1 ;;; GNU Guix --- Functional package management for GNU
2 ;;; Copyright © 2015 Taylan Ulrich Bayırlı/Kammer <taylanbayirli@gmail.com>
3 ;;; Copyright © 2015, 2016, 2017 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 Efraim Flashner <efraim@flashner.co.il>
9 ;;; Copyright © 2016 ng0 <ng0@we.make.ritual.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 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 Tobias Geerinckx-Rice <me@tobias.gr>
18 ;;; This file is part of GNU Guix.
20 ;;; GNU Guix is free software; you can redistribute it and/or modify it
21 ;;; under the terms of the GNU General Public License as published by
22 ;;; the Free Software Foundation; either version 3 of the License, or (at
23 ;;; your option) any later version.
25 ;;; GNU Guix is distributed in the hope that it will be useful, but
26 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
27 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
28 ;;; GNU General Public License for more details.
30 ;;; You should have received a copy of the GNU General Public License
31 ;;; along with GNU Guix. If not, see <http://www.gnu.org/licenses/>.
33 (define-module (gnu packages textutils)
34 #:use-module ((guix licenses) #:prefix license:)
35 #:use-module (guix packages)
36 #:use-module (guix download)
37 #:use-module (guix git-download)
38 #:use-module (guix build-system ant)
39 #:use-module (guix build-system gnu)
40 #:use-module (guix build-system go)
41 #:use-module (guix build-system cmake)
42 #:use-module (guix build-system python)
43 #:use-module (gnu packages)
44 #:use-module (gnu packages autotools)
45 #:use-module (gnu packages compression)
46 #:use-module (gnu packages gettext)
47 #:use-module (gnu packages java)
48 #:use-module (gnu packages ncurses)
49 #:use-module (gnu packages perl)
50 #:use-module (gnu packages pkg-config)
51 #:use-module (gnu packages python)
52 #:use-module (gnu packages readline)
53 #:use-module (gnu packages slang))
55 (define-public dos2unix
62 (uri (string-append "https://waterlan.home.xs4all.nl/" name "/"
63 name "-" version ".tar.gz"))
65 (base32 "12h4c61g376bhq03y5g2xszkrkrj5hwd928rly3xsp6rvfmnbixs"))))
66 (build-system gnu-build-system)
68 '(#:make-flags (list "CC=gcc"
69 (string-append "prefix=" (assoc-ref %outputs "out")))
71 (modify-phases %standard-phases
72 (delete 'configure)))) ; no configure script
74 `(("gettext" ,gettext-minimal)
76 (home-page "https://waterlan.home.xs4all.nl/dos2unix.html")
77 (synopsis "DOS/Mac to Unix and vice versa text file format converter")
79 "dos2unix is a tool to convert line breaks in a text file from Unix format
80 to DOS format and vice versa.")
81 (license license:bsd-2)))
90 (uri (string-append "https://github.com/rrthomas/recode/releases/"
91 "download/v" version "/" name "-" version ".tar.gz"))
94 "0r4yhf7i7zp2nl2apyzz7r3i2in12n385hmr8zcfr18ly0ly530q"))
95 (modules '((guix build utils)))
97 (delete-file "tests/Recode.c")
99 (build-system gnu-build-system)
101 `(("python" ,python-2)
102 ("python2-cython" ,python2-cython)))
103 (home-page "https://github.com/rrthomas/recode")
104 (synopsis "Text encoding converter")
105 (description "The Recode library converts files between character sets and
106 usages. It recognises or produces over 200 different character sets (or about
107 300 if combined with an iconv library) and transliterates files between almost
108 any pair. When exact transliteration are not possible, it gets rid of
109 offending characters or falls back on approximations. The recode program is a
110 handy front-end to the library.")
111 (license license:gpl3+)))
121 "https://github.com/nijel/enca/archive/" version ".tar.gz"))
123 (base32 "099z526i7qgij7q1w3lvhl88iv3jc3nqxca2i09h6s08ghyrmzf4"))
124 (file-name (string-append name "-" version ".tar.gz"))))
125 (build-system gnu-build-system)
126 ;; enca-1.19 tests fail with recent recode.
127 ;(inputs `(("recode" ,recode)))
128 (home-page "https://github.com/nijel/enca")
129 (synopsis "Text encoding detection tool")
130 (description "Enca (Extremely Naive Charset Analyser) consists of libenca,
131 an encoding detection library, and enca, a command line frontend, integrating
132 libenca and several charset conversion libraries and tools.")
133 (license license:gpl2)))
135 (define-public utf8proc
143 "https://github.com/JuliaLang/utf8proc/archive/v"
145 (file-name (string-append name "-" version ".tar.gz"))
147 (base32 "0q1jhdkk4f9b0zb8s2ql3sba3br5nvjsmbsaybmgj064k9hwbk15"))))
148 (build-system gnu-build-system)
149 (inputs ;test data that is otherwise downloaded with curl
150 `(("NormalizationTest.txt"
153 (uri (string-append "http://www.unicode.org/Public/9.0.0/ucd/"
154 "NormalizationTest.txt"))
156 (base32 "1fxrz0bilsbwl685336aqi88k62i6nqhm62rvy4zhg3bcm4dhj1d"))))
157 ("GraphemeBreakTest.txt"
160 (uri (string-append "http://www.unicode.org/Public/9.0.0/ucd/"
161 "auxiliary/GraphemeBreakTest.txt"))
163 (base32 "0qbhyhmf0778lc2hcwlpizrvmdxwpk959v2q2wb8abv09ba7wvn7"))))))
165 '(#:make-flags (list "CC=gcc"
166 (string-append "prefix=" (assoc-ref %outputs "out")))
168 (modify-phases %standard-phases
170 (add-before 'check 'check-data
171 (lambda* (#:key inputs #:allow-other-keys)
172 (for-each (lambda (i)
173 (copy-file (assoc-ref inputs i)
174 (string-append "data/" i)))
175 '("NormalizationTest.txt" "GraphemeBreakTest.txt"))
176 (substitute* "data/GraphemeBreakTest.txt"
179 (home-page "http://julialang.org/utf8proc/")
180 (synopsis "C library for processing UTF-8 Unicode data")
181 (description "utf8proc is a small C library that provides Unicode
182 normalization, case-folding, and other operations for data in the UTF-8
183 encoding, supporting Unicode version 9.0.0.")
184 (license license:expat)))
186 (define-public libgtextutils
188 (name "libgtextutils")
194 "https://github.com/agordon/libgtextutils/releases/download/"
195 version "/libgtextutils-" version ".tar.gz"))
197 (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
198 (build-system gnu-build-system)
201 (modify-phases %standard-phases
202 (add-after 'unpack 'autoreconf
203 (lambda _ (zero? (system* "autoreconf" "-vif")))))))
205 `(("autoconf" ,autoconf)
206 ("automake" ,automake)
207 ("libtool" ,libtool)))
208 (home-page "https://github.com/agordon/libgtextutils")
209 (synopsis "Gordon's text utils library")
211 "libgtextutils is a text utilities library used by the fastx toolkit from
213 (license license:agpl3+)))
215 (define-public cityhash
216 (let ((commit "8af9b8c"))
219 (version (string-append "1.1-2." commit))
223 (url "https://github.com/google/cityhash.git")
225 (file-name (string-append name "-" version ".tar.gz"))
228 "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
229 (build-system gnu-build-system)
231 '(#:make-flags (list "CXXFLAGS=-g -O3")
233 (modify-phases %standard-phases
234 ;; citycrc is not installed by default but is used by some
236 (add-after 'install 'install-citycrc
237 (lambda* (#:key outputs #:allow-other-keys)
238 (let* ((out (assoc-ref outputs "out"))
239 (include (string-append out "/include")))
240 (install-file "src/citycrc.h" include))
242 (home-page "https://github.com/google/cityhash")
243 (synopsis "C++ hash functions for strings")
245 "CityHash provides hash functions for strings. The functions mix the
246 input bits thoroughly but are not suitable for cryptography.")
247 (license license:expat))))
255 (uri (string-append "http://www.and.org/ustr/" version
256 "/ustr-" version ".tar.bz2"))
259 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))
260 (patches (search-patches "ustr-fix-build-with-gcc-5.patch"))))
261 (build-system gnu-build-system)
266 ;; Override "/sbin/ldconfig" with "echo" because we don't need
269 (string-append "prefix=" (assoc-ref %outputs "out"))
272 (modify-phases %standard-phases
273 (add-after 'unpack 'disable-check-for-stdint
275 ;; Of course we have stdint.h, just not in /usr/include
276 (substitute* '("Makefile"
278 (("-f \"/usr/include/stdint.h\"") "-z \"\""))
280 ;; No configure script
281 (delete 'configure))))
282 (home-page "http://www.and.org/ustr/")
283 (synopsis "String library with very low memory overhead")
285 "Ustr is a string library for C with very low memory overhead.")
286 ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
287 ;; LGPL, etc. ... if you need another license to help compatibility, just
288 ;; ask for it. It's basically public domain, without all the legal
289 ;; problems for everyone that trying to make something public domain
291 (license license:public-domain)))
293 (define-public libconfig
299 (uri (string-append "http://www.hyperrealm.com/libconfig/"
300 "libconfig-" version ".tar.gz"))
303 "1xh3hzk63v4y8815lc5209m3s6ms2cpgw4h5hg462i4f1lwsl7g3"))))
304 (build-system gnu-build-system)
305 (home-page "http://www.hyperrealm.com/libconfig/")
306 (synopsis "C/C++ configuration file library")
308 "Libconfig is a simple library for manipulating structured configuration
309 files. This file format is more compact and more readable than XML. And
310 unlike XML, it is type-aware, so it is not necessary to do string parsing in
312 (license license:lgpl2.1+)))
320 (uri (string-append "https://github.com/pfff/pfff/archive/v"
322 (file-name (string-append name "-" version ".tar.gz"))
325 "00m553aa277iarxj6dalmklyb64r7ias49bfwzbacsfg8h3kar8m"))))
326 (build-system cmake-build-system)
327 (home-page "http://biit.cs.ut.ee/pfff/")
328 (synopsis "Probabilistic fast file fingerprinting tool")
330 "pfff is a tool for calculating a compact digital fingerprint of a file
331 by sampling randomly from the file instead of reading it in full.
332 Consequently, the computation has a flat performance characteristic,
333 correlated with data variation rather than file size. pfff can be as reliable
334 as existing hashing techniques, with provably negligible risk of collisions.")
335 (license license:bsd-3)))
337 (define-public oniguruma
343 (uri (string-append "https://github.com/kkos/"
344 "oniguruma/releases/download/v" version
345 "/onig-" version ".tar.gz"))
348 "19s79vsclqn170mw0ajwv7j37qsbn4f1yjz3yavnhvva6c820r6m"))))
349 (build-system gnu-build-system)
350 (home-page "https://github.com/kkos/oniguruma")
351 (synopsis "Regular expression library")
352 (description "Oniguruma is a regular expressions library. The special
353 characteristic of this library is that different character encoding for every
354 regular expression object can be specified.")
355 (license license:bsd-2)))
357 (define-public antiword
363 (uri (string-append "http://www.winfield.demon.nl/linux"
364 "/antiword-" version ".tar.gz"))
367 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
368 (patches (search-patches "antiword-CVE-2014-8123.patch"))))
369 (build-system gnu-build-system)
371 `(#:tests? #f ; There are no tests
373 (list "-f" "Makefile.Linux"
374 (string-append "GLOBAL_INSTALL_DIR="
375 (assoc-ref %outputs "out") "/bin")
376 (string-append "GLOBAL_RESOURCES_DIR="
377 (assoc-ref %outputs "out") "/share/antiword"))
379 (modify-phases %standard-phases
381 (lambda* (#:key outputs #:allow-other-keys)
382 ;; Ensure that mapping files can be found in the actual package
384 (substitute* "antiword.h"
385 (("/usr/share/antiword")
386 (string-append (assoc-ref outputs "out") "/share/antiword")))
389 (lambda* (#:key make-flags #:allow-other-keys)
390 (zero? (apply system* "make" `("global_install" ,@make-flags))))))))
391 (home-page "http://www.winfield.demon.nl/")
392 (synopsis "Microsoft Word document reader")
393 (description "Antiword is an application for displaying Microsoft Word
394 documents. It can also convert the document to PostScript or XML. Only
395 documents made by MS Word version 2 and version 6 or later are supported. The
396 name comes from: \"The antidote against people who send Microsoft Word files
397 to everybody, because they believe that everybody runs Windows and therefore
399 (license license:gpl2+)))
401 (define-public catdoc
407 (uri (string-append "http://ftp.wagner.pp.ru/pub/catdoc/"
408 "catdoc-" version ".tar.gz"))
409 (patches (search-patches "catdoc-CVE-2017-11110.patch"))
412 "15h7v3bmwfk4z8r78xs5ih6vd0pskn0rj90xghvbzdjj0cc88jji"))))
413 (build-system gnu-build-system)
414 ;; TODO: Also build `wordview` which requires `tk` – make a separate
417 '(#:tests? #f ; There are no tests
418 #:configure-flags '("--disable-wordview")
420 (modify-phases %standard-phases
421 (add-before 'install 'fix-install
422 (lambda* (#:key outputs #:allow-other-keys)
423 (let ((out (assoc-ref outputs "out")))
424 (mkdir-p (string-append out "/share/man/man1"))))))))
425 (home-page "http://www.wagner.pp.ru/~vitus/software/catdoc/")
426 (synopsis "MS-Word to TeX or plain text converter")
427 (description "@command{catdoc} extracts text from MS-Word files, trying to
428 preserve as many special printable characters as possible. It supports
429 everything up to Word-97. Also supported are MS Write documents and RTF files.
431 @command{catdoc} does not preserve complex word formatting, but it can
432 translate some non-ASCII characters into TeX escape codes. It's goal is to
433 extract plain text and allow you to read it and, probably, reformat with TeX,
434 according to TeXnical rules.
436 This package also provides @command{xls2csv}, which extracts data from Excel
437 spreadsheets and outputs it in comma-separated-value format, and
438 @command{catppt}, which extracts data from PowerPoint presentations.")
439 (license license:gpl2+)))
441 (define-public utfcpp
448 (string-append "https://github.com/nemtrif/utfcpp/archive/v"
450 (file-name (string-append name "-" version ".tar.gz"))
453 "0gcqcfw19kfim8xw29xdp91l310yfjyrqdj2zsx8xx02dkpy1zzk"))))
454 (build-system cmake-build-system)
456 `(#:out-of-source? #f
458 (modify-phases %standard-phases
459 (replace 'install ; no install target
460 (lambda* (#:key outputs #:allow-other-keys)
461 (let* ((out (assoc-ref outputs "out"))
462 (include (string-append out "/include"))
463 (doc (string-append out "/share/doc/" ,name)))
464 (copy-recursively "source" include)
465 (install-file "README.md" doc)
467 (home-page "https://github.com/nemtrif/utfcpp")
468 (synopsis "Portable C++ library for handling UTF-8")
469 (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
471 (license license:boost1.0)))
480 (uri (string-append "http://www.lbreyer.com/gpl/"
481 name "-" version ".tar.gz"))
484 "0224g6x71hyvy7jikfxmgcwww1r5lvk0jx36cva319cb9nmrbrq7"))))
485 (build-system gnu-build-system)
489 (string-append "-I" (assoc-ref %build-inputs "slang")
491 (string-append "-I" (assoc-ref %build-inputs "ncurses")
494 (modify-phases %standard-phases
495 (add-after 'unpack 'delete-sample6-and-japanese
497 (substitute* "doc/Makefile.am"
499 (("japanese.txt") ""))
500 (delete-file "doc/sample6.txt")
501 (delete-file "doc/japanese.txt")
502 (substitute* (list "src/tests/Makefile.am"
503 "src/tests/Makefile.in")
504 (("dbacl-jap.shin") "")
505 (("dbacl-jap.sh") ""))
507 (add-after 'unpack 'delete-test
508 ;; See comments about the license.
510 (delete-file "src/tests/dbacl-jap.shin")
512 (add-after 'delete-sample6-and-japanese 'autoreconf
514 (invoke "autoreconf" "-vif")
516 (add-after 'unpack 'fix-test-files
517 (lambda* (#:key inputs outputs #:allow-other-keys)
518 (let* ((out (assoc-ref outputs "out"))
519 (bin (string-append out "/bin")))
520 (substitute* (find-files "src/tests/" "\\.shin$")
521 (("PATH=/bin:/usr/bin")
522 "#PATH=/bin:/usr/bin")
523 (("diff") (string-append (which "diff")))
524 (("tr") (string-append (which "tr"))))
527 `(("ncurses" ,ncurses)
529 ("readline" ,readline)
532 `(("libtool" ,libtool)
533 ("autoconf" ,autoconf)
534 ("automake" ,automake)
535 ("pkg-config" ,pkg-config)))
536 (home-page "https://www.lbreyer.com/dbacl.html")
537 (synopsis "Bayesian text and email classifier")
539 "dbacl is a fast Bayesian text and email classifier. It builds a variety
540 of language models using maximum entropy (minimum divergence) principles, and
541 these can then be used to categorize input data automatically among multiple
543 ;; The software is licensed as GPLv3 or later, but
544 ;; includes various sample texts in the doc dir:
545 ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
547 ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
548 ;; - sample6.txt is a forwarded email, copyright unknown.
549 ;; Guix does exclude sample6.txt.
550 ;; - japanese.txt is a Japanese unoffical translation of the
551 ;; GNU General Public License, (c) by the Free Software Foundation.
552 ;; Guix excludes this file.
553 (license (list license:gpl3+ license:public-domain))))
555 (define-public dotconf
562 "https://github.com/williamh/dotconf/archive/v"
564 (file-name (string-append name "-" version ".tar.gz"))
567 "0lsnh0yaw44psmx59hq94cj1932gscp5h8d3cnh05l0svr0cy7kz"))))
568 (build-system gnu-build-system)
570 `(#:tests? #f ; FIXME maketest.sh does not work.
572 (modify-phases %standard-phases
573 (add-after 'unpack 'autoreconf
575 (zero? (system* "autoreconf" "-vif")))))))
577 `(("autoconf" ,autoconf)
578 ("automake" ,automake)
579 ("libtool" ,libtool)))
580 (home-page "https://github.com/williamh/dotconf")
581 (synopsis "Configuration file parser library")
583 "C library for creating and parsing configuration files.")
584 (license (list license:lgpl2.1 ; Main distribution.
585 license:asl1.1)))) ; src/readdir.{c,h}
587 (define-public java-rsyntaxtextarea
589 (name "java-rsyntaxtextarea")
593 (uri (string-append "https://github.com/bobbylight/"
594 "RSyntaxTextArea/archive/"
596 (file-name (string-append name "-" version ".tar.gz"))
599 "0c5mqg2klj5rvf8fhycrli8rf6s37l9p7a8knw9gpp65r1c120q2"))))
600 (build-system ant-build-system)
602 `(;; FIXME: some tests fail because locale resources cannot be found.
603 ;; Even when I add them to the class path,
604 ;; RSyntaxTextAreaEditorKitDumbCompleteWordActionTest fails.
606 #:jar-name "rsyntaxtextarea.jar"))
608 `(("java-junit" ,java-junit)
609 ("java-hamcrest-core" ,java-hamcrest-core)))
610 (home-page "https://bobbylight.github.io/RSyntaxTextArea/")
611 (synopsis "Syntax highlighting text component for Java Swing")
612 (description "RSyntaxTextArea is a syntax highlighting, code folding text
613 component for Java Swing. It extends @code{JTextComponent} so it integrates
614 completely with the standard @code{javax.swing.text} package. It is fast and
615 efficient, and can be used in any application that needs to edit or view
617 (license license:bsd-3)))
619 ;; We use the sources from git instead of the tarball from pypi, because the
620 ;; latter does not include the Cython source file from which bycython.cpp is
622 (define-public python-editdistance
623 (let ((commit "3ea84a7dd3258c76aa3be851ef3d50e59c886846")
626 (name "python-editdistance")
627 (version (string-append "0.3.1-" revision "." (string-take commit 7)))
632 (url "https://github.com/aflc/editdistance.git")
634 (file-name (git-file-name name version))
637 "1l43svsv12crvzphrgi6x435z6xg8m086c64armp8wzb4l8ccm7g"))))
638 (build-system python-build-system)
641 (modify-phases %standard-phases
642 (add-after 'unpack 'build-cython-code
644 (with-directory-excursion "editdistance"
645 (delete-file "bycython.cpp")
646 (invoke "cython" "--cplus" "bycython.pyx")))))))
648 `(("python-cython" ,python-cython)))
649 (home-page "https://www.github.com/aflc/editdistance")
650 (synopsis "Fast implementation of the edit distance (Levenshtein distance)")
652 "This library simply implements Levenshtein distance algorithm with C++
654 (license license:expat))))
656 (define-public go-github.com-mattn-go-runewidth
658 (name "go-github.com-mattn-go-runewidth")
663 (url "https://github.com/mattn/go-runewidth")
664 (commit (string-append "v" version))))
665 (file-name (git-file-name name version))
668 "0vkrfrz3fzn5n6ix4k8s0cg0b448459sldq8bp4riavsxm932jzb"))))
669 (build-system go-build-system)
671 '(#:import-path "github.com/mattn/go-runewidth"))
672 (synopsis "@code{runewidth} provides Go functions to work with string widths")
674 "The @code{runewidth} library provides Go functions for padding,
675 measuring and checking the width of strings, with support east asian text.")
676 (home-page "https://github.com/jessevdk/go-flags")
677 (license license:expat)))