "double-conversion"
1543                     "farmhash"
1544                     "fft2d"
1545                     "highwayhash"
1546                     "nsync"
1547                     "re2"))
1548
1549              (rename-file "../build/cub/src/cub/cub-1.8.0/"
1550                           "../build/cub/src/cub/cub/")
1551              #t))
1552          (add-after 'unpack 'fix-python-build
1553            (lambda* (#:key inputs outputs #:allow-other-keys)
1554              (mkdir-p "protobuf-src")
1555              (invoke "tar" "xf" (assoc-ref inputs "protobuf:src")
1556                      "-C" "protobuf-src" "--strip-components=1")
1557              (mkdir-p "eigen-src")
1558              (invoke "tar" "xf" (assoc-ref inputs "eigen:src")
1559                      "-C" "eigen-src" "--strip-components=1")
1560
1561              (substitute* "tensorflow/contrib/cmake/tf_python.cmake"
1562                ;; Ensure that all Python dependencies can be found at build time.
1563                (("PYTHONPATH=\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/tf_python" m)
1564                 (string-append m ":" (getenv "PYTHONPATH")))
1565                ;; Take protobuf source files from our source package.
1566                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/protobuf/src/protobuf/src/google")
1567                 (string-append (getcwd) "/protobuf-src/src/google")))
1568
1569              (substitute* '("tensorflow/contrib/cmake/tf_shared_lib.cmake"
1570                             "tensorflow/contrib/cmake/tf_python.cmake")
1571                ;; Take Eigen source files from our source package.
1572                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/eigen/src/eigen/")
1573                 (string-append (getcwd) "/eigen-src/"))
1574                ;; Take Eigen headers from our own package.
1575                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/external/eigen_archive")
1576                 (string-append (assoc-ref inputs "eigen") "/include/eigen3")))
1577
1578              ;; Correct the RUNPATH of ops libraries generated for Python.
1579              ;; TODO: this doesn't work :(
1580              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1581              ;; warning: RUNPATH contains bogus entries: ("/tmp/guix-build-tensorflow-1.9.0.drv-0/source/tensorflow/contrib/build")
1582              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1583              ;; error: depends on 'libpywrap_tensorflow_internal.so', which
1584              ;; cannot be found in RUNPATH ...
1585              (substitute* "tensorflow/contrib/cmake/tf_cc_ops.cmake"
1586                (("set_target_properties.*")
1587                 (string-append "set_target_properties(${_AT_TARGET} PROPERTIES \
1588 COMPILE_FLAGS ${target_compile_flags} \
1589 INSTALL_RPATH_USE_LINK_PATH TRUE \
1590 INSTALL_RPATH " (assoc-ref outputs "out") "/lib)\n")))
1591              #t))
1592          (add-after 'build 'build-pip-package
1593            (lambda* (#:key outputs #:allow-other-keys)
1594              (setenv "LDFLAGS"
1595                      (string-append "-Wl,-rpath="
1596                                     (assoc-ref outputs "out") "/lib"))
1597              (invoke "make" "tf_python_build_pip_package")
1598              #t))
1599          (add-after 'build-pip-package 'install-python
1600            (lambda* (#:key outputs #:allow-other-keys)
1601              (let ((out (assoc-ref outputs "out"))
1602                    (wheel (car (find-files "../build/tf_python/dist/" "\\.whl$"))))
1603                (invoke "python" "-m" "pip" "install" wheel
1604                        (string-append "--prefix=" out))
1605
1606                ;; XXX: broken RUNPATH, see fix-python-build phase.
1607                (delete-file
1608                 (string-append
1609                  out "/lib/python3.7/site-packages/tensorflow/contrib/"
1610                  "seq2seq/python/ops/lib_beam_search_ops.so"))
1611                #t))))))
1612     (native-inputs
1613      `(("pkg-config" ,pkg-config)
1614        ("protobuf:native" ,protobuf-3.6) ; protoc
1615        ("protobuf:src" ,(package-source protobuf-3.6))
1616        ("eigen:src" ,(package-source eigen-for-tensorflow))
1617        ;; install_pip_packages.sh wants setuptools 39.1.0 specifically.
1618        ("python-setuptools" ,python-setuptools-for-tensorflow)
1619
1620        ;; The commit hashes and URLs for third-party source code are taken
1621        ;; from "tensorflow/workspace.bzl".
1622        ("boringssl-src"
1623         ,(let ((commit "ee7aa02")
1624                (revision "1"))
1625            (origin
1626              (method git-fetch)
1627              (uri (git-reference
1628                    (url "https://boringssl.googlesource.com/boringssl")
1629                    (commit commit)))
1630              (file-name (string-append "boringssl-0-" revision
1631                                        (string-take commit 7)
1632                                        "-checkout"))
1633              (sha256
1634               (base32
1635                "1jf693q0nw0adsic6cgmbdx6g7wr4rj4vxa8j1hpn792fqhd8wgw")))))
1636        ("cub-src"
1637         ,(let ((version "1.8.0"))
1638            (origin
1639              (method url-fetch)
1640              (uri (string-append "https://mirror.bazel.build/github.com/NVlabs/"
1641                                  "cub/archive/" version ".zip"))
1642              (file-name (string-append "cub-" version ".zip"))
1643              (sha256
1644               (base32
1645                "1hsqikqridb90dkxkjr2918dcry6pfh46ccnwrzawl56aamhdykb")))))
1646        ("double-conversion-src"
1647         ,(let ((commit "5664746")
1648                (revision "1"))
1649            (origin
1650              (method git-fetch)
1651              (uri (git-reference
1652                    (url "https://github.com/google/double-conversion.git")
1653                    (commit commit)))
1654              (file-name
1655               (git-file-name "double-conversion"
1656                              (string-append "0-" revision "."
1657                                             (string-take commit 7))))
1658              (sha256
1659               (base32
1660                "1h5lppqqxcvdg5jq42i5msgwx20ryij3apvmndflngrgdpc04gn1")))))
1661        ("farmhash-src"
1662         ,(let ((commit "816a4ae622e964763ca0862d9dbd19324a1eaf45"))
1663            (origin
1664              (method url-fetch)
1665              (uri (string-append
1666                    "https://mirror.bazel.build/github.com/google/farmhash/archive/"
1667                    commit ".tar.gz"))
1668              (file-name (string-append "farmhash-0-" (string-take commit 7)
1669                                        ".tar.gz"))
1670              (sha256
1671               (base32
1672                "185b2xdxl4d4cnsnv6abg8s22gxvx8673jq2yaq85bz4cdy58q35")))))
1673        ;; The license notice on the home page at
1674        ;; http://www.kurims.kyoto-u.ac.jp/~ooura/fft.html says:
1675        ;;   Copyright Takuya OOURA, 1996-2001
1676        ;;
1677        ;;   You may use, copy, modify and distribute this code for any purpose
1678        ;;   (include commercial use) and without fee. Please refer to this
1679        ;;   package when you modify this code.
1680        ;;
1681        ;; We take the identical tarball from the Bazel mirror, because the URL
1682        ;; at the home page is not versioned and might change.
1683        ("fft2d-src"
1684         ,(origin
1685            (method url-fetch)
1686            (uri "https://mirror.bazel.build/www.kurims.kyoto-u.ac.jp/~ooura/fft.tgz")
1687            (file-name "fft2d.tar.gz")
1688            (sha256
1689             (base32
1690              "15jjkfvhqvl2c0753d2di8hz0pyzn598g74wqy79awdrf1y67fsj"))))
1691        ("highwayhash-src"
1692         ,(let ((commit "be5edafc2e1a455768e260ccd68ae7317b6690ee")
1693                (revision "1"))
1694            (origin
1695              (method git-fetch)
1696              (uri (git-reference
1697                    (url "https://github.com/google/highwayhash.git")
1698                    (commit commit)))
1699              (file-name (string-append "highwayhash-0-" revision
1700                                        (string-take commit 7)
1701                                        "-checkout"))
1702              (sha256
1703               (base32
1704                "154jwf98cyy54hldr94pgjn85zynly3abpnc1avmb8a18lzwjyb6")))))
1705        ("nsync-src"
1706         ,(let ((version "0559ce013feac8db639ee1bf776aca0325d28777")
1707                (revision "1"))
1708            (origin
1709              (method url-fetch)
1710              (uri (string-append "https://mirror.bazel.build/"
1711                                  "github.com/google/nsync/archive/"
1712                                  version ".tar.gz"))
1713              (file-name (string-append "nsync-0." revision
1714                                        "-" (string-take version 7)
1715                                        ".tar.gz"))
1716              (sha256
1717               (base32
1718                "0qdkyqym34x739mmzv97ah5r7ph462v5xkxqxvidmcfqbi64b132")))))
1719        ("re2-src"
1720         ,(let ((commit "e7efc48")
1721                (revision "1"))
1722            (origin
1723              (method git-fetch)
1724              (uri (git-reference
1725                    (url "https://github.com/google/re2")
1726                    (commit commit)))
1727              (file-name (string-append "re2-0-" revision
1728                                        (string-take commit 7)
1729                                        "-checkout"))
1730              (sha256
1731               (base32
1732                "161g9841rjfsy5pn52fcis0s9hdr7rxvb06pad38j5rppfihvign")))))
1733        ("googletest" ,googletest)
1734        ("swig" ,swig)
1735        ("unzip" ,unzip)))
1736     (propagated-inputs
1737      `(("python-absl-py" ,python-absl-py)
1738        ("python-astor" ,python-astor)
1739        ("python-gast" ,python-gast)
1740        ("python-grpcio" ,python-grpcio)
1741        ("python-numpy" ,python-numpy)
1742        ("python-protobuf" ,python-protobuf-3.6)
1743        ("python-six" ,python-six)
1744        ("python-termcolo" ,python-termcolor)
1745        ("python-wheel" ,python-wheel)))
1746     (inputs
1747      `(("c-ares" ,c-ares)
1748        ("eigen" ,eigen-for-tensorflow)
1749        ("gemmlowp" ,gemmlowp-for-tensorflow)
1750        ("lmdb" ,lmdb)
1751        ("libjpeg" ,libjpeg-turbo)
1752        ("libpng" ,libpng)
1753        ("giflib" ,giflib)
1754        ("grpc" ,grpc)
1755        ("jsoncpp" ,jsoncpp-for-tensorflow)
1756        ("snappy" ,snappy)
1757        ("sqlite" ,sqlite)
1758        ("protobuf" ,protobuf-3.6)
1759        ("python" ,python-wrapper)
1760        ("zlib" ,zlib)))
1761     (home-page "https://tensorflow.org")
1762     (synopsis "Machine learning framework")
1763     (description
1764      "TensorFlow is a flexible platform for building and training machine
1765 learning models.  It provides a library for high performance numerical
1766 computation and includes high level Python APIs, including both a sequential
1767 API for beginners that allows users to build models quickly by plugging
1768 together building blocks and a subclassing API with an imperative style for
1769 advanced research.")
1770     (license license:asl2.0)))
1771
1772 (define-public python-iml
1773   (package
1774     (name "python-iml")
1775     (version "0.6.2")
1776     (source
1777      (origin
1778        (method url-fetch)
1779        (uri (pypi-uri "iml" version))
1780        (sha256
1781         (base32
1782          "1k8szlpm19rcwcxdny9qdm3gmaqq8akb4xlvrzyz8c2d679aak6l"))))
1783     (build-system python-build-system)
1784     (propagated-inputs
1785      `(("ipython" ,python-ipython)
1786        ("nose" ,python-nose)
1787        ("numpy" ,python-numpy)
1788        ("pandas" ,python-pandas)
1789        ("scipy" ,python-scipy)))
1790     (home-page "http://github.com/interpretable-ml/iml")
1791     (synopsis "Interpretable Machine Learning (iML) package")
1792     (description "Interpretable ML (iML) is a set of data type objects,
1793 visualizations, and interfaces that can be used by any method designed to
1794 explain the predictions of machine learning models (or really the output of
1795 any function).  It currently contains the interface and IO code from the Shap
1796 project, and it will potentially also do the same for the Lime project.")
1797     (license license:expat)))
1798
1799 (define-public python-keras-applications
1800   (package
1801     (name "python-keras-applications")
1802     (version "1.0.8")
1803     (source
1804      (origin
1805        (method url-fetch)
1806        (uri (pypi-uri "Keras_Applications" version))
1807        (sha256
1808         (base32
1809          "1rcz31ca4axa6kzhjx4lwqxbg4wvlljkj8qj9a7p9sfd5fhzjyam"))))
1810     (build-system python-build-system)
1811     ;; The tests require Keras, but this package is needed to build Keras.
1812     (arguments '(#:tests? #f))
1813     (propagated-inputs
1814      `(("python-h5py" ,python-h5py)
1815        ("python-numpy" ,python-numpy)))
1816     (native-inputs
1817      `(("python-pytest" ,python-pytest)
1818        ("python-pytest-cov" ,python-pytest-cov)
1819        ("python-pytest-pep8" ,python-pytest-pep8)
1820        ("python-pytest-xdist" ,python-pytest-xdist)))
1821     (home-page "https://github.com/keras-team/keras-applications")
1822     (synopsis "Reference implementations of popular deep learning models")
1823     (description
1824      "This package provides reference implementations of popular deep learning
1825 models for use with the Keras deep learning framework.")
1826     (license license:expat)))
1827
1828 (define-public python-keras-preprocessing
1829   (package
1830     (name "python-keras-preprocessing")
1831     (version "1.1.0")
1832     (source
1833      (origin
1834        (method url-fetch)
1835        (uri (pypi-uri "Keras_Preprocessing" version))
1836        (sha256
1837         (base32
1838          "1r98nm4k1svsqjyaqkfk23i31bl1kcfcyp7094yyj3c43phfp3as"))))
1839     (build-system python-build-system)
1840     (propagated-inputs
1841      `(("python-numpy" ,python-numpy)
1842        ("python-six" ,python-six)))
1843     (native-inputs
1844      `(("python-pandas" ,python-pandas)
1845        ("python-pillow" ,python-pillow)
1846        ("python-pytest" ,python-pytest)
1847        ("python-pytest-cov" ,python-pytest-cov)
1848        ("python-pytest-xdist" ,python-pytest-xdist)
1849        ("tensorflow" ,tensorflow)))
1850     (home-page "https://github.com/keras-team/keras-preprocessing/")
1851     (synopsis "Data preprocessing and augmentation for deep learning models")
1852     (description
1853      "Keras Preprocessing is the data preprocessing and data augmentation
1854 module of the Keras deep learning library.  It provides utilities for working
1855 with image data, text data, and sequence data.")
1856     (license license:expat)))
1857
1858 (define-public python-keras
1859   (package
1860     (name "python-keras")
1861     (version "2.2.4")
1862     (source
1863      (origin
1864        (method url-fetch)
1865        (uri (pypi-uri "Keras" version))
1866        (patches (search-patches "python-keras-integration-test.patch"))
1867        (sha256
1868         (base32
1869          "1j8bsqzh49vjdxy6l1k4iwax5vpjzniynyd041xjavdzvfii1dlh"))))
1870     (build-system python-build-system)
1871     (arguments
1872      `(#:phases
1873        (modify-phases %standard-phases
1874          (add-after 'unpack 'remove-tests-for-unavailable-features
1875            (lambda _
1876              (delete-file "keras/backend/theano_backend.py")
1877              (delete-file "keras/backend/cntk_backend.py")
1878              (delete-file "tests/keras/backend/backend_test.py")
1879
1880              ;; FIXME: This doesn't work because Tensorflow is missing the
1881              ;; coder ops library.
1882              (delete-file "tests/keras/test_callbacks.py")
1883              #t))
1884          (replace 'check
1885            (lambda _
1886              ;; These tests attempt to download data files from the internet.
1887              (delete-file "tests/integration_tests/test_datasets.py")
1888              (delete-file "tests/integration_tests/imagenet_utils_test.py")
1889
1890              (setenv "PYTHONPATH"
1891                      (string-append (getcwd) "/build/lib:"
1892                                     (getenv "PYTHONPATH")))
1893              (invoke "py.test" "-v"
1894                      "-p" "no:cacheprovider"
1895                      "--ignore" "keras/utils"))))))
1896     (propagated-inputs
1897      `(("python-h5py" ,python-h5py)
1898        ("python-keras-applications" ,python-keras-applications)
1899        ("python-keras-preprocessing" ,python-keras-preprocessing)
1900        ("python-numpy" ,python-numpy)
1901        ("python-pydot" ,python-pydot)
1902        ("python-pyyaml" ,python-pyyaml)
1903        ("python-scipy" ,python-scipy)
1904        ("python-six" ,python-six)
1905        ("tensorflow" ,tensorflow)
1906        ("graphviz" ,graphviz)))
1907     (native-inputs
1908      `(("python-pandas" ,python-pandas)
1909        ("python-pytest" ,python-pytest)
1910        ("python-pytest-cov" ,python-pytest-cov)
1911        ("python-pytest-pep8" ,python-pytest-pep8)
1912        ("python-pytest-timeout" ,python-pytest-timeout)
1913        ("python-pytest-xdist" ,python-pytest-xdist)
1914        ("python-sphinx" ,python-sphinx)
1915        ("python-requests" ,python-requests)))
1916     (home-page "https://github.com/keras-team/keras")
1917     (synopsis "High-level deep learning framework")
1918     (description "Keras is a high-level neural networks API, written in Python
1919 and capable of running on top of TensorFlow.  It was developed with a focus on
1920 enabling fast experimentation.  Use Keras if you need a deep learning library
1921 that:
1922
1923 @itemize
1924 @item Allows for easy and fast prototyping (through user friendliness,
1925   modularity, and extensibility).
1926 @item Supports both convolutional networks and recurrent networks, as well as
1927   combinations of the two.
1928 @item Runs seamlessly on CPU and GPU.
1929 @end itemize\n")
1930     (license license:expat)))
1931
1932 (define-public sbcl-cl-libsvm-format
1933   (let ((commit "3300f84fd8d9f5beafc114f543f9d83417c742fb")
1934         (revision "0"))
1935     (package
1936       (name "sbcl-cl-libsvm-format")
1937       (version (git-version "0.1.0" revision commit))
1938       (source
1939        (origin
1940          (method git-fetch)
1941          (uri (git-reference
1942                (url "https://github.com/masatoi/cl-libsvm-format.git")
1943                (commit commit)))
1944          (file-name (git-file-name name version))
1945          (sha256
1946           (base32
1947            "0284aj84xszhkhlivaigf9qj855fxad3mzmv3zfr0qzb5k0nzwrg"))))
1948       (build-system asdf-build-system/sbcl)
1949       (native-inputs
1950        `(("prove" ,sbcl-prove)
1951          ("prove-asdf" ,sbcl-prove-asdf)))
1952       (inputs
1953        `(("alexandria" ,sbcl-alexandria)))
1954       (synopsis "LibSVM data format reader for Common Lisp")
1955       (description
1956        "This Common Lisp library provides a fast reader for data in LibSVM
1957 format.")
1958       (home-page "https://github.com/masatoi/cl-libsvm-format")
1959       (license license:expat))))
1960
1961 (define-public cl-libsvm-format
1962   (sbcl-package->cl-source-package sbcl-cl-libsvm-format))
1963
1964 (define-public ecl-cl-libsvm-format
1965   (sbcl-package->ecl-package sbcl-cl-libsvm-format))
1966
1967 (define-public sbcl-cl-online-learning
1968   (let ((commit "fc7a34f4f161cd1c7dd747d2ed8f698947781423")
1969         (revision "0"))
1970     (package
1971       (name "sbcl-cl-online-learning")
1972       (version (git-version "0.5" revision commit))
1973       (source
1974        (origin
1975          (method git-fetch)
1976          (uri (git-reference
1977                (url "https://github.com/masatoi/cl-online-learning.git")
1978                (commit commit)))
1979          (file-name (git-file-name name version))
1980          (sha256
1981           (base32
1982            "14x95rlg80ay5hv645ki57pqvy12v28hz4k1w0f6bsfi2rmpxchq"))))
1983       (build-system asdf-build-system/sbcl)
1984       (native-inputs
1985        `(("prove" ,sbcl-prove)
1986          ("prove-asdf" ,sbcl-prove-asdf)))
1987       (inputs
1988        `(("cl-libsvm-format" ,sbcl-cl-libsvm-format)
1989          ("cl-store" ,sbcl-cl-store)))
1990       (arguments
1991        `(;; FIXME: Tests pass but then the check phase crashes
1992          #:tests? #f))
1993       (synopsis "Online Machine Learning for Common Lisp")
1994       (description
1995        "This library contains a collection of machine learning algorithms for
1996 online linear classification written in Common Lisp.")
1997       (home-page "https://github.com/masatoi/cl-online-learning")
1998       (license license:expat))))
1999
2000 (define-public cl-online-learning
2001   (sbcl-package->cl-source-package sbcl-cl-online-learning))
2002
2003 (define-public ecl-cl-online-learning
2004   (sbcl-package->ecl-package sbcl-cl-online-learning))
2005
2006 (define-public sbcl-cl-random-forest
2007   (let ((commit "85fbdd4596d40e824f70f1b7cf239cf544e49d51")
2008         (revision "0"))
2009     (package
2010       (name "sbcl-cl-random-forest")
2011       (version (git-version "0.1" revision commit))
2012       (source
2013        (origin
2014          (method git-fetch)
2015          (uri (git-reference
2016                (url "https://github.com/masatoi/cl-random-forest.git")
2017                (commit commit)))
2018          (file-name (git-file-name name version))
2019          (sha256
2020           (base32
2021            "097xv60i1ndz68sg9p4pc7c5gvyp9i1xgw966b4wwfq3x6hbz421"))))
2022       (build-system asdf-build-system/sbcl)
2023       (native-inputs
2024        `(("prove" ,sbcl-prove)
2025          ("prove-asdf" ,sbcl-prove-asdf)
2026          ("trivial-garbage" ,sbcl-trivial-garbage)))
2027       (inputs
2028        `(("alexandria" ,sbcl-alexandria)
2029          ("cl-libsvm-format" ,sbcl-cl-libsvm-format)
2030          ("cl-online-learning" ,sbcl-cl-online-learning)
2031          ("lparallel" ,sbcl-lparallel)))
2032       (arguments
2033        `(;; The tests download data from the Internet
2034          #:tests? #f
2035          #:phases
2036          (modify-phases %standard-phases
2037            (add-after 'unpack 'add-sb-cltl2-dependency
2038              (lambda _
2039                ;; sb-cltl2 is required by lparallel when using sbcl, but it is
2040                ;; not loaded automatically.
2041                (substitute* "cl-random-forest.asd"
2042                  (("\\(in-package :cl-user\\)")
2043                   "(in-package :cl-user) #+sbcl (require :sb-cltl2)"))
2044                #t)))))
2045       (synopsis "Random Forest and Global Refinement for Common Lisp")
2046       (description
2047        "CL-random-forest is an implementation of Random Forest for multiclass
2048 classification and univariate regression written in Common Lisp.  It also
2049 includes an implementation of Global Refinement of Random Forest.")
2050       (home-page "https://github.com/masatoi/cl-random-forest")
2051       (license license:expat))))
2052
2053 (define-public cl-random-forest
2054   (sbcl-package->cl-source-package sbcl-cl-random-forest))
2055
2056 (define-public ecl-cl-random-forest
2057   (sbcl-package->ecl-package sbcl-cl-random-forest))
2058
2059 (define-public gloo
2060   (let ((version "0.0.0") ; no proper version tag
2061         (commit "ca528e32fea9ca8f2b16053cff17160290fc84ce")
2062         (revision "0"))
2063     (package
2064       (name "gloo")
2065       (version (git-version version revision commit))
2066       (source
2067        (origin
2068          (method git-fetch)
2069          (uri (git-reference
2070                (url "https://github.com/facebookincubator/gloo.git")
2071                (commit commit)))
2072          (file-name (git-file-name name version))
2073          (sha256
2074           (base32
2075            "1q9f80zy75f6njrzrqkmhc0g3qxs4gskr7ns2jdqanxa2ww7a99w"))))
2076       (build-system cmake-build-system)
2077       (native-inputs
2078        `(("googletest" ,googletest)))
2079       (arguments
2080        `(#:configure-flags '("-DBUILD_TEST=1")
2081          #:phases
2082          (modify-phases %standard-phases
2083            (replace 'check
2084              (lambda _
2085                (invoke "make" "gloo_test")
2086                #t)))))
2087       (synopsis "Collective communications library")
2088       (description
2089        "Gloo is a collective communications library.  It comes with a
2090 number of collective algorithms useful for machine learning applications.
2091 These include a barrier, broadcast, and allreduce.")
2092       (home-page "https://github.com/facebookincubator/gloo")
2093       (license license:bsd-3))))
2094
2095 (define-public python-umap-learn
2096   (package
2097     (name "python-umap-learn")
2098     (version "0.3.10")
2099     (source
2100      (origin
2101        (method url-fetch)
2102        (uri (pypi-uri "umap-learn" version))
2103        (sha256
2104         (base32
2105          "02ada2yy6km6zgk2836kg1c97yrcpalvan34p8c57446finnpki1"))))
2106     (build-system python-build-system)
2107     (native-inputs
2108      `(("python-nose" ,python-nose)))
2109     (propagated-inputs
2110      `(("python-numba" ,python-numba)
2111        ("python-numpy" ,python-numpy)
2112        ("python-scikit-learn" ,python-scikit-learn)
2113        ("python-scipy" ,python-scipy)))
2114     (home-page "https://github.com/lmcinnes/umap")
2115     (synopsis
2116      "Uniform Manifold Approximation and Projection")
2117     (description
2118      "Uniform Manifold Approximation and Projection is a dimension reduction
2119 technique that can be used for visualisation similarly to t-SNE, but also for
2120 general non-linear dimension reduction.")
2121     (license license:bsd-3)))