gentoo resync : 25.04.2020

2 files changed, 185 insertions, 0 deletions

diff --git a/dev-python/nltk-data/Manifest b/dev-python/nltk-data/Manifest
index c671c3327b17..147e0105722e 100644
--- a/dev-python/nltk-data/Manifest
+++ b/dev-python/nltk-data/Manifest
@@ -105,5 +105,6 @@ DIST nltk-wordnet-20200312.zip 10775600 BLAKE2B dac56a8fb1fa6882b1871c394ad2acb2
 DIST nltk-wordnet_ic-20200312.zip 12056682 BLAKE2B c2dc2a646015b23699a72f636b588ec5718c70e6941d9d56863257e1e0396c8cf59ac1dc6ed74e5d7f0c2ee9129d63221a03967bf66a3d335e99160f295ed44d SHA512 1c94451a13af6c76bff60a0cab2e70402a3d9abd2e8fc62a5473f24ab4229feb0afe4faa8d389734697a6cf86d2c8b1dc700bb3afa3cbc279b75d7e0ec19fc6d
 DIST nltk-words-20200312.zip 757777 BLAKE2B eaaaaab6c26e206e9b6ce45daf779e3cc6706a06132afeabf013026d0009caee2d678f3c4ea9125b9654f7143bef29ec7a5706b79e5650ea556c6821b7754e6f SHA512 2810f05d3fc7ee6b6f8636fa1ff7b4e8c8cdac12b415cc54d15c69102290122ea138ec4fa36cb483f790c1ac10b0f83ae4c2c3e0e8df7e67e90e962ee5dbb0be
 DIST nltk-ycoe-20200312.zip 477 BLAKE2B 574835aa011a06a06363e26facd6a6f583a1dc1cac2de39adff59d8ab48eefac030b43d935a2f79af855259f2a9a571193dae2811589483af97406ff05c76c9e SHA512 e39ce165074d10ff63cb84ea52905d7ecb937797c8123ed113c5609afe1f63ac44d04d48a681002c4eac21dc9076ac74164b886c6f9ce42f3a102c38d1e8e756
+EBUILD nltk-data-20200312-r1.ebuild 3849 BLAKE2B dbceac6bb6c111f6953607ba9520ae42c08b809a594e76dbf05b472f0e392b6dbbf7c036cf207a2512079c9215d8db46f7f11e06302fe40997ac204baea4684c SHA512 7574fc9b41eeef51649a557be0f1e1ce06f6ac8163b3d18f7abee50fa0027760190be4c8349df4d51e450890b6d286f90e5d46bd8e53cef7a985ccc79bb4db47
 EBUILD nltk-data-20200312.ebuild 3849 BLAKE2B a2c28c659c7f6451c20641adb5c2d9dfbfc54612ce7edf2a865170fd31bf1f823b7aa7fef272e2207e722ca60670f732c764b7475c2ce1ca530b9da1642a7405 SHA512 59a1e5bf2276d913f2800db5f1f8a71d45cc0d63ca701b39cdd5f3b5b9ce72b460a01d5db57ad346917fa11b180617449ce7cff3804222406859855106e4b6a3
 MISC metadata.xml 389 BLAKE2B e9ecfd76b7c3cf4c8482ecbd809fa2dc94862b8315a3893afa7a57097000f4a048e990121e271d89655c77ad5a800669ce2b1a6fd5d01ad27ca6c3941c342e65 SHA512 1c9392dc21cf4cf203d368be88d5d7f3302570668a04caa2383038668542fb8c4e25c907aa8b86edd5e9e1f969c96aa1637f98e3538113c13e4e7c1084f4be5b
diff --git a/dev-python/nltk-data/nltk-data-20200312-r1.ebuild b/dev-python/nltk-data/nltk-data-20200312-r1.ebuild
new file mode 100644
index 000000000000..c8f12c580e60
--- /dev/null
+++ b/dev-python/nltk-data/nltk-data-20200312-r1.ebuild
@@ -0,0 +1,184 @@
+# Copyright 2020 Gentoo Authors
+# Distributed under the terms of the GNU General Public License v2
+
+EAPI=7
+
+inherit check-reqs
+
+DESCRIPTION="Data files for NLTK"
+HOMEPAGE="https://www.nltk.org/nltk_data/"
+
+# at least some of the files have poorly documented licenses
+# TODO: create a USE flag for free-ish subset
+LICENSE="all-rights-reserved"
+SLOT="0"
+KEYWORDS="~amd64 ~x86"
+IUSE="extra"
+RESTRICT="bindist mirror"
+
+BDEPEND="app-arch/unzip"
+
+PACKAGES_ZIP=(
+	# wget -O - https://www.nltk.org/nltk_data/ | xml sel -t -m '//package[@unzip=0]' -v @subdir -o "/" -v @id -n - | sort
+	corpora/comtrans
+	corpora/conll2007
+	corpora/jeita
+	corpora/knbc
+	corpora/machado
+	corpora/masc_tagged
+	corpora/nombank.1.0
+	corpora/panlex_swadesh
+	corpora/propbank
+	corpora/reuters
+	corpora/semcor
+	corpora/universal_treebanks_v20
+	sentiment/vader_lexicon
+	stemmers/snowball_data
+)
+
+PACKAGES_UNPACK=(
+	# wget -O - https://www.nltk.org/nltk_data/ | xml sel -t -m '//package[@unzip=1]' -v @subdir -o "/" -v @id -n - | sort
+	corpora/abc
+	corpora/alpino
+	corpora/brown
+	corpora/cess_cat
+	corpora/cess_esp
+	corpora/chat80
+	corpora/city_database
+	corpora/cmudict
+	corpora/comparative_sentences
+	corpora/conll2000
+	corpora/conll2002
+	corpora/crubadan
+	corpora/dependency_treebank
+	corpora/dolch
+	corpora/europarl_raw
+	corpora/floresta
+	corpora/framenet_v15
+	corpora/framenet_v17
+	corpora/gazetteers
+	corpora/genesis
+	corpora/gutenberg
+	corpora/ieer
+	corpora/inaugural
+	corpora/indian
+	corpora/lin_thesaurus
+	corpora/mac_morpho
+	corpora/movie_reviews
+	corpora/mte_teip5
+	corpora/names
+	corpora/nonbreaking_prefixes
+	corpora/nps_chat
+	corpora/omw
+	corpora/opinion_lexicon
+	corpora/pl196x
+	corpora/ppattach
+	corpora/product_reviews_1
+	corpora/product_reviews_2
+	corpora/pros_cons
+	corpora/ptb
+	corpora/qc
+	corpora/rte
+	corpora/senseval
+	corpora/sentence_polarity
+	corpora/sentiwordnet
+	corpora/shakespeare
+	corpora/sinica_treebank
+	corpora/state_union
+	corpora/stopwords
+	corpora/subjectivity
+	corpora/swadesh
+	corpora/switchboard
+	corpora/timit
+	corpora/toolbox
+	corpora/treebank
+	corpora/twitter_samples
+	corpora/udhr
+	corpora/udhr2
+	corpora/verbnet
+	corpora/webtext
+	corpora/wordnet
+	corpora/wordnet_ic
+	corpora/words
+	grammars/book_grammars
+	grammars/large_grammars
+	grammars/sample_grammars
+	misc/perluniprops
+	models/bllip_wsj_no_aux
+	models/moses_sample
+	models/wmt15_eval
+	models/word2vec_sample
+	stemmers/porter_test
+	stemmers/rslp
+	taggers/averaged_perceptron_tagger
+	taggers/averaged_perceptron_tagger_ru
+	taggers/universal_tagset
+	tokenizers/punkt
+)
+
+PACKAGES_UNPACK_EXTRA=(
+	chunkers/maxent_ne_chunker
+	corpora/biocreative_ppi
+	corpora/brown_tei
+	corpora/kimmo
+	corpora/paradigms
+	corpora/pe08
+	corpora/pil
+	corpora/problem_reports
+	corpora/smultron
+	corpora/unicode_samples
+	corpora/verbnet3
+	corpora/ycoe
+	grammars/basque_grammars
+	grammars/spanish_grammars
+	help/tagsets
+	misc/mwa_ppdb
+	taggers/maxent_treebank_pos_tagger
+)
+
+add_data() {
+	local x
+	for x; do
+		SRC_URI+="
+			https://raw.githubusercontent.com/nltk/nltk_data/gh-pages/packages/${x}.zip
+				-> nltk-${x#*/}-${PV}.zip"
+	done
+}
+
+add_data "${PACKAGES_ZIP[@]}" "${PACKAGES_UNPACK[@]}"
+SRC_URI+="
+	extra? ("
+add_data "${PACKAGES_UNPACK_EXTRA[@]}"
+SRC_URI+="
+	)"
+
+CHECKREQS_DISK_USR=3G
+CHECKREQS_DISK_BUILD=${CHECKREQS_DISK_USR}
+
+src_unpack() {
+	local x
+	local to_unpack=( "${PACKAGES_UNPACK[@]}" )
+	use extra && to_unpack+=( "${PACKAGES_UNPACK_EXTRA[@]}" )
+	for x in "${to_unpack[@]}"; do
+		local cat=${x%/*}
+		local pkg=${x#*/}
+
+		mkdir -p "${S}/${cat}" || die
+		cd "${S}/${cat}" || die
+		unpack "nltk-${pkg}-${PV}.zip"
+	done
+}
+
+src_install() {
+	dodir /usr/share/nltk_data
+	mv * "${ED}/usr/share/nltk_data/" || die
+
+	local x
+	for x in "${PACKAGES_ZIP[@]}"; do
+		local cat=${x%/*}
+		local pkg=${x#*/}
+
+		insinto "/usr/share/nltk_data/${cat}"
+		newins "${DISTDIR}/nltk-${pkg}-${PV}.zip" "${pkg}.zip"
+	done
+}