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