gnu/packages/textutils.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2015 Taylan Ulrich Bayırlı/Kammer <taylanbayirli@gmail.com>
   3 ;;; Copyright © 2015, 2016, 2017 Ricardo Wurmus <rekado@elephly.net>
   4 ;;; Copyright © 2015, 2016 Ben Woodcroft <donttrustben@gmail.com>
   5 ;;; Copyright © 2015 Roel Janssen <roel@gnu.org>
   6 ;;; Copyright © 2016 Jelle Licht <jlicht@fsfe.org>
   7 ;;; Copyright © 2016 Alex Griffin <a@ajgrf.com>
   8 ;;; Copyright © 2016, 2018 Efraim Flashner <efraim@flashner.co.il>
   9 ;;; Copyright © 2016 Nils Gillmann <ng0@n0.is>
  10 ;;; Copyright © 2016 Marius Bakke <mbakke@fastmail.com>
  11 ;;; Copyright © 2017 Eric Bavier <bavier@member.fsf.org>
  12 ;;; Copyright © 2017 Rene Saavedra <rennes@openmailbox.org>
  13 ;;; Copyright © 2017 Hartmut Goebel <h.goebel@crazy-compilers.com>
  14 ;;; Copyright © 2017 Kei Kebreau <kkebreau@posteo.net>
  15 ;;; Copyright © 2017 Alex Vong <alexvong1995@gmail.com>
  16 ;;; Copyright © 2018 Tobias Geerinckx-Rice <me@tobias.gr>
  17 ;;;
  18 ;;; This file is part of GNU Guix.
  19 ;;;
  20 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  21 ;;; under the terms of the GNU General Public License as published by
  22 ;;; the Free Software Foundation; either version 3 of the License, or (at
  23 ;;; your option) any later version.
  24 ;;;
  25 ;;; GNU Guix is distributed in the hope that it will be useful, but
  26 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  27 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  28 ;;; GNU General Public License for more details.
  29 ;;;
  30 ;;; You should have received a copy of the GNU General Public License
  31 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  32
  33 (define-module (gnu packages textutils)
  34   #:use-module ((guix licenses) #:prefix license:)
  35   #:use-module (guix packages)
  36   #:use-module (guix download)
  37   #:use-module (guix git-download)
  38   #:use-module (guix build-system ant)
  39   #:use-module (guix build-system gnu)
  40   #:use-module (guix build-system go)
  41   #:use-module (guix build-system cmake)
  42   #:use-module (guix build-system python)
  43   #:use-module (gnu packages)
  44   #:use-module (gnu packages autotools)
  45   #:use-module (gnu packages compression)
  46   #:use-module (gnu packages gettext)
  47   #:use-module (gnu packages java)
  48   #:use-module (gnu packages ncurses)
  49   #:use-module (gnu packages perl)
  50   #:use-module (gnu packages pkg-config)
  51   #:use-module (gnu packages python)
  52   #:use-module (gnu packages readline)
  53   #:use-module (gnu packages slang))
  54
  55 (define-public dos2unix
  56   (package
  57     (name "dos2unix")
  58     (version "7.4.0")
  59     (source
  60      (origin
  61        (method url-fetch)
  62        (uri (string-append "https://waterlan.home.xs4all.nl/" name "/"
  63                            name "-" version ".tar.gz"))
  64        (sha256
  65         (base32 "12h4c61g376bhq03y5g2xszkrkrj5hwd928rly3xsp6rvfmnbixs"))))
  66     (build-system gnu-build-system)
  67     (arguments
  68      '(#:make-flags (list "CC=gcc"
  69                           (string-append "prefix=" (assoc-ref %outputs "out")))
  70        #:phases
  71        (modify-phases %standard-phases
  72          (delete 'configure)))) ; no configure script
  73     (native-inputs
  74      `(("gettext" ,gettext-minimal)
  75        ("perl" ,perl)))
  76     (home-page "https://waterlan.home.xs4all.nl/dos2unix.html")
  77     (synopsis "DOS/Mac to Unix and vice versa text file format converter")
  78     (description
  79      "dos2unix is a tool to convert line breaks in a text file from Unix format
  80 to DOS format and vice versa.")
  81     (license license:bsd-2)))
  82
  83 (define-public recode
  84   (package
  85     (name "recode")
  86     (version "3.7")
  87     (source
  88      (origin
  89        (method url-fetch)
  90        (uri (string-append "https://github.com/rrthomas/recode/releases/"
  91                            "download/v" version "/" name "-" version ".tar.gz"))
  92        (sha256
  93         (base32
  94          "0r4yhf7i7zp2nl2apyzz7r3i2in12n385hmr8zcfr18ly0ly530q"))
  95        (modules '((guix build utils)))
  96        (snippet
  97         `(begin
  98            (delete-file "tests/Recode.c")
  99            #t))))
 100     (build-system gnu-build-system)
 101     (native-inputs
 102      `(("python" ,python-2)
 103        ("python2-cython" ,python2-cython)))
 104     (home-page "https://github.com/rrthomas/recode")
 105     (synopsis "Text encoding converter")
 106     (description "The Recode library converts files between character sets and
 107 usages.  It recognises or produces over 200 different character sets (or about
 108 300 if combined with an iconv library) and transliterates files between almost
 109 any pair.  When exact transliteration are not possible, it gets rid of
 110 offending characters or falls back on approximations.  The recode program is a
 111 handy front-end to the library.")
 112     (license license:gpl3+)))
 113
 114 (define-public enca
 115   (package
 116     (name "enca")
 117     (version "1.19")
 118     (source
 119      (origin
 120        (method url-fetch)
 121        (uri (string-append
 122              "https://github.com/nijel/enca/archive/" version ".tar.gz"))
 123        (sha256
 124         (base32 "099z526i7qgij7q1w3lvhl88iv3jc3nqxca2i09h6s08ghyrmzf4"))
 125        (file-name (string-append name "-" version ".tar.gz"))))
 126     (build-system gnu-build-system)
 127     ;; enca-1.19 tests fail with recent recode.
 128     ;(inputs `(("recode" ,recode)))
 129     (home-page "https://github.com/nijel/enca")
 130     (synopsis "Text encoding detection tool")
 131     (description "Enca (Extremely Naive Charset Analyser) consists of libenca,
 132 an encoding detection library, and enca, a command line frontend, integrating
 133 libenca and several charset conversion libraries and tools.")
 134     (license license:gpl2)))
 135
 136 (define-public utf8proc
 137   (package
 138     (name "utf8proc")
 139     (version "2.1.0")
 140     (source
 141      (origin
 142        (method url-fetch)
 143        (uri (string-append
 144              "https://github.com/JuliaLang/utf8proc/archive/v"
 145              version ".tar.gz"))
 146        (file-name (string-append name "-" version ".tar.gz"))
 147        (sha256
 148         (base32 "0q1jhdkk4f9b0zb8s2ql3sba3br5nvjsmbsaybmgj064k9hwbk15"))))
 149     (build-system gnu-build-system)
 150     (inputs                  ;test data that is otherwise downloaded with curl
 151      `(("NormalizationTest.txt"
 152         ,(origin
 153            (method url-fetch)
 154            (uri (string-append "http://www.unicode.org/Public/9.0.0/ucd/"
 155                                "NormalizationTest.txt"))
 156            (sha256
 157             (base32 "1fxrz0bilsbwl685336aqi88k62i6nqhm62rvy4zhg3bcm4dhj1d"))))
 158        ("GraphemeBreakTest.txt"
 159         ,(origin
 160            (method url-fetch)
 161            (uri (string-append "http://www.unicode.org/Public/9.0.0/ucd/"
 162                                "auxiliary/GraphemeBreakTest.txt"))
 163            (sha256
 164             (base32 "0qbhyhmf0778lc2hcwlpizrvmdxwpk959v2q2wb8abv09ba7wvn7"))))))
 165     (arguments
 166      '(#:make-flags (list "CC=gcc"
 167                           (string-append "prefix=" (assoc-ref %outputs "out")))
 168        #:phases
 169        (modify-phases %standard-phases
 170          (delete 'configure)
 171          (add-before 'check 'check-data
 172            (lambda* (#:key inputs #:allow-other-keys)
 173              (for-each (lambda (i)
 174                          (copy-file (assoc-ref inputs i)
 175                                     (string-append "data/" i)))
 176                        '("NormalizationTest.txt" "GraphemeBreakTest.txt"))
 177              (substitute* "data/GraphemeBreakTest.txt"
 178                (("÷") "/")
 179                (("×") "+")))))))
 180     (home-page "http://julialang.org/utf8proc/")
 181     (synopsis "C library for processing UTF-8 Unicode data")
 182     (description "utf8proc is a small C library that provides Unicode
 183 normalization, case-folding, and other operations for data in the UTF-8
 184 encoding, supporting Unicode version 9.0.0.")
 185     (license license:expat)))
 186
 187 (define-public libgtextutils
 188   (package
 189     (name "libgtextutils")
 190     (version "0.7")
 191     (source
 192      (origin
 193        (method url-fetch)
 194        (uri (string-append
 195              "https://github.com/agordon/libgtextutils/releases/download/"
 196              version "/libgtextutils-" version ".tar.gz"))
 197        (sha256
 198         (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
 199     (build-system gnu-build-system)
 200     (arguments
 201      '(#:phases
 202        (modify-phases %standard-phases
 203          (add-after 'unpack 'autoreconf
 204            (lambda _ (zero? (system* "autoreconf" "-vif")))))))
 205     (native-inputs
 206      `(("autoconf" ,autoconf)
 207        ("automake" ,automake)
 208        ("libtool" ,libtool)))
 209     (home-page "https://github.com/agordon/libgtextutils")
 210     (synopsis "Gordon's text utils library")
 211     (description
 212      "libgtextutils is a text utilities library used by the fastx toolkit from
 213 the Hannon Lab.")
 214     (license license:agpl3+)))
 215
 216 (define-public cityhash
 217   (let ((commit "8af9b8c"))
 218     (package
 219       (name "cityhash")
 220       (version (string-append "1.1-2." commit))
 221       (source (origin
 222                 (method git-fetch)
 223                 (uri (git-reference
 224                       (url "https://github.com/google/cityhash.git")
 225                       (commit commit)))
 226                 (file-name (string-append name "-" version ".tar.gz"))
 227                 (sha256
 228                  (base32
 229                   "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
 230       (build-system gnu-build-system)
 231       (arguments
 232        '(#:make-flags (list "CXXFLAGS=-g -O3")
 233          #:phases
 234          (modify-phases %standard-phases
 235            ;; citycrc is not installed by default but is used by some
 236            ;; programs.
 237            (add-after 'install 'install-citycrc
 238              (lambda* (#:key outputs #:allow-other-keys)
 239                (let* ((out (assoc-ref outputs "out"))
 240                       (include (string-append out "/include")))
 241                  (install-file "src/citycrc.h" include))
 242                #t)))))
 243       (home-page "https://github.com/google/cityhash")
 244       (synopsis "C++ hash functions for strings")
 245       (description
 246        "CityHash provides hash functions for strings.  The functions mix the
 247 input bits thoroughly but are not suitable for cryptography.")
 248       (license license:expat))))
 249
 250 (define-public ustr
 251   (package
 252     (name "ustr")
 253     (version "1.0.4")
 254     (source (origin
 255               (method url-fetch)
 256               (uri (string-append "http://www.and.org/ustr/" version
 257                                   "/ustr-" version ".tar.bz2"))
 258               (sha256
 259                (base32
 260                 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
 261               (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
 262     (build-system gnu-build-system)
 263     (arguments
 264      `(#:make-flags
 265        (list "CC=gcc"
 266              "HIDE="
 267              ;; Override "/sbin/ldconfig" with "echo" because we don't need
 268              ;; "ldconfig".
 269              "LDCONFIG=echo"
 270              (string-append "prefix=" (assoc-ref %outputs "out"))
 271              "all-shared")
 272        #:phases
 273        (modify-phases %standard-phases
 274          (add-after 'unpack 'disable-check-for-stdint
 275            (lambda _
 276              ;; Of course we have stdint.h, just not in /usr/include
 277              (substitute* '("Makefile"
 278                             "ustr-import.in")
 279                (("-f \"/usr/include/stdint.h\"") "-z \"\""))
 280              #t))
 281          ;; No configure script
 282          (delete 'configure))))
 283     (home-page "http://www.and.org/ustr/")
 284     (synopsis "String library with very low memory overhead")
 285     (description
 286      "Ustr is a string library for C with very low memory overhead.")
 287     ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
 288     ;; LGPL, etc. ... if you need another license to help compatibility, just
 289     ;; ask for it.  It's basically public domain, without all the legal
 290     ;; problems for everyone that trying to make something public domain
 291     ;; entails."
 292     (license license:public-domain)))
 293
 294 (define-public libconfig
 295   (package
 296     (name "libconfig")
 297     (version "1.5")
 298     (source (origin
 299               (method url-fetch)
 300               (uri (string-append "http://www.hyperrealm.com/libconfig/"
 301                                   "libconfig-" version ".tar.gz"))
 302               (sha256
 303                (base32
 304                 "1xh3hzk63v4y8815lc5209m3s6ms2cpgw4h5hg462i4f1lwsl7g3"))))
 305     (build-system gnu-build-system)
 306     (home-page "http://www.hyperrealm.com/libconfig/")
 307     (synopsis "C/C++ configuration file library")
 308     (description
 309      "Libconfig is a simple library for manipulating structured configuration
 310 files.  This file format is more compact and more readable than XML.  And
 311 unlike XML, it is type-aware, so it is not necessary to do string parsing in
 312 application code.")
 313     (license license:lgpl2.1+)))
 314
 315 (define-public pfff
 316   (package
 317     (name "pfff")
 318     (version "1.0")
 319     (source (origin
 320               (method url-fetch)
 321               (uri (string-append "https://github.com/pfff/pfff/archive/v"
 322                                   version ".tar.gz"))
 323               (file-name (string-append name "-" version ".tar.gz"))
 324               (sha256
 325                (base32
 326                 "00m553aa277iarxj6dalmklyb64r7ias49bfwzbacsfg8h3kar8m"))))
 327     (build-system cmake-build-system)
 328     (home-page "http://biit.cs.ut.ee/pfff/")
 329     (synopsis "Probabilistic fast file fingerprinting tool")
 330     (description
 331      "pfff is a tool for calculating a compact digital fingerprint of a file
 332 by sampling randomly from the file instead of reading it in full.
 333 Consequently, the computation has a flat performance characteristic,
 334 correlated with data variation rather than file size.  pfff can be as reliable
 335 as existing hashing techniques, with provably negligible risk of collisions.")
 336     (license license:bsd-3)))
 337
 338 (define-public oniguruma
 339   (package
 340     (name "oniguruma")
 341     (version "5.9.6")
 342     (source (origin
 343               (method url-fetch)
 344               (uri (string-append "https://github.com/kkos/"
 345                                   "oniguruma/releases/download/v" version
 346                                   "/onig-" version ".tar.gz"))
 347               (sha256
 348                (base32
 349                 "19s79vsclqn170mw0ajwv7j37qsbn4f1yjz3yavnhvva6c820r6m"))))
 350     (build-system gnu-build-system)
 351     (home-page "https://github.com/kkos/oniguruma")
 352     (synopsis "Regular expression library")
 353     (description "Oniguruma is a regular expressions library.  The special
 354 characteristic of this library is that different character encoding for every
 355 regular expression object can be specified.")
 356     (license license:bsd-2)))
 357
 358 (define-public antiword
 359   (package
 360     (name "antiword")
 361     (version "0.37")
 362     (source (origin
 363               (method url-fetch)
 364               (uri (string-append "http://www.winfield.demon.nl/linux"
 365                                   "/antiword-" version ".tar.gz"))
 366               (sha256
 367                (base32
 368                 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
 369               (patches (search-patches "antiword-CVE-2014-8123.patch"))))
 370     (build-system gnu-build-system)
 371     (arguments
 372      `(#:tests? #f ; There are no tests
 373        #:make-flags
 374        (list "-f" "Makefile.Linux"
 375              (string-append "GLOBAL_INSTALL_DIR="
 376                             (assoc-ref %outputs "out") "/bin")
 377              (string-append "GLOBAL_RESOURCES_DIR="
 378                             (assoc-ref %outputs "out") "/share/antiword"))
 379        #:phases
 380        (modify-phases %standard-phases
 381          (replace 'configure
 382            (lambda* (#:key outputs #:allow-other-keys)
 383              ;; Ensure that mapping files can be found in the actual package
 384              ;; data directory.
 385              (substitute* "antiword.h"
 386                (("/usr/share/antiword")
 387                 (string-append (assoc-ref outputs "out") "/share/antiword")))
 388              #t))
 389          (replace 'install
 390            (lambda* (#:key make-flags #:allow-other-keys)
 391              (zero? (apply system* "make" `("global_install" ,@make-flags))))))))
 392     (home-page "http://www.winfield.demon.nl/")
 393     (synopsis "Microsoft Word document reader")
 394     (description "Antiword is an application for displaying Microsoft Word
 395 documents.  It can also convert the document to PostScript or XML.  Only
 396 documents made by MS Word version 2 and version 6 or later are supported.  The
 397 name comes from: \"The antidote against people who send Microsoft Word files
 398 to everybody, because they believe that everybody runs Windows and therefore
 399 runs Word\".")
 400     (license license:gpl2+)))
 401
 402 (define-public catdoc
 403   (package
 404     (name "catdoc")
 405     (version "0.95")
 406     (source (origin
 407               (method url-fetch)
 408               (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
 409                                   "catdoc-" version ".tar.gz"))
 410               (patches (search-patches "catdoc-CVE-2017-11110.patch"))
 411               (sha256
 412                (base32
 413                 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
 414     (build-system gnu-build-system)
 415     ;; TODO: Also build `wordview` which requires `tk` – make a separate
 416     ;; package for this.
 417     (arguments
 418      '(#:tests? #f ; There are no tests
 419        #:configure-flags '("--disable-wordview")
 420        #:phases
 421        (modify-phases %standard-phases
 422          (add-before 'install 'fix-install
 423            (lambda* (#:key outputs #:allow-other-keys)
 424              (let ((out (assoc-ref outputs "out")))
 425                (mkdir-p (string-append out "/share/man/man1"))))))))
 426     (home-page "http://www.wagner.pp.ru/~vitus/software/catdoc/")
 427     (synopsis "MS-Word to TeX or plain text converter")
 428     (description "@command{catdoc} extracts text from MS-Word files, trying to
 429 preserve as many special printable characters as possible.  It supports
 430 everything up to Word-97. Also supported are MS Write documents and RTF files.
 431
 432 @command{catdoc} does not preserve complex word formatting, but it can
 433 translate some non-ASCII characters into TeX escape codes.  It's goal is to
 434 extract plain text and allow you to read it and, probably, reformat with TeX,
 435 according to TeXnical rules.
 436
 437 This package also provides @command{xls2csv}, which extracts data from Excel
 438 spreadsheets and outputs it in comma-separated-value format, and
 439 @command{catppt}, which extracts data from PowerPoint presentations.")
 440     (license license:gpl2+)))
 441
 442 (define-public utfcpp
 443   (package
 444     (name "utfcpp")
 445     (version "2.3.5")
 446     (source (origin
 447               (method url-fetch)
 448               (uri
 449                (string-append "https://github.com/nemtrif/utfcpp/archive/v"
 450                               version ".tar.gz"))
 451               (file-name (string-append name "-" version ".tar.gz"))
 452               (sha256
 453                (base32
 454                 "0gcqcfw19kfim8xw29xdp91l310yfjyrqdj2zsx8xx02dkpy1zzk"))))
 455     (build-system cmake-build-system)
 456     (arguments
 457      `(#:out-of-source? #f
 458        #:phases
 459        (modify-phases %standard-phases
 460          (replace 'install              ; no install target
 461            (lambda* (#:key outputs #:allow-other-keys)
 462              (let* ((out (assoc-ref outputs "out"))
 463                     (include (string-append out "/include"))
 464                     (doc (string-append out "/share/doc/" ,name)))
 465                (copy-recursively "source" include)
 466                (install-file "README.md" doc)
 467                #t))))))
 468     (home-page "https://github.com/nemtrif/utfcpp")
 469     (synopsis "Portable C++ library for handling UTF-8")
 470     (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
 471 in a portable way.")
 472     (license license:boost1.0)))
 473
 474 (define-public dbacl
 475   (package
 476     (name "dbacl")
 477     (version "1.14")
 478     (source
 479      (origin
 480        (method url-fetch)
 481        (uri (string-append "http://www.lbreyer.com/gpl/"
 482                            name "-" version ".tar.gz"))
 483        (sha256
 484         (base32
 485          "0224g6x71hyvy7jikfxmgcwww1r5lvk0jx36cva319cb9nmrbrq7"))))
 486     (build-system gnu-build-system)
 487     (arguments
 488      `(#:make-flags
 489        (list
 490         (string-append "-I" (assoc-ref %build-inputs "slang")
 491                        "/include/slang")
 492         (string-append "-I" (assoc-ref %build-inputs "ncurses")
 493                        "/include/ncurses"))
 494        #:phases
 495        (modify-phases %standard-phases
 496          (add-after 'unpack 'delete-sample6-and-japanese
 497            (lambda _
 498              (substitute* "doc/Makefile.am"
 499                (("sample6.txt") "")
 500                (("japanese.txt") ""))
 501              (delete-file "doc/sample6.txt")
 502              (delete-file "doc/japanese.txt")
 503              (substitute* (list "src/tests/Makefile.am"
 504                                 "src/tests/Makefile.in")
 505                (("dbacl-jap.shin") "")
 506                (("dbacl-jap.sh") ""))
 507              #t))
 508          (add-after 'unpack 'delete-test
 509            ;; See comments about the license.
 510            (lambda _
 511              (delete-file "src/tests/dbacl-jap.shin")
 512              #t))
 513          (add-after 'delete-sample6-and-japanese 'autoreconf
 514            (lambda _
 515              (invoke "autoreconf" "-vif")
 516              #t))
 517          (add-after 'unpack 'fix-test-files
 518            (lambda* (#:key inputs outputs #:allow-other-keys)
 519              (let* ((out (assoc-ref outputs "out"))
 520                     (bin (string-append out "/bin")))
 521                (substitute* (find-files "src/tests/" "\\.shin$")
 522                  (("PATH=/bin:/usr/bin")
 523                   "#PATH=/bin:/usr/bin")
 524                  (("diff") (string-append (which "diff")))
 525                  (("tr") (string-append (which "tr"))))
 526                #t))))))
 527     (inputs
 528      `(("ncurses" ,ncurses)
 529        ("perl" ,perl)
 530        ("readline" ,readline)
 531        ("slang" ,slang)))
 532     (native-inputs
 533      `(("libtool" ,libtool)
 534        ("autoconf" ,autoconf)
 535        ("automake" ,automake)
 536        ("pkg-config" ,pkg-config)))
 537     (home-page "https://www.lbreyer.com/dbacl.html")
 538     (synopsis "Bayesian text and email classifier")
 539     (description
 540      "dbacl is a fast Bayesian text and email classifier.  It builds a variety
 541 of language models using maximum entropy (minimum divergence) principles, and
 542 these can then be used to categorize input data automatically among multiple
 543 categories.")
 544     ;; The software is licensed as GPLv3 or later, but
 545     ;; includes various sample texts in the doc dir:
 546     ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
 547     ;;   by Mark Twain.
 548     ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
 549     ;; - sample6.txt is a forwarded email, copyright unknown.
 550     ;;   Guix does exclude sample6.txt.
 551     ;; - japanese.txt is a Japanese unoffical translation of the
 552     ;;   GNU General Public License, (c) by the Free Software Foundation.
 553     ;;   Guix excludes this file.
 554     (license (list license:gpl3+ license:public-domain))))
 555
 556 (define-public dotconf
 557   (package
 558     (name "dotconf")
 559     (version "1.3")
 560     (source (origin
 561               (method url-fetch)
 562               (uri (string-append
 563                     "https://github.com/williamh/dotconf/archive/v"
 564                     version ".tar.gz"))
 565               (file-name (string-append name "-" version ".tar.gz"))
 566               (sha256
 567                (base32
 568                 "0lsnh0yaw44psmx59hq94cj1932gscp5h8d3cnh05l0svr0cy7kz"))))
 569     (build-system gnu-build-system)
 570     (arguments
 571      `(#:tests? #f ; FIXME maketest.sh does not work.
 572        #:phases
 573        (modify-phases %standard-phases
 574          (add-after 'unpack 'autoreconf
 575            (lambda _
 576              (zero? (system* "autoreconf" "-vif")))))))
 577     (native-inputs
 578      `(("autoconf" ,autoconf)
 579        ("automake" ,automake)
 580        ("libtool" ,libtool)))
 581     (home-page "https://github.com/williamh/dotconf")
 582     (synopsis "Configuration file parser library")
 583     (description
 584      "C library for creating and parsing configuration files.")
 585     (license (list license:lgpl2.1         ; Main distribution.
 586                    license:asl1.1))))      ; src/readdir.{c,h}
 587
 588 (define-public java-rsyntaxtextarea
 589   (package
 590     (name "java-rsyntaxtextarea")
 591     (version "2.6.1")
 592     (source (origin
 593               (method url-fetch)
 594               (uri (string-append "https://github.com/bobbylight/"
 595                                   "RSyntaxTextArea/archive/"
 596                                   version ".tar.gz"))
 597               (file-name (string-append name "-" version ".tar.gz"))
 598               (sha256
 599                (base32
 600                 "0c5mqg2klj5rvf8fhycrli8rf6s37l9p7a8knw9gpp65r1c120q2"))))
 601     (build-system ant-build-system)
 602     (arguments
 603      `(;; FIXME: some tests fail because locale resources cannot be found.
 604        ;; Even when I add them to the class path,
 605        ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
 606        #:tests? #f
 607        #:jar-name "rsyntaxtextarea.jar"))
 608     (native-inputs
 609      `(("java-junit" ,java-junit)
 610        ("java-hamcrest-core" ,java-hamcrest-core)))
 611     (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
 612     (synopsis "Syntax highlighting text component for Java Swing")
 613     (description "RSyntaxTextArea is a syntax highlighting, code folding text
 614 component for Java Swing.  It extends @code{JTextComponent} so it integrates
 615 completely with the standard @code{javax.swing.text} package.  It is fast and
 616 efficient, and can be used in any application that needs to edit or view
 617 source code.")
 618     (license license:bsd-3)))
 619
 620 ;; We use the sources from git instead of the tarball from pypi, because the
 621 ;; latter does not include the Cython source file from which bycython.cpp is
 622 ;; generated.
 623 (define-public python-editdistance
 624   (let ((commit "3ea84a7dd3258c76aa3be851ef3d50e59c886846")
 625         (revision "1"))
 626     (package
 627       (name "python-editdistance")
 628       (version (string-append "0.3.1-" revision "." (string-take commit 7)))
 629       (source
 630        (origin
 631          (method git-fetch)
 632          (uri (git-reference
 633                (url "https://github.com/aflc/editdistance.git")
 634                (commit commit)))
 635          (file-name (git-file-name name version))
 636          (sha256
 637           (base32
 638            "1l43svsv12crvzphrgi6x435z6xg8m086c64armp8wzb4l8ccm7g"))))
 639       (build-system python-build-system)
 640       (arguments
 641        `(#:phases
 642          (modify-phases %standard-phases
 643            (add-after 'unpack 'build-cython-code
 644              (lambda _
 645                (with-directory-excursion "editdistance"
 646                  (delete-file "bycython.cpp")
 647                  (invoke "cython" "--cplus" "bycython.pyx")))))))
 648       (native-inputs
 649        `(("python-cython" ,python-cython)))
 650       (home-page "https://www.github.com/aflc/editdistance")
 651       (synopsis "Fast implementation of the edit distance (Levenshtein distance)")
 652       (description
 653        "This library simply implements Levenshtein distance algorithm with C++
 654 and Cython.")
 655       (license license:expat))))
 656
 657 (define-public go-github.com-mattn-go-runewidth
 658   (package
 659     (name "go-github.com-mattn-go-runewidth")
 660     (version "0.0.2")
 661     (source (origin
 662               (method git-fetch)
 663               (uri (git-reference
 664                     (url "https://github.com/mattn/go-runewidth")
 665                     (commit (string-append "v" version))))
 666               (file-name (git-file-name name version))
 667               (sha256
 668                (base32
 669                 "0vkrfrz3fzn5n6ix4k8s0cg0b448459sldq8bp4riavsxm932jzb"))))
 670     (build-system go-build-system)
 671     (arguments
 672      '(#:import-path "github.com/mattn/go-runewidth"))
 673     (synopsis "@code{runewidth} provides Go functions to work with string widths")
 674     (description
 675      "The @code{runewidth} library provides Go functions for padding,
 676 measuring and checking the width of strings, with support east asian text.")
 677     (home-page "https://github.com/jessevdk/go-flags")
 678     (license license:expat)))