gnu/packages/textutils.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2015 Taylan Ulrich Bayırlı/Kammer <taylanbayirli@gmail.com>
   3 ;;; Copyright © 2015, 2016 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 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 ;;;
  13 ;;; This file is part of GNU Guix.
  14 ;;;
  15 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  16 ;;; under the terms of the GNU General Public License as published by
  17 ;;; the Free Software Foundation; either version 3 of the License, or (at
  18 ;;; your option) any later version.
  19 ;;;
  20 ;;; GNU Guix is distributed in the hope that it will be useful, but
  21 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  22 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  23 ;;; GNU General Public License for more details.
  24 ;;;
  25 ;;; You should have received a copy of the GNU General Public License
  26 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  27
  28 (define-module (gnu packages textutils)
  29   #:use-module ((guix licenses) #:prefix license:)
  30   #:use-module (guix packages)
  31   #:use-module (guix download)
  32   #:use-module (guix git-download)
  33   #:use-module (guix build-system gnu)
  34   #:use-module (guix build-system cmake)
  35   #:use-module (guix build-system trivial)
  36   #:use-module (gnu packages)
  37   #:use-module (gnu packages autotools)
  38   #:use-module (gnu packages ncurses)
  39   #:use-module (gnu packages perl)
  40   #:use-module (gnu packages pkg-config)
  41   #:use-module (gnu packages python)
  42   #:use-module (gnu packages readline)
  43   #:use-module (gnu packages slang)
  44   #:use-module (gnu packages zip))
  45
  46 (define-public recode
  47   (package
  48     (name "recode")
  49     ;; Last beta release (3.7-beta2) is from 2008; last commit from Feb 2014.
  50     ;; So we use that commit instead.
  51     (version "3.7.0.201402")
  52     (source
  53      (origin
  54        (method git-fetch)
  55        (uri (git-reference
  56              (url "https://github.com/pinard/Recode.git")
  57              (commit "2d7092a9999194fc0e9449717a8048c8d8e26c18")))
  58        (sha256
  59         (base32 "1wssv8z6g3ryrw33sksz4rjhlnhgvvdqszw1ggl4rcwks34n86zm"))
  60        (file-name (string-append name "-" version "-checkout"))))
  61     (build-system gnu-build-system)
  62     (native-inputs `(("python" ,python-2)))
  63     (arguments
  64      '(#:phases
  65        (alist-cons-before
  66         'check 'pre-check
  67         (lambda _
  68           (substitute* "tests/setup.py"
  69             (("([[:space:]]*)include_dirs=.*" all space)
  70              (string-append all space "library_dirs=['../src/.libs'],\n")))
  71           ;; The test extension 'Recode.so' lacks RUNPATH for 'librecode.so'.
  72           (setenv "LD_LIBRARY_PATH" (string-append (getcwd) "/src/.libs")))
  73         %standard-phases)))
  74     (home-page "https://github.com/pinard/Recode")
  75     (synopsis "Text encoding converter")
  76     (description "The Recode library converts files between character sets and
  77 usages.  It recognises or produces over 200 different character sets (or about
  78 300 if combined with an iconv library) and transliterates files between almost
  79 any pair.  When exact transliteration are not possible, it gets rid of
  80 offending characters or falls back on approximations.  The recode program is a
  81 handy front-end to the library.")
  82     (license license:gpl2+)))
  83
  84 (define-public enca
  85   (package
  86     (name "enca")
  87     (version "1.16")
  88     (source
  89      (origin
  90        (method url-fetch)
  91        (uri (string-append
  92              "https://github.com/nijel/enca/archive/" version ".tar.gz"))
  93        (sha256
  94         (base32 "1xik00x0yvhswsw2isnclabhv536xk1s42cf5z54gfbpbhc7ni8l"))
  95        (file-name (string-append name "-" version ".tar.gz"))))
  96     (build-system gnu-build-system)
  97     (inputs `(("recode" ,recode)))
  98
  99     ;; Both 'test-convert-64.sh' and 'test-convert-filter.sh' manipulate a
 100     ;; 'test.tmp' file, so they have to run in sequence.
 101     (arguments '(#:parallel-tests? #f))
 102
 103     (home-page "https://github.com/nijel/enca")
 104     (synopsis "Text encoding detection tool")
 105     (description "Enca (Extremely Naive Charset Analyser) consists of libenca,
 106 an encoding detection library, and enca, a command line frontend, integrating
 107 libenca and several charset conversion libraries and tools.")
 108     (license license:gpl2)))
 109
 110 (define-public utf8proc
 111   (package
 112     (name "utf8proc")
 113     (version "2.1.0")
 114     (source
 115      (origin
 116        (method url-fetch)
 117        (uri (string-append
 118              "https://github.com/JuliaLang/utf8proc/archive/v"
 119              version ".tar.gz"))
 120        (file-name (string-append name "-" version ".tar.gz"))
 121        (sha256
 122         (base32 "0q1jhdkk4f9b0zb8s2ql3sba3br5nvjsmbsaybmgj064k9hwbk15"))))
 123     (build-system gnu-build-system)
 124     (inputs                  ;test data that is otherwise downloaded with curl
 125      `(("NormalizationTest.txt"
 126         ,(origin
 127            (method url-fetch)
 128            (uri (string-append "http://www.unicode.org/Public/9.0.0/ucd/"
 129                                "NormalizationTest.txt"))
 130            (sha256
 131             (base32 "1fxrz0bilsbwl685336aqi88k62i6nqhm62rvy4zhg3bcm4dhj1d"))))
 132        ("GraphemeBreakTest.txt"
 133         ,(origin
 134            (method url-fetch)
 135            (uri (string-append "http://www.unicode.org/Public/9.0.0/ucd/"
 136                                "auxiliary/GraphemeBreakTest.txt"))
 137            (sha256
 138             (base32 "0qbhyhmf0778lc2hcwlpizrvmdxwpk959v2q2wb8abv09ba7wvn7"))))))
 139     (arguments
 140      '(#:make-flags (list "CC=gcc"
 141                           (string-append "prefix=" (assoc-ref %outputs "out")))
 142        #:phases
 143        (modify-phases %standard-phases
 144          (delete 'configure)
 145          (add-before 'check 'check-data
 146            (lambda* (#:key inputs #:allow-other-keys)
 147              (for-each (lambda (i)
 148                          (copy-file (assoc-ref inputs i)
 149                                     (string-append "data/" i)))
 150                        '("NormalizationTest.txt" "GraphemeBreakTest.txt"))
 151              (substitute* "data/GraphemeBreakTest.txt"
 152                (("÷") "/")
 153                (("×") "+")))))))
 154     (home-page "http://julialang.org/utf8proc/")
 155     (synopsis "C library for processing UTF-8 Unicode data")
 156     (description "utf8proc is a small C library that provides Unicode
 157 normalization, case-folding, and other operations for data in the UTF-8
 158 encoding, supporting Unicode version 9.0.0.")
 159     (license license:expat)))
 160
 161 (define-public libgtextutils
 162   (package
 163     (name "libgtextutils")
 164     (version "0.7")
 165     (source
 166      (origin
 167        (method url-fetch)
 168        (uri (string-append
 169              "https://github.com/agordon/libgtextutils/releases/download/"
 170              version "/libgtextutils-" version ".tar.gz"))
 171        (sha256
 172         (base32 "0jiybkb2z58wa2msvllnphr4js2hvjvh988pavb3mzkgr6ihwbkr"))))
 173     (build-system gnu-build-system)
 174     (arguments
 175      '(#:phases
 176        (alist-cons-after
 177         'unpack 'autoreconf
 178         (lambda _ (zero? (system* "autoreconf" "-vif")))
 179         %standard-phases)))
 180     (native-inputs
 181      `(("autoconf" ,autoconf)
 182        ("automake" ,automake)
 183        ("libtool" ,libtool)))
 184     (home-page "https://github.com/agordon/libgtextutils")
 185     (synopsis "Gordon's text utils library")
 186     (description
 187      "libgtextutils is a text utilities library used by the fastx toolkit from
 188 the Hannon Lab.")
 189     (license license:agpl3+)))
 190
 191 (define-public cityhash
 192   (let ((commit "8af9b8c"))
 193     (package
 194       (name "cityhash")
 195       (version (string-append "1.1-2." commit))
 196       (source (origin
 197                 (method git-fetch)
 198                 (uri (git-reference
 199                       (url "https://github.com/google/cityhash.git")
 200                       (commit commit)))
 201                 (file-name (string-append name "-" version ".tar.gz"))
 202                 (sha256
 203                  (base32
 204                   "0n6skf5dv8yfl1ckax8dqhvsbslkwc9158zf2ims0xqdvzsahbi6"))))
 205       (build-system gnu-build-system)
 206       (arguments
 207        '(#:make-flags (list "CXXFLAGS=-g -O3")
 208          #:phases
 209          (modify-phases %standard-phases
 210            ;; citycrc is not installed by default but is used by some
 211            ;; programs.
 212            (add-after 'install 'install-citycrc
 213              (lambda* (#:key outputs #:allow-other-keys)
 214                (let* ((out (assoc-ref outputs "out"))
 215                       (include (string-append out "/include")))
 216                  (install-file "src/citycrc.h" include))
 217                #t)))))
 218       (home-page "https://github.com/google/cityhash")
 219       (synopsis "C++ hash functions for strings")
 220       (description
 221        "CityHash provides hash functions for strings.  The functions mix the
 222 input bits thoroughly but are not suitable for cryptography.")
 223       (license license:expat))))
 224
 225 (define-public ustr
 226   (package
 227     (name "ustr")
 228     (version "1.0.4")
 229     (source (origin
 230               (method url-fetch)
 231               (uri (string-append "http://www.and.org/ustr/" version
 232                                   "/ustr-" version ".tar.bz2"))
 233               (sha256
 234                (base32
 235                 "1i623ygdj7rkizj7985q9d6vj5amwg686aqb5j3ixpkqkyp6xbrx"))))
 236     (build-system gnu-build-system)
 237     (arguments
 238      `(#:make-flags
 239        (list "CC=gcc"
 240              "HIDE="
 241              ;; Override "/sbin/ldconfig" with "echo" because we don't need
 242              ;; "ldconfig".
 243              "LDCONFIG=echo"
 244              (string-append "prefix=" (assoc-ref %outputs "out"))
 245              "all-shared")
 246        #:phases
 247        (modify-phases %standard-phases
 248          (add-after 'unpack 'disable-check-for-stdint
 249            (lambda _
 250              ;; Of course we have stdint.h, just not in /usr/include
 251              (substitute* '("Makefile"
 252                             "ustr-import.in")
 253                (("-f \"/usr/include/stdint.h\"") "-z \"\""))
 254              #t))
 255          ;; No configure script
 256          (delete 'configure))))
 257     (home-page "http://www.and.org/ustr/")
 258     (synopsis "String library with very low memory overhead")
 259     (description
 260      "Ustr is a string library for C with very low memory overhead.")
 261     ;; Quoted from the home page: "The License for the code is MIT, new-BSD,
 262     ;; LGPL, etc. ... if you need another license to help compatibility, just
 263     ;; ask for it.  It's basically public domain, without all the legal
 264     ;; problems for everyone that trying to make something public domain
 265     ;; entails."
 266     (license license:public-domain)))
 267
 268 (define-public libconfig
 269   (package
 270     (name "libconfig")
 271     (version "1.5")
 272     (source (origin
 273               (method url-fetch)
 274               (uri (string-append "http://www.hyperrealm.com/libconfig/"
 275                                   "libconfig-" version ".tar.gz"))
 276               (sha256
 277                (base32
 278                 "1xh3hzk63v4y8815lc5209m3s6ms2cpgw4h5hg462i4f1lwsl7g3"))))
 279     (build-system gnu-build-system)
 280     (home-page "http://www.hyperrealm.com/libconfig/")
 281     (synopsis "C/C++ configuration file library")
 282     (description
 283      "Libconfig is a simple library for manipulating structured configuration
 284 files.  This file format is more compact and more readable than XML.  And
 285 unlike XML, it is type-aware, so it is not necessary to do string parsing in
 286 application code.")
 287     (license license:lgpl2.1+)))
 288
 289 (define-public pfff
 290   (package
 291     (name "pfff")
 292     (version "1.0")
 293     (source (origin
 294               (method url-fetch)
 295               (uri (string-append "https://github.com/pfff/pfff/archive/v"
 296                                   version ".tar.gz"))
 297               (file-name (string-append name "-" version ".tar.gz"))
 298               (sha256
 299                (base32
 300                 "00m553aa277iarxj6dalmklyb64r7ias49bfwzbacsfg8h3kar8m"))))
 301     (build-system cmake-build-system)
 302     (home-page "http://biit.cs.ut.ee/pfff/")
 303     (synopsis "Probabilistic fast file fingerprinting tool")
 304     (description
 305      "pfff is a tool for calculating a compact digital fingerprint of a file
 306 by sampling randomly from the file instead of reading it in full.
 307 Consequently, the computation has a flat performance characteristic,
 308 correlated with data variation rather than file size.  pfff can be as reliable
 309 as existing hashing techniques, with provably negligible risk of collisions.")
 310     (license license:bsd-3)))
 311
 312 (define-public oniguruma
 313   (package
 314     (name "oniguruma")
 315     (version "5.9.6")
 316     (source (origin
 317               (method url-fetch)
 318               (uri (string-append "https://github.com/kkos/"
 319                                   "oniguruma/releases/download/v" version
 320                                   "/onig-" version ".tar.gz"))
 321               (sha256
 322                (base32
 323                 "19s79vsclqn170mw0ajwv7j37qsbn4f1yjz3yavnhvva6c820r6m"))))
 324     (build-system gnu-build-system)
 325     (home-page "https://github.com/kkos/oniguruma")
 326     (synopsis "Regular expression library")
 327     (description "Oniguruma is a regular expressions library.  The special
 328 characteristic of this library is that different character encoding for every
 329 regular expression object can be specified.")
 330     (license license:bsd-2)))
 331
 332 (define-public antiword
 333   (package
 334     (name "antiword")
 335     (version "0.37")
 336     (source (origin
 337               (method url-fetch)
 338               (uri (string-append "http://www.winfield.demon.nl/linux"
 339                                   "/antiword-" version ".tar.gz"))
 340               (sha256
 341                (base32
 342                 "1b7mi1l20jhj09kyh0bq14qzz8vdhhyf35gzwsq43mn6rc7h0b4f"))
 343               (patches (search-patches "antiword-CVE-2014-8123.patch"))))
 344     (build-system gnu-build-system)
 345     (arguments
 346      `(#:tests? #f ; There are no tests
 347        #:make-flags
 348        (list "-f" "Makefile.Linux"
 349              (string-append "GLOBAL_INSTALL_DIR="
 350                             (assoc-ref %outputs "out") "/bin")
 351              (string-append "GLOBAL_RESOURCES_DIR="
 352                             (assoc-ref %outputs "out") "/share/antiword"))
 353        #:phases
 354        (modify-phases %standard-phases
 355          (delete 'configure)
 356          (replace 'install
 357            (lambda* (#:key make-flags #:allow-other-keys)
 358              (zero? (apply system* "make" `("global_install" ,@make-flags))))))))
 359     (home-page "http://www.winfield.demon.nl/")
 360     (synopsis "Microsoft Word document reader")
 361     (description "Antiword is an application for displaying Microsoft Word
 362 documents.  It can also convert the document to PostScript or XML.  Only
 363 documents made by MS Word version 2 and version 6 or later are supported.  The
 364 name comes from: \"The antidote against people who send Microsoft Word files
 365 to everybody, because they believe that everybody runs Windows and therefore
 366 runs Word\".")
 367     (license license:gpl2+)))
 368
 369 (define-public utfcpp
 370   (package
 371     (name "utfcpp")
 372     (version "2.3.4")
 373     (source (origin
 374               (method url-fetch)
 375               (uri
 376                (string-append
 377                 "mirror://sourceforge/utfcpp/utf8cpp_2x/Release%20"
 378                 version "/utf8_v"
 379                 (string-map (lambda (x) (if (eq? x #\.) #\_ x)) version)
 380                 ".zip"))
 381               (file-name (string-append name "-" version ".zip"))
 382               (sha256
 383                (base32
 384                 "1vqhs0aipcvvdrwcs7h3jsryg6mgbmc4s34n5cm6d36q4nxwwwrk"))))
 385     (build-system trivial-build-system)
 386     (arguments
 387      `(#:modules ((guix build utils))
 388        #:builder
 389        (begin
 390          (use-modules (guix build utils))
 391          (let ((source (assoc-ref %build-inputs "source"))
 392                (out    (assoc-ref %outputs "out"))
 393                (unzip  (string-append (assoc-ref %build-inputs "unzip")
 394                                       "/bin/unzip")))
 395            (mkdir-p out)
 396            (with-directory-excursion out
 397              (system* unzip source)
 398              (mkdir-p "share/doc")
 399              (rename-file "doc" "share/doc/utfcpp")
 400              (rename-file "source" "include"))))))
 401     (native-inputs `(("unzip" ,unzip)))
 402     (home-page "https://github.com/nemtrif/utfcpp")
 403     (synopsis "Portable C++ library for handling UTF-8")
 404     (description "UTF8-CPP is a C++ library for handling UTF-8 encoded text
 405 in a portable way.")
 406     (license license:boost1.0)))
 407
 408 (define-public dbacl
 409   (package
 410     (name "dbacl")
 411     (version "1.14")
 412     (source
 413      (origin
 414        (method url-fetch)
 415        (uri (string-append "http://www.lbreyer.com/gpl/"
 416                            name "-" version ".tar.gz"))
 417        (sha256
 418         (base32
 419          "0224g6x71hyvy7jikfxmgcwww1r5lvk0jx36cva319cb9nmrbrq7"))))
 420     (build-system gnu-build-system)
 421     (arguments
 422      `(#:make-flags
 423        (list
 424         (string-append "-I" (assoc-ref %build-inputs "slang")
 425                        "/include/slang")
 426         (string-append "-I" (assoc-ref %build-inputs "ncurses")
 427                        "/include/ncurses"))
 428        #:phases
 429        (modify-phases %standard-phases
 430          (add-after 'unpack 'delete-sample6-and-japanese
 431            (lambda _
 432              (substitute* "doc/Makefile.am"
 433                (("sample6.txt") "")
 434                (("japanese.txt") ""))
 435              (delete-file "doc/sample6.txt")
 436              (delete-file "doc/japanese.txt")
 437              (substitute* (list "src/tests/Makefile.am"
 438                                 "src/tests/Makefile.in")
 439                (("dbacl-jap.shin") "")
 440                (("dbacl-jap.sh") ""))
 441              #t))
 442          (add-after 'unpack 'delete-test
 443            ;; See comments about the license.
 444            (lambda _
 445              (delete-file "src/tests/dbacl-jap.shin")))
 446          (add-after 'delete-sample6-and-japanese 'autoreconf
 447            (lambda _
 448              (zero? (system* "autoreconf" "-vif"))))
 449          (add-after 'unpack 'fix-test-files
 450            (lambda* (#:key inputs outputs #:allow-other-keys)
 451              (let* ((out (assoc-ref outputs "out"))
 452                     (bin (string-append out "/bin")))
 453                (substitute* (find-files "src/tests/" "\\.shin$")
 454                  (("PATH=/bin:/usr/bin")
 455                   "#PATH=/bin:/usr/bin")
 456                  (("diff") (string-append (which "diff")))
 457                  (("tr") (string-append (which "tr"))))
 458                #t))))))
 459     (inputs
 460      `(("ncurses" ,ncurses)
 461        ("perl" ,perl)
 462        ("readline" ,readline)
 463        ("slang" ,slang)))
 464     (native-inputs
 465      `(("libtool" ,libtool)
 466        ("autoconf" ,autoconf)
 467        ("automake" ,automake)
 468        ("pkg-config" ,pkg-config)))
 469     (home-page "http://www.lbreyer.com/dbacl.html")
 470     (synopsis "Bayesian text and email classifier")
 471     (description
 472      "dbacl is a fast Bayesian text and email classifier.  It builds a variety
 473 of language models using maximum entropy (minimum divergence) principles, and
 474 these can then be used to categorize input data automatically among multiple
 475 categories.")
 476     ;; The software is licensed as GPLv3 or later, but
 477     ;; includes various sample texts in the doc dir:
 478     ;; - sample1.txt, sample3 and sampe5.txt are in the public domain,
 479     ;;   by Mark Twain.
 480     ;; - sample2.txt, sample4.txt are in the public domain, by Aristotle.
 481     ;; - sample6.txt is a forwarded email, copyright unknown.
 482     ;;   Guix does exclude sample6.txt.
 483     ;; - japanese.txt is a Japanese unoffical translation of the
 484     ;;   GNU General Public License, (c) by the Free Software Foundation.
 485     ;;   Guix excludes this file.
 486     (license (list license:gpl3+ license:public-domain))))
 487
 488 (define-public dotconf
 489   (package
 490     (name "dotconf")
 491     (version "1.3")
 492     (source (origin
 493               (method url-fetch)
 494               (uri (string-append
 495                     "https://github.com/williamh/dotconf/archive/v"
 496                     version ".tar.gz"))
 497               (file-name (string-append name "-" version ".tar.gz"))
 498               (sha256
 499                (base32
 500                 "0lsnh0yaw44psmx59hq94cj1932gscp5h8d3cnh05l0svr0cy7kz"))))
 501     (build-system gnu-build-system)
 502     (arguments
 503      `(#:tests? #f ; FIXME maketest.sh does not work.
 504        #:phases
 505        (modify-phases %standard-phases
 506          (add-before 'configure 'autoreconf
 507            (lambda _
 508              (zero? (system* "autoreconf" "-vif")))))))
 509     (native-inputs
 510      `(("autoconf" ,autoconf)
 511        ("automake" ,automake)
 512        ("libtool" ,libtool)))
 513     (home-page "https://github.com/williamh/dotconf")
 514     (synopsis "Configuration file parser library")
 515     (description
 516      "C library for creating and parsing configuration files.")
 517     (license (list license:lgpl2.1         ; Main distribution.
 518                    license:asl1.1))))      ; src/readdir.{c,h}