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>
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 (guix utils)
50 #:use-module (gnu packages)
51 #:use-module (gnu packages autotools)
52 #:use-module (gnu packages base)
53 #:use-module (gnu packages compression)
54 #:use-module (gnu packages gcc)
55 #:use-module (gnu packages golang)
56 #:use-module (gnu packages gettext)
57 #:use-module (gnu packages java)
58 #:use-module (gnu packages ncurses)
59 #:use-module (gnu packages pcre)
60 #:use-module (gnu packages perl)
61 #:use-module (gnu packages pkg-config)
62 #:use-module (gnu packages python)
63 #:use-module (gnu packages python-build)
64 #:use-module (gnu packages python-xyz)
65 #:use-module (gnu packages readline)
66 #:use-module (gnu packages slang)
67 #:use-module (gnu packages web))
69 (define-public dos2unix
76 (uri (string-append "https://waterlan.home.xs4all.nl/dos2unix/"
77 "dos2unix-" version ".tar.gz"))
79 (base32 "00dfsf4rfyjb5j12gan8xjiirm0asshdz6dmd3l34a7ays6wadb0"))))
80 (build-system gnu-build-system)
83 (list (string-append "CC=" ,(cc-for-target))
84 (string-append "prefix=" (assoc-ref %outputs "out")))
86 (modify-phases %standard-phases
87 (delete 'configure)))) ; no configure script
89 `(("gettext" ,gettext-minimal)
91 (home-page "https://waterlan.home.xs4all.nl/dos2unix.html")
92 (synopsis "DOS/Mac to Unix and vice versa text file format converter")
94 "dos2unix is a tool to convert line breaks in a text file from Unix format
95 to DOS format and vice versa.")
96 (license license:bsd-2)))
105 (uri (string-append "https://github.com/rrthomas/recode/releases/"
106 "download/v" version "/recode-" version ".tar.gz"))
108 (base32 "0m59sd1ca0zw1aydpc3m8sw03nc885knmccqryg7byzmqs585ia6"))))
109 (build-system gnu-build-system)
112 ("python-cython" ,python-cython)))
113 (home-page "https://github.com/rrthomas/recode")
114 (synopsis "Text encoding converter")
115 (description "The Recode library converts files between character sets and
116 usages. It recognises or produces over 200 different character sets (or about
117 300 if combined with an iconv library) and transliterates files between almost
118 any pair. When exact transliteration are not possible, it gets rid of
119 offending characters or falls back on approximations. The recode program is a
120 handy front-end to the library.")
121 (license license:gpl3+)))
131 (url "https://github.com/nijel/enca")
133 (file-name (git-file-name name version))
135 (base32 "19q7cwwxmmk5j9438bsqdpjvdjawsd3zmw1zyqgi7s4m0rasr3ah"))))
136 (build-system gnu-build-system)
137 ;; enca-1.19 tests fail with recent recode.
138 ;(inputs `(("recode" ,recode)))
139 (home-page "https://github.com/nijel/enca")
140 (synopsis "Text encoding detection tool")
141 (description "Enca (Extremely Naive Charset Analyser) consists of libenca,
142 an encoding detection library, and enca, a command line frontend, integrating
143 libenca and several charset conversion libraries and tools.")
144 (license license:gpl2)))
146 (define-public utf8proc
154 (url "https://github.com/JuliaStrings/utf8proc")
155 (commit (string-append "v" version))))
156 (file-name (git-file-name name version))
158 (base32 "1xlkazhdnja4lksn5c9nf4bln5gjqa35a8gwlam5r0728w0h83qq"))))
159 (build-system gnu-build-system)
161 (let ((UNICODE_VERSION "13.0.0")) ; defined in data/Makefile
162 ;; Test data that is otherwise downloaded with curl.
163 `(("NormalizationTest.txt"
166 (uri (string-append "https://www.unicode.org/Public/"
167 UNICODE_VERSION "/ucd/NormalizationTest.txt"))
169 (base32 "07g0ya4f6zfzvpp24ccxkb2yq568kh83gls85rjl950nv5fya3nn"))))
170 ("GraphemeBreakTest.txt"
173 (uri (string-append "https://www.unicode.org/Public/"
175 "/ucd/auxiliary/GraphemeBreakTest.txt"))
177 (base32 "07f8rrvcsq4pibdz6zxggxy8w7zjjqyw2ggclqlhalyv45yv7prj"))))
182 '(#:make-flags (list "CC=gcc"
183 (string-append "prefix=" (assoc-ref %outputs "out")))
185 (modify-phases %standard-phases
187 (add-before 'check 'check-data
188 (lambda* (#:key inputs #:allow-other-keys)
189 (for-each (lambda (i)
190 (copy-file (assoc-ref inputs i)
191 (string-append "data/" i)))
192 '("NormalizationTest.txt" "GraphemeBreakTest.txt"))
193 (substitute* "data/GraphemeBreakTest.txt"
197 (home-page "https://juliastrings.github.io/utf8proc/")
198 (synopsis "C library for processing UTF-8 Unicode data")
199 (description "utf8proc is a small C library that provides Unicode
200 normalization, case-folding, and other operations for data in the UTF-8
201 encoding, supporting Unicode version 9.0.0.")
202 (license license:expat)))
204 (define-public libconfuse
211 (uri (string-append "https://github.com/martinh/libconfuse/"
212 "releases/download/v" version
213 "/confuse-" version ".tar.xz"))
215 (base32 "043hqqykpprgrkw9s2hbdlxr308a7yxwsgxj4m8aadg1401hmm8x"))))
216 (build-system gnu-build-system)
218 '(#:configure-flags '("--disable-static")))
219 (home-page "https://github.com/martinh/libconfuse")
220 (synopsis "Configuration file parser library")
221 (description "libconfuse is a configuration file parser library. It
222 supports sections and (lists of) values (strings, integers, floats, booleans
223 or other sections), as well as some other features (such as
224 single/double-quoted strings, environment variable expansion, functions and
225 nested include statements).")
226 (license license:isc)))
228 (define-public libgtextutils
230 (name "libgtextutils")
236 "https://github.com/agordon/libgtextutils/releases/download/"
237 version "/libgtextutils-" version ".tar.gz"))
239 (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
240 (build-system gnu-build-system)
243 (modify-phases %standard-phases
245 (lambda _ (invoke "sh" "reconf")))
246 (add-after 'set-paths 'hide-default-gcc
247 (lambda* (#:key inputs #:allow-other-keys)
248 (let ((gcc (assoc-ref inputs "gcc")))
249 ;; Remove the default GCC from CPLUS_INCLUDE_PATH to prevent
250 ;; conflicts with the GCC 5 input.
251 (setenv "CPLUS_INCLUDE_PATH"
253 (delete (string-append gcc "/include/c++")
254 (string-split (getenv "CPLUS_INCLUDE_PATH") #\:))
258 `(("autoconf" ,autoconf)
259 ("automake" ,automake)
260 ("gcc@5" ,gcc-5) ; doesn't build with later versions
261 ("libtool" ,libtool)))
262 (home-page "https://github.com/agordon/libgtextutils")
263 (synopsis "Gordon's text utils library")
265 "libgtextutils is a text utilities library used by the fastx toolkit from
267 (license license:agpl3+)))
269 (define-public cityhash
270 (let ((commit "8af9b8c"))
273 (version (string-append "1.1-2." commit))
277 (url "https://github.com/google/cityhash")
279 (file-name (string-append name "-" version ".tar.gz"))
282 "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
283 (build-system gnu-build-system)
285 '(#:make-flags (list "CXXFLAGS=-g -O3")
287 (modify-phases %standard-phases
288 ;; citycrc is not installed by default but is used by some
290 (add-after 'install 'install-citycrc
291 (lambda* (#:key outputs #:allow-other-keys)
292 (let* ((out (assoc-ref outputs "out"))
293 (include (string-append out "/include")))
294 (install-file "src/citycrc.h" include))
296 (home-page "https://github.com/google/cityhash")
297 (synopsis "C++ hash functions for strings")
299 "CityHash provides hash functions for strings. The functions mix the
300 input bits thoroughly but are not suitable for cryptography.")
301 (license license:expat))))
309 (uri (string-append "http://www.and.org/ustr/" version
310 "/ustr-" version ".tar.bz2"))
313 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
314 (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
315 (build-system gnu-build-system)
320 ;; Override "/sbin/ldconfig" with "echo" because we don't need
323 (string-append "prefix=" (assoc-ref %outputs "out"))
326 (modify-phases %standard-phases
327 (add-after 'unpack 'disable-check-for-stdint
329 ;; Of course we have stdint.h, just not in /usr/include
330 (substitute* '("Makefile"
332 (("-f \"/usr/include/stdint.h\"") "-z \"\""))
334 ;; No configure script
335 (delete 'configure))))
336 (home-page "http://www.and.org/ustr/")
337 (synopsis "String library with very low memory overhead")
339 "Ustr is a string library for C with very low memory overhead.")
340 ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
341 ;; LGPL, etc. ... if you need another license to help compatibility, just
342 ;; ask for it. It's basically public domain, without all the legal
343 ;; problems for everyone that trying to make something public domain
345 (license license:public-domain)))
347 (define-public ascii2binary
349 (name "ascii2binary")
354 (uri (string-append "http://billposer.org/Software/Downloads/"
355 "ascii2binary-" version ".tar.bz2"))
357 (base32 "0dc9fxcdmppbs9s06jvq61zbk552laxps0xyk098gj41697ihd96"))))
358 (build-system gnu-build-system)
360 `(("gettext" ,gettext-minimal)))
361 (home-page "https://billposer.org/Software/a2b.html")
362 (synopsis "Convert between ASCII, hexadecimal and binary representations")
363 (description "The two programs are useful for generating test data, for
364 inspecting binary files, and for interfacing programs that generate textual
365 output to programs that require binary input and conversely. They can also be
366 useful when it is desired to reformat numbers.
370 @item @command{ascii2binary} reads input consisting of ascii or hexadecimal
371 representation numbers separated by whitespace and produces as output
372 the binary equivalents. The type and precision of the binary output
373 is selected using command line flags.
375 @item @command{binary2ascii} reads input consisting of binary numbers
376 and converts them to their ascii or hexadecimal representation.
377 Command line flags specify the type and size of the binary numbers
378 and provide control over the format of the output.
379 Unsigned integers may be written out in binary, octal, decimal,
382 Signed integers may be written out only in binary or decimal. Floating
383 point numbers may be written out only decimal, either in standard or
384 scientific notation. (If you want to examine the binary representation
385 of floating point numbers, just treat the input as a sequence of unsigned
389 (license license:gpl3)))
391 (define-public uniutils
398 (uri (string-append "http://billposer.org/Software/Downloads/"
399 "uniutils-" version ".tar.bz2"))
401 (base32 "19w1510w87gx7n4qy3zsb0m467a4rn5scvh4ajajg7jh6x5xri08"))))
402 (build-system gnu-build-system)
404 '(#:configure-flags '("--disable-dependency-tracking")
406 (modify-phases %standard-phases
407 (add-after 'build 'fix-paths
408 (lambda* (#:key outputs inputs #:allow-other-keys)
409 (let ((out (assoc-ref outputs "out"))
410 (a2b (assoc-ref inputs "ascii2binary"))
411 (iconv (assoc-ref inputs "libiconv")))
412 (substitute* "utf8lookup"
413 (("^ascii2binary ") (string-append a2b "/bin/ascii2binary "))
414 (("^uniname ") (string-append out "/bin/uniname "))
415 (("^iconv ") (string-append iconv "/bin/iconv ")))
418 `(("ascii2binary" ,ascii2binary)
419 ("libiconv" ,libiconv)))
420 (home-page "https://billposer.org/Software/unidesc.html")
421 (synopsis "Find out what is in a Unicode file")
422 (description "Useful tools when working with Unicode files when one
423 doesn't know the writing system, doesn't have the necessary font, needs to
424 inspect invisible characters, needs to find out whether characters have been
425 combined or in what order they occur, or needs statistics on which characters
430 @item @command{uniname} defaults to printing the character offset of each
431 character, its byte offset, its hex code value, its encoding, the glyph
432 itself, and its name. It may also be used to validate UTF-8 input.
434 @item @command{unidesc} reports the character ranges to which different
435 portions of the text belong. It can also be used to identify Unicode encodings
436 (e.g. UTF-16be) flagged by magic numbers.
438 @item @command{unihist} generates a histogram of the characters in its input.
440 @item @command{ExplicateUTF8} is intended for debugging or for learning about
441 Unicode. It determines and explains the validity of a sequence of bytes as a
444 @item @command{utf8lookup} provides a handy way to look up Unicode characters
445 from the command line.
447 @item @command{unireverse} reverse each line of UTF-8 input
448 character-by-character.
451 (license license:gpl3)))
453 (define-public libconfig
457 (home-page "https://hyperrealm.github.io/libconfig/")
460 (uri (string-append home-page "/dist/libconfig-"
464 "1ngs2qx3cx5cbwinc5mvadly0b5n7s86zsc68c404czzfff7lg3w"))))
465 (build-system gnu-build-system)
466 (synopsis "C/C++ configuration file library")
468 "Libconfig is a simple library for manipulating structured configuration
469 files. This file format is more compact and more readable than XML. And
470 unlike XML, it is type-aware, so it is not necessary to do string parsing in
472 (license license:lgpl2.1+)))
481 (url "https://github.com/pfff/pfff")
482 (commit (string-append "v" version))))
483 (file-name (git-file-name name version))
486 "1nxkfm7zliq3rmr7yp871sppwfnz71iz364m2sgazny71pzykggc"))))
487 (build-system cmake-build-system)
488 (home-page "https://biit.cs.ut.ee/pfff/")
489 (synopsis "Probabilistic fast file fingerprinting tool")
491 "pfff is a tool for calculating a compact digital fingerprint of a file
492 by sampling randomly from the file instead of reading it in full.
493 Consequently, the computation has a flat performance characteristic,
494 correlated with data variation rather than file size. pfff can be as reliable
495 as existing hashing techniques, with provably negligible risk of collisions.")
496 (license license:bsd-3)))
498 (define-public oniguruma
501 (version "6.9.5-rev1")
504 (uri (string-append "https://github.com/kkos/"
505 "oniguruma/releases/download/v"
506 ;; If there is a "-" in the version, convert
507 ;; to underscore for this part of the URI.
508 (string-map (lambda (c) (if (char=? #\- c) #\_ c))
510 "/onig-" version ".tar.gz"))
513 "17m92k1n6bvza6m35fpd5g36zwpwm3hfz3478iwj5bvj2sfq8g6k"))))
514 (build-system gnu-build-system)
515 (arguments '(#:configure-flags '("--disable-static")))
516 (home-page "https://github.com/kkos/oniguruma")
517 (synopsis "Regular expression library")
518 (description "Oniguruma is a regular expressions library. The special
519 characteristic of this library is that different character encoding for every
520 regular expression object can be specified.")
521 (license license:bsd-2)))
523 (define-public antiword
529 (uri (string-append "http://www.winfield.demon.nl/linux"
530 "/antiword-" version ".tar.gz"))
533 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
534 (patches (search-patches "antiword-CVE-2014-8123.patch"))))
535 (build-system gnu-build-system)
537 `(#:tests? #f ; There are no tests
539 (list "-f" "Makefile.Linux"
540 (string-append "GLOBAL_INSTALL_DIR="
541 (assoc-ref %outputs "out") "/bin")
542 (string-append "GLOBAL_RESOURCES_DIR="
543 (assoc-ref %outputs "out") "/share/antiword"))
545 (modify-phases %standard-phases
547 (lambda* (#:key outputs #:allow-other-keys)
548 ;; Ensure that mapping files can be found in the actual package
550 (substitute* "antiword.h"
551 (("/usr/share/antiword")
552 (string-append (assoc-ref outputs "out") "/share/antiword")))
555 (lambda* (#:key make-flags #:allow-other-keys)
556 (apply invoke "make" `("global_install" ,@make-flags)))))))
557 (home-page "http://www.winfield.demon.nl/")
558 (synopsis "Microsoft Word document reader")
559 (description "Antiword is an application for displaying Microsoft Word
560 documents. It can also convert the document to PostScript or XML. Only
561 documents made by MS Word version 2 and version 6 or later are supported. The
562 name comes from: \"The antidote against people who send Microsoft Word files
563 to everybody, because they believe that everybody runs Windows and therefore
565 (license license:gpl2+)))
567 (define-public catdoc
573 (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
574 "catdoc-" version ".tar.gz"))
575 (patches (search-patches "catdoc-CVE-2017-11110.patch"))
578 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
579 (build-system gnu-build-system)
580 ;; TODO: Also build `wordview` which requires `tk` – make a separate
583 '(#:tests? #f ; There are no tests
584 #:configure-flags '("--disable-wordview")
586 (modify-phases %standard-phases
587 (add-before 'install 'fix-install
588 (lambda* (#:key outputs #:allow-other-keys)
589 (let ((out (assoc-ref outputs "out")))
590 (mkdir-p (string-append out "/share/man/man1"))))))))
591 (home-page "https://www.wagner.pp.ru/~vitus/software/catdoc/")
592 (synopsis "MS-Word to TeX or plain text converter")
593 (description "@command{catdoc} extracts text from MS-Word files, trying to
594 preserve as many special printable characters as possible. It supports
595 everything up to Word-97. Also supported are MS Write documents and RTF files.
597 @command{catdoc} does not preserve complex word formatting, but it can
598 translate some non-ASCII characters into TeX escape codes. It's goal is to
599 extract plain text and allow you to read it and, probably, reformat with TeX,
600 according to TeXnical rules.
602 This package also provides @command{xls2csv}, which extracts data from Excel
603 spreadsheets and outputs it in comma-separated-value format, and
604 @command{catppt}, which extracts data from PowerPoint presentations.")
605 (license license:gpl2+)))
607 (define-public utfcpp
614 (url "https://github.com/nemtrif/utfcpp")
615 (commit (string-append "v" version))))
616 (file-name (git-file-name name version))
619 "1gr98d826z6wa58r1s5i7rz7q2x3r31v7zj0pjjlrc7gfxwklr4s"))))
620 (build-system cmake-build-system)
622 `(#:out-of-source? #f
624 (modify-phases %standard-phases
625 (replace 'install ; no install target
626 (lambda* (#:key outputs #:allow-other-keys)
627 (let* ((out (assoc-ref outputs "out"))
628 (include (string-append out "/include"))
629 (doc (string-append out "/share/doc/" ,name)))
630 (copy-recursively "source" include)
631 (install-file "README.md" doc)
633 (home-page "https://github.com/nemtrif/utfcpp")
634 (synopsis "Portable C++ library for handling UTF-8")
635 (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
637 (license license:boost1.0)))
646 (uri (string-append "mirror://sourceforge/dbacl/dbacl/" version "/"
647 "dbacl-" version ".tar.gz"))
649 (base32 "1gas0112wqjvwn9qg3hxnawk7h3prr0w9b2h68f3p1ifd1kzn3gz"))
650 (patches (search-patches "dbacl-include-locale.h.patch"))))
651 (build-system gnu-build-system)
655 (string-append "-I" (assoc-ref %build-inputs "slang")
657 (string-append "-I" (assoc-ref %build-inputs "ncurses")
660 (modify-phases %standard-phases
661 (add-after 'unpack 'delete-sample6-and-japanese
663 (substitute* "doc/Makefile.am"
665 (("japanese.txt") ""))
666 (delete-file "doc/sample6.txt")
667 (delete-file "doc/japanese.txt")
668 (substitute* (list "src/tests/Makefile.am"
669 "src/tests/Makefile.in")
670 (("dbacl-jap.shin") "")
671 (("dbacl-jap.sh") ""))
673 (add-after 'unpack 'delete-test
674 ;; See comments about the license.
676 (delete-file "src/tests/dbacl-jap.shin")
678 (add-after 'unpack 'fix-test-files
679 (lambda* (#:key inputs outputs #:allow-other-keys)
680 (let* ((out (assoc-ref outputs "out"))
681 (bin (string-append out "/bin")))
682 (substitute* (find-files "src/tests/" "\\.shin$")
683 (("PATH=/bin:/usr/bin")
684 "#PATH=/bin:/usr/bin")
685 (("diff") (string-append (which "diff")))
686 (("tr") (string-append (which "tr"))))
690 (invoke "autoreconf" "-vif")
693 `(("ncurses" ,ncurses)
695 ("readline" ,readline)
698 `(("libtool" ,libtool)
699 ("autoconf" ,autoconf)
700 ("automake" ,automake)
701 ("pkg-config" ,pkg-config)))
702 (home-page "https://www.lbreyer.com/dbacl.html")
703 (synopsis "Bayesian text and email classifier")
705 "dbacl is a fast Bayesian text and email classifier. It builds a variety
706 of language models using maximum entropy (minimum divergence) principles, and
707 these can then be used to categorize input data automatically among multiple
709 ;; The software is licensed as GPLv3 or later, but
710 ;; includes various sample texts in the doc dir:
711 ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
713 ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
714 ;; - sample6.txt is a forwarded email, copyright unknown.
715 ;; Guix does exclude sample6.txt.
716 ;; - japanese.txt is a Japanese unoffical translation of the
717 ;; GNU General Public License, (c) by the Free Software Foundation.
718 ;; Guix excludes this file.
719 (license (list license:gpl3+ license:public-domain))))
721 (define-public dotconf
728 (url "https://github.com/williamh/dotconf")
729 (commit (string-append "v" version))))
730 (file-name (git-file-name name version))
733 "1sc95hw5k2xagpafny0v35filmcn05k1ds5ghkldfpf6xw4hakp7"))))
734 (build-system gnu-build-system)
735 (arguments `(#:tests? #f)) ; FIXME maketest.sh does not work.
737 `(("autoconf" ,autoconf)
738 ("automake" ,automake)
739 ("libtool" ,libtool)))
740 (home-page "https://github.com/williamh/dotconf")
741 (synopsis "Configuration file parser library")
743 "C library for creating and parsing configuration files.")
744 (license (list license:lgpl2.1 ; Main distribution.
745 license:asl1.1)))) ; src/readdir.{c,h}
747 (define-public drm-tools
753 (uri (string-append "mirror://sourceforge/drmtools/drm_tools-"
757 "187zbxw21zcg8gpyc13gxlycfw0n05a6rmqq6im5wr9zk1v1wj80"))))
758 (build-system cmake-build-system)
760 `(#:tests? #f ;the test suite fails
761 #:phases (modify-phases %standard-phases
762 (add-after 'unpack 'set-install-prefixes
763 (lambda* (#:key outputs #:allow-other-keys)
764 (let* ((out (assoc-ref outputs "out")))
765 (substitute* "CMakeLists.txt"
771 (add-after 'unpack 'adjust-test-paths
773 (substitute* '("test_extract_increment.sh"
774 "test_extract_features.sh"
775 "test_extract_features2.sh"
777 (("\\./extract") "extract")
778 (("\\./dmath") "dmath")
779 (("/usr/local/bin/") "")
784 ;; The produced binaries are written directly to %output/bin.
786 (add-after 'build 'check
787 (lambda* (#:key outputs tests? #:allow-other-keys)
789 (let* ((out (assoc-ref outputs "out"))
790 (bin (string-append out "/bin")))
791 (setenv "PATH" (string-append bin ":"
793 (with-directory-excursion
794 (format #f "../drm_tools-~a" ,version)
795 (invoke "sh" "test_all.sh")))))))))
796 (native-inputs `(("which" ,which))) ;for tests
797 (inputs `(("pcre" ,pcre)))
798 (home-page "http://drmtools.sourceforge.net/")
799 (synopsis "Utilities to manipulate text and binary files")
800 (description "The drm_tools package contains the following commands:
803 An extended version of the \"date\" program that has sub-second accuracy.
805 Format complex binary data into text.
807 Load data into a binary file using simple commands from the input.
809 Sort, merge, search, retrieve or generate test data consisting of fixed size
812 Find or find/replace in binary files.
814 Split test data consisting of fixed size binary records into one or more
817 Find changes between two files at the character level. Unlike \"diff\", it
818 lists just the characters that differ, so if the 40,000th character is
819 different only that one character will be shown, not the entire line.
821 Add columns of integers, decimals, and/or times.
823 A utility for formatting binary data dumps.
825 Double precision interactive command line math calculator.
827 Extract and emit data from text files based on character or token position.
829 A utility that reads from STDIN and executes each line as a command in a
832 A utility for rapid retrieval of text by line numbers, in any order, from a
837 Create message queues and send/receive messages.
840 Multiple buffer in and out. Used for buffering a lot of data between a slow
841 device and a fast device. Mostly for buffering streaming tape drives for use
842 with slower network connections, so that streaming is maintained as much as
843 possible to minimize wear on the tape device.
845 Corrupt data streams - useful for testing error correction and data recovery.
847 Filter, list, or split a tar file.
849 (license license:gpl2+)))
851 (define-public java-rsyntaxtextarea
853 (name "java-rsyntaxtextarea")
858 (url "https://github.com/bobbylight/RSyntaxTextArea")
860 (file-name (git-file-name name version))
863 "0dyflzvxq2wvs0rgqfyi5yzzrb6r4bzw2dm8cl304dakxk38ddys"))))
864 (build-system ant-build-system)
866 `(;; FIXME: some tests fail because locale resources cannot be found.
867 ;; Even when I add them to the class path,
868 ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
870 #:jar-name "rsyntaxtextarea.jar"))
872 `(("java-junit" ,java-junit)
873 ("java-hamcrest-core" ,java-hamcrest-core)))
874 (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
875 (synopsis "Syntax highlighting text component for Java Swing")
876 (description "RSyntaxTextArea is a syntax highlighting, code folding text
877 component for Java Swing. It extends @code{JTextComponent} so it integrates
878 completely with the standard @code{javax.swing.text} package. It is fast and
879 efficient, and can be used in any application that needs to edit or view
881 (license license:bsd-3)))
883 ;; We use the sources from git instead of the tarball from pypi, because the
884 ;; latter does not include the Cython source file from which bycython.cpp is
886 (define-public python-editdistance
887 (let ((commit "3ea84a7dd3258c76aa3be851ef3d50e59c886846")
890 (name "python-editdistance")
891 (version (string-append "0.3.1-" revision "." (string-take commit 7)))
896 (url "https://github.com/aflc/editdistance")
898 (file-name (git-file-name name version))
901 "1l43svsv12crvzphrgi6x435z6xg8m086c64armp8wzb4l8ccm7g"))))
902 (build-system python-build-system)
905 (modify-phases %standard-phases
906 (add-after 'unpack 'build-cython-code
908 (with-directory-excursion "editdistance"
909 (delete-file "bycython.cpp")
910 (invoke "cython" "--cplus" "bycython.pyx")))))))
912 `(("python-cython" ,python-cython)))
913 (home-page "https://www.github.com/aflc/editdistance")
914 (synopsis "Fast implementation of the edit distance (Levenshtein distance)")
916 "This library simply implements Levenshtein distance algorithm with C++
918 (license license:expat))))
920 (define-public go-github.com-mattn-go-runewidth
921 (let ((commit "703b5e6b11ae25aeb2af9ebb5d5fdf8fa2575211")
925 (name "go-github.com-mattn-go-runewidth")
926 (version (git-version version revision commit))
931 (url "https://github.com/mattn/runewidth")
933 (file-name (git-file-name name version))
936 "0znpyz71gajx3g0j2zp63nhjj2c07g16885vxv4ykwnrfmzbgk4w"))))
937 (build-system go-build-system)
939 '(#:import-path "github.com/mattn/go-runewidth"))
940 (synopsis "@code{runewidth} provides Go functions to work with string widths")
942 "The @code{runewidth} library provides Go functions for padding,
943 measuring and checking the width of strings, with support for East Asian
945 (home-page "https://github.com/mattn/runewidth")
946 (license license:expat))))
948 (define-public docx2txt
955 "mirror://sourceforge/docx2txt/docx2txt/v"
956 version "/docx2txt-" version ".tgz"))
959 "06vdikjvpj6qdb41d8wzfnyj44jpnknmlgbhbr1w215420lpb5xj"))))
960 (build-system gnu-build-system)
965 `(#:tests? #f ; No tests.
966 #:make-flags (list (string-append "BINDIR="
967 (assoc-ref %outputs "out") "/bin")
968 (string-append "CONFIGDIR="
969 (assoc-ref %outputs "out") "/etc")
970 ;; Makefile seems to be a bit dumb at guessing.
971 (string-append "INSTALL=install")
972 (string-append "PERL=perl"))
974 (modify-phases %standard-phases
976 (add-after 'install 'fix-install
977 (lambda* (#:key outputs inputs #:allow-other-keys)
978 (let* ((out (assoc-ref outputs "out"))
979 (bin (string-append out "/bin"))
980 (config (string-append out "/etc/docx2txt.config"))
981 (unzip (assoc-ref inputs "unzip")))
982 ;; According to INSTALL, the .sh wrapper can be skipped.
983 (delete-file (string-append bin "/docx2txt.sh"))
984 (rename-file (string-append bin "/docx2txt.pl")
985 (string-append bin "/docx2txt"))
987 (("config_unzip => '/usr/bin/unzip',")
988 (string-append "config_unzip => '"
991 ;; Makefile is wrong.
994 (synopsis "Recover text from @file{.docx} files, with good formatting")
996 "@command{docx2txt} is a Perl based command line utility to convert
997 Microsoft Office @file{.docx} documents to equivalent text documents. Latest
998 version supports following features during text extraction.
1001 @item Character conversions; currency characters are converted to respective
1003 @item Capitalisation of text blocks.
1004 @item Center and right justification of text fitting in a line of
1005 (configurable) 80 columns.
1006 @item Horizontal ruler, line breaks, paragraphs separation, tabs.
1007 @item Indicating hyperlinked text along with the hyperlink (configurable).
1008 @item Handling (bullet, decimal, letter, roman) lists along with (attempt at)
1011 (home-page "http://docx2txt.sourceforge.net")
1012 (license license:gpl3+)))
1014 (define-public odt2txt
1022 (url "https://github.com/dstosberg/odt2txt/")
1023 (commit (string-append "v" version))))
1024 (file-name (git-file-name name version))
1027 "0im3kzvhxkjlx57w6h13mc9584c74ma1dyymgvpq2y61av3gc35v"))))
1028 (build-system gnu-build-system)
1030 `(#:tests? #f ; no make check
1031 #:make-flags (list "CC=gcc"
1032 (string-append "DESTDIR=" (assoc-ref %outputs "out")))
1034 (modify-phases %standard-phases
1035 ;; no configure script
1036 (delete 'configure))))
1039 (home-page "https://github.com/dstosberg/odt2txt/")
1040 (synopsis "Converter from OpenDocument Text to plain text")
1041 (description "odt2txt is a command-line tool which extracts the text out
1042 of OpenDocument Texts, as produced by OpenOffice.org, KOffice, StarOffice and
1045 odt2txt can also extract text from some file formats similar to OpenDocument
1046 Text, such as OpenOffice.org XML (*.sxw), which was used by OpenOffice.org
1047 version 1.x and older StarOffice versions. To a lesser extent, odt2txt may be
1048 useful to extract content from OpenDocument spreadsheets (*.ods) and
1049 OpenDocument presentations (*.odp).")
1050 (license license:gpl2)))
1052 (define-public opencc
1060 (url "https://github.com/BYVoid/OpenCC")
1061 (commit (string-append "ver." version))))
1062 (file-name (git-file-name name version))
1065 "07y0pvxjlmpcnym229j87qcfwnfm7vi10dad0a20xy6as55a9j3d"))
1066 (modules '((guix build utils)))
1069 ;; TODO: Unbundle tclap, darts-clone, gtest
1070 (delete-file-recursively "deps/rapidjson-1.1.0") #t))))
1071 (build-system cmake-build-system)
1073 ;; Required to locate the install script properly.
1074 `(#:out-of-source? #f
1075 #:parallel-build? #f ;occasionally failed.
1077 (,@%cmake-build-system-modules
1078 (guix build python-build-system))
1079 #:modules ((guix build cmake-build-system)
1080 ((guix build python-build-system) #:prefix python:)
1083 (modify-phases %standard-phases
1084 (add-after 'unpack 'prevent-rebuild-during-installation
1086 (substitute* "python/setup.py"
1087 (("'build_py': BuildPyCommand,") ""))
1089 (add-after 'unpack 'patch-3rd-party-references
1090 (lambda* (#:key inputs #:allow-other-keys)
1091 (let ((rapidjson (assoc-ref inputs "rapidjson")))
1092 (substitute* "src/CMakeLists.txt"
1093 (("../deps/rapidjson-1.1.0")
1094 (string-append rapidjson "/include/rapidjson")))
1096 (add-before 'configure 'patch-python-binding-installation
1097 (lambda* (#:key outputs inputs #:allow-other-keys)
1098 (let ((out (assoc-ref outputs "out")))
1099 (substitute* "python/opencc/__init__.py"
1100 (("(_libopenccfile =).*$" _ prefix)
1101 (format #f "~a os.path.join('~a/lib', _libopenccfilename)~%"
1103 (("(_opencc_share_dir =).*$" _ prefix)
1104 (format #f "~a '~a/share/opencc'~%" prefix out))))
1106 (add-after 'install 'install-python-binding
1107 (lambda* (#:key outputs inputs #:allow-other-keys)
1108 (let* ((out (assoc-ref outputs "out"))
1109 (dist (string-append
1111 ,(version-major+minor (package-version python))
1115 (setenv "PYTHONPATH"
1116 (string-append dist ":" (getenv "PYTHONPATH")))
1117 (invoke "python" "setup.py" "install"
1118 "--root=/" "--single-version-externally-managed"
1119 (string-append "--prefix=" out))
1121 (add-before 'install-python-binding 'enable-bytecode-determinism
1122 (assoc-ref python:%standard-phases 'enable-bytecode-determinism)))))
1124 `(("python" ,python-wrapper)
1125 ("rapidjson" ,rapidjson)
1126 ("python-setuptools" ,python-setuptools)
1127 ("python-wheel" ,python-wheel)))
1128 (home-page "https://github.com/BYVoid/OpenCC")
1129 (synopsis "Convert between Traditional Chinese and Simplified Chinese")
1130 (description "Open Chinese Convert (OpenCC) converts between Traditional
1131 Chinese and Simplified Chinese, supporting character-level conversion,
1132 phrase-level conversion, variant conversion, and regional idioms among
1133 Mainland China, Taiwan, and Hong-Kong.")
1134 (license license:asl2.0)))
1137 (let ((commit "08043eadf4abdddcf277842217e3c77a24740dc2")
1141 ;; The commits corresponding to specific versions are published
1143 ;; https://ja.osdn.net/projects/nkf/scm/git/nkf/
1148 (url "https://github.com/nurse/nkf")
1150 (file-name (git-file-name name version))
1153 "0anw0knr1iy4p9w3d3b3pbwzh1c43p1i2q4c28kw9zviw8kx2rly"))))
1154 (build-system gnu-build-system)
1156 `(#:tests? #f ; test for perl module
1157 #:make-flags (list "CC=gcc" "CFLAGS=-O2 -Wall -pedantic"
1158 (string-append "prefix=" %output)
1161 (modify-phases %standard-phases
1162 (delete 'configure)))) ; No ./configure script
1163 (home-page "https://ja.osdn.net/projects/nkf/")
1164 (synopsis "Network Kanji Filter")
1165 (description "Nkf is yet another kanji code converter among networks,
1166 hosts and terminals. It converts input kanji code to designated kanji code
1167 such as ISO-2022-JP, Shift_JIS, EUC-JP, UTF-8, UTF-16 or UTF-32.")
1168 (license license:zlib))))
1170 (define-public python-pandocfilters
1172 (name "python-pandocfilters")
1177 (uri (pypi-uri "pandocfilters" version))
1179 (base32 "1sq675dg4barb5949xxz4d5gk2ly524hi1p1xgwb3d1l0nsznqxw"))))
1180 (build-system python-build-system)
1181 (home-page "https://github.com/jgm/pandocfilters")
1182 (synopsis "Python module for writing Pandoc filters")
1183 (description "Pandoc is a powerful utility to transform various
1184 input formats into a wide range of output formats. To alter the
1185 exported output document, Pandoc allows the usage of filters, which
1186 are pipes that read a JSON serialization of the Pandoc AST from stdin,
1187 transform it in some way, and write it to stdout. It allows therefore
1188 to alter the processing of Pandoc's supported input formats, for
1189 instance one can add new syntax elements to markdown, etc.
1191 This package provides Python bindings.")
1192 (license license:bsd-3)))
1202 (url "https://github.com/theZiz/aha")
1205 (base32 "1gywad0rvvz3c5balz8cxsnx0562hj2ngzqyr8zsy2mb4pn0lpgv"))
1206 (file-name (git-file-name name version))))
1207 (build-system gnu-build-system)
1210 (modify-phases %standard-phases
1211 (delete 'configure))
1212 #:make-flags (list (string-append "CC=" ,(cc-for-target))
1213 (string-append "PREFIX="
1214 (assoc-ref %outputs "out")))
1215 #:tests? #f)) ; no test suite
1216 (home-page "https://github.com/theZiz/aha")
1217 (synopsis "Converts terminal escape sequences to HTML")
1218 (description "@command{aha} (Ansi Html Adapter) converts ANSI escape sequences
1219 of a Unix terminal to HTML code.")
1220 (license (list license:lgpl2.0+ license:mpl1.1))))
1222 (define-public go-github-com-errata-ai-vale
1224 (name "go-github-com-errata-ai-vale")
1230 (url "https://github.com/errata-ai/vale")
1231 (commit (string-append "v" version))))
1233 (base32 "0d07fwha2220m8j24h527xl0gnl3svvyaywflgk5292d6g49ach2"))
1234 (file-name (git-file-name name version))))
1235 (build-system go-build-system)
1237 `(("github.com/sp13/afero" ,go-github-com-spf13-afero)
1238 ("github.com/urfave/cli" ,go-github-com-urfave-cli)
1239 ("github.com/olekukonko/tablewriter" ,go-github-com-olekukonko-tablewriter)
1240 ("github.com/mitchellh/mapstructure" ,go-github-com-mitchellh-mapstructure)))
1242 `(#:import-path "github.com/errata-ai/vale"
1243 #:install-source? #f))
1244 (home-page "https://github.com/errata-ai/vale")
1245 (synopsis "Fully customizable syntax-aware linter that focuses on your style")
1247 "Vale is a fully extensible linter that focuses on your own writing style
1248 by making use of rules in individual YAML files. It is syntax-aware on markup
1249 languages such as HTML, Markdown, Asciidoc, and reStructuredText. The community
1250 around it also has a list of style guides implemented with Vale in
1251 @url{https://github.com/errata-ai/styles, their styles repo}.")
1252 (license license:expat)))