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 Efraim Flashner <efraim@flashner.co.il>
   9 ;;; Copyright © 2016 ng0 <ng0@we.make.ritual.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 ;;;
  17 ;;; This file is part of GNU Guix.
  18 ;;;
  19 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  20 ;;; under the terms of the GNU General Public License as published by
  21 ;;; the Free Software Foundation; either version 3 of the License, or (at
  22 ;;; your option) any later version.
  23 ;;;
  24 ;;; GNU Guix is distributed in the hope that it will be useful, but
  25 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  26 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  27 ;;; GNU General Public License for more details.
  28 ;;;
  29 ;;; You should have received a copy of the GNU General Public License
  30 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  31
  32 (define-module (gnu packages textutils)
  33   #:use-module ((guix licenses) #:prefix license:)
  34   #:use-module (guix packages)
  35   #:use-module (guix download)
  36   #:use-module (guix git-download)
  37   #:use-module (guix build-system ant)
  38   #:use-module (guix build-system gnu)
  39   #:use-module (guix build-system cmake)
  40   #:use-module (guix build-system trivial)
  41   #:use-module (gnu packages)
  42   #:use-module (gnu packages autotools)
  43   #:use-module (gnu packages compression)
  44   #:use-module (gnu packages gettext)
  45   #:use-module (gnu packages java)
  46   #:use-module (gnu packages ncurses)
  47   #:use-module (gnu packages perl)
  48   #:use-module (gnu packages pkg-config)
  49   #:use-module (gnu packages python)
  50   #:use-module (gnu packages readline)
  51   #:use-module (gnu packages slang))
  52
  53 (define-public dos2unix
  54   (package
  55     (name "dos2unix")
  56     (version "7.3.4")
  57     (source
  58      (origin
  59        (method url-fetch)
  60        (uri (string-append "https://waterlan.home.xs4all.nl/" name "/"
  61                            name "-" version ".tar.gz"))
  62        (sha256
  63         (base32 "1i9hbxn0br7xa18z4bjpkdv7mrzmbfxhm44mzpd07yd2qnxsgkcc"))))
  64     (build-system gnu-build-system)
  65     (arguments
  66      '(#:make-flags (list "CC=gcc"
  67                           (string-append "prefix=" (assoc-ref %outputs "out")))
  68        #:phases
  69        (modify-phases %standard-phases
  70          (delete 'configure)))) ; no configure script
  71     (native-inputs
  72      `(("gettext" ,gettext-minimal)
  73        ("perl" ,perl)))
  74     (home-page "https://waterlan.home.xs4all.nl/dos2unix.html")
  75     (synopsis "DOS/Mac to Unix and vice versa text file format converter")
  76     (description
  77      "dos2unix is a tool to convert line breaks in a text file from Unix format
  78 to DOS format and vice versa.")
  79     (license license:bsd-2)))
  80
  81 (define-public recode
  82   (package
  83     (name "recode")
  84     ;; Last beta release (3.7-beta2) is from 2008; last commit from Feb 2014.
  85     ;; So we use that commit instead.
  86     (version "3.7.0.201402")
  87     (source
  88      (origin
  89        (method git-fetch)
  90        (uri (git-reference
  91              (url "https://github.com/pinard/Recode.git")
  92              (commit "2d7092a9999194fc0e9449717a8048c8d8e26c18")))
  93        (sha256
  94         (base32 "1wssv8z6g3ryrw33sksz4rjhlnhgvvdqszw1ggl4rcwks34n86zm"))
  95        (file-name (string-append name "-" version "-checkout"))))
  96     (build-system gnu-build-system)
  97     (native-inputs `(("python" ,python-2)))
  98     (arguments
  99      '(#:phases
 100        (alist-cons-before
 101         'check 'pre-check
 102         (lambda _
 103           (substitute* "tests/setup.py"
 104             (("([[:space:]]*)include_dirs=.*" all space)
 105              (string-append all space "library_dirs=['../src/.libs'],\n")))
 106           ;; The test extension 'Recode.so' lacks RUNPATH for 'librecode.so'.
 107           (setenv "LD_LIBRARY_PATH" (string-append (getcwd) "/src/.libs")))
 108         %standard-phases)))
 109     (home-page "https://github.com/pinard/Recode")
 110     (synopsis "Text encoding converter")
 111     (description "The Recode library converts files between character sets and
 112 usages.  It recognises or produces over 200 different character sets (or about
 113 300 if combined with an iconv library) and transliterates files between almost
 114 any pair.  When exact transliteration are not possible, it gets rid of
 115 offending characters or falls back on approximations.  The recode program is a
 116 handy front-end to the library.")
 117     (license license:gpl2+)))
 118
 119 (define-public enca
 120   (package
 121     (name "enca")
 122     (version "1.16")
 123     (source
 124      (origin
 125        (method url-fetch)
 126        (uri (string-append
 127              "https://github.com/nijel/enca/archive/" version ".tar.gz"))
 128        (sha256
 129         (base32 "1xik00x0yvhswsw2isnclabhv536xk1s42cf5z54gfbpbhc7ni8l"))
 130        (file-name (string-append name "-" version ".tar.gz"))))
 131     (build-system gnu-build-system)
 132     (inputs `(("recode" ,recode)))
 133
 134     ;; Both 'test-convert-64.sh' and 'test-convert-filter.sh' manipulate a
 135     ;; 'test.tmp' file, so they have to run in sequence.
 136     (arguments '(#:parallel-tests? #f))
 137
 138     (home-page "https://github.com/nijel/enca")
 139     (synopsis "Text encoding detection tool")
 140     (description "Enca (Extremely Naive Charset Analyser) consists of libenca,
 141 an encoding detection library, and enca, a command line frontend, integrating
 142 libenca and several charset conversion libraries and tools.")
 143     (license license:gpl2)))
 144
 145 (define-public utf8proc
 146   (package
 147     (name "utf8proc")
 148     (version "2.1.0")
 149     (source
 150      (origin
 151        (method url-fetch)
 152        (uri (string-append
 153              "https://github.com/JuliaLang/utf8proc/archive/v"
 154              version ".tar.gz"))
 155        (file-name (string-append name "-" version ".tar.gz"))
 156        (sha256
 157         (base32 "0q1jhdkk4f9b0zb8s2ql3sba3br5nvjsmbsaybmgj064k9hwbk15"))))
 158     (build-system gnu-build-system)
 159     (inputs                  ;test data that is otherwise downloaded with curl
 160      `(("NormalizationTest.txt"
 161         ,(origin
 162            (method url-fetch)
 163            (uri (string-append "http://www.unicode.org/Public/9.0.0/ucd/"
 164                                "NormalizationTest.txt"))
 165            (sha256
 166             (base32 "1fxrz0bilsbwl685336aqi88k62i6nqhm62rvy4zhg3bcm4dhj1d"))))
 167        ("GraphemeBreakTest.txt"
 168         ,(origin
 169            (method url-fetch)
 170            (uri (string-append "http://www.unicode.org/Public/9.0.0/ucd/"
 171                                "auxiliary/GraphemeBreakTest.txt"))
 172            (sha256
 173             (base32 "0qbhyhmf0778lc2hcwlpizrvmdxwpk959v2q2wb8abv09ba7wvn7"))))))
 174     (arguments
 175      '(#:make-flags (list "CC=gcc"
 176                           (string-append "prefix=" (assoc-ref %outputs "out")))
 177        #:phases
 178        (modify-phases %standard-phases
 179          (delete 'configure)
 180          (add-before 'check 'check-data
 181            (lambda* (#:key inputs #:allow-other-keys)
 182              (for-each (lambda (i)
 183                          (copy-file (assoc-ref inputs i)
 184                                     (string-append "data/" i)))
 185                        '("NormalizationTest.txt" "GraphemeBreakTest.txt"))
 186              (substitute* "data/GraphemeBreakTest.txt"
 187                (("÷") "/")
 188                (("×") "+")))))))
 189     (home-page "http://julialang.org/utf8proc/")
 190     (synopsis "C library for processing UTF-8 Unicode data")
 191     (description "utf8proc is a small C library that provides Unicode
 192 normalization, case-folding, and other operations for data in the UTF-8
 193 encoding, supporting Unicode version 9.0.0.")
 194     (license license:expat)))
 195
 196 (define-public libgtextutils
 197   (package
 198     (name "libgtextutils")
 199     (version "0.7")
 200     (source
 201      (origin
 202        (method url-fetch)
 203        (uri (string-append
 204              "https://github.com/agordon/libgtextutils/releases/download/"
 205              version "/libgtextutils-" version ".tar.gz"))
 206        (sha256
 207         (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
 208     (build-system gnu-build-system)
 209     (arguments
 210      '(#:phases
 211        (alist-cons-after
 212         'unpack 'autoreconf
 213         (lambda _ (zero? (system* "autoreconf" "-vif")))
 214         %standard-phases)))
 215     (native-inputs
 216      `(("autoconf" ,autoconf)
 217        ("automake" ,automake)
 218        ("libtool" ,libtool)))
 219     (home-page "https://github.com/agordon/libgtextutils")
 220     (synopsis "Gordon's text utils library")
 221     (description
 222      "libgtextutils is a text utilities library used by the fastx toolkit from
 223 the Hannon Lab.")
 224     (license license:agpl3+)))
 225
 226 (define-public cityhash
 227   (let ((commit "8af9b8c"))
 228     (package
 229       (name "cityhash")
 230       (version (string-append "1.1-2." commit))
 231       (source (origin
 232                 (method git-fetch)
 233                 (uri (git-reference
 234                       (url "https://github.com/google/cityhash.git")
 235                       (commit commit)))
 236                 (file-name (string-append name "-" version ".tar.gz"))
 237                 (sha256
 238                  (base32
 239                   "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
 240       (build-system gnu-build-system)
 241       (arguments
 242        '(#:make-flags (list "CXXFLAGS=-g -O3")
 243          #:phases
 244          (modify-phases %standard-phases
 245            ;; citycrc is not installed by default but is used by some
 246            ;; programs.
 247            (add-after 'install 'install-citycrc
 248              (lambda* (#:key outputs #:allow-other-keys)
 249                (let* ((out (assoc-ref outputs "out"))
 250                       (include (string-append out "/include")))
 251                  (install-file "src/citycrc.h" include))
 252                #t)))))
 253       (home-page "https://github.com/google/cityhash")
 254       (synopsis "C++ hash functions for strings")
 255       (description
 256        "CityHash provides hash functions for strings.  The functions mix the
 257 input bits thoroughly but are not suitable for cryptography.")
 258       (license license:expat))))
 259
 260 (define-public ustr
 261   (package
 262     (name "ustr")
 263     (version "1.0.4")
 264     (source (origin
 265               (method url-fetch)
 266               (uri (string-append "http://www.and.org/ustr/" version
 267                                   "/ustr-" version ".tar.bz2"))
 268               (sha256
 269                (base32
 270                 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
 271               (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
 272     (build-system gnu-build-system)
 273     (arguments
 274      `(#:make-flags
 275        (list "CC=gcc"
 276              "HIDE="
 277              ;; Override "/sbin/ldconfig" with "echo" because we don't need
 278              ;; "ldconfig".
 279              "LDCONFIG=echo"
 280              (string-append "prefix=" (assoc-ref %outputs "out"))
 281              "all-shared")
 282        #:phases
 283        (modify-phases %standard-phases
 284          (add-after 'unpack 'disable-check-for-stdint
 285            (lambda _
 286              ;; Of course we have stdint.h, just not in /usr/include
 287              (substitute* '("Makefile"
 288                             "ustr-import.in")
 289                (("-f \"/usr/include/stdint.h\"") "-z \"\""))
 290              #t))
 291          ;; No configure script
 292          (delete 'configure))))
 293     (home-page "http://www.and.org/ustr/")
 294     (synopsis "String library with very low memory overhead")
 295     (description
 296      "Ustr is a string library for C with very low memory overhead.")
 297     ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
 298     ;; LGPL, etc. ... if you need another license to help compatibility, just
 299     ;; ask for it.  It's basically public domain, without all the legal
 300     ;; problems for everyone that trying to make something public domain
 301     ;; entails."
 302     (license license:public-domain)))
 303
 304 (define-public libconfig
 305   (package
 306     (name "libconfig")
 307     (version "1.5")
 308     (source (origin
 309               (method url-fetch)
 310               (uri (string-append "http://www.hyperrealm.com/libconfig/"
 311                                   "libconfig-" version ".tar.gz"))
 312               (sha256
 313                (base32
 314                 "1xh3hzk63v4y8815lc5209m3s6ms2cpgw4h5hg462i4f1lwsl7g3"))))
 315     (build-system gnu-build-system)
 316     (home-page "http://www.hyperrealm.com/libconfig/")
 317     (synopsis "C/C++ configuration file library")
 318     (description
 319      "Libconfig is a simple library for manipulating structured configuration
 320 files.  This file format is more compact and more readable than XML.  And
 321 unlike XML, it is type-aware, so it is not necessary to do string parsing in
 322 application code.")
 323     (license license:lgpl2.1+)))
 324
 325 (define-public pfff
 326   (package
 327     (name "pfff")
 328     (version "1.0")
 329     (source (origin
 330               (method url-fetch)
 331               (uri (string-append "https://github.com/pfff/pfff/archive/v"
 332                                   version ".tar.gz"))
 333               (file-name (string-append name "-" version ".tar.gz"))
 334               (sha256
 335                (base32
 336                 "00m553aa277iarxj6dalmklyb64r7ias49bfwzbacsfg8h3kar8m"))))
 337     (build-system cmake-build-system)
 338     (home-page "http://biit.cs.ut.ee/pfff/")
 339     (synopsis "Probabilistic fast file fingerprinting tool")
 340     (description
 341      "pfff is a tool for calculating a compact digital fingerprint of a file
 342 by sampling randomly from the file instead of reading it in full.
 343 Consequently, the computation has a flat performance characteristic,
 344 correlated with data variation rather than file size.  pfff can be as reliable
 345 as existing hashing techniques, with provably negligible risk of collisions.")
 346     (license license:bsd-3)))
 347
 348 (define-public oniguruma
 349   (package
 350     (name "oniguruma")
 351     (version "5.9.6")
 352     (source (origin
 353               (method url-fetch)
 354               (uri (string-append "https://github.com/kkos/"
 355                                   "oniguruma/releases/download/v" version
 356                                   "/onig-" version ".tar.gz"))
 357               (sha256
 358                (base32
 359                 "19s79vsclqn170mw0ajwv7j37qsbn4f1yjz3yavnhvva6c820r6m"))))
 360     (build-system gnu-build-system)
 361     (home-page "https://github.com/kkos/oniguruma")
 362     (synopsis "Regular expression library")
 363     (description "Oniguruma is a regular expressions library.  The special
 364 characteristic of this library is that different character encoding for every
 365 regular expression object can be specified.")
 366     (license license:bsd-2)))
 367
 368 (define-public antiword
 369   (package
 370     (name "antiword")
 371     (version "0.37")
 372     (source (origin
 373               (method url-fetch)
 374               (uri (string-append "http://www.winfield.demon.nl/linux"
 375                                   "/antiword-" version ".tar.gz"))
 376               (sha256
 377                (base32
 378                 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
 379               (patches (search-patches "antiword-CVE-2014-8123.patch"))))
 380     (build-system gnu-build-system)
 381     (arguments
 382      `(#:tests? #f ; There are no tests
 383        #:make-flags
 384        (list "-f" "Makefile.Linux"
 385              (string-append "GLOBAL_INSTALL_DIR="
 386                             (assoc-ref %outputs "out") "/bin")
 387              (string-append "GLOBAL_RESOURCES_DIR="
 388                             (assoc-ref %outputs "out") "/share/antiword"))
 389        #:phases
 390        (modify-phases %standard-phases
 391          (delete 'configure)
 392          (replace 'install
 393            (lambda* (#:key make-flags #:allow-other-keys)
 394              (zero? (apply system* "make" `("global_install" ,@make-flags))))))))
 395     (home-page "http://www.winfield.demon.nl/")
 396     (synopsis "Microsoft Word document reader")
 397     (description "Antiword is an application for displaying Microsoft Word
 398 documents.  It can also convert the document to PostScript or XML.  Only
 399 documents made by MS Word version 2 and version 6 or later are supported.  The
 400 name comes from: \"The antidote against people who send Microsoft Word files
 401 to everybody, because they believe that everybody runs Windows and therefore
 402 runs Word\".")
 403     (license license:gpl2+)))
 404
 405 (define-public catdoc
 406   (package
 407     (name "catdoc")
 408     (version "0.95")
 409     (source (origin
 410               (method url-fetch)
 411               (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
 412                                   "catdoc-" version ".tar.gz"))
 413               (patches (search-patches "catdoc-CVE-2017-11110.patch"))
 414               (sha256
 415                (base32
 416                 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
 417     (build-system gnu-build-system)
 418     ;; TODO: Also build `wordview` which requires `tk` – make a separate
 419     ;; package for this.
 420     (arguments
 421      '(#:tests? #f ; There are no tests
 422        #:configure-flags '("--disable-wordview")
 423        #:phases
 424        (modify-phases %standard-phases
 425          (add-before 'install 'fix-install
 426            (lambda* (#:key outputs #:allow-other-keys)
 427              (let ((out (assoc-ref outputs "out")))
 428                (mkdir-p (string-append out "/share/man/man1"))))))))
 429     (home-page "http://www.wagner.pp.ru/~vitus/software/catdoc/")
 430     (synopsis "MS-Word to TeX or plain text converter")
 431     (description "@command{catdoc} extracts text from MS-Word files, trying to
 432 preserve as many special printable characters as possible.  It supports
 433 everything up to Word-97. Also supported are MS Write documents and RTF files.
 434
 435 @command{catdoc} does not preserve complex word formatting, but it can
 436 translate some non-ASCII characters into TeX escape codes.  It's goal is to
 437 extract plain text and allow you to read it and, probably, reformat with TeX,
 438 according to TeXnical rules.
 439
 440 This package also provides @command{xls2csv}, which extracts data from Excel
 441 spreadsheets and outputs it in comma-separated-value format, and
 442 @command{catppt}, which extracts data from PowerPoint presentations.")
 443     (license license:gpl2+)))
 444
 445 (define-public utfcpp
 446   (package
 447     (name "utfcpp")
 448     (version "2.3.4")
 449     (source (origin
 450               (method url-fetch)
 451               (uri
 452                (string-append
 453                 "mirror://sourceforge/utfcpp/utf8cpp_2x/Release%20"
 454                 version "/utf8_v"
 455                 (string-map (lambda (x) (if (eq? x #\.) #\_ x)) version)
 456                 ".zip"))
 457               (file-name (string-append name "-" version ".zip"))
 458               (sha256
 459                (base32
 460                 "1vqhs0aipcvvdrwcs7h3jsryg6mgbmc4s34n5cm6d36q4nxwwwrk"))))
 461     (build-system trivial-build-system)
 462     (arguments
 463      `(#:modules ((guix build utils))
 464        #:builder
 465        (begin
 466          (use-modules (guix build utils))
 467          (let ((source (assoc-ref %build-inputs "source"))
 468                (out    (assoc-ref %outputs "out"))
 469                (unzip  (string-append (assoc-ref %build-inputs "unzip")
 470                                       "/bin/unzip")))
 471            (mkdir-p out)
 472            (with-directory-excursion out
 473              (system* unzip source)
 474              (mkdir-p "share/doc")
 475              (rename-file "doc" "share/doc/utfcpp")
 476              (rename-file "source" "include"))))))
 477     (native-inputs `(("unzip" ,unzip)))
 478     (home-page "https://github.com/nemtrif/utfcpp")
 479     (synopsis "Portable C++ library for handling UTF-8")
 480     (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
 481 in a portable way.")
 482     (license license:boost1.0)))
 483
 484 (define-public dbacl
 485   (package
 486     (name "dbacl")
 487     (version "1.14")
 488     (source
 489      (origin
 490        (method url-fetch)
 491        (uri (string-append "http://www.lbreyer.com/gpl/"
 492                            name "-" version ".tar.gz"))
 493        (sha256
 494         (base32
 495          "0224g6x71hyvy7jikfxmgcwww1r5lvk0jx36cva319cb9nmrbrq7"))))
 496     (build-system gnu-build-system)
 497     (arguments
 498      `(#:make-flags
 499        (list
 500         (string-append "-I" (assoc-ref %build-inputs "slang")
 501                        "/include/slang")
 502         (string-append "-I" (assoc-ref %build-inputs "ncurses")
 503                        "/include/ncurses"))
 504        #:phases
 505        (modify-phases %standard-phases
 506          (add-after 'unpack 'delete-sample6-and-japanese
 507            (lambda _
 508              (substitute* "doc/Makefile.am"
 509                (("sample6.txt") "")
 510                (("japanese.txt") ""))
 511              (delete-file "doc/sample6.txt")
 512              (delete-file "doc/japanese.txt")
 513              (substitute* (list "src/tests/Makefile.am"
 514                                 "src/tests/Makefile.in")
 515                (("dbacl-jap.shin") "")
 516                (("dbacl-jap.sh") ""))
 517              #t))
 518          (add-after 'unpack 'delete-test
 519            ;; See comments about the license.
 520            (lambda _
 521              (delete-file "src/tests/dbacl-jap.shin")))
 522          (add-after 'delete-sample6-and-japanese 'autoreconf
 523            (lambda _
 524              (zero? (system* "autoreconf" "-vif"))))
 525          (add-after 'unpack 'fix-test-files
 526            (lambda* (#:key inputs outputs #:allow-other-keys)
 527              (let* ((out (assoc-ref outputs "out"))
 528                     (bin (string-append out "/bin")))
 529                (substitute* (find-files "src/tests/" "\\.shin$")
 530                  (("PATH=/bin:/usr/bin")
 531                   "#PATH=/bin:/usr/bin")
 532                  (("diff") (string-append (which "diff")))
 533                  (("tr") (string-append (which "tr"))))
 534                #t))))))
 535     (inputs
 536      `(("ncurses" ,ncurses)
 537        ("perl" ,perl)
 538        ("readline" ,readline)
 539        ("slang" ,slang)))
 540     (native-inputs
 541      `(("libtool" ,libtool)
 542        ("autoconf" ,autoconf)
 543        ("automake" ,automake)
 544        ("pkg-config" ,pkg-config)))
 545     (home-page "http://www.lbreyer.com/dbacl.html")
 546     (synopsis "Bayesian text and email classifier")
 547     (description
 548      "dbacl is a fast Bayesian text and email classifier.  It builds a variety
 549 of language models using maximum entropy (minimum divergence) principles, and
 550 these can then be used to categorize input data automatically among multiple
 551 categories.")
 552     ;; The software is licensed as GPLv3 or later, but
 553     ;; includes various sample texts in the doc dir:
 554     ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
 555     ;;   by Mark Twain.
 556     ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
 557     ;; - sample6.txt is a forwarded email, copyright unknown.
 558     ;;   Guix does exclude sample6.txt.
 559     ;; - japanese.txt is a Japanese unoffical translation of the
 560     ;;   GNU General Public License, (c) by the Free Software Foundation.
 561     ;;   Guix excludes this file.
 562     (license (list license:gpl3+ license:public-domain))))
 563
 564 (define-public dotconf
 565   (package
 566     (name "dotconf")
 567     (version "1.3")
 568     (source (origin
 569               (method url-fetch)
 570               (uri (string-append
 571                     "https://github.com/williamh/dotconf/archive/v"
 572                     version ".tar.gz"))
 573               (file-name (string-append name "-" version ".tar.gz"))
 574               (sha256
 575                (base32
 576                 "0lsnh0yaw44psmx59hq94cj1932gscp5h8d3cnh05l0svr0cy7kz"))))
 577     (build-system gnu-build-system)
 578     (arguments
 579      `(#:tests? #f ; FIXME maketest.sh does not work.
 580        #:phases
 581        (modify-phases %standard-phases
 582          (add-after 'unpack 'autoreconf
 583            (lambda _
 584              (zero? (system* "autoreconf" "-vif")))))))
 585     (native-inputs
 586      `(("autoconf" ,autoconf)
 587        ("automake" ,automake)
 588        ("libtool" ,libtool)))
 589     (home-page "https://github.com/williamh/dotconf")
 590     (synopsis "Configuration file parser library")
 591     (description
 592      "C library for creating and parsing configuration files.")
 593     (license (list license:lgpl2.1         ; Main distribution.
 594                    license:asl1.1))))      ; src/readdir.{c,h}
 595
 596 (define-public java-rsyntaxtextarea
 597   (package
 598     (name "java-rsyntaxtextarea")
 599     (version "2.6.1")
 600     (source (origin
 601               (method url-fetch)
 602               (uri (string-append "https://github.com/bobbylight/"
 603                                   "RSyntaxTextArea/archive/"
 604                                   version ".tar.gz"))
 605               (file-name (string-append name "-" version ".tar.gz"))
 606               (sha256
 607                (base32
 608                 "0c5mqg2klj5rvf8fhycrli8rf6s37l9p7a8knw9gpp65r1c120q2"))))
 609     (build-system ant-build-system)
 610     (arguments
 611      `(;; FIXME: some tests fail because locale resources cannot be found.
 612        ;; Even when I add them to the class path,
 613        ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
 614        #:tests? #f
 615        #:jar-name "rsyntaxtextarea.jar"))
 616     (native-inputs
 617      `(("java-junit" ,java-junit)
 618        ("java-hamcrest-core" ,java-hamcrest-core)))
 619     (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
 620     (synopsis "Syntax highlighting text component for Java Swing")
 621     (description "RSyntaxTextArea is a syntax highlighting, code folding text
 622 component for Java Swing.  It extends @code{JTextComponent} so it integrates
 623 completely with the standard @code{javax.swing.text} package.  It is fast and
 624 efficient, and can be used in any application that needs to edit or view
 625 source code.")
 626     (license license:bsd-3)))