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 ;;; Copyright © 2019 Mădălin Ionel Patrașcu <madalinionel.patrascu@mdc-berlin.de>
  21 ;;;
  22 ;;; This file is part of GNU Guix.
  23 ;;;
  24 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  25 ;;; under the terms of the GNU General Public License as published by
  26 ;;; the Free Software Foundation; either version 3 of the License, or (at
  27 ;;; your option) any later version.
  28 ;;;
  29 ;;; GNU Guix is distributed in the hope that it will be useful, but
  30 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  31 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  32 ;;; GNU General Public License for more details.
  33 ;;;
  34 ;;; You should have received a copy of the GNU General Public License
  35 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  36
  37 (define-module (gnu packages textutils)
  38   #:use-module ((guix licenses) #:prefix license:)
  39   #:use-module (guix packages)
  40   #:use-module (guix download)
  41   #:use-module (guix git-download)
  42   #:use-module (guix build-system ant)
  43   #:use-module (guix build-system gnu)
  44   #:use-module (guix build-system go)
  45   #:use-module (guix build-system cmake)
  46   #:use-module (guix build-system python)
  47   #:use-module (gnu packages)
  48   #:use-module (gnu packages autotools)
  49   #:use-module (gnu packages base)
  50   #:use-module (gnu packages compression)
  51   #:use-module (gnu packages gcc)
  52   #:use-module (gnu packages gettext)
  53   #:use-module (gnu packages java)
  54   #:use-module (gnu packages ncurses)
  55   #:use-module (gnu packages perl)
  56   #:use-module (gnu packages pkg-config)
  57   #:use-module (gnu packages python)
  58   #:use-module (gnu packages python-xyz)
  59   #:use-module (gnu packages readline)
  60   #:use-module (gnu packages slang)
  61   #:use-module (gnu packages web))
  62
  63 (define-public dos2unix
  64   (package
  65     (name "dos2unix")
  66     (version "7.4.1")
  67     (source
  68      (origin
  69        (method url-fetch)
  70        (uri (string-append "https://waterlan.home.xs4all.nl/dos2unix/"
  71                            "dos2unix-" version ".tar.gz"))
  72        (sha256
  73         (base32 "08w6yywzirsxq8bh87jycvvw922ybhc2l426j2iqzliyn1h8mm8w"))))
  74     (build-system gnu-build-system)
  75     (arguments
  76      '(#:make-flags (list "CC=gcc"
  77                           (string-append "prefix=" (assoc-ref %outputs "out")))
  78        #:phases
  79        (modify-phases %standard-phases
  80          (delete 'configure)))) ; no configure script
  81     (native-inputs
  82      `(("gettext" ,gettext-minimal)
  83        ("perl" ,perl)))
  84     (home-page "https://waterlan.home.xs4all.nl/dos2unix.html")
  85     (synopsis "DOS/Mac to Unix and vice versa text file format converter")
  86     (description
  87      "dos2unix is a tool to convert line breaks in a text file from Unix format
  88 to DOS format and vice versa.")
  89     (license license:bsd-2)))
  90
  91 (define-public recode
  92   (package
  93     (name "recode")
  94     (version "3.7.6")
  95     (source
  96      (origin
  97        (method url-fetch)
  98        (uri (string-append "https://github.com/rrthomas/recode/releases/"
  99                            "download/v" version "/recode-" version ".tar.gz"))
 100        (sha256
 101         (base32 "0m59sd1ca0zw1aydpc3m8sw03nc885knmccqryg7byzmqs585ia6"))))
 102     (build-system gnu-build-system)
 103     (native-inputs
 104      `(("python" ,python)
 105        ("python-cython" ,python-cython)))
 106     (home-page "https://github.com/rrthomas/recode")
 107     (synopsis "Text encoding converter")
 108     (description "The Recode library converts files between character sets and
 109 usages.  It recognises or produces over 200 different character sets (or about
 110 300 if combined with an iconv library) and transliterates files between almost
 111 any pair.  When exact transliteration are not possible, it gets rid of
 112 offending characters or falls back on approximations.  The recode program is a
 113 handy front-end to the library.")
 114     (license license:gpl3+)))
 115
 116 (define-public enca
 117   (package
 118     (name "enca")
 119     (version "1.19")
 120     (source
 121      (origin
 122        (method git-fetch)
 123        (uri (git-reference
 124               (url "https://github.com/nijel/enca")
 125               (commit version)))
 126        (file-name (git-file-name name version))
 127        (sha256
 128         (base32 "19q7cwwxmmk5j9438bsqdpjvdjawsd3zmw1zyqgi7s4m0rasr3ah"))))
 129     (build-system gnu-build-system)
 130     ;; enca-1.19 tests fail with recent recode.
 131     ;(inputs `(("recode" ,recode)))
 132     (home-page "https://github.com/nijel/enca")
 133     (synopsis "Text encoding detection tool")
 134     (description "Enca (Extremely Naive Charset Analyser) consists of libenca,
 135 an encoding detection library, and enca, a command line frontend, integrating
 136 libenca and several charset conversion libraries and tools.")
 137     (license license:gpl2)))
 138
 139 (define-public utf8proc
 140   (package
 141     (name "utf8proc")
 142     (version "2.4.0")
 143     (source
 144      (origin
 145        (method git-fetch)
 146        (uri (git-reference
 147              (url "https://github.com/JuliaStrings/utf8proc")
 148              (commit (string-append "v" version))))
 149        (file-name (git-file-name name version))
 150        (sha256
 151         (base32 "1i42hqwc8znqii9brangwkxk5cyc2lk95ip405fg88zr7z2ncr34"))))
 152     (build-system gnu-build-system)
 153     (native-inputs           ;test data that is otherwise downloaded with curl
 154      `(("NormalizationTest.txt"
 155         ,(origin
 156            (method url-fetch)
 157            (uri (string-append "https://www.unicode.org/Public/12.1.0/ucd/"
 158                                "NormalizationTest.txt"))
 159            (sha256
 160             (base32 "0hb97k9xv1lr847hwz0719ksqy39s47xw6k01dgs1368jdibvawc"))))
 161        ("GraphemeBreakTest.txt"
 162         ,(origin
 163            (method url-fetch)
 164            (uri (string-append "https://www.unicode.org/Public/12.1.0/ucd/"
 165                                "auxiliary/GraphemeBreakTest.txt"))
 166            (sha256
 167             (base32 "0qc90ppmrwfn3y9cdn8jcjrn7qpdf0fhxkwh945yp4rvh37mbgcm"))))
 168
 169        ;; For tests.
 170        ("perl" ,perl)))
 171     (arguments
 172      '(#:make-flags (list "CC=gcc"
 173                           (string-append "prefix=" (assoc-ref %outputs "out")))
 174        #:phases
 175        (modify-phases %standard-phases
 176          (delete 'configure)
 177          (add-before 'check 'check-data
 178            (lambda* (#:key inputs #:allow-other-keys)
 179              (for-each (lambda (i)
 180                          (copy-file (assoc-ref inputs i)
 181                                     (string-append "data/" i)))
 182                        '("NormalizationTest.txt" "GraphemeBreakTest.txt"))
 183              (substitute* "data/GraphemeBreakTest.txt"
 184                (("÷") "/")
 185                (("×") "+"))
 186              #t)))))
 187     (home-page "https://juliastrings.github.io/utf8proc/")
 188     (synopsis "C library for processing UTF-8 Unicode data")
 189     (description "utf8proc is a small C library that provides Unicode
 190 normalization, case-folding, and other operations for data in the UTF-8
 191 encoding, supporting Unicode version 9.0.0.")
 192     (license license:expat)))
 193
 194 (define-public libconfuse
 195   (package
 196     (name "libconfuse")
 197     (version "3.2.2")
 198     (source
 199      (origin
 200        (method url-fetch)
 201        (uri (string-append "https://github.com/martinh/libconfuse/"
 202                            "releases/download/v" version
 203                            "/confuse-" version ".tar.xz"))
 204        (sha256
 205         (base32 "02r1mmzik2m0iigbc2da3y754vj24i18r3ml5p2wzs027mjhn959"))))
 206     (build-system gnu-build-system)
 207     (home-page "https://github.com/martinh/libconfuse")
 208     (synopsis "Configuration file parser library")
 209     (description "libconfuse is a configuration file parser library.  It
 210 supports sections and (lists of) values (strings, integers, floats, booleans
 211 or other sections), as well as some other features (such as
 212 single/double-quoted strings, environment variable expansion, functions and
 213 nested include statements).")
 214     (license license:isc)))
 215
 216 (define-public libgtextutils
 217   (package
 218     (name "libgtextutils")
 219     (version "0.7")
 220     (source
 221      (origin
 222        (method url-fetch)
 223        (uri (string-append
 224              "https://github.com/agordon/libgtextutils/releases/download/"
 225              version "/libgtextutils-" version ".tar.gz"))
 226        (sha256
 227         (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
 228     (build-system gnu-build-system)
 229     (arguments
 230      '(#:phases
 231        (modify-phases %standard-phases
 232          (replace 'bootstrap
 233            (lambda _ (invoke "sh" "reconf"))))))
 234     (native-inputs
 235      `(("autoconf" ,autoconf)
 236        ("automake" ,automake)
 237        ("gcc" ,gcc-5) ;; doesn't build with later versions
 238        ("libtool" ,libtool)))
 239     (home-page "https://github.com/agordon/libgtextutils")
 240     (synopsis "Gordon's text utils library")
 241     (description
 242      "libgtextutils is a text utilities library used by the fastx toolkit from
 243 the Hannon Lab.")
 244     (license license:agpl3+)))
 245
 246 (define-public cityhash
 247   (let ((commit "8af9b8c"))
 248     (package
 249       (name "cityhash")
 250       (version (string-append "1.1-2." commit))
 251       (source (origin
 252                 (method git-fetch)
 253                 (uri (git-reference
 254                       (url "https://github.com/google/cityhash.git")
 255                       (commit commit)))
 256                 (file-name (string-append name "-" version ".tar.gz"))
 257                 (sha256
 258                  (base32
 259                   "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
 260       (build-system gnu-build-system)
 261       (arguments
 262        '(#:make-flags (list "CXXFLAGS=-g -O3")
 263          #:phases
 264          (modify-phases %standard-phases
 265            ;; citycrc is not installed by default but is used by some
 266            ;; programs.
 267            (add-after 'install 'install-citycrc
 268              (lambda* (#:key outputs #:allow-other-keys)
 269                (let* ((out (assoc-ref outputs "out"))
 270                       (include (string-append out "/include")))
 271                  (install-file "src/citycrc.h" include))
 272                #t)))))
 273       (home-page "https://github.com/google/cityhash")
 274       (synopsis "C++ hash functions for strings")
 275       (description
 276        "CityHash provides hash functions for strings.  The functions mix the
 277 input bits thoroughly but are not suitable for cryptography.")
 278       (license license:expat))))
 279
 280 (define-public ustr
 281   (package
 282     (name "ustr")
 283     (version "1.0.4")
 284     (source (origin
 285               (method url-fetch)
 286               (uri (string-append "http://www.and.org/ustr/" version
 287                                   "/ustr-" version ".tar.bz2"))
 288               (sha256
 289                (base32
 290                 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
 291               (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
 292     (build-system gnu-build-system)
 293     (arguments
 294      `(#:make-flags
 295        (list "CC=gcc"
 296              "HIDE="
 297              ;; Override "/sbin/ldconfig" with "echo" because we don't need
 298              ;; "ldconfig".
 299              "LDCONFIG=echo"
 300              (string-append "prefix=" (assoc-ref %outputs "out"))
 301              "all-shared")
 302        #:phases
 303        (modify-phases %standard-phases
 304          (add-after 'unpack 'disable-check-for-stdint
 305            (lambda _
 306              ;; Of course we have stdint.h, just not in /usr/include
 307              (substitute* '("Makefile"
 308                             "ustr-import.in")
 309                (("-f \"/usr/include/stdint.h\"") "-z \"\""))
 310              #t))
 311          ;; No configure script
 312          (delete 'configure))))
 313     (home-page "http://www.and.org/ustr/")
 314     (synopsis "String library with very low memory overhead")
 315     (description
 316      "Ustr is a string library for C with very low memory overhead.")
 317     ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
 318     ;; LGPL, etc. ... if you need another license to help compatibility, just
 319     ;; ask for it.  It's basically public domain, without all the legal
 320     ;; problems for everyone that trying to make something public domain
 321     ;; entails."
 322     (license license:public-domain)))
 323
 324 (define-public ascii2binary
 325   (package
 326     (name "ascii2binary")
 327     (version "2.14")
 328     (source
 329      (origin
 330        (method url-fetch)
 331        (uri (string-append "http://billposer.org/Software/Downloads/"
 332                            "ascii2binary-" version ".tar.bz2"))
 333        (sha256
 334         (base32 "0dc9fxcdmppbs9s06jvq61zbk552laxps0xyk098gj41697ihd96"))))
 335     (build-system gnu-build-system)
 336     (native-inputs
 337      `(("gettext" ,gettext-minimal)))
 338     (home-page "https://billposer.org/Software/a2b.html")
 339     (synopsis "Convert between ASCII, hexadecimal and binary representations")
 340     (description "The two programs are useful for generating test data, for
 341 inspecting binary files, and for interfacing programs that generate textual
 342 output to programs that require binary input and conversely.  They can also be
 343 useful when it is desired to reformat numbers.
 344
 345 @itemize
 346
 347 @item @command{ascii2binary} reads input consisting of ascii or hexadecimal
 348    representation numbers separated by whitespace and produces as output
 349    the binary equivalents.  The type and precision of the binary output
 350    is selected using command line flags.
 351
 352 @item @command{binary2ascii} reads input consisting of binary numbers
 353    and converts them to their ascii or hexadecimal representation.
 354    Command line flags specify the type and size of the binary numbers
 355    and provide control over the format of the output.
 356    Unsigned integers may be written out in binary, octal, decimal,
 357    or hexadecimal.
 358
 359    Signed integers may be written out only in binary or decimal.  Floating
 360    point numbers may be written out only decimal, either in standard or
 361    scientific notation.  (If you want to examine the binary representation
 362    of floating point numbers, just treat the input as a sequence of unsigned
 363    characters.)
 364
 365 @end itemize")
 366     (license license:gpl3)))
 367
 368 (define-public uniutils
 369   (package
 370     (name "uniutils")
 371     (version "2.27")
 372     (source
 373      (origin
 374        (method url-fetch)
 375        (uri (string-append "http://billposer.org/Software/Downloads/"
 376                            "uniutils-" version ".tar.bz2"))
 377        (sha256
 378         (base32 "19w1510w87gx7n4qy3zsb0m467a4rn5scvh4ajajg7jh6x5xri08"))))
 379     (build-system gnu-build-system)
 380     (arguments
 381      '(#:configure-flags '("--disable-dependency-tracking")
 382        #:phases
 383        (modify-phases %standard-phases
 384          (add-after 'build 'fix-paths
 385            (lambda* (#:key outputs inputs #:allow-other-keys)
 386              (let ((out (assoc-ref outputs "out"))
 387                    (a2b (assoc-ref inputs "ascii2binary"))
 388                    (iconv (assoc-ref inputs "libiconv")))
 389                (substitute* "utf8lookup"
 390                  (("^ascii2binary ") (string-append a2b "/bin/ascii2binary "))
 391                  (("^uniname ") (string-append out "/bin/uniname "))
 392                  (("^iconv ") (string-append iconv "/bin/iconv ")))
 393              #t))))))
 394     (inputs
 395      `(("ascii2binary" ,ascii2binary)
 396        ("libiconv" ,libiconv)))
 397     (home-page "https://billposer.org/Software/unidesc.html")
 398     (synopsis "Find out what is in a Unicode file")
 399     (description "Useful tools when working with Unicode files when one
 400 doesn't know the writing system, doesn't have the necessary font, needs to
 401 inspect invisible characters, needs to find out whether characters have been
 402 combined or in what order they occur, or needs statistics on which characters
 403 occur.
 404
 405 @itemize
 406
 407 @item @command{uniname} defaults to printing the character offset of each
 408 character, its byte offset, its hex code value, its encoding, the glyph
 409 itself, and its name.  It may also be used to validate UTF-8 input.
 410
 411 @item @command{unidesc} reports the character ranges to which different
 412 portions of the text belong.   It can also be used to identify Unicode encodings
 413 (e.g. UTF-16be) flagged by magic numbers.
 414
 415 @item @command{unihist} generates a histogram of the characters in its input.
 416
 417 @item @command{ExplicateUTF8} is intended for debugging or for learning about
 418 Unicode.  It determines and explains the validity of a sequence of bytes as a
 419 UTF8 encoding.
 420
 421 @item @command{utf8lookup} provides a handy way to look up Unicode characters
 422 from the command line.
 423
 424 @item @command{unireverse} reverse each line of UTF-8 input
 425 character-by-character.
 426
 427 @end itemize")
 428     (license license:gpl3)))
 429
 430 (define-public libconfig
 431   (package
 432     (name "libconfig")
 433     (version "1.7.2")
 434     (home-page "https://hyperrealm.github.io/libconfig/")
 435     (source (origin
 436               (method url-fetch)
 437               (uri (string-append home-page "/dist/libconfig-"
 438                                   version ".tar.gz"))
 439               (sha256
 440                (base32
 441                 "1ngs2qx3cx5cbwinc5mvadly0b5n7s86zsc68c404czzfff7lg3w"))))
 442     (build-system gnu-build-system)
 443     (synopsis "C/C++ configuration file library")
 444     (description
 445      "Libconfig is a simple library for manipulating structured configuration
 446 files.  This file format is more compact and more readable than XML.  And
 447 unlike XML, it is type-aware, so it is not necessary to do string parsing in
 448 application code.")
 449     (license license:lgpl2.1+)))
 450
 451 (define-public pfff
 452   (package
 453     (name "pfff")
 454     (version "1.0")
 455     (source (origin
 456               (method git-fetch)
 457               (uri (git-reference
 458                      (url "https://github.com/pfff/pfff")
 459                      (commit (string-append "v" version))))
 460               (file-name (git-file-name name version))
 461               (sha256
 462                (base32
 463                 "1nxkfm7zliq3rmr7yp871sppwfnz71iz364m2sgazny71pzykggc"))))
 464     (build-system cmake-build-system)
 465     (home-page "http://biit.cs.ut.ee/pfff/")
 466     (synopsis "Probabilistic fast file fingerprinting tool")
 467     (description
 468      "pfff is a tool for calculating a compact digital fingerprint of a file
 469 by sampling randomly from the file instead of reading it in full.
 470 Consequently, the computation has a flat performance characteristic,
 471 correlated with data variation rather than file size.  pfff can be as reliable
 472 as existing hashing techniques, with provably negligible risk of collisions.")
 473     (license license:bsd-3)))
 474
 475 (define-public oniguruma
 476   (package
 477     (name "oniguruma")
 478     (version "6.9.4")
 479     (source (origin
 480               (method url-fetch)
 481               (uri (string-append "https://github.com/kkos/"
 482                                   "oniguruma/releases/download/v" version
 483                                   "/onig-" version ".tar.gz"))
 484               (sha256
 485                (base32
 486                 "0lvd1rpp49i0k1icblb0i76lj2cwmhf1c5p1jdz2m6g0ywpx4sa6"))))
 487     (build-system gnu-build-system)
 488     (home-page "https://github.com/kkos/oniguruma")
 489     (synopsis "Regular expression library")
 490     (description "Oniguruma is a regular expressions library.  The special
 491 characteristic of this library is that different character encoding for every
 492 regular expression object can be specified.")
 493     (license license:bsd-2)))
 494
 495 (define-public antiword
 496   (package
 497     (name "antiword")
 498     (version "0.37")
 499     (source (origin
 500               (method url-fetch)
 501               (uri (string-append "http://www.winfield.demon.nl/linux"
 502                                   "/antiword-" version ".tar.gz"))
 503               (sha256
 504                (base32
 505                 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
 506               (patches (search-patches "antiword-CVE-2014-8123.patch"))))
 507     (build-system gnu-build-system)
 508     (arguments
 509      `(#:tests? #f ; There are no tests
 510        #:make-flags
 511        (list "-f" "Makefile.Linux"
 512              (string-append "GLOBAL_INSTALL_DIR="
 513                             (assoc-ref %outputs "out") "/bin")
 514              (string-append "GLOBAL_RESOURCES_DIR="
 515                             (assoc-ref %outputs "out") "/share/antiword"))
 516        #:phases
 517        (modify-phases %standard-phases
 518          (replace 'configure
 519            (lambda* (#:key outputs #:allow-other-keys)
 520              ;; Ensure that mapping files can be found in the actual package
 521              ;; data directory.
 522              (substitute* "antiword.h"
 523                (("/usr/share/antiword")
 524                 (string-append (assoc-ref outputs "out") "/share/antiword")))
 525              #t))
 526          (replace 'install
 527            (lambda* (#:key make-flags #:allow-other-keys)
 528              (apply invoke "make" `("global_install" ,@make-flags)))))))
 529     (home-page "http://www.winfield.demon.nl/")
 530     (synopsis "Microsoft Word document reader")
 531     (description "Antiword is an application for displaying Microsoft Word
 532 documents.  It can also convert the document to PostScript or XML.  Only
 533 documents made by MS Word version 2 and version 6 or later are supported.  The
 534 name comes from: \"The antidote against people who send Microsoft Word files
 535 to everybody, because they believe that everybody runs Windows and therefore
 536 runs Word\".")
 537     (license license:gpl2+)))
 538
 539 (define-public catdoc
 540   (package
 541     (name "catdoc")
 542     (version "0.95")
 543     (source (origin
 544               (method url-fetch)
 545               (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
 546                                   "catdoc-" version ".tar.gz"))
 547               (patches (search-patches "catdoc-CVE-2017-11110.patch"))
 548               (sha256
 549                (base32
 550                 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
 551     (build-system gnu-build-system)
 552     ;; TODO: Also build `wordview` which requires `tk` – make a separate
 553     ;; package for this.
 554     (arguments
 555      '(#:tests? #f ; There are no tests
 556        #:configure-flags '("--disable-wordview")
 557        #:phases
 558        (modify-phases %standard-phases
 559          (add-before 'install 'fix-install
 560            (lambda* (#:key outputs #:allow-other-keys)
 561              (let ((out (assoc-ref outputs "out")))
 562                (mkdir-p (string-append out "/share/man/man1"))))))))
 563     (home-page "http://www.wagner.pp.ru/~vitus/software/catdoc/")
 564     (synopsis "MS-Word to TeX or plain text converter")
 565     (description "@command{catdoc} extracts text from MS-Word files, trying to
 566 preserve as many special printable characters as possible.  It supports
 567 everything up to Word-97. Also supported are MS Write documents and RTF files.
 568
 569 @command{catdoc} does not preserve complex word formatting, but it can
 570 translate some non-ASCII characters into TeX escape codes.  It's goal is to
 571 extract plain text and allow you to read it and, probably, reformat with TeX,
 572 according to TeXnical rules.
 573
 574 This package also provides @command{xls2csv}, which extracts data from Excel
 575 spreadsheets and outputs it in comma-separated-value format, and
 576 @command{catppt}, which extracts data from PowerPoint presentations.")
 577     (license license:gpl2+)))
 578
 579 (define-public utfcpp
 580   (package
 581     (name "utfcpp")
 582     (version "2.3.5")
 583     (source (origin
 584               (method git-fetch)
 585               (uri (git-reference
 586                      (url "https://github.com/nemtrif/utfcpp")
 587                      (commit (string-append "v" version))))
 588               (file-name (git-file-name name version))
 589               (sha256
 590                (base32
 591                 "1gr98d826z6wa58r1s5i7rz7q2x3r31v7zj0pjjlrc7gfxwklr4s"))))
 592     (build-system cmake-build-system)
 593     (arguments
 594      `(#:out-of-source? #f
 595        #:phases
 596        (modify-phases %standard-phases
 597          (replace 'install              ; no install target
 598            (lambda* (#:key outputs #:allow-other-keys)
 599              (let* ((out (assoc-ref outputs "out"))
 600                     (include (string-append out "/include"))
 601                     (doc (string-append out "/share/doc/" ,name)))
 602                (copy-recursively "source" include)
 603                (install-file "README.md" doc)
 604                #t))))))
 605     (home-page "https://github.com/nemtrif/utfcpp")
 606     (synopsis "Portable C++ library for handling UTF-8")
 607     (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
 608 in a portable way.")
 609     (license license:boost1.0)))
 610
 611 (define-public dbacl
 612   (package
 613     (name "dbacl")
 614     (version "1.14")
 615     (source
 616      (origin
 617        (method url-fetch)
 618        (uri (string-append "http://www.lbreyer.com/gpl/"
 619                            "dbacl-" version ".tar.gz"))
 620        (sha256
 621         (base32
 622          "0224g6x71hyvy7jikfxmgcwww1r5lvk0jx36cva319cb9nmrbrq7"))))
 623     (build-system gnu-build-system)
 624     (arguments
 625      `(#:make-flags
 626        (list
 627         (string-append "-I" (assoc-ref %build-inputs "slang")
 628                        "/include/slang")
 629         (string-append "-I" (assoc-ref %build-inputs "ncurses")
 630                        "/include/ncurses"))
 631        #:phases
 632        (modify-phases %standard-phases
 633          (add-after 'unpack 'delete-sample6-and-japanese
 634            (lambda _
 635              (substitute* "doc/Makefile.am"
 636                (("sample6.txt") "")
 637                (("japanese.txt") ""))
 638              (delete-file "doc/sample6.txt")
 639              (delete-file "doc/japanese.txt")
 640              (substitute* (list "src/tests/Makefile.am"
 641                                 "src/tests/Makefile.in")
 642                (("dbacl-jap.shin") "")
 643                (("dbacl-jap.sh") ""))
 644              #t))
 645          (add-after 'unpack 'delete-test
 646            ;; See comments about the license.
 647            (lambda _
 648              (delete-file "src/tests/dbacl-jap.shin")
 649              #t))
 650          (add-after 'delete-sample6-and-japanese 'autoreconf
 651            (lambda _
 652              (invoke "autoreconf" "-vif")
 653              #t))
 654          (add-after 'unpack 'fix-test-files
 655            (lambda* (#:key inputs outputs #:allow-other-keys)
 656              (let* ((out (assoc-ref outputs "out"))
 657                     (bin (string-append out "/bin")))
 658                (substitute* (find-files "src/tests/" "\\.shin$")
 659                  (("PATH=/bin:/usr/bin")
 660                   "#PATH=/bin:/usr/bin")
 661                  (("diff") (string-append (which "diff")))
 662                  (("tr") (string-append (which "tr"))))
 663                #t))))))
 664     (inputs
 665      `(("ncurses" ,ncurses)
 666        ("perl" ,perl)
 667        ("readline" ,readline)
 668        ("slang" ,slang)))
 669     (native-inputs
 670      `(("libtool" ,libtool)
 671        ("autoconf" ,autoconf)
 672        ("automake" ,automake)
 673        ("pkg-config" ,pkg-config)))
 674     (home-page "https://www.lbreyer.com/dbacl.html")
 675     (synopsis "Bayesian text and email classifier")
 676     (description
 677      "dbacl is a fast Bayesian text and email classifier.  It builds a variety
 678 of language models using maximum entropy (minimum divergence) principles, and
 679 these can then be used to categorize input data automatically among multiple
 680 categories.")
 681     ;; The software is licensed as GPLv3 or later, but
 682     ;; includes various sample texts in the doc dir:
 683     ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
 684     ;;   by Mark Twain.
 685     ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
 686     ;; - sample6.txt is a forwarded email, copyright unknown.
 687     ;;   Guix does exclude sample6.txt.
 688     ;; - japanese.txt is a Japanese unoffical translation of the
 689     ;;   GNU General Public License, (c) by the Free Software Foundation.
 690     ;;   Guix excludes this file.
 691     (license (list license:gpl3+ license:public-domain))))
 692
 693 (define-public dotconf
 694   (package
 695     (name "dotconf")
 696     (version "1.3")
 697     (source (origin
 698               (method git-fetch)
 699               (uri (git-reference
 700                     (url "https://github.com/williamh/dotconf.git")
 701                     (commit (string-append "v" version))))
 702               (file-name (git-file-name name version))
 703               (sha256
 704                (base32
 705                 "1sc95hw5k2xagpafny0v35filmcn05k1ds5ghkldfpf6xw4hakp7"))))
 706     (build-system gnu-build-system)
 707     (arguments `(#:tests? #f))  ; FIXME maketest.sh does not work.
 708     (native-inputs
 709      `(("autoconf" ,autoconf)
 710        ("automake" ,automake)
 711        ("libtool" ,libtool)))
 712     (home-page "https://github.com/williamh/dotconf")
 713     (synopsis "Configuration file parser library")
 714     (description
 715      "C library for creating and parsing configuration files.")
 716     (license (list license:lgpl2.1         ; Main distribution.
 717                    license:asl1.1))))      ; src/readdir.{c,h}
 718
 719 (define-public java-rsyntaxtextarea
 720   (package
 721     (name "java-rsyntaxtextarea")
 722     (version "2.6.1")
 723     (source (origin
 724               (method url-fetch)
 725               (uri (string-append "https://github.com/bobbylight/"
 726                                   "RSyntaxTextArea/archive/"
 727                                   version ".tar.gz"))
 728               (file-name (string-append name "-" version ".tar.gz"))
 729               (sha256
 730                (base32
 731                 "0c5mqg2klj5rvf8fhycrli8rf6s37l9p7a8knw9gpp65r1c120q2"))))
 732     (build-system ant-build-system)
 733     (arguments
 734      `(;; FIXME: some tests fail because locale resources cannot be found.
 735        ;; Even when I add them to the class path,
 736        ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
 737        #:tests? #f
 738        #:jar-name "rsyntaxtextarea.jar"))
 739     (native-inputs
 740      `(("java-junit" ,java-junit)
 741        ("java-hamcrest-core" ,java-hamcrest-core)))
 742     (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
 743     (synopsis "Syntax highlighting text component for Java Swing")
 744     (description "RSyntaxTextArea is a syntax highlighting, code folding text
 745 component for Java Swing.  It extends @code{JTextComponent} so it integrates
 746 completely with the standard @code{javax.swing.text} package.  It is fast and
 747 efficient, and can be used in any application that needs to edit or view
 748 source code.")
 749     (license license:bsd-3)))
 750
 751 ;; We use the sources from git instead of the tarball from pypi, because the
 752 ;; latter does not include the Cython source file from which bycython.cpp is
 753 ;; generated.
 754 (define-public python-editdistance
 755   (let ((commit "3ea84a7dd3258c76aa3be851ef3d50e59c886846")
 756         (revision "1"))
 757     (package
 758       (name "python-editdistance")
 759       (version (string-append "0.3.1-" revision "." (string-take commit 7)))
 760       (source
 761        (origin
 762          (method git-fetch)
 763          (uri (git-reference
 764                (url "https://github.com/aflc/editdistance.git")
 765                (commit commit)))
 766          (file-name (git-file-name name version))
 767          (sha256
 768           (base32
 769            "1l43svsv12crvzphrgi6x435z6xg8m086c64armp8wzb4l8ccm7g"))))
 770       (build-system python-build-system)
 771       (arguments
 772        `(#:phases
 773          (modify-phases %standard-phases
 774            (add-after 'unpack 'build-cython-code
 775              (lambda _
 776                (with-directory-excursion "editdistance"
 777                  (delete-file "bycython.cpp")
 778                  (invoke "cython" "--cplus" "bycython.pyx")))))))
 779       (native-inputs
 780        `(("python-cython" ,python-cython)))
 781       (home-page "https://www.github.com/aflc/editdistance")
 782       (synopsis "Fast implementation of the edit distance (Levenshtein distance)")
 783       (description
 784        "This library simply implements Levenshtein distance algorithm with C++
 785 and Cython.")
 786       (license license:expat))))
 787
 788 (define-public go-github.com-mattn-go-runewidth
 789   (let ((commit "703b5e6b11ae25aeb2af9ebb5d5fdf8fa2575211")
 790         (version "0.0.4")
 791         (revision "1"))
 792     (package
 793       (name "go-github.com-mattn-go-runewidth")
 794       (version (git-version version revision commit))
 795       (source
 796        (origin
 797          (method git-fetch)
 798          (uri (git-reference
 799                (url "https://github.com/mattn/runewidth")
 800                (commit commit)))
 801          (file-name (git-file-name name version))
 802          (sha256
 803           (base32
 804            "0znpyz71gajx3g0j2zp63nhjj2c07g16885vxv4ykwnrfmzbgk4w"))))
 805       (build-system go-build-system)
 806       (arguments
 807        '(#:import-path "github.com/mattn/go-runewidth"))
 808       (synopsis "@code{runewidth} provides Go functions to work with string widths")
 809       (description
 810        "The @code{runewidth} library provides Go functions for padding,
 811 measuring and checking the width of strings, with support for East Asian
 812 text.")
 813       (home-page "https://github.com/mattn/runewidth")
 814       (license license:expat))))
 815
 816 (define-public docx2txt
 817   (package
 818     (name "docx2txt")
 819     (version "1.4")
 820     (source (origin
 821               (method url-fetch)
 822               (uri (string-append
 823                     "mirror://sourceforge/docx2txt/docx2txt/v"
 824                     version "/docx2txt-" version ".tgz"))
 825               (sha256
 826                (base32
 827                 "06vdikjvpj6qdb41d8wzfnyj44jpnknmlgbhbr1w215420lpb5xj"))))
 828     (build-system gnu-build-system)
 829     (inputs
 830      `(("unzip" ,unzip)
 831        ("perl" ,perl)))
 832     (arguments
 833      `(#:tests? #f                      ; No tests.
 834        #:make-flags (list (string-append "BINDIR="
 835                                          (assoc-ref %outputs "out") "/bin")
 836                           (string-append "CONFIGDIR="
 837                                          (assoc-ref %outputs "out") "/etc")
 838                           ;; Makefile seems to be a bit dumb at guessing.
 839                           (string-append "INSTALL=install")
 840                           (string-append "PERL=perl"))
 841        #:phases
 842        (modify-phases %standard-phases
 843          (delete 'configure)
 844          (add-after 'install 'fix-install
 845            (lambda* (#:key outputs inputs #:allow-other-keys)
 846              (let* ((out (assoc-ref outputs "out"))
 847                     (bin (string-append out "/bin"))
 848                     (config (string-append out "/etc/docx2txt.config"))
 849                     (unzip (assoc-ref inputs "unzip")))
 850                ;; According to INSTALL, the .sh wrapper can be skipped.
 851                (delete-file (string-append bin "/docx2txt.sh"))
 852                (rename-file (string-append bin "/docx2txt.pl")
 853                             (string-append bin "/docx2txt"))
 854                (substitute* config
 855                  (("config_unzip         => '/usr/bin/unzip',")
 856                   (string-append "config_unzip         => '"
 857                                  unzip
 858                                  "/bin/unzip',")))
 859                ;; Makefile is wrong.
 860                (chmod config #o644)
 861                #t))))))
 862     (synopsis "Recover text from @file{.docx} files, with good formatting")
 863     (description
 864      "@command{docx2txt} is a Perl based command line utility to convert
 865 Microsoft Office @file{.docx} documents to equivalent text documents.  Latest
 866 version supports following features during text extraction.
 867
 868 @itemize
 869 @item Character conversions; currency characters are converted to respective
 870 names like Euro.
 871 @item Capitalisation of text blocks.
 872 @item Center and right justification of text fitting in a line of
 873 (configurable) 80 columns.
 874 @item Horizontal ruler, line breaks, paragraphs separation, tabs.
 875 @item Indicating hyperlinked text along with the hyperlink (configurable).
 876 @item Handling (bullet, decimal, letter, roman) lists along with (attempt at)
 877 indentation.
 878 @end itemize\n")
 879     (home-page "http://docx2txt.sourceforge.net")
 880     (license license:gpl3+)))
 881
 882 (define-public odt2txt
 883   (package
 884     (name "odt2txt")
 885     (version "0.5")
 886     (source
 887       (origin
 888         (method git-fetch)
 889         (uri (git-reference
 890                (url "https://github.com/dstosberg/odt2txt/")
 891                (commit (string-append "v" version))))
 892         (file-name (git-file-name name version))
 893         (sha256
 894          (base32
 895           "0im3kzvhxkjlx57w6h13mc9584c74ma1dyymgvpq2y61av3gc35v"))))
 896     (build-system gnu-build-system)
 897     (arguments
 898      `(#:tests? #f ; no make check
 899        #:make-flags (list "CC=gcc"
 900                           (string-append "DESTDIR=" (assoc-ref %outputs "out")))
 901        #:phases
 902        (modify-phases %standard-phases
 903          ;; no configure script
 904          (delete 'configure))))
 905     (inputs
 906      `(("zlib" ,zlib)))
 907     (home-page "https://github.com/dstosberg/odt2txt/")
 908     (synopsis "Converter from OpenDocument Text to plain text")
 909     (description "odt2txt is a command-line tool which extracts the text out
 910 of OpenDocument Texts, as produced by OpenOffice.org, KOffice, StarOffice and
 911 others.
 912
 913 odt2txt can also extract text from some file formats similar to OpenDocument
 914 Text, such as OpenOffice.org XML (*.sxw), which was used by OpenOffice.org
 915 version 1.x and older StarOffice versions.  To a lesser extent, odt2txt may be
 916 useful to extract content from OpenDocument spreadsheets (*.ods) and
 917 OpenDocument presentations (*.odp).")
 918     (license license:gpl2)))
 919
 920 (define-public opencc
 921   (package
 922     (name "opencc")
 923     (version "1.0.5")
 924     (source
 925      (origin
 926        (method git-fetch)
 927        (uri (git-reference
 928               (url "https://github.com/BYVoid/OpenCC")
 929               (commit (string-append "ver." version))))
 930        (file-name (git-file-name name version))
 931        (sha256
 932         (base32
 933          "1pv5md225qwhbn8ql932zdg6gh1qlx3paiajaks8gfsa07yzvhr4"))
 934        (modules '((guix build utils)))
 935        (snippet
 936         '(begin
 937            ;; TODO: Unbundle tclap, darts-clone, gtest
 938            (delete-file-recursively "deps/rapidjson-0.11") #t))))
 939     (build-system cmake-build-system)
 940     (arguments
 941      '(#:phases
 942        (modify-phases %standard-phases
 943          (add-after 'unpack 'patch-3rd-party-references
 944            (lambda* (#:key inputs #:allow-other-keys)
 945              (let ((rapidjson (assoc-ref inputs "rapidjson")))
 946                (substitute* "src/CMakeLists.txt"
 947                  (("../deps/rapidjson-0.11")
 948                   (string-append rapidjson "/include/rapidjson")))
 949              #t))))))
 950     (native-inputs
 951      `(("python" ,python-wrapper)
 952        ("rapidjson" ,rapidjson)))
 953     (home-page "https://github.com/BYVoid/OpenCC")
 954     (synopsis "Convert between Traditional Chinese and Simplified Chinese")
 955     (description "Open Chinese Convert (OpenCC) converts between Traditional
 956 Chinese and Simplified Chinese, supporting character-level conversion,
 957 phrase-level conversion, variant conversion, and regional idioms among
 958 Mainland China, Taiwan, and Hong-Kong.")
 959     (license license:asl2.0)))
 960
 961 (define-public nkf
 962   (let ((commit "08043eadf4abdddcf277842217e3c77a24740dc2")
 963         (revision "1"))
 964     (package
 965       (name "nkf")
 966       ;; The commits corresponding to specific versions are published
 967       ;; here:
 968       ;; https://ja.osdn.net/projects/nkf/scm/git/nkf/
 969       (version "2.1.5")
 970       (source (origin
 971                 (method git-fetch)
 972                 (uri (git-reference
 973                       (url "https://github.com/nurse/nkf.git")
 974                       (commit commit)))
 975                 (file-name (git-file-name name version))
 976                 (sha256
 977                  (base32
 978                   "0anw0knr1iy4p9w3d3b3pbwzh1c43p1i2q4c28kw9zviw8kx2rly"))))
 979       (build-system gnu-build-system)
 980       (arguments
 981        `(#:tests? #f ; test for perl module
 982          #:make-flags (list "CC=gcc" "CFLAGS=-O2 -Wall -pedantic"
 983                             (string-append "prefix=" %output)
 984                             "MKDIR=mkdir -p")
 985          #:phases
 986          (modify-phases %standard-phases
 987            (delete 'configure)))) ; No ./configure script
 988       (home-page "https://ja.osdn.net/projects/nkf/")
 989       (synopsis "Network Kanji Filter")
 990       (description "Nkf is yet another kanji code converter among networks,
 991 hosts and terminals.  It converts input kanji code to designated kanji code
 992 such as ISO-2022-JP, Shift_JIS, EUC-JP, UTF-8, UTF-16 or UTF-32.")
 993       (license license:zlib))))
 994
 995 (define-public python-pandocfilters
 996   (package
 997     (name "python-pandocfilters")
 998     (version "1.4.2")
 999     (source
1000      (origin
1001        (method url-fetch)
1002        (uri (pypi-uri "pandocfilters" version))
1003        (sha256
1004         (base32
1005          "1a8d9b7s48gmq9zj0pmbyv2sivn5i7m6mybgpkk4jm5vd7hp1pdk"))))
1006     (build-system python-build-system)
1007     (home-page "https://github.com/jgm/pandocfilters")
1008     (synopsis "Python module for writing Pandoc filters")
1009     (description "Pandoc is a powerful utility to transform various
1010 input formats into a wide range of output formats.  To alter the
1011 exported output document, Pandoc allows the usage of filters, which
1012 are pipes that read a JSON serialization of the Pandoc AST from stdin,
1013 transform it in some way, and write it to stdout.  It allows therefore
1014 to alter the processing of Pandoc's supported input formats, for
1015 instance one can add new syntax elements to markdown, etc.
1016
1017 This package provides Python bindings.")
1018     (license license:bsd-3)))