gnu/packages/machine-learning.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2015, 2016, 2017, 2018, 2019, 2020 Ricardo Wurmus <rekado@elephly.net>
   3 ;;; Copyright © 2016, 2020 Efraim Flashner <efraim@flashner.co.il>
   4 ;;; Copyright © 2016, 2017, 2020 Marius Bakke <mbakke@fastmail.com>
   5 ;;; Copyright © 2016 Hartmut Goebel <h.goebel@crazy-compilers.com>
   6 ;;; Copyright © 2018, 2019, 2020 Tobias Geerinckx-Rice <me@tobias.gr>
   7 ;;; Copyright © 2018 Kei Kebreau <kkebreau@posteo.net>
   8 ;;; Copyright © 2018 Mark Meyer <mark@ofosos.org>
   9 ;;; Copyright © 2018 Ben Woodcroft <donttrustben@gmail.com>
  10 ;;; Copyright © 2018 Fis Trivial <ybbs.daans@hotmail.com>
  11 ;;; Copyright © 2018 Julien Lepiller <julien@lepiller.eu>
  12 ;;; Copyright © 2018 Björn Höfling <bjoern.hoefling@bjoernhoefling.de>
  13 ;;; Copyright © 2019 Nicolas Goaziou <mail@nicolasgoaziou.fr>
  14 ;;; Copyright © 2019, 2020 Guillaume Le Vaillant <glv@posteo.net>
  15 ;;; Copyright © 2019 Brett Gilio <brettg@gnu.org>
  16 ;;; Copyright © 2020 Konrad Hinsen <konrad.hinsen@fastmail.net>
  17 ;;; Copyright © 2020 Edouard Klein <edk@beaver-labs.com>
  18 ;;; Copyright © 2020 Vinicius Monego <monego@posteo.net>
  19 ;;;
  20 ;;; This file is part of GNU Guix.
  21 ;;;
  22 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  23 ;;; under the terms of the GNU General Public License as published by
  24 ;;; the Free Software Foundation; either version 3 of the License, or (at
  25 ;;; your option) any later version.
  26 ;;;
  27 ;;; GNU Guix is distributed in the hope that it will be useful, but
  28 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  29 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  30 ;;; GNU General Public License for more details.
  31 ;;;
  32 ;;; You should have received a copy of the GNU General Public License
  33 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  34
  35 (define-module (gnu packages machine-learning)
  36   #:use-module ((guix licenses) #:prefix license:)
  37   #:use-module (guix packages)
  38   #:use-module (guix utils)
  39   #:use-module (guix download)
  40   #:use-module (guix svn-download)
  41   #:use-module (guix build-system cmake)
  42   #:use-module (guix build-system gnu)
  43   #:use-module (guix build-system ocaml)
  44   #:use-module (guix build-system python)
  45   #:use-module (guix build-system r)
  46   #:use-module (guix git-download)
  47   #:use-module (gnu packages)
  48   #:use-module (gnu packages adns)
  49   #:use-module (gnu packages algebra)
  50   #:use-module (gnu packages audio)
  51   #:use-module (gnu packages autotools)
  52   #:use-module (gnu packages base)
  53   #:use-module (gnu packages bash)
  54   #:use-module (gnu packages boost)
  55   #:use-module (gnu packages check)
  56   #:use-module (gnu packages compression)
  57   #:use-module (gnu packages cmake)
  58   #:use-module (gnu packages cran)
  59   #:use-module (gnu packages databases)
  60   #:use-module (gnu packages dejagnu)
  61   #:use-module (gnu packages gcc)
  62   #:use-module (gnu packages glib)
  63   #:use-module (gnu packages graphviz)
  64   #:use-module (gnu packages gstreamer)
  65   #:use-module (gnu packages image)
  66   #:use-module (gnu packages linux)
  67   #:use-module (gnu packages maths)
  68   #:use-module (gnu packages mpi)
  69   #:use-module (gnu packages ocaml)
  70   #:use-module (gnu packages onc-rpc)
  71   #:use-module (gnu packages perl)
  72   #:use-module (gnu packages pkg-config)
  73   #:use-module (gnu packages protobuf)
  74   #:use-module (gnu packages python)
  75   #:use-module (gnu packages python-check)
  76   #:use-module (gnu packages python-science)
  77   #:use-module (gnu packages python-web)
  78   #:use-module (gnu packages python-xyz)
  79   #:use-module (gnu packages rpc)
  80   #:use-module (gnu packages serialization)
  81   #:use-module (gnu packages sphinx)
  82   #:use-module (gnu packages statistics)
  83   #:use-module (gnu packages sqlite)
  84   #:use-module (gnu packages swig)
  85   #:use-module (gnu packages web)
  86   #:use-module (gnu packages xml)
  87   #:use-module (gnu packages xorg)
  88   #:use-module (ice-9 match))
  89
  90 (define-public fann
  91   ;; The last release is >100 commits behind, so we package from git.
  92   (let ((commit "d71d54788bee56ba4cf7522801270152da5209d7"))
  93     (package
  94       (name "fann")
  95       (version (string-append "2.2.0-1." (string-take commit 8)))
  96       (source (origin
  97                 (method git-fetch)
  98                 (uri (git-reference
  99                       (url "https://github.com/libfann/fann")
 100                       (commit commit)))
 101                 (file-name (string-append name "-" version "-checkout"))
 102                 (sha256
 103                  (base32
 104                   "0ibwpfrjs6q2lijs8slxjgzb2llcl6rk3v2ski4r6215g5jjhg3x"))))
 105       (build-system cmake-build-system)
 106       (arguments
 107        `(#:phases
 108          (modify-phases %standard-phases
 109            (replace 'check
 110              (lambda* (#:key outputs #:allow-other-keys)
 111                (let* ((out (assoc-ref outputs "out")))
 112                  (with-directory-excursion (string-append (getcwd) "/tests")
 113                    (invoke "./fann_tests"))))))))
 114       (home-page "http://leenissen.dk/fann/wp/")
 115       (synopsis "Fast Artificial Neural Network")
 116       (description
 117        "FANN is a neural network library, which implements multilayer
 118 artificial neural networks in C with support for both fully connected and
 119 sparsely connected networks.")
 120       (license license:lgpl2.1))))
 121
 122 (define-public libsvm
 123   (package
 124     (name "libsvm")
 125     (version "3.23")
 126     (source
 127      (origin
 128        (method url-fetch)
 129        (uri (string-append "https://www.csie.ntu.edu.tw/~cjlin/libsvm/"
 130                            name "-" version ".tar.gz"))
 131        (sha256
 132         (base32 "0jpaq0rr92x38p4nk3gjan79ip67m6p80anb28z1d8601miysyi5"))))
 133     (build-system gnu-build-system)
 134     (arguments
 135      `(#:tests? #f                      ; no "check" target
 136        #:phases (modify-phases %standard-phases
 137                   (delete 'configure)
 138                   (replace
 139                    'install             ; no ‘install’ target
 140                    (lambda* (#:key outputs #:allow-other-keys)
 141                      (let* ((out (assoc-ref outputs "out"))
 142                             (bin (string-append out "/bin/")))
 143                        (mkdir-p bin)
 144                        (for-each (lambda (file)
 145                                    (copy-file file (string-append bin file)))
 146                                  '("svm-train"
 147                                    "svm-predict"
 148                                    "svm-scale")))
 149                      #t)))))
 150     (home-page "https://www.csie.ntu.edu.tw/~cjlin/libsvm/")
 151     (synopsis "Library for Support Vector Machines")
 152     (description
 153      "LIBSVM is a machine learning library for support vector
 154 classification, (C-SVC, nu-SVC), regression (epsilon-SVR, nu-SVR) and
 155 distribution estimation (one-class SVM).  It supports multi-class
 156 classification.")
 157     (license license:bsd-3)))
 158
 159 (define-public python-libsvm
 160   (package (inherit libsvm)
 161     (name "python-libsvm")
 162     (build-system gnu-build-system)
 163     (arguments
 164      `(#:tests? #f                      ; no "check" target
 165        #:make-flags '("-C" "python")
 166        #:phases
 167        (modify-phases %standard-phases
 168          (delete 'configure)
 169          (replace
 170           'install                      ; no ‘install’ target
 171           (lambda* (#:key inputs outputs #:allow-other-keys)
 172             (let ((site (string-append (assoc-ref outputs "out")
 173                                        "/lib/python"
 174                                        (string-take
 175                                         (string-take-right
 176                                          (assoc-ref inputs "python") 5) 3)
 177                                        "/site-packages/")))
 178               (substitute* "python/svm.py"
 179                 (("../libsvm.so.2") "libsvm.so.2"))
 180               (mkdir-p site)
 181               (for-each (lambda (file)
 182                           (copy-file file (string-append site (basename file))))
 183                         (find-files "python" "\\.py"))
 184               (copy-file "libsvm.so.2"
 185                          (string-append site "libsvm.so.2")))
 186             #t)))))
 187     (inputs
 188      `(("python" ,python)))
 189     (synopsis "Python bindings of libSVM")))
 190
 191 (define-public ghmm
 192   ;; The latest release candidate is several years and a couple of fixes have
 193   ;; been published since.  This is why we download the sources from the SVN
 194   ;; repository.
 195   (let ((svn-revision 2341))
 196     (package
 197       (name "ghmm")
 198       (version (string-append "0.9-rc3-0." (number->string svn-revision)))
 199       (source (origin
 200                 (method svn-fetch)
 201                 (uri (svn-reference
 202                       (url "http://svn.code.sf.net/p/ghmm/code/trunk")
 203                       (revision svn-revision)))
 204                 (file-name (string-append name "-" version "-checkout"))
 205                 (sha256
 206                  (base32
 207                   "0qbq1rqp94l530f043qzp8aw5lj7dng9wq0miffd7spd1ff638wq"))))
 208       (build-system gnu-build-system)
 209       (arguments
 210        `(#:imported-modules (,@%gnu-build-system-modules
 211                              (guix build python-build-system))
 212          #:modules          ((guix build python-build-system)
 213                              ,@%gnu-build-system-modules)
 214          #:phases
 215          (modify-phases %standard-phases
 216            (add-after 'unpack 'enter-dir
 217              (lambda _ (chdir "ghmm") #t))
 218            (delete 'check)
 219            (add-after 'install 'check
 220              (assoc-ref %standard-phases 'check))
 221            (add-before 'check 'fix-PYTHONPATH
 222              (lambda* (#:key inputs outputs #:allow-other-keys)
 223                (let ((python-version (python-version
 224                                       (assoc-ref inputs "python"))))
 225                  (setenv "PYTHONPATH"
 226                          (string-append (getenv "PYTHONPATH")
 227                                         ":" (assoc-ref outputs "out")
 228                                         "/lib/python" python-version
 229                                         "/site-packages")))
 230                #t))
 231            (add-after 'enter-dir 'fix-runpath
 232              (lambda* (#:key outputs #:allow-other-keys)
 233                (substitute* "ghmmwrapper/setup.py"
 234                  (("^(.*)extra_compile_args = \\[" line indent)
 235                   (string-append indent
 236                                  "extra_link_args = [\"-Wl,-rpath="
 237                                  (assoc-ref outputs "out") "/lib\"],\n"
 238                                  line
 239                                  "\"-Wl,-rpath="
 240                                  (assoc-ref outputs "out")
 241                                  "/lib\", ")))
 242                #t))
 243            (add-after 'enter-dir 'disable-broken-tests
 244              (lambda _
 245                (substitute* "tests/Makefile.am"
 246                  ;; GHMM_SILENT_TESTS is assumed to be a command.
 247                  (("TESTS_ENVIRONMENT.*") "")
 248                  ;; Do not build broken tests.
 249                  (("chmm .*") "")
 250                  (("read_fa .*") "")
 251                  (("mcmc .*") "")
 252                  (("label_higher_order_test.*$")
 253                   "label_higher_order_test\n"))
 254
 255                ;; These Python unittests are broken as there is no gato.
 256                ;; See https://sourceforge.net/p/ghmm/support-requests/3/
 257                (substitute* "ghmmwrapper/ghmmunittests.py"
 258                  (("^(.*)def (testNewXML|testMultipleTransitionClasses|testNewXML)"
 259                    line indent)
 260                   (string-append indent
 261                                  "@unittest.skip(\"Disabled by Guix\")\n"
 262                                  line)))
 263                #t)))))
 264       (inputs
 265        `(("python" ,python-2) ; only Python 2 is supported
 266          ("libxml2" ,libxml2)))
 267       (native-inputs
 268        `(("pkg-config" ,pkg-config)
 269          ("dejagnu" ,dejagnu)
 270          ("swig" ,swig)
 271          ("autoconf" ,autoconf)
 272          ("automake" ,automake)
 273          ("libtool" ,libtool)))
 274       (home-page "http://ghmm.org")
 275       (synopsis "Hidden Markov Model library")
 276       (description
 277        "The General Hidden Markov Model library (GHMM) is a C library with
 278 additional Python bindings implementing a wide range of types of @dfn{Hidden
 279 Markov Models} (HMM) and algorithms: discrete, continuous emissions, basic
 280 training, HMM clustering, HMM mixtures.")
 281       (license license:lgpl2.0+))))
 282
 283 (define-public mcl
 284   (package
 285     (name "mcl")
 286     (version "14.137")
 287     (source (origin
 288               (method url-fetch)
 289               (uri (string-append
 290                     "http://micans.org/mcl/src/mcl-"
 291                     (string-replace-substring version "." "-")
 292                     ".tar.gz"))
 293               (sha256
 294                (base32
 295                 "15xlax3z31lsn62vlg94hkm75nm40q4679amnfg13jm8m2bnhy5m"))))
 296     (build-system gnu-build-system)
 297     (arguments
 298      `(#:configure-flags (list "--enable-blast")))
 299     (inputs
 300      `(("perl" ,perl)))
 301     (home-page "http://micans.org/mcl/")
 302     (synopsis "Clustering algorithm for graphs")
 303     (description
 304      "The MCL algorithm is short for the @dfn{Markov Cluster Algorithm}, a
 305 fast and scalable unsupervised cluster algorithm for graphs (also known as
 306 networks) based on simulation of (stochastic) flow in graphs.")
 307     ;; In the LICENCE file and web page it says "The software is licensed
 308     ;; under the GNU General Public License, version 3.", but in several of
 309     ;; the source code files it suggests GPL3 or later.
 310     ;; http://listserver.ebi.ac.uk/pipermail/mcl-users/2016/000376.html
 311     (license license:gpl3)))
 312
 313 (define-public ocaml-mcl
 314   (package
 315     (name "ocaml-mcl")
 316     (version "12-068oasis4")
 317     (source
 318      (origin
 319        (method git-fetch)
 320        (uri (git-reference
 321              (url "https://github.com/fhcrc/mcl")
 322              (commit version)))
 323        (file-name (git-file-name name version))
 324        (sha256
 325         (base32
 326          "0009dc3h2jp3qg5val452wngpqnbfyhbcxylghq0mrjqxx0jdq5p"))))
 327     (build-system ocaml-build-system)
 328     (arguments
 329      `(#:phases
 330        (modify-phases %standard-phases
 331          (add-before 'configure 'patch-paths
 332            (lambda _
 333              (substitute* "configure"
 334                (("/bin/sh") (which "sh")))
 335              (substitute* "setup.ml"
 336                (("LDFLAGS=-fPIC")
 337                 (string-append "LDFLAGS=-fPIC\"; \"SHELL=" (which "sh")))
 338                (("-std=c89") "-std=gnu99")
 339
 340                ;; This is a mutable string, which is no longer supported.  Use
 341                ;; a byte buffer instead.
 342                (("String.make \\(String.length s\\)")
 343                 "Bytes.make (String.length s)")
 344
 345                ;; These two belong together.
 346                (("OASISString.replace_chars")
 347                 "Bytes.to_string (OASISString.replace_chars")
 348                ((" s;")
 349                 " s);"))
 350              (substitute* "myocamlbuild.ml"
 351                (("std=c89") "std=gnu99"))
 352              ;; Since we build with a more recent OCaml, we have to use C99 or
 353              ;; later.  This causes problems with the old C code.
 354              (substitute* "src/impala/matrix.c"
 355                (("restrict") "restrict_"))
 356              #t)))))
 357     (native-inputs
 358      `(("ocamlbuild" ,ocamlbuild)))
 359     (home-page "https://github.com/fhcrc/mcl")
 360     (synopsis "OCaml wrappers around MCL")
 361     (description
 362      "This package provides OCaml bindings for the MCL graph clustering
 363 algorithm.")
 364     (license license:gpl3)))
 365
 366 (define-public randomjungle
 367   (package
 368     (name "randomjungle")
 369     (version "2.1.0")
 370     (source
 371      (origin
 372        (method url-fetch)
 373        (uri (string-append
 374              "https://www.imbs.uni-luebeck.de/fileadmin/files/Software"
 375              "/randomjungle/randomjungle-" version ".tar_.gz"))
 376        (patches (search-patches "randomjungle-disable-static-build.patch"))
 377        (sha256
 378         (base32
 379          "12c8rf30cla71swx2mf4ww9mfd8jbdw5lnxd7dxhyw1ygrvg6y4w"))))
 380     (build-system gnu-build-system)
 381     (arguments
 382      `(#:configure-flags
 383        (list "--disable-static"
 384              (string-append "--with-boost="
 385                             (assoc-ref %build-inputs "boost")))
 386        #:phases
 387        (modify-phases %standard-phases
 388          (add-before
 389           'configure 'set-CXXFLAGS
 390           (lambda _
 391             (setenv "CXXFLAGS" "-fpermissive ")
 392             #t)))))
 393     (inputs
 394      `(("boost" ,boost)
 395        ("gsl" ,gsl)
 396        ("libxml2" ,libxml2)
 397        ("zlib" ,zlib)))
 398     (native-inputs
 399      `(("gfortran" ,gfortran)
 400        ("gfortran:lib" ,gfortran "lib")))
 401     ;; Non-portable assembly instructions are used so building fails on
 402     ;; platforms other than x86_64 or i686.
 403     (supported-systems '("x86_64-linux" "i686-linux"))
 404     (home-page "https://www.imbs.uni-luebeck.de/forschung/software/details.html#c224")
 405     (synopsis "Implementation of the Random Forests machine learning method")
 406     (description
 407      "Random Jungle is an implementation of Random Forests.  It is supposed to
 408 analyse high dimensional data.  In genetics, it can be used for analysing big
 409 Genome Wide Association (GWA) data.  Random Forests is a powerful machine
 410 learning method.  Most interesting features are variable selection, missing
 411 value imputation, classifier creation, generalization error estimation and
 412 sample proximities between pairs of cases.")
 413     (license license:gpl3+)))
 414
 415 (define-public openfst
 416   (package
 417     (name "openfst")
 418     (version "1.7.2")
 419     (source (origin
 420               (method url-fetch)
 421               (uri (string-append "http://www.openfst.org/twiki/pub/FST/"
 422                                   "FstDownload/openfst-" version ".tar.gz"))
 423               (sha256
 424                (base32
 425                 "0fqgk8195kz21is09gwzwnrg7fr9526bi9mh4apyskapz27pbhr1"))))
 426     (build-system gnu-build-system)
 427     (home-page "http://www.openfst.org")
 428     (synopsis "Library for weighted finite-state transducers")
 429     (description "OpenFst is a library for constructing, combining,
 430 optimizing, and searching weighted finite-state transducers (FSTs).")
 431     (license license:asl2.0)))
 432
 433 (define-public shogun
 434   (package
 435     (name "shogun")
 436     (version "6.1.3")
 437     (source
 438      (origin
 439        (method url-fetch)
 440        (uri (string-append
 441              "ftp://shogun-toolbox.org/shogun/releases/"
 442              (version-major+minor version)
 443              "/sources/shogun-" version ".tar.bz2"))
 444        (sha256
 445         (base32
 446          "1rn9skm3nw6hr7mr3lgp2gfqhi7ii0lyxck7qmqnf8avq349s5jp"))
 447        (modules '((guix build utils)
 448                   (ice-9 rdelim)))
 449        (snippet
 450         '(begin
 451            ;; Remove non-free sources and files referencing them
 452            (for-each delete-file
 453                      (find-files "src/shogun/classifier/svm/"
 454                                  "SVMLight\\.(cpp|h)"))
 455            (for-each delete-file
 456                      (find-files "examples/undocumented/libshogun/"
 457                                  (string-append
 458                                   "(classifier_.*svmlight.*|"
 459                                   "evaluation_cross_validation_locked_comparison).cpp")))
 460            ;; Remove non-free functions.
 461            (define (delete-ifdefs file)
 462              (with-atomic-file-replacement file
 463                (lambda (in out)
 464                  (let loop ((line (read-line in 'concat))
 465                             (skipping? #f))
 466                    (if (eof-object? line)
 467                        #t
 468                        (let ((skip-next?
 469                               (or (and skipping?
 470                                        (not (string-prefix?
 471                                              "#endif //USE_SVMLIGHT" line)))
 472                                   (string-prefix?
 473                                    "#ifdef USE_SVMLIGHT" line))))
 474                          (when (or (not skipping?)
 475                                    (and skipping? (not skip-next?)))
 476                            (display line out))
 477                          (loop (read-line in 'concat) skip-next?)))))))
 478            (for-each delete-ifdefs
 479                      (append
 480                       (find-files "src/shogun/classifier/mkl"
 481                                   "^MKLClassification\\.cpp")
 482                       (find-files "src/shogun/classifier/svm"
 483                                   "^SVMLightOneClass\\.(cpp|h)")
 484                       (find-files "src/shogun/multiclass"
 485                                   "^ScatterSVM\\.(cpp|h)")
 486                       (find-files "src/shogun/kernel/"
 487                                   "^(Kernel|CombinedKernel|ProductKernel)\\.(cpp|h)")
 488                       (find-files "src/shogun/regression/svr"
 489                                   "^(MKLRegression|SVRLight)\\.(cpp|h)")
 490                       (find-files "src/shogun/transfer/domain_adaptation"
 491                                   "^DomainAdaptationSVM\\.(cpp|h)")))
 492            #t))))
 493     (build-system cmake-build-system)
 494     (arguments
 495      '(#:tests? #f ;no check target
 496        #:phases
 497        (modify-phases %standard-phases
 498          (add-after 'unpack 'delete-broken-symlinks
 499            (lambda _
 500              (for-each delete-file '("applications/arts/data"
 501                                      "applications/asp/data"
 502                                      "applications/easysvm/data"
 503                                      "applications/msplicer/data"
 504                                      "applications/ocr/data"
 505                                      "examples/meta/data"
 506                                      "examples/undocumented/data"))
 507              #t))
 508          (add-after 'unpack 'change-R-target-path
 509            (lambda* (#:key outputs #:allow-other-keys)
 510              (substitute* '("src/interfaces/r/CMakeLists.txt"
 511                             "examples/meta/r/CMakeLists.txt")
 512                (("\\$\\{R_COMPONENT_LIB_PATH\\}")
 513                 (string-append (assoc-ref outputs "out")
 514                                "/lib/R/library/")))
 515              #t))
 516          (add-after 'unpack 'fix-octave-modules
 517            (lambda* (#:key outputs #:allow-other-keys)
 518              (substitute* "src/interfaces/octave/CMakeLists.txt"
 519                (("^include_directories\\(\\$\\{OCTAVE_INCLUDE_DIRS\\}")
 520                 "include_directories(${OCTAVE_INCLUDE_DIRS} ${OCTAVE_INCLUDE_DIRS}/octave")
 521                ;; change target directory
 522                (("\\$\\{OCTAVE_OCT_LOCAL_API_FILE_DIR\\}")
 523                 (string-append (assoc-ref outputs "out")
 524                                "/share/octave/packages")))
 525              (substitute* '("src/interfaces/octave/swig_typemaps.i"
 526                             "src/interfaces/octave/sg_print_functions.cpp")
 527                ;; "octave/config.h" and "octave/oct-obj.h" deprecated in Octave.
 528                (("octave/config\\.h") "octave/octave-config.h")
 529                (("octave/oct-obj.h") "octave/ovl.h"))
 530              #t))
 531          (add-after 'unpack 'move-rxcpp
 532            (lambda* (#:key inputs #:allow-other-keys)
 533              (let ((rxcpp-dir "shogun/third-party/rxcpp"))
 534                (mkdir-p rxcpp-dir)
 535                (install-file (assoc-ref inputs "rxcpp") rxcpp-dir)
 536                #t)))
 537          (add-before 'build 'set-HOME
 538            ;; $HOME needs to be set at some point during the build phase
 539            (lambda _ (setenv "HOME" "/tmp") #t)))
 540        #:configure-flags
 541        (list "-DCMAKE_BUILD_WITH_INSTALL_RPATH=TRUE"
 542              "-DUSE_SVMLIGHT=OFF" ;disable proprietary SVMLIGHT
 543              "-DBUILD_META_EXAMPLES=OFF" ;requires unpackaged ctags
 544              ;;"-DINTERFACE_JAVA=ON" ;requires unpackaged jblas
 545              ;;"-DINTERFACE_RUBY=ON" ;requires unpackaged ruby-narray
 546              ;;"-DINTERFACE_PERL=ON" ;"FindPerlLibs" does not exist
 547              ;;"-DINTERFACE_LUA=ON"  ;fails because lua doesn't build pkgconfig file
 548              "-DINTERFACE_OCTAVE=ON"
 549              "-DINTERFACE_PYTHON=ON"
 550              "-DINTERFACE_R=ON")))
 551     (inputs
 552      `(("python" ,python)
 553        ("numpy" ,python-numpy)
 554        ("r-minimal" ,r-minimal)
 555        ("octave" ,octave-cli)
 556        ("swig" ,swig)
 557        ("eigen" ,eigen)
 558        ("hdf5" ,hdf5)
 559        ("atlas" ,atlas)
 560        ("arpack" ,arpack-ng)
 561        ("lapack" ,lapack)
 562        ("glpk" ,glpk)
 563        ("libxml2" ,libxml2)
 564        ("lzo" ,lzo)
 565        ("zlib" ,zlib)))
 566     (native-inputs
 567      `(("pkg-config" ,pkg-config)
 568        ("rxcpp" ,rxcpp)))
 569     ;; Non-portable SSE instructions are used so building fails on platforms
 570     ;; other than x86_64.
 571     (supported-systems '("x86_64-linux"))
 572     (home-page "https://shogun-toolbox.org/")
 573     (synopsis "Machine learning toolbox")
 574     (description
 575      "The Shogun Machine learning toolbox provides a wide range of unified and
 576 efficient Machine Learning (ML) methods.  The toolbox seamlessly
 577 combines multiple data representations, algorithm classes, and general purpose
 578 tools.  This enables both rapid prototyping of data pipelines and extensibility
 579 in terms of new algorithms.")
 580     (license license:gpl3+)))
 581
 582 (define-public python-onnx
 583   (package
 584     (name "python-onnx")
 585     (version "1.7.0")
 586     (source
 587      (origin
 588        (method url-fetch)
 589        (uri (pypi-uri "onnx" version))
 590        ;; ONNX will build googletest from a git checkout.  Patch CMake
 591        ;; to use googletest from Guix and enable tests by default.
 592        (patches (search-patches "python-onnx-use-system-googletest.patch"))
 593        (sha256
 594         (base32 "0j6rgfbhsw3a8id8pyg18y93k68lbjbj1kq6qia36h69f6pvlyjy"))))
 595     (build-system python-build-system)
 596     (native-inputs
 597      `(("cmake" ,cmake)
 598        ("googletest" ,googletest)
 599        ("pybind11" ,pybind11)
 600        ("python-coverage" ,python-coverage)
 601        ("python-nbval" ,python-nbval)
 602        ("python-pytest" ,python-pytest)
 603        ("python-pytest-runner" ,python-pytest-runner)))
 604     (inputs
 605      `(("protobuf" ,protobuf)))
 606     (propagated-inputs
 607      `(("python-numpy" ,python-numpy)
 608        ("python-protobuf" ,python-protobuf)
 609        ("python-six" ,python-six)
 610        ("python-tabulate" ,python-tabulate)
 611        ("python-typing-extensions"
 612         ,python-typing-extensions)))
 613     (home-page "https://onnx.ai/")
 614     (synopsis "Open Neural Network Exchange")
 615     (description
 616      "Open Neural Network Exchange (ONNX) provides an open source format for
 617 AI models, both deep learning and traditional ML.  It defines an extensible
 618 computation graph model, as well as definitions of built-in operators and
 619 standard data types.")
 620     (license license:expat)))
 621
 622 (define-public rxcpp
 623   (package
 624     (name "rxcpp")
 625     (version "4.1.0")
 626     (source
 627      (origin
 628        (method git-fetch)
 629        (uri (git-reference
 630              (url "https://github.com/ReactiveX/RxCpp")
 631              (commit (string-append "v" version))))
 632        (sha256
 633         (base32 "1rdpa3jlc181jd08nk437aar085h28i45s6nzrv65apb3xyyz0ij"))
 634        (file-name (git-file-name name version))))
 635     (build-system cmake-build-system)
 636     (arguments
 637      `(#:phases
 638        (modify-phases %standard-phases
 639          (add-after 'unpack 'remove-werror
 640            (lambda _
 641              (substitute* (find-files ".")
 642                (("-Werror") ""))
 643              #t))
 644          (replace 'check
 645            (lambda _
 646              (invoke "ctest"))))))
 647     (native-inputs
 648      `(("catch" ,catch-framework)))
 649     (home-page "http://reactivex.io/")
 650     (synopsis "Reactive Extensions for C++")
 651     (description
 652      "The Reactive Extensions for C++ (RxCpp) is a library of algorithms for
 653 values-distributed-in-time.  ReactiveX is a library for composing asynchronous
 654 and event-based programs by using observable sequences.
 655
 656 It extends the observer pattern to support sequences of data and/or events and
 657 adds operators that allow you to compose sequences together declaratively while
 658 abstracting away concerns about things like low-level threading,
 659 synchronization, thread-safety, concurrent data structures, and non-blocking
 660 I/O.")
 661     (license license:asl2.0)))
 662
 663 (define-public r-adaptivesparsity
 664   (package
 665     (name "r-adaptivesparsity")
 666     (version "1.6")
 667     (source (origin
 668               (method url-fetch)
 669               (uri (cran-uri "AdaptiveSparsity" version))
 670               (sha256
 671                (base32
 672                 "0imr5m8mll9j6n4icsv6z9rl5kbnwsp9wvzrg7n90nnmcxq2cz91"))))
 673     (properties
 674      `((upstream-name . "AdaptiveSparsity")))
 675     (build-system r-build-system)
 676     (arguments
 677      `(#:phases
 678        (modify-phases %standard-phases
 679          (add-after 'unpack 'link-against-armadillo
 680            (lambda _
 681              (substitute* "src/Makevars"
 682                (("PKG_LIBS=" prefix)
 683                 (string-append prefix "-larmadillo"))))))))
 684     (propagated-inputs
 685      `(("r-mass" ,r-mass)
 686        ("r-matrix" ,r-matrix)
 687        ("r-rcpp" ,r-rcpp)
 688        ("r-rcpparmadillo" ,r-rcpparmadillo)))
 689     (inputs
 690      `(("armadillo" ,armadillo)))
 691     (home-page "https://cran.r-project.org/web/packages/AdaptiveSparsity")
 692     (synopsis "Adaptive sparsity models")
 693     (description
 694      "This package implements the Figueiredo machine learning algorithm for
 695 adaptive sparsity and the Wong algorithm for adaptively sparse gaussian
 696 geometric models.")
 697     (license license:lgpl3+)))
 698
 699 (define-public gemmlowp-for-tensorflow
 700   ;; The commit hash is taken from "tensorflow/workspace.bzl".
 701   (let ((commit "38ebac7b059e84692f53e5938f97a9943c120d98")
 702         (revision "2"))
 703     (package
 704       (name "gemmlowp")
 705       (version (git-version "0" revision commit))
 706       (source (origin
 707                 (method url-fetch)
 708                 (uri (string-append "https://mirror.bazel.build/"
 709                                     "github.com/google/gemmlowp/archive/"
 710                                     commit ".zip"))
 711                 (file-name (string-append "gemmlowp-" version ".zip"))
 712                 (sha256
 713                  (base32
 714                   "0n56s2g8hrssm4w8qj1v58gfm56a04n9v992ixkmvk6zjiralzxq"))))
 715       (build-system cmake-build-system)
 716       (arguments
 717        `(#:configure-flags
 718          (list ,@(match (%current-system)
 719                    ((or "x86_64-linux" "i686-linux")
 720                     '("-DCMAKE_CXX_FLAGS=-msse2"))
 721                    (_ '())))
 722          #:phases
 723          (modify-phases %standard-phases
 724            ;; This directory contains the CMakeLists.txt.
 725            (add-after 'unpack 'chdir
 726              (lambda _ (chdir "contrib") #t))
 727            ;; There is no install target
 728            (replace 'install
 729              (lambda* (#:key outputs #:allow-other-keys)
 730                (let* ((out (assoc-ref outputs "out"))
 731                       (lib (string-append out "/lib/"))
 732                       (inc (string-append out "/include/")))
 733                  (install-file "../build/libeight_bit_int_gemm.so" lib)
 734                  (for-each (lambda (dir)
 735                              (let ((target (string-append inc "/" dir)))
 736                                (mkdir-p target)
 737                                (for-each (lambda (h)
 738                                            (install-file h target))
 739                                          (find-files (string-append "../" dir)
 740                                                      "\\.h$"))))
 741                            '("meta" "profiling" "public" "fixedpoint"
 742                              "eight_bit_int_gemm" "internal"))
 743                  #t))))))
 744       (native-inputs
 745        `(("unzip" ,unzip)))
 746       (home-page "https://github.com/google/gemmlowp")
 747       (synopsis "Small self-contained low-precision GEMM library")
 748       (description
 749        "This is a small self-contained low-precision @dfn{general matrix
 750 multiplication} (GEMM) library.  It is not a full linear algebra library.
 751 Low-precision means that the input and output matrix entries are integers on
 752 at most 8 bits.  To avoid overflow, results are internally accumulated on more
 753 than 8 bits, and at the end only some significant 8 bits are kept.")
 754       (license license:asl2.0))))
 755
 756 (define-public dlib
 757   (package
 758     (name "dlib")
 759     (version "19.20")
 760     (source (origin
 761               (method url-fetch)
 762               (uri (string-append
 763                     "http://dlib.net/files/dlib-" version ".tar.bz2"))
 764               (sha256
 765                (base32
 766                 "139jyi19qz37wwmmy48gil9d1kkh2r3w3bwdzabha6ayxmba96nz"))
 767               (modules '((guix build utils)))
 768               (snippet
 769                '(begin
 770                   ;; Delete ~13MB of bundled dependencies.
 771                   (delete-file-recursively "dlib/external")
 772                   (delete-file-recursively "docs/dlib/external")
 773                   #t))))
 774     (build-system cmake-build-system)
 775     (arguments
 776      `(#:phases
 777        (modify-phases %standard-phases
 778          (add-after 'unpack 'disable-asserts
 779            (lambda _
 780              ;; config.h recommends explicitly enabling or disabling asserts
 781              ;; when building as a shared library. By default neither is set.
 782              (substitute* "dlib/config.h"
 783                (("^//#define DLIB_DISABLE_ASSERTS") "#define DLIB_DISABLE_ASSERTS"))
 784              #t))
 785          (add-after 'disable-asserts 'disable-failing-tests
 786            (lambda _
 787              ;; One test times out on MIPS, so we need to disable it.
 788              ;; Others are flaky on some platforms.
 789              (let* ((system ,(or (%current-target-system)
 790                                  (%current-system)))
 791                     (disabled-tests (cond
 792                                      ((string-prefix? "mips64" system)
 793                                       '("object_detector" ; timeout
 794                                         "data_io"))
 795                                      ((string-prefix? "armhf" system)
 796                                       '("learning_to_track"))
 797                                      ((string-prefix? "i686" system)
 798                                       '("optimization"))
 799                                      (else '()))))
 800                (for-each
 801                 (lambda (test)
 802                   (substitute* "dlib/test/makefile"
 803                     (((string-append "SRC \\+= " test "\\.cpp")) "")))
 804                 disabled-tests)
 805                #t)))
 806          (replace 'check
 807            (lambda _
 808              ;; No test target, so we build and run the unit tests here.
 809              (let ((test-dir (string-append "../dlib-" ,version "/dlib/test")))
 810                (with-directory-excursion test-dir
 811                  (invoke "make" "-j" (number->string (parallel-job-count)))
 812                  (invoke "./dtest" "--runall"))
 813                #t)))
 814          (add-after 'install 'delete-static-library
 815            (lambda* (#:key outputs #:allow-other-keys)
 816              (delete-file (string-append (assoc-ref outputs "out")
 817                                          "/lib/libdlib.a"))
 818              #t)))))
 819     (native-inputs
 820      `(("pkg-config" ,pkg-config)
 821        ;; For tests.
 822        ("libnsl" ,libnsl)))
 823     (inputs
 824      `(("giflib" ,giflib)
 825        ("lapack" ,lapack)
 826        ("libjpeg" ,libjpeg-turbo)
 827        ("libpng" ,libpng)
 828        ("libx11" ,libx11)
 829        ("openblas" ,openblas)
 830        ("zlib" ,zlib)))
 831     (synopsis
 832      "Toolkit for making machine learning and data analysis applications in C++")
 833     (description
 834      "Dlib is a modern C++ toolkit containing machine learning algorithms and
 835 tools.  It is used in both industry and academia in a wide range of domains
 836 including robotics, embedded devices, mobile phones, and large high performance
 837 computing environments.")
 838     (home-page "http://dlib.net")
 839     (license license:boost1.0)))
 840
 841 (define-public python-scikit-learn
 842   (package
 843     (name "python-scikit-learn")
 844     (version "0.22.1")
 845     (source
 846      (origin
 847        (method git-fetch)
 848        (uri (git-reference
 849              (url "https://github.com/scikit-learn/scikit-learn")
 850              (commit version)))
 851        (file-name (git-file-name name version))
 852        (sha256
 853         (base32
 854          "1xqxv210gsmjw094vc5ghq2y9lmm74qkk22pq6flcjzj51b86jxf"))))
 855     (build-system python-build-system)
 856     (arguments
 857      `(#:phases
 858        (modify-phases %standard-phases
 859          (add-after 'build 'build-ext
 860            (lambda _ (invoke "python" "setup.py" "build_ext" "--inplace") #t))
 861          (replace 'check
 862            (lambda _
 863              ;; Restrict OpenBLAS threads to prevent segfaults while testing!
 864              (setenv "OPENBLAS_NUM_THREADS" "1")
 865
 866              ;; Some tests require write access to $HOME.
 867              (setenv "HOME" "/tmp")
 868
 869              (invoke "pytest" "sklearn" "-m" "not network")))
 870          (add-before 'reset-gzip-timestamps 'make-files-writable
 871            (lambda* (#:key outputs #:allow-other-keys)
 872              ;; Make sure .gz files are writable so that the
 873              ;; 'reset-gzip-timestamps' phase can do its work.
 874              (let ((out (assoc-ref outputs "out")))
 875                (for-each make-file-writable
 876                          (find-files out "\\.gz$"))
 877                #t))))))
 878     (inputs
 879      `(("openblas" ,openblas)))
 880     (native-inputs
 881      `(("python-pytest" ,python-pytest)
 882        ("python-pandas" ,python-pandas) ;for tests
 883        ("python-cython" ,python-cython)))
 884     (propagated-inputs
 885      `(("python-numpy" ,python-numpy)
 886        ("python-scipy" ,python-scipy)
 887        ("python-joblib" ,python-joblib)))
 888     (home-page "https://scikit-learn.org/")
 889     (synopsis "Machine Learning in Python")
 890     (description
 891      "Scikit-learn provides simple and efficient tools for data mining and
 892 data analysis.")
 893     (properties `((python2-variant . ,(delay python2-scikit-learn))))
 894     (license license:bsd-3)))
 895
 896 ;; scikit-learn 0.22 and later only supports Python 3, so we stick with
 897 ;; an older version here.
 898 (define-public python2-scikit-learn
 899   (let ((base (package-with-python2 (strip-python2-variant python-scikit-learn))))
 900     (package
 901       (inherit base)
 902       (version "0.20.4")
 903       (source (origin
 904                 (method git-fetch)
 905                 (uri (git-reference
 906                       (url "https://github.com/scikit-learn/scikit-learn")
 907                       (commit version)))
 908                 (file-name (git-file-name "python-scikit-learn" version))
 909                 (sha256
 910                  (base32
 911                   "08zbzi8yx5wdlxfx9jap61vg1malc9ajf576w7a0liv6jvvrxlpj")))))))
 912
 913 (define-public python-scikit-rebate
 914   (package
 915     (name "python-scikit-rebate")
 916     (version "0.6")
 917     (source (origin
 918               (method url-fetch)
 919               (uri (pypi-uri "skrebate" version))
 920               (sha256
 921                (base32
 922                 "1h7qs9gjxpzqabzhb8rmpv3jpmi5iq41kqdibg48299h94iikiw7"))))
 923     (build-system python-build-system)
 924     ;; Pandas is only needed to run the tests.
 925     (native-inputs
 926      `(("python-pandas" ,python-pandas)))
 927     (propagated-inputs
 928      `(("python-numpy" ,python-numpy)
 929        ("python-scipy" ,python-scipy)
 930        ("python-scikit-learn" ,python-scikit-learn)
 931        ("python-joblib" ,python-joblib)))
 932     (home-page "https://epistasislab.github.io/scikit-rebate/")
 933     (synopsis "Relief-based feature selection algorithms for Python")
 934     (description "Scikit-rebate is a scikit-learn-compatible Python
 935 implementation of ReBATE, a suite of Relief-based feature selection algorithms
 936 for Machine Learning.  These algorithms excel at identifying features that are
 937 predictive of the outcome in supervised learning problems, and are especially
 938 good at identifying feature interactions that are normally overlooked by
 939 standard feature selection algorithms.")
 940     (license license:expat)))
 941
 942 (define-public python-autograd
 943   (let* ((commit "442205dfefe407beffb33550846434baa90c4de7")
 944          (revision "0")
 945          (version (git-version "0.0.0" revision commit)))
 946     (package
 947       (name "python-autograd")
 948       (home-page "https://github.com/HIPS/autograd")
 949       (source (origin
 950                 (method git-fetch)
 951                 (uri (git-reference
 952                       (url home-page)
 953                       (commit commit)))
 954                 (sha256
 955                  (base32
 956                   "189sv2xb0mwnjawa9z7mrgdglc1miaq93pnck26r28fi1jdwg0z4"))
 957                 (file-name (git-file-name name version))))
 958       (version version)
 959       (build-system python-build-system)
 960       (native-inputs
 961        `(("python-nose" ,python-nose)
 962          ("python-pytest" ,python-pytest)))
 963       (propagated-inputs
 964        `(("python-future" ,python-future)
 965          ("python-numpy" ,python-numpy)))
 966       (arguments
 967        `(#:phases (modify-phases %standard-phases
 968                     (replace 'check
 969                       (lambda _
 970                         (invoke "py.test" "-v"))))))
 971       (synopsis "Efficiently computes derivatives of NumPy code")
 972       (description "Autograd can automatically differentiate native Python and
 973 NumPy code.  It can handle a large subset of Python's features, including loops,
 974 ifs, recursion and closures, and it can even take derivatives of derivatives
 975 of derivatives.  It supports reverse-mode differentiation
 976 (a.k.a. backpropagation), which means it can efficiently take gradients of
 977 scalar-valued functions with respect to array-valued arguments, as well as
 978 forward-mode differentiation, and the two can be composed arbitrarily.  The
 979 main intended application of Autograd is gradient-based optimization.")
 980       (license license:expat))))
 981
 982 (define-public python2-autograd
 983   (package-with-python2 python-autograd))
 984
 985 (define-public lightgbm
 986   (package
 987     (name "lightgbm")
 988     (version "2.0.12")
 989     (source (origin
 990               (method git-fetch)
 991               (uri (git-reference
 992                      (url "https://github.com/Microsoft/LightGBM")
 993                      (commit (string-append "v" version))))
 994               (sha256
 995                (base32
 996                 "0jlvyn7k81dzrh9ij3zw576wbgiwmmr26rzpdxjn1dbpc3njpvzi"))
 997               (file-name (git-file-name name version))))
 998     (native-inputs
 999      `(("python-pytest" ,python-pytest)
1000        ("python-nose" ,python-nose)))
1001     (inputs
1002      `(("openmpi" ,openmpi)))
1003     (propagated-inputs
1004      `(("python-numpy" ,python-numpy)
1005        ("python-scipy" ,python-scipy)))
1006     (arguments
1007      `(#:configure-flags
1008        '("-DUSE_MPI=ON")
1009        #:phases
1010        (modify-phases %standard-phases
1011          (replace 'check
1012            (lambda _
1013              (with-directory-excursion "../source"
1014                (invoke "pytest" "tests/c_api_test/test_.py")))))))
1015     (build-system cmake-build-system)
1016     (home-page "https://github.com/Microsoft/LightGBM")
1017     (synopsis "Gradient boosting framework based on decision tree algorithms")
1018     (description "LightGBM is a gradient boosting framework that uses tree
1019 based learning algorithms.  It is designed to be distributed and efficient with
1020 the following advantages:
1021
1022 @itemize
1023 @item Faster training speed and higher efficiency
1024 @item Lower memory usage
1025 @item Better accuracy
1026 @item Parallel and GPU learning supported (not enabled in this package)
1027 @item Capable of handling large-scale data
1028 @end itemize\n")
1029     (license license:expat)))
1030
1031 (define-public vowpal-wabbit
1032   ;; Language bindings not included.
1033   (package
1034     (name "vowpal-wabbit")
1035     (version "8.5.0")
1036     (source (origin
1037               (method git-fetch)
1038               (uri (git-reference
1039                      (url "https://github.com/JohnLangford/vowpal_wabbit")
1040                      (commit version)))
1041               (sha256
1042                (base32
1043                 "04bwzk6ifgnz3fmzid8b7avxf9n5pnx9xcjm61nkjng1vv0bpj8x"))
1044               (file-name (git-file-name name version))))
1045     (inputs
1046      `(("boost" ,boost)
1047        ("zlib" ,zlib)))
1048     (arguments
1049      `(#:configure-flags
1050        (list (string-append "--with-boost="
1051                             (assoc-ref %build-inputs "boost")))
1052        #:phases
1053        (modify-phases %standard-phases
1054          (add-after 'unpack 'make-files-writable
1055            (lambda _
1056              (for-each make-file-writable (find-files "." ".*")) #t)))))
1057     (build-system gnu-build-system)
1058     (home-page "https://github.com/JohnLangford/vowpal_wabbit")
1059     (synopsis "Fast machine learning library for online learning")
1060     (description "Vowpal Wabbit is a machine learning system with techniques
1061 such as online, hashing, allreduce, reductions, learning2search, active, and
1062 interactive learning.")
1063     (license license:bsd-3)))
1064
1065 (define-public python2-fastlmm
1066   (package
1067     (name "python2-fastlmm")
1068     (version "0.2.21")
1069     (source
1070      (origin
1071        (method url-fetch)
1072        (uri (pypi-uri "fastlmm" version ".zip"))
1073        (sha256
1074         (base32
1075          "1q8c34rpmwkfy3r4d5172pzdkpfryj561897z9r3x22gq7813x1m"))))
1076     (build-system python-build-system)
1077     (arguments
1078      `(#:tests? #f ; some test files are missing
1079        #:python ,python-2)) ; only Python 2.7 is supported
1080     (propagated-inputs
1081      `(("python2-numpy" ,python2-numpy)
1082        ("python2-scipy" ,python2-scipy)
1083        ("python2-matplotlib" ,python2-matplotlib)
1084        ("python2-pandas" ,python2-pandas)
1085        ("python2-scikit-learn" ,python2-scikit-learn)
1086        ("python2-pysnptools" ,python2-pysnptools)))
1087     (native-inputs
1088      `(("unzip" ,unzip)
1089        ("python2-cython" ,python2-cython)
1090        ("python2-mock" ,python2-mock)
1091        ("python2-nose" ,python2-nose)))
1092     (home-page "http://research.microsoft.com/en-us/um/redmond/projects/mscompbio/fastlmm/")
1093     (synopsis "Perform genome-wide association studies on large data sets")
1094     (description
1095      "FaST-LMM, which stands for Factored Spectrally Transformed Linear Mixed
1096 Models, is a program for performing both single-SNP and SNP-set genome-wide
1097 association studies (GWAS) on extremely large data sets.")
1098     (license license:asl2.0)))
1099
1100 ;; There have been no proper releases yet.
1101 (define-public kaldi
1102   (let ((commit "d4791c0f3fc1a09c042dac365e120899ee2ad21e")
1103         (revision "2"))
1104     (package
1105       (name "kaldi")
1106       (version (git-version "0" revision commit))
1107       (source (origin
1108                 (method git-fetch)
1109                 (uri (git-reference
1110                       (url "https://github.com/kaldi-asr/kaldi")
1111                       (commit commit)))
1112                 (file-name (git-file-name name version))
1113                 (sha256
1114                  (base32
1115                   "07k80my6f19mhrkwbzhjsnpf9871wmrwkl0ym468i830w67qyjrz"))))
1116       (build-system gnu-build-system)
1117       (arguments
1118        `(#:test-target "test"
1119          #:phases
1120          (modify-phases %standard-phases
1121            (add-after 'unpack 'chdir
1122              (lambda _ (chdir "src") #t))
1123            (replace 'configure
1124              (lambda* (#:key build system inputs outputs #:allow-other-keys)
1125                (when (not (or (string-prefix? "x86_64" system)
1126                               (string-prefix? "i686" system)))
1127                  (substitute* "makefiles/linux_openblas.mk"
1128                    (("-msse -msse2") "")))
1129                (substitute* "makefiles/default_rules.mk"
1130                  (("/bin/bash") (which "bash")))
1131                (substitute* "Makefile"
1132                  (("ext_depend: check_portaudio")
1133                   "ext_depend:"))
1134                (substitute* '("online/Makefile"
1135                               "onlinebin/Makefile"
1136                               "gst-plugin/Makefile")
1137                  (("../../tools/portaudio/install")
1138                   (assoc-ref inputs "portaudio")))
1139
1140                ;; This `configure' script doesn't support variables passed as
1141                ;; arguments, nor does it support "prefix".
1142                (let ((out (assoc-ref outputs "out"))
1143                      (openblas (assoc-ref inputs "openblas"))
1144                      (openfst (assoc-ref inputs "openfst")))
1145                  (substitute* "configure"
1146                    (("check_for_slow_expf;") "")
1147                    ;; This affects the RPATH and also serves as the installation
1148                    ;; directory.
1149                    (("KALDILIBDIR=`pwd`/lib")
1150                     (string-append "KALDILIBDIR=" out "/lib")))
1151                  (mkdir-p out) ; must exist
1152                  (setenv "CONFIG_SHELL" (which "bash"))
1153                  (setenv "OPENFST_VER" ,(package-version openfst))
1154                  (invoke "./configure"
1155                          "--use-cuda=no"
1156                          "--shared"
1157                          (string-append "--openblas-root=" openblas)
1158                          (string-append "--fst-root=" openfst)))))
1159            (add-after 'build 'build-ext-and-gstreamer-plugin
1160              (lambda _
1161                (invoke "make" "-C" "online" "depend")
1162                (invoke "make" "-C" "online")
1163                (invoke "make" "-C" "onlinebin" "depend")
1164                (invoke "make" "-C" "onlinebin")
1165                (invoke "make" "-C" "gst-plugin" "depend")
1166                (invoke "make" "-C" "gst-plugin")
1167                #t))
1168            ;; TODO: also install the executables.
1169            (replace 'install
1170              (lambda* (#:key outputs #:allow-other-keys)
1171                (let* ((out (assoc-ref outputs "out"))
1172                       (inc (string-append out "/include"))
1173                       (lib (string-append out "/lib")))
1174                  (mkdir-p lib)
1175                  ;; The build phase installed symlinks to the actual
1176                  ;; libraries.  Install the actual targets.
1177                  (for-each (lambda (file)
1178                              (let ((target (readlink file)))
1179                                (delete-file file)
1180                                (install-file target lib)))
1181                            (find-files lib "\\.so"))
1182                  ;; Install headers
1183                  (for-each (lambda (file)
1184                              (let ((target-dir (string-append inc "/" (dirname file))))
1185                                (install-file file target-dir)))
1186                            (find-files "." "\\.h"))
1187                  (install-file "gst-plugin/libgstonlinegmmdecodefaster.so"
1188                                (string-append lib "/gstreamer-1.0"))
1189                  #t))))))
1190       (inputs
1191        `(("alsa-lib" ,alsa-lib)
1192          ("gfortran" ,gfortran "lib")
1193          ("glib" ,glib)
1194          ("gstreamer" ,gstreamer)
1195          ("jack" ,jack-1)
1196          ("openblas" ,openblas)
1197          ("openfst" ,openfst)
1198          ("portaudio" ,portaudio)
1199          ("python" ,python)))
1200       (native-inputs
1201        `(("glib" ,glib "bin")             ; glib-genmarshal
1202          ("grep" ,grep)
1203          ("sed" ,sed)
1204          ("pkg-config" ,pkg-config)
1205          ("which" ,which)))
1206       (home-page "https://kaldi-asr.org/")
1207       (synopsis "Speech recognition toolkit")
1208       (description "Kaldi is an extensible toolkit for speech recognition
1209 written in C++.")
1210       (license license:asl2.0))))
1211
1212 (define-public gst-kaldi-nnet2-online
1213   (let ((commit "cb227ef43b66a9835c14eb0ad39e08ee03c210ad")
1214         (revision "2"))
1215     (package
1216       (name "gst-kaldi-nnet2-online")
1217       (version (git-version "0" revision commit))
1218       (source (origin
1219                 (method git-fetch)
1220                 (uri (git-reference
1221                       (url "https://github.com/alumae/gst-kaldi-nnet2-online")
1222                       (commit commit)))
1223                 (file-name (git-file-name name version))
1224                 (sha256
1225                  (base32
1226                   "1i6ffwiavxx07ri0lxix6s8q0r31x7i4xxvhys5jxkixf5q34w8g"))))
1227       (build-system gnu-build-system)
1228       (arguments
1229        `(#:tests? #f                    ; there are none
1230          #:make-flags
1231          (list (string-append "SHELL="
1232                               (assoc-ref %build-inputs "bash") "/bin/bash")
1233                (string-append "KALDI_ROOT="
1234                               (assoc-ref %build-inputs "kaldi-src"))
1235                (string-append "KALDILIBDIR="
1236                               (assoc-ref %build-inputs "kaldi") "/lib")
1237                "KALDI_FLAVOR=dynamic")
1238          #:phases
1239          (modify-phases %standard-phases
1240            (add-after 'unpack 'chdir
1241              (lambda _ (chdir "src") #t))
1242            (replace 'configure
1243              (lambda* (#:key inputs #:allow-other-keys)
1244                (let ((glib (assoc-ref inputs "glib")))
1245                  (setenv "CXXFLAGS" "-fPIC")
1246                  (setenv "CPLUS_INCLUDE_PATH"
1247                          (string-append glib "/include/glib-2.0:"
1248                                         glib "/lib/glib-2.0/include:"
1249                                         (assoc-ref inputs "gstreamer")
1250                                         "/include/gstreamer-1.0")))
1251                (substitute* "Makefile"
1252                  (("include \\$\\(KALDI_ROOT\\)/src/kaldi.mk") "")
1253                  (("\\$\\(error Cannot find") "#"))
1254                #t))
1255            (add-before 'build 'build-depend
1256              (lambda* (#:key make-flags #:allow-other-keys)
1257                (apply invoke "make" "depend" make-flags)))
1258            (replace 'install
1259              (lambda* (#:key outputs #:allow-other-keys)
1260                (let* ((out (assoc-ref outputs "out"))
1261                       (lib (string-append out "/lib/gstreamer-1.0")))
1262                  (install-file "libgstkaldinnet2onlinedecoder.so" lib)
1263                  #t))))))
1264       (inputs
1265        `(("glib" ,glib)
1266          ("gstreamer" ,gstreamer)
1267          ("jansson" ,jansson)
1268          ("openfst" ,openfst)
1269          ("kaldi" ,kaldi)))
1270       (native-inputs
1271        `(("bash" ,bash)
1272          ("glib:bin" ,glib "bin")       ; glib-genmarshal
1273          ("kaldi-src" ,(package-source kaldi))
1274          ("pkg-config" ,pkg-config)))
1275       (home-page "https://kaldi-asr.org/")
1276       (synopsis "Gstreamer plugin for decoding speech")
1277       (description "This package provides a GStreamer plugin that wraps
1278 Kaldi's @code{SingleUtteranceNnet2Decoder}.  It requires iVector-adapted DNN
1279 acoustic models.  The iVectors are adapted to the current audio stream
1280 automatically.")
1281       (license license:asl2.0))))
1282
1283 (define-public kaldi-gstreamer-server
1284   ;; This is the tip of the py3 branch
1285   (let ((commit "f68cab490be7eb0da2af1475fbc16655f50a60cb")
1286         (revision "2"))
1287     (package
1288       (name "kaldi-gstreamer-server")
1289       (version (git-version "0" revision commit))
1290       (source (origin
1291                 (method git-fetch)
1292                 (uri (git-reference
1293                       (url "https://github.com/alumae/kaldi-gstreamer-server")
1294                       (commit commit)))
1295                 (file-name (git-file-name name version))
1296                 (sha256
1297                  (base32
1298                   "17lh1368vkg8ngrcbn2phvigzlmalrqg6djx2gg61qq1a0nj87dm"))))
1299       (build-system gnu-build-system)
1300       (arguments
1301        `(#:tests? #f ; there are no tests that can be run automatically
1302          #:modules ((guix build utils)
1303                     (guix build gnu-build-system)
1304                     (srfi srfi-26))
1305          #:phases
1306          (modify-phases %standard-phases
1307            (delete 'configure)
1308            (replace 'build
1309              (lambda* (#:key outputs #:allow-other-keys)
1310                ;; Disable hash randomization to ensure the generated .pycs
1311                ;; are reproducible.
1312                (setenv "PYTHONHASHSEED" "0")
1313                (with-directory-excursion "kaldigstserver"
1314                  ;; See https://github.com/alumae/kaldi-gstreamer-server/issues/232
1315                  (substitute* "master_server.py"
1316                    (("\\.replace\\('\\\\.*") ")"))
1317
1318                  ;; This is a Python 2 file
1319                  (delete-file "decoder_test.py")
1320                  (delete-file "test-buffer.py")
1321
1322                  (for-each (lambda (file)
1323                              (apply invoke
1324                                     `("python"
1325                                       "-m" "compileall"
1326                                       "-f" ; force rebuild
1327                                       ,file)))
1328                            (find-files "." "\\.py$")))
1329                #t))
1330            (replace 'install
1331              (lambda* (#:key inputs outputs #:allow-other-keys)
1332                (let* ((out (assoc-ref outputs "out"))
1333                       (bin (string-append out "/bin"))
1334                       (share (string-append out "/share/kaldi-gstreamer-server/")))
1335                  ;; Install Python files
1336                  (with-directory-excursion "kaldigstserver"
1337                    (for-each (cut install-file <> share)
1338                              (find-files "." ".*")))
1339
1340                  ;; Install sample configuration files
1341                  (for-each (cut install-file <> share)
1342                            (find-files "." "\\.yaml"))
1343
1344                  ;; Install executables
1345                  (mkdir-p bin)
1346                  (let* ((server (string-append bin "/kaldi-gst-server"))
1347                         (client (string-append bin "/kaldi-gst-client"))
1348                         (worker (string-append bin "/kaldi-gst-worker"))
1349                         (PYTHONPATH (getenv "PYTHONPATH"))
1350                         (GST_PLUGIN_PATH (string-append
1351                                           (assoc-ref inputs "gst-kaldi-nnet2-online")
1352                                           "/lib/gstreamer-1.0:${GST_PLUGIN_PATH}"))
1353                         (wrap (lambda (wrapper what)
1354                                 (with-output-to-file wrapper
1355                                   (lambda _
1356                                     (format #t
1357                                             "#!~a
1358 export PYTHONPATH=~a
1359 export GST_PLUGIN_PATH=~a
1360 exec ~a ~a/~a \"$@\"~%"
1361                                             (which "bash") PYTHONPATH GST_PLUGIN_PATH
1362                                             (which "python") share what)))
1363                                 (chmod wrapper #o555))))
1364                    (for-each wrap
1365                              (list server client worker)
1366                              (list "master_server.py"
1367                                    "client.py"
1368                                    "worker.py")))
1369                  #t))))))
1370       (inputs
1371        `(("gst-kaldi-nnet2-online" ,gst-kaldi-nnet2-online)
1372          ("python" ,python-wrapper)
1373          ("python-pygobject" ,python-pygobject)
1374          ("python-pyyaml" ,python-pyyaml)
1375          ("python-tornado" ,python-tornado-6)))
1376       (home-page "https://github.com/alumae/kaldi-gstreamer-server")
1377       (synopsis "Real-time full-duplex speech recognition server")
1378       (description "This is a real-time full-duplex speech recognition server,
1379 based on the Kaldi toolkit and the GStreamer framework and implemented in
1380 Python.")
1381       (license license:bsd-2))))
1382
1383 ;; Note that Tensorflow includes a "third_party" directory, which seems to not
1384 ;; only contain modified subsets of upstream library source code, but also
1385 ;; adapter headers provided by Google (such as the fft.h header, which is not
1386 ;; part of the upstream project code).  The Tensorflow code includes headers
1387 ;; from the "third_party" directory.  It does not look like we can replace
1388 ;; these headers with unmodified upstream files, so we keep them.
1389 (define-public tensorflow
1390   (package
1391     (name "tensorflow")
1392     (version "1.9.0")
1393     (source
1394      (origin
1395        (method git-fetch)
1396        (uri (git-reference
1397              (url "https://github.com/tensorflow/tensorflow")
1398              (commit (string-append "v" version))))
1399        (file-name (string-append "tensorflow-" version "-checkout"))
1400        (sha256
1401         (base32
1402          "0a9kwha395g3wgxfwln5j8vn9nkspmd75xldrlqdq540w996g8xa"))))
1403     (build-system cmake-build-system)
1404     (arguments
1405      `(#:tests? #f                      ; no "check" target
1406        #:build-type "Release"
1407        #:configure-flags
1408        (let ((protobuf (assoc-ref %build-inputs "protobuf"))
1409              (protobuf:native (assoc-ref %build-inputs "protobuf:native"))
1410              (jsoncpp (assoc-ref %build-inputs "jsoncpp"))
1411              (snappy (assoc-ref %build-inputs "snappy"))
1412              (sqlite (assoc-ref %build-inputs "sqlite")))
1413          (list
1414           ;; Use protobuf from Guix
1415           (string-append "-Dprotobuf_STATIC_LIBRARIES="
1416                          protobuf "/lib/libprotobuf.so")
1417           (string-append "-DPROTOBUF_PROTOC_EXECUTABLE="
1418                          protobuf:native "/bin/protoc")
1419
1420           ;; Use snappy from Guix
1421           (string-append "-Dsnappy_STATIC_LIBRARIES="
1422                          snappy "/lib/libsnappy.so")
1423           ;; Yes, this is not actually the include directory but a prefix...
1424           (string-append "-Dsnappy_INCLUDE_DIR=" snappy)
1425
1426           ;; Use jsoncpp from Guix
1427           (string-append "-Djsoncpp_STATIC_LIBRARIES="
1428                          jsoncpp "/lib/libjsoncpp.so")
1429           ;; Yes, this is not actually the include directory but a prefix...
1430           (string-append "-Djsoncpp_INCLUDE_DIR=" jsoncpp)
1431
1432           ;; Use sqlite from Guix
1433           (string-append "-Dsqlite_STATIC_LIBRARIES="
1434                          sqlite "/lib/libsqlite.a")
1435
1436           ;; Use system libraries wherever possible.  Currently, this
1437           ;; only affects zlib.
1438           "-Dsystemlib_ALL=ON"
1439           "-Dtensorflow_ENABLE_POSITION_INDEPENDENT_CODE=ON"
1440           "-Dtensorflow_BUILD_SHARED_LIB=ON"
1441           "-Dtensorflow_OPTIMIZE_FOR_NATIVE_ARCH=OFF"
1442           "-Dtensorflow_ENABLE_SSL_SUPPORT=OFF"
1443           "-Dtensorflow_BUILD_CONTRIB_KERNELS=OFF"))
1444        #:make-flags
1445        (list "CC=gcc")
1446        #:modules ((ice-9 ftw)
1447                   (guix build utils)
1448                   (guix build cmake-build-system)
1449                   ((guix build python-build-system)
1450                    #:select (python-version)))
1451        #:imported-modules (,@%cmake-build-system-modules
1452                             (guix build python-build-system))
1453        #:phases
1454        (modify-phases %standard-phases
1455          (add-after 'unpack 'set-source-file-times-to-1980
1456            ;; At the end of the tf_python_build_pip_package target, a ZIP
1457            ;; archive should be generated via bdist_wheel, but it fails with
1458            ;; "ZIP does not support timestamps before 1980".  Luckily,
1459            ;; SOURCE_DATE_EPOCH is respected, which we set to some time in
1460            ;; 1980.
1461            (lambda _ (setenv "SOURCE_DATE_EPOCH" "315532800") #t))
1462          ;; See https://github.com/tensorflow/tensorflow/issues/20517#issuecomment-406373913
1463          (add-after 'unpack 'python3.7-compatibility
1464            (lambda _
1465              (substitute* '("tensorflow/python/eager/pywrap_tfe_src.cc"
1466                             "tensorflow/python/lib/core/ndarray_tensor.cc"
1467                             "tensorflow/python/lib/core/py_func.cc")
1468                (("PyUnicode_AsUTF8") "(char *)PyUnicode_AsUTF8"))
1469              (substitute* "tensorflow/c/eager/c_api.h"
1470                (("unsigned char async")
1471                 "unsigned char is_async"))
1472
1473              ;; Remove dependency on tensorboard, a complicated but probably
1474              ;; optional package.
1475              (substitute* "tensorflow/tools/pip_package/setup.py"
1476                ((".*'tensorboard >.*") ""))
1477
1478              ;; Fix the build with python-3.8, taken from rejected upstream patch:
1479              ;; https://github.com/tensorflow/tensorflow/issues/34197
1480              (substitute* (find-files "tensorflow/python" ".*\\.cc$")
1481                (("(nullptr,)(\\ +/. tp_print)" _ _ tp_print)
1482                 (string-append "NULL,   " tp_print)))
1483              #t))
1484          (add-after 'python3.7-compatibility 'chdir
1485            (lambda _ (chdir "tensorflow/contrib/cmake") #t))
1486          (add-after 'chdir 'disable-downloads
1487            (lambda* (#:key inputs #:allow-other-keys)
1488              (substitute* (find-files "external" "\\.cmake$")
1489                (("GIT_REPOSITORY.*") "")
1490                (("GIT_TAG.*") "")
1491                (("PREFIX ")
1492                 "DOWNLOAD_COMMAND \"\"\nPREFIX "))
1493
1494              ;; Use packages from Guix
1495              (let ((grpc (assoc-ref inputs "grpc")))
1496                (substitute* "CMakeLists.txt"
1497                  ;; Sqlite
1498                  (("include\\(sqlite\\)") "")
1499                  (("\\$\\{sqlite_STATIC_LIBRARIES\\}")
1500                   (string-append (assoc-ref inputs "sqlite")
1501                                  "/lib/libsqlite3.so"))
1502                  (("sqlite_copy_headers_to_destination") "")
1503
1504                  ;; PNG
1505                  (("include\\(png\\)") "")
1506                  (("\\$\\{png_STATIC_LIBRARIES\\}")
1507                   (string-append (assoc-ref inputs "libpng")
1508                                  "/lib/libpng16.so"))
1509                  (("png_copy_headers_to_destination") "")
1510
1511                  ;; JPEG
1512                  (("include\\(jpeg\\)") "")
1513                  (("\\$\\{jpeg_STATIC_LIBRARIES\\}")
1514                   (string-append (assoc-ref inputs "libjpeg")
1515                                  "/lib/libjpeg.so"))
1516                  (("jpeg_copy_headers_to_destination") "")
1517
1518                  ;; GIF
1519                  (("include\\(gif\\)") "")
1520                  (("\\$\\{gif_STATIC_LIBRARIES\\}")
1521                   (string-append (assoc-ref inputs "giflib")
1522                                  "/lib/libgif.so"))
1523                  (("gif_copy_headers_to_destination") "")
1524
1525                  ;; lmdb
1526                  (("include\\(lmdb\\)") "")
1527                  (("\\$\\{lmdb_STATIC_LIBRARIES\\}")
1528                   (string-append (assoc-ref inputs "lmdb")
1529                                  "/lib/liblmdb.so"))
1530                  (("lmdb_copy_headers_to_destination") "")
1531
1532                  ;; Protobuf
1533                  (("include\\(protobuf\\)") "")
1534                  (("protobuf_copy_headers_to_destination") "")
1535                  (("^ +protobuf") "")
1536
1537                  ;; gRPC
1538                  (("include\\(grpc\\)")
1539                   "find_package(grpc REQUIRED NAMES gRPC)")
1540                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES grpc\\)") "")
1541
1542                  ;; Eigen
1543                  (("include\\(eigen\\)")
1544                   (string-append "find_package(eigen REQUIRED NAMES Eigen3)
1545 set(eigen_INCLUDE_DIRS ${CMAKE_CURRENT_BINARY_DIR}/external/eigen_archive "
1546                                  (assoc-ref inputs "eigen") "/include/eigen3)"))
1547                  (("^ +eigen") "")
1548
1549                  ;; snappy
1550                  (("include\\(snappy\\)")
1551                   "add_definitions(-DTF_USE_SNAPPY)")
1552                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES snappy\\)") "")
1553
1554                  ;; jsoncpp
1555                  (("include\\(jsoncpp\\)") "")
1556                  (("^ +jsoncpp") ""))
1557
1558                (substitute* "tf_core_framework.cmake"
1559                  ((" grpc") "")
1560                  (("\\$\\{GRPC_BUILD\\}/grpc_cpp_plugin")
1561                   (which "grpc_cpp_plugin"))
1562                  ;; Link with gRPC libraries
1563                  (("add_library\\(tf_protos_cc.*" m)
1564                   (string-append m
1565                                  (format #f "\ntarget_link_libraries(tf_protos_cc PRIVATE \
1566 ~a/lib/libgrpc++_unsecure.a \
1567 ~a/lib/libgrpc_unsecure.a \
1568 ~a/lib/libaddress_sorting.a \
1569 ~a/lib/libgpr.a \
1570 ~a//lib/libcares.so
1571 )\n"
1572                                          grpc grpc grpc grpc
1573                                          (assoc-ref inputs "c-ares"))))))
1574              (substitute* "tf_tools.cmake"
1575                (("add_dependencies\\(\\$\\{proto_text.*") ""))
1576              ;; Remove dependency on bundled grpc
1577              (substitute* "tf_core_distributed_runtime.cmake"
1578                (("tf_core_cpu grpc") "tf_core_cpu"))
1579
1580              ;; This directory is a dependency of many targets.
1581              (mkdir-p "protobuf")
1582              #t))
1583          (add-after 'configure 'unpack-third-party-sources
1584            (lambda* (#:key inputs #:allow-other-keys)
1585              ;; This is needed to configure bundled packages properly.
1586              (setenv "CONFIG_SHELL" (which "bash"))
1587              (for-each
1588               (lambda (name)
1589                 (let* ((what  (assoc-ref inputs (string-append name "-src")))
1590                        (name* (string-map (lambda (c)
1591                                             (if (char=? c #\-)
1592                                                 #\_ c)) name))
1593                        (where (string-append "../build/" name* "/src/" name*)))
1594                   (cond
1595                    ((string-suffix? ".zip" what)
1596                     (mkdir-p where)
1597                     (with-directory-excursion where
1598                       (invoke "unzip" what)))
1599                    ((string-suffix? ".tar.gz" what)
1600                     (mkdir-p where)
1601                     (invoke "tar" "xf" what
1602                             "-C" where "--strip-components=1"))
1603                    (else
1604                     (let ((parent (dirname where)))
1605                       (mkdir-p parent)
1606                       (with-directory-excursion parent
1607                         (when (file-exists? name*)
1608                           (delete-file-recursively name*))
1609                         (copy-recursively what name*)
1610                         (map make-file-writable
1611                              (find-files name* ".*"))))))))
1612               (list "boringssl"
1613                     "cub"
1614                     "double-conversion"
1615                     "farmhash"
1616                     "fft2d"
1617                     "highwayhash"
1618                     "nsync"
1619                     "re2"))
1620
1621              (rename-file "../build/cub/src/cub/cub-1.8.0/"
1622                           "../build/cub/src/cub/cub/")
1623              #t))
1624          (add-after 'unpack 'fix-python-build
1625            (lambda* (#:key inputs outputs #:allow-other-keys)
1626              (mkdir-p "protobuf-src")
1627              (invoke "tar" "xf" (assoc-ref inputs "protobuf:src")
1628                      "-C" "protobuf-src" "--strip-components=1")
1629              (mkdir-p "eigen-src")
1630              (invoke "tar" "xf" (assoc-ref inputs "eigen:src")
1631                      "-C" "eigen-src" "--strip-components=1")
1632
1633              (substitute* "tensorflow/contrib/cmake/tf_python.cmake"
1634                ;; Ensure that all Python dependencies can be found at build time.
1635                (("PYTHONPATH=\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/tf_python" m)
1636                 (string-append m ":" (getenv "PYTHONPATH")))
1637                ;; Take protobuf source files from our source package.
1638                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/protobuf/src/protobuf/src/google")
1639                 (string-append (getcwd) "/protobuf-src/src/google")))
1640
1641              (substitute* '("tensorflow/contrib/cmake/tf_shared_lib.cmake"
1642                             "tensorflow/contrib/cmake/tf_python.cmake")
1643                ;; Take Eigen source files from our source package.
1644                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/eigen/src/eigen/")
1645                 (string-append (getcwd) "/eigen-src/"))
1646                ;; Take Eigen headers from our own package.
1647                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/external/eigen_archive")
1648                 (string-append (assoc-ref inputs "eigen") "/include/eigen3")))
1649
1650              ;; Correct the RUNPATH of ops libraries generated for Python.
1651              ;; TODO: this doesn't work :(
1652              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1653              ;; warning: RUNPATH contains bogus entries: ("/tmp/guix-build-tensorflow-1.9.0.drv-0/source/tensorflow/contrib/build")
1654              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1655              ;; error: depends on 'libpywrap_tensorflow_internal.so', which
1656              ;; cannot be found in RUNPATH ...
1657              (substitute* "tensorflow/contrib/cmake/tf_cc_ops.cmake"
1658                (("set_target_properties.*")
1659                 (string-append "set_target_properties(${_AT_TARGET} PROPERTIES \
1660 COMPILE_FLAGS ${target_compile_flags} \
1661 INSTALL_RPATH_USE_LINK_PATH TRUE \
1662 INSTALL_RPATH " (assoc-ref outputs "out") "/lib)\n")))
1663              #t))
1664          (add-after 'build 'build-pip-package
1665            (lambda* (#:key outputs #:allow-other-keys)
1666              (setenv "LDFLAGS"
1667                      (string-append "-Wl,-rpath="
1668                                     (assoc-ref outputs "out") "/lib"))
1669              (invoke "make" "tf_python_build_pip_package")
1670              #t))
1671          (add-after 'build-pip-package 'install-python
1672            (lambda* (#:key inputs outputs #:allow-other-keys)
1673              (let ((out (assoc-ref outputs "out"))
1674                    (wheel (car (find-files "../build/tf_python/dist/" "\\.whl$")))
1675                    (python-version (python-version
1676                                      (assoc-ref inputs "python"))))
1677                (invoke "python" "-m" "pip" "install" wheel
1678                        (string-append "--prefix=" out))
1679
1680                ;; XXX: broken RUNPATH, see fix-python-build phase.
1681                (delete-file
1682                 (string-append
1683                  out "/lib/python" python-version
1684                  "/site-packages/tensorflow/contrib/"
1685                  "seq2seq/python/ops/lib_beam_search_ops.so"))
1686                #t))))))
1687     (native-inputs
1688      `(("pkg-config" ,pkg-config)
1689        ("protobuf:native" ,protobuf-3.6) ; protoc
1690        ("protobuf:src" ,(package-source protobuf-3.6))
1691        ("eigen:src" ,(package-source eigen-for-tensorflow))
1692        ;; install_pip_packages.sh wants setuptools 39.1.0 specifically.
1693        ("python-setuptools" ,python-setuptools-for-tensorflow)
1694
1695        ;; The commit hashes and URLs for third-party source code are taken
1696        ;; from "tensorflow/workspace.bzl".
1697        ("boringssl-src"
1698         ,(let ((commit "ee7aa02")
1699                (revision "1"))
1700            (origin
1701              (method git-fetch)
1702              (uri (git-reference
1703                    (url "https://boringssl.googlesource.com/boringssl")
1704                    (commit commit)))
1705              (file-name (string-append "boringssl-0-" revision
1706                                        (string-take commit 7)
1707                                        "-checkout"))
1708              (sha256
1709               (base32
1710                "1jf693q0nw0adsic6cgmbdx6g7wr4rj4vxa8j1hpn792fqhd8wgw")))))
1711        ("cub-src"
1712         ,(let ((version "1.8.0"))
1713            (origin
1714              (method url-fetch)
1715              (uri (string-append "https://mirror.bazel.build/github.com/NVlabs/"
1716                                  "cub/archive/" version ".zip"))
1717              (file-name (string-append "cub-" version ".zip"))
1718              (sha256
1719               (base32
1720                "1hsqikqridb90dkxkjr2918dcry6pfh46ccnwrzawl56aamhdykb")))))
1721        ("double-conversion-src"
1722         ,(let ((commit "5664746")
1723                (revision "1"))
1724            (origin
1725              (method git-fetch)
1726              (uri (git-reference
1727                    (url "https://github.com/google/double-conversion")
1728                    (commit commit)))
1729              (file-name
1730               (git-file-name "double-conversion"
1731                              (string-append "0-" revision "."
1732                                             (string-take commit 7))))
1733              (sha256
1734               (base32
1735                "1h5lppqqxcvdg5jq42i5msgwx20ryij3apvmndflngrgdpc04gn1")))))
1736        ("farmhash-src"
1737         ,(let ((commit "816a4ae622e964763ca0862d9dbd19324a1eaf45"))
1738            (origin
1739              (method url-fetch)
1740              (uri (string-append
1741                    "https://mirror.bazel.build/github.com/google/farmhash/archive/"
1742                    commit ".tar.gz"))
1743              (file-name (string-append "farmhash-0-" (string-take commit 7)
1744                                        ".tar.gz"))
1745              (sha256
1746               (base32
1747                "185b2xdxl4d4cnsnv6abg8s22gxvx8673jq2yaq85bz4cdy58q35")))))
1748        ;; The license notice on the home page at
1749        ;; http://www.kurims.kyoto-u.ac.jp/~ooura/fft.html says:
1750        ;;   Copyright Takuya OOURA, 1996-2001
1751        ;;
1752        ;;   You may use, copy, modify and distribute this code for any purpose
1753        ;;   (include commercial use) and without fee. Please refer to this
1754        ;;   package when you modify this code.
1755        ;;
1756        ;; We take the identical tarball from the Bazel mirror, because the URL
1757        ;; at the home page is not versioned and might change.
1758        ("fft2d-src"
1759         ,(origin
1760            (method url-fetch)
1761            (uri "https://mirror.bazel.build/www.kurims.kyoto-u.ac.jp/~ooura/fft.tgz")
1762            (file-name "fft2d.tar.gz")
1763            (sha256
1764             (base32
1765              "15jjkfvhqvl2c0753d2di8hz0pyzn598g74wqy79awdrf1y67fsj"))))
1766        ("highwayhash-src"
1767         ,(let ((commit "be5edafc2e1a455768e260ccd68ae7317b6690ee")
1768                (revision "1"))
1769            (origin
1770              (method git-fetch)
1771              (uri (git-reference
1772                    (url "https://github.com/google/highwayhash")
1773                    (commit commit)))
1774              (file-name (string-append "highwayhash-0-" revision
1775                                        (string-take commit 7)
1776                                        "-checkout"))
1777              (sha256
1778               (base32
1779                "154jwf98cyy54hldr94pgjn85zynly3abpnc1avmb8a18lzwjyb6")))))
1780        ("nsync-src"
1781         ,(let ((version "0559ce013feac8db639ee1bf776aca0325d28777")
1782                (revision "1"))
1783            (origin
1784              (method url-fetch)
1785              (uri (string-append "https://mirror.bazel.build/"
1786                                  "github.com/google/nsync/archive/"
1787                                  version ".tar.gz"))
1788              (file-name (string-append "nsync-0." revision
1789                                        "-" (string-take version 7)
1790                                        ".tar.gz"))
1791              (sha256
1792               (base32
1793                "0qdkyqym34x739mmzv97ah5r7ph462v5xkxqxvidmcfqbi64b132")))))
1794        ("re2-src"
1795         ,(let ((commit "e7efc48")
1796                (revision "1"))
1797            (origin
1798              (method git-fetch)
1799              (uri (git-reference
1800                    (url "https://github.com/google/re2")
1801                    (commit commit)))
1802              (file-name (string-append "re2-0-" revision
1803                                        (string-take commit 7)
1804                                        "-checkout"))
1805              (sha256
1806               (base32
1807                "161g9841rjfsy5pn52fcis0s9hdr7rxvb06pad38j5rppfihvign")))))
1808        ("googletest" ,googletest)
1809        ("swig" ,swig)
1810        ("unzip" ,unzip)))
1811     (propagated-inputs
1812      `(("python-absl-py" ,python-absl-py)
1813        ("python-astor" ,python-astor)
1814        ("python-gast" ,python-gast)
1815        ("python-grpcio" ,python-grpcio)
1816        ("python-numpy" ,python-numpy)
1817        ("python-protobuf" ,python-protobuf-3.6)
1818        ("python-six" ,python-six)
1819        ("python-termcolo" ,python-termcolor)
1820        ("python-wheel" ,python-wheel)))
1821     (inputs
1822      `(("c-ares" ,c-ares)
1823        ("eigen" ,eigen-for-tensorflow)
1824        ("gemmlowp" ,gemmlowp-for-tensorflow)
1825        ("lmdb" ,lmdb)
1826        ("libjpeg" ,libjpeg-turbo)
1827        ("libpng" ,libpng)
1828        ("giflib" ,giflib)
1829        ("grpc" ,grpc-1.16.1 "static")
1830        ("grpc:bin" ,grpc-1.16.1)
1831        ("jsoncpp" ,jsoncpp-for-tensorflow)
1832        ("snappy" ,snappy)
1833        ("sqlite" ,sqlite)
1834        ("protobuf" ,protobuf-3.6)
1835        ("python" ,python-wrapper)
1836        ("zlib" ,zlib)))
1837     (home-page "https://tensorflow.org")
1838     (synopsis "Machine learning framework")
1839     (description
1840      "TensorFlow is a flexible platform for building and training machine
1841 learning models.  It provides a library for high performance numerical
1842 computation and includes high level Python APIs, including both a sequential
1843 API for beginners that allows users to build models quickly by plugging
1844 together building blocks and a subclassing API with an imperative style for
1845 advanced research.")
1846     (license license:asl2.0)))
1847
1848 (define-public python-iml
1849   (package
1850     (name "python-iml")
1851     (version "0.6.2")
1852     (source
1853      (origin
1854        (method url-fetch)
1855        (uri (pypi-uri "iml" version))
1856        (sha256
1857         (base32
1858          "1k8szlpm19rcwcxdny9qdm3gmaqq8akb4xlvrzyz8c2d679aak6l"))))
1859     (build-system python-build-system)
1860     (propagated-inputs
1861      `(("ipython" ,python-ipython)
1862        ("numpy" ,python-numpy)
1863        ("pandas" ,python-pandas)
1864        ("scipy" ,python-scipy)))
1865     (native-inputs
1866      `(("nose" ,python-nose)))
1867     (home-page "https://github.com/interpretable-ml/iml")
1868     (synopsis "Interpretable Machine Learning (iML) package")
1869     (description "Interpretable ML (iML) is a set of data type objects,
1870 visualizations, and interfaces that can be used by any method designed to
1871 explain the predictions of machine learning models (or really the output of
1872 any function).  It currently contains the interface and IO code from the Shap
1873 project, and it will potentially also do the same for the Lime project.")
1874     (license license:expat)))
1875
1876 (define-public python-keras-applications
1877   (package
1878     (name "python-keras-applications")
1879     (version "1.0.8")
1880     (source
1881      (origin
1882        (method url-fetch)
1883        (uri (pypi-uri "Keras_Applications" version))
1884        (sha256
1885         (base32
1886          "1rcz31ca4axa6kzhjx4lwqxbg4wvlljkj8qj9a7p9sfd5fhzjyam"))))
1887     (build-system python-build-system)
1888     ;; The tests require Keras, but this package is needed to build Keras.
1889     (arguments '(#:tests? #f))
1890     (propagated-inputs
1891      `(("python-h5py" ,python-h5py)
1892        ("python-numpy" ,python-numpy)))
1893     (native-inputs
1894      `(("python-pytest" ,python-pytest)
1895        ("python-pytest-cov" ,python-pytest-cov)
1896        ("python-pytest-pep8" ,python-pytest-pep8)
1897        ("python-pytest-xdist" ,python-pytest-xdist)))
1898     (home-page "https://github.com/keras-team/keras-applications")
1899     (synopsis "Reference implementations of popular deep learning models")
1900     (description
1901      "This package provides reference implementations of popular deep learning
1902 models for use with the Keras deep learning framework.")
1903     (license license:expat)))
1904
1905 (define-public python-keras-preprocessing
1906   (package
1907     (name "python-keras-preprocessing")
1908     (version "1.1.0")
1909     (source
1910      (origin
1911        (method url-fetch)
1912        (uri (pypi-uri "Keras_Preprocessing" version))
1913        (sha256
1914         (base32
1915          "1r98nm4k1svsqjyaqkfk23i31bl1kcfcyp7094yyj3c43phfp3as"))))
1916     (build-system python-build-system)
1917     (propagated-inputs
1918      `(("python-numpy" ,python-numpy)
1919        ("python-six" ,python-six)))
1920     (native-inputs
1921      `(("python-pandas" ,python-pandas)
1922        ("python-pillow" ,python-pillow)
1923        ("python-pytest" ,python-pytest)
1924        ("python-pytest-cov" ,python-pytest-cov)
1925        ("python-pytest-xdist" ,python-pytest-xdist)
1926        ("tensorflow" ,tensorflow)))
1927     (home-page "https://github.com/keras-team/keras-preprocessing/")
1928     (synopsis "Data preprocessing and augmentation for deep learning models")
1929     (description
1930      "Keras Preprocessing is the data preprocessing and data augmentation
1931 module of the Keras deep learning library.  It provides utilities for working
1932 with image data, text data, and sequence data.")
1933     (license license:expat)))
1934
1935 (define-public python-keras
1936   (package
1937     (name "python-keras")
1938     (version "2.2.4")
1939     (source
1940      (origin
1941        (method url-fetch)
1942        (uri (pypi-uri "Keras" version))
1943        (patches (search-patches "python-keras-integration-test.patch"))
1944        (sha256
1945         (base32
1946          "1j8bsqzh49vjdxy6l1k4iwax5vpjzniynyd041xjavdzvfii1dlh"))))
1947     (build-system python-build-system)
1948     (arguments
1949      `(#:phases
1950        (modify-phases %standard-phases
1951          (add-after 'unpack 'remove-tests-for-unavailable-features
1952            (lambda _
1953              (delete-file "keras/backend/theano_backend.py")
1954              (delete-file "keras/backend/cntk_backend.py")
1955              (delete-file "tests/keras/backend/backend_test.py")
1956
1957              ;; FIXME: This doesn't work because Tensorflow is missing the
1958              ;; coder ops library.
1959              (delete-file "tests/keras/test_callbacks.py")
1960              #t))
1961          (replace 'check
1962            (lambda _
1963              ;; These tests attempt to download data files from the internet.
1964              (delete-file "tests/integration_tests/test_datasets.py")
1965              (delete-file "tests/integration_tests/imagenet_utils_test.py")
1966
1967              (setenv "PYTHONPATH"
1968                      (string-append (getcwd) "/build/lib:"
1969                                     (getenv "PYTHONPATH")))
1970              (invoke "py.test" "-v"
1971                      "-p" "no:cacheprovider"
1972                      "--ignore" "keras/utils"))))))
1973     (propagated-inputs
1974      `(("python-h5py" ,python-h5py)
1975        ("python-keras-applications" ,python-keras-applications)
1976        ("python-keras-preprocessing" ,python-keras-preprocessing)
1977        ("python-numpy" ,python-numpy)
1978        ("python-pydot" ,python-pydot)
1979        ("python-pyyaml" ,python-pyyaml)
1980        ("python-scipy" ,python-scipy)
1981        ("python-six" ,python-six)
1982        ("tensorflow" ,tensorflow)
1983        ("graphviz" ,graphviz)))
1984     (native-inputs
1985      `(("python-pandas" ,python-pandas)
1986        ("python-pytest" ,python-pytest)
1987        ("python-pytest-cov" ,python-pytest-cov)
1988        ("python-pytest-pep8" ,python-pytest-pep8)
1989        ("python-pytest-timeout" ,python-pytest-timeout)
1990        ("python-pytest-xdist" ,python-pytest-xdist)
1991        ("python-sphinx" ,python-sphinx)
1992        ("python-requests" ,python-requests)))
1993     (home-page "https://github.com/keras-team/keras")
1994     (synopsis "High-level deep learning framework")
1995     (description "Keras is a high-level neural networks API, written in Python
1996 and capable of running on top of TensorFlow.  It was developed with a focus on
1997 enabling fast experimentation.  Use Keras if you need a deep learning library
1998 that:
1999
2000 @itemize
2001 @item Allows for easy and fast prototyping (through user friendliness,
2002   modularity, and extensibility).
2003 @item Supports both convolutional networks and recurrent networks, as well as
2004   combinations of the two.
2005 @item Runs seamlessly on CPU and GPU.
2006 @end itemize\n")
2007     (license license:expat)))
2008
2009 (define-public gloo
2010   (let ((version "0.0.0") ; no proper version tag
2011         (commit "ca528e32fea9ca8f2b16053cff17160290fc84ce")
2012         (revision "0"))
2013     (package
2014       (name "gloo")
2015       (version (git-version version revision commit))
2016       (source
2017        (origin
2018          (method git-fetch)
2019          (uri (git-reference
2020                (url "https://github.com/facebookincubator/gloo")
2021                (commit commit)))
2022          (file-name (git-file-name name version))
2023          (sha256
2024           (base32
2025            "1q9f80zy75f6njrzrqkmhc0g3qxs4gskr7ns2jdqanxa2ww7a99w"))))
2026       (build-system cmake-build-system)
2027       (native-inputs
2028        `(("googletest" ,googletest)))
2029       (arguments
2030        `(#:configure-flags '("-DBUILD_TEST=1")
2031          #:phases
2032          (modify-phases %standard-phases
2033            (replace 'check
2034              (lambda _
2035                (invoke "make" "gloo_test")
2036                #t)))))
2037       (synopsis "Collective communications library")
2038       (description
2039        "Gloo is a collective communications library.  It comes with a
2040 number of collective algorithms useful for machine learning applications.
2041 These include a barrier, broadcast, and allreduce.")
2042       (home-page "https://github.com/facebookincubator/gloo")
2043       (license license:bsd-3))))
2044
2045 (define-public python-umap-learn
2046   (package
2047     (name "python-umap-learn")
2048     (version "0.3.10")
2049     (source
2050      (origin
2051        (method url-fetch)
2052        (uri (pypi-uri "umap-learn" version))
2053        (sha256
2054         (base32
2055          "02ada2yy6km6zgk2836kg1c97yrcpalvan34p8c57446finnpki1"))))
2056     (build-system python-build-system)
2057     (native-inputs
2058      `(("python-joblib" ,python-joblib)
2059        ("python-nose" ,python-nose)))
2060     (propagated-inputs
2061      `(("python-numba" ,python-numba)
2062        ("python-numpy" ,python-numpy)
2063        ("python-scikit-learn" ,python-scikit-learn)
2064        ("python-scipy" ,python-scipy)))
2065     (home-page "https://github.com/lmcinnes/umap")
2066     (synopsis
2067      "Uniform Manifold Approximation and Projection")
2068     (description
2069      "Uniform Manifold Approximation and Projection is a dimension reduction
2070 technique that can be used for visualisation similarly to t-SNE, but also for
2071 general non-linear dimension reduction.")
2072     (license license:bsd-3)))