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 Nikita <nikita@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, 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 ;;;
  24 ;;; This file is part of GNU Guix.
  25 ;;;
  26 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  27 ;;; under the terms of the GNU General Public License as published by
  28 ;;; the Free Software Foundation; either version 3 of the License, or (at
  29 ;;; your option) any later version.
  30 ;;;
  31 ;;; GNU Guix is distributed in the hope that it will be useful, but
  32 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  33 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  34 ;;; GNU General Public License for more details.
  35 ;;;
  36 ;;; You should have received a copy of the GNU General Public License
  37 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  38
  39 (define-module (gnu packages textutils)
  40   #:use-module ((guix licenses) #:prefix license:)
  41   #:use-module (guix packages)
  42   #:use-module (guix download)
  43   #:use-module (guix git-download)
  44   #:use-module (guix build-system ant)
  45   #:use-module (guix build-system gnu)
  46   #:use-module (guix build-system go)
  47   #:use-module (guix build-system cmake)
  48   #:use-module (guix build-system python)
  49   #:use-module (gnu packages)
  50   #:use-module (gnu packages autotools)
  51   #:use-module (gnu packages base)
  52   #:use-module (gnu packages compression)
  53   #:use-module (gnu packages gcc)
  54   #:use-module (gnu packages gettext)
  55   #:use-module (gnu packages java)
  56   #:use-module (gnu packages ncurses)
  57   #:use-module (gnu packages pcre)
  58   #:use-module (gnu packages perl)
  59   #:use-module (gnu packages pkg-config)
  60   #:use-module (gnu packages python)
  61   #:use-module (gnu packages python-xyz)
  62   #:use-module (gnu packages readline)
  63   #:use-module (gnu packages slang)
  64   #:use-module (gnu packages web))
  65
  66 (define-public dos2unix
  67   (package
  68     (name "dos2unix")
  69     (version "7.4.1")
  70     (source
  71      (origin
  72        (method url-fetch)
  73        (uri (string-append "https://waterlan.home.xs4all.nl/dos2unix/"
  74                            "dos2unix-" version ".tar.gz"))
  75        (sha256
  76         (base32 "08w6yywzirsxq8bh87jycvvw922ybhc2l426j2iqzliyn1h8mm8w"))))
  77     (build-system gnu-build-system)
  78     (arguments
  79      '(#:make-flags (list "CC=gcc"
  80                           (string-append "prefix=" (assoc-ref %outputs "out")))
  81        #:phases
  82        (modify-phases %standard-phases
  83          (delete 'configure)))) ; no configure script
  84     (native-inputs
  85      `(("gettext" ,gettext-minimal)
  86        ("perl" ,perl)))
  87     (home-page "https://waterlan.home.xs4all.nl/dos2unix.html")
  88     (synopsis "DOS/Mac to Unix and vice versa text file format converter")
  89     (description
  90      "dos2unix is a tool to convert line breaks in a text file from Unix format
  91 to DOS format and vice versa.")
  92     (license license:bsd-2)))
  93
  94 (define-public recode
  95   (package
  96     (name "recode")
  97     (version "3.7.6")
  98     (source
  99      (origin
 100        (method url-fetch)
 101        (uri (string-append "https://github.com/rrthomas/recode/releases/"
 102                            "download/v" version "/recode-" version ".tar.gz"))
 103        (sha256
 104         (base32 "0m59sd1ca0zw1aydpc3m8sw03nc885knmccqryg7byzmqs585ia6"))))
 105     (build-system gnu-build-system)
 106     (native-inputs
 107      `(("python" ,python)
 108        ("python-cython" ,python-cython)))
 109     (home-page "https://github.com/rrthomas/recode")
 110     (synopsis "Text encoding converter")
 111     (description "The Recode library converts files between character sets and
 112 usages.  It recognises or produces over 200 different character sets (or about
 113 300 if combined with an iconv library) and transliterates files between almost
 114 any pair.  When exact transliteration are not possible, it gets rid of
 115 offending characters or falls back on approximations.  The recode program is a
 116 handy front-end to the library.")
 117     (license license:gpl3+)))
 118
 119 (define-public enca
 120   (package
 121     (name "enca")
 122     (version "1.19")
 123     (source
 124      (origin
 125        (method git-fetch)
 126        (uri (git-reference
 127               (url "https://github.com/nijel/enca")
 128               (commit version)))
 129        (file-name (git-file-name name version))
 130        (sha256
 131         (base32 "19q7cwwxmmk5j9438bsqdpjvdjawsd3zmw1zyqgi7s4m0rasr3ah"))))
 132     (build-system gnu-build-system)
 133     ;; enca-1.19 tests fail with recent recode.
 134     ;(inputs `(("recode" ,recode)))
 135     (home-page "https://github.com/nijel/enca")
 136     (synopsis "Text encoding detection tool")
 137     (description "Enca (Extremely Naive Charset Analyser) consists of libenca,
 138 an encoding detection library, and enca, a command line frontend, integrating
 139 libenca and several charset conversion libraries and tools.")
 140     (license license:gpl2)))
 141
 142 (define-public utf8proc
 143   (package
 144     (name "utf8proc")
 145     (version "2.5.0")
 146     (source
 147      (origin
 148        (method git-fetch)
 149        (uri (git-reference
 150              (url "https://github.com/JuliaStrings/utf8proc")
 151              (commit (string-append "v" version))))
 152        (file-name (git-file-name name version))
 153        (sha256
 154         (base32 "1xlkazhdnja4lksn5c9nf4bln5gjqa35a8gwlam5r0728w0h83qq"))))
 155     (build-system gnu-build-system)
 156     (native-inputs
 157      (let ((UNICODE_VERSION "13.0.0"))  ; defined in data/Makefile
 158        ;; Test data that is otherwise downloaded with curl.
 159        `(("NormalizationTest.txt"
 160           ,(origin
 161              (method url-fetch)
 162              (uri (string-append "https://www.unicode.org/Public/"
 163                                  UNICODE_VERSION "/ucd/NormalizationTest.txt"))
 164              (sha256
 165               (base32 "07g0ya4f6zfzvpp24ccxkb2yq568kh83gls85rjl950nv5fya3nn"))))
 166          ("GraphemeBreakTest.txt"
 167           ,(origin
 168              (method url-fetch)
 169              (uri (string-append "https://www.unicode.org/Public/"
 170                                  UNICODE_VERSION
 171                                  "/ucd/auxiliary/GraphemeBreakTest.txt"))
 172              (sha256
 173               (base32 "07f8rrvcsq4pibdz6zxggxy8w7zjjqyw2ggclqlhalyv45yv7prj"))))
 174
 175          ;; For tests.
 176          ("perl" ,perl))))
 177     (arguments
 178      '(#:make-flags (list "CC=gcc"
 179                           (string-append "prefix=" (assoc-ref %outputs "out")))
 180        #:phases
 181        (modify-phases %standard-phases
 182          (delete 'configure)
 183          (add-before 'check 'check-data
 184            (lambda* (#:key inputs #:allow-other-keys)
 185              (for-each (lambda (i)
 186                          (copy-file (assoc-ref inputs i)
 187                                     (string-append "data/" i)))
 188                        '("NormalizationTest.txt" "GraphemeBreakTest.txt"))
 189              (substitute* "data/GraphemeBreakTest.txt"
 190                (("÷") "/")
 191                (("×") "+"))
 192              #t)))))
 193     (home-page "https://juliastrings.github.io/utf8proc/")
 194     (synopsis "C library for processing UTF-8 Unicode data")
 195     (description "utf8proc is a small C library that provides Unicode
 196 normalization, case-folding, and other operations for data in the UTF-8
 197 encoding, supporting Unicode version 9.0.0.")
 198     (license license:expat)))
 199
 200 (define-public libconfuse
 201   (package
 202     (name "libconfuse")
 203     (version "3.2.2")
 204     (source
 205      (origin
 206        (method url-fetch)
 207        (uri (string-append "https://github.com/martinh/libconfuse/"
 208                            "releases/download/v" version
 209                            "/confuse-" version ".tar.xz"))
 210        (sha256
 211         (base32 "02r1mmzik2m0iigbc2da3y754vj24i18r3ml5p2wzs027mjhn959"))))
 212     (build-system gnu-build-system)
 213     (home-page "https://github.com/martinh/libconfuse")
 214     (synopsis "Configuration file parser library")
 215     (description "libconfuse is a configuration file parser library.  It
 216 supports sections and (lists of) values (strings, integers, floats, booleans
 217 or other sections), as well as some other features (such as
 218 single/double-quoted strings, environment variable expansion, functions and
 219 nested include statements).")
 220     (license license:isc)))
 221
 222 (define-public libgtextutils
 223   (package
 224     (name "libgtextutils")
 225     (version "0.7")
 226     (source
 227      (origin
 228        (method url-fetch)
 229        (uri (string-append
 230              "https://github.com/agordon/libgtextutils/releases/download/"
 231              version "/libgtextutils-" version ".tar.gz"))
 232        (sha256
 233         (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
 234     (build-system gnu-build-system)
 235     (arguments
 236      '(#:phases
 237        (modify-phases %standard-phases
 238          (replace 'bootstrap
 239            (lambda _ (invoke "sh" "reconf"))))))
 240     (native-inputs
 241      `(("autoconf" ,autoconf)
 242        ("automake" ,automake)
 243        ("gcc" ,gcc-5) ;; doesn't build with later versions
 244        ("libtool" ,libtool)))
 245     (home-page "https://github.com/agordon/libgtextutils")
 246     (synopsis "Gordon's text utils library")
 247     (description
 248      "libgtextutils is a text utilities library used by the fastx toolkit from
 249 the Hannon Lab.")
 250     (license license:agpl3+)))
 251
 252 (define-public cityhash
 253   (let ((commit "8af9b8c"))
 254     (package
 255       (name "cityhash")
 256       (version (string-append "1.1-2." commit))
 257       (source (origin
 258                 (method git-fetch)
 259                 (uri (git-reference
 260                       (url "https://github.com/google/cityhash.git")
 261                       (commit commit)))
 262                 (file-name (string-append name "-" version ".tar.gz"))
 263                 (sha256
 264                  (base32
 265                   "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
 266       (build-system gnu-build-system)
 267       (arguments
 268        '(#:make-flags (list "CXXFLAGS=-g -O3")
 269          #:phases
 270          (modify-phases %standard-phases
 271            ;; citycrc is not installed by default but is used by some
 272            ;; programs.
 273            (add-after 'install 'install-citycrc
 274              (lambda* (#:key outputs #:allow-other-keys)
 275                (let* ((out (assoc-ref outputs "out"))
 276                       (include (string-append out "/include")))
 277                  (install-file "src/citycrc.h" include))
 278                #t)))))
 279       (home-page "https://github.com/google/cityhash")
 280       (synopsis "C++ hash functions for strings")
 281       (description
 282        "CityHash provides hash functions for strings.  The functions mix the
 283 input bits thoroughly but are not suitable for cryptography.")
 284       (license license:expat))))
 285
 286 (define-public ustr
 287   (package
 288     (name "ustr")
 289     (version "1.0.4")
 290     (source (origin
 291               (method url-fetch)
 292               (uri (string-append "http://www.and.org/ustr/" version
 293                                   "/ustr-" version ".tar.bz2"))
 294               (sha256
 295                (base32
 296                 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
 297               (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
 298     (build-system gnu-build-system)
 299     (arguments
 300      `(#:make-flags
 301        (list "CC=gcc"
 302              "HIDE="
 303              ;; Override "/sbin/ldconfig" with "echo" because we don't need
 304              ;; "ldconfig".
 305              "LDCONFIG=echo"
 306              (string-append "prefix=" (assoc-ref %outputs "out"))
 307              "all-shared")
 308        #:phases
 309        (modify-phases %standard-phases
 310          (add-after 'unpack 'disable-check-for-stdint
 311            (lambda _
 312              ;; Of course we have stdint.h, just not in /usr/include
 313              (substitute* '("Makefile"
 314                             "ustr-import.in")
 315                (("-f \"/usr/include/stdint.h\"") "-z \"\""))
 316              #t))
 317          ;; No configure script
 318          (delete 'configure))))
 319     (home-page "http://www.and.org/ustr/")
 320     (synopsis "String library with very low memory overhead")
 321     (description
 322      "Ustr is a string library for C with very low memory overhead.")
 323     ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
 324     ;; LGPL, etc. ... if you need another license to help compatibility, just
 325     ;; ask for it.  It's basically public domain, without all the legal
 326     ;; problems for everyone that trying to make something public domain
 327     ;; entails."
 328     (license license:public-domain)))
 329
 330 (define-public ascii2binary
 331   (package
 332     (name "ascii2binary")
 333     (version "2.14")
 334     (source
 335      (origin
 336        (method url-fetch)
 337        (uri (string-append "http://billposer.org/Software/Downloads/"
 338                            "ascii2binary-" version ".tar.bz2"))
 339        (sha256
 340         (base32 "0dc9fxcdmppbs9s06jvq61zbk552laxps0xyk098gj41697ihd96"))))
 341     (build-system gnu-build-system)
 342     (native-inputs
 343      `(("gettext" ,gettext-minimal)))
 344     (home-page "https://billposer.org/Software/a2b.html")
 345     (synopsis "Convert between ASCII, hexadecimal and binary representations")
 346     (description "The two programs are useful for generating test data, for
 347 inspecting binary files, and for interfacing programs that generate textual
 348 output to programs that require binary input and conversely.  They can also be
 349 useful when it is desired to reformat numbers.
 350
 351 @itemize
 352
 353 @item @command{ascii2binary} reads input consisting of ascii or hexadecimal
 354    representation numbers separated by whitespace and produces as output
 355    the binary equivalents.  The type and precision of the binary output
 356    is selected using command line flags.
 357
 358 @item @command{binary2ascii} reads input consisting of binary numbers
 359    and converts them to their ascii or hexadecimal representation.
 360    Command line flags specify the type and size of the binary numbers
 361    and provide control over the format of the output.
 362    Unsigned integers may be written out in binary, octal, decimal,
 363    or hexadecimal.
 364
 365    Signed integers may be written out only in binary or decimal.  Floating
 366    point numbers may be written out only decimal, either in standard or
 367    scientific notation.  (If you want to examine the binary representation
 368    of floating point numbers, just treat the input as a sequence of unsigned
 369    characters.)
 370
 371 @end itemize")
 372     (license license:gpl3)))
 373
 374 (define-public uniutils
 375   (package
 376     (name "uniutils")
 377     (version "2.27")
 378     (source
 379      (origin
 380        (method url-fetch)
 381        (uri (string-append "http://billposer.org/Software/Downloads/"
 382                            "uniutils-" version ".tar.bz2"))
 383        (sha256
 384         (base32 "19w1510w87gx7n4qy3zsb0m467a4rn5scvh4ajajg7jh6x5xri08"))))
 385     (build-system gnu-build-system)
 386     (arguments
 387      '(#:configure-flags '("--disable-dependency-tracking")
 388        #:phases
 389        (modify-phases %standard-phases
 390          (add-after 'build 'fix-paths
 391            (lambda* (#:key outputs inputs #:allow-other-keys)
 392              (let ((out (assoc-ref outputs "out"))
 393                    (a2b (assoc-ref inputs "ascii2binary"))
 394                    (iconv (assoc-ref inputs "libiconv")))
 395                (substitute* "utf8lookup"
 396                  (("^ascii2binary ") (string-append a2b "/bin/ascii2binary "))
 397                  (("^uniname ") (string-append out "/bin/uniname "))
 398                  (("^iconv ") (string-append iconv "/bin/iconv ")))
 399              #t))))))
 400     (inputs
 401      `(("ascii2binary" ,ascii2binary)
 402        ("libiconv" ,libiconv)))
 403     (home-page "https://billposer.org/Software/unidesc.html")
 404     (synopsis "Find out what is in a Unicode file")
 405     (description "Useful tools when working with Unicode files when one
 406 doesn't know the writing system, doesn't have the necessary font, needs to
 407 inspect invisible characters, needs to find out whether characters have been
 408 combined or in what order they occur, or needs statistics on which characters
 409 occur.
 410
 411 @itemize
 412
 413 @item @command{uniname} defaults to printing the character offset of each
 414 character, its byte offset, its hex code value, its encoding, the glyph
 415 itself, and its name.  It may also be used to validate UTF-8 input.
 416
 417 @item @command{unidesc} reports the character ranges to which different
 418 portions of the text belong.   It can also be used to identify Unicode encodings
 419 (e.g. UTF-16be) flagged by magic numbers.
 420
 421 @item @command{unihist} generates a histogram of the characters in its input.
 422
 423 @item @command{ExplicateUTF8} is intended for debugging or for learning about
 424 Unicode.  It determines and explains the validity of a sequence of bytes as a
 425 UTF8 encoding.
 426
 427 @item @command{utf8lookup} provides a handy way to look up Unicode characters
 428 from the command line.
 429
 430 @item @command{unireverse} reverse each line of UTF-8 input
 431 character-by-character.
 432
 433 @end itemize")
 434     (license license:gpl3)))
 435
 436 (define-public libconfig
 437   (package
 438     (name "libconfig")
 439     (version "1.7.2")
 440     (home-page "https://hyperrealm.github.io/libconfig/")
 441     (source (origin
 442               (method url-fetch)
 443               (uri (string-append home-page "/dist/libconfig-"
 444                                   version ".tar.gz"))
 445               (sha256
 446                (base32
 447                 "1ngs2qx3cx5cbwinc5mvadly0b5n7s86zsc68c404czzfff7lg3w"))))
 448     (build-system gnu-build-system)
 449     (synopsis "C/C++ configuration file library")
 450     (description
 451      "Libconfig is a simple library for manipulating structured configuration
 452 files.  This file format is more compact and more readable than XML.  And
 453 unlike XML, it is type-aware, so it is not necessary to do string parsing in
 454 application code.")
 455     (license license:lgpl2.1+)))
 456
 457 (define-public pfff
 458   (package
 459     (name "pfff")
 460     (version "1.0")
 461     (source (origin
 462               (method git-fetch)
 463               (uri (git-reference
 464                      (url "https://github.com/pfff/pfff")
 465                      (commit (string-append "v" version))))
 466               (file-name (git-file-name name version))
 467               (sha256
 468                (base32
 469                 "1nxkfm7zliq3rmr7yp871sppwfnz71iz364m2sgazny71pzykggc"))))
 470     (build-system cmake-build-system)
 471     (home-page "https://biit.cs.ut.ee/pfff/")
 472     (synopsis "Probabilistic fast file fingerprinting tool")
 473     (description
 474      "pfff is a tool for calculating a compact digital fingerprint of a file
 475 by sampling randomly from the file instead of reading it in full.
 476 Consequently, the computation has a flat performance characteristic,
 477 correlated with data variation rather than file size.  pfff can be as reliable
 478 as existing hashing techniques, with provably negligible risk of collisions.")
 479     (license license:bsd-3)))
 480
 481 (define-public oniguruma
 482   (package
 483     (name "oniguruma")
 484     (version "6.9.5-rev1")
 485     (source (origin
 486               (method url-fetch)
 487               (uri (string-append "https://github.com/kkos/"
 488                                   "oniguruma/releases/download/v"
 489                                   ;; If there is a "-" in the version, convert
 490                                   ;; to underscore for this part of the URI.
 491                                   (string-map (lambda (c) (if (char=? #\- c) #\_ c))
 492                                               version)
 493                                   "/onig-" version ".tar.gz"))
 494               (sha256
 495                (base32
 496                 "17m92k1n6bvza6m35fpd5g36zwpwm3hfz3478iwj5bvj2sfq8g6k"))))
 497     (build-system gnu-build-system)
 498     (arguments '(#:configure-flags '("--disable-static")))
 499     (home-page "https://github.com/kkos/oniguruma")
 500     (synopsis "Regular expression library")
 501     (description "Oniguruma is a regular expressions library.  The special
 502 characteristic of this library is that different character encoding for every
 503 regular expression object can be specified.")
 504     (license license:bsd-2)))
 505
 506 (define-public antiword
 507   (package
 508     (name "antiword")
 509     (version "0.37")
 510     (source (origin
 511               (method url-fetch)
 512               (uri (string-append "http://www.winfield.demon.nl/linux"
 513                                   "/antiword-" version ".tar.gz"))
 514               (sha256
 515                (base32
 516                 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
 517               (patches (search-patches "antiword-CVE-2014-8123.patch"))))
 518     (build-system gnu-build-system)
 519     (arguments
 520      `(#:tests? #f ; There are no tests
 521        #:make-flags
 522        (list "-f" "Makefile.Linux"
 523              (string-append "GLOBAL_INSTALL_DIR="
 524                             (assoc-ref %outputs "out") "/bin")
 525              (string-append "GLOBAL_RESOURCES_DIR="
 526                             (assoc-ref %outputs "out") "/share/antiword"))
 527        #:phases
 528        (modify-phases %standard-phases
 529          (replace 'configure
 530            (lambda* (#:key outputs #:allow-other-keys)
 531              ;; Ensure that mapping files can be found in the actual package
 532              ;; data directory.
 533              (substitute* "antiword.h"
 534                (("/usr/share/antiword")
 535                 (string-append (assoc-ref outputs "out") "/share/antiword")))
 536              #t))
 537          (replace 'install
 538            (lambda* (#:key make-flags #:allow-other-keys)
 539              (apply invoke "make" `("global_install" ,@make-flags)))))))
 540     (home-page "http://www.winfield.demon.nl/")
 541     (synopsis "Microsoft Word document reader")
 542     (description "Antiword is an application for displaying Microsoft Word
 543 documents.  It can also convert the document to PostScript or XML.  Only
 544 documents made by MS Word version 2 and version 6 or later are supported.  The
 545 name comes from: \"The antidote against people who send Microsoft Word files
 546 to everybody, because they believe that everybody runs Windows and therefore
 547 runs Word\".")
 548     (license license:gpl2+)))
 549
 550 (define-public catdoc
 551   (package
 552     (name "catdoc")
 553     (version "0.95")
 554     (source (origin
 555               (method url-fetch)
 556               (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
 557                                   "catdoc-" version ".tar.gz"))
 558               (patches (search-patches "catdoc-CVE-2017-11110.patch"))
 559               (sha256
 560                (base32
 561                 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
 562     (build-system gnu-build-system)
 563     ;; TODO: Also build `wordview` which requires `tk` – make a separate
 564     ;; package for this.
 565     (arguments
 566      '(#:tests? #f ; There are no tests
 567        #:configure-flags '("--disable-wordview")
 568        #:phases
 569        (modify-phases %standard-phases
 570          (add-before 'install 'fix-install
 571            (lambda* (#:key outputs #:allow-other-keys)
 572              (let ((out (assoc-ref outputs "out")))
 573                (mkdir-p (string-append out "/share/man/man1"))))))))
 574     (home-page "https://www.wagner.pp.ru/~vitus/software/catdoc/")
 575     (synopsis "MS-Word to TeX or plain text converter")
 576     (description "@command{catdoc} extracts text from MS-Word files, trying to
 577 preserve as many special printable characters as possible.  It supports
 578 everything up to Word-97. Also supported are MS Write documents and RTF files.
 579
 580 @command{catdoc} does not preserve complex word formatting, but it can
 581 translate some non-ASCII characters into TeX escape codes.  It's goal is to
 582 extract plain text and allow you to read it and, probably, reformat with TeX,
 583 according to TeXnical rules.
 584
 585 This package also provides @command{xls2csv}, which extracts data from Excel
 586 spreadsheets and outputs it in comma-separated-value format, and
 587 @command{catppt}, which extracts data from PowerPoint presentations.")
 588     (license license:gpl2+)))
 589
 590 (define-public utfcpp
 591   (package
 592     (name "utfcpp")
 593     (version "2.3.5")
 594     (source (origin
 595               (method git-fetch)
 596               (uri (git-reference
 597                      (url "https://github.com/nemtrif/utfcpp")
 598                      (commit (string-append "v" version))))
 599               (file-name (git-file-name name version))
 600               (sha256
 601                (base32
 602                 "1gr98d826z6wa58r1s5i7rz7q2x3r31v7zj0pjjlrc7gfxwklr4s"))))
 603     (build-system cmake-build-system)
 604     (arguments
 605      `(#:out-of-source? #f
 606        #:phases
 607        (modify-phases %standard-phases
 608          (replace 'install              ; no install target
 609            (lambda* (#:key outputs #:allow-other-keys)
 610              (let* ((out (assoc-ref outputs "out"))
 611                     (include (string-append out "/include"))
 612                     (doc (string-append out "/share/doc/" ,name)))
 613                (copy-recursively "source" include)
 614                (install-file "README.md" doc)
 615                #t))))))
 616     (home-page "https://github.com/nemtrif/utfcpp")
 617     (synopsis "Portable C++ library for handling UTF-8")
 618     (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
 619 in a portable way.")
 620     (license license:boost1.0)))
 621
 622 (define-public dbacl
 623   (package
 624     (name "dbacl")
 625     (version "1.14.1")
 626     (source
 627      (origin
 628        (method url-fetch)
 629        (uri (string-append "mirror://sourceforge/dbacl/dbacl/" version "/"
 630                            "dbacl-" version ".tar.gz"))
 631        (sha256
 632         (base32 "1gas0112wqjvwn9qg3hxnawk7h3prr0w9b2h68f3p1ifd1kzn3gz"))
 633        (patches (search-patches "dbacl-include-locale.h.patch"))))
 634     (build-system gnu-build-system)
 635     (arguments
 636      `(#:make-flags
 637        (list
 638         (string-append "-I" (assoc-ref %build-inputs "slang")
 639                        "/include/slang")
 640         (string-append "-I" (assoc-ref %build-inputs "ncurses")
 641                        "/include/ncurses"))
 642        #:phases
 643        (modify-phases %standard-phases
 644          (add-after 'unpack 'delete-sample6-and-japanese
 645            (lambda _
 646              (substitute* "doc/Makefile.am"
 647                (("sample6.txt") "")
 648                (("japanese.txt") ""))
 649              (delete-file "doc/sample6.txt")
 650              (delete-file "doc/japanese.txt")
 651              (substitute* (list "src/tests/Makefile.am"
 652                                 "src/tests/Makefile.in")
 653                (("dbacl-jap.shin") "")
 654                (("dbacl-jap.sh") ""))
 655              #t))
 656          (add-after 'unpack 'delete-test
 657            ;; See comments about the license.
 658            (lambda _
 659              (delete-file "src/tests/dbacl-jap.shin")
 660              #t))
 661          (add-after 'unpack 'fix-test-files
 662            (lambda* (#:key inputs outputs #:allow-other-keys)
 663              (let* ((out (assoc-ref outputs "out"))
 664                     (bin (string-append out "/bin")))
 665                (substitute* (find-files "src/tests/" "\\.shin$")
 666                  (("PATH=/bin:/usr/bin")
 667                   "#PATH=/bin:/usr/bin")
 668                  (("diff") (string-append (which "diff")))
 669                  (("tr") (string-append (which "tr"))))
 670                #t)))
 671          (replace 'bootstrap
 672            (lambda _
 673              (invoke "autoreconf" "-vif")
 674              #t)))))
 675     (inputs
 676      `(("ncurses" ,ncurses)
 677        ("perl" ,perl)
 678        ("readline" ,readline)
 679        ("slang" ,slang)))
 680     (native-inputs
 681      `(("libtool" ,libtool)
 682        ("autoconf" ,autoconf)
 683        ("automake" ,automake)
 684        ("pkg-config" ,pkg-config)))
 685     (home-page "https://www.lbreyer.com/dbacl.html")
 686     (synopsis "Bayesian text and email classifier")
 687     (description
 688      "dbacl is a fast Bayesian text and email classifier.  It builds a variety
 689 of language models using maximum entropy (minimum divergence) principles, and
 690 these can then be used to categorize input data automatically among multiple
 691 categories.")
 692     ;; The software is licensed as GPLv3 or later, but
 693     ;; includes various sample texts in the doc dir:
 694     ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
 695     ;;   by Mark Twain.
 696     ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
 697     ;; - sample6.txt is a forwarded email, copyright unknown.
 698     ;;   Guix does exclude sample6.txt.
 699     ;; - japanese.txt is a Japanese unoffical translation of the
 700     ;;   GNU General Public License, (c) by the Free Software Foundation.
 701     ;;   Guix excludes this file.
 702     (license (list license:gpl3+ license:public-domain))))
 703
 704 (define-public dotconf
 705   (package
 706     (name "dotconf")
 707     (version "1.3")
 708     (source (origin
 709               (method git-fetch)
 710               (uri (git-reference
 711                     (url "https://github.com/williamh/dotconf.git")
 712                     (commit (string-append "v" version))))
 713               (file-name (git-file-name name version))
 714               (sha256
 715                (base32
 716                 "1sc95hw5k2xagpafny0v35filmcn05k1ds5ghkldfpf6xw4hakp7"))))
 717     (build-system gnu-build-system)
 718     (arguments `(#:tests? #f))  ; FIXME maketest.sh does not work.
 719     (native-inputs
 720      `(("autoconf" ,autoconf)
 721        ("automake" ,automake)
 722        ("libtool" ,libtool)))
 723     (home-page "https://github.com/williamh/dotconf")
 724     (synopsis "Configuration file parser library")
 725     (description
 726      "C library for creating and parsing configuration files.")
 727     (license (list license:lgpl2.1         ; Main distribution.
 728                    license:asl1.1))))      ; src/readdir.{c,h}
 729
 730 (define-public drm-tools
 731   (package
 732     (name "drm-tools")
 733     (version "1.1.33")
 734     (source (origin
 735               (method url-fetch)
 736               (uri (string-append "mirror://sourceforge/drmtools/drm_tools-"
 737                                   version ".tar.gz"))
 738               (sha256
 739                (base32
 740                 "187zbxw21zcg8gpyc13gxlycfw0n05a6rmqq6im5wr9zk1v1wj80"))))
 741     (build-system cmake-build-system)
 742     (arguments
 743      `(#:tests? #f                      ;the test suite fails
 744        #:phases (modify-phases %standard-phases
 745                   (add-after 'unpack 'set-install-prefixes
 746                     (lambda* (#:key outputs #:allow-other-keys)
 747                       (let* ((out (assoc-ref outputs "out")))
 748                         (substitute* "CMakeLists.txt"
 749                           (("tmp/testinstall")
 750                            (string-drop out 1))
 751                           (("/man/man1")
 752                            "/share/man/man1"))
 753                         #t)))
 754                   (add-after 'unpack 'adjust-test-paths
 755                     (lambda _
 756                       (substitute* '("test_extract_increment.sh"
 757                                      "test_extract_features.sh"
 758                                      "test_extract_features2.sh"
 759                                      "test_dmath.sh")
 760                         (("\\./extract") "extract")
 761                         (("\\./dmath") "dmath")
 762                         (("/usr/local/bin/") "")
 763                         (("/bin/rm") "rm")
 764                         (("/bin/cp") "cp"))
 765                       #t))
 766                   (delete 'check)
 767                   ;; The produced binaries are written directly to %output/bin.
 768                   (delete 'install)
 769                   (add-after 'build 'check
 770                     (lambda* (#:key outputs tests? #:allow-other-keys)
 771                       (when tests?
 772                         (let* ((out (assoc-ref outputs "out"))
 773                                (bin (string-append out "/bin")))
 774                           (setenv "PATH" (string-append bin ":"
 775                                                         (getenv "PATH")))
 776                           (with-directory-excursion
 777                               (format #f "../drm_tools-~a" ,version)
 778                             (invoke "sh" "test_all.sh")))))))))
 779     (native-inputs `(("which" ,which))) ;for tests
 780     (inputs `(("pcre" ,pcre)))
 781     (home-page "http://drmtools.sourceforge.net/")
 782     (synopsis "Utilities to manipulate text and binary files")
 783     (description "The drm_tools package contains the following commands:
 784 @table @command
 785 @item accudate
 786 An extended version of the \"date\" program that has sub-second accuracy.
 787 @item binformat
 788 Format complex binary data into text.
 789 @item binload
 790 Load data into a binary file using simple commands from the input.
 791 @item binorder
 792 Sort, merge, search, retrieve or generate test data consisting of fixed size
 793 binary records.
 794 @item binreplace
 795 Find or find/replace in binary files.
 796 @item binsplit
 797 Split test data consisting of fixed size binary records into one or more
 798 output streams.
 799 @item chardiff
 800 Find changes between two files at the character level.  Unlike \"diff\", it
 801 lists just the characters that differ, so if the 40,000th character is
 802 different only that one character will be shown, not the entire line.
 803 @item columnadd
 804 Add columns of integers, decimals, and/or times.
 805 @item datasniffer
 806 A utility for formatting binary data dumps.
 807 @item dmath
 808 Double precision interactive command line math calculator.
 809 @item extract
 810 Extract and emit data from text files based on character or token position.
 811 @item execinput
 812 A utility that reads from STDIN and executes each line as a command in a
 813 sub-process.
 814 @item indexed_text
 815 A utility for rapid retrieval of text by line numbers, in any order, from a
 816 text file.
 817 @item mdump
 818 Format binary data.
 819 @item msgqueue
 820 Create message queues and send/receive messages.
 821 @item mbin
 822 @itemx mbout
 823 Multiple buffer in and out.  Used for buffering a lot of data between a slow
 824 device and a fast device.  Mostly for buffering streaming tape drives for use
 825 with slower network connections, so that streaming is maintained as much as
 826 possible to minimize wear on the tape device.
 827 @item pockmark
 828 Corrupt data streams - useful for testing error correction and data recovery.
 829 @item tarsieve
 830 Filter, list, or split a tar file.
 831 @end table")
 832     (license license:gpl2+)))
 833
 834 (define-public java-rsyntaxtextarea
 835   (package
 836     (name "java-rsyntaxtextarea")
 837     (version "2.6.1")
 838     (source (origin
 839               (method url-fetch)
 840               (uri (string-append "https://github.com/bobbylight/"
 841                                   "RSyntaxTextArea/archive/"
 842                                   version ".tar.gz"))
 843               (file-name (string-append name "-" version ".tar.gz"))
 844               (sha256
 845                (base32
 846                 "0c5mqg2klj5rvf8fhycrli8rf6s37l9p7a8knw9gpp65r1c120q2"))))
 847     (build-system ant-build-system)
 848     (arguments
 849      `(;; FIXME: some tests fail because locale resources cannot be found.
 850        ;; Even when I add them to the class path,
 851        ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
 852        #:tests? #f
 853        #:jar-name "rsyntaxtextarea.jar"))
 854     (native-inputs
 855      `(("java-junit" ,java-junit)
 856        ("java-hamcrest-core" ,java-hamcrest-core)))
 857     (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
 858     (synopsis "Syntax highlighting text component for Java Swing")
 859     (description "RSyntaxTextArea is a syntax highlighting, code folding text
 860 component for Java Swing.  It extends @code{JTextComponent} so it integrates
 861 completely with the standard @code{javax.swing.text} package.  It is fast and
 862 efficient, and can be used in any application that needs to edit or view
 863 source code.")
 864     (license license:bsd-3)))
 865
 866 ;; We use the sources from git instead of the tarball from pypi, because the
 867 ;; latter does not include the Cython source file from which bycython.cpp is
 868 ;; generated.
 869 (define-public python-editdistance
 870   (let ((commit "3ea84a7dd3258c76aa3be851ef3d50e59c886846")
 871         (revision "1"))
 872     (package
 873       (name "python-editdistance")
 874       (version (string-append "0.3.1-" revision "." (string-take commit 7)))
 875       (source
 876        (origin
 877          (method git-fetch)
 878          (uri (git-reference
 879                (url "https://github.com/aflc/editdistance.git")
 880                (commit commit)))
 881          (file-name (git-file-name name version))
 882          (sha256
 883           (base32
 884            "1l43svsv12crvzphrgi6x435z6xg8m086c64armp8wzb4l8ccm7g"))))
 885       (build-system python-build-system)
 886       (arguments
 887        `(#:phases
 888          (modify-phases %standard-phases
 889            (add-after 'unpack 'build-cython-code
 890              (lambda _
 891                (with-directory-excursion "editdistance"
 892                  (delete-file "bycython.cpp")
 893                  (invoke "cython" "--cplus" "bycython.pyx")))))))
 894       (native-inputs
 895        `(("python-cython" ,python-cython)))
 896       (home-page "https://www.github.com/aflc/editdistance")
 897       (synopsis "Fast implementation of the edit distance (Levenshtein distance)")
 898       (description
 899        "This library simply implements Levenshtein distance algorithm with C++
 900 and Cython.")
 901       (license license:expat))))
 902
 903 (define-public go-github.com-mattn-go-runewidth
 904   (let ((commit "703b5e6b11ae25aeb2af9ebb5d5fdf8fa2575211")
 905         (version "0.0.4")
 906         (revision "1"))
 907     (package
 908       (name "go-github.com-mattn-go-runewidth")
 909       (version (git-version version revision commit))
 910       (source
 911        (origin
 912          (method git-fetch)
 913          (uri (git-reference
 914                (url "https://github.com/mattn/runewidth")
 915                (commit commit)))
 916          (file-name (git-file-name name version))
 917          (sha256
 918           (base32
 919            "0znpyz71gajx3g0j2zp63nhjj2c07g16885vxv4ykwnrfmzbgk4w"))))
 920       (build-system go-build-system)
 921       (arguments
 922        '(#:import-path "github.com/mattn/go-runewidth"))
 923       (synopsis "@code{runewidth} provides Go functions to work with string widths")
 924       (description
 925        "The @code{runewidth} library provides Go functions for padding,
 926 measuring and checking the width of strings, with support for East Asian
 927 text.")
 928       (home-page "https://github.com/mattn/runewidth")
 929       (license license:expat))))
 930
 931 (define-public docx2txt
 932   (package
 933     (name "docx2txt")
 934     (version "1.4")
 935     (source (origin
 936               (method url-fetch)
 937               (uri (string-append
 938                     "mirror://sourceforge/docx2txt/docx2txt/v"
 939                     version "/docx2txt-" version ".tgz"))
 940               (sha256
 941                (base32
 942                 "06vdikjvpj6qdb41d8wzfnyj44jpnknmlgbhbr1w215420lpb5xj"))))
 943     (build-system gnu-build-system)
 944     (inputs
 945      `(("unzip" ,unzip)
 946        ("perl" ,perl)))
 947     (arguments
 948      `(#:tests? #f                      ; No tests.
 949        #:make-flags (list (string-append "BINDIR="
 950                                          (assoc-ref %outputs "out") "/bin")
 951                           (string-append "CONFIGDIR="
 952                                          (assoc-ref %outputs "out") "/etc")
 953                           ;; Makefile seems to be a bit dumb at guessing.
 954                           (string-append "INSTALL=install")
 955                           (string-append "PERL=perl"))
 956        #:phases
 957        (modify-phases %standard-phases
 958          (delete 'configure)
 959          (add-after 'install 'fix-install
 960            (lambda* (#:key outputs inputs #:allow-other-keys)
 961              (let* ((out (assoc-ref outputs "out"))
 962                     (bin (string-append out "/bin"))
 963                     (config (string-append out "/etc/docx2txt.config"))
 964                     (unzip (assoc-ref inputs "unzip")))
 965                ;; According to INSTALL, the .sh wrapper can be skipped.
 966                (delete-file (string-append bin "/docx2txt.sh"))
 967                (rename-file (string-append bin "/docx2txt.pl")
 968                             (string-append bin "/docx2txt"))
 969                (substitute* config
 970                  (("config_unzip         => '/usr/bin/unzip',")
 971                   (string-append "config_unzip         => '"
 972                                  unzip
 973                                  "/bin/unzip',")))
 974                ;; Makefile is wrong.
 975                (chmod config #o644)
 976                #t))))))
 977     (synopsis "Recover text from @file{.docx} files, with good formatting")
 978     (description
 979      "@command{docx2txt} is a Perl based command line utility to convert
 980 Microsoft Office @file{.docx} documents to equivalent text documents.  Latest
 981 version supports following features during text extraction.
 982
 983 @itemize
 984 @item Character conversions; currency characters are converted to respective
 985 names like Euro.
 986 @item Capitalisation of text blocks.
 987 @item Center and right justification of text fitting in a line of
 988 (configurable) 80 columns.
 989 @item Horizontal ruler, line breaks, paragraphs separation, tabs.
 990 @item Indicating hyperlinked text along with the hyperlink (configurable).
 991 @item Handling (bullet, decimal, letter, roman) lists along with (attempt at)
 992 indentation.
 993 @end itemize\n")
 994     (home-page "http://docx2txt.sourceforge.net")
 995     (license license:gpl3+)))
 996
 997 (define-public odt2txt
 998   (package
 999     (name "odt2txt")
1000     (version "0.5")
1001     (source
1002       (origin
1003         (method git-fetch)
1004         (uri (git-reference
1005                (url "https://github.com/dstosberg/odt2txt/")
1006                (commit (string-append "v" version))))
1007         (file-name (git-file-name name version))
1008         (sha256
1009          (base32
1010           "0im3kzvhxkjlx57w6h13mc9584c74ma1dyymgvpq2y61av3gc35v"))))
1011     (build-system gnu-build-system)
1012     (arguments
1013      `(#:tests? #f ; no make check
1014        #:make-flags (list "CC=gcc"
1015                           (string-append "DESTDIR=" (assoc-ref %outputs "out")))
1016        #:phases
1017        (modify-phases %standard-phases
1018          ;; no configure script
1019          (delete 'configure))))
1020     (inputs
1021      `(("zlib" ,zlib)))
1022     (home-page "https://github.com/dstosberg/odt2txt/")
1023     (synopsis "Converter from OpenDocument Text to plain text")
1024     (description "odt2txt is a command-line tool which extracts the text out
1025 of OpenDocument Texts, as produced by OpenOffice.org, KOffice, StarOffice and
1026 others.
1027
1028 odt2txt can also extract text from some file formats similar to OpenDocument
1029 Text, such as OpenOffice.org XML (*.sxw), which was used by OpenOffice.org
1030 version 1.x and older StarOffice versions.  To a lesser extent, odt2txt may be
1031 useful to extract content from OpenDocument spreadsheets (*.ods) and
1032 OpenDocument presentations (*.odp).")
1033     (license license:gpl2)))
1034
1035 (define-public opencc
1036   (package
1037     (name "opencc")
1038     (version "1.0.5")
1039     (source
1040      (origin
1041        (method git-fetch)
1042        (uri (git-reference
1043               (url "https://github.com/BYVoid/OpenCC")
1044               (commit (string-append "ver." version))))
1045        (file-name (git-file-name name version))
1046        (sha256
1047         (base32
1048          "1pv5md225qwhbn8ql932zdg6gh1qlx3paiajaks8gfsa07yzvhr4"))
1049        (modules '((guix build utils)))
1050        (snippet
1051         '(begin
1052            ;; TODO: Unbundle tclap, darts-clone, gtest
1053            (delete-file-recursively "deps/rapidjson-0.11") #t))))
1054     (build-system cmake-build-system)
1055     (arguments
1056      '(#:phases
1057        (modify-phases %standard-phases
1058          (add-after 'unpack 'patch-3rd-party-references
1059            (lambda* (#:key inputs #:allow-other-keys)
1060              (let ((rapidjson (assoc-ref inputs "rapidjson")))
1061                (substitute* "src/CMakeLists.txt"
1062                  (("../deps/rapidjson-0.11")
1063                   (string-append rapidjson "/include/rapidjson")))
1064              #t))))))
1065     (native-inputs
1066      `(("python" ,python-wrapper)
1067        ("rapidjson" ,rapidjson)))
1068     (home-page "https://github.com/BYVoid/OpenCC")
1069     (synopsis "Convert between Traditional Chinese and Simplified Chinese")
1070     (description "Open Chinese Convert (OpenCC) converts between Traditional
1071 Chinese and Simplified Chinese, supporting character-level conversion,
1072 phrase-level conversion, variant conversion, and regional idioms among
1073 Mainland China, Taiwan, and Hong-Kong.")
1074     (license license:asl2.0)))
1075
1076 (define-public nkf
1077   (let ((commit "08043eadf4abdddcf277842217e3c77a24740dc2")
1078         (revision "1"))
1079     (package
1080       (name "nkf")
1081       ;; The commits corresponding to specific versions are published
1082       ;; here:
1083       ;; https://ja.osdn.net/projects/nkf/scm/git/nkf/
1084       (version "2.1.5")
1085       (source (origin
1086                 (method git-fetch)
1087                 (uri (git-reference
1088                       (url "https://github.com/nurse/nkf.git")
1089                       (commit commit)))
1090                 (file-name (git-file-name name version))
1091                 (sha256
1092                  (base32
1093                   "0anw0knr1iy4p9w3d3b3pbwzh1c43p1i2q4c28kw9zviw8kx2rly"))))
1094       (build-system gnu-build-system)
1095       (arguments
1096        `(#:tests? #f ; test for perl module
1097          #:make-flags (list "CC=gcc" "CFLAGS=-O2 -Wall -pedantic"
1098                             (string-append "prefix=" %output)
1099                             "MKDIR=mkdir -p")
1100          #:phases
1101          (modify-phases %standard-phases
1102            (delete 'configure)))) ; No ./configure script
1103       (home-page "https://ja.osdn.net/projects/nkf/")
1104       (synopsis "Network Kanji Filter")
1105       (description "Nkf is yet another kanji code converter among networks,
1106 hosts and terminals.  It converts input kanji code to designated kanji code
1107 such as ISO-2022-JP, Shift_JIS, EUC-JP, UTF-8, UTF-16 or UTF-32.")
1108       (license license:zlib))))
1109
1110 (define-public python-pandocfilters
1111   (package
1112     (name "python-pandocfilters")
1113     (version "1.4.2")
1114     (source
1115      (origin
1116        (method url-fetch)
1117        (uri (pypi-uri "pandocfilters" version))
1118        (sha256
1119         (base32
1120          "1a8d9b7s48gmq9zj0pmbyv2sivn5i7m6mybgpkk4jm5vd7hp1pdk"))))
1121     (build-system python-build-system)
1122     (home-page "https://github.com/jgm/pandocfilters")
1123     (synopsis "Python module for writing Pandoc filters")
1124     (description "Pandoc is a powerful utility to transform various
1125 input formats into a wide range of output formats.  To alter the
1126 exported output document, Pandoc allows the usage of filters, which
1127 are pipes that read a JSON serialization of the Pandoc AST from stdin,
1128 transform it in some way, and write it to stdout.  It allows therefore
1129 to alter the processing of Pandoc's supported input formats, for
1130 instance one can add new syntax elements to markdown, etc.
1131
1132 This package provides Python bindings.")
1133     (license license:bsd-3)))
1134
1135 (define-public aha
1136   (package
1137     (name "aha")
1138     (version "0.5")
1139     (source
1140      (origin
1141        (method git-fetch)
1142        (uri (git-reference
1143              (url "https://github.com/theZiz/aha")
1144              (commit version)))
1145        (sha256
1146         (base32
1147          "0byml4rmpiaalwx69jcixl3yvpvwmwiss1jzgsqwshilb2p4qnmz"))
1148        (file-name (git-file-name name version))))
1149     (build-system gnu-build-system)
1150     (arguments
1151      '(#:phases
1152        (modify-phases %standard-phases
1153          (delete 'configure))
1154        #:make-flags (list "CC=gcc"
1155                           (string-append "PREFIX="
1156                                          (assoc-ref %outputs "out")))
1157        ;; no check target
1158        #:tests? #f))
1159     (home-page "https://github.com/theZiz/aha")
1160     (synopsis "Converts terminal escape sequences to HTML")
1161     (description "@command{aha} (Ansi Html Adapter) converts ANSI escape sequences
1162 of a Unix terminal to HTML code.")
1163     (license (list license:lgpl2.0+ license:mpl1.1))))