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        (modify-phases %standard-phases
 101          (add-before '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              #t)))))
 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        (modify-phases %standard-phases
 212          (add-after 'unpack 'autoreconf
 213            (lambda _ (zero? (system* "autoreconf" "-vif")))))))
 214     (native-inputs
 215      `(("autoconf" ,autoconf)
 216        ("automake" ,automake)
 217        ("libtool" ,libtool)))
 218     (home-page "https://github.com/agordon/libgtextutils")
 219     (synopsis "Gordon's text utils library")
 220     (description
 221      "libgtextutils is a text utilities library used by the fastx toolkit from
 222 the Hannon Lab.")
 223     (license license:agpl3+)))
 224
 225 (define-public cityhash
 226   (let ((commit "8af9b8c"))
 227     (package
 228       (name "cityhash")
 229       (version (string-append "1.1-2." commit))
 230       (source (origin
 231                 (method git-fetch)
 232                 (uri (git-reference
 233                       (url "https://github.com/google/cityhash.git")
 234                       (commit commit)))
 235                 (file-name (string-append name "-" version ".tar.gz"))
 236                 (sha256
 237                  (base32
 238                   "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
 239       (build-system gnu-build-system)
 240       (arguments
 241        '(#:make-flags (list "CXXFLAGS=-g -O3")
 242          #:phases
 243          (modify-phases %standard-phases
 244            ;; citycrc is not installed by default but is used by some
 245            ;; programs.
 246            (add-after 'install 'install-citycrc
 247              (lambda* (#:key outputs #:allow-other-keys)
 248                (let* ((out (assoc-ref outputs "out"))
 249                       (include (string-append out "/include")))
 250                  (install-file "src/citycrc.h" include))
 251                #t)))))
 252       (home-page "https://github.com/google/cityhash")
 253       (synopsis "C++ hash functions for strings")
 254       (description
 255        "CityHash provides hash functions for strings.  The functions mix the
 256 input bits thoroughly but are not suitable for cryptography.")
 257       (license license:expat))))
 258
 259 (define-public ustr
 260   (package
 261     (name "ustr")
 262     (version "1.0.4")
 263     (source (origin
 264               (method url-fetch)
 265               (uri (string-append "http://www.and.org/ustr/" version
 266                                   "/ustr-" version ".tar.bz2"))
 267               (sha256
 268                (base32
 269                 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
 270               (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
 271     (build-system gnu-build-system)
 272     (arguments
 273      `(#:make-flags
 274        (list "CC=gcc"
 275              "HIDE="
 276              ;; Override "/sbin/ldconfig" with "echo" because we don't need
 277              ;; "ldconfig".
 278              "LDCONFIG=echo"
 279              (string-append "prefix=" (assoc-ref %outputs "out"))
 280              "all-shared")
 281        #:phases
 282        (modify-phases %standard-phases
 283          (add-after 'unpack 'disable-check-for-stdint
 284            (lambda _
 285              ;; Of course we have stdint.h, just not in /usr/include
 286              (substitute* '("Makefile"
 287                             "ustr-import.in")
 288                (("-f \"/usr/include/stdint.h\"") "-z \"\""))
 289              #t))
 290          ;; No configure script
 291          (delete 'configure))))
 292     (home-page "http://www.and.org/ustr/")
 293     (synopsis "String library with very low memory overhead")
 294     (description
 295      "Ustr is a string library for C with very low memory overhead.")
 296     ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
 297     ;; LGPL, etc. ... if you need another license to help compatibility, just
 298     ;; ask for it.  It's basically public domain, without all the legal
 299     ;; problems for everyone that trying to make something public domain
 300     ;; entails."
 301     (license license:public-domain)))
 302
 303 (define-public libconfig
 304   (package
 305     (name "libconfig")
 306     (version "1.5")
 307     (source (origin
 308               (method url-fetch)
 309               (uri (string-append "http://www.hyperrealm.com/libconfig/"
 310                                   "libconfig-" version ".tar.gz"))
 311               (sha256
 312                (base32
 313                 "1xh3hzk63v4y8815lc5209m3s6ms2cpgw4h5hg462i4f1lwsl7g3"))))
 314     (build-system gnu-build-system)
 315     (home-page "http://www.hyperrealm.com/libconfig/")
 316     (synopsis "C/C++ configuration file library")
 317     (description
 318      "Libconfig is a simple library for manipulating structured configuration
 319 files.  This file format is more compact and more readable than XML.  And
 320 unlike XML, it is type-aware, so it is not necessary to do string parsing in
 321 application code.")
 322     (license license:lgpl2.1+)))
 323
 324 (define-public pfff
 325   (package
 326     (name "pfff")
 327     (version "1.0")
 328     (source (origin
 329               (method url-fetch)
 330               (uri (string-append "https://github.com/pfff/pfff/archive/v"
 331                                   version ".tar.gz"))
 332               (file-name (string-append name "-" version ".tar.gz"))
 333               (sha256
 334                (base32
 335                 "00m553aa277iarxj6dalmklyb64r7ias49bfwzbacsfg8h3kar8m"))))
 336     (build-system cmake-build-system)
 337     (home-page "http://biit.cs.ut.ee/pfff/")
 338     (synopsis "Probabilistic fast file fingerprinting tool")
 339     (description
 340      "pfff is a tool for calculating a compact digital fingerprint of a file
 341 by sampling randomly from the file instead of reading it in full.
 342 Consequently, the computation has a flat performance characteristic,
 343 correlated with data variation rather than file size.  pfff can be as reliable
 344 as existing hashing techniques, with provably negligible risk of collisions.")
 345     (license license:bsd-3)))
 346
 347 (define-public oniguruma
 348   (package
 349     (name "oniguruma")
 350     (version "5.9.6")
 351     (source (origin
 352               (method url-fetch)
 353               (uri (string-append "https://github.com/kkos/"
 354                                   "oniguruma/releases/download/v" version
 355                                   "/onig-" version ".tar.gz"))
 356               (sha256
 357                (base32
 358                 "19s79vsclqn170mw0ajwv7j37qsbn4f1yjz3yavnhvva6c820r6m"))))
 359     (build-system gnu-build-system)
 360     (home-page "https://github.com/kkos/oniguruma")
 361     (synopsis "Regular expression library")
 362     (description "Oniguruma is a regular expressions library.  The special
 363 characteristic of this library is that different character encoding for every
 364 regular expression object can be specified.")
 365     (license license:bsd-2)))
 366
 367 (define-public antiword
 368   (package
 369     (name "antiword")
 370     (version "0.37")
 371     (source (origin
 372               (method url-fetch)
 373               (uri (string-append "http://www.winfield.demon.nl/linux"
 374                                   "/antiword-" version ".tar.gz"))
 375               (sha256
 376                (base32
 377                 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
 378               (patches (search-patches "antiword-CVE-2014-8123.patch"))))
 379     (build-system gnu-build-system)
 380     (arguments
 381      `(#:tests? #f ; There are no tests
 382        #:make-flags
 383        (list "-f" "Makefile.Linux"
 384              (string-append "GLOBAL_INSTALL_DIR="
 385                             (assoc-ref %outputs "out") "/bin")
 386              (string-append "GLOBAL_RESOURCES_DIR="
 387                             (assoc-ref %outputs "out") "/share/antiword"))
 388        #:phases
 389        (modify-phases %standard-phases
 390          (delete 'configure)
 391          (replace 'install
 392            (lambda* (#:key make-flags #:allow-other-keys)
 393              (zero? (apply system* "make" `("global_install" ,@make-flags))))))))
 394     (home-page "http://www.winfield.demon.nl/")
 395     (synopsis "Microsoft Word document reader")
 396     (description "Antiword is an application for displaying Microsoft Word
 397 documents.  It can also convert the document to PostScript or XML.  Only
 398 documents made by MS Word version 2 and version 6 or later are supported.  The
 399 name comes from: \"The antidote against people who send Microsoft Word files
 400 to everybody, because they believe that everybody runs Windows and therefore
 401 runs Word\".")
 402     (license license:gpl2+)))
 403
 404 (define-public catdoc
 405   (package
 406     (name "catdoc")
 407     (version "0.95")
 408     (source (origin
 409               (method url-fetch)
 410               (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
 411                                   "catdoc-" version ".tar.gz"))
 412               (patches (search-patches "catdoc-CVE-2017-11110.patch"))
 413               (sha256
 414                (base32
 415                 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
 416     (build-system gnu-build-system)
 417     ;; TODO: Also build `wordview` which requires `tk` – make a separate
 418     ;; package for this.
 419     (arguments
 420      '(#:tests? #f ; There are no tests
 421        #:configure-flags '("--disable-wordview")
 422        #:phases
 423        (modify-phases %standard-phases
 424          (add-before 'install 'fix-install
 425            (lambda* (#:key outputs #:allow-other-keys)
 426              (let ((out (assoc-ref outputs "out")))
 427                (mkdir-p (string-append out "/share/man/man1"))))))))
 428     (home-page "http://www.wagner.pp.ru/~vitus/software/catdoc/")
 429     (synopsis "MS-Word to TeX or plain text converter")
 430     (description "@command{catdoc} extracts text from MS-Word files, trying to
 431 preserve as many special printable characters as possible.  It supports
 432 everything up to Word-97. Also supported are MS Write documents and RTF files.
 433
 434 @command{catdoc} does not preserve complex word formatting, but it can
 435 translate some non-ASCII characters into TeX escape codes.  It's goal is to
 436 extract plain text and allow you to read it and, probably, reformat with TeX,
 437 according to TeXnical rules.
 438
 439 This package also provides @command{xls2csv}, which extracts data from Excel
 440 spreadsheets and outputs it in comma-separated-value format, and
 441 @command{catppt}, which extracts data from PowerPoint presentations.")
 442     (license license:gpl2+)))
 443
 444 (define-public utfcpp
 445   (package
 446     (name "utfcpp")
 447     (version "2.3.4")
 448     (source (origin
 449               (method url-fetch)
 450               (uri
 451                (string-append
 452                 "mirror://sourceforge/utfcpp/utf8cpp_2x/Release%20"
 453                 version "/utf8_v"
 454                 (string-map (lambda (x) (if (eq? x #\.) #\_ x)) version)
 455                 ".zip"))
 456               (file-name (string-append name "-" version ".zip"))
 457               (sha256
 458                (base32
 459                 "1vqhs0aipcvvdrwcs7h3jsryg6mgbmc4s34n5cm6d36q4nxwwwrk"))))
 460     (build-system trivial-build-system)
 461     (arguments
 462      `(#:modules ((guix build utils))
 463        #:builder
 464        (begin
 465          (use-modules (guix build utils))
 466          (let ((source (assoc-ref %build-inputs "source"))
 467                (out    (assoc-ref %outputs "out"))
 468                (unzip  (string-append (assoc-ref %build-inputs "unzip")
 469                                       "/bin/unzip")))
 470            (mkdir-p out)
 471            (with-directory-excursion out
 472              (system* unzip source)
 473              (mkdir-p "share/doc")
 474              (rename-file "doc" "share/doc/utfcpp")
 475              (rename-file "source" "include"))))))
 476     (native-inputs `(("unzip" ,unzip)))
 477     (home-page "https://github.com/nemtrif/utfcpp")
 478     (synopsis "Portable C++ library for handling UTF-8")
 479     (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
 480 in a portable way.")
 481     (license license:boost1.0)))
 482
 483 (define-public dbacl
 484   (package
 485     (name "dbacl")
 486     (version "1.14")
 487     (source
 488      (origin
 489        (method url-fetch)
 490        (uri (string-append "http://www.lbreyer.com/gpl/"
 491                            name "-" version ".tar.gz"))
 492        (sha256
 493         (base32
 494          "0224g6x71hyvy7jikfxmgcwww1r5lvk0jx36cva319cb9nmrbrq7"))))
 495     (build-system gnu-build-system)
 496     (arguments
 497      `(#:make-flags
 498        (list
 499         (string-append "-I" (assoc-ref %build-inputs "slang")
 500                        "/include/slang")
 501         (string-append "-I" (assoc-ref %build-inputs "ncurses")
 502                        "/include/ncurses"))
 503        #:phases
 504        (modify-phases %standard-phases
 505          (add-after 'unpack 'delete-sample6-and-japanese
 506            (lambda _
 507              (substitute* "doc/Makefile.am"
 508                (("sample6.txt") "")
 509                (("japanese.txt") ""))
 510              (delete-file "doc/sample6.txt")
 511              (delete-file "doc/japanese.txt")
 512              (substitute* (list "src/tests/Makefile.am"
 513                                 "src/tests/Makefile.in")
 514                (("dbacl-jap.shin") "")
 515                (("dbacl-jap.sh") ""))
 516              #t))
 517          (add-after 'unpack 'delete-test
 518            ;; See comments about the license.
 519            (lambda _
 520              (delete-file "src/tests/dbacl-jap.shin")))
 521          (add-after 'delete-sample6-and-japanese 'autoreconf
 522            (lambda _
 523              (zero? (system* "autoreconf" "-vif"))))
 524          (add-after 'unpack 'fix-test-files
 525            (lambda* (#:key inputs outputs #:allow-other-keys)
 526              (let* ((out (assoc-ref outputs "out"))
 527                     (bin (string-append out "/bin")))
 528                (substitute* (find-files "src/tests/" "\\.shin$")
 529                  (("PATH=/bin:/usr/bin")
 530                   "#PATH=/bin:/usr/bin")
 531                  (("diff") (string-append (which "diff")))
 532                  (("tr") (string-append (which "tr"))))
 533                #t))))))
 534     (inputs
 535      `(("ncurses" ,ncurses)
 536        ("perl" ,perl)
 537        ("readline" ,readline)
 538        ("slang" ,slang)))
 539     (native-inputs
 540      `(("libtool" ,libtool)
 541        ("autoconf" ,autoconf)
 542        ("automake" ,automake)
 543        ("pkg-config" ,pkg-config)))
 544     (home-page "http://www.lbreyer.com/dbacl.html")
 545     (synopsis "Bayesian text and email classifier")
 546     (description
 547      "dbacl is a fast Bayesian text and email classifier.  It builds a variety
 548 of language models using maximum entropy (minimum divergence) principles, and
 549 these can then be used to categorize input data automatically among multiple
 550 categories.")
 551     ;; The software is licensed as GPLv3 or later, but
 552     ;; includes various sample texts in the doc dir:
 553     ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
 554     ;;   by Mark Twain.
 555     ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
 556     ;; - sample6.txt is a forwarded email, copyright unknown.
 557     ;;   Guix does exclude sample6.txt.
 558     ;; - japanese.txt is a Japanese unoffical translation of the
 559     ;;   GNU General Public License, (c) by the Free Software Foundation.
 560     ;;   Guix excludes this file.
 561     (license (list license:gpl3+ license:public-domain))))
 562
 563 (define-public dotconf
 564   (package
 565     (name "dotconf")
 566     (version "1.3")
 567     (source (origin
 568               (method url-fetch)
 569               (uri (string-append
 570                     "https://github.com/williamh/dotconf/archive/v"
 571                     version ".tar.gz"))
 572               (file-name (string-append name "-" version ".tar.gz"))
 573               (sha256
 574                (base32
 575                 "0lsnh0yaw44psmx59hq94cj1932gscp5h8d3cnh05l0svr0cy7kz"))))
 576     (build-system gnu-build-system)
 577     (arguments
 578      `(#:tests? #f ; FIXME maketest.sh does not work.
 579        #:phases
 580        (modify-phases %standard-phases
 581          (add-after 'unpack 'autoreconf
 582            (lambda _
 583              (zero? (system* "autoreconf" "-vif")))))))
 584     (native-inputs
 585      `(("autoconf" ,autoconf)
 586        ("automake" ,automake)
 587        ("libtool" ,libtool)))
 588     (home-page "https://github.com/williamh/dotconf")
 589     (synopsis "Configuration file parser library")
 590     (description
 591      "C library for creating and parsing configuration files.")
 592     (license (list license:lgpl2.1         ; Main distribution.
 593                    license:asl1.1))))      ; src/readdir.{c,h}
 594
 595 (define-public java-rsyntaxtextarea
 596   (package
 597     (name "java-rsyntaxtextarea")
 598     (version "2.6.1")
 599     (source (origin
 600               (method url-fetch)
 601               (uri (string-append "https://github.com/bobbylight/"
 602                                   "RSyntaxTextArea/archive/"
 603                                   version ".tar.gz"))
 604               (file-name (string-append name "-" version ".tar.gz"))
 605               (sha256
 606                (base32
 607                 "0c5mqg2klj5rvf8fhycrli8rf6s37l9p7a8knw9gpp65r1c120q2"))))
 608     (build-system ant-build-system)
 609     (arguments
 610      `(;; FIXME: some tests fail because locale resources cannot be found.
 611        ;; Even when I add them to the class path,
 612        ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
 613        #:tests? #f
 614        #:jar-name "rsyntaxtextarea.jar"))
 615     (native-inputs
 616      `(("java-junit" ,java-junit)
 617        ("java-hamcrest-core" ,java-hamcrest-core)))
 618     (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
 619     (synopsis "Syntax highlighting text component for Java Swing")
 620     (description "RSyntaxTextArea is a syntax highlighting, code folding text
 621 component for Java Swing.  It extends @code{JTextComponent} so it integrates
 622 completely with the standard @code{javax.swing.text} package.  It is fast and
 623 efficient, and can be used in any application that needs to edit or view
 624 source code.")
 625     (license license:bsd-3)))