| 1 |
commit: bc7f4eb883bfc42ec54af84e582aa186f49bac0e |
| 2 |
Author: Michał Górny <mgorny <AT> gentoo <DOT> org> |
| 3 |
AuthorDate: Tue Dec 21 07:19:06 2021 +0000 |
| 4 |
Commit: Michał Górny <mgorny <AT> gentoo <DOT> org> |
| 5 |
CommitDate: Tue Dec 21 09:42:47 2021 +0000 |
| 6 |
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=bc7f4eb8 |
| 7 |
|
| 8 |
dev-python/nltk-data: Add new data (20211221) |
| 9 |
|
| 10 |
This one goes straight to stable since it's mirror-restricted |
| 11 |
and upstream replaced the files in place again! |
| 12 |
|
| 13 |
Signed-off-by: Michał Górny <mgorny <AT> gentoo.org> |
| 14 |
|
| 15 |
dev-python/nltk-data/Manifest | 5 + |
| 16 |
dev-python/nltk-data/nltk-data-20211221.ebuild | 210 +++++++++++++++++++++++++ |
| 17 |
2 files changed, 215 insertions(+) |
| 18 |
|
| 19 |
diff --git a/dev-python/nltk-data/Manifest b/dev-python/nltk-data/Manifest |
| 20 |
index 0a6f8624fbaa..2a75eb265207 100644 |
| 21 |
--- a/dev-python/nltk-data/Manifest |
| 22 |
+++ b/dev-python/nltk-data/Manifest |
| 23 |
@@ -30,6 +30,7 @@ DIST nltk-genesis-20200312.zip 473239 BLAKE2B 5234c2bcc31ae738bbd6cdfe1ee231a75c |
| 24 |
DIST nltk-gutenberg-20200312.zip 4251829 BLAKE2B 54448e919010f12faa92c0ffd4adb31a7d1d76c262d5ac99c10129d188830107621f90e4ba611ffc1541c25037fc916b451169a8ae1df276747890fea2b5e06f SHA512 7b4442d18b89f9bb8261f23ad2f4d513b470461255633cbebe2b0c310f003effe819f7a86b39a62a6cde1b3fdbf099dabbdaaed520268c99fd1bb933baa97ead |
| 25 |
DIST nltk-ieer-20200312.zip 166156 BLAKE2B 4c63be6f0f16e59584b7f97c21da0cf60a1efd3678ce662d683d8a6a9d05d333690b26ebb037d3047b26e3b51163ffa666f048ad8a89ee6c851ddd3edd2faf58 SHA512 b55545f0ac03282fb85c4a86bb69fd9988db8cc504114bff7b763007e20f559efa61b8ae5b7955c251c1c2dd9d23246e23515b1cd5d8d28225c20ec236c05cd8 |
| 26 |
DIST nltk-inaugural-20200312.zip 329806 BLAKE2B 222e22742d93156b852ddd72ffccdd5be5c907f43973b0ceee8d8d383bc5387575d1f974b29fd6d59499ec86483f56270251e6a5e48d7840c577207849baed59 SHA512 096591769373019883b9a2f85d32bc86021aefb74deb26dc9005233ae7f317d587c247edc53f1a956b8e29911656419d9677c271fb59ded4ca4edf2a80ff86b7 |
| 27 |
+DIST nltk-inaugural-20211221.zip 346476 BLAKE2B 6f14827fcd8edf9300725c6ffcfa899f5878537aaefc40b165581af60866ca7e7b0ed316a2b8dd87ff858667ce7a5f4035090a1cc3a404afa64b28d34dcf8747 SHA512 5440bff001209e42ee86c2501d2cd628cf639734c194c11097a862f6759f09a6adef1da0702b77211774f84d5d9d4ee55e89085b39d8a911811def1bd07ac8e2 |
| 28 |
DIST nltk-indian-20200312.zip 199187 BLAKE2B 9f70f1f28bcb96d828ca2ed96d6f89cd8eca8391b64842c742e940aaf59f31e142a4919ab5a21bce954bbe2ed85add162ac67c9549758f9aa44b3e90d45d44a7 SHA512 08e9342567bcc607e9007aea8415af2cad1f8a289104b47a61bf569f1391629ebbe25f19aacf53e2b14725edfdf18076331f044e1c2e42d07c96d5c875ff8f9e |
| 29 |
DIST nltk-jeita-20200312.zip 16531215 BLAKE2B 7cf98b389d5c4b41d9d613c14c5116b83c520cfe7418ef982a6acc1294d8ecdb8a5d6d396409c9f13ca83df999793619d5eb5483612e87874766afe565c68362 SHA512 491e6dc27d70ebfd139be4dcd179b02c383f38edbe4290aed85638ecd7c987259b025e4f6f3f1d5bfe2b7aa3eff7188109205417d0acb64759988150f2753129 |
| 30 |
DIST nltk-kimmo-20200312.zip 186958 BLAKE2B ca288f48c09a33b86449c780e5c8612c3521e22bfe70918629bf3144b75b5566a942c3cc1440a13c2d81e3a69c77928c336850cc9fac4c232deb77293bfbe4e6 SHA512 fe38d5298ffb6dd62223c2002278d9ddf2695f341a29477f44434644bc6766cfe2c73091067f9eb8aea201735eef5012f96ba267f2e6ea3aec41ac206a43cac2 |
| 31 |
@@ -49,6 +50,7 @@ DIST nltk-names-20200312.zip 21326 BLAKE2B 37389f44337724fac26eaad5a4c310ea244b4 |
| 32 |
DIST nltk-nombank.1.0-20200312.zip 6728397 BLAKE2B 4dfb61fb651817086ad20957d02a39c99dff83ec45dc1a01e1dac055e0d7d9ef8108ef8df5218ce57b2b0866b748f5f76663a58e08f5e1731272173888e714e9 SHA512 81bcf29898ba1205ccbc8486d14518e95b0189b97b172a54580d4cb4584f4ef32a5284fe0abed0b954e67a318de8f10de5df1a20e6c08976df09cd2c9c66cb32 |
| 33 |
DIST nltk-nonbreaking_prefixes-20200312.zip 25437 BLAKE2B 9f26efbcc429f2d36a5584014604d2ca9d4de3864aa6192a114e9e2cd55ff1df50d4321a8c40dbe96225fad6be51f7be66f8a4d58146e370a1d33a9305f7f95f SHA512 4cddfdacf1c7fe4e93a31333413a83ebea2ea7907f4e6fe99e1c6ff4aed8efc951f4399478cdb88e5d5dc01cfb734b6672f4007b2a17c4784896fecaf4a93cdf |
| 34 |
DIST nltk-nps_chat-20200312.zip 301366 BLAKE2B eadf0d0ee54c982765a124321fe6de161bfbfbfb0385d5d7e171666945ef313b960ba7b820e4fed02885ffc213173e2a7c97822513ae6e81e68858af21d9f424 SHA512 a716f3aaba3aedbbdaeaff1c6f1ef607476aa989eaef1c2971b827215f3167952477b203f1543d0fd27e5594c3efa01349c6a0efc1ac423048808f18eae94643 |
| 35 |
+DIST nltk-omw-1.4-20211221.zip 26634772 BLAKE2B 4d5fe648987b9aaa589d612f439560287200fe947a80182b941b0660a97a739ddda00a472209a26715aa72c44f37cdb5dbce9df89cb7bfbcdb7f4fa2908bb96e SHA512 d25c591c894fedeaf8649ef26653c51b74282e8af43bd0a8867eb16f79a89c4b567328a7c83f7651ec9e41dc746cba87b0c8bdbbe736db7f18c62a56820a92e6 |
| 36 |
DIST nltk-omw-20200312.zip 12110409 BLAKE2B 5ae634fdd89fcd2be2dea1511b2c434755e16b92804ea6bb5e2139113c644885d142e6290ab1d00c418f9df5241ebf6b7e227da120dc95d1e907d8f62a3d0cb8 SHA512 05b1b014e55738c49b34ea0371b95bd1d1aadfb0cc57325ec110ffbbb1250690e3a4aab45e877eb5433de08c7afe3ffa161e29976a0e28db9803fb637b24c843 |
| 37 |
DIST nltk-opinion_lexicon-20200312.zip 24947 BLAKE2B 47f5a381647075f4bea0230939098a6e83b47860165595e72f076a5436e2aad78148606d86ec331b1cc323963cb2a7500669c6fc2aff2e889bbaec8fd8571722 SHA512 75b98afa24c4e0284b2858770d7655c5c399066f56c1b9aa03dd51d80f211a694856695f8192570f67f02f0827ab02c3f3b65feb5a56c04bebfae3b58ca56ba1 |
| 38 |
DIST nltk-panlex_swadesh-20200312.zip 2861668 BLAKE2B e9ee3b2296c8738319a86dfb39d34fc5bcd346e725e0ad0df74571106ef05b71126aff70914e914cff436674c0d70c71ca36121331ab0327932505a2bdb26be1 SHA512 d096be1c6ea54814e448f61048f05df9107c0ca5f75eb3051a009efd4643c94188dde03505b0a4ae29d301268ffc5153eeee7012aae18310f9288fd4ccdb2287 |
| 39 |
@@ -77,6 +79,7 @@ DIST nltk-sentence_polarity-20200312.zip 490256 BLAKE2B 56a1b4668875d5c73b220710 |
| 40 |
DIST nltk-sentiwordnet-20200312.zip 4686546 BLAKE2B 98a45d9a5be971bd9e76ad5c154359c1800aa96ab1e8e301215e435cae6d659cc0fb38604435da4e9a391741a72ef7fcb842a1cf119348a7039f717843035f12 SHA512 cd0ff31abec3b85e3114546918c9152360e3da45fbbaf7b4c76a13a6d288f1d9ea7a4633638080cd0205cc293b4123445106e9b5daddb2ae379d6a08d1c992a5 |
| 41 |
DIST nltk-shakespeare-20200312.zip 475458 BLAKE2B c60acbf2d90f5fffdae43bdf85191dbd694ec5ae19e37b145ad04af6d8481988f0c2ce644baff2e31707e86d95a9b3b03c1da4faaa46be92da2291cce565955e SHA512 72cbbdb237b78df1f6bf27b5504606f805433746b9862dd1ce544ab712783a41fb165f3a4f282a7f1bd58c7dfe6036c14ecb07a82c109f6d326eb95e470002fb |
| 42 |
DIST nltk-sinica_treebank-20200312.zip 899237 BLAKE2B e074dbdc7f1fe5958109f8ea019bead946bd2e3cd1c0849cf3ad49eaa34a1989baee7935a70e9103c121e5d375954d1dc53e4a1599e394ce6c2fda4e98769645 SHA512 724fb0905cbd0f903c3a75d70aee8a3e6f29e62971c1501efd8bfb3f4c6619e5fd8a285648e7792ea8e86bbda0a2afc6eecae45ad9f361983bf83b756ab52960 |
| 43 |
+DIST nltk-sinica_treebank-20211221.zip 906706 BLAKE2B 0c00bc7d39765b86eb99f55e82522d390cf0270fd2e2688d5825481b6f11e54aaa71b3f9ac95af170be543353746777da130bb6b294b0529330c712f9edaaab8 SHA512 22d44e23e89691f775a0e54971fada692aecd7cfa5181d49e9a53d0a96dc3be1beaa1e1ec6840f9b2756c5c16c93f96b571f6b08c175701485afc0c32261cac8 |
| 44 |
DIST nltk-smultron-20200312.zip 166207 BLAKE2B d0c3e75dd108965e260d913e0c02137da8707a751a4f53a1f59ff6a26af9997e2f0193980b5087277b87b74d9aab3e3c8d066d7264311cec2a94f0f9d21caf44 SHA512 8872717994269caff4ccf3f52cef273b2ef007efeb8ab89392cb4db3b8e3ba4b8859ab2fd70762b8941bee4558aebba3021869772be17a64e1056968aa605e19 |
| 45 |
DIST nltk-snowball_data-20200312.zip 6785405 BLAKE2B 44c10439b142540ac7eece967efa1431fd8f45342f0a90875dacf29ad374fe4c7d30af11d42ba45e0f1ec1836d56b2ff684ee352c5e8536cfb5db5eb7632285b SHA512 6c8a9259d88f6f7f499867d83b731de99d7fa4e8827ecedf836f653fc1a810efa9f6c5c6e2720a9e6610bc00978956b6a119bd08b70e3e241c4e9faccddd81d8 |
| 46 |
DIST nltk-spanish_grammars-20200312.zip 4047 BLAKE2B d8a8dbb558850a6a60f1fe5ab0f617f3a0f3c64bc7d49980cf793d374c6679d1bd42afb7e61776737b5eec162f2520abf2ee3acc92ea9ee0f397c3089b3b5b28 SHA512 4513347156e9351c259c0e2448198d68354bbc95e0a54561c31a88f13f333ebcba3e294c820cb62036665f2904eb6a7137546cd580e361c0423c30a8aed950b2 |
| 47 |
@@ -102,7 +105,9 @@ DIST nltk-webtext-20200312.zip 646297 BLAKE2B ca072fc38c144b659c76c36c9161641c91 |
| 48 |
DIST nltk-wmt15_eval-20200312.zip 383096 BLAKE2B 119943db4240171077569b3302c678644c2c9547ba67bfd055751059e0a3ad3ab6a19e4eedf9108d313d46dcd36cf19e11d973981da8c70a01c4cb790a7bd739 SHA512 362395d4c77ebe92f4c19fa8c2000082dc7a2343acc19cccb596ca00db6c40c231b904d807f46e2691cc4c4a0c79d14873b2a1983a494f2ca1485d540d787ceb |
| 49 |
DIST nltk-word2vec_sample-20200312.zip 49396025 BLAKE2B 0512b9bb7121a528190079f578e82b4e8f8021bfc0062cfa5613d260f3eee17460aecfeaacb65d950e79d27a653c78633a88c3638ec16377e2dbc3006387ebaa SHA512 09c30a4ab8f9fb6a5b36974b5953260d01cb4f285827fb90a374d054ad775ac978602ab56c452f46f4f8601312e232fd739d2f54dafa44ab8b7b01831cf0d9d1 |
| 50 |
DIST nltk-wordnet-20200312.zip 10775600 BLAKE2B dac56a8fb1fa6882b1871c394ad2acb2d3be739c424570e27c89fb6983df5f896a8f359092ba82752ddfc0531d83563a219e85f80124202f29bda93181efe4dc SHA512 1923a8bcd56fa0b9a9de91f53070dce28c3a7efbab11d2ef55c87134b1bf30de0f40abab59c39eb15dce54aec9491d8a5a259de212ff4cb25cde0ad09317009a |
| 51 |
+DIST nltk-wordnet2021-20211221.zip 11332750 BLAKE2B f4296728b803ef5a48cf312e2dd1471e8a240a0486e0c45b13c4c9998b8577ac30f87d151ffb2957e319e2b7b82706eba82b7b21530bfb3600d7cff2f4aefa0b SHA512 f27e847e0fd3fda83848af8bf94793d0e89c5f89b7868e222ba60be59122587f524e124e9d84f0b904b3a48808aae1c7d849f5851e2f3119ec99b3c30baac400 |
| 52 |
DIST nltk-wordnet31-20211023.zip 11055271 BLAKE2B e41a1951af5a71c9506d1e948b860574c94ab0ef31c1789a7e7bfb29c6dccea5b1d8895007631f7b595e9f90306365b5042e7a80dc6e1364fdbf4a5f0cba3b28 SHA512 a86091bd55e3a706892550b232be8f5199092623f1f8305d8c9be967a8527fe7d4ecb6250c369b229fdf52b6f3008106b758adc355fa2ad08b5b0cf2a458c173 |
| 53 |
+DIST nltk-wordnet31-20211221.zip 11058667 BLAKE2B fb2dca62f602e05226dd62802f42e5f74d11cc535e71b1dc634a202a8c52a7912b1eb0b71edb3c564520c7c05251d9a9928abe88398b816a43035a6db4bda1b3 SHA512 dffea47ded52a042594dced94bc877686d78e6dc9d5a9db5f2a1718190f411cc991b04028d0a4407a8354afd695258e2a5ae881d5f6d9eaf0c3d30de1a8cbf96 |
| 54 |
DIST nltk-wordnet_ic-20200312.zip 12056682 BLAKE2B c2dc2a646015b23699a72f636b588ec5718c70e6941d9d56863257e1e0396c8cf59ac1dc6ed74e5d7f0c2ee9129d63221a03967bf66a3d335e99160f295ed44d SHA512 1c94451a13af6c76bff60a0cab2e70402a3d9abd2e8fc62a5473f24ab4229feb0afe4faa8d389734697a6cf86d2c8b1dc700bb3afa3cbc279b75d7e0ec19fc6d |
| 55 |
DIST nltk-words-20200312.zip 757777 BLAKE2B eaaaaab6c26e206e9b6ce45daf779e3cc6706a06132afeabf013026d0009caee2d678f3c4ea9125b9654f7143bef29ec7a5706b79e5650ea556c6821b7754e6f SHA512 2810f05d3fc7ee6b6f8636fa1ff7b4e8c8cdac12b415cc54d15c69102290122ea138ec4fa36cb483f790c1ac10b0f83ae4c2c3e0e8df7e67e90e962ee5dbb0be |
| 56 |
DIST nltk-ycoe-20200312.zip 477 BLAKE2B 574835aa011a06a06363e26facd6a6f583a1dc1cac2de39adff59d8ab48eefac030b43d935a2f79af855259f2a9a571193dae2811589483af97406ff05c76c9e SHA512 e39ce165074d10ff63cb84ea52905d7ecb937797c8123ed113c5609afe1f63ac44d04d48a681002c4eac21dc9076ac74164b886c6f9ce42f3a102c38d1e8e756 |
| 57 |
|
| 58 |
diff --git a/dev-python/nltk-data/nltk-data-20211221.ebuild b/dev-python/nltk-data/nltk-data-20211221.ebuild |
| 59 |
new file mode 100644 |
| 60 |
index 000000000000..104917b4e5a0 |
| 61 |
--- /dev/null |
| 62 |
+++ b/dev-python/nltk-data/nltk-data-20211221.ebuild |
| 63 |
@@ -0,0 +1,210 @@ |
| 64 |
+# Copyright 2020-2021 Gentoo Authors |
| 65 |
+# Distributed under the terms of the GNU General Public License v2 |
| 66 |
+ |
| 67 |
+EAPI=8 |
| 68 |
+ |
| 69 |
+inherit check-reqs |
| 70 |
+ |
| 71 |
+DESCRIPTION="Data files for NLTK" |
| 72 |
+HOMEPAGE="https://www.nltk.org/nltk_data/" |
| 73 |
+ |
| 74 |
+# at least some of the files have poorly documented licenses |
| 75 |
+# TODO: create a USE flag for free-ish subset |
| 76 |
+LICENSE="all-rights-reserved" |
| 77 |
+SLOT="0" |
| 78 |
+KEYWORDS="amd64 x86" |
| 79 |
+IUSE="extra" |
| 80 |
+RESTRICT="bindist mirror" |
| 81 |
+ |
| 82 |
+BDEPEND="app-arch/unzip" |
| 83 |
+ |
| 84 |
+PACKAGES_ZIP_2020=( |
| 85 |
+ # wget -O - https://www.nltk.org/nltk_data/ | xml sel -t -m '//package[@unzip=0]' -v @subdir -o "/" -v @id -n - | sort |
| 86 |
+ corpora/comtrans |
| 87 |
+ corpora/conll2007 |
| 88 |
+ corpora/jeita |
| 89 |
+ corpora/knbc |
| 90 |
+ corpora/machado |
| 91 |
+ corpora/masc_tagged |
| 92 |
+ corpora/nombank.1.0 |
| 93 |
+ corpora/panlex_swadesh |
| 94 |
+ corpora/propbank |
| 95 |
+ corpora/reuters |
| 96 |
+ corpora/semcor |
| 97 |
+ corpora/universal_treebanks_v20 |
| 98 |
+ sentiment/vader_lexicon |
| 99 |
+ stemmers/snowball_data |
| 100 |
+) |
| 101 |
+ |
| 102 |
+PACKAGES_UNPACK_2020=( |
| 103 |
+ # wget -O - https://www.nltk.org/nltk_data/ | xml sel -t -m '//package[@unzip=1]' -v @subdir -o "/" -v @id -n - | sort |
| 104 |
+ corpora/abc |
| 105 |
+ corpora/alpino |
| 106 |
+ corpora/brown |
| 107 |
+ corpora/cess_cat |
| 108 |
+ corpora/cess_esp |
| 109 |
+ corpora/chat80 |
| 110 |
+ corpora/city_database |
| 111 |
+ corpora/cmudict |
| 112 |
+ corpora/comparative_sentences |
| 113 |
+ corpora/conll2000 |
| 114 |
+ corpora/conll2002 |
| 115 |
+ corpora/crubadan |
| 116 |
+ corpora/dependency_treebank |
| 117 |
+ corpora/dolch |
| 118 |
+ corpora/europarl_raw |
| 119 |
+ corpora/floresta |
| 120 |
+ corpora/framenet_v15 |
| 121 |
+ corpora/framenet_v17 |
| 122 |
+ corpora/gazetteers |
| 123 |
+ corpora/genesis |
| 124 |
+ corpora/gutenberg |
| 125 |
+ corpora/ieer |
| 126 |
+ corpora/indian |
| 127 |
+ corpora/lin_thesaurus |
| 128 |
+ corpora/mac_morpho |
| 129 |
+ corpora/movie_reviews |
| 130 |
+ corpora/mte_teip5 |
| 131 |
+ corpora/names |
| 132 |
+ corpora/nonbreaking_prefixes |
| 133 |
+ corpora/nps_chat |
| 134 |
+ corpora/omw |
| 135 |
+ corpora/opinion_lexicon |
| 136 |
+ corpora/pl196x |
| 137 |
+ corpora/ppattach |
| 138 |
+ corpora/product_reviews_1 |
| 139 |
+ corpora/product_reviews_2 |
| 140 |
+ corpora/pros_cons |
| 141 |
+ corpora/ptb |
| 142 |
+ corpora/qc |
| 143 |
+ corpora/rte |
| 144 |
+ corpora/senseval |
| 145 |
+ corpora/sentence_polarity |
| 146 |
+ corpora/sentiwordnet |
| 147 |
+ corpora/shakespeare |
| 148 |
+ corpora/state_union |
| 149 |
+ corpora/subjectivity |
| 150 |
+ corpora/swadesh |
| 151 |
+ corpora/switchboard |
| 152 |
+ corpora/timit |
| 153 |
+ corpora/toolbox |
| 154 |
+ corpora/treebank |
| 155 |
+ corpora/twitter_samples |
| 156 |
+ corpora/udhr |
| 157 |
+ corpora/udhr2 |
| 158 |
+ corpora/verbnet |
| 159 |
+ corpora/webtext |
| 160 |
+ corpora/wordnet |
| 161 |
+ corpora/wordnet_ic |
| 162 |
+ corpora/words |
| 163 |
+ grammars/book_grammars |
| 164 |
+ grammars/large_grammars |
| 165 |
+ grammars/sample_grammars |
| 166 |
+ misc/perluniprops |
| 167 |
+ models/bllip_wsj_no_aux |
| 168 |
+ models/moses_sample |
| 169 |
+ models/wmt15_eval |
| 170 |
+ models/word2vec_sample |
| 171 |
+ stemmers/porter_test |
| 172 |
+ stemmers/rslp |
| 173 |
+ taggers/averaged_perceptron_tagger |
| 174 |
+ taggers/averaged_perceptron_tagger_ru |
| 175 |
+ taggers/universal_tagset |
| 176 |
+ tokenizers/punkt |
| 177 |
+) |
| 178 |
+ |
| 179 |
+PACKAGES_UNPACK_2021=( |
| 180 |
+ corpora/stopwords |
| 181 |
+) |
| 182 |
+ |
| 183 |
+PACKAGES_UNPACK_2021_12=( |
| 184 |
+ corpora/inaugural |
| 185 |
+ corpora/omw-1.4 |
| 186 |
+ corpora/sinica_treebank |
| 187 |
+ corpora/wordnet2021 |
| 188 |
+ corpora/wordnet31 |
| 189 |
+) |
| 190 |
+ |
| 191 |
+PACKAGES_UNPACK_EXTRA_2020=( |
| 192 |
+ chunkers/maxent_ne_chunker |
| 193 |
+ corpora/biocreative_ppi |
| 194 |
+ corpora/brown_tei |
| 195 |
+ corpora/kimmo |
| 196 |
+ corpora/paradigms |
| 197 |
+ corpora/pe08 |
| 198 |
+ corpora/pil |
| 199 |
+ corpora/problem_reports |
| 200 |
+ corpora/smultron |
| 201 |
+ corpora/unicode_samples |
| 202 |
+ corpora/verbnet3 |
| 203 |
+ corpora/ycoe |
| 204 |
+ grammars/basque_grammars |
| 205 |
+ grammars/spanish_grammars |
| 206 |
+ help/tagsets |
| 207 |
+ misc/mwa_ppdb |
| 208 |
+ taggers/maxent_treebank_pos_tagger |
| 209 |
+) |
| 210 |
+ |
| 211 |
+add_data() { |
| 212 |
+ local x version=${1} |
| 213 |
+ shift |
| 214 |
+ |
| 215 |
+ for x; do |
| 216 |
+ SRC_URI+=" |
| 217 |
+ https://raw.githubusercontent.com/nltk/nltk_data/gh-pages/packages/${x}.zip |
| 218 |
+ -> nltk-${x#*/}-${version}.zip" |
| 219 |
+ done |
| 220 |
+} |
| 221 |
+ |
| 222 |
+add_data 20200312 "${PACKAGES_ZIP_2020[@]}" "${PACKAGES_UNPACK_2020[@]}" |
| 223 |
+add_data 20211023 "${PACKAGES_UNPACK_2021[@]}" |
| 224 |
+add_data 20211221 "${PACKAGES_UNPACK_2021_12[@]}" |
| 225 |
+SRC_URI+=" |
| 226 |
+ extra? (" |
| 227 |
+add_data 20200312 "${PACKAGES_UNPACK_EXTRA_2020[@]}" |
| 228 |
+SRC_URI+=" |
| 229 |
+ )" |
| 230 |
+ |
| 231 |
+CHECKREQS_DISK_USR=3G |
| 232 |
+CHECKREQS_DISK_BUILD=${CHECKREQS_DISK_USR} |
| 233 |
+ |
| 234 |
+unpack_data() { |
| 235 |
+ local x version=${1} |
| 236 |
+ shift |
| 237 |
+ |
| 238 |
+ for x; do |
| 239 |
+ local cat=${x%/*} |
| 240 |
+ local pkg=${x#*/} |
| 241 |
+ |
| 242 |
+ mkdir -p "${S}/${cat}" || die |
| 243 |
+ cd "${S}/${cat}" || die |
| 244 |
+ unpack "nltk-${pkg}-${version}.zip" |
| 245 |
+ done |
| 246 |
+} |
| 247 |
+ |
| 248 |
+src_unpack() { |
| 249 |
+ unpack_data 20200312 "${PACKAGES_UNPACK_2020[@]}" |
| 250 |
+ unpack_data 20211023 "${PACKAGES_UNPACK_2021[@]}" |
| 251 |
+ unpack_data 20211221 "${PACKAGES_UNPACK_2021_12[@]}" |
| 252 |
+ use extra && unpack_data 20200312 "${PACKAGES_UNPACK_EXTRA_2020[@]}" |
| 253 |
+} |
| 254 |
+ |
| 255 |
+install_zips() { |
| 256 |
+ local x version=${1} |
| 257 |
+ shift |
| 258 |
+ |
| 259 |
+ for x; do |
| 260 |
+ local cat=${x%/*} |
| 261 |
+ local pkg=${x#*/} |
| 262 |
+ |
| 263 |
+ insinto "/usr/share/nltk_data/${cat}" |
| 264 |
+ newins "${DISTDIR}/nltk-${pkg}-${version}.zip" "${pkg}.zip" |
| 265 |
+ done |
| 266 |
+} |
| 267 |
+ |
| 268 |
+src_install() { |
| 269 |
+ dodir /usr/share/nltk_data |
| 270 |
+ mv * "${ED}/usr/share/nltk_data/" || die |
| 271 |
+ |
| 272 |
+ install_zips 20200312 "${PACKAGES_ZIP_2020[@]}" |
| 273 |
+} |
Archives