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, 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>
24 ;;; This file is part of GNU Guix.
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.
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.
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/>.
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))
66 (define-public dos2unix
73 (uri (string-append "https://waterlan.home.xs4all.nl/dos2unix/"
74 "dos2unix-" version ".tar.gz"))
76 (base32 "08w6yywzirsxq8bh87jycvvw922ybhc2l426j2iqzliyn1h8mm8w"))))
77 (build-system gnu-build-system)
79 '(#:make-flags (list "CC=gcc"
80 (string-append "prefix=" (assoc-ref %outputs "out")))
82 (modify-phases %standard-phases
83 (delete 'configure)))) ; no configure script
85 `(("gettext" ,gettext-minimal)
87 (home-page "https://waterlan.home.xs4all.nl/dos2unix.html")
88 (synopsis "DOS/Mac to Unix and vice versa text file format converter")
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)))
101 (uri (string-append "https://github.com/rrthomas/recode/releases/"
102 "download/v" version "/recode-" version ".tar.gz"))
104 (base32 "0m59sd1ca0zw1aydpc3m8sw03nc885knmccqryg7byzmqs585ia6"))))
105 (build-system gnu-build-system)
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+)))
127 (url "https://github.com/nijel/enca")
129 (file-name (git-file-name name version))
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)))
142 (define-public utf8proc
150 (url "https://github.com/JuliaStrings/utf8proc")
151 (commit (string-append "v" version))))
152 (file-name (git-file-name name version))
154 (base32 "1xlkazhdnja4lksn5c9nf4bln5gjqa35a8gwlam5r0728w0h83qq"))))
155 (build-system gnu-build-system)
157 (let ((UNICODE_VERSION "13.0.0")) ; defined in data/Makefile
158 ;; Test data that is otherwise downloaded with curl.
159 `(("NormalizationTest.txt"
162 (uri (string-append "https://www.unicode.org/Public/"
163 UNICODE_VERSION "/ucd/NormalizationTest.txt"))
165 (base32 "07g0ya4f6zfzvpp24ccxkb2yq568kh83gls85rjl950nv5fya3nn"))))
166 ("GraphemeBreakTest.txt"
169 (uri (string-append "https://www.unicode.org/Public/"
171 "/ucd/auxiliary/GraphemeBreakTest.txt"))
173 (base32 "07f8rrvcsq4pibdz6zxggxy8w7zjjqyw2ggclqlhalyv45yv7prj"))))
178 '(#:make-flags (list "CC=gcc"
179 (string-append "prefix=" (assoc-ref %outputs "out")))
181 (modify-phases %standard-phases
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"
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)))
200 (define-public libconfuse
207 (uri (string-append "https://github.com/martinh/libconfuse/"
208 "releases/download/v" version
209 "/confuse-" version ".tar.xz"))
211 (base32 "043hqqykpprgrkw9s2hbdlxr308a7yxwsgxj4m8aadg1401hmm8x"))))
212 (build-system gnu-build-system)
214 '(#:configure-flags '("--disable-static")))
215 (home-page "https://github.com/martinh/libconfuse")
216 (synopsis "Configuration file parser library")
217 (description "libconfuse is a configuration file parser library. It
218 supports sections and (lists of) values (strings, integers, floats, booleans
219 or other sections), as well as some other features (such as
220 single/double-quoted strings, environment variable expansion, functions and
221 nested include statements).")
222 (license license:isc)))
224 (define-public libgtextutils
226 (name "libgtextutils")
232 "https://github.com/agordon/libgtextutils/releases/download/"
233 version "/libgtextutils-" version ".tar.gz"))
235 (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
236 (build-system gnu-build-system)
239 (modify-phases %standard-phases
241 (lambda _ (invoke "sh" "reconf"))))))
243 `(("autoconf" ,autoconf)
244 ("automake" ,automake)
245 ("gcc" ,gcc-5) ;; doesn't build with later versions
246 ("libtool" ,libtool)))
247 (home-page "https://github.com/agordon/libgtextutils")
248 (synopsis "Gordon's text utils library")
250 "libgtextutils is a text utilities library used by the fastx toolkit from
252 (license license:agpl3+)))
254 (define-public cityhash
255 (let ((commit "8af9b8c"))
258 (version (string-append "1.1-2." commit))
262 (url "https://github.com/google/cityhash")
264 (file-name (string-append name "-" version ".tar.gz"))
267 "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
268 (build-system gnu-build-system)
270 '(#:make-flags (list "CXXFLAGS=-g -O3")
272 (modify-phases %standard-phases
273 ;; citycrc is not installed by default but is used by some
275 (add-after 'install 'install-citycrc
276 (lambda* (#:key outputs #:allow-other-keys)
277 (let* ((out (assoc-ref outputs "out"))
278 (include (string-append out "/include")))
279 (install-file "src/citycrc.h" include))
281 (home-page "https://github.com/google/cityhash")
282 (synopsis "C++ hash functions for strings")
284 "CityHash provides hash functions for strings. The functions mix the
285 input bits thoroughly but are not suitable for cryptography.")
286 (license license:expat))))
294 (uri (string-append "http://www.and.org/ustr/" version
295 "/ustr-" version ".tar.bz2"))
298 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
299 (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
300 (build-system gnu-build-system)
305 ;; Override "/sbin/ldconfig" with "echo" because we don't need
308 (string-append "prefix=" (assoc-ref %outputs "out"))
311 (modify-phases %standard-phases
312 (add-after 'unpack 'disable-check-for-stdint
314 ;; Of course we have stdint.h, just not in /usr/include
315 (substitute* '("Makefile"
317 (("-f \"/usr/include/stdint.h\"") "-z \"\""))
319 ;; No configure script
320 (delete 'configure))))
321 (home-page "http://www.and.org/ustr/")
322 (synopsis "String library with very low memory overhead")
324 "Ustr is a string library for C with very low memory overhead.")
325 ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
326 ;; LGPL, etc. ... if you need another license to help compatibility, just
327 ;; ask for it. It's basically public domain, without all the legal
328 ;; problems for everyone that trying to make something public domain
330 (license license:public-domain)))
332 (define-public ascii2binary
334 (name "ascii2binary")
339 (uri (string-append "http://billposer.org/Software/Downloads/"
340 "ascii2binary-" version ".tar.bz2"))
342 (base32 "0dc9fxcdmppbs9s06jvq61zbk552laxps0xyk098gj41697ihd96"))))
343 (build-system gnu-build-system)
345 `(("gettext" ,gettext-minimal)))
346 (home-page "https://billposer.org/Software/a2b.html")
347 (synopsis "Convert between ASCII, hexadecimal and binary representations")
348 (description "The two programs are useful for generating test data, for
349 inspecting binary files, and for interfacing programs that generate textual
350 output to programs that require binary input and conversely. They can also be
351 useful when it is desired to reformat numbers.
355 @item @command{ascii2binary} reads input consisting of ascii or hexadecimal
356 representation numbers separated by whitespace and produces as output
357 the binary equivalents. The type and precision of the binary output
358 is selected using command line flags.
360 @item @command{binary2ascii} reads input consisting of binary numbers
361 and converts them to their ascii or hexadecimal representation.
362 Command line flags specify the type and size of the binary numbers
363 and provide control over the format of the output.
364 Unsigned integers may be written out in binary, octal, decimal,
367 Signed integers may be written out only in binary or decimal. Floating
368 point numbers may be written out only decimal, either in standard or
369 scientific notation. (If you want to examine the binary representation
370 of floating point numbers, just treat the input as a sequence of unsigned
374 (license license:gpl3)))
376 (define-public uniutils
383 (uri (string-append "http://billposer.org/Software/Downloads/"
384 "uniutils-" version ".tar.bz2"))
386 (base32 "19w1510w87gx7n4qy3zsb0m467a4rn5scvh4ajajg7jh6x5xri08"))))
387 (build-system gnu-build-system)
389 '(#:configure-flags '("--disable-dependency-tracking")
391 (modify-phases %standard-phases
392 (add-after 'build 'fix-paths
393 (lambda* (#:key outputs inputs #:allow-other-keys)
394 (let ((out (assoc-ref outputs "out"))
395 (a2b (assoc-ref inputs "ascii2binary"))
396 (iconv (assoc-ref inputs "libiconv")))
397 (substitute* "utf8lookup"
398 (("^ascii2binary ") (string-append a2b "/bin/ascii2binary "))
399 (("^uniname ") (string-append out "/bin/uniname "))
400 (("^iconv ") (string-append iconv "/bin/iconv ")))
403 `(("ascii2binary" ,ascii2binary)
404 ("libiconv" ,libiconv)))
405 (home-page "https://billposer.org/Software/unidesc.html")
406 (synopsis "Find out what is in a Unicode file")
407 (description "Useful tools when working with Unicode files when one
408 doesn't know the writing system, doesn't have the necessary font, needs to
409 inspect invisible characters, needs to find out whether characters have been
410 combined or in what order they occur, or needs statistics on which characters
415 @item @command{uniname} defaults to printing the character offset of each
416 character, its byte offset, its hex code value, its encoding, the glyph
417 itself, and its name. It may also be used to validate UTF-8 input.
419 @item @command{unidesc} reports the character ranges to which different
420 portions of the text belong. It can also be used to identify Unicode encodings
421 (e.g. UTF-16be) flagged by magic numbers.
423 @item @command{unihist} generates a histogram of the characters in its input.
425 @item @command{ExplicateUTF8} is intended for debugging or for learning about
426 Unicode. It determines and explains the validity of a sequence of bytes as a
429 @item @command{utf8lookup} provides a handy way to look up Unicode characters
430 from the command line.
432 @item @command{unireverse} reverse each line of UTF-8 input
433 character-by-character.
436 (license license:gpl3)))
438 (define-public libconfig
442 (home-page "https://hyperrealm.github.io/libconfig/")
445 (uri (string-append home-page "/dist/libconfig-"
449 "1ngs2qx3cx5cbwinc5mvadly0b5n7s86zsc68c404czzfff7lg3w"))))
450 (build-system gnu-build-system)
451 (synopsis "C/C++ configuration file library")
453 "Libconfig is a simple library for manipulating structured configuration
454 files. This file format is more compact and more readable than XML. And
455 unlike XML, it is type-aware, so it is not necessary to do string parsing in
457 (license license:lgpl2.1+)))
466 (url "https://github.com/pfff/pfff")
467 (commit (string-append "v" version))))
468 (file-name (git-file-name name version))
471 "1nxkfm7zliq3rmr7yp871sppwfnz71iz364m2sgazny71pzykggc"))))
472 (build-system cmake-build-system)
473 (home-page "https://biit.cs.ut.ee/pfff/")
474 (synopsis "Probabilistic fast file fingerprinting tool")
476 "pfff is a tool for calculating a compact digital fingerprint of a file
477 by sampling randomly from the file instead of reading it in full.
478 Consequently, the computation has a flat performance characteristic,
479 correlated with data variation rather than file size. pfff can be as reliable
480 as existing hashing techniques, with provably negligible risk of collisions.")
481 (license license:bsd-3)))
483 (define-public oniguruma
486 (version "6.9.5-rev1")
489 (uri (string-append "https://github.com/kkos/"
490 "oniguruma/releases/download/v"
491 ;; If there is a "-" in the version, convert
492 ;; to underscore for this part of the URI.
493 (string-map (lambda (c) (if (char=? #\- c) #\_ c))
495 "/onig-" version ".tar.gz"))
498 "17m92k1n6bvza6m35fpd5g36zwpwm3hfz3478iwj5bvj2sfq8g6k"))))
499 (build-system gnu-build-system)
500 (arguments '(#:configure-flags '("--disable-static")))
501 (home-page "https://github.com/kkos/oniguruma")
502 (synopsis "Regular expression library")
503 (description "Oniguruma is a regular expressions library. The special
504 characteristic of this library is that different character encoding for every
505 regular expression object can be specified.")
506 (license license:bsd-2)))
508 (define-public antiword
514 (uri (string-append "http://www.winfield.demon.nl/linux"
515 "/antiword-" version ".tar.gz"))
518 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
519 (patches (search-patches "antiword-CVE-2014-8123.patch"))))
520 (build-system gnu-build-system)
522 `(#:tests? #f ; There are no tests
524 (list "-f" "Makefile.Linux"
525 (string-append "GLOBAL_INSTALL_DIR="
526 (assoc-ref %outputs "out") "/bin")
527 (string-append "GLOBAL_RESOURCES_DIR="
528 (assoc-ref %outputs "out") "/share/antiword"))
530 (modify-phases %standard-phases
532 (lambda* (#:key outputs #:allow-other-keys)
533 ;; Ensure that mapping files can be found in the actual package
535 (substitute* "antiword.h"
536 (("/usr/share/antiword")
537 (string-append (assoc-ref outputs "out") "/share/antiword")))
540 (lambda* (#:key make-flags #:allow-other-keys)
541 (apply invoke "make" `("global_install" ,@make-flags)))))))
542 (home-page "http://www.winfield.demon.nl/")
543 (synopsis "Microsoft Word document reader")
544 (description "Antiword is an application for displaying Microsoft Word
545 documents. It can also convert the document to PostScript or XML. Only
546 documents made by MS Word version 2 and version 6 or later are supported. The
547 name comes from: \"The antidote against people who send Microsoft Word files
548 to everybody, because they believe that everybody runs Windows and therefore
550 (license license:gpl2+)))
552 (define-public catdoc
558 (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
559 "catdoc-" version ".tar.gz"))
560 (patches (search-patches "catdoc-CVE-2017-11110.patch"))
563 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
564 (build-system gnu-build-system)
565 ;; TODO: Also build `wordview` which requires `tk` – make a separate
568 '(#:tests? #f ; There are no tests
569 #:configure-flags '("--disable-wordview")
571 (modify-phases %standard-phases
572 (add-before 'install 'fix-install
573 (lambda* (#:key outputs #:allow-other-keys)
574 (let ((out (assoc-ref outputs "out")))
575 (mkdir-p (string-append out "/share/man/man1"))))))))
576 (home-page "https://www.wagner.pp.ru/~vitus/software/catdoc/")
577 (synopsis "MS-Word to TeX or plain text converter")
578 (description "@command{catdoc} extracts text from MS-Word files, trying to
579 preserve as many special printable characters as possible. It supports
580 everything up to Word-97. Also supported are MS Write documents and RTF files.
582 @command{catdoc} does not preserve complex word formatting, but it can
583 translate some non-ASCII characters into TeX escape codes. It's goal is to
584 extract plain text and allow you to read it and, probably, reformat with TeX,
585 according to TeXnical rules.
587 This package also provides @command{xls2csv}, which extracts data from Excel
588 spreadsheets and outputs it in comma-separated-value format, and
589 @command{catppt}, which extracts data from PowerPoint presentations.")
590 (license license:gpl2+)))
592 (define-public utfcpp
599 (url "https://github.com/nemtrif/utfcpp")
600 (commit (string-append "v" version))))
601 (file-name (git-file-name name version))
604 "1gr98d826z6wa58r1s5i7rz7q2x3r31v7zj0pjjlrc7gfxwklr4s"))))
605 (build-system cmake-build-system)
607 `(#:out-of-source? #f
609 (modify-phases %standard-phases
610 (replace 'install ; no install target
611 (lambda* (#:key outputs #:allow-other-keys)
612 (let* ((out (assoc-ref outputs "out"))
613 (include (string-append out "/include"))
614 (doc (string-append out "/share/doc/" ,name)))
615 (copy-recursively "source" include)
616 (install-file "README.md" doc)
618 (home-page "https://github.com/nemtrif/utfcpp")
619 (synopsis "Portable C++ library for handling UTF-8")
620 (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
622 (license license:boost1.0)))
631 (uri (string-append "mirror://sourceforge/dbacl/dbacl/" version "/"
632 "dbacl-" version ".tar.gz"))
634 (base32 "1gas0112wqjvwn9qg3hxnawk7h3prr0w9b2h68f3p1ifd1kzn3gz"))
635 (patches (search-patches "dbacl-include-locale.h.patch"))))
636 (build-system gnu-build-system)
640 (string-append "-I" (assoc-ref %build-inputs "slang")
642 (string-append "-I" (assoc-ref %build-inputs "ncurses")
645 (modify-phases %standard-phases
646 (add-after 'unpack 'delete-sample6-and-japanese
648 (substitute* "doc/Makefile.am"
650 (("japanese.txt") ""))
651 (delete-file "doc/sample6.txt")
652 (delete-file "doc/japanese.txt")
653 (substitute* (list "src/tests/Makefile.am"
654 "src/tests/Makefile.in")
655 (("dbacl-jap.shin") "")
656 (("dbacl-jap.sh") ""))
658 (add-after 'unpack 'delete-test
659 ;; See comments about the license.
661 (delete-file "src/tests/dbacl-jap.shin")
663 (add-after 'unpack 'fix-test-files
664 (lambda* (#:key inputs outputs #:allow-other-keys)
665 (let* ((out (assoc-ref outputs "out"))
666 (bin (string-append out "/bin")))
667 (substitute* (find-files "src/tests/" "\\.shin$")
668 (("PATH=/bin:/usr/bin")
669 "#PATH=/bin:/usr/bin")
670 (("diff") (string-append (which "diff")))
671 (("tr") (string-append (which "tr"))))
675 (invoke "autoreconf" "-vif")
678 `(("ncurses" ,ncurses)
680 ("readline" ,readline)
683 `(("libtool" ,libtool)
684 ("autoconf" ,autoconf)
685 ("automake" ,automake)
686 ("pkg-config" ,pkg-config)))
687 (home-page "https://www.lbreyer.com/dbacl.html")
688 (synopsis "Bayesian text and email classifier")
690 "dbacl is a fast Bayesian text and email classifier. It builds a variety
691 of language models using maximum entropy (minimum divergence) principles, and
692 these can then be used to categorize input data automatically among multiple
694 ;; The software is licensed as GPLv3 or later, but
695 ;; includes various sample texts in the doc dir:
696 ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
698 ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
699 ;; - sample6.txt is a forwarded email, copyright unknown.
700 ;; Guix does exclude sample6.txt.
701 ;; - japanese.txt is a Japanese unoffical translation of the
702 ;; GNU General Public License, (c) by the Free Software Foundation.
703 ;; Guix excludes this file.
704 (license (list license:gpl3+ license:public-domain))))
706 (define-public dotconf
713 (url "https://github.com/williamh/dotconf")
714 (commit (string-append "v" version))))
715 (file-name (git-file-name name version))
718 "1sc95hw5k2xagpafny0v35filmcn05k1ds5ghkldfpf6xw4hakp7"))))
719 (build-system gnu-build-system)
720 (arguments `(#:tests? #f)) ; FIXME maketest.sh does not work.
722 `(("autoconf" ,autoconf)
723 ("automake" ,automake)
724 ("libtool" ,libtool)))
725 (home-page "https://github.com/williamh/dotconf")
726 (synopsis "Configuration file parser library")
728 "C library for creating and parsing configuration files.")
729 (license (list license:lgpl2.1 ; Main distribution.
730 license:asl1.1)))) ; src/readdir.{c,h}
732 (define-public drm-tools
738 (uri (string-append "mirror://sourceforge/drmtools/drm_tools-"
742 "187zbxw21zcg8gpyc13gxlycfw0n05a6rmqq6im5wr9zk1v1wj80"))))
743 (build-system cmake-build-system)
745 `(#:tests? #f ;the test suite fails
746 #:phases (modify-phases %standard-phases
747 (add-after 'unpack 'set-install-prefixes
748 (lambda* (#:key outputs #:allow-other-keys)
749 (let* ((out (assoc-ref outputs "out")))
750 (substitute* "CMakeLists.txt"
756 (add-after 'unpack 'adjust-test-paths
758 (substitute* '("test_extract_increment.sh"
759 "test_extract_features.sh"
760 "test_extract_features2.sh"
762 (("\\./extract") "extract")
763 (("\\./dmath") "dmath")
764 (("/usr/local/bin/") "")
769 ;; The produced binaries are written directly to %output/bin.
771 (add-after 'build 'check
772 (lambda* (#:key outputs tests? #:allow-other-keys)
774 (let* ((out (assoc-ref outputs "out"))
775 (bin (string-append out "/bin")))
776 (setenv "PATH" (string-append bin ":"
778 (with-directory-excursion
779 (format #f "../drm_tools-~a" ,version)
780 (invoke "sh" "test_all.sh")))))))))
781 (native-inputs `(("which" ,which))) ;for tests
782 (inputs `(("pcre" ,pcre)))
783 (home-page "http://drmtools.sourceforge.net/")
784 (synopsis "Utilities to manipulate text and binary files")
785 (description "The drm_tools package contains the following commands:
788 An extended version of the \"date\" program that has sub-second accuracy.
790 Format complex binary data into text.
792 Load data into a binary file using simple commands from the input.
794 Sort, merge, search, retrieve or generate test data consisting of fixed size
797 Find or find/replace in binary files.
799 Split test data consisting of fixed size binary records into one or more
802 Find changes between two files at the character level. Unlike \"diff\", it
803 lists just the characters that differ, so if the 40,000th character is
804 different only that one character will be shown, not the entire line.
806 Add columns of integers, decimals, and/or times.
808 A utility for formatting binary data dumps.
810 Double precision interactive command line math calculator.
812 Extract and emit data from text files based on character or token position.
814 A utility that reads from STDIN and executes each line as a command in a
817 A utility for rapid retrieval of text by line numbers, in any order, from a
822 Create message queues and send/receive messages.
825 Multiple buffer in and out. Used for buffering a lot of data between a slow
826 device and a fast device. Mostly for buffering streaming tape drives for use
827 with slower network connections, so that streaming is maintained as much as
828 possible to minimize wear on the tape device.
830 Corrupt data streams - useful for testing error correction and data recovery.
832 Filter, list, or split a tar file.
834 (license license:gpl2+)))
836 (define-public java-rsyntaxtextarea
838 (name "java-rsyntaxtextarea")
842 (uri (string-append "https://github.com/bobbylight/"
843 "RSyntaxTextArea/archive/"
845 (file-name (string-append name "-" version ".tar.gz"))
848 "0c5mqg2klj5rvf8fhycrli8rf6s37l9p7a8knw9gpp65r1c120q2"))))
849 (build-system ant-build-system)
851 `(;; FIXME: some tests fail because locale resources cannot be found.
852 ;; Even when I add them to the class path,
853 ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
855 #:jar-name "rsyntaxtextarea.jar"))
857 `(("java-junit" ,java-junit)
858 ("java-hamcrest-core" ,java-hamcrest-core)))
859 (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
860 (synopsis "Syntax highlighting text component for Java Swing")
861 (description "RSyntaxTextArea is a syntax highlighting, code folding text
862 component for Java Swing. It extends @code{JTextComponent} so it integrates
863 completely with the standard @code{javax.swing.text} package. It is fast and
864 efficient, and can be used in any application that needs to edit or view
866 (license license:bsd-3)))
868 ;; We use the sources from git instead of the tarball from pypi, because the
869 ;; latter does not include the Cython source file from which bycython.cpp is
871 (define-public python-editdistance
872 (let ((commit "3ea84a7dd3258c76aa3be851ef3d50e59c886846")
875 (name "python-editdistance")
876 (version (string-append "0.3.1-" revision "." (string-take commit 7)))
881 (url "https://github.com/aflc/editdistance")
883 (file-name (git-file-name name version))
886 "1l43svsv12crvzphrgi6x435z6xg8m086c64armp8wzb4l8ccm7g"))))
887 (build-system python-build-system)
890 (modify-phases %standard-phases
891 (add-after 'unpack 'build-cython-code
893 (with-directory-excursion "editdistance"
894 (delete-file "bycython.cpp")
895 (invoke "cython" "--cplus" "bycython.pyx")))))))
897 `(("python-cython" ,python-cython)))
898 (home-page "https://www.github.com/aflc/editdistance")
899 (synopsis "Fast implementation of the edit distance (Levenshtein distance)")
901 "This library simply implements Levenshtein distance algorithm with C++
903 (license license:expat))))
905 (define-public go-github.com-mattn-go-runewidth
906 (let ((commit "703b5e6b11ae25aeb2af9ebb5d5fdf8fa2575211")
910 (name "go-github.com-mattn-go-runewidth")
911 (version (git-version version revision commit))
916 (url "https://github.com/mattn/runewidth")
918 (file-name (git-file-name name version))
921 "0znpyz71gajx3g0j2zp63nhjj2c07g16885vxv4ykwnrfmzbgk4w"))))
922 (build-system go-build-system)
924 '(#:import-path "github.com/mattn/go-runewidth"))
925 (synopsis "@code{runewidth} provides Go functions to work with string widths")
927 "The @code{runewidth} library provides Go functions for padding,
928 measuring and checking the width of strings, with support for East Asian
930 (home-page "https://github.com/mattn/runewidth")
931 (license license:expat))))
933 (define-public docx2txt
940 "mirror://sourceforge/docx2txt/docx2txt/v"
941 version "/docx2txt-" version ".tgz"))
944 "06vdikjvpj6qdb41d8wzfnyj44jpnknmlgbhbr1w215420lpb5xj"))))
945 (build-system gnu-build-system)
950 `(#:tests? #f ; No tests.
951 #:make-flags (list (string-append "BINDIR="
952 (assoc-ref %outputs "out") "/bin")
953 (string-append "CONFIGDIR="
954 (assoc-ref %outputs "out") "/etc")
955 ;; Makefile seems to be a bit dumb at guessing.
956 (string-append "INSTALL=install")
957 (string-append "PERL=perl"))
959 (modify-phases %standard-phases
961 (add-after 'install 'fix-install
962 (lambda* (#:key outputs inputs #:allow-other-keys)
963 (let* ((out (assoc-ref outputs "out"))
964 (bin (string-append out "/bin"))
965 (config (string-append out "/etc/docx2txt.config"))
966 (unzip (assoc-ref inputs "unzip")))
967 ;; According to INSTALL, the .sh wrapper can be skipped.
968 (delete-file (string-append bin "/docx2txt.sh"))
969 (rename-file (string-append bin "/docx2txt.pl")
970 (string-append bin "/docx2txt"))
972 (("config_unzip => '/usr/bin/unzip',")
973 (string-append "config_unzip => '"
976 ;; Makefile is wrong.
979 (synopsis "Recover text from @file{.docx} files, with good formatting")
981 "@command{docx2txt} is a Perl based command line utility to convert
982 Microsoft Office @file{.docx} documents to equivalent text documents. Latest
983 version supports following features during text extraction.
986 @item Character conversions; currency characters are converted to respective
988 @item Capitalisation of text blocks.
989 @item Center and right justification of text fitting in a line of
990 (configurable) 80 columns.
991 @item Horizontal ruler, line breaks, paragraphs separation, tabs.
992 @item Indicating hyperlinked text along with the hyperlink (configurable).
993 @item Handling (bullet, decimal, letter, roman) lists along with (attempt at)
996 (home-page "http://docx2txt.sourceforge.net")
997 (license license:gpl3+)))
999 (define-public odt2txt
1007 (url "https://github.com/dstosberg/odt2txt/")
1008 (commit (string-append "v" version))))
1009 (file-name (git-file-name name version))
1012 "0im3kzvhxkjlx57w6h13mc9584c74ma1dyymgvpq2y61av3gc35v"))))
1013 (build-system gnu-build-system)
1015 `(#:tests? #f ; no make check
1016 #:make-flags (list "CC=gcc"
1017 (string-append "DESTDIR=" (assoc-ref %outputs "out")))
1019 (modify-phases %standard-phases
1020 ;; no configure script
1021 (delete 'configure))))
1024 (home-page "https://github.com/dstosberg/odt2txt/")
1025 (synopsis "Converter from OpenDocument Text to plain text")
1026 (description "odt2txt is a command-line tool which extracts the text out
1027 of OpenDocument Texts, as produced by OpenOffice.org, KOffice, StarOffice and
1030 odt2txt can also extract text from some file formats similar to OpenDocument
1031 Text, such as OpenOffice.org XML (*.sxw), which was used by OpenOffice.org
1032 version 1.x and older StarOffice versions. To a lesser extent, odt2txt may be
1033 useful to extract content from OpenDocument spreadsheets (*.ods) and
1034 OpenDocument presentations (*.odp).")
1035 (license license:gpl2)))
1037 (define-public opencc
1045 (url "https://github.com/BYVoid/OpenCC")
1046 (commit (string-append "ver." version))))
1047 (file-name (git-file-name name version))
1050 "1pv5md225qwhbn8ql932zdg6gh1qlx3paiajaks8gfsa07yzvhr4"))
1051 (modules '((guix build utils)))
1054 ;; TODO: Unbundle tclap, darts-clone, gtest
1055 (delete-file-recursively "deps/rapidjson-0.11") #t))))
1056 (build-system cmake-build-system)
1059 (modify-phases %standard-phases
1060 (add-after 'unpack 'patch-3rd-party-references
1061 (lambda* (#:key inputs #:allow-other-keys)
1062 (let ((rapidjson (assoc-ref inputs "rapidjson")))
1063 (substitute* "src/CMakeLists.txt"
1064 (("../deps/rapidjson-0.11")
1065 (string-append rapidjson "/include/rapidjson")))
1068 `(("python" ,python-wrapper)
1069 ("rapidjson" ,rapidjson)))
1070 (home-page "https://github.com/BYVoid/OpenCC")
1071 (synopsis "Convert between Traditional Chinese and Simplified Chinese")
1072 (description "Open Chinese Convert (OpenCC) converts between Traditional
1073 Chinese and Simplified Chinese, supporting character-level conversion,
1074 phrase-level conversion, variant conversion, and regional idioms among
1075 Mainland China, Taiwan, and Hong-Kong.")
1076 (license license:asl2.0)))
1079 (let ((commit "08043eadf4abdddcf277842217e3c77a24740dc2")
1083 ;; The commits corresponding to specific versions are published
1085 ;; https://ja.osdn.net/projects/nkf/scm/git/nkf/
1090 (url "https://github.com/nurse/nkf")
1092 (file-name (git-file-name name version))
1095 "0anw0knr1iy4p9w3d3b3pbwzh1c43p1i2q4c28kw9zviw8kx2rly"))))
1096 (build-system gnu-build-system)
1098 `(#:tests? #f ; test for perl module
1099 #:make-flags (list "CC=gcc" "CFLAGS=-O2 -Wall -pedantic"
1100 (string-append "prefix=" %output)
1103 (modify-phases %standard-phases
1104 (delete 'configure)))) ; No ./configure script
1105 (home-page "https://ja.osdn.net/projects/nkf/")
1106 (synopsis "Network Kanji Filter")
1107 (description "Nkf is yet another kanji code converter among networks,
1108 hosts and terminals. It converts input kanji code to designated kanji code
1109 such as ISO-2022-JP, Shift_JIS, EUC-JP, UTF-8, UTF-16 or UTF-32.")
1110 (license license:zlib))))
1112 (define-public python-pandocfilters
1114 (name "python-pandocfilters")
1119 (uri (pypi-uri "pandocfilters" version))
1122 "1a8d9b7s48gmq9zj0pmbyv2sivn5i7m6mybgpkk4jm5vd7hp1pdk"))))
1123 (build-system python-build-system)
1124 (home-page "https://github.com/jgm/pandocfilters")
1125 (synopsis "Python module for writing Pandoc filters")
1126 (description "Pandoc is a powerful utility to transform various
1127 input formats into a wide range of output formats. To alter the
1128 exported output document, Pandoc allows the usage of filters, which
1129 are pipes that read a JSON serialization of the Pandoc AST from stdin,
1130 transform it in some way, and write it to stdout. It allows therefore
1131 to alter the processing of Pandoc's supported input formats, for
1132 instance one can add new syntax elements to markdown, etc.
1134 This package provides Python bindings.")
1135 (license license:bsd-3)))
1145 (url "https://github.com/theZiz/aha")
1149 "0byml4rmpiaalwx69jcixl3yvpvwmwiss1jzgsqwshilb2p4qnmz"))
1150 (file-name (git-file-name name version))))
1151 (build-system gnu-build-system)
1154 (modify-phases %standard-phases
1155 (delete 'configure))
1156 #:make-flags (list "CC=gcc"
1157 (string-append "PREFIX="
1158 (assoc-ref %outputs "out")))
1161 (home-page "https://github.com/theZiz/aha")
1162 (synopsis "Converts terminal escape sequences to HTML")
1163 (description "@command{aha} (Ansi Html Adapter) converts ANSI escape sequences
1164 of a Unix terminal to HTML code.")
1165 (license (list license:lgpl2.0+ license:mpl1.1))))