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, 2020 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, 2020 Efraim Flashner <efraim@flashner.co.il>
   9 ;;; Copyright © 2016 Nikita <nikita@n0.is>
  10 ;;; Copyright © 2016, 2020 Marius Bakke <marius@gnu.org>
  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, 2020 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 ;;; Copyright © 2019 Wiktor Żelazny <wzelazny@vurv.cz>
  22 ;;; Copyright © 2020 Maxim Cournoyer <maxim.cournoyer@gmail.com>
  23 ;;; Copyright © 2021 Jean-Baptiste Volatier <jbv@pm.me>
  24 ;;; Copyright © 2021 Simon Tournier <zimon.toutoune@gmail.com>
  25 ;;; Copyright © 2021 Felix Gruber <felgru@posteo.net>
  26 ;;;
  27 ;;; This file is part of GNU Guix.
  28 ;;;
  29 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  30 ;;; under the terms of the GNU General Public License as published by
  31 ;;; the Free Software Foundation; either version 3 of the License, or (at
  32 ;;; your option) any later version.
  33 ;;;
  34 ;;; GNU Guix is distributed in the hope that it will be useful, but
  35 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  36 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  37 ;;; GNU General Public License for more details.
  38 ;;;
  39 ;;; You should have received a copy of the GNU General Public License
  40 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  41
  42 (define-module (gnu packages textutils)
  43   #:use-module ((guix licenses) #:prefix license:)
  44   #:use-module (guix packages)
  45   #:use-module (guix download)
  46   #:use-module (guix git-download)
  47   #:use-module (guix build-system ant)
  48   #:use-module (guix build-system gnu)
  49   #:use-module (guix build-system go)
  50   #:use-module (guix build-system cmake)
  51   #:use-module (guix build-system python)
  52   #:use-module (guix utils)
  53   #:use-module (gnu packages)
  54   #:use-module (gnu packages autotools)
  55   #:use-module (gnu packages base)
  56   #:use-module (gnu packages compression)
  57   #:use-module (gnu packages gcc)
  58   #:use-module (gnu packages golang)
  59   #:use-module (gnu packages gettext)
  60   #:use-module (gnu packages java)
  61   #:use-module (gnu packages ncurses)
  62   #:use-module (gnu packages pcre)
  63   #:use-module (gnu packages perl)
  64   #:use-module (gnu packages pkg-config)
  65   #:use-module (gnu packages python)
  66   #:use-module (gnu packages python-build)
  67   #:use-module (gnu packages python-xyz)
  68   #:use-module (gnu packages readline)
  69   #:use-module (gnu packages ruby)
  70   #:use-module (gnu packages slang)
  71   #:use-module (gnu packages web))
  72
  73 (define-public dos2unix
  74   (package
  75     (name "dos2unix")
  76     (version "7.4.2")
  77     (source
  78      (origin
  79        (method url-fetch)
  80        (uri (string-append "https://waterlan.home.xs4all.nl/dos2unix/"
  81                            "dos2unix-" version ".tar.gz"))
  82        (sha256
  83         (base32 "00dfsf4rfyjb5j12gan8xjiirm0asshdz6dmd3l34a7ays6wadb0"))))
  84     (build-system gnu-build-system)
  85     (arguments
  86      `(#:make-flags
  87        (list (string-append "CC=" ,(cc-for-target))
  88              (string-append "prefix=" (assoc-ref %outputs "out")))
  89        #:phases
  90        (modify-phases %standard-phases
  91          (delete 'configure)))) ; no configure script
  92     (native-inputs
  93      `(("gettext" ,gettext-minimal)
  94        ("perl" ,perl)))
  95     (home-page "https://waterlan.home.xs4all.nl/dos2unix.html")
  96     (synopsis "DOS/Mac to Unix and vice versa text file format converter")
  97     (description
  98      "dos2unix is a tool to convert line breaks in a text file from Unix format
  99 to DOS format and vice versa.")
 100     (license license:bsd-2)))
 101
 102 (define-public recode
 103   (package
 104     (name "recode")
 105     (version "3.7.6")
 106     (source
 107      (origin
 108        (method url-fetch)
 109        (uri (string-append "https://github.com/rrthomas/recode/releases/"
 110                            "download/v" version "/recode-" version ".tar.gz"))
 111        (sha256
 112         (base32 "0m59sd1ca0zw1aydpc3m8sw03nc885knmccqryg7byzmqs585ia6"))))
 113     (build-system gnu-build-system)
 114     (native-inputs
 115      `(("python" ,python)
 116        ("python-cython" ,python-cython)))
 117     (home-page "https://github.com/rrthomas/recode")
 118     (synopsis "Text encoding converter")
 119     (description "The Recode library converts files between character sets and
 120 usages.  It recognises or produces over 200 different character sets (or about
 121 300 if combined with an iconv library) and transliterates files between almost
 122 any pair.  When exact transliteration are not possible, it gets rid of
 123 offending characters or falls back on approximations.  The recode program is a
 124 handy front-end to the library.")
 125     (license license:gpl3+)))
 126
 127 (define-public enca
 128   (package
 129     (name "enca")
 130     (version "1.19")
 131     (source
 132      (origin
 133        (method git-fetch)
 134        (uri (git-reference
 135               (url "https://github.com/nijel/enca")
 136               (commit version)))
 137        (file-name (git-file-name name version))
 138        (sha256
 139         (base32 "19q7cwwxmmk5j9438bsqdpjvdjawsd3zmw1zyqgi7s4m0rasr3ah"))))
 140     (build-system gnu-build-system)
 141     ;; enca-1.19 tests fail with recent recode.
 142     ;(inputs `(("recode" ,recode)))
 143     (home-page "https://github.com/nijel/enca")
 144     (synopsis "Text encoding detection tool")
 145     (description "Enca (Extremely Naive Charset Analyser) consists of libenca,
 146 an encoding detection library, and enca, a command line frontend, integrating
 147 libenca and several charset conversion libraries and tools.")
 148     (license license:gpl2)))
 149
 150 (define-public utf8proc
 151   (package
 152     (name "utf8proc")
 153     (version "2.5.0")
 154     (source
 155      (origin
 156        (method git-fetch)
 157        (uri (git-reference
 158              (url "https://github.com/JuliaStrings/utf8proc")
 159              (commit (string-append "v" version))))
 160        (file-name (git-file-name name version))
 161        (sha256
 162         (base32 "1xlkazhdnja4lksn5c9nf4bln5gjqa35a8gwlam5r0728w0h83qq"))))
 163     (build-system gnu-build-system)
 164     (native-inputs
 165      (let ((UNICODE_VERSION "13.0.0"))  ; defined in data/Makefile
 166        ;; Test data that is otherwise downloaded with curl.
 167        `(("NormalizationTest.txt"
 168           ,(origin
 169              (method url-fetch)
 170              (uri (string-append "https://www.unicode.org/Public/"
 171                                  UNICODE_VERSION "/ucd/NormalizationTest.txt"))
 172              (sha256
 173               (base32 "07g0ya4f6zfzvpp24ccxkb2yq568kh83gls85rjl950nv5fya3nn"))))
 174          ("GraphemeBreakTest.txt"
 175           ,(origin
 176              (method url-fetch)
 177              (uri (string-append "https://www.unicode.org/Public/"
 178                                  UNICODE_VERSION
 179                                  "/ucd/auxiliary/GraphemeBreakTest.txt"))
 180              (sha256
 181               (base32 "07f8rrvcsq4pibdz6zxggxy8w7zjjqyw2ggclqlhalyv45yv7prj"))))
 182
 183          ;; For tests.
 184          ("perl" ,perl))))
 185     (arguments
 186      '(#:make-flags (list "CC=gcc"
 187                           (string-append "prefix=" (assoc-ref %outputs "out")))
 188        #:phases
 189        (modify-phases %standard-phases
 190          (delete 'configure)
 191          (add-before 'check 'check-data
 192            (lambda* (#:key inputs #:allow-other-keys)
 193              (for-each (lambda (i)
 194                          (copy-file (assoc-ref inputs i)
 195                                     (string-append "data/" i)))
 196                        '("NormalizationTest.txt" "GraphemeBreakTest.txt"))
 197              (substitute* "data/GraphemeBreakTest.txt"
 198                (("÷") "/")
 199                (("×") "+"))
 200              #t)))))
 201     (home-page "https://juliastrings.github.io/utf8proc/")
 202     (synopsis "C library for processing UTF-8 Unicode data")
 203     (description "utf8proc is a small C library that provides Unicode
 204 normalization, case-folding, and other operations for data in the UTF-8
 205 encoding, supporting Unicode version 9.0.0.")
 206     (license license:expat)))
 207
 208 (define-public utf8proc-2.6.1
 209   (package
 210     (inherit utf8proc)
 211     (name "utf8proc")
 212     (version "2.6.1")
 213     (source
 214      (origin
 215        (method git-fetch)
 216        (uri (git-reference
 217              (url "https://github.com/JuliaStrings/utf8proc")
 218              (commit (string-append "v" version))))
 219        (file-name (git-file-name name version))
 220        (sha256
 221         (base32 "1zqc6airkzkssbjxanx5v8blfk90180gc9id0dx8ncs54f1ib8w7"))))
 222     (arguments
 223      (substitute-keyword-arguments (package-arguments utf8proc)
 224        ((#:phases phases)
 225         `(modify-phases ,phases
 226            (replace 'check-data
 227              (lambda* (#:key inputs native-inputs #:allow-other-keys)
 228                (display native-inputs)
 229                (for-each (lambda (i)
 230                            (copy-file (assoc-ref (or native-inputs inputs) i)
 231                                       (string-append "data/" i)))
 232                          '("NormalizationTest.txt" "GraphemeBreakTest.txt"
 233                            "DerivedCoreProperties.txt"))))))))
 234     (native-inputs
 235      (append
 236       (package-native-inputs utf8proc)
 237       (let ((UNICODE_VERSION "13.0.0"))
 238         `(("DerivedCoreProperties.txt"
 239            ,(origin
 240               (method url-fetch)
 241               (uri (string-append "https://www.unicode.org/Public/"
 242                                   UNICODE_VERSION "/ucd/DerivedCoreProperties.txt"))
 243               (sha256
 244                (base32 "0j12x112cd8fpgazkc8izxnhhpia44p1m36ff8yapslxndcmzm55"))))
 245           ;; For tests
 246           ("ruby" ,ruby)))))))
 247
 248 (define-public libconfuse
 249   (package
 250     (name "libconfuse")
 251     (version "3.3")
 252     (source
 253      (origin
 254        (method url-fetch)
 255        (uri (string-append "https://github.com/martinh/libconfuse/"
 256                            "releases/download/v" version
 257                            "/confuse-" version ".tar.xz"))
 258        (sha256
 259         (base32 "043hqqykpprgrkw9s2hbdlxr308a7yxwsgxj4m8aadg1401hmm8x"))))
 260     (build-system gnu-build-system)
 261     (arguments
 262      '(#:configure-flags '("--disable-static")))
 263     (home-page "https://github.com/martinh/libconfuse")
 264     (synopsis "Configuration file parser library")
 265     (description "libconfuse is a configuration file parser library.  It
 266 supports sections and (lists of) values (strings, integers, floats, booleans
 267 or other sections), as well as some other features (such as
 268 single/double-quoted strings, environment variable expansion, functions and
 269 nested include statements).")
 270     (license license:isc)))
 271
 272 (define-public libgtextutils
 273   (package
 274     (name "libgtextutils")
 275     (version "0.7")
 276     (source
 277      (origin
 278        (method url-fetch)
 279        (uri (string-append
 280              "https://github.com/agordon/libgtextutils/releases/download/"
 281              version "/libgtextutils-" version ".tar.gz"))
 282        (sha256
 283         (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
 284     (build-system gnu-build-system)
 285     (arguments
 286      `(#:phases
 287        (modify-phases %standard-phases
 288          (replace 'bootstrap
 289            (lambda _ (invoke "sh" "reconf")))
 290          (add-after 'set-paths 'hide-default-gcc
 291            (lambda* (#:key inputs #:allow-other-keys)
 292              (let ((gcc (assoc-ref inputs "gcc")))
 293                ;; Remove the default GCC from CPLUS_INCLUDE_PATH to prevent
 294                ;; conflicts with the GCC 5 input.
 295                (setenv "CPLUS_INCLUDE_PATH"
 296                        (string-join
 297                         (delete (string-append gcc "/include/c++")
 298                                 (string-split (getenv "CPLUS_INCLUDE_PATH") #\:))
 299                         ":"))
 300                #t))))))
 301     (native-inputs
 302      `(("autoconf" ,autoconf)
 303        ("automake" ,automake)
 304        ("gcc@5" ,gcc-5) ; doesn't build with later versions
 305        ("libtool" ,libtool)))
 306     (home-page "https://github.com/agordon/libgtextutils")
 307     (synopsis "Gordon's text utils library")
 308     (description
 309      "libgtextutils is a text utilities library used by the fastx toolkit from
 310 the Hannon Lab.")
 311     (license license:agpl3+)))
 312
 313 (define-public cityhash
 314   (let ((commit "8af9b8c"))
 315     (package
 316       (name "cityhash")
 317       (version (string-append "1.1-2." commit))
 318       (source (origin
 319                 (method git-fetch)
 320                 (uri (git-reference
 321                       (url "https://github.com/google/cityhash")
 322                       (commit commit)))
 323                 (file-name (string-append name "-" version ".tar.gz"))
 324                 (sha256
 325                  (base32
 326                   "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
 327       (build-system gnu-build-system)
 328       (arguments
 329        '(#:make-flags (list "CXXFLAGS=-g -O3")
 330          #:phases
 331          (modify-phases %standard-phases
 332            ;; citycrc is not installed by default but is used by some
 333            ;; programs.
 334            (add-after 'install 'install-citycrc
 335              (lambda* (#:key outputs #:allow-other-keys)
 336                (let* ((out (assoc-ref outputs "out"))
 337                       (include (string-append out "/include")))
 338                  (install-file "src/citycrc.h" include))
 339                #t)))))
 340       (home-page "https://github.com/google/cityhash")
 341       (synopsis "C++ hash functions for strings")
 342       (description
 343        "CityHash provides hash functions for strings.  The functions mix the
 344 input bits thoroughly but are not suitable for cryptography.")
 345       (license license:expat))))
 346
 347 (define-public ustr
 348   (package
 349     (name "ustr")
 350     (version "1.0.4")
 351     (source (origin
 352               (method url-fetch)
 353               (uri (string-append "http://www.and.org/ustr/" version
 354                                   "/ustr-" version ".tar.bz2"))
 355               (sha256
 356                (base32
 357                 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
 358               (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
 359     (build-system gnu-build-system)
 360     (arguments
 361      `(#:make-flags
 362        (list (string-append "CC=" ,(cc-for-target))
 363              "HIDE="
 364              ;; Override "/sbin/ldconfig" with "echo" because we don't need
 365              ;; "ldconfig".
 366              "LDCONFIG=echo"
 367              (string-append "prefix=" (assoc-ref %outputs "out"))
 368              "all-shared")
 369        #:phases
 370        (modify-phases %standard-phases
 371          (add-after 'unpack 'fix-cross-compilation
 372            ;; The Makefile contains more insults than cross-compilation support.
 373            ;; It poorly reinvents autotools by compiling C programmes with $CC,
 374            ;; then tries to run them during the build.  Hard-code the results.
 375            (lambda _
 376              (substitute* "Makefile"
 377                (("\\./autoconf_64b")
 378                 ,(if (target-64bit? (or (%current-target-system)
 379                                         (%current-system)))
 380                      "echo 1"
 381                      "echo 0"))
 382                (("\\./autoconf_vsnprintf") "echo 0"))))
 383          (add-after 'unpack 'omit-static-libraries
 384            ;; These are needed to build & test ustr, but don't install them.
 385            (lambda _
 386              (substitute* "Makefile"
 387                ((".*install.*LIB_STATIC.*") ""))))
 388          (add-after 'unpack 'disable-check-for-stdint
 389            (lambda _
 390              ;; Of course we have stdint.h, just not in /usr/include
 391              (substitute* '("Makefile"
 392                             "ustr-import.in")
 393                (("-f \"/usr/include/stdint.h\"") "-z \"\""))
 394              #t))
 395          ;; No configure script
 396          (delete 'configure))))
 397     (home-page "http://www.and.org/ustr/")
 398     (synopsis "String library with very low memory overhead")
 399     (description
 400      "Ustr is a string library for C with very low memory overhead.")
 401     ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
 402     ;; LGPL, etc. ... if you need another license to help compatibility, just
 403     ;; ask for it.  It's basically public domain, without all the legal
 404     ;; problems for everyone that trying to make something public domain
 405     ;; entails."
 406     (license license:public-domain)))
 407
 408 (define-public ascii2binary
 409   (package
 410     (name "ascii2binary")
 411     (version "2.14")
 412     (source
 413      (origin
 414        (method url-fetch)
 415        (uri (string-append "http://billposer.org/Software/Downloads/"
 416                            "ascii2binary-" version ".tar.bz2"))
 417        (sha256
 418         (base32 "0dc9fxcdmppbs9s06jvq61zbk552laxps0xyk098gj41697ihd96"))))
 419     (build-system gnu-build-system)
 420     (native-inputs
 421      `(("gettext" ,gettext-minimal)))
 422     (home-page "https://billposer.org/Software/a2b.html")
 423     (synopsis "Convert between ASCII, hexadecimal and binary representations")
 424     (description "The two programs are useful for generating test data, for
 425 inspecting binary files, and for interfacing programs that generate textual
 426 output to programs that require binary input and conversely.  They can also be
 427 useful when it is desired to reformat numbers.
 428
 429 @itemize
 430
 431 @item @command{ascii2binary} reads input consisting of ascii or hexadecimal
 432    representation numbers separated by whitespace and produces as output
 433    the binary equivalents.  The type and precision of the binary output
 434    is selected using command line flags.
 435
 436 @item @command{binary2ascii} reads input consisting of binary numbers
 437    and converts them to their ascii or hexadecimal representation.
 438    Command line flags specify the type and size of the binary numbers
 439    and provide control over the format of the output.
 440    Unsigned integers may be written out in binary, octal, decimal,
 441    or hexadecimal.
 442
 443    Signed integers may be written out only in binary or decimal.  Floating
 444    point numbers may be written out only decimal, either in standard or
 445    scientific notation.  (If you want to examine the binary representation
 446    of floating point numbers, just treat the input as a sequence of unsigned
 447    characters.)
 448
 449 @end itemize")
 450     (license license:gpl3)))
 451
 452 (define-public uniutils
 453   (package
 454     (name "uniutils")
 455     (version "2.27")
 456     (source
 457      (origin
 458        (method url-fetch)
 459        (uri (string-append "http://billposer.org/Software/Downloads/"
 460                            "uniutils-" version ".tar.bz2"))
 461        (sha256
 462         (base32 "19w1510w87gx7n4qy3zsb0m467a4rn5scvh4ajajg7jh6x5xri08"))))
 463     (build-system gnu-build-system)
 464     (arguments
 465      '(#:configure-flags '("--disable-dependency-tracking")
 466        #:phases
 467        (modify-phases %standard-phases
 468          (add-after 'build 'fix-paths
 469            (lambda* (#:key outputs inputs #:allow-other-keys)
 470              (let ((out (assoc-ref outputs "out"))
 471                    (a2b (assoc-ref inputs "ascii2binary"))
 472                    (iconv (assoc-ref inputs "libiconv")))
 473                (substitute* "utf8lookup"
 474                  (("^ascii2binary ") (string-append a2b "/bin/ascii2binary "))
 475                  (("^uniname ") (string-append out "/bin/uniname "))
 476                  (("^iconv ") (string-append iconv "/bin/iconv ")))
 477              #t))))))
 478     (inputs
 479      `(("ascii2binary" ,ascii2binary)
 480        ("libiconv" ,libiconv)))
 481     (home-page "https://billposer.org/Software/unidesc.html")
 482     (synopsis "Find out what is in a Unicode file")
 483     (description "Useful tools when working with Unicode files when one
 484 doesn't know the writing system, doesn't have the necessary font, needs to
 485 inspect invisible characters, needs to find out whether characters have been
 486 combined or in what order they occur, or needs statistics on which characters
 487 occur.
 488
 489 @itemize
 490
 491 @item @command{uniname} defaults to printing the character offset of each
 492 character, its byte offset, its hex code value, its encoding, the glyph
 493 itself, and its name.  It may also be used to validate UTF-8 input.
 494
 495 @item @command{unidesc} reports the character ranges to which different
 496 portions of the text belong.   It can also be used to identify Unicode encodings
 497 (e.g. UTF-16be) flagged by magic numbers.
 498
 499 @item @command{unihist} generates a histogram of the characters in its input.
 500
 501 @item @command{ExplicateUTF8} is intended for debugging or for learning about
 502 Unicode.  It determines and explains the validity of a sequence of bytes as a
 503 UTF8 encoding.
 504
 505 @item @command{utf8lookup} provides a handy way to look up Unicode characters
 506 from the command line.
 507
 508 @item @command{unireverse} reverse each line of UTF-8 input
 509 character-by-character.
 510
 511 @end itemize")
 512     (license license:gpl3)))
 513
 514 (define-public libconfig
 515   (package
 516     (name "libconfig")
 517     (version "1.7.3")
 518     (home-page "https://hyperrealm.github.io/libconfig/")
 519     (source (origin
 520               (method url-fetch)
 521               (uri (string-append home-page "/dist/libconfig-"
 522                                   version ".tar.gz"))
 523               (sha256
 524                (base32
 525                 "1rqynfxl1zxwk4b42sniz9xlw285aidcrsfih51p8dy0rbb6clal"))))
 526     (build-system gnu-build-system)
 527     (synopsis "C/C++ configuration file library")
 528     (description
 529      "Libconfig is a simple library for manipulating structured configuration
 530 files.  This file format is more compact and more readable than XML.  And
 531 unlike XML, it is type-aware, so it is not necessary to do string parsing in
 532 application code.")
 533     (license license:lgpl2.1+)))
 534
 535 (define-public pfff
 536   (package
 537     (name "pfff")
 538     (version "1.0")
 539     (source (origin
 540               (method git-fetch)
 541               (uri (git-reference
 542                      (url "https://github.com/pfff/pfff")
 543                      (commit (string-append "v" version))))
 544               (file-name (git-file-name name version))
 545               (sha256
 546                (base32
 547                 "1nxkfm7zliq3rmr7yp871sppwfnz71iz364m2sgazny71pzykggc"))))
 548     (build-system cmake-build-system)
 549     (home-page "https://biit.cs.ut.ee/pfff/")
 550     (synopsis "Probabilistic fast file fingerprinting tool")
 551     (description
 552      "pfff is a tool for calculating a compact digital fingerprint of a file
 553 by sampling randomly from the file instead of reading it in full.
 554 Consequently, the computation has a flat performance characteristic,
 555 correlated with data variation rather than file size.  pfff can be as reliable
 556 as existing hashing techniques, with provably negligible risk of collisions.")
 557     (license license:bsd-3)))
 558
 559 (define-public oniguruma
 560   (package
 561     (name "oniguruma")
 562     (version "6.9.5-rev1")
 563     (source (origin
 564               (method url-fetch)
 565               (uri (string-append "https://github.com/kkos/"
 566                                   "oniguruma/releases/download/v"
 567                                   ;; If there is a "-" in the version, convert
 568                                   ;; to underscore for this part of the URI.
 569                                   (string-map (lambda (c) (if (char=? #\- c) #\_ c))
 570                                               version)
 571                                   "/onig-" version ".tar.gz"))
 572               (sha256
 573                (base32
 574                 "17m92k1n6bvza6m35fpd5g36zwpwm3hfz3478iwj5bvj2sfq8g6k"))))
 575     (build-system gnu-build-system)
 576     (arguments '(#:configure-flags '("--disable-static")))
 577     (home-page "https://github.com/kkos/oniguruma")
 578     (synopsis "Regular expression library")
 579     (description "Oniguruma is a regular expressions library.  The special
 580 characteristic of this library is that different character encoding for every
 581 regular expression object can be specified.")
 582     (license license:bsd-2)))
 583
 584 (define-public antiword
 585   (package
 586     (name "antiword")
 587     (version "0.37")
 588     (source (origin
 589               (method url-fetch)
 590               (uri (string-append "http://www.winfield.demon.nl/linux"
 591                                   "/antiword-" version ".tar.gz"))
 592               (sha256
 593                (base32
 594                 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
 595               (patches (search-patches "antiword-CVE-2014-8123.patch"))))
 596     (build-system gnu-build-system)
 597     (arguments
 598      `(#:tests? #f ; There are no tests
 599        #:make-flags
 600        (list "-f" "Makefile.Linux"
 601              (string-append "GLOBAL_INSTALL_DIR="
 602                             (assoc-ref %outputs "out") "/bin")
 603              (string-append "GLOBAL_RESOURCES_DIR="
 604                             (assoc-ref %outputs "out") "/share/antiword"))
 605        #:phases
 606        (modify-phases %standard-phases
 607          (replace 'configure
 608            (lambda* (#:key outputs #:allow-other-keys)
 609              ;; Ensure that mapping files can be found in the actual package
 610              ;; data directory.
 611              (substitute* "antiword.h"
 612                (("/usr/share/antiword")
 613                 (string-append (assoc-ref outputs "out") "/share/antiword")))
 614              #t))
 615          (replace 'install
 616            (lambda* (#:key make-flags #:allow-other-keys)
 617              (apply invoke "make" `("global_install" ,@make-flags)))))))
 618     (home-page "http://www.winfield.demon.nl/")
 619     (synopsis "Microsoft Word document reader")
 620     (description "Antiword is an application for displaying Microsoft Word
 621 documents.  It can also convert the document to PostScript or XML.  Only
 622 documents made by MS Word version 2 and version 6 or later are supported.  The
 623 name comes from: \"The antidote against people who send Microsoft Word files
 624 to everybody, because they believe that everybody runs Windows and therefore
 625 runs Word\".")
 626     (license license:gpl2+)))
 627
 628 (define-public catdoc
 629   (package
 630     (name "catdoc")
 631     (version "0.95")
 632     (source (origin
 633               (method url-fetch)
 634               (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
 635                                   "catdoc-" version ".tar.gz"))
 636               (patches (search-patches "catdoc-CVE-2017-11110.patch"))
 637               (sha256
 638                (base32
 639                 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
 640     (build-system gnu-build-system)
 641     ;; TODO: Also build `wordview` which requires `tk` – make a separate
 642     ;; package for this.
 643     (arguments
 644      '(#:tests? #f ; There are no tests
 645        #:configure-flags '("--disable-wordview")
 646        #:phases
 647        (modify-phases %standard-phases
 648          (add-before 'install 'fix-install
 649            (lambda* (#:key outputs #:allow-other-keys)
 650              (let ((out (assoc-ref outputs "out")))
 651                (mkdir-p (string-append out "/share/man/man1"))))))))
 652     (home-page "https://www.wagner.pp.ru/~vitus/software/catdoc/")
 653     (synopsis "MS-Word to TeX or plain text converter")
 654     (description "@command{catdoc} extracts text from MS-Word files, trying to
 655 preserve as many special printable characters as possible.  It supports
 656 everything up to Word-97. Also supported are MS Write documents and RTF files.
 657
 658 @command{catdoc} does not preserve complex word formatting, but it can
 659 translate some non-ASCII characters into TeX escape codes.  It's goal is to
 660 extract plain text and allow you to read it and, probably, reformat with TeX,
 661 according to TeXnical rules.
 662
 663 This package also provides @command{xls2csv}, which extracts data from Excel
 664 spreadsheets and outputs it in comma-separated-value format, and
 665 @command{catppt}, which extracts data from PowerPoint presentations.")
 666     (license license:gpl2+)))
 667
 668 (define-public utfcpp
 669   (package
 670     (name "utfcpp")
 671     (version "2.3.5")
 672     (source (origin
 673               (method git-fetch)
 674               (uri (git-reference
 675                      (url "https://github.com/nemtrif/utfcpp")
 676                      (commit (string-append "v" version))))
 677               (file-name (git-file-name name version))
 678               (sha256
 679                (base32
 680                 "1gr98d826z6wa58r1s5i7rz7q2x3r31v7zj0pjjlrc7gfxwklr4s"))))
 681     (build-system cmake-build-system)
 682     (arguments
 683      `(#:out-of-source? #f
 684        #:phases
 685        (modify-phases %standard-phases
 686          (replace 'install              ; no install target
 687            (lambda* (#:key outputs #:allow-other-keys)
 688              (let* ((out (assoc-ref outputs "out"))
 689                     (include (string-append out "/include"))
 690                     (doc (string-append out "/share/doc/" ,name)))
 691                (copy-recursively "source" include)
 692                (install-file "README.md" doc)
 693                #t))))))
 694     (home-page "https://github.com/nemtrif/utfcpp")
 695     (synopsis "Portable C++ library for handling UTF-8")
 696     (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
 697 in a portable way.")
 698     (license license:boost1.0)))
 699
 700 (define-public dbacl
 701   (package
 702     (name "dbacl")
 703     (version "1.14.1")
 704     (source
 705      (origin
 706        (method url-fetch)
 707        (uri (string-append "mirror://sourceforge/dbacl/dbacl/" version "/"
 708                            "dbacl-" version ".tar.gz"))
 709        (sha256
 710         (base32 "1gas0112wqjvwn9qg3hxnawk7h3prr0w9b2h68f3p1ifd1kzn3gz"))
 711        (patches (search-patches "dbacl-include-locale.h.patch"))))
 712     (build-system gnu-build-system)
 713     (arguments
 714      `(#:make-flags
 715        (list
 716         (string-append "-I" (assoc-ref %build-inputs "slang")
 717                        "/include/slang")
 718         (string-append "-I" (assoc-ref %build-inputs "ncurses")
 719                        "/include/ncurses"))
 720        #:phases
 721        (modify-phases %standard-phases
 722          (add-after 'unpack 'delete-sample6-and-japanese
 723            (lambda _
 724              (substitute* "doc/Makefile.am"
 725                (("sample6.txt") "")
 726                (("japanese.txt") ""))
 727              (delete-file "doc/sample6.txt")
 728              (delete-file "doc/japanese.txt")
 729              (substitute* (list "src/tests/Makefile.am"
 730                                 "src/tests/Makefile.in")
 731                (("dbacl-jap.shin") "")
 732                (("dbacl-jap.sh") ""))
 733              #t))
 734          (add-after 'unpack 'delete-test
 735            ;; See comments about the license.
 736            (lambda _
 737              (delete-file "src/tests/dbacl-jap.shin")
 738              #t))
 739          (add-after 'unpack 'fix-test-files
 740            (lambda* (#:key inputs outputs #:allow-other-keys)
 741              (let* ((out (assoc-ref outputs "out"))
 742                     (bin (string-append out "/bin")))
 743                (substitute* (find-files "src/tests/" "\\.shin$")
 744                  (("PATH=/bin:/usr/bin")
 745                   "#PATH=/bin:/usr/bin")
 746                  (("diff") (string-append (which "diff")))
 747                  (("tr") (string-append (which "tr"))))
 748                #t)))
 749          (replace 'bootstrap
 750            (lambda _
 751              (invoke "autoreconf" "-vif")
 752              #t)))))
 753     (inputs
 754      `(("ncurses" ,ncurses)
 755        ("perl" ,perl)
 756        ("readline" ,readline)
 757        ("slang" ,slang)))
 758     (native-inputs
 759      `(("libtool" ,libtool)
 760        ("autoconf" ,autoconf)
 761        ("automake" ,automake)
 762        ("pkg-config" ,pkg-config)))
 763     (home-page "https://www.lbreyer.com/dbacl.html")
 764     (synopsis "Bayesian text and email classifier")
 765     (description
 766      "dbacl is a fast Bayesian text and email classifier.  It builds a variety
 767 of language models using maximum entropy (minimum divergence) principles, and
 768 these can then be used to categorize input data automatically among multiple
 769 categories.")
 770     ;; The software is licensed as GPLv3 or later, but
 771     ;; includes various sample texts in the doc dir:
 772     ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
 773     ;;   by Mark Twain.
 774     ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
 775     ;; - sample6.txt is a forwarded email, copyright unknown.
 776     ;;   Guix does exclude sample6.txt.
 777     ;; - japanese.txt is a Japanese unoffical translation of the
 778     ;;   GNU General Public License, (c) by the Free Software Foundation.
 779     ;;   Guix excludes this file.
 780     (license (list license:gpl3+ license:public-domain))))
 781
 782 (define-public dotconf
 783   (package
 784     (name "dotconf")
 785     (version "1.3")
 786     (source (origin
 787               (method git-fetch)
 788               (uri (git-reference
 789                     (url "https://github.com/williamh/dotconf")
 790                     (commit (string-append "v" version))))
 791               (file-name (git-file-name name version))
 792               (sha256
 793                (base32
 794                 "1sc95hw5k2xagpafny0v35filmcn05k1ds5ghkldfpf6xw4hakp7"))))
 795     (build-system gnu-build-system)
 796     (arguments `(#:tests? #f))  ; FIXME maketest.sh does not work.
 797     (native-inputs
 798      `(("autoconf" ,autoconf)
 799        ("automake" ,automake)
 800        ("libtool" ,libtool)))
 801     (home-page "https://github.com/williamh/dotconf")
 802     (synopsis "Configuration file parser library")
 803     (description
 804      "C library for creating and parsing configuration files.")
 805     (license (list license:lgpl2.1         ; Main distribution.
 806                    license:asl1.1))))      ; src/readdir.{c,h}
 807
 808 (define-public drm-tools
 809   (package
 810     (name "drm-tools")
 811     (version "1.1.33")
 812     (source (origin
 813               (method url-fetch)
 814               (uri (string-append "mirror://sourceforge/drmtools/drm_tools-"
 815                                   version ".tar.gz"))
 816               (sha256
 817                (base32
 818                 "187zbxw21zcg8gpyc13gxlycfw0n05a6rmqq6im5wr9zk1v1wj80"))))
 819     (build-system cmake-build-system)
 820     (arguments
 821      `(#:tests? #f                      ;the test suite fails
 822        #:phases (modify-phases %standard-phases
 823                   (add-after 'unpack 'set-install-prefixes
 824                     (lambda* (#:key outputs #:allow-other-keys)
 825                       (let* ((out (assoc-ref outputs "out")))
 826                         (substitute* "CMakeLists.txt"
 827                           (("tmp/testinstall")
 828                            (string-drop out 1))
 829                           (("/man/man1")
 830                            "/share/man/man1"))
 831                         #t)))
 832                   (add-after 'unpack 'adjust-test-paths
 833                     (lambda _
 834                       (substitute* '("test_extract_increment.sh"
 835                                      "test_extract_features.sh"
 836                                      "test_extract_features2.sh"
 837                                      "test_dmath.sh")
 838                         (("\\./extract") "extract")
 839                         (("\\./dmath") "dmath")
 840                         (("/usr/local/bin/") "")
 841                         (("/bin/rm") "rm")
 842                         (("/bin/cp") "cp"))
 843                       #t))
 844                   (delete 'check)
 845                   ;; The produced binaries are written directly to %output/bin.
 846                   (delete 'install)
 847                   (add-after 'build 'check
 848                     (lambda* (#:key outputs tests? #:allow-other-keys)
 849                       (when tests?
 850                         (let* ((out (assoc-ref outputs "out"))
 851                                (bin (string-append out "/bin")))
 852                           (setenv "PATH" (string-append bin ":"
 853                                                         (getenv "PATH")))
 854                           (with-directory-excursion
 855                               (format #f "../drm_tools-~a" ,version)
 856                             (invoke "sh" "test_all.sh")))))))))
 857     (native-inputs `(("which" ,which))) ;for tests
 858     (inputs `(("pcre" ,pcre)))
 859     (home-page "http://drmtools.sourceforge.net/")
 860     (synopsis "Utilities to manipulate text and binary files")
 861     (description "The drm_tools package contains the following commands:
 862 @table @command
 863 @item accudate
 864 An extended version of the \"date\" program that has sub-second accuracy.
 865 @item binformat
 866 Format complex binary data into text.
 867 @item binload
 868 Load data into a binary file using simple commands from the input.
 869 @item binorder
 870 Sort, merge, search, retrieve or generate test data consisting of fixed size
 871 binary records.
 872 @item binreplace
 873 Find or find/replace in binary files.
 874 @item binsplit
 875 Split test data consisting of fixed size binary records into one or more
 876 output streams.
 877 @item chardiff
 878 Find changes between two files at the character level.  Unlike \"diff\", it
 879 lists just the characters that differ, so if the 40,000th character is
 880 different only that one character will be shown, not the entire line.
 881 @item columnadd
 882 Add columns of integers, decimals, and/or times.
 883 @item datasniffer
 884 A utility for formatting binary data dumps.
 885 @item dmath
 886 Double precision interactive command line math calculator.
 887 @item extract
 888 Extract and emit data from text files based on character or token position.
 889 @item execinput
 890 A utility that reads from STDIN and executes each line as a command in a
 891 sub-process.
 892 @item indexed_text
 893 A utility for rapid retrieval of text by line numbers, in any order, from a
 894 text file.
 895 @item mdump
 896 Format binary data.
 897 @item msgqueue
 898 Create message queues and send/receive messages.
 899 @item mbin
 900 @itemx mbout
 901 Multiple buffer in and out.  Used for buffering a lot of data between a slow
 902 device and a fast device.  Mostly for buffering streaming tape drives for use
 903 with slower network connections, so that streaming is maintained as much as
 904 possible to minimize wear on the tape device.
 905 @item pockmark
 906 Corrupt data streams - useful for testing error correction and data recovery.
 907 @item tarsieve
 908 Filter, list, or split a tar file.
 909 @end table")
 910     (license license:gpl2+)))
 911
 912 (define-public java-rsyntaxtextarea
 913   (package
 914     (name "java-rsyntaxtextarea")
 915     (version "2.6.1")
 916     (source (origin
 917               (method git-fetch)
 918               (uri (git-reference
 919                      (url "https://github.com/bobbylight/RSyntaxTextArea")
 920                      (commit version)))
 921               (file-name (git-file-name name version))
 922               (sha256
 923                (base32
 924                 "0dyflzvxq2wvs0rgqfyi5yzzrb6r4bzw2dm8cl304dakxk38ddys"))))
 925     (build-system ant-build-system)
 926     (arguments
 927      `(;; FIXME: some tests fail because locale resources cannot be found.
 928        ;; Even when I add them to the class path,
 929        ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
 930        #:tests? #f
 931        #:jar-name "rsyntaxtextarea.jar"))
 932     (native-inputs
 933      `(("java-junit" ,java-junit)
 934        ("java-hamcrest-core" ,java-hamcrest-core)))
 935     (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
 936     (synopsis "Syntax highlighting text component for Java Swing")
 937     (description "RSyntaxTextArea is a syntax highlighting, code folding text
 938 component for Java Swing.  It extends @code{JTextComponent} so it integrates
 939 completely with the standard @code{javax.swing.text} package.  It is fast and
 940 efficient, and can be used in any application that needs to edit or view
 941 source code.")
 942     (license license:bsd-3)))
 943
 944 ;; We use the sources from git instead of the tarball from pypi, because the
 945 ;; latter does not include the Cython source file from which bycython.cpp is
 946 ;; generated.
 947 (define-public python-editdistance
 948   (let ((commit "3ea84a7dd3258c76aa3be851ef3d50e59c886846")
 949         (revision "1"))
 950     (package
 951       (name "python-editdistance")
 952       (version (string-append "0.3.1-" revision "." (string-take commit 7)))
 953       (source
 954        (origin
 955          (method git-fetch)
 956          (uri (git-reference
 957                (url "https://github.com/aflc/editdistance")
 958                (commit commit)))
 959          (file-name (git-file-name name version))
 960          (sha256
 961           (base32
 962            "1l43svsv12crvzphrgi6x435z6xg8m086c64armp8wzb4l8ccm7g"))))
 963       (build-system python-build-system)
 964       (arguments
 965        `(#:phases
 966          (modify-phases %standard-phases
 967            (add-after 'unpack 'build-cython-code
 968              (lambda _
 969                (with-directory-excursion "editdistance"
 970                  (delete-file "bycython.cpp")
 971                  (invoke "cython" "--cplus" "bycython.pyx")))))))
 972       (native-inputs
 973        `(("python-cython" ,python-cython)))
 974       (home-page "https://www.github.com/aflc/editdistance")
 975       (synopsis "Fast implementation of the edit distance (Levenshtein distance)")
 976       (description
 977        "This library simply implements Levenshtein distance algorithm with C++
 978 and Cython.")
 979       (license license:expat))))
 980
 981 (define-public go-github.com-mattn-go-runewidth
 982   (let ((commit "703b5e6b11ae25aeb2af9ebb5d5fdf8fa2575211")
 983         (version "0.0.4")
 984         (revision "1"))
 985     (package
 986       (name "go-github.com-mattn-go-runewidth")
 987       (version (git-version version revision commit))
 988       (source
 989        (origin
 990          (method git-fetch)
 991          (uri (git-reference
 992                (url "https://github.com/mattn/runewidth")
 993                (commit commit)))
 994          (file-name (git-file-name name version))
 995          (sha256
 996           (base32
 997            "0znpyz71gajx3g0j2zp63nhjj2c07g16885vxv4ykwnrfmzbgk4w"))))
 998       (build-system go-build-system)
 999       (arguments
1000        '(#:import-path "github.com/mattn/go-runewidth"))
1001       (synopsis "@code{runewidth} provides Go functions to work with string widths")
1002       (description
1003        "The @code{runewidth} library provides Go functions for padding,
1004 measuring and checking the width of strings, with support for East Asian
1005 text.")
1006       (home-page "https://github.com/mattn/runewidth")
1007       (license license:expat))))
1008
1009 (define-public docx2txt
1010   (package
1011     (name "docx2txt")
1012     (version "1.4")
1013     (source (origin
1014               (method url-fetch)
1015               (uri (string-append
1016                     "mirror://sourceforge/docx2txt/docx2txt/v"
1017                     version "/docx2txt-" version ".tgz"))
1018               (sha256
1019                (base32
1020                 "06vdikjvpj6qdb41d8wzfnyj44jpnknmlgbhbr1w215420lpb5xj"))))
1021     (build-system gnu-build-system)
1022     (inputs
1023      `(("unzip" ,unzip)
1024        ("perl" ,perl)))
1025     (arguments
1026      `(#:tests? #f                      ; No tests.
1027        #:make-flags (list (string-append "BINDIR="
1028                                          (assoc-ref %outputs "out") "/bin")
1029                           (string-append "CONFIGDIR="
1030                                          (assoc-ref %outputs "out") "/etc")
1031                           ;; Makefile seems to be a bit dumb at guessing.
1032                           (string-append "INSTALL=install")
1033                           (string-append "PERL=perl"))
1034        #:phases
1035        (modify-phases %standard-phases
1036          (delete 'configure)
1037          (add-after 'install 'fix-install
1038            (lambda* (#:key outputs inputs #:allow-other-keys)
1039              (let* ((out (assoc-ref outputs "out"))
1040                     (bin (string-append out "/bin"))
1041                     (config (string-append out "/etc/docx2txt.config"))
1042                     (unzip (assoc-ref inputs "unzip")))
1043                ;; According to INSTALL, the .sh wrapper can be skipped.
1044                (delete-file (string-append bin "/docx2txt.sh"))
1045                (rename-file (string-append bin "/docx2txt.pl")
1046                             (string-append bin "/docx2txt"))
1047                (substitute* config
1048                  (("config_unzip         => '/usr/bin/unzip',")
1049                   (string-append "config_unzip         => '"
1050                                  unzip
1051                                  "/bin/unzip',")))
1052                ;; Makefile is wrong.
1053                (chmod config #o644)
1054                #t))))))
1055     (synopsis "Recover text from @file{.docx} files, with good formatting")
1056     (description
1057      "@command{docx2txt} is a Perl based command line utility to convert
1058 Microsoft Office @file{.docx} documents to equivalent text documents.  Latest
1059 version supports following features during text extraction.
1060
1061 @itemize
1062 @item Character conversions; currency characters are converted to respective
1063 names like Euro.
1064 @item Capitalisation of text blocks.
1065 @item Center and right justification of text fitting in a line of
1066 (configurable) 80 columns.
1067 @item Horizontal ruler, line breaks, paragraphs separation, tabs.
1068 @item Indicating hyperlinked text along with the hyperlink (configurable).
1069 @item Handling (bullet, decimal, letter, roman) lists along with (attempt at)
1070 indentation.
1071 @end itemize\n")
1072     (home-page "http://docx2txt.sourceforge.net")
1073     (license license:gpl3+)))
1074
1075 (define-public html2text
1076   ;; Use commit directly to get the fixes to the installation phase
1077   ;; that are not in a release yet.
1078   (let ((commit "05364c1028026a87d6f45130a8e86e1ee67704d2")
1079         (revision "1"))
1080     (package
1081       (name "html2text")
1082       (version (git-version "2.0.1_pre" revision commit))
1083       (source
1084        (origin
1085          (method git-fetch)
1086          (uri (git-reference
1087                (url "https://github.com/grobian/html2text")
1088                (commit (string-append commit))))
1089          (file-name (git-file-name name version))
1090          (sha256
1091           (base32 "0n6pl0nijcn4z3p0dvf3gmvvpjq261pagnk84s9f78c4c55bw5cm"))))
1092       (build-system gnu-build-system)
1093       (arguments
1094        `(#:make-flags
1095          (list (string-append "PREFIX=" (assoc-ref %outputs "out")))
1096          #:phases
1097          (modify-phases %standard-phases
1098            (replace 'configure
1099              ;; The configure script is not from autotools and does not accept
1100              ;; ‘--style’ options.  There is no proper error handling.
1101              (lambda* (#:key outputs #:allow-other-keys)
1102                (invoke "./configure"
1103                        (string-append "CXX=" ,(cxx-for-target))))))))
1104       (home-page "https://github.com/grobian/html2text")
1105       (synopsis "HTML to plain text converter")
1106       (description
1107        "@code{html2text} is a command line utility that converts HTML
1108 documents into plain text.")
1109       (license (list license:bsd-4      ; cmp_nocase.cpp & sgml.h
1110                      license:gpl2+))))) ; everything else
1111
1112 (define-public odt2txt
1113   (package
1114     (name "odt2txt")
1115     (version "0.5")
1116     (source
1117       (origin
1118         (method git-fetch)
1119         (uri (git-reference
1120                (url "https://github.com/dstosberg/odt2txt/")
1121                (commit (string-append "v" version))))
1122         (file-name (git-file-name name version))
1123         (sha256
1124          (base32
1125           "0im3kzvhxkjlx57w6h13mc9584c74ma1dyymgvpq2y61av3gc35v"))))
1126     (build-system gnu-build-system)
1127     (arguments
1128      `(#:tests? #f ; no make check
1129        #:make-flags (list "CC=gcc"
1130                           (string-append "DESTDIR=" (assoc-ref %outputs "out")))
1131        #:phases
1132        (modify-phases %standard-phases
1133          ;; no configure script
1134          (delete 'configure))))
1135     (inputs
1136      `(("zlib" ,zlib)))
1137     (home-page "https://github.com/dstosberg/odt2txt/")
1138     (synopsis "Converter from OpenDocument Text to plain text")
1139     (description "odt2txt is a command-line tool which extracts the text out
1140 of OpenDocument Texts, as produced by OpenOffice.org, KOffice, StarOffice and
1141 others.
1142
1143 odt2txt can also extract text from some file formats similar to OpenDocument
1144 Text, such as OpenOffice.org XML (*.sxw), which was used by OpenOffice.org
1145 version 1.x and older StarOffice versions.  To a lesser extent, odt2txt may be
1146 useful to extract content from OpenDocument spreadsheets (*.ods) and
1147 OpenDocument presentations (*.odp).")
1148     (license license:gpl2)))
1149
1150 (define-public opencc
1151   (package
1152     (name "opencc")
1153     (version "1.1.1")
1154     (source
1155      (origin
1156        (method git-fetch)
1157        (uri (git-reference
1158              (url "https://github.com/BYVoid/OpenCC")
1159              (commit (string-append "ver." version))))
1160        (file-name (git-file-name name version))
1161        (sha256
1162         (base32
1163          "07y0pvxjlmpcnym229j87qcfwnfm7vi10dad0a20xy6as55a9j3d"))
1164        (modules '((guix build utils)))
1165        (snippet
1166         '(begin
1167            ;; TODO: Unbundle tclap, darts-clone, gtest
1168            (delete-file-recursively "deps/rapidjson-1.1.0") #t))))
1169     (build-system cmake-build-system)
1170     (arguments
1171      ;; Required to locate the install script properly.
1172      `(#:out-of-source? #f
1173        #:parallel-build? #f             ;occasionally failed.
1174        #:imported-modules
1175        (,@%cmake-build-system-modules
1176         (guix build python-build-system))
1177        #:modules ((guix build cmake-build-system)
1178                   ((guix build python-build-system) #:prefix python:)
1179                   (guix build utils))
1180        #:phases
1181        (modify-phases %standard-phases
1182          (add-after 'unpack 'prevent-rebuild-during-installation
1183            (lambda _
1184              (substitute* "python/setup.py"
1185                (("'build_py': BuildPyCommand,") ""))
1186              #t))
1187          (add-after 'unpack 'patch-3rd-party-references
1188            (lambda* (#:key inputs #:allow-other-keys)
1189              (let ((rapidjson (assoc-ref inputs "rapidjson")))
1190                (substitute* "src/CMakeLists.txt"
1191                  (("../deps/rapidjson-1.1.0")
1192                   (string-append rapidjson "/include/rapidjson")))
1193                #t)))
1194          (add-before 'configure 'patch-python-binding-installation
1195            (lambda* (#:key outputs inputs #:allow-other-keys)
1196              (let ((out (assoc-ref outputs "out")))
1197                (substitute* "python/opencc/__init__.py"
1198                  (("(_libopenccfile =).*$" _ prefix)
1199                   (format #f "~a os.path.join('~a/lib', _libopenccfilename)~%"
1200                           prefix out))
1201                  (("(_opencc_share_dir =).*$" _ prefix)
1202                   (format #f "~a '~a/share/opencc'~%" prefix out))))
1203              #t))
1204          (add-after 'install 'install-python-binding
1205            (lambda* (#:key outputs inputs #:allow-other-keys)
1206              (let* ((out (assoc-ref outputs "out"))
1207                     (dist (string-append
1208                            out "/lib/python"
1209                            ,(version-major+minor (package-version python))
1210                            "/site-packages")))
1211                (chdir "python")
1212                (mkdir-p dist)
1213                (setenv "PYTHONPATH"
1214                        (string-append dist ":" (getenv "GUIX_PYTHONPATH")))
1215                (invoke "python" "setup.py" "install"
1216                        "--root=/" "--single-version-externally-managed"
1217                        (string-append "--prefix=" out))
1218                #t)))
1219          (add-before 'install-python-binding 'enable-bytecode-determinism
1220            (assoc-ref python:%standard-phases 'enable-bytecode-determinism)))))
1221     (native-inputs
1222      `(("python" ,python-wrapper)
1223        ("rapidjson" ,rapidjson)
1224        ("python-setuptools" ,python-setuptools)
1225        ("python-wheel" ,python-wheel)))
1226     (home-page "https://github.com/BYVoid/OpenCC")
1227     (synopsis "Convert between Traditional Chinese and Simplified Chinese")
1228     (description "Open Chinese Convert (OpenCC) converts between Traditional
1229 Chinese and Simplified Chinese, supporting character-level conversion,
1230 phrase-level conversion, variant conversion, and regional idioms among
1231 Mainland China, Taiwan, and Hong-Kong.")
1232     (license license:asl2.0)))
1233
1234 (define-public nkf
1235   (let ((commit "08043eadf4abdddcf277842217e3c77a24740dc2")
1236         (revision "1"))
1237     (package
1238       (name "nkf")
1239       ;; The commits corresponding to specific versions are published
1240       ;; here:
1241       ;; https://ja.osdn.net/projects/nkf/scm/git/nkf/
1242       (version "2.1.5")
1243       (source (origin
1244                 (method git-fetch)
1245                 (uri (git-reference
1246                       (url "https://github.com/nurse/nkf")
1247                       (commit commit)))
1248                 (file-name (git-file-name name version))
1249                 (sha256
1250                  (base32
1251                   "0anw0knr1iy4p9w3d3b3pbwzh1c43p1i2q4c28kw9zviw8kx2rly"))))
1252       (build-system gnu-build-system)
1253       (arguments
1254        `(#:tests? #f ; test for perl module
1255          #:make-flags (list "CC=gcc" "CFLAGS=-O2 -Wall -pedantic"
1256                             (string-append "prefix=" %output)
1257                             "MKDIR=mkdir -p")
1258          #:phases
1259          (modify-phases %standard-phases
1260            (delete 'configure)))) ; No ./configure script
1261       (home-page "https://ja.osdn.net/projects/nkf/")
1262       (synopsis "Network Kanji Filter")
1263       (description "Nkf is yet another kanji code converter among networks,
1264 hosts and terminals.  It converts input kanji code to designated kanji code
1265 such as ISO-2022-JP, Shift_JIS, EUC-JP, UTF-8, UTF-16 or UTF-32.")
1266       (license license:zlib))))
1267
1268 (define-public python-pandocfilters
1269   (package
1270     (name "python-pandocfilters")
1271     (version "1.4.3")
1272     (source
1273      (origin
1274        (method url-fetch)
1275        (uri (pypi-uri "pandocfilters" version))
1276        (sha256
1277         (base32 "1sq675dg4barb5949xxz4d5gk2ly524hi1p1xgwb3d1l0nsznqxw"))))
1278     (build-system python-build-system)
1279     (home-page "https://github.com/jgm/pandocfilters")
1280     (synopsis "Python module for writing Pandoc filters")
1281     (description "Pandoc is a powerful utility to transform various
1282 input formats into a wide range of output formats.  To alter the
1283 exported output document, Pandoc allows the usage of filters, which
1284 are pipes that read a JSON serialization of the Pandoc AST from stdin,
1285 transform it in some way, and write it to stdout.  It allows therefore
1286 to alter the processing of Pandoc's supported input formats, for
1287 instance one can add new syntax elements to markdown, etc.
1288
1289 This package provides Python bindings.")
1290     (license license:bsd-3)))
1291
1292 (define-public aha
1293   (package
1294     (name "aha")
1295     (version "0.5.1")
1296     (source
1297      (origin
1298        (method git-fetch)
1299        (uri (git-reference
1300              (url "https://github.com/theZiz/aha")
1301              (commit version)))
1302        (sha256
1303         (base32 "1gywad0rvvz3c5balz8cxsnx0562hj2ngzqyr8zsy2mb4pn0lpgv"))
1304        (file-name (git-file-name name version))))
1305     (build-system gnu-build-system)
1306     (arguments
1307      `(#:phases
1308        (modify-phases %standard-phases
1309          (delete 'configure))
1310        #:make-flags (list (string-append "CC=" ,(cc-for-target))
1311                           (string-append "PREFIX="
1312                                          (assoc-ref %outputs "out")))
1313        #:tests? #f))                    ; no test suite
1314     (home-page "https://github.com/theZiz/aha")
1315     (synopsis "Converts terminal escape sequences to HTML")
1316     (description "@command{aha} (Ansi Html Adapter) converts ANSI escape sequences
1317 of a Unix terminal to HTML code.")
1318     (license (list license:lgpl2.0+ license:mpl1.1))))
1319
1320 (define-public go-github-com-errata-ai-vale
1321   (package
1322     (name "go-github-com-errata-ai-vale")
1323     (version "2.4.0")
1324     (source
1325      (origin
1326        (method git-fetch)
1327        (uri (git-reference
1328              (url "https://github.com/errata-ai/vale")
1329              (commit (string-append "v" version))))
1330        (sha256
1331         (base32 "0d07fwha2220m8j24h527xl0gnl3svvyaywflgk5292d6g49ach2"))
1332        (file-name (git-file-name name version))))
1333     (build-system go-build-system)
1334     (native-inputs
1335      `(("github.com/sp13/afero" ,go-github-com-spf13-afero)
1336        ("github.com/urfave/cli" ,go-github-com-urfave-cli)
1337        ("github.com/olekukonko/tablewriter" ,go-github-com-olekukonko-tablewriter)
1338        ("github.com/mitchellh/mapstructure" ,go-github-com-mitchellh-mapstructure)))
1339     (arguments
1340      `(#:import-path "github.com/errata-ai/vale"
1341        #:install-source? #f))
1342     (home-page "https://github.com/errata-ai/vale")
1343     (synopsis "Fully customizable syntax-aware linter that focuses on your style")
1344     (description
1345      "Vale is a fully extensible linter that focuses on your own writing style
1346 by making use of rules in individual YAML files.  It is syntax-aware on markup
1347 languages such as HTML, Markdown, Asciidoc, and reStructuredText.  The community
1348 around it also has a list of style guides implemented with Vale in
1349 @url{https://github.com/errata-ai/styles, their styles repo}.")
1350     (license license:expat)))