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, 2018, 2019 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 ng0 <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 ;;; Copyright © 2018 Pierre Neidhardt <mail@ambrevar.xyz>
  18 ;;; Copyright © 2018 Meiyo Peng <meiyo.peng@gmail.com>
  19 ;;; Copyright © 2019 Yoshinori Arai <kumagusu08@gmail.com>
  20 ;;;
  21 ;;; This file is part of GNU Guix.
  22 ;;;
  23 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  24 ;;; under the terms of the GNU General Public License as published by
  25 ;;; the Free Software Foundation; either version 3 of the License, or (at
  26 ;;; your option) any later version.
  27 ;;;
  28 ;;; GNU Guix is distributed in the hope that it will be useful, but
  29 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  30 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  31 ;;; GNU General Public License for more details.
  32 ;;;
  33 ;;; You should have received a copy of the GNU General Public License
  34 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  35
  36 (define-module (gnu packages textutils)
  37   #:use-module ((guix licenses) #:prefix license:)
  38   #:use-module (guix packages)
  39   #:use-module (guix download)
  40   #:use-module (guix git-download)
  41   #:use-module (guix build-system ant)
  42   #:use-module (guix build-system gnu)
  43   #:use-module (guix build-system go)
  44   #:use-module (guix build-system cmake)
  45   #:use-module (guix build-system python)
  46   #:use-module (gnu packages)
  47   #:use-module (gnu packages autotools)
  48   #:use-module (gnu packages compression)
  49   #:use-module (gnu packages gettext)
  50   #:use-module (gnu packages java)
  51   #:use-module (gnu packages ncurses)
  52   #:use-module (gnu packages perl)
  53   #:use-module (gnu packages pkg-config)
  54   #:use-module (gnu packages python)
  55   #:use-module (gnu packages python-xyz)
  56   #:use-module (gnu packages readline)
  57   #:use-module (gnu packages slang)
  58   #:use-module (gnu packages web))
  59
  60 (define-public dos2unix
  61   (package
  62     (name "dos2unix")
  63     (version "7.4.0")
  64     (source
  65      (origin
  66        (method url-fetch)
  67        (uri (string-append "https://waterlan.home.xs4all.nl/" name "/"
  68                            name "-" version ".tar.gz"))
  69        (sha256
  70         (base32 "12h4c61g376bhq03y5g2xszkrkrj5hwd928rly3xsp6rvfmnbixs"))))
  71     (build-system gnu-build-system)
  72     (arguments
  73      '(#:make-flags (list "CC=gcc"
  74                           (string-append "prefix=" (assoc-ref %outputs "out")))
  75        #:phases
  76        (modify-phases %standard-phases
  77          (delete 'configure)))) ; no configure script
  78     (native-inputs
  79      `(("gettext" ,gettext-minimal)
  80        ("perl" ,perl)))
  81     (home-page "https://waterlan.home.xs4all.nl/dos2unix.html")
  82     (synopsis "DOS/Mac to Unix and vice versa text file format converter")
  83     (description
  84      "dos2unix is a tool to convert line breaks in a text file from Unix format
  85 to DOS format and vice versa.")
  86     (license license:bsd-2)))
  87
  88 (define-public recode
  89   (package
  90     (name "recode")
  91     (version "3.7.1")
  92     (source
  93      (origin
  94        (method url-fetch)
  95        (uri (string-append "https://github.com/rrthomas/recode/releases/"
  96                            "download/v" version "/" name "-" version ".tar.gz"))
  97        (sha256
  98         (base32
  99          "0215hfj0rhlh0grg91qfx75pp6z09bpv8211qdxqihniw7y9a4fs"))
 100        (modules '((guix build utils)))
 101        (snippet '(begin
 102                    (delete-file "tests/Recode.c")
 103                    #t))))
 104     (build-system gnu-build-system)
 105     (native-inputs
 106      `(("python" ,python-2)
 107        ("python2-cython" ,python2-cython)))
 108     (home-page "https://github.com/rrthomas/recode")
 109     (synopsis "Text encoding converter")
 110     (description "The Recode library converts files between character sets and
 111 usages.  It recognises or produces over 200 different character sets (or about
 112 300 if combined with an iconv library) and transliterates files between almost
 113 any pair.  When exact transliteration are not possible, it gets rid of
 114 offending characters or falls back on approximations.  The recode program is a
 115 handy front-end to the library.")
 116     (license license:gpl3+)))
 117
 118 (define-public enca
 119   (package
 120     (name "enca")
 121     (version "1.19")
 122     (source
 123      (origin
 124        (method url-fetch)
 125        (uri (string-append
 126              "https://github.com/nijel/enca/archive/" version ".tar.gz"))
 127        (sha256
 128         (base32 "099z526i7qgij7q1w3lvhl88iv3jc3nqxca2i09h6s08ghyrmzf4"))
 129        (file-name (string-append name "-" version ".tar.gz"))))
 130     (build-system gnu-build-system)
 131     ;; enca-1.19 tests fail with recent recode.
 132     ;(inputs `(("recode" ,recode)))
 133     (home-page "https://github.com/nijel/enca")
 134     (synopsis "Text encoding detection tool")
 135     (description "Enca (Extremely Naive Charset Analyser) consists of libenca,
 136 an encoding detection library, and enca, a command line frontend, integrating
 137 libenca and several charset conversion libraries and tools.")
 138     (license license:gpl2)))
 139
 140 (define-public utf8proc
 141   (package
 142     (name "utf8proc")
 143     (version "2.1.1")
 144     (source
 145      (origin
 146        (method url-fetch)
 147        (uri (string-append
 148              "https://github.com/JuliaLang/utf8proc/archive/v"
 149              version ".tar.gz"))
 150        (file-name (string-append name "-" version ".tar.gz"))
 151        (sha256
 152         (base32 "1cnpigrazhslw65s4j1a56j7p6d7d61wsxxjf1218i9mkwv2yw17"))))
 153     (build-system gnu-build-system)
 154     (inputs                 ; test data that is otherwise downloaded with curl
 155      `(("NormalizationTest.txt"
 156         ,(origin
 157            (method url-fetch)
 158            (uri (string-append "http://www.unicode.org/Public/9.0.0/ucd/"
 159                                "NormalizationTest.txt"))
 160            (sha256
 161             (base32 "1fxrz0bilsbwl685336aqi88k62i6nqhm62rvy4zhg3bcm4dhj1d"))))
 162        ("GraphemeBreakTest.txt"
 163         ,(origin
 164            (method url-fetch)
 165            (uri (string-append "http://www.unicode.org/Public/9.0.0/ucd/"
 166                                "auxiliary/GraphemeBreakTest.txt"))
 167            (sha256
 168             (base32 "0qbhyhmf0778lc2hcwlpizrvmdxwpk959v2q2wb8abv09ba7wvn7"))))))
 169     (arguments
 170      '(#:make-flags (list "CC=gcc"
 171                           (string-append "prefix=" (assoc-ref %outputs "out")))
 172        #:phases
 173        (modify-phases %standard-phases
 174          (delete 'configure)
 175          (add-before 'check 'check-data
 176            (lambda* (#:key inputs #:allow-other-keys)
 177              (for-each (lambda (i)
 178                          (copy-file (assoc-ref inputs i)
 179                                     (string-append "data/" i)))
 180                        '("NormalizationTest.txt" "GraphemeBreakTest.txt"))
 181              (substitute* "data/GraphemeBreakTest.txt"
 182                (("÷") "/")
 183                (("×") "+")))))))
 184     (home-page "https://julialang.org/utf8proc/")
 185     (synopsis "C library for processing UTF-8 Unicode data")
 186     (description "utf8proc is a small C library that provides Unicode
 187 normalization, case-folding, and other operations for data in the UTF-8
 188 encoding, supporting Unicode version 9.0.0.")
 189     (license license:expat)))
 190
 191 (define-public libconfuse
 192   (package
 193     (name "libconfuse")
 194     (version "3.2.2")
 195     (source
 196      (origin
 197        (method url-fetch)
 198        (uri (string-append "https://github.com/martinh/libconfuse/"
 199                            "releases/download/v" version
 200                            "/confuse-" version ".tar.xz"))
 201        (sha256
 202         (base32 "02r1mmzik2m0iigbc2da3y754vj24i18r3ml5p2wzs027mjhn959"))))
 203     (build-system gnu-build-system)
 204     (home-page "https://github.com/martinh/libconfuse")
 205     (synopsis "Configuration file parser library")
 206     (description "libconfuse is a configuration file parser library.  It
 207 supports sections and (lists of) values (strings, integers, floats, booleans
 208 or other sections), as well as some other features (such as
 209 single/double-quoted strings, environment variable expansion, functions and
 210 nested include statements).")
 211     (license license:isc)))
 212
 213 (define-public libgtextutils
 214   (package
 215     (name "libgtextutils")
 216     (version "0.7")
 217     (source
 218      (origin
 219        (method url-fetch)
 220        (uri (string-append
 221              "https://github.com/agordon/libgtextutils/releases/download/"
 222              version "/libgtextutils-" version ".tar.gz"))
 223        (sha256
 224         (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
 225     (build-system gnu-build-system)
 226     (arguments
 227      '(#:phases
 228        (modify-phases %standard-phases
 229          (replace 'bootstrap
 230            (lambda _ (invoke "sh" "reconf"))))))
 231     (native-inputs
 232      `(("autoconf" ,autoconf)
 233        ("automake" ,automake)
 234        ("libtool" ,libtool)))
 235     (home-page "https://github.com/agordon/libgtextutils")
 236     (synopsis "Gordon's text utils library")
 237     (description
 238      "libgtextutils is a text utilities library used by the fastx toolkit from
 239 the Hannon Lab.")
 240     (license license:agpl3+)))
 241
 242 (define-public cityhash
 243   (let ((commit "8af9b8c"))
 244     (package
 245       (name "cityhash")
 246       (version (string-append "1.1-2." commit))
 247       (source (origin
 248                 (method git-fetch)
 249                 (uri (git-reference
 250                       (url "https://github.com/google/cityhash.git")
 251                       (commit commit)))
 252                 (file-name (string-append name "-" version ".tar.gz"))
 253                 (sha256
 254                  (base32
 255                   "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
 256       (build-system gnu-build-system)
 257       (arguments
 258        '(#:make-flags (list "CXXFLAGS=-g -O3")
 259          #:phases
 260          (modify-phases %standard-phases
 261            ;; citycrc is not installed by default but is used by some
 262            ;; programs.
 263            (add-after 'install 'install-citycrc
 264              (lambda* (#:key outputs #:allow-other-keys)
 265                (let* ((out (assoc-ref outputs "out"))
 266                       (include (string-append out "/include")))
 267                  (install-file "src/citycrc.h" include))
 268                #t)))))
 269       (home-page "https://github.com/google/cityhash")
 270       (synopsis "C++ hash functions for strings")
 271       (description
 272        "CityHash provides hash functions for strings.  The functions mix the
 273 input bits thoroughly but are not suitable for cryptography.")
 274       (license license:expat))))
 275
 276 (define-public ustr
 277   (package
 278     (name "ustr")
 279     (version "1.0.4")
 280     (source (origin
 281               (method url-fetch)
 282               (uri (string-append "http://www.and.org/ustr/" version
 283                                   "/ustr-" version ".tar.bz2"))
 284               (sha256
 285                (base32
 286                 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
 287               (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
 288     (build-system gnu-build-system)
 289     (arguments
 290      `(#:make-flags
 291        (list "CC=gcc"
 292              "HIDE="
 293              ;; Override "/sbin/ldconfig" with "echo" because we don't need
 294              ;; "ldconfig".
 295              "LDCONFIG=echo"
 296              (string-append "prefix=" (assoc-ref %outputs "out"))
 297              "all-shared")
 298        #:phases
 299        (modify-phases %standard-phases
 300          (add-after 'unpack 'disable-check-for-stdint
 301            (lambda _
 302              ;; Of course we have stdint.h, just not in /usr/include
 303              (substitute* '("Makefile"
 304                             "ustr-import.in")
 305                (("-f \"/usr/include/stdint.h\"") "-z \"\""))
 306              #t))
 307          ;; No configure script
 308          (delete 'configure))))
 309     (home-page "http://www.and.org/ustr/")
 310     (synopsis "String library with very low memory overhead")
 311     (description
 312      "Ustr is a string library for C with very low memory overhead.")
 313     ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
 314     ;; LGPL, etc. ... if you need another license to help compatibility, just
 315     ;; ask for it.  It's basically public domain, without all the legal
 316     ;; problems for everyone that trying to make something public domain
 317     ;; entails."
 318     (license license:public-domain)))
 319
 320 (define-public libconfig
 321   (package
 322     (name "libconfig")
 323     (version "1.7.2")
 324     (home-page "https://hyperrealm.github.io/libconfig/")
 325     (source (origin
 326               (method url-fetch)
 327               (uri (string-append home-page "/dist/libconfig-"
 328                                   version ".tar.gz"))
 329               (sha256
 330                (base32
 331                 "1ngs2qx3cx5cbwinc5mvadly0b5n7s86zsc68c404czzfff7lg3w"))))
 332     (build-system gnu-build-system)
 333     (synopsis "C/C++ configuration file library")
 334     (description
 335      "Libconfig is a simple library for manipulating structured configuration
 336 files.  This file format is more compact and more readable than XML.  And
 337 unlike XML, it is type-aware, so it is not necessary to do string parsing in
 338 application code.")
 339     (license license:lgpl2.1+)))
 340
 341 (define-public pfff
 342   (package
 343     (name "pfff")
 344     (version "1.0")
 345     (source (origin
 346               (method url-fetch)
 347               (uri (string-append "https://github.com/pfff/pfff/archive/v"
 348                                   version ".tar.gz"))
 349               (file-name (string-append name "-" version ".tar.gz"))
 350               (sha256
 351                (base32
 352                 "00m553aa277iarxj6dalmklyb64r7ias49bfwzbacsfg8h3kar8m"))))
 353     (build-system cmake-build-system)
 354     (home-page "http://biit.cs.ut.ee/pfff/")
 355     (synopsis "Probabilistic fast file fingerprinting tool")
 356     (description
 357      "pfff is a tool for calculating a compact digital fingerprint of a file
 358 by sampling randomly from the file instead of reading it in full.
 359 Consequently, the computation has a flat performance characteristic,
 360 correlated with data variation rather than file size.  pfff can be as reliable
 361 as existing hashing techniques, with provably negligible risk of collisions.")
 362     (license license:bsd-3)))
 363
 364 (define-public oniguruma
 365   (package
 366     (name "oniguruma")
 367     (version "6.9.2")
 368     (source (origin
 369               (method url-fetch)
 370               (uri (string-append "https://github.com/kkos/"
 371                                   "oniguruma/releases/download/v" version
 372                                   "/onig-" version ".tar.gz"))
 373               (sha256
 374                (base32
 375                 "0slp4hpw9qxk4xhn7abyw7065sd355xwkyfq72glxczcjsqxsynv"))))
 376     (build-system gnu-build-system)
 377     (home-page "https://github.com/kkos/oniguruma")
 378     (synopsis "Regular expression library")
 379     (description "Oniguruma is a regular expressions library.  The special
 380 characteristic of this library is that different character encoding for every
 381 regular expression object can be specified.")
 382     (license license:bsd-2)))
 383
 384 ;; PHP < 7.3.0 requires this old version.  Remove once no longer needed.
 385 (define-public oniguruma-5
 386   (package
 387     (inherit oniguruma)
 388     (version "5.9.6")
 389     (source (origin
 390               (method url-fetch)
 391               (uri (string-append "https://github.com/kkos/"
 392                                   "oniguruma/releases/download/v" version
 393                                   "/onig-" version ".tar.gz"))
 394               (sha256
 395                (base32
 396                 "19s79vsclqn170mw0ajwv7j37qsbn4f1yjz3yavnhvva6c820r6m"))))))
 397
 398 (define-public antiword
 399   (package
 400     (name "antiword")
 401     (version "0.37")
 402     (source (origin
 403               (method url-fetch)
 404               (uri (string-append "http://www.winfield.demon.nl/linux"
 405                                   "/antiword-" version ".tar.gz"))
 406               (sha256
 407                (base32
 408                 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
 409               (patches (search-patches "antiword-CVE-2014-8123.patch"))))
 410     (build-system gnu-build-system)
 411     (arguments
 412      `(#:tests? #f ; There are no tests
 413        #:make-flags
 414        (list "-f" "Makefile.Linux"
 415              (string-append "GLOBAL_INSTALL_DIR="
 416                             (assoc-ref %outputs "out") "/bin")
 417              (string-append "GLOBAL_RESOURCES_DIR="
 418                             (assoc-ref %outputs "out") "/share/antiword"))
 419        #:phases
 420        (modify-phases %standard-phases
 421          (replace 'configure
 422            (lambda* (#:key outputs #:allow-other-keys)
 423              ;; Ensure that mapping files can be found in the actual package
 424              ;; data directory.
 425              (substitute* "antiword.h"
 426                (("/usr/share/antiword")
 427                 (string-append (assoc-ref outputs "out") "/share/antiword")))
 428              #t))
 429          (replace 'install
 430            (lambda* (#:key make-flags #:allow-other-keys)
 431              (apply invoke "make" `("global_install" ,@make-flags)))))))
 432     (home-page "http://www.winfield.demon.nl/")
 433     (synopsis "Microsoft Word document reader")
 434     (description "Antiword is an application for displaying Microsoft Word
 435 documents.  It can also convert the document to PostScript or XML.  Only
 436 documents made by MS Word version 2 and version 6 or later are supported.  The
 437 name comes from: \"The antidote against people who send Microsoft Word files
 438 to everybody, because they believe that everybody runs Windows and therefore
 439 runs Word\".")
 440     (license license:gpl2+)))
 441
 442 (define-public catdoc
 443   (package
 444     (name "catdoc")
 445     (version "0.95")
 446     (source (origin
 447               (method url-fetch)
 448               (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
 449                                   "catdoc-" version ".tar.gz"))
 450               (patches (search-patches "catdoc-CVE-2017-11110.patch"))
 451               (sha256
 452                (base32
 453                 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
 454     (build-system gnu-build-system)
 455     ;; TODO: Also build `wordview` which requires `tk` – make a separate
 456     ;; package for this.
 457     (arguments
 458      '(#:tests? #f ; There are no tests
 459        #:configure-flags '("--disable-wordview")
 460        #:phases
 461        (modify-phases %standard-phases
 462          (add-before 'install 'fix-install
 463            (lambda* (#:key outputs #:allow-other-keys)
 464              (let ((out (assoc-ref outputs "out")))
 465                (mkdir-p (string-append out "/share/man/man1"))))))))
 466     (home-page "http://www.wagner.pp.ru/~vitus/software/catdoc/")
 467     (synopsis "MS-Word to TeX or plain text converter")
 468     (description "@command{catdoc} extracts text from MS-Word files, trying to
 469 preserve as many special printable characters as possible.  It supports
 470 everything up to Word-97. Also supported are MS Write documents and RTF files.
 471
 472 @command{catdoc} does not preserve complex word formatting, but it can
 473 translate some non-ASCII characters into TeX escape codes.  It's goal is to
 474 extract plain text and allow you to read it and, probably, reformat with TeX,
 475 according to TeXnical rules.
 476
 477 This package also provides @command{xls2csv}, which extracts data from Excel
 478 spreadsheets and outputs it in comma-separated-value format, and
 479 @command{catppt}, which extracts data from PowerPoint presentations.")
 480     (license license:gpl2+)))
 481
 482 (define-public utfcpp
 483   (package
 484     (name "utfcpp")
 485     (version "2.3.5")
 486     (source (origin
 487               (method url-fetch)
 488               (uri
 489                (string-append "https://github.com/nemtrif/utfcpp/archive/v"
 490                               version ".tar.gz"))
 491               (file-name (string-append name "-" version ".tar.gz"))
 492               (sha256
 493                (base32
 494                 "0gcqcfw19kfim8xw29xdp91l310yfjyrqdj2zsx8xx02dkpy1zzk"))))
 495     (build-system cmake-build-system)
 496     (arguments
 497      `(#:out-of-source? #f
 498        #:phases
 499        (modify-phases %standard-phases
 500          (replace 'install              ; no install target
 501            (lambda* (#:key outputs #:allow-other-keys)
 502              (let* ((out (assoc-ref outputs "out"))
 503                     (include (string-append out "/include"))
 504                     (doc (string-append out "/share/doc/" ,name)))
 505                (copy-recursively "source" include)
 506                (install-file "README.md" doc)
 507                #t))))))
 508     (home-page "https://github.com/nemtrif/utfcpp")
 509     (synopsis "Portable C++ library for handling UTF-8")
 510     (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
 511 in a portable way.")
 512     (license license:boost1.0)))
 513
 514 (define-public dbacl
 515   (package
 516     (name "dbacl")
 517     (version "1.14")
 518     (source
 519      (origin
 520        (method url-fetch)
 521        (uri (string-append "http://www.lbreyer.com/gpl/"
 522                            "dbacl-" version ".tar.gz"))
 523        (sha256
 524         (base32
 525          "0224g6x71hyvy7jikfxmgcwww1r5lvk0jx36cva319cb9nmrbrq7"))))
 526     (build-system gnu-build-system)
 527     (arguments
 528      `(#:make-flags
 529        (list
 530         (string-append "-I" (assoc-ref %build-inputs "slang")
 531                        "/include/slang")
 532         (string-append "-I" (assoc-ref %build-inputs "ncurses")
 533                        "/include/ncurses"))
 534        #:phases
 535        (modify-phases %standard-phases
 536          (add-after 'unpack 'delete-sample6-and-japanese
 537            (lambda _
 538              (substitute* "doc/Makefile.am"
 539                (("sample6.txt") "")
 540                (("japanese.txt") ""))
 541              (delete-file "doc/sample6.txt")
 542              (delete-file "doc/japanese.txt")
 543              (substitute* (list "src/tests/Makefile.am"
 544                                 "src/tests/Makefile.in")
 545                (("dbacl-jap.shin") "")
 546                (("dbacl-jap.sh") ""))
 547              #t))
 548          (add-after 'unpack 'delete-test
 549            ;; See comments about the license.
 550            (lambda _
 551              (delete-file "src/tests/dbacl-jap.shin")
 552              #t))
 553          (add-after 'delete-sample6-and-japanese 'autoreconf
 554            (lambda _
 555              (invoke "autoreconf" "-vif")
 556              #t))
 557          (add-after 'unpack 'fix-test-files
 558            (lambda* (#:key inputs outputs #:allow-other-keys)
 559              (let* ((out (assoc-ref outputs "out"))
 560                     (bin (string-append out "/bin")))
 561                (substitute* (find-files "src/tests/" "\\.shin$")
 562                  (("PATH=/bin:/usr/bin")
 563                   "#PATH=/bin:/usr/bin")
 564                  (("diff") (string-append (which "diff")))
 565                  (("tr") (string-append (which "tr"))))
 566                #t))))))
 567     (inputs
 568      `(("ncurses" ,ncurses)
 569        ("perl" ,perl)
 570        ("readline" ,readline)
 571        ("slang" ,slang)))
 572     (native-inputs
 573      `(("libtool" ,libtool)
 574        ("autoconf" ,autoconf)
 575        ("automake" ,automake)
 576        ("pkg-config" ,pkg-config)))
 577     (home-page "https://www.lbreyer.com/dbacl.html")
 578     (synopsis "Bayesian text and email classifier")
 579     (description
 580      "dbacl is a fast Bayesian text and email classifier.  It builds a variety
 581 of language models using maximum entropy (minimum divergence) principles, and
 582 these can then be used to categorize input data automatically among multiple
 583 categories.")
 584     ;; The software is licensed as GPLv3 or later, but
 585     ;; includes various sample texts in the doc dir:
 586     ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
 587     ;;   by Mark Twain.
 588     ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
 589     ;; - sample6.txt is a forwarded email, copyright unknown.
 590     ;;   Guix does exclude sample6.txt.
 591     ;; - japanese.txt is a Japanese unoffical translation of the
 592     ;;   GNU General Public License, (c) by the Free Software Foundation.
 593     ;;   Guix excludes this file.
 594     (license (list license:gpl3+ license:public-domain))))
 595
 596 (define-public dotconf
 597   (package
 598     (name "dotconf")
 599     (version "1.3")
 600     (source (origin
 601               (method git-fetch)
 602               (uri (git-reference
 603                     (url "https://github.com/williamh/dotconf.git")
 604                     (commit (string-append "v" version))))
 605               (file-name (git-file-name name version))
 606               (sha256
 607                (base32
 608                 "1sc95hw5k2xagpafny0v35filmcn05k1ds5ghkldfpf6xw4hakp7"))))
 609     (build-system gnu-build-system)
 610     (arguments `(#:tests? #f))  ; FIXME maketest.sh does not work.
 611     (native-inputs
 612      `(("autoconf" ,autoconf)
 613        ("automake" ,automake)
 614        ("libtool" ,libtool)))
 615     (home-page "https://github.com/williamh/dotconf")
 616     (synopsis "Configuration file parser library")
 617     (description
 618      "C library for creating and parsing configuration files.")
 619     (license (list license:lgpl2.1         ; Main distribution.
 620                    license:asl1.1))))      ; src/readdir.{c,h}
 621
 622 (define-public java-rsyntaxtextarea
 623   (package
 624     (name "java-rsyntaxtextarea")
 625     (version "2.6.1")
 626     (source (origin
 627               (method url-fetch)
 628               (uri (string-append "https://github.com/bobbylight/"
 629                                   "RSyntaxTextArea/archive/"
 630                                   version ".tar.gz"))
 631               (file-name (string-append name "-" version ".tar.gz"))
 632               (sha256
 633                (base32
 634                 "0c5mqg2klj5rvf8fhycrli8rf6s37l9p7a8knw9gpp65r1c120q2"))))
 635     (build-system ant-build-system)
 636     (arguments
 637      `(;; FIXME: some tests fail because locale resources cannot be found.
 638        ;; Even when I add them to the class path,
 639        ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
 640        #:tests? #f
 641        #:jar-name "rsyntaxtextarea.jar"))
 642     (native-inputs
 643      `(("java-junit" ,java-junit)
 644        ("java-hamcrest-core" ,java-hamcrest-core)))
 645     (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
 646     (synopsis "Syntax highlighting text component for Java Swing")
 647     (description "RSyntaxTextArea is a syntax highlighting, code folding text
 648 component for Java Swing.  It extends @code{JTextComponent} so it integrates
 649 completely with the standard @code{javax.swing.text} package.  It is fast and
 650 efficient, and can be used in any application that needs to edit or view
 651 source code.")
 652     (license license:bsd-3)))
 653
 654 ;; We use the sources from git instead of the tarball from pypi, because the
 655 ;; latter does not include the Cython source file from which bycython.cpp is
 656 ;; generated.
 657 (define-public python-editdistance
 658   (let ((commit "3ea84a7dd3258c76aa3be851ef3d50e59c886846")
 659         (revision "1"))
 660     (package
 661       (name "python-editdistance")
 662       (version (string-append "0.3.1-" revision "." (string-take commit 7)))
 663       (source
 664        (origin
 665          (method git-fetch)
 666          (uri (git-reference
 667                (url "https://github.com/aflc/editdistance.git")
 668                (commit commit)))
 669          (file-name (git-file-name name version))
 670          (sha256
 671           (base32
 672            "1l43svsv12crvzphrgi6x435z6xg8m086c64armp8wzb4l8ccm7g"))))
 673       (build-system python-build-system)
 674       (arguments
 675        `(#:phases
 676          (modify-phases %standard-phases
 677            (add-after 'unpack 'build-cython-code
 678              (lambda _
 679                (with-directory-excursion "editdistance"
 680                  (delete-file "bycython.cpp")
 681                  (invoke "cython" "--cplus" "bycython.pyx")))))))
 682       (native-inputs
 683        `(("python-cython" ,python-cython)))
 684       (home-page "https://www.github.com/aflc/editdistance")
 685       (synopsis "Fast implementation of the edit distance (Levenshtein distance)")
 686       (description
 687        "This library simply implements Levenshtein distance algorithm with C++
 688 and Cython.")
 689       (license license:expat))))
 690
 691 (define-public go-github.com-mattn-go-runewidth
 692   (package
 693     (name "go-github.com-mattn-go-runewidth")
 694     (version "0.0.2")
 695     (source (origin
 696               (method git-fetch)
 697               (uri (git-reference
 698                     (url "https://github.com/mattn/go-runewidth")
 699                     (commit (string-append "v" version))))
 700               (file-name (git-file-name name version))
 701               (sha256
 702                (base32
 703                 "0vkrfrz3fzn5n6ix4k8s0cg0b448459sldq8bp4riavsxm932jzb"))))
 704     (build-system go-build-system)
 705     (arguments
 706      '(#:import-path "github.com/mattn/go-runewidth"))
 707     (synopsis "@code{runewidth} provides Go functions to work with string widths")
 708     (description
 709      "The @code{runewidth} library provides Go functions for padding,
 710 measuring and checking the width of strings, with support east asian text.")
 711     (home-page "https://github.com/jessevdk/go-flags")
 712     (license license:expat)))
 713
 714 (define-public docx2txt
 715   (package
 716     (name "docx2txt")
 717     (version "1.4")
 718     (source (origin
 719               (method url-fetch)
 720               (uri (string-append
 721                     "mirror://sourceforge/docx2txt/docx2txt/v"
 722                     version "/docx2txt-" version ".tgz"))
 723               (sha256
 724                (base32
 725                 "06vdikjvpj6qdb41d8wzfnyj44jpnknmlgbhbr1w215420lpb5xj"))))
 726     (build-system gnu-build-system)
 727     (inputs
 728      `(("unzip" ,unzip)
 729        ("perl" ,perl)))
 730     (arguments
 731      `(#:tests? #f                      ; No tests.
 732        #:make-flags (list (string-append "BINDIR="
 733                                          (assoc-ref %outputs "out") "/bin")
 734                           (string-append "CONFIGDIR="
 735                                          (assoc-ref %outputs "out") "/etc")
 736                           ;; Makefile seems to be a bit dumb at guessing.
 737                           (string-append "INSTALL=install")
 738                           (string-append "PERL=perl"))
 739        #:phases
 740        (modify-phases %standard-phases
 741          (delete 'configure)
 742          (add-after 'install 'fix-install
 743            (lambda* (#:key outputs inputs #:allow-other-keys)
 744              (let* ((out (assoc-ref outputs "out"))
 745                     (bin (string-append out "/bin"))
 746                     (config (string-append out "/etc/docx2txt.config"))
 747                     (unzip (assoc-ref inputs "unzip")))
 748                ;; According to INSTALL, the .sh wrapper can be skipped.
 749                (delete-file (string-append bin "/docx2txt.sh"))
 750                (rename-file (string-append bin "/docx2txt.pl")
 751                             (string-append bin "/docx2txt"))
 752                (substitute* config
 753                  (("config_unzip         => '/usr/bin/unzip',")
 754                   (string-append "config_unzip         => '"
 755                                  unzip
 756                                  "/bin/unzip',")))
 757                ;; Makefile is wrong.
 758                (chmod config #o644)
 759                #t))))))
 760     (synopsis "Recover text from @file{.docx} files, with good formatting")
 761     (description
 762      "@command{docx2txt} is a Perl based command line utility to convert
 763 Microsoft Office @file{.docx} documents to equivalent text documents.  Latest
 764 version supports following features during text extraction.
 765
 766 @itemize
 767 @item Character conversions; currency characters are converted to respective
 768 names like Euro.
 769 @item Capitalisation of text blocks.
 770 @item Center and right justification of text fitting in a line of
 771 (configurable) 80 columns.
 772 @item Horizontal ruler, line breaks, paragraphs separation, tabs.
 773 @item Indicating hyperlinked text along with the hyperlink (configurable).
 774 @item Handling (bullet, decimal, letter, roman) lists along with (attempt at)
 775 indentation.
 776 @end itemize\n")
 777     (home-page "http://docx2txt.sourceforge.net")
 778     (license license:gpl3+)))
 779
 780 (define-public opencc
 781   (package
 782     (name "opencc")
 783     (version "1.0.5")
 784     (source
 785      (origin
 786        (method git-fetch)
 787        (uri (git-reference
 788               (url "https://github.com/BYVoid/OpenCC")
 789               (commit (string-append "ver." version))))
 790        (file-name (git-file-name name version))
 791        (sha256
 792         (base32
 793          "1pv5md225qwhbn8ql932zdg6gh1qlx3paiajaks8gfsa07yzvhr4"))
 794        (modules '((guix build utils)))
 795        (snippet
 796         '(begin
 797            ;; TODO: Unbundle tclap, darts-clone, gtest
 798            (delete-file-recursively "deps/rapidjson-0.11") #t))))
 799     (build-system cmake-build-system)
 800     (arguments
 801      '(#:phases
 802        (modify-phases %standard-phases
 803          (add-after 'unpack 'patch-3rd-party-references
 804            (lambda* (#:key inputs #:allow-other-keys)
 805              (let ((rapidjson (assoc-ref inputs "rapidjson")))
 806                (substitute* "src/CMakeLists.txt"
 807                  (("../deps/rapidjson-0.11")
 808                   (string-append rapidjson "/include/rapidjson")))
 809              #t))))))
 810     (native-inputs
 811      `(("python" ,python-wrapper)
 812        ("rapidjson" ,rapidjson)))
 813     (home-page "https://github.com/BYVoid/OpenCC")
 814     (synopsis "Convert between Traditional Chinese and Simplified Chinese")
 815     (description "Open Chinese Convert (OpenCC) converts between Traditional
 816 Chinese and Simplified Chinese, supporting character-level conversion,
 817 phrase-level conversion, variant conversion, and regional idioms among
 818 Mainland China, Taiwan, and Hong-Kong.")
 819     (license license:asl2.0)))
 820
 821 (define-public nkf
 822   (let ((commit "08043eadf4abdddcf277842217e3c77a24740dc2")
 823         (revision "1"))
 824     (package
 825       (name "nkf")
 826       ;; The commits corresponding to specific versions are published
 827       ;; here:
 828       ;; https://ja.osdn.net/projects/nkf/scm/git/nkf/
 829       (version "2.1.5")
 830       (source (origin
 831                 (method git-fetch)
 832                 (uri (git-reference
 833                       (url "https://github.com/nurse/nkf.git")
 834                       (commit commit)))
 835                 (file-name (git-file-name name version))
 836                 (sha256
 837                  (base32
 838                   "0anw0knr1iy4p9w3d3b3pbwzh1c43p1i2q4c28kw9zviw8kx2rly"))))
 839       (build-system gnu-build-system)
 840       (arguments
 841        `(#:tests? #f ; test for perl module
 842          #:make-flags (list "CC=gcc" "CFLAGS=-O2 -Wall -pedantic"
 843                             (string-append "prefix=" %output)
 844                             "MKDIR=mkdir -p")
 845          #:phases
 846          (modify-phases %standard-phases
 847            (delete 'configure)))) ; No ./configure script
 848       (home-page "https://ja.osdn.net/projects/nkf/")
 849       (synopsis "Network Kanji Filter")
 850       (description "Nkf is yet another kanji code converter among networks,
 851 hosts and terminals.  It converts input kanji code to designated kanji code
 852 such as ISO-2022-JP, Shift_JIS, EUC-JP, UTF-8, UTF-16 or UTF-32.")
 853       (license license:zlib))))
 854
 855 (define-public python-pandocfilters
 856   (package
 857     (name "python-pandocfilters")
 858     (version "1.4.2")
 859     (source
 860      (origin
 861        (method url-fetch)
 862        (uri (pypi-uri "pandocfilters" version))
 863        (sha256
 864         (base32
 865          "1a8d9b7s48gmq9zj0pmbyv2sivn5i7m6mybgpkk4jm5vd7hp1pdk"))))
 866     (build-system python-build-system)
 867     (home-page "https://github.com/jgm/pandocfilters")
 868     (synopsis "Python module for writing Pandoc filters")
 869     (description "Pandoc is a powerful utility to transform various
 870 input formats into a wide range of output formats.  To alter the
 871 exported output document, Pandoc allows the usage of filters, which
 872 are pipes that read a JSON serialization of the Pandoc AST from stdin,
 873 transform it in some way, and write it to stdout.  It allows therefore
 874 to alter the processing of Pandoc's supported input formats, for
 875 instance one can add new syntax elements to markdown, etc.
 876
 877 This package provides Python bindings.")
 878     (license license:bsd-3)))