gnu/packages/language.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2015, 2016 Eric Bavier <bavier@member.fsf.org>
   3 ;;; Copyright © 2017, 2018 Tobias Geerinckx-Rice <me@tobias.gr>
   4 ;;; Copyright © 2018 Nikita <nikita@n0.is>
   5 ;;; Copyright © 2019 Alex Vong <alexvong1995@gmail.com>
   6 ;;; Copyright © 2020 Ricardo Wurmus <rekado@elephly.net>
   7 ;;; Copyright © 2020 Julien Lepiller <julien@lepiller.eu>
   8 ;;;
   9 ;;; This file is part of GNU Guix.
  10 ;;;
  11 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  12 ;;; under the terms of the GNU General Public License as published by
  13 ;;; the Free Software Foundation; either version 3 of the License, or (at
  14 ;;; your option) any later version.
  15 ;;;
  16 ;;; GNU Guix is distributed in the hope that it will be useful, but
  17 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 ;;; GNU General Public License for more details.
  20 ;;;
  21 ;;; You should have received a copy of the GNU General Public License
  22 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  23
  24 (define-module (gnu packages language)
  25   #:use-module (gnu packages)
  26   #:use-module (gnu packages autotools)
  27   #:use-module (gnu packages audio)
  28   #:use-module (gnu packages glib)
  29   #:use-module (gnu packages gtk)
  30   #:use-module (gnu packages java)
  31   #:use-module (gnu packages linux)
  32   #:use-module (gnu packages llvm)
  33   #:use-module (gnu packages man)
  34   #:use-module (gnu packages ocr)
  35   #:use-module (gnu packages perl)
  36   #:use-module (gnu packages pkg-config)
  37   #:use-module (gnu packages pulseaudio)
  38   #:use-module (gnu packages python)
  39   #:use-module (gnu packages perl-check)
  40   #:use-module (gnu packages swig)
  41   #:use-module (gnu packages texinfo)
  42   #:use-module (gnu packages web)
  43   #:use-module (gnu packages xml)
  44   #:use-module (gnu packages xorg)
  45   #:use-module (guix packages)
  46   #:use-module (guix build-system gnu)
  47   #:use-module (guix build-system perl)
  48   #:use-module (guix build-system python)
  49   #:use-module ((guix licenses)
  50                 #:select
  51                 (bsd-3 gpl2 gpl2+ gpl3 gpl3+ lgpl2.1 lgpl2.1+ lgpl3+ perl-license zpl2.1))
  52   #:use-module (guix download)
  53   #:use-module (guix git-download)
  54   #:use-module (guix utils))
  55
  56 (define-public liblouis
  57   (package
  58     (name "liblouis")
  59     (version "3.14.0")
  60     (source
  61      (origin
  62        (method git-fetch)
  63        (uri
  64         (git-reference
  65          (url "https://github.com/liblouis/liblouis.git")
  66          (commit (string-append "v" version))))
  67        (file-name (git-file-name name version))
  68        (sha256
  69         (base32 "0v6w8b9r994mkkbm2gqgd7k5yfmdhgbabh0j1gmn375nyvhy4qqh"))))
  70     (build-system gnu-build-system)
  71     (outputs '("out" "bin" "doc" "python"))
  72     (arguments
  73      `(#:configure-flags
  74        (list
  75         "--disable-static"
  76         "--enable-ucs4")
  77        #:phases
  78        (modify-phases %standard-phases
  79          ;; To install the sub-package "python".
  80          (add-after 'install 'install-python-extension
  81            (lambda* (#:key outputs #:allow-other-keys)
  82              (with-directory-excursion "python"
  83                (invoke "python" "setup.py" "install"
  84                        (string-append "--prefix="
  85                                       (assoc-ref outputs "python"))
  86                        "--root=/")))))))
  87     (native-inputs
  88      `(("autoconf" ,autoconf)
  89        ("automake" ,automake)
  90        ("clang-format" ,clang)
  91        ("help2man" ,help2man)
  92        ("libtool" ,libtool)
  93        ("libyaml" ,libyaml)
  94        ("makeinfo" ,texinfo)
  95        ("perl" ,perl)
  96        ("pkg-config" ,pkg-config)
  97        ("python" ,python-wrapper)))
  98     (synopsis "Braille translator and back-translator")
  99     (description "Liblouis is a braille translator and back-translator named in
 100 honor of Louis Braille.  It features support for computer and literary braille,
 101 supports contracted and uncontracted translation for many languages and has
 102 support for hyphenation.  New languages can easily be added through tables that
 103 support a rule- or dictionary based approach.  Tools for testing and debugging
 104 tables are also included.  Liblouis also supports math braille, Nemeth and
 105 Marburg.")
 106     (home-page "http://liblouis.org/")
 107     (license
 108      (list
 109       ;; Library
 110       lgpl2.1+
 111       ;; Tools
 112       gpl3+))))
 113
 114 (define-public liblouisutdml
 115   (package
 116     (name "liblouisutdml")
 117     (version "2.8.0")
 118     (source
 119      (origin
 120        (method git-fetch)
 121        (uri
 122         (git-reference
 123          (url "https://github.com/liblouis/liblouisutdml.git")
 124          (commit (string-append "v" version))))
 125        (file-name (git-file-name name version))
 126        (sha256
 127         (base32 "11xxky3crjm8bidfljzpqaz1m1i1m1yskmdpavf9b4jpf87nzjj2"))
 128        (patches
 129         (search-patches "liblouisutdml-fix-tests.patch"))))
 130     (build-system gnu-build-system)
 131     (outputs '("out" "bin" "doc"))
 132     (arguments
 133      `(#:configure-flags
 134        (list
 135         "--disable-static")))
 136     (native-inputs
 137      `(("autoconf" ,autoconf)
 138        ("automake" ,automake)
 139        ("help2man" ,help2man)
 140        ("jdk" ,icedtea "jdk")
 141        ("libtool" ,libtool)
 142        ("makeinfo" ,texinfo)
 143        ("pkg-config" ,pkg-config)))
 144     (inputs
 145      `(("libxml2" ,libxml2)))
 146     (propagated-inputs
 147      `(("liblouis" ,liblouis)
 148        ("liblouis:bin" ,liblouis "bin")))
 149     (synopsis "Braille transcription services")
 150     (description "Liblouisutdml is a library providing complete braille
 151 transcription services for xml, html and text documents.  It translates into
 152 appropriate braille codes and formats according to its style sheet and the
 153 specifications in the document.")
 154     (home-page "http://liblouis.org/")
 155     (license
 156      (list
 157       ;; Library
 158       lgpl3+
 159       ;; Tools
 160       gpl3+))))
 161
 162 (define-public libstemmer
 163   (package
 164     (name "libstemmer")
 165     (version "2.0.0")
 166     (source
 167      (origin
 168        (method url-fetch)
 169        (uri "https://snowballstem.org/dist/libstemmer_c.tgz")
 170        (sha256
 171         (base32 "1z2xvrjsaaypc04lwz7dg8mjm5cq1gzmn0l544pn6y2ll3r7ckh5"))))
 172     (build-system gnu-build-system)
 173     (arguments
 174      `(#:tests? #f                      ; No tests exist
 175        #:make-flags
 176        (list
 177         (string-append "CC=" ,(cc-for-target))
 178         "CFLAGS=-fPIC")
 179        #:phases
 180        (modify-phases %standard-phases
 181          (delete 'configure)
 182          (replace 'install
 183            (lambda* (#:key outputs #:allow-other-keys)
 184              (let* ((out (assoc-ref outputs "out"))
 185                     (out-bin (string-append out "/bin"))
 186                     (out-include (string-append out "/include"))
 187                     (out-lib (string-append out "/lib")))
 188                (install-file "stemwords" out-bin)
 189                (install-file "include/libstemmer.h" out-include)
 190                (rename-file "libstemmer.o" "libstemmer.a")
 191                (install-file "libstemmer.a" out-lib)
 192                #t))))))
 193     (synopsis "Stemming Library")
 194     (description "LibStemmer provides stemming library, supporting several
 195 languages.")
 196     (home-page "https://snowballstem.org/")
 197     (license bsd-3)))
 198
 199 (define-public perl-lingua-en-findnumber
 200   (package
 201     (name "perl-lingua-en-findnumber")
 202     (version "1.32")
 203     (source
 204      (origin
 205        (method url-fetch)
 206        (uri (string-append "mirror://cpan/authors/id/N/NE/NEILB/"
 207                            "Lingua-EN-FindNumber-" version ".tar.gz"))
 208        (sha256
 209         (base32
 210          "015ximzdp42v824llwlg2pd77vd0d172lb4xs55q9f9zhqf6s5qx"))))
 211     (build-system perl-build-system)
 212     (propagated-inputs
 213      `(("perl-lingua-en-words2nums" ,perl-lingua-en-words2nums)))
 214     (home-page "https://metacpan.org/release/Lingua-EN-FindNumber")
 215     (synopsis "Locate (written) numbers in English text ")
 216     (description "This module provides a regular expression for finding
 217 numbers in English text.  It also provides functions for extracting and
 218 manipulating such numbers.")
 219     (license perl-license)))
 220
 221 (define-public perl-lingua-en-inflect
 222   (package
 223     (name "perl-lingua-en-inflect")
 224     (version "1.903")
 225     (source
 226      (origin
 227        (method url-fetch)
 228        (uri (string-append "mirror://cpan/authors/id/D/DC/DCONWAY/"
 229                            "Lingua-EN-Inflect-" version ".tar.gz"))
 230        (sha256
 231         (base32
 232          "0j8d1f1wvmgc11d71pc8xp8fv5a1nb2yfw1dgd19xhscn1klpvzw"))))
 233     (build-system perl-build-system)
 234     (native-inputs `(("perl-module-build" ,perl-module-build)))
 235     (home-page "https://metacpan.org/release/Lingua-EN-Inflect")
 236     (synopsis "Convert singular to plural")
 237     (description "Lingua::EN::Inflect provides plural inflections,
 238 \"a\"/\"an\" selection for English words, and manipulation of numbers as
 239 words.  Plural forms of all nouns, most verbs, and some adjectives are
 240 provided.  Where appropriate, \"classical\" variants (for example: \"brother\"
 241 -> \"brethren\", \"dogma\" -> \"dogmata\", etc.) are also provided.")
 242     (license perl-license)))
 243
 244 (define-public perl-lingua-en-inflect-number
 245   (package
 246     (name "perl-lingua-en-inflect-number")
 247     (version "1.12")
 248     (source
 249      (origin
 250        (method url-fetch)
 251        (uri (string-append "mirror://cpan/authors/id/N/NE/NEILB/"
 252                            "Lingua-EN-Inflect-Number-" version ".tar.gz"))
 253        (sha256
 254         (base32
 255          "1gxccynkaqav43ww43jp4rzkyr36x97jd03yb5f6yx0jhn1k7yv6"))))
 256     (build-system perl-build-system)
 257     (propagated-inputs
 258      `(("perl-lingua-en-inflect" ,perl-lingua-en-inflect)))
 259     (home-page "https://metacpan.org/release/Lingua-EN-Inflect-Number")
 260     (synopsis "Force number of words to singular or plural")
 261     (description "This module extends the functionality of Lingua::EN::Inflect
 262 with three new functions for determining plurality of a word and forcefully
 263 converting a word to singular or plural.")
 264     (license perl-license)))
 265
 266 (define-public perl-lingua-en-inflect-phrase
 267   (package
 268     (name "perl-lingua-en-inflect-phrase")
 269     (version "0.20")
 270     (source
 271      (origin
 272        (method url-fetch)
 273        (uri (string-append "mirror://cpan/authors/id/R/RK/RKITOVER/"
 274                            "Lingua-EN-Inflect-Phrase-" version ".tar.gz"))
 275        (sha256
 276         (base32
 277          "1a6y1l2pjim2242wcpgz066di4pbzfgsjjdl7vg5a5wzm48qj1am"))))
 278     (build-system perl-build-system)
 279     (native-inputs
 280      `(("perl-test-nowarnings" ,perl-test-nowarnings)))
 281     (propagated-inputs
 282      `(("perl-lingua-en-findnumber" ,perl-lingua-en-findnumber)
 283        ("perl-lingua-en-inflect" ,perl-lingua-en-inflect)
 284        ("perl-lingua-en-inflect-number" ,perl-lingua-en-inflect-number)
 285        ("perl-lingua-en-number-isordinal" ,perl-lingua-en-number-isordinal)
 286        ("perl-lingua-en-tagger" ,perl-lingua-en-tagger)))
 287     (home-page "https://metacpan.org/release/Lingua-EN-Inflect-Phrase")
 288     (synopsis "Inflect short English phrases")
 289     (description "This module attempts to pluralize or singularize short
 290 English phrases.")
 291     (license perl-license)))
 292
 293 (define-public perl-lingua-en-number-isordinal
 294   (package
 295     (name "perl-lingua-en-number-isordinal")
 296     (version "0.05")
 297     (source
 298      (origin
 299        (method url-fetch)
 300        (uri (string-append "mirror://cpan/authors/id/R/RK/RKITOVER/"
 301                            "Lingua-EN-Number-IsOrdinal-" version ".tar.gz"))
 302        (sha256
 303         (base32
 304          "1mhqjvh2ad30gjab5b3a6mbr4aysyrscp4wp42yy5x6001a6km98"))))
 305     (build-system perl-build-system)
 306     (native-inputs
 307      `(("perl-try-tiny" ,perl-try-tiny)
 308        ("perl-test-fatal" ,perl-test-fatal)))
 309     (propagated-inputs
 310      `(("perl-lingua-en-findnumber" ,perl-lingua-en-findnumber)))
 311     (home-page "https://metacpan.org/release/Lingua-EN-Number-IsOrdinal")
 312     (synopsis "Detect if English number is ordinal or cardinal")
 313     (description "This module will tell you if a number, either in words or as
 314 digits, is a cardinal or ordinal number.")
 315     (license perl-license)))
 316
 317 (define-public perl-lingua-en-tagger
 318   (package
 319     (name "perl-lingua-en-tagger")
 320     (version "0.30")
 321     (source
 322      (origin
 323        (method url-fetch)
 324        (uri (string-append "mirror://cpan/authors/id/A/AC/ACOBURN/"
 325                            "Lingua-EN-Tagger-" version ".tar.gz"))
 326        (sha256
 327         (base32
 328          "0nrnkvsf9f0a7lp82sanmy89ms2nqq1lvjqicvsagsvzp513bl5b"))))
 329     (build-system perl-build-system)
 330     (propagated-inputs
 331      `(("perl-memoize-expirelru" ,perl-memoize-expirelru)
 332        ("perl-lingua-stem" ,perl-lingua-stem)
 333        ("perl-html-parser" ,perl-html-parser)
 334        ("perl-html-tagset" ,perl-html-tagset)))
 335     (home-page "https://metacpan.org/release/Lingua-EN-Tagger")
 336     (synopsis "Part-of-speech tagger for English natural language processing")
 337     (description "This module is a probability based, corpus-trained tagger
 338 that assigns part-of-speech tags to English text based on a lookup dictionary
 339 and a set of probability values.  The tagger assigns appropriate tags based on
 340 conditional probabilities - it examines the preceding tag to determine the
 341 appropriate tag for the current word.  Unknown words are classified according
 342 to word morphology or can be set to be treated as nouns or other parts of
 343 speech.  The tagger also extracts as many nouns and noun phrases as it can,
 344 using a set of regular expressions.")
 345     (license gpl3)))
 346
 347 (define-public perl-lingua-en-words2nums
 348   (package
 349     (name "perl-lingua-en-words2nums")
 350     (version "0.18")
 351     (source
 352      (origin
 353        (method url-fetch)
 354        (uri (string-append "mirror://cpan/authors/id/J/JO/JOEY/"
 355                            "Lingua-EN-Words2Nums-" version ".tar.gz"))
 356        (sha256
 357         (base32
 358          "118xx8qr1zbx30psv7ic55w65h15mc1vz6zicshfm96jgiwmcrb8"))))
 359     (build-system perl-build-system)
 360     (home-page "https://metacpan.org/release/Lingua-EN-Words2Nums")
 361     (synopsis "Convert English text to numbers")
 362     (description "This module converts English text into numbers.  It supports
 363 both ordinal and cardinal numbers, negative numbers, and very large numbers.")
 364     (license perl-license)))
 365
 366 (define-public perl-lingua-pt-stemmer
 367   (package
 368     (name "perl-lingua-pt-stemmer")
 369     (version "0.02")
 370     (source
 371      (origin
 372        (method url-fetch)
 373        (uri (string-append "mirror://cpan/authors/id/N/NE/NEILB/"
 374                            "Lingua-PT-Stemmer-" version ".tar.gz"))
 375        (sha256
 376         (base32
 377          "17c48sfbgwd2ivlgf59sr6jdhwa3aim8750f8pyzz7xpi8gz0var"))))
 378     (build-system perl-build-system)
 379     (home-page "https://metacpan.org/release/Lingua-PT-Stemmer")
 380     (synopsis "Portuguese language stemming")
 381     (description "This module implements a Portuguese stemming algorithm
 382 proposed in the paper A Stemming Algorithm for the Portuguese Language by
 383 Moreira, V. and Huyck, C.")
 384     (license perl-license)))
 385
 386 (define-public perl-lingua-stem
 387   (package
 388     (name "perl-lingua-stem")
 389     (version "0.84")
 390     (source
 391      (origin
 392        (method url-fetch)
 393        (uri (string-append "mirror://cpan/authors/id/S/SN/SNOWHARE/"
 394                            "Lingua-Stem-" version ".tar.gz"))
 395        (sha256
 396         (base32
 397          "12avh2mnnc7llmmshrr5bgb473fvydxnlqrqbl2815mf2dp4pxcg"))))
 398     (build-system perl-build-system)
 399     (native-inputs
 400      `(("perl-module-build" ,perl-module-build)))
 401     (propagated-inputs
 402      `(("perl-lingua-pt-stemmer" ,perl-lingua-pt-stemmer)
 403        ("perl-lingua-stem-fr" ,perl-lingua-stem-fr)
 404        ("perl-lingua-stem-it" ,perl-lingua-stem-it)
 405        ("perl-lingua-stem-ru" ,perl-lingua-stem-ru)
 406        ("perl-lingua-stem-snowball-da" ,perl-lingua-stem-snowball-da)
 407        ("perl-snowball-norwegian" ,perl-snowball-norwegian)
 408        ("perl-snowball-swedish" ,perl-snowball-swedish)
 409        ("perl-text-german" ,perl-text-german)))
 410     (home-page "https://metacpan.org/release/Lingua-Stem")
 411     (synopsis "Stemming of words in various languages")
 412     (description "This routine applies stemming algorithms to its parameters,
 413 returning the stemmed words as appropriate to the selected locale.")
 414     (license perl-license)))
 415
 416 (define-public perl-lingua-stem-fr
 417   (package
 418     (name "perl-lingua-stem-fr")
 419     (version "0.02")
 420     (source
 421      (origin
 422        (method url-fetch)
 423        (uri (string-append "mirror://cpan/authors/id/S/SD/SDP/"
 424                            "Lingua-Stem-Fr-" version ".tar.gz"))
 425        (sha256
 426         (base32
 427          "0vyrspwzaqjxm5mqshf4wvwa3938mkajd1918d9ii2l9m2rn8kwx"))))
 428     (build-system perl-build-system)
 429     (home-page "https://metacpan.org/release/Lingua-Stem-Fr")
 430     (synopsis "Porter's stemming algorithm for French")
 431     (description "This module uses a modified version of the Porter Stemming
 432 Algorithm to return a stemmed French word.")
 433     (license perl-license)))
 434
 435 (define-public perl-lingua-stem-it
 436   (package
 437     (name "perl-lingua-stem-it")
 438     (version "0.02")
 439     (source
 440      (origin
 441        (method url-fetch)
 442        (uri (string-append "mirror://cpan/authors/id/A/AC/ACALPINI/"
 443                            "Lingua-Stem-It-" version ".tar.gz"))
 444        (sha256
 445         (base32
 446          "1207r183s5hlh4mfwa6p46vzm0dhvrs2dnss5s41a0gyfkxp7riq"))))
 447     (build-system perl-build-system)
 448     (home-page "https://metacpan.org/release/Lingua-Stem-It")
 449     (synopsis "Porter's stemming algorithm for Italian")
 450     (description "This module applies the Porter Stemming Algorithm to its
 451 parameters, returning the stemmed Italian word.")
 452     (license perl-license)))
 453
 454 (define-public perl-lingua-stem-ru
 455   (package
 456     (name "perl-lingua-stem-ru")
 457     (version "0.04")
 458     (source
 459      (origin
 460        (method url-fetch)
 461        (uri (string-append "mirror://cpan/authors/id/N/NE/NEILB/"
 462                            "Lingua-Stem-Ru-" version ".tar.gz"))
 463        (sha256
 464         (base32
 465          "0a2jmdz7jn32qj5hyiw5kbv8fvlpmws8i00a6xcbkzb48yvwww0j"))))
 466     (build-system perl-build-system)
 467     (home-page "https://metacpan.org/release/Lingua-Stem-Ru")
 468     (synopsis "Porter's stemming algorithm for Russian")
 469     (description "This module applies the Porter Stemming Algorithm to its
 470 parameters, returning the stemmed Russian (KOI8-R only) word.")
 471     (license perl-license)))
 472
 473 (define-public perl-lingua-stem-snowball-da
 474   (package
 475     (name "perl-lingua-stem-snowball-da")
 476     (version "1.01")
 477     (source
 478      (origin
 479        (method url-fetch)
 480        (uri (string-append "mirror://cpan/authors/id/C/CI/CINE/"
 481                            "Lingua-Stem-Snowball-Da-" version ".tar.gz"))
 482        (sha256
 483         (base32
 484          "0mm0m7glm1s6i9f6a78jslw6wh573208arxhq93yriqmw17bwf9f"))))
 485     (build-system perl-build-system)
 486     (home-page "https://metacpan.org/release/Lingua-Stem-Snowball-Da")
 487     (synopsis "Porters stemming algorithm for Danish")
 488     (description "Lingua::Stem::Snowball::Da is a perl port of the danish
 489 stemmer at http://snowball.sourceforge.net, it was originally altered from the
 490 Lingua::Stem::Snowball::Se.")
 491     (license gpl2)))
 492
 493 (define-public perl-snowball-norwegian
 494   (package
 495     (name "perl-snowball-norwegian")
 496     (version "1.2")
 497     (source
 498      (origin
 499        (method url-fetch)
 500        (uri (string-append "mirror://cpan/authors/id/A/AS/ASKSH/"
 501                            "Snowball-Norwegian-" version ".tar.gz"))
 502        (sha256
 503         (base32
 504          "0675v45bbsh7vr7kpf36xs2q79g02iq1kmfw22h20xdk4rzqvkqx"))))
 505     (build-system perl-build-system)
 506     (native-inputs `(("perl-module-build" ,perl-module-build)))
 507     (home-page "https://metacpan.org/release/Snowball-Norwegian")
 508     (synopsis "Porters stemming algorithm for Norwegian")
 509     (description "Lingua::Stem::Snowball::No is a perl port of the norwegian
 510 stemmer at http://snowball.tartarus.org.")
 511     (license perl-license)))
 512
 513 (define-public perl-snowball-swedish
 514   (package
 515     (name "perl-snowball-swedish")
 516     (version "1.2")
 517     (source
 518      (origin
 519        (method url-fetch)
 520        (uri (string-append "mirror://cpan/authors/id/A/AS/ASKSH/"
 521                            "Snowball-Swedish-" version ".tar.gz"))
 522        (sha256
 523         (base32
 524          "0agwc12jk5kmabnpsplw3wf4ii5w1zb159cpin44x3srb0sr5apg"))))
 525     (build-system perl-build-system)
 526     (native-inputs `(("perl-module-build" ,perl-module-build)))
 527     (home-page "https://metacpan.org/release/Snowball-Swedish")
 528     (synopsis "Porters stemming algorithm for Swedish")
 529     (description "Lingua::Stem::Snowball::Se is a perl port of the swedish
 530 stemmer at http://snowball.sourceforge.net.")
 531     (license perl-license)))
 532
 533 (define-public perl-string-toidentifier-en
 534   (package
 535     (name "perl-string-toidentifier-en")
 536     (version "0.12")
 537     (source
 538      (origin
 539        (method url-fetch)
 540        (uri (string-append "mirror://cpan/authors/id/R/RK/RKITOVER/"
 541                            "String-ToIdentifier-EN-" version ".tar.gz"))
 542        (sha256
 543         (base32
 544          "12nw7h2yiybhdw0vnnpc7bif8ylhsn6kqf6s39dsrf9h54iq9yrs"))))
 545     (build-system perl-build-system)
 546     (propagated-inputs
 547      `(("perl-lingua-en-inflect-phrase" ,perl-lingua-en-inflect-phrase)
 548        ("perl-text-unidecode" ,perl-text-unidecode)
 549        ("perl-namespace-clean" ,perl-namespace-clean)))
 550     (home-page "https://metacpan.org/release/String-ToIdentifier-EN")
 551     (synopsis "Convert strings to English program identifiers")
 552     (description "This module provides a utility method, \"to_identifier\" for
 553 converting an arbitrary string into a readable representation using the ASCII
 554 subset of \"\\w\" for use as an identifier in a computer program.  The intent
 555 is to make unique identifier names from which the content of the original
 556 string can be easily inferred by a human just by reading the identifier.")
 557     (license perl-license)))
 558
 559 (define-public perl-text-german
 560   (package
 561     (name "perl-text-german")
 562     (version "0.06")
 563     (source
 564      (origin
 565        (method url-fetch)
 566        (uri (string-append "mirror://cpan/authors/id/U/UL/ULPFR/"
 567                            "Text-German-" version ".tar.gz"))
 568        (sha256
 569         (base32
 570          "1p87pgap99lw0nv62i3ghvsi7yg90lhn8vsa3yqp75rd04clybcj"))))
 571     (build-system perl-build-system)
 572     (home-page "https://metacpan.org/release/Text-German")
 573     (synopsis "German grundform reduction")
 574     (description "This module is a rather incomplete implementation of work
 575 done by Gudrun Putze-Meier.")
 576     (license perl-license)))
 577
 578 (define* (tegaki-release-uri proj version
 579                              #:optional (ext "tar.gz"))
 580   (string-append "https://github.com/tegaki/tegaki/releases/download"
 581                  "/v" version "/" proj "-" version "." ext))
 582
 583 (define remove-pre-compiled-files
 584   (lambda exts
 585     "Return snippet for removing pre-compiled files matching one of the
 586 extensions in EXTS."
 587     `(begin (for-each delete-file
 588                       (find-files "."
 589                                   (lambda (name _)
 590                                     (any (cut string-suffix? <> name)
 591                                          (map (cut string-append "." <>)
 592                                               ',exts)))))
 593             #t)))
 594
 595 ;;; modules required for the above snippet
 596 (define remove-pre-compiled-files-modules
 597   '((guix build utils)
 598     (srfi srfi-1)
 599     (srfi srfi-26)))
 600
 601 (define-public python2-tegaki-wagomu
 602   (package
 603     (name "python2-tegaki-wagomu")
 604     (version "0.3.1")
 605     (source
 606      (origin
 607        (method url-fetch)
 608        (uri (tegaki-release-uri "tegaki-wagomu" version))
 609        (sha256
 610         (base32
 611          "1pzdiq4zy1nyylaj9i6v2h4h0r05klahskzpafpp367p4rysi1x9"))
 612        (modules remove-pre-compiled-files-modules)
 613        (snippet (remove-pre-compiled-files "pyc"))))
 614     (build-system python-build-system)
 615     (arguments
 616      `(#:python ,python-2 ; only Python 2 is supported
 617        #:phases
 618        (modify-phases %standard-phases
 619          (add-after 'unpack 'fix-recognizer
 620            (lambda* (#:key inputs #:allow-other-keys)
 621              ;; fix missing module and function
 622              (substitute* "tegakiwagomu.py"
 623                (("import Results,")
 624                 "import ")
 625                (("def _recognize")
 626                 "def recognize")
 627                (("Results\\(candidates\\)")
 628                 "candidates"))
 629              #t)))))
 630     (inputs
 631      `(("glib" ,glib)))
 632     (native-inputs
 633      `(("pkg-config" ,pkg-config)
 634        ("swig" ,swig)))
 635     (home-page "https://tegaki.github.io/")
 636     (synopsis
 637      "Chinese and Japanese Handwriting Recognition (Recognition engine)")
 638     (description
 639      "Tegaki is an ongoing project which aims to develop a free and open-source
 640 modern implementation of handwriting recognition software, specifically
 641 designed for Chinese (simplified and traditional) and Japanese, and that is
 642 suitable for both the desktop and mobile devices.")
 643     (license gpl2+))) ; all files
 644
 645 (define-public python2-tegaki-python
 646   (package
 647     (inherit python2-tegaki-wagomu)
 648     (name "python2-tegaki-python")
 649     (version "0.3.1")
 650     (source
 651      (origin
 652        (method url-fetch)
 653        (uri (tegaki-release-uri "tegaki-python" version))
 654        (sha256
 655         (base32
 656          "0x93k7pw9nh0ywd97pr8pm7jv3f94nw044i5k0zvzhdpsjqvak7p"))
 657        (modules remove-pre-compiled-files-modules)
 658        (snippet (remove-pre-compiled-files "pyc"))))
 659     (arguments
 660      (substitute-keyword-arguments (package-arguments python2-tegaki-wagomu)
 661        ((#:phases _)
 662         `(modify-phases %standard-phases
 663            (add-after 'unpack 'pre-configure
 664              (lambda* (#:key inputs #:allow-other-keys)
 665                ;; Always convert string to unicode to avoid the following error
 666                ;; when running "tegaki-build" in python2-tegaki-tools:
 667                ;;
 668                ;; sqlite3.ProgrammingError: You must not use 8-bit bytestrings
 669                ;; unless you use a text_factory that can interpret 8-bit
 670                ;; bytestrings (like text_factory = str).
 671                ;; It is highly recommended that you instead just switch your
 672                ;; application to Unicode strings.
 673                (substitute* "tegaki/charcol.py"
 674                  (("sqlite3.OptimizedUnicode")
 675                   "lambda s: unicode(s, 'utf-8')"))
 676                (substitute* "tegaki/engine.py"
 677                  (("/usr(/local)?")
 678                   (assoc-ref inputs "python2-tegaki-wagomu")))
 679                #t))))))
 680     ;; override inherited inputs
 681     (inputs '())
 682     (native-inputs '())
 683     (propagated-inputs
 684      `(("python2-tegaki-wagomu" ,python2-tegaki-wagomu)
 685        ("python2-zinnia" ,python2-zinnia)))
 686     (synopsis
 687      "Chinese and Japanese Handwriting Recognition (Base python library)")
 688     (license (list gpl2+ ; all files except...
 689                    bsd-3 ; dictutils.py
 690                    zpl2.1)))) ; minjson.py
 691
 692 (define-public python2-tegaki-pygtk
 693   (package
 694     (inherit python2-tegaki-wagomu)
 695     (name "python2-tegaki-pygtk")
 696     (version "0.3.1")
 697     (source
 698      (origin
 699        (method url-fetch)
 700        (uri (tegaki-release-uri "tegaki-pygtk" version))
 701        (sha256
 702         (base32
 703          "1cip0azxhjdj2dg2z85cp1z3lz4qwx3w1j7z4xmcm7npapmsaqs2"))
 704        (modules remove-pre-compiled-files-modules)
 705        (snippet (remove-pre-compiled-files "pyc"))))
 706     (arguments
 707      (substitute-keyword-arguments (package-arguments python2-tegaki-wagomu)
 708        ((#:phases _)
 709         `(modify-phases %standard-phases
 710            (add-after 'unpack 'fix-paths
 711              (lambda* (#:key inputs #:allow-other-keys)
 712                (substitute* "tegakigtk/fakekey.py"
 713                  (("libX11.so.6" so)
 714                   (string-append (assoc-ref inputs "libx11") "/lib/" so))
 715                  (("libXtst.so.6" so)
 716                   (string-append (assoc-ref inputs "libxtst") "/lib/" so)))
 717                #t))))))
 718     (inputs ; required for sending key strokes
 719      `(("libx11" ,libx11)
 720        ("libxtst" ,libxtst)))
 721     (native-inputs '()) ; override inherited inputs
 722     (propagated-inputs
 723      `(("python2-pygtk" ,python2-pygtk)
 724        ("python2-tegaki-python" ,python2-tegaki-python)))
 725     (synopsis "Chinese and Japanese Handwriting Recognition (Base UI library)")
 726     (license gpl2+))) ; all files
 727
 728 (define-public python2-tegaki-tools
 729   (package
 730     (inherit python2-tegaki-wagomu)
 731     (name "python2-tegaki-tools")
 732     (version "0.3.1")
 733     (source
 734      (origin
 735        (method url-fetch)
 736        (uri (tegaki-release-uri "tegaki-tools" version))
 737        (sha256
 738         (base32
 739          "0xxv97ggh2jgldw3r7y59lv3fhz733r6l7mdn6nh4m0gvb0ja971"))
 740        (modules remove-pre-compiled-files-modules)
 741        (snippet (remove-pre-compiled-files "pyc"))))
 742     (arguments
 743      (substitute-keyword-arguments (package-arguments python2-tegaki-wagomu)
 744        ((#:phases _) '%standard-phases)))
 745     (inputs
 746      `(("python2-tegaki-pygtk" ,python2-tegaki-pygtk)))
 747     ;; override inherited inputs
 748     (native-inputs '())
 749     (propagated-inputs '())
 750     (synopsis "Chinese and Japanese Handwriting Recognition (Advanced tools)")
 751     ;; Files in gifenc/ are licensed under gpl3+ while other files are licensed
 752     ;; under gpl2+. Therefore, the combined work is licensed under gpl3+.
 753     (license gpl3+)))
 754
 755 (define-public python2-tegaki-recognize
 756   (let ((commit "eceec69fe651d0733c8c8752dae569d2283d0f3c")
 757         (revision "1"))
 758     (package
 759       (inherit python2-tegaki-tools)
 760       (name "python2-tegaki-recognize")
 761       ;; version copied from <https://github.com/tegaki/tegaki/releases>
 762       (version (git-version "0.3.1" revision commit))
 763       (source
 764        (origin
 765          ;; We use GIT-FETCH because 'tegaki-recognize.desktop.in' and
 766          ;; 'tegaki-recognize.in' are missing in the tarball.
 767          (method git-fetch)
 768          (uri (git-reference
 769                (url "https://github.com/tegaki/tegaki")
 770                (commit commit)))
 771          (sha256
 772           (base32
 773            "09mw2if9p885phbgah5f95q3fwy7s5b46qlmpxqyzfcnj6g7afr5"))
 774          (file-name (git-file-name name version))
 775          (modules `((guix build utils)
 776                     (ice-9 ftw)
 777                     (srfi srfi-26)
 778                     ,@remove-pre-compiled-files-modules))
 779          (snippet
 780           `(begin
 781              ;; remove unnecessary files with potentially different license
 782              (for-each delete-file-recursively
 783                        (scandir "."
 784                                 (negate (cut member <> '("tegaki-recognize"
 785                                                          "." "..")))))
 786              ,(remove-pre-compiled-files "pyc")
 787              #t))))
 788       (arguments
 789        (substitute-keyword-arguments (package-arguments python2-tegaki-tools)
 790          ((#:phases _)
 791           `(modify-phases %standard-phases
 792              (add-after 'unpack 'chdir
 793                (lambda _
 794                  (chdir "tegaki-recognize")
 795                  #t))
 796              ;; 'setup.py' script does not support one of the Python build
 797              ;; system's default flags, "--single-version-externally-managed"
 798              (replace 'install
 799                (lambda* (#:key outputs #:allow-other-keys)
 800                  (invoke "python" "setup.py" "install"
 801                          (string-append "--prefix=" (assoc-ref outputs "out"))
 802                          "--root=/")
 803                  #t))))))
 804       (synopsis "Chinese and Japanese Handwriting Recognition (Main program)")
 805       (license gpl2+)))) ; all files
 806
 807 (define-public tegaki-zinnia-japanese
 808   (package
 809     (inherit python2-tegaki-wagomu)
 810     (name "tegaki-zinnia-japanese")
 811     (version "0.3")
 812     (source
 813      (origin
 814        (method url-fetch)
 815        (uri (tegaki-release-uri name version "zip"))
 816        (sha256
 817         (base32
 818          "1nmg9acxhcqly9gwkyb9m0hpy76fll91ywk4b1q4xms0ajxip1h7"))
 819        (modules remove-pre-compiled-files-modules)
 820        (snippet (remove-pre-compiled-files "model"))))
 821     (build-system gnu-build-system)
 822     (arguments
 823      `(#:tests? #f ; no tests
 824        #:phases
 825        (modify-phases %standard-phases
 826          (replace 'configure
 827            (lambda* (#:key outputs #:allow-other-keys)
 828              (substitute* "Makefile"
 829                (("/usr/local")
 830                 (assoc-ref outputs "out")))
 831              #t)))))
 832     ;; override inherited inputs
 833     (inputs '())
 834     (native-inputs
 835      `(("python2-tegaki-tools" ,python2-tegaki-tools)))
 836     (propagated-inputs '())
 837     (native-search-paths
 838      (list (search-path-specification
 839             (variable "TEGAKI_MODEL_PATH")
 840             (files '("share/tegaki/models")))))
 841     (synopsis "Chinese and Japanese Handwriting Recognition (Model)")
 842     (license lgpl2.1))) ; all files
 843
 844 (define-public tegaki-zinnia-japanese-light
 845   (package
 846     (inherit tegaki-zinnia-japanese)
 847     (name "tegaki-zinnia-japanese-light")
 848     (version "0.3")
 849     (source
 850      (origin
 851        (method url-fetch)
 852        (uri (tegaki-release-uri name version "zip"))
 853        (sha256
 854         (base32
 855          "0x0fs29ylqzxd6xvg51h7rigpbisd7q8v11df425ib2j792yfyf8"))
 856        (modules remove-pre-compiled-files-modules)
 857        (snippet (remove-pre-compiled-files "model"))))
 858     (license lgpl2.1))) ; all files
 859
 860 (define-public tegaki-zinnia-japanese-kyoiku
 861   (package
 862     (inherit tegaki-zinnia-japanese)
 863     (name "tegaki-zinnia-japanese-kyoiku")
 864     (version "0.3")
 865     (source
 866      (origin
 867        (method url-fetch)
 868        (uri (tegaki-release-uri name version "zip"))
 869        (sha256
 870         (base32
 871          "0am94bcpmbzplxdnwn9gk15sgaizvcfhmv13mk14jjvx3419cvvx"))
 872        (modules remove-pre-compiled-files-modules)
 873        (snippet (remove-pre-compiled-files "model"))))
 874     (license lgpl2.1))) ; all files
 875
 876 (define-public tegaki-zinnia-japanese-joyo
 877   (package
 878     (inherit tegaki-zinnia-japanese)
 879     (name "tegaki-zinnia-japanese-joyo")
 880     (version "0.3")
 881     (source
 882      (origin
 883        (method url-fetch)
 884        (uri (tegaki-release-uri name version "zip"))
 885        (sha256
 886         (base32
 887          "1v0j40lzdyiz01ayws0b8r7fsdy2mr32658382kz4wyk883wzx2z"))
 888        (modules remove-pre-compiled-files-modules)
 889        (snippet (remove-pre-compiled-files "model"))))
 890     (license lgpl2.1))) ; all files
 891
 892 (define-public tegaki-zinnia-simplified-chinese
 893   (package
 894     (inherit tegaki-zinnia-japanese)
 895     (name "tegaki-zinnia-simplified-chinese")
 896     (version "0.3")
 897     (source
 898      (origin
 899        (method url-fetch)
 900        (uri (tegaki-release-uri name version "zip"))
 901        (sha256
 902         (base32
 903          "18wq0jccv7lpnrfnzspyc110d6pj2v1i21xcx4fmgzz1lnln3fs5"))
 904        (modules remove-pre-compiled-files-modules)
 905        (snippet (remove-pre-compiled-files "model"))))
 906     (license lgpl2.1))) ; all files
 907
 908 (define-public tegaki-zinnia-simplified-chinese-light
 909   (package
 910     (inherit tegaki-zinnia-japanese)
 911     (name "tegaki-zinnia-simplified-chinese-light")
 912     (version "0.3")
 913     (source
 914      (origin
 915        (method url-fetch)
 916        (uri (tegaki-release-uri name version "zip"))
 917        (sha256
 918         (base32
 919          "0v24yf0w0p03lb7fyx128a75mwzad166bigvlbrzqnad789qg1sr"))
 920        (modules remove-pre-compiled-files-modules)
 921        (snippet (remove-pre-compiled-files "model"))))
 922     (license lgpl2.1))) ; all files
 923
 924 (define-public tegaki-zinnia-traditional-chinese
 925   (package
 926     (inherit tegaki-zinnia-japanese)
 927     (name "tegaki-zinnia-traditional-chinese")
 928     (version "0.3")
 929     (source
 930      (origin
 931        (method url-fetch)
 932        (uri (tegaki-release-uri name version "zip"))
 933        (sha256
 934         (base32
 935          "140nlp6hynrai2svs5670jjfw1za6ayflhyj2dl0bzsfgbk3447l"))
 936        (modules remove-pre-compiled-files-modules)
 937        (snippet (remove-pre-compiled-files "model"))))
 938     (license lgpl2.1))) ; all files
 939
 940 (define-public tegaki-zinnia-traditional-chinese-light
 941   (package
 942     (inherit tegaki-zinnia-japanese)
 943     (name "tegaki-zinnia-traditional-chinese-light")
 944     (version "0.3")
 945     (source
 946      (origin
 947        (method url-fetch)
 948        (uri (tegaki-release-uri name version "zip"))
 949        (sha256
 950         (base32
 951          "1m6yk6a57vs9wg5y50qciwi1ahhmklp2mgsjysbj4mnyzv6yhcr2"))
 952        (modules remove-pre-compiled-files-modules)
 953        (snippet (remove-pre-compiled-files "model"))))
 954     (license lgpl2.1))) ; all files
 955
 956 (define-public tegaki-wagomu-japanese
 957   (package
 958     (inherit tegaki-zinnia-japanese)
 959     (name "tegaki-wagomu-japanese")
 960     (version "0.3")
 961     (source
 962      (origin
 963        (method url-fetch)
 964        (uri (tegaki-release-uri name version "zip"))
 965        (sha256
 966         (base32
 967          "0flj5id8xwsn7csrrzqz9prdikswnwm2wms0as2vzdpxzph1az4k"))
 968        (modules remove-pre-compiled-files-modules)
 969        (snippet (remove-pre-compiled-files "model"))))
 970     (license lgpl2.1))) ; all files
 971
 972 (define-public tegaki-wagomu-japanese-kyoiku
 973   (package
 974     (inherit tegaki-zinnia-japanese)
 975     (name "tegaki-wagomu-japanese-kyoiku")
 976     (version "0.3")
 977     (source
 978      (origin
 979        (method url-fetch)
 980        (uri (tegaki-release-uri name version "zip"))
 981        (sha256
 982         (base32
 983          "0v8crfh8rdf6ndp16g52s5jlrrlwh73xp38zjn5i9dlacx8kfqg1"))
 984        (modules remove-pre-compiled-files-modules)
 985        (snippet (remove-pre-compiled-files "model"))))
 986     (license lgpl2.1))) ; all files
 987
 988 (define-public tegaki-wagomu-japanese-joyo
 989   (package
 990     (inherit tegaki-zinnia-japanese)
 991     (name "tegaki-wagomu-japanese-joyo")
 992     (version "0.3")
 993     (source
 994      (origin
 995        (method url-fetch)
 996        (uri (tegaki-release-uri name version "zip"))
 997        (sha256
 998         (base32
 999          "0wk8shpr963zp328g991qs6abpnacq4242003m687z2d6yp7nph2"))
1000        (modules remove-pre-compiled-files-modules)
1001        (snippet (remove-pre-compiled-files "model"))))
1002     (license lgpl2.1))) ; all files
1003
1004 (define-public tegaki-wagomu-simplified-chinese
1005   (package
1006     (inherit tegaki-zinnia-japanese)
1007     (name "tegaki-wagomu-simplified-chinese")
1008     (version "0.3")
1009     (source
1010      (origin
1011        (method url-fetch)
1012        (uri (tegaki-release-uri name version "zip"))
1013        (sha256
1014         (base32
1015          "0wqprynigqxqxv128i1smh81gxvmjj056d9qpznxa3n9f5ymlbj6"))
1016        (modules remove-pre-compiled-files-modules)
1017        (snippet (remove-pre-compiled-files "model"))))
1018     (license lgpl2.1))) ; all files
1019
1020 ;;; Upstream does not provide the source for tegaki-wagomu-traditional-chinese.
1021 ;;; Therefore, we use the source for tegaki-zinnia-traditional-chinese and
1022 ;;; patch the Makefile accordingly.
1023 (define-public tegaki-wagomu-traditional-chinese
1024   (package
1025     (inherit tegaki-zinnia-traditional-chinese)
1026     (name "tegaki-wagomu-traditional-chinese")
1027     (arguments
1028      (substitute-keyword-arguments
1029          (package-arguments tegaki-zinnia-traditional-chinese)
1030        ((#:phases phases '%standard-phases)
1031         `(modify-phases ,phases
1032            (replace 'configure
1033              (lambda args
1034                (let ((configure (assq-ref ,phases 'configure)))
1035                  (apply configure args))
1036                (substitute* "Makefile"
1037                  (("zinnia") "wagomu"))
1038                #t))))))
1039     (license lgpl2.1))) ; all files
1040
1041 (define-public link-grammar
1042   (package
1043     (name "link-grammar")
1044     (version "5.7.0")
1045     (source (origin
1046               (method url-fetch)
1047               (uri (string-append "https://www.abisource.com/downloads/"
1048                                   "link-grammar/" version
1049                                   "/link-grammar-" version ".tar.gz"))
1050               (sha256
1051                (base32
1052                 "0ak1v469k56v3511kxxkxvx1nw6zcxcl0f1kcvc82ffacqbr4y96"))))
1053     (build-system gnu-build-system)
1054     (home-page "https://www.abisource.com/projects/link-grammar/")
1055     (synopsis "Link grammar parser")
1056     (description "The Link Grammar Parser is a syntactic parser of English,
1057 Russian, Arabic and Persian (and other languages as well), based on Link
1058 Grammar, an original theory of syntax and morphology.  Given a sentence, the
1059 system assigns to it a syntactic structure, which consists of a set of
1060 labelled links connecting pairs of words.  The parser also produces a
1061 \"constituent\" (HPSG style phrase tree) representation of a sentence (showing
1062 noun phrases, verb phrases, etc.).")
1063     (license bsd-3)))
1064
1065 (define-public praat
1066   (package
1067     (name "praat")
1068     (version "6.1.16")
1069     (source (origin
1070               (method git-fetch)
1071               (uri (git-reference
1072                      (url "https://github.com/praat/praat")
1073                      (commit (string-append "v" version))))
1074               (file-name (git-file-name name version))
1075               (sha256
1076                (base32
1077                 "1rx4qvl0dd85x0r6pl0zk4bysx9ykxl05kywjr4pyvv6dvpswkrm"))))
1078     (build-system gnu-build-system)
1079     (arguments
1080      `(#:tests? #f; no test target
1081        #:phases
1082        (modify-phases %standard-phases
1083          (replace 'configure
1084            (lambda _
1085              (copy-file "makefiles/makefile.defs.linux.pulse" "makefile.defs")
1086              #t))
1087          (replace 'install
1088            (lambda* (#:key outputs #:allow-other-keys)
1089              (let* ((out (assoc-ref outputs "out"))
1090                     (bin (string-append out "/bin")))
1091                (mkdir-p bin)
1092                (copy-file "praat" (string-append bin "/praat")))
1093              #t)))))
1094     (inputs
1095      `(("alsa-lib" ,alsa-lib)
1096        ("gtk" ,gtk+-2)
1097        ("jack" ,jack-1)
1098        ("publesaudio" ,pulseaudio)))
1099     (native-inputs
1100      `(("pkg-config" ,pkg-config)))
1101     (home-page "http://www.fon.hum.uva.nl/praat/")
1102     (synopsis "Doing phonetics by computer")
1103     (description "Praat is a tool to perform phonetics tasks.  It can do speach
1104 analysis (pitch, formant, intensity, ...), speach synthesis, labeling, segmenting
1105 and manipulation.")
1106     (license gpl2+)))