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, 2019 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 © 2017,2019 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, 2019 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 base)
  49   #:use-module (gnu packages compression)
  50   #:use-module (gnu packages gettext)
  51   #:use-module (gnu packages java)
  52   #:use-module (gnu packages ncurses)
  53   #:use-module (gnu packages perl)
  54   #:use-module (gnu packages pkg-config)
  55   #:use-module (gnu packages python)
  56   #:use-module (gnu packages python-xyz)
  57   #:use-module (gnu packages readline)
  58   #:use-module (gnu packages slang)
  59   #:use-module (gnu packages web))
  60
  61 (define-public dos2unix
  62   (package
  63     (name "dos2unix")
  64     (version "7.4.0")
  65     (source
  66      (origin
  67        (method url-fetch)
  68        (uri (string-append "https://waterlan.home.xs4all.nl/" name "/"
  69                            name "-" version ".tar.gz"))
  70        (sha256
  71         (base32 "12h4c61g376bhq03y5g2xszkrkrj5hwd928rly3xsp6rvfmnbixs"))))
  72     (build-system gnu-build-system)
  73     (arguments
  74      '(#:make-flags (list "CC=gcc"
  75                           (string-append "prefix=" (assoc-ref %outputs "out")))
  76        #:phases
  77        (modify-phases %standard-phases
  78          (delete 'configure)))) ; no configure script
  79     (native-inputs
  80      `(("gettext" ,gettext-minimal)
  81        ("perl" ,perl)))
  82     (home-page "https://waterlan.home.xs4all.nl/dos2unix.html")
  83     (synopsis "DOS/Mac to Unix and vice versa text file format converter")
  84     (description
  85      "dos2unix is a tool to convert line breaks in a text file from Unix format
  86 to DOS format and vice versa.")
  87     (license license:bsd-2)))
  88
  89 (define-public recode
  90   (package
  91     (name "recode")
  92     (version "3.7.6")
  93     (source
  94      (origin
  95        (method url-fetch)
  96        (uri (string-append "https://github.com/rrthomas/recode/releases/"
  97                            "download/v" version "/recode-" version ".tar.gz"))
  98        (sha256
  99         (base32 "0m59sd1ca0zw1aydpc3m8sw03nc885knmccqryg7byzmqs585ia6"))))
 100     (build-system gnu-build-system)
 101     (native-inputs
 102      `(("python" ,python)
 103        ("python-cython" ,python-cython)))
 104     (home-page "https://github.com/rrthomas/recode")
 105     (synopsis "Text encoding converter")
 106     (description "The Recode library converts files between character sets and
 107 usages.  It recognises or produces over 200 different character sets (or about
 108 300 if combined with an iconv library) and transliterates files between almost
 109 any pair.  When exact transliteration are not possible, it gets rid of
 110 offending characters or falls back on approximations.  The recode program is a
 111 handy front-end to the library.")
 112     (license license:gpl3+)))
 113
 114 (define-public enca
 115   (package
 116     (name "enca")
 117     (version "1.19")
 118     (source
 119      (origin
 120        (method git-fetch)
 121        (uri (git-reference
 122               (url "https://github.com/nijel/enca")
 123               (commit version)))
 124        (file-name (git-file-name name version))
 125        (sha256
 126         (base32 "19q7cwwxmmk5j9438bsqdpjvdjawsd3zmw1zyqgi7s4m0rasr3ah"))))
 127     (build-system gnu-build-system)
 128     ;; enca-1.19 tests fail with recent recode.
 129     ;(inputs `(("recode" ,recode)))
 130     (home-page "https://github.com/nijel/enca")
 131     (synopsis "Text encoding detection tool")
 132     (description "Enca (Extremely Naive Charset Analyser) consists of libenca,
 133 an encoding detection library, and enca, a command line frontend, integrating
 134 libenca and several charset conversion libraries and tools.")
 135     (license license:gpl2)))
 136
 137 (define-public utf8proc
 138   (package
 139     (name "utf8proc")
 140     (version "2.4.0")
 141     (source
 142      (origin
 143        (method git-fetch)
 144        (uri (git-reference
 145              (url "https://github.com/JuliaStrings/utf8proc")
 146              (commit (string-append "v" version))))
 147        (file-name (git-file-name name version))
 148        (sha256
 149         (base32 "1i42hqwc8znqii9brangwkxk5cyc2lk95ip405fg88zr7z2ncr34"))))
 150     (build-system gnu-build-system)
 151     (native-inputs           ;test data that is otherwise downloaded with curl
 152      `(("NormalizationTest.txt"
 153         ,(origin
 154            (method url-fetch)
 155            (uri (string-append "https://www.unicode.org/Public/12.1.0/ucd/"
 156                                "NormalizationTest.txt"))
 157            (sha256
 158             (base32 "0hb97k9xv1lr847hwz0719ksqy39s47xw6k01dgs1368jdibvawc"))))
 159        ("GraphemeBreakTest.txt"
 160         ,(origin
 161            (method url-fetch)
 162            (uri (string-append "https://www.unicode.org/Public/12.1.0/ucd/"
 163                                "auxiliary/GraphemeBreakTest.txt"))
 164            (sha256
 165             (base32 "0qc90ppmrwfn3y9cdn8jcjrn7qpdf0fhxkwh945yp4rvh37mbgcm"))))
 166
 167        ;; For tests.
 168        ("perl" ,perl)))
 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              #t)))))
 185     (home-page "https://juliastrings.github.io/utf8proc/")
 186     (synopsis "C library for processing UTF-8 Unicode data")
 187     (description "utf8proc is a small C library that provides Unicode
 188 normalization, case-folding, and other operations for data in the UTF-8
 189 encoding, supporting Unicode version 9.0.0.")
 190     (license license:expat)))
 191
 192 (define-public libconfuse
 193   (package
 194     (name "libconfuse")
 195     (version "3.2.2")
 196     (source
 197      (origin
 198        (method url-fetch)
 199        (uri (string-append "https://github.com/martinh/libconfuse/"
 200                            "releases/download/v" version
 201                            "/confuse-" version ".tar.xz"))
 202        (sha256
 203         (base32 "02r1mmzik2m0iigbc2da3y754vj24i18r3ml5p2wzs027mjhn959"))))
 204     (build-system gnu-build-system)
 205     (home-page "https://github.com/martinh/libconfuse")
 206     (synopsis "Configuration file parser library")
 207     (description "libconfuse is a configuration file parser library.  It
 208 supports sections and (lists of) values (strings, integers, floats, booleans
 209 or other sections), as well as some other features (such as
 210 single/double-quoted strings, environment variable expansion, functions and
 211 nested include statements).")
 212     (license license:isc)))
 213
 214 (define-public libgtextutils
 215   (package
 216     (name "libgtextutils")
 217     (version "0.7")
 218     (source
 219      (origin
 220        (method url-fetch)
 221        (uri (string-append
 222              "https://github.com/agordon/libgtextutils/releases/download/"
 223              version "/libgtextutils-" version ".tar.gz"))
 224        (sha256
 225         (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
 226     (build-system gnu-build-system)
 227     (arguments
 228      '(#:phases
 229        (modify-phases %standard-phases
 230          (replace 'bootstrap
 231            (lambda _ (invoke "sh" "reconf"))))))
 232     (native-inputs
 233      `(("autoconf" ,autoconf)
 234        ("automake" ,automake)
 235        ("libtool" ,libtool)))
 236     (home-page "https://github.com/agordon/libgtextutils")
 237     (synopsis "Gordon's text utils library")
 238     (description
 239      "libgtextutils is a text utilities library used by the fastx toolkit from
 240 the Hannon Lab.")
 241     (license license:agpl3+)))
 242
 243 (define-public cityhash
 244   (let ((commit "8af9b8c"))
 245     (package
 246       (name "cityhash")
 247       (version (string-append "1.1-2." commit))
 248       (source (origin
 249                 (method git-fetch)
 250                 (uri (git-reference
 251                       (url "https://github.com/google/cityhash.git")
 252                       (commit commit)))
 253                 (file-name (string-append name "-" version ".tar.gz"))
 254                 (sha256
 255                  (base32
 256                   "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
 257       (build-system gnu-build-system)
 258       (arguments
 259        '(#:make-flags (list "CXXFLAGS=-g -O3")
 260          #:phases
 261          (modify-phases %standard-phases
 262            ;; citycrc is not installed by default but is used by some
 263            ;; programs.
 264            (add-after 'install 'install-citycrc
 265              (lambda* (#:key outputs #:allow-other-keys)
 266                (let* ((out (assoc-ref outputs "out"))
 267                       (include (string-append out "/include")))
 268                  (install-file "src/citycrc.h" include))
 269                #t)))))
 270       (home-page "https://github.com/google/cityhash")
 271       (synopsis "C++ hash functions for strings")
 272       (description
 273        "CityHash provides hash functions for strings.  The functions mix the
 274 input bits thoroughly but are not suitable for cryptography.")
 275       (license license:expat))))
 276
 277 (define-public ustr
 278   (package
 279     (name "ustr")
 280     (version "1.0.4")
 281     (source (origin
 282               (method url-fetch)
 283               (uri (string-append "http://www.and.org/ustr/" version
 284                                   "/ustr-" version ".tar.bz2"))
 285               (sha256
 286                (base32
 287                 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
 288               (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
 289     (build-system gnu-build-system)
 290     (arguments
 291      `(#:make-flags
 292        (list "CC=gcc"
 293              "HIDE="
 294              ;; Override "/sbin/ldconfig" with "echo" because we don't need
 295              ;; "ldconfig".
 296              "LDCONFIG=echo"
 297              (string-append "prefix=" (assoc-ref %outputs "out"))
 298              "all-shared")
 299        #:phases
 300        (modify-phases %standard-phases
 301          (add-after 'unpack 'disable-check-for-stdint
 302            (lambda _
 303              ;; Of course we have stdint.h, just not in /usr/include
 304              (substitute* '("Makefile"
 305                             "ustr-import.in")
 306                (("-f \"/usr/include/stdint.h\"") "-z \"\""))
 307              #t))
 308          ;; No configure script
 309          (delete 'configure))))
 310     (home-page "http://www.and.org/ustr/")
 311     (synopsis "String library with very low memory overhead")
 312     (description
 313      "Ustr is a string library for C with very low memory overhead.")
 314     ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
 315     ;; LGPL, etc. ... if you need another license to help compatibility, just
 316     ;; ask for it.  It's basically public domain, without all the legal
 317     ;; problems for everyone that trying to make something public domain
 318     ;; entails."
 319     (license license:public-domain)))
 320
 321 (define-public ascii2binary
 322   (package
 323     (name "ascii2binary")
 324     (version "2.14")
 325     (source
 326      (origin
 327        (method url-fetch)
 328        (uri (string-append "http://billposer.org/Software/Downloads/"
 329                            "ascii2binary-" version ".tar.bz2"))
 330        (sha256
 331         (base32 "0dc9fxcdmppbs9s06jvq61zbk552laxps0xyk098gj41697ihd96"))))
 332     (build-system gnu-build-system)
 333     (native-inputs
 334      `(("gettext" ,gettext-minimal)))
 335     (home-page "https://billposer.org/Software/a2b.html")
 336     (synopsis "Convert between ASCII, hexadecimal and binary representations")
 337     (description "The two programs are useful for generating test data, for
 338 inspecting binary files, and for interfacing programs that generate textual
 339 output to programs that require binary input and conversely.  They can also be
 340 useful when it is desired to reformat numbers.
 341
 342 @itemize
 343
 344 @item @command{ascii2binary} reads input consisting of ascii or hexadecimal
 345    representation numbers separated by whitespace and produces as output
 346    the binary equivalents.  The type and precision of the binary output
 347    is selected using command line flags.
 348
 349 @item @command{binary2ascii} reads input consisting of binary numbers
 350    and converts them to their ascii or hexadecimal representation.
 351    Command line flags specify the type and size of the binary numbers
 352    and provide control over the format of the output.
 353    Unsigned integers may be written out in binary, octal, decimal,
 354    or hexadecimal.
 355
 356    Signed integers may be written out only in binary or decimal.  Floating
 357    point numbers may be written out only decimal, either in standard or
 358    scientific notation.  (If you want to examine the binary representation
 359    of floating point numbers, just treat the input as a sequence of unsigned
 360    characters.)
 361
 362 @end itemize")
 363     (license license:gpl3)))
 364
 365 (define-public uniutils
 366   (package
 367     (name "uniutils")
 368     (version "2.27")
 369     (source
 370      (origin
 371        (method url-fetch)
 372        (uri (string-append "http://billposer.org/Software/Downloads/"
 373                            "uniutils-" version ".tar.bz2"))
 374        (sha256
 375         (base32 "19w1510w87gx7n4qy3zsb0m467a4rn5scvh4ajajg7jh6x5xri08"))))
 376     (build-system gnu-build-system)
 377     (arguments
 378      '(#:configure-flags '("--disable-dependency-tracking")
 379        #:phases
 380        (modify-phases %standard-phases
 381          (add-after 'build 'fix-paths
 382            (lambda* (#:key outputs inputs #:allow-other-keys)
 383              (let ((out (assoc-ref outputs "out"))
 384                    (a2b (assoc-ref inputs "ascii2binary"))
 385                    (iconv (assoc-ref inputs "libiconv")))
 386                (substitute* "utf8lookup"
 387                  (("^ascii2binary ") (string-append a2b "/bin/ascii2binary "))
 388                  (("^uniname ") (string-append out "/bin/uniname "))
 389                  (("^iconv ") (string-append iconv "/bin/iconv ")))
 390              #t))))))
 391     (inputs
 392      `(("ascii2binary" ,ascii2binary)
 393        ("libiconv" ,libiconv)))
 394     (home-page "https://billposer.org/Software/unidesc.html")
 395     (synopsis "Find out what is in a Unicode file")
 396     (description "Useful tools when working with Unicode files when one
 397 doesn't know the writing system, doesn't have the necessary font, needs to
 398 inspect invisible characters, needs to find out whether characters have been
 399 combined or in what order they occur, or needs statistics on which characters
 400 occur.
 401
 402 @itemize
 403
 404 @item @command{uniname} defaults to printing the character offset of each
 405 character, its byte offset, its hex code value, its encoding, the glyph
 406 itself, and its name.  It may also be used to validate UTF-8 input.
 407
 408 @item @command{unidesc} reports the character ranges to which different
 409 portions of the text belong.   It can also be used to identify Unicode encodings
 410 (e.g. UTF-16be) flagged by magic numbers.
 411
 412 @item @command{unihist} generates a histogram of the characters in its input.
 413
 414 @item @command{ExplicateUTF8} is intended for debugging or for learning about
 415 Unicode.  It determines and explains the validity of a sequence of bytes as a
 416 UTF8 encoding.
 417
 418 @item @command{utf8lookup} provides a handy way to look up Unicode characters
 419 from the command line.
 420
 421 @item @command{unireverse} reverse each line of UTF-8 input
 422 character-by-character.
 423
 424 @end itemize")
 425     (license license:gpl3)))
 426
 427 (define-public libconfig
 428   (package
 429     (name "libconfig")
 430     (version "1.7.2")
 431     (home-page "https://hyperrealm.github.io/libconfig/")
 432     (source (origin
 433               (method url-fetch)
 434               (uri (string-append home-page "/dist/libconfig-"
 435                                   version ".tar.gz"))
 436               (sha256
 437                (base32
 438                 "1ngs2qx3cx5cbwinc5mvadly0b5n7s86zsc68c404czzfff7lg3w"))))
 439     (build-system gnu-build-system)
 440     (synopsis "C/C++ configuration file library")
 441     (description
 442      "Libconfig is a simple library for manipulating structured configuration
 443 files.  This file format is more compact and more readable than XML.  And
 444 unlike XML, it is type-aware, so it is not necessary to do string parsing in
 445 application code.")
 446     (license license:lgpl2.1+)))
 447
 448 (define-public pfff
 449   (package
 450     (name "pfff")
 451     (version "1.0")
 452     (source (origin
 453               (method git-fetch)
 454               (uri (git-reference
 455                      (url "https://github.com/pfff/pfff")
 456                      (commit (string-append "v" version))))
 457               (file-name (git-file-name name version))
 458               (sha256
 459                (base32
 460                 "1nxkfm7zliq3rmr7yp871sppwfnz71iz364m2sgazny71pzykggc"))))
 461     (build-system cmake-build-system)
 462     (home-page "http://biit.cs.ut.ee/pfff/")
 463     (synopsis "Probabilistic fast file fingerprinting tool")
 464     (description
 465      "pfff is a tool for calculating a compact digital fingerprint of a file
 466 by sampling randomly from the file instead of reading it in full.
 467 Consequently, the computation has a flat performance characteristic,
 468 correlated with data variation rather than file size.  pfff can be as reliable
 469 as existing hashing techniques, with provably negligible risk of collisions.")
 470     (license license:bsd-3)))
 471
 472 (define-public oniguruma
 473   (package
 474     (name "oniguruma")
 475     (version "6.9.3")
 476     (source (origin
 477               (method url-fetch)
 478               (uri (string-append "https://github.com/kkos/"
 479                                   "oniguruma/releases/download/v" version
 480                                   "/onig-" version ".tar.gz"))
 481               (sha256
 482                (base32
 483                 "0pvj37r1rd5h5vw99mdk8z4k44gq1ldwrapkamdiicksdfkr4ndb"))))
 484     (build-system gnu-build-system)
 485     (home-page "https://github.com/kkos/oniguruma")
 486     (synopsis "Regular expression library")
 487     (description "Oniguruma is a regular expressions library.  The special
 488 characteristic of this library is that different character encoding for every
 489 regular expression object can be specified.")
 490     (license license:bsd-2)))
 491
 492 (define-public antiword
 493   (package
 494     (name "antiword")
 495     (version "0.37")
 496     (source (origin
 497               (method url-fetch)
 498               (uri (string-append "http://www.winfield.demon.nl/linux"
 499                                   "/antiword-" version ".tar.gz"))
 500               (sha256
 501                (base32
 502                 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
 503               (patches (search-patches "antiword-CVE-2014-8123.patch"))))
 504     (build-system gnu-build-system)
 505     (arguments
 506      `(#:tests? #f ; There are no tests
 507        #:make-flags
 508        (list "-f" "Makefile.Linux"
 509              (string-append "GLOBAL_INSTALL_DIR="
 510                             (assoc-ref %outputs "out") "/bin")
 511              (string-append "GLOBAL_RESOURCES_DIR="
 512                             (assoc-ref %outputs "out") "/share/antiword"))
 513        #:phases
 514        (modify-phases %standard-phases
 515          (replace 'configure
 516            (lambda* (#:key outputs #:allow-other-keys)
 517              ;; Ensure that mapping files can be found in the actual package
 518              ;; data directory.
 519              (substitute* "antiword.h"
 520                (("/usr/share/antiword")
 521                 (string-append (assoc-ref outputs "out") "/share/antiword")))
 522              #t))
 523          (replace 'install
 524            (lambda* (#:key make-flags #:allow-other-keys)
 525              (apply invoke "make" `("global_install" ,@make-flags)))))))
 526     (home-page "http://www.winfield.demon.nl/")
 527     (synopsis "Microsoft Word document reader")
 528     (description "Antiword is an application for displaying Microsoft Word
 529 documents.  It can also convert the document to PostScript or XML.  Only
 530 documents made by MS Word version 2 and version 6 or later are supported.  The
 531 name comes from: \"The antidote against people who send Microsoft Word files
 532 to everybody, because they believe that everybody runs Windows and therefore
 533 runs Word\".")
 534     (license license:gpl2+)))
 535
 536 (define-public catdoc
 537   (package
 538     (name "catdoc")
 539     (version "0.95")
 540     (source (origin
 541               (method url-fetch)
 542               (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
 543                                   "catdoc-" version ".tar.gz"))
 544               (patches (search-patches "catdoc-CVE-2017-11110.patch"))
 545               (sha256
 546                (base32
 547                 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
 548     (build-system gnu-build-system)
 549     ;; TODO: Also build `wordview` which requires `tk` – make a separate
 550     ;; package for this.
 551     (arguments
 552      '(#:tests? #f ; There are no tests
 553        #:configure-flags '("--disable-wordview")
 554        #:phases
 555        (modify-phases %standard-phases
 556          (add-before 'install 'fix-install
 557            (lambda* (#:key outputs #:allow-other-keys)
 558              (let ((out (assoc-ref outputs "out")))
 559                (mkdir-p (string-append out "/share/man/man1"))))))))
 560     (home-page "http://www.wagner.pp.ru/~vitus/software/catdoc/")
 561     (synopsis "MS-Word to TeX or plain text converter")
 562     (description "@command{catdoc} extracts text from MS-Word files, trying to
 563 preserve as many special printable characters as possible.  It supports
 564 everything up to Word-97. Also supported are MS Write documents and RTF files.
 565
 566 @command{catdoc} does not preserve complex word formatting, but it can
 567 translate some non-ASCII characters into TeX escape codes.  It's goal is to
 568 extract plain text and allow you to read it and, probably, reformat with TeX,
 569 according to TeXnical rules.
 570
 571 This package also provides @command{xls2csv}, which extracts data from Excel
 572 spreadsheets and outputs it in comma-separated-value format, and
 573 @command{catppt}, which extracts data from PowerPoint presentations.")
 574     (license license:gpl2+)))
 575
 576 (define-public utfcpp
 577   (package
 578     (name "utfcpp")
 579     (version "2.3.5")
 580     (source (origin
 581               (method git-fetch)
 582               (uri (git-reference
 583                      (url "https://github.com/nemtrif/utfcpp")
 584                      (commit (string-append "v" version))))
 585               (file-name (git-file-name name version))
 586               (sha256
 587                (base32
 588                 "1gr98d826z6wa58r1s5i7rz7q2x3r31v7zj0pjjlrc7gfxwklr4s"))))
 589     (build-system cmake-build-system)
 590     (arguments
 591      `(#:out-of-source? #f
 592        #:phases
 593        (modify-phases %standard-phases
 594          (replace 'install              ; no install target
 595            (lambda* (#:key outputs #:allow-other-keys)
 596              (let* ((out (assoc-ref outputs "out"))
 597                     (include (string-append out "/include"))
 598                     (doc (string-append out "/share/doc/" ,name)))
 599                (copy-recursively "source" include)
 600                (install-file "README.md" doc)
 601                #t))))))
 602     (home-page "https://github.com/nemtrif/utfcpp")
 603     (synopsis "Portable C++ library for handling UTF-8")
 604     (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
 605 in a portable way.")
 606     (license license:boost1.0)))
 607
 608 (define-public dbacl
 609   (package
 610     (name "dbacl")
 611     (version "1.14")
 612     (source
 613      (origin
 614        (method url-fetch)
 615        (uri (string-append "http://www.lbreyer.com/gpl/"
 616                            "dbacl-" version ".tar.gz"))
 617        (sha256
 618         (base32
 619          "0224g6x71hyvy7jikfxmgcwww1r5lvk0jx36cva319cb9nmrbrq7"))))
 620     (build-system gnu-build-system)
 621     (arguments
 622      `(#:make-flags
 623        (list
 624         (string-append "-I" (assoc-ref %build-inputs "slang")
 625                        "/include/slang")
 626         (string-append "-I" (assoc-ref %build-inputs "ncurses")
 627                        "/include/ncurses"))
 628        #:phases
 629        (modify-phases %standard-phases
 630          (add-after 'unpack 'delete-sample6-and-japanese
 631            (lambda _
 632              (substitute* "doc/Makefile.am"
 633                (("sample6.txt") "")
 634                (("japanese.txt") ""))
 635              (delete-file "doc/sample6.txt")
 636              (delete-file "doc/japanese.txt")
 637              (substitute* (list "src/tests/Makefile.am"
 638                                 "src/tests/Makefile.in")
 639                (("dbacl-jap.shin") "")
 640                (("dbacl-jap.sh") ""))
 641              #t))
 642          (add-after 'unpack 'delete-test
 643            ;; See comments about the license.
 644            (lambda _
 645              (delete-file "src/tests/dbacl-jap.shin")
 646              #t))
 647          (add-after 'delete-sample6-and-japanese 'autoreconf
 648            (lambda _
 649              (invoke "autoreconf" "-vif")
 650              #t))
 651          (add-after 'unpack 'fix-test-files
 652            (lambda* (#:key inputs outputs #:allow-other-keys)
 653              (let* ((out (assoc-ref outputs "out"))
 654                     (bin (string-append out "/bin")))
 655                (substitute* (find-files "src/tests/" "\\.shin$")
 656                  (("PATH=/bin:/usr/bin")
 657                   "#PATH=/bin:/usr/bin")
 658                  (("diff") (string-append (which "diff")))
 659                  (("tr") (string-append (which "tr"))))
 660                #t))))))
 661     (inputs
 662      `(("ncurses" ,ncurses)
 663        ("perl" ,perl)
 664        ("readline" ,readline)
 665        ("slang" ,slang)))
 666     (native-inputs
 667      `(("libtool" ,libtool)
 668        ("autoconf" ,autoconf)
 669        ("automake" ,automake)
 670        ("pkg-config" ,pkg-config)))
 671     (home-page "https://www.lbreyer.com/dbacl.html")
 672     (synopsis "Bayesian text and email classifier")
 673     (description
 674      "dbacl is a fast Bayesian text and email classifier.  It builds a variety
 675 of language models using maximum entropy (minimum divergence) principles, and
 676 these can then be used to categorize input data automatically among multiple
 677 categories.")
 678     ;; The software is licensed as GPLv3 or later, but
 679     ;; includes various sample texts in the doc dir:
 680     ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
 681     ;;   by Mark Twain.
 682     ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
 683     ;; - sample6.txt is a forwarded email, copyright unknown.
 684     ;;   Guix does exclude sample6.txt.
 685     ;; - japanese.txt is a Japanese unoffical translation of the
 686     ;;   GNU General Public License, (c) by the Free Software Foundation.
 687     ;;   Guix excludes this file.
 688     (license (list license:gpl3+ license:public-domain))))
 689
 690 (define-public dotconf
 691   (package
 692     (name "dotconf")
 693     (version "1.3")
 694     (source (origin
 695               (method git-fetch)
 696               (uri (git-reference
 697                     (url "https://github.com/williamh/dotconf.git")
 698                     (commit (string-append "v" version))))
 699               (file-name (git-file-name name version))
 700               (sha256
 701                (base32
 702                 "1sc95hw5k2xagpafny0v35filmcn05k1ds5ghkldfpf6xw4hakp7"))))
 703     (build-system gnu-build-system)
 704     (arguments `(#:tests? #f))  ; FIXME maketest.sh does not work.
 705     (native-inputs
 706      `(("autoconf" ,autoconf)
 707        ("automake" ,automake)
 708        ("libtool" ,libtool)))
 709     (home-page "https://github.com/williamh/dotconf")
 710     (synopsis "Configuration file parser library")
 711     (description
 712      "C library for creating and parsing configuration files.")
 713     (license (list license:lgpl2.1         ; Main distribution.
 714                    license:asl1.1))))      ; src/readdir.{c,h}
 715
 716 (define-public java-rsyntaxtextarea
 717   (package
 718     (name "java-rsyntaxtextarea")
 719     (version "2.6.1")
 720     (source (origin
 721               (method url-fetch)
 722               (uri (string-append "https://github.com/bobbylight/"
 723                                   "RSyntaxTextArea/archive/"
 724                                   version ".tar.gz"))
 725               (file-name (string-append name "-" version ".tar.gz"))
 726               (sha256
 727                (base32
 728                 "0c5mqg2klj5rvf8fhycrli8rf6s37l9p7a8knw9gpp65r1c120q2"))))
 729     (build-system ant-build-system)
 730     (arguments
 731      `(;; FIXME: some tests fail because locale resources cannot be found.
 732        ;; Even when I add them to the class path,
 733        ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
 734        #:tests? #f
 735        #:jar-name "rsyntaxtextarea.jar"))
 736     (native-inputs
 737      `(("java-junit" ,java-junit)
 738        ("java-hamcrest-core" ,java-hamcrest-core)))
 739     (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
 740     (synopsis "Syntax highlighting text component for Java Swing")
 741     (description "RSyntaxTextArea is a syntax highlighting, code folding text
 742 component for Java Swing.  It extends @code{JTextComponent} so it integrates
 743 completely with the standard @code{javax.swing.text} package.  It is fast and
 744 efficient, and can be used in any application that needs to edit or view
 745 source code.")
 746     (license license:bsd-3)))
 747
 748 ;; We use the sources from git instead of the tarball from pypi, because the
 749 ;; latter does not include the Cython source file from which bycython.cpp is
 750 ;; generated.
 751 (define-public python-editdistance
 752   (let ((commit "3ea84a7dd3258c76aa3be851ef3d50e59c886846")
 753         (revision "1"))
 754     (package
 755       (name "python-editdistance")
 756       (version (string-append "0.3.1-" revision "." (string-take commit 7)))
 757       (source
 758        (origin
 759          (method git-fetch)
 760          (uri (git-reference
 761                (url "https://github.com/aflc/editdistance.git")
 762                (commit commit)))
 763          (file-name (git-file-name name version))
 764          (sha256
 765           (base32
 766            "1l43svsv12crvzphrgi6x435z6xg8m086c64armp8wzb4l8ccm7g"))))
 767       (build-system python-build-system)
 768       (arguments
 769        `(#:phases
 770          (modify-phases %standard-phases
 771            (add-after 'unpack 'build-cython-code
 772              (lambda _
 773                (with-directory-excursion "editdistance"
 774                  (delete-file "bycython.cpp")
 775                  (invoke "cython" "--cplus" "bycython.pyx")))))))
 776       (native-inputs
 777        `(("python-cython" ,python-cython)))
 778       (home-page "https://www.github.com/aflc/editdistance")
 779       (synopsis "Fast implementation of the edit distance (Levenshtein distance)")
 780       (description
 781        "This library simply implements Levenshtein distance algorithm with C++
 782 and Cython.")
 783       (license license:expat))))
 784
 785 (define-public go-github.com-mattn-go-runewidth
 786   (let ((commit "703b5e6b11ae25aeb2af9ebb5d5fdf8fa2575211")
 787         (version "0.0.4")
 788         (revision "1"))
 789     (package
 790       (name "go-github.com-mattn-go-runewidth")
 791       (version (git-version version revision commit))
 792       (source
 793        (origin
 794          (method git-fetch)
 795          (uri (git-reference
 796                (url "https://github.com/mattn/runewidth")
 797                (commit commit)))
 798          (file-name (git-file-name name version))
 799          (sha256
 800           (base32
 801            "0znpyz71gajx3g0j2zp63nhjj2c07g16885vxv4ykwnrfmzbgk4w"))))
 802       (build-system go-build-system)
 803       (arguments
 804        '(#:import-path "github.com/mattn/go-runewidth"))
 805       (synopsis "@code{runewidth} provides Go functions to work with string widths")
 806       (description
 807        "The @code{runewidth} library provides Go functions for padding,
 808 measuring and checking the width of strings, with support for East Asian
 809 text.")
 810       (home-page "https://github.com/mattn/runewidth")
 811       (license license:expat))))
 812
 813 (define-public docx2txt
 814   (package
 815     (name "docx2txt")
 816     (version "1.4")
 817     (source (origin
 818               (method url-fetch)
 819               (uri (string-append
 820                     "mirror://sourceforge/docx2txt/docx2txt/v"
 821                     version "/docx2txt-" version ".tgz"))
 822               (sha256
 823                (base32
 824                 "06vdikjvpj6qdb41d8wzfnyj44jpnknmlgbhbr1w215420lpb5xj"))))
 825     (build-system gnu-build-system)
 826     (inputs
 827      `(("unzip" ,unzip)
 828        ("perl" ,perl)))
 829     (arguments
 830      `(#:tests? #f                      ; No tests.
 831        #:make-flags (list (string-append "BINDIR="
 832                                          (assoc-ref %outputs "out") "/bin")
 833                           (string-append "CONFIGDIR="
 834                                          (assoc-ref %outputs "out") "/etc")
 835                           ;; Makefile seems to be a bit dumb at guessing.
 836                           (string-append "INSTALL=install")
 837                           (string-append "PERL=perl"))
 838        #:phases
 839        (modify-phases %standard-phases
 840          (delete 'configure)
 841          (add-after 'install 'fix-install
 842            (lambda* (#:key outputs inputs #:allow-other-keys)
 843              (let* ((out (assoc-ref outputs "out"))
 844                     (bin (string-append out "/bin"))
 845                     (config (string-append out "/etc/docx2txt.config"))
 846                     (unzip (assoc-ref inputs "unzip")))
 847                ;; According to INSTALL, the .sh wrapper can be skipped.
 848                (delete-file (string-append bin "/docx2txt.sh"))
 849                (rename-file (string-append bin "/docx2txt.pl")
 850                             (string-append bin "/docx2txt"))
 851                (substitute* config
 852                  (("config_unzip         => '/usr/bin/unzip',")
 853                   (string-append "config_unzip         => '"
 854                                  unzip
 855                                  "/bin/unzip',")))
 856                ;; Makefile is wrong.
 857                (chmod config #o644)
 858                #t))))))
 859     (synopsis "Recover text from @file{.docx} files, with good formatting")
 860     (description
 861      "@command{docx2txt} is a Perl based command line utility to convert
 862 Microsoft Office @file{.docx} documents to equivalent text documents.  Latest
 863 version supports following features during text extraction.
 864
 865 @itemize
 866 @item Character conversions; currency characters are converted to respective
 867 names like Euro.
 868 @item Capitalisation of text blocks.
 869 @item Center and right justification of text fitting in a line of
 870 (configurable) 80 columns.
 871 @item Horizontal ruler, line breaks, paragraphs separation, tabs.
 872 @item Indicating hyperlinked text along with the hyperlink (configurable).
 873 @item Handling (bullet, decimal, letter, roman) lists along with (attempt at)
 874 indentation.
 875 @end itemize\n")
 876     (home-page "http://docx2txt.sourceforge.net")
 877     (license license:gpl3+)))
 878
 879 (define-public odt2txt
 880   (package
 881     (name "odt2txt")
 882     (version "0.5")
 883     (source
 884       (origin
 885         (method git-fetch)
 886         (uri (git-reference
 887                (url "https://github.com/dstosberg/odt2txt/")
 888                (commit (string-append "v" version))))
 889         (file-name (git-file-name name version))
 890         (sha256
 891          (base32
 892           "0im3kzvhxkjlx57w6h13mc9584c74ma1dyymgvpq2y61av3gc35v"))))
 893     (build-system gnu-build-system)
 894     (arguments
 895      `(#:tests? #f ; no make check
 896        #:make-flags (list "CC=gcc"
 897                           (string-append "DESTDIR=" (assoc-ref %outputs "out")))
 898        #:phases
 899        (modify-phases %standard-phases
 900          ;; no configure script
 901          (delete 'configure))))
 902     (inputs
 903      `(("zlib" ,zlib)))
 904     (home-page "https://github.com/dstosberg/odt2txt/")
 905     (synopsis "Converter from OpenDocument Text to plain text")
 906     (description "odt2txt is a command-line tool which extracts the text out
 907 of OpenDocument Texts, as produced by OpenOffice.org, KOffice, StarOffice and
 908 others.
 909
 910 odt2txt can also extract text from some file formats similar to OpenDocument
 911 Text, such as OpenOffice.org XML (*.sxw), which was used by OpenOffice.org
 912 version 1.x and older StarOffice versions.  To a lesser extent, odt2txt may be
 913 useful to extract content from OpenDocument spreadsheets (*.ods) and
 914 OpenDocument presentations (*.odp).")
 915     (license license:gpl2)))
 916
 917 (define-public opencc
 918   (package
 919     (name "opencc")
 920     (version "1.0.5")
 921     (source
 922      (origin
 923        (method git-fetch)
 924        (uri (git-reference
 925               (url "https://github.com/BYVoid/OpenCC")
 926               (commit (string-append "ver." version))))
 927        (file-name (git-file-name name version))
 928        (sha256
 929         (base32
 930          "1pv5md225qwhbn8ql932zdg6gh1qlx3paiajaks8gfsa07yzvhr4"))
 931        (modules '((guix build utils)))
 932        (snippet
 933         '(begin
 934            ;; TODO: Unbundle tclap, darts-clone, gtest
 935            (delete-file-recursively "deps/rapidjson-0.11") #t))))
 936     (build-system cmake-build-system)
 937     (arguments
 938      '(#:phases
 939        (modify-phases %standard-phases
 940          (add-after 'unpack 'patch-3rd-party-references
 941            (lambda* (#:key inputs #:allow-other-keys)
 942              (let ((rapidjson (assoc-ref inputs "rapidjson")))
 943                (substitute* "src/CMakeLists.txt"
 944                  (("../deps/rapidjson-0.11")
 945                   (string-append rapidjson "/include/rapidjson")))
 946              #t))))))
 947     (native-inputs
 948      `(("python" ,python-wrapper)
 949        ("rapidjson" ,rapidjson)))
 950     (home-page "https://github.com/BYVoid/OpenCC")
 951     (synopsis "Convert between Traditional Chinese and Simplified Chinese")
 952     (description "Open Chinese Convert (OpenCC) converts between Traditional
 953 Chinese and Simplified Chinese, supporting character-level conversion,
 954 phrase-level conversion, variant conversion, and regional idioms among
 955 Mainland China, Taiwan, and Hong-Kong.")
 956     (license license:asl2.0)))
 957
 958 (define-public nkf
 959   (let ((commit "08043eadf4abdddcf277842217e3c77a24740dc2")
 960         (revision "1"))
 961     (package
 962       (name "nkf")
 963       ;; The commits corresponding to specific versions are published
 964       ;; here:
 965       ;; https://ja.osdn.net/projects/nkf/scm/git/nkf/
 966       (version "2.1.5")
 967       (source (origin
 968                 (method git-fetch)
 969                 (uri (git-reference
 970                       (url "https://github.com/nurse/nkf.git")
 971                       (commit commit)))
 972                 (file-name (git-file-name name version))
 973                 (sha256
 974                  (base32
 975                   "0anw0knr1iy4p9w3d3b3pbwzh1c43p1i2q4c28kw9zviw8kx2rly"))))
 976       (build-system gnu-build-system)
 977       (arguments
 978        `(#:tests? #f ; test for perl module
 979          #:make-flags (list "CC=gcc" "CFLAGS=-O2 -Wall -pedantic"
 980                             (string-append "prefix=" %output)
 981                             "MKDIR=mkdir -p")
 982          #:phases
 983          (modify-phases %standard-phases
 984            (delete 'configure)))) ; No ./configure script
 985       (home-page "https://ja.osdn.net/projects/nkf/")
 986       (synopsis "Network Kanji Filter")
 987       (description "Nkf is yet another kanji code converter among networks,
 988 hosts and terminals.  It converts input kanji code to designated kanji code
 989 such as ISO-2022-JP, Shift_JIS, EUC-JP, UTF-8, UTF-16 or UTF-32.")
 990       (license license:zlib))))
 991
 992 (define-public python-pandocfilters
 993   (package
 994     (name "python-pandocfilters")
 995     (version "1.4.2")
 996     (source
 997      (origin
 998        (method url-fetch)
 999        (uri (pypi-uri "pandocfilters" version))
1000        (sha256
1001         (base32
1002          "1a8d9b7s48gmq9zj0pmbyv2sivn5i7m6mybgpkk4jm5vd7hp1pdk"))))
1003     (build-system python-build-system)
1004     (home-page "https://github.com/jgm/pandocfilters")
1005     (synopsis "Python module for writing Pandoc filters")
1006     (description "Pandoc is a powerful utility to transform various
1007 input formats into a wide range of output formats.  To alter the
1008 exported output document, Pandoc allows the usage of filters, which
1009 are pipes that read a JSON serialization of the Pandoc AST from stdin,
1010 transform it in some way, and write it to stdout.  It allows therefore
1011 to alter the processing of Pandoc's supported input formats, for
1012 instance one can add new syntax elements to markdown, etc.
1013
1014 This package provides Python bindings.")
1015     (license license:bsd-3)))