From 51f398613c3716559daacd64de9364124d39ee35 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Dani=C3=ABl=20de=20Kok?= Date: Sun, 16 May 2021 10:02:43 +0200 Subject: [PATCH] python3Packages.spacy_models: 2.3.0 -> 3.0.0 Update models for compatibility with spaCy 3.0.0. --- .../spacy/annotation-test/annotate.py | 10 +- .../python-modules/spacy/models.json | 596 ++++++++++-------- .../python-modules/spacy/models.nix | 20 +- 3 files changed, 355 insertions(+), 271 deletions(-) diff --git a/pkgs/development/python-modules/spacy/annotation-test/annotate.py b/pkgs/development/python-modules/spacy/annotation-test/annotate.py index 822eb8ac0743..eb6288084883 100644 --- a/pkgs/development/python-modules/spacy/annotation-test/annotate.py +++ b/pkgs/development/python-modules/spacy/annotation-test/annotate.py @@ -26,10 +26,11 @@ def test_entities(doc_en_core_web_sm): assert entities == [ ('Sebastian Thrun', 'PERSON'), - ('Google', 'ORG'), ('2007', 'DATE'), + ('2007', 'DATE'), ('American', 'NORP'), - ('Thrun', 'ORG'), - ('earlier this week', 'DATE') + ('Thrun', 'PERSON'), + ('Recode', 'PERSON'), + ('earlier this week', 'DATE'), ] @@ -60,10 +61,9 @@ def test_verbs(doc_en_core_web_sm): 'work', 'drive', 'take', - 'can', 'tell', - 'would', 'shake', 'turn', + 'be', 'talk', 'say'] diff --git a/pkgs/development/python-modules/spacy/models.json b/pkgs/development/python-modules/spacy/models.json index 3a427f8932a5..a2989b6da502 100644 --- a/pkgs/development/python-modules/spacy/models.json +++ b/pkgs/development/python-modules/spacy/models.json @@ -1,264 +1,332 @@ -[{ - "pname": "da_core_news_lg", - "version": "2.3.0", - "sha256": "18y1jany1ha27jkwb9563haxsbylm0axkh5c8009lsfxc8y2w9hr", - "license": "cc-by-sa-40" -}, -{ - "pname": "da_core_news_md", - "version": "2.3.0", - "sha256": "06nm5grj5jdx0rja7vw1f91vvd69p6vhafrwpfr1npqk24j6cacb", - "license": "cc-by-sa-40" -}, -{ - "pname": "da_core_news_sm", - "version": "2.3.0", - "sha256": "00byhlrcbg4wxplr473g9b3126pvk4vwy0q34xg0zx4994qb6rgn", - "license": "cc-by-sa-40" -}, -{ - "pname": "de_core_news_lg", - "version": "2.3.0", - "sha256": "0rixhgdz4z7sq6f2b1w3n5cn1645cr37g40hbd9xzlvdzdf5cg6d", - "license": "mit" -}, -{ - "pname": "de_core_news_md", - "version": "2.3.0", - "sha256": "0kxir1w000r5fn1kpa38m7688xinkn2mk1m82aiwqlck3r72jdi6", - "license": "mit" -}, -{ - "pname": "de_core_news_sm", - "version": "2.3.0", - "sha256": "00cbmrf4njg28laysapdnp4rv4lw4yw03rxkynw1ain5fwb0izl7", - "license": "mit" -}, -{ - "pname": "el_core_news_lg", - "version": "2.3.0", - "sha256": "001c19dd1kirlvzbdv2i89zw8nf6c4icv2w0chm7rd6x9m3i13gd", - "license": "cc-by-nc-sa-30" -}, -{ - "pname": "el_core_news_md", - "version": "2.3.0", - "sha256": "170x8bzm5nf02mhkxyxjk58yk2639hsjb5b9prcc69500c0vmnp0", - "license": "cc-by-nc-sa-30" -}, -{ - "pname": "el_core_news_sm", - "version": "2.3.0", - "sha256": "10mh3za4jvr07rawzk8ps642rp11s3smraj9xvrxflik4fqkz18b", - "license": "cc-by-nc-sa-30" -}, -{ - "pname": "en_core_web_lg", - "version": "2.3.0", - "sha256": "0mfa5wz31ya295jhyj489gb4qy806zmpq1zc11bvv5alv2m35if2", - "license": "mit" -}, -{ - "pname": "en_core_web_md", - "version": "2.3.0", - "sha256": "1ys8sqkhiap1mq6mhbkbq8bc07lvl68xngbx725xkwvirzl5gabh", - "license": "mit" -}, -{ - "pname": "en_core_web_sm", - "version": "2.3.0", - "sha256": "04icv9qf4pj53ll8vqxcjl2a723q1k00i7lifk8wx5saif28g37a", - "license": "mit" -}, -{ - "pname": "en_vectors_web_lg", - "version": "2.3.0", - "sha256": "13g012rwh0bcxx3ii5mmygqzyryah1y3zd000zhidnacc1x1g743", - "license": "cc-by-sa-30" -}, -{ - "pname": "es_core_news_lg", - "version": "2.3.0", - "sha256": "1r0pr0lzs3j9w7rd5z9nw87iayjm36v7f7gamvaiiphs6bc5p7ls", - "license": "gpl3" -}, -{ - "pname": "es_core_news_md", - "version": "2.3.0", - "sha256": "0nz33bmpr3rxqbnv6vb1id8pkfsvh8ii8vqplwgb3b8772kmpzy2", - "license": "gpl3" -}, -{ - "pname": "es_core_news_sm", - "version": "2.3.0", - "sha256": "02xqhg4m0gg5r9yibvl02zixkll6w0nsmbdhp07y5yyaqjarc90d", - "license": "gpl3" -}, -{ - "pname": "fr_core_news_lg", - "version": "2.3.0", - "sha256": "1yliamws8nqqjhpk9gr2dzlk0dms2mr958zbj21biv8fimbq60ik", - "license": "lgpllr" -}, -{ - "pname": "fr_core_news_md", - "version": "2.3.0", - "sha256": "04fk212ksac3bp9dj7dmzsdcnbqmbsgymsic6ddcv9zbfdv5d0db", - "license": "lgpllr" -}, -{ - "pname": "fr_core_news_sm", - "version": "2.3.0", - "sha256": "0kldww855z67qfc9maa9z1lsvdf5vj5vc8gj0x3h68kv5n1xr4h0", - "license": "lgpllr" -}, -{ - "pname": "it_core_news_lg", - "version": "2.3.0", - "sha256": "192rdmqnwl3ajxzhnw4r1cqv5bkziv0yc2bbzckmzqss64wk7k70", - "license": "cc-by-nc-sa-30" -}, -{ - "pname": "it_core_news_md", - "version": "2.3.0", - "sha256": "019ih4vwq1w6j38j0wc8pyyg1an6yy37wxq2w4amwppynmmcnd5w", - "license": "cc-by-nc-sa-30" -}, -{ - "pname": "it_core_news_sm", - "version": "2.3.0", - "sha256": "1c3ywqa8li0j7cyvd1xqbb096y61978hd6qv7rc6cxxjdhmkrrds", - "license": "cc-by-nc-sa-40" -}, -{ - "pname": "lt_core_news_lg", - "version": "2.3.0", - "sha256": "0hn5w8n7mgv33i6gvnaxl1j44n9gz4j86gg1a9jjlgdw5z98n0p2", - "license": "cc-by-sa-40" -}, -{ - "pname": "lt_core_news_md", - "version": "2.3.0", - "sha256": "1xya79cz2xd5vgzg6qg0ww5j2bmv7kppdk3mdjf6zpwrlzwdbk5d", - "license": "cc-by-sa-40" -}, -{ - "pname": "lt_core_news_sm", - "version": "2.3.0", - "sha256": "0r3rbqgz4897wyhz5jli30lryb45039f4rlvn4q0364cg1pm92g9", - "license": "cc-by-sa-40" -}, -{ - "pname": "nb_core_news_lg", - "version": "2.3.0", - "sha256": "18mblypw3c82004qz5w1p3262iqwn99wl9b781dq7aqwxskr02d6", - "license": "mit" -}, -{ - "pname": "nb_core_news_md", - "version": "2.3.0", - "sha256": "0iw97k9glxbar8mrpvnmmcb1nffgdhb83akn99p53pwmqbzxy9p3", - "license": "mit" -}, -{ - "pname": "nb_core_news_sm", - "version": "2.3.0", - "sha256": "07b7xri2q3m7fvn9a2gjc1044a3f14231vr32hrw96h7k6vg95h7", - "license": "mit" -}, -{ - "pname": "nl_core_news_lg", - "version": "2.3.0", - "sha256": "016166kzpgi0p3m0x3k308a0r60a28yz7npagjvmpl1dfm9lzhnv", - "license": "cc-by-sa-40" -}, -{ - "pname": "nl_core_news_md", - "version": "2.3.0", - "sha256": "1anfhig531k9k14s0cbgsvmvifp3h50qi1h8dhx894kjmq10k2lg", - "license": "cc-by-sa-40" -}, -{ - "pname": "nl_core_news_sm", - "version": "2.3.0", - "sha256": "0alvz7pn7cj0yax8h5gp71vrdblh3mcsmyhzgiddsd44ry35nxnj", - "license": "cc-by-sa-40" -}, -{ - "pname": "pl_core_news_lg", - "version": "2.3.0", - "sha256": "1acchp8pv1h4c6cwvxz07lh4ychn6aw809zfg3mbbsxgsgd2ahjr", - "license": "gpl3" -}, -{ - "pname": "pl_core_news_md", - "version": "2.3.0", - "sha256": "19jjjjvbys3ayibkm3cx497b4bh63ll39hfq04wx116rj4ajpwwg", - "license": "gpl3" -}, -{ - "pname": "pl_core_news_sm", - "version": "2.3.0", - "sha256": "166mqlfkgiszcc6hwg2mr6sir9y88y22nd81a1nidq0fiif5lfji", - "license": "gpl3" -}, -{ - "pname": "pt_core_news_lg", - "version": "2.3.0", - "sha256": "18gvdmfwyy2sbq6206imglhghyagd6a4gb6wcfkwhm7lzbnq714d", - "license": "cc-by-sa-40" -}, -{ - "pname": "pt_core_news_md", - "version": "2.3.0", - "sha256": "1yxnpwby2aq6ydvd35lylc4fs141fisfnzlx8pl88pp2b2gxijvl", - "license": "cc-by-sa-40" -}, -{ - "pname": "pt_core_news_sm", - "version": "2.3.0", - "sha256": "1vcvzdg9f93x0vaafkk9l9xhpmaavfj0cf0l3p06c5kx2d76f9ph", - "license": "cc-by-sa-40" -}, -{ - "pname": "ro_core_news_sm", - "version": "2.3.0", - "sha256": "0lsmbdwsaczv37y5sa1vvgwszy2hs8jp24a0nvc5qm8vb71rxj8w", - "license": "cc-by-sa-40" -}, -{ - "pname": "ro_core_news_md", - "version": "2.3.0", - "sha256": "1igwkz3yd1117gi2g78yilh9ln8n5yrdimas4prfxjgzwid3q8bc", - "license": "cc-by-sa-40" -}, -{ - "pname": "ro_core_news_lg", - "version": "2.3.0", - "sha256": "0id1y32kjfans7llh1i55rgr4n2x6xn208y4qf6yl3pbc17i0n9z", - "license": "cc-by-sa-40" -}, -{ - "pname": "xx_ent_wiki_sm", - "version": "2.3.0", - "sha256": "0x3zmmybl5kh4dn5prkfmr4q5j9bh13p40qc3rhdfi0i3jxc11pn", - "license": "mit" -}, -{ - "pname": "zh_core_web_lg", - "version": "2.3.1", - "sha256": "17zxk7cz47k07yb3qaigc3sx4dj4zwilr1lsn2jq6w7jc8k2h1ll", - "license": "mit" -}, -{ - "pname": "zh_core_web_md", - "version": "2.3.1", - "sha256": "1n4iwxyam4ykn0f9jdzwkhczack8r9c3kkbyga3c4h6iwqsflzcj", - "license": "mit" -}, -{ - "pname": "zh_core_web_sm", - "version": "2.3.1", - "sha256": "1lj5nwhx38cpwnvajwxlfkf84dr1xx2h6wwbg3scycsh459i9fpc", - "license": "mit" -}] +[ + { + "pname": "da_core_news_lg", + "version": "3.0.0", + "sha256": "0l0wljc1lm9a72ngfd4aa90laz4zcc37ix9nsiaqlw004v01z7wj", + "license": "cc-by-sa-40" + }, + { + "pname": "da_core_news_md", + "version": "3.0.0", + "sha256": "14h3ym22224aimfk2kj88pmn83hkb57w402i0x6pd7ra86n372lh", + "license": "cc-by-sa-40" + }, + { + "pname": "da_core_news_sm", + "version": "3.0.0", + "sha256": "05893dpmx76waqnlysnkq8hz9271rkk30xf6hy98gka6244l9a1l", + "license": "cc-by-sa-40" + }, + { + "pname": "de_core_news_lg", + "version": "3.0.0", + "sha256": "0s7vfpr9gv22pvh697ffg35fppxkjhw23ynf4bpz73hl9jikdqvj", + "license": "mit" + }, + { + "pname": "de_core_news_md", + "version": "3.0.0", + "sha256": "09vvlm3rxmyiima81y4bvcyxhn9bjxrqlkbmglzmwhrhxm84nkmx", + "license": "mit" + }, + { + "pname": "de_core_news_sm", + "version": "3.0.0", + "sha256": "1w5aqfzknnnxpsi9i6kn6bki58j0mp24d4gr2203bf6g5kahiq03", + "license": "mit" + }, + { + "pname": "de_dep_news_trf", + "version": "3.0.0", + "sha256": "1snkm911jn73mqfz0y0anr12r6j3gdi6wd8qmd5alwm623x4s6hm", + "license": "mit" + }, + { + "pname": "el_core_news_lg", + "version": "3.0.0", + "sha256": "011lqmx3f3laf2vvqp0hxp5y105pn54kwdh1xzw4hs2pj6fac9p5", + "license": "cc-by-nc-sa-30" + }, + { + "pname": "el_core_news_md", + "version": "3.0.0", + "sha256": "0p75c18sg38j9dj79ykmm5kzcwjxccpgrcw4cjcscb6ad6wwvcjx", + "license": "cc-by-nc-sa-30" + }, + { + "pname": "el_core_news_sm", + "version": "3.0.0", + "sha256": "0gvisa7yg1w49hpfp79ahy50v64l3rmk56k0k7zkgc8ml1gn892r", + "license": "cc-by-nc-sa-30" + }, + { + "pname": "en_core_web_lg", + "version": "3.0.0", + "sha256": "0sdb85zvfb12d14k3wy23dfryy3xwc9ag79zq71qnxwpvvygmc8y", + "license": "mit" + }, + { + "pname": "en_core_web_md", + "version": "3.0.0", + "sha256": "0c669b1vsp3z28n52lfsijmkn9465r8zjjygjml5rlf9lf1paxa5", + "license": "mit" + }, + { + "pname": "en_core_web_sm", + "version": "3.0.0", + "sha256": "0risizvzkicffw7vgrj92z23dfb7zvvzihqgvjcrx8989b7b6wq6", + "license": "mit" + }, + { + "pname": "en_core_web_trf", + "version": "3.0.0", + "sha256": "0plmg77rv1spr0swn4jakci16mbqsxm32mz9nnwc9ylynbvgrhmn", + "license": "mit" + }, + { + "pname": "es_core_news_lg", + "version": "3.0.0", + "sha256": "0832w8qmg0fp2q8329ndlbbzpfkpbw9v38ny7106a45xaz0rn2xc", + "license": "gpl3" + }, + { + "pname": "es_core_news_md", + "version": "3.0.0", + "sha256": "01is980r63a5418jq917scapzkl9xydj56lrsxbr16fya0hh8qnn", + "license": "gpl3" + }, + { + "pname": "es_core_news_sm", + "version": "3.0.0", + "sha256": "1wgya0f25dgix57pb60fyl4hf2msma16d1f6cf617ypk6g3v80rb", + "license": "gpl3" + }, + { + "pname": "es_dep_news_trf", + "version": "3.0.0", + "sha256": "07lim35p0mxb75qiym79wcrak3j7wcan393260haxgwrj29rzpvv", + "license": "gpl3" + }, + { + "pname": "fr_core_news_lg", + "version": "3.0.0", + "sha256": "1frg734lb64gkm7pagqp1mj7lqpwsxxj5vyjm10yja0rkdi6kcca", + "license": "lgpllr" + }, + { + "pname": "fr_core_news_md", + "version": "3.0.0", + "sha256": "1xshr9r639hdb8vkj5nribk4lkm3a5fb7zrxj3y3p678dr53xalz", + "license": "lgpllr" + }, + { + "pname": "fr_core_news_sm", + "version": "3.0.0", + "sha256": "0n23c9rbg1b44c8yjlf6cc0g8ccj6x0rmfjg76ddmpkjaj83jwv1", + "license": "lgpllr" + }, + { + "pname": "fr_dep_news_trf", + "version": "3.0.0", + "sha256": "192l6n5yxn1ndc4fk8k759j2d5hryj9mfkpy2aminaxr4dmp2imr", + "license": "lgpllr" + }, + { + "pname": "it_core_news_lg", + "version": "3.0.0", + "sha256": "121nki732wrnfyzcflvsv54nvrz3l3hx55hkd72hlhrvjw0kvkv5", + "license": "cc-by-nc-sa-30" + }, + { + "pname": "it_core_news_md", + "version": "3.0.0", + "sha256": "0m168wrf1p6dz5kc4n5ga2h8c0d6jzxx876i3ndrg6b7z418hhi5", + "license": "cc-by-nc-sa-30" + }, + { + "pname": "it_core_news_sm", + "version": "3.0.0", + "sha256": "132v06cah8l7q4caxg6n4nw34v9jd8y8cqp20njryx4nirm9c36l", + "license": "cc-by-nc-sa-30" + }, + { + "pname": "lt_core_news_lg", + "version": "3.0.0", + "sha256": "034qycqpbdiyzhpzjz92kpnx6z2nai70dddz75r48hylzlw1d92h", + "license": "cc-by-sa-40" + }, + { + "pname": "lt_core_news_md", + "version": "3.0.0", + "sha256": "18mb2lmrjwnsc7s4yaq3yvdbh8p8p1k0xpm8cqn521hanpr0jqj3", + "license": "cc-by-sa-40" + }, + { + "pname": "lt_core_news_sm", + "version": "3.0.0", + "sha256": "1p998h9lnp16czj3gg8781gywg17dap2h9f8qc6f87daxyc9bdjs", + "license": "cc-by-sa-40" + }, + { + "pname": "mk_core_news_lg", + "version": "3.0.0", + "sha256": "1fmrpgq9plndid7402wkybidpi0phnybb3031jxppan99ihr3hfj", + "license": "cc-by-sa-40" + }, + { + "pname": "mk_core_news_md", + "version": "3.0.0", + "sha256": "1mnabkyjxph2xa4g2an5rqp24d4gbq969ln27zpjycyiwxlkz7vl", + "license": "cc-by-sa-40" + }, + { + "pname": "mk_core_news_sm", + "version": "3.0.0", + "sha256": "1ax6pl61n0p4qf4wcd6c7d42zqjrgh3vhlpl6xby57a78547asxr", + "license": "cc-by-sa-40" + }, + { + "pname": "nb_core_news_lg", + "version": "3.0.0", + "sha256": "07a8nsfswlpb2jc2afzf201bjkl2nlz40kqmjx3dbva8jphj3ljs", + "license": "mit" + }, + { + "pname": "nb_core_news_md", + "version": "3.0.0", + "sha256": "0y1vydhhgb6cifq7k4vc7dy4jl6wb1z6pklbv65v6nxl7rhn76fz", + "license": "mit" + }, + { + "pname": "nb_core_news_sm", + "version": "3.0.0", + "sha256": "1lk1869cb2176j6lvd8lraclfl706p12m1gvvf1ixm99ra8zkxhs", + "license": "mit" + }, + { + "pname": "nl_core_news_lg", + "version": "3.0.0", + "sha256": "0iq4ayzh9g9gil4l8kcl5qcm0l16hymavsqgiczf3ddvamciqaxs", + "license": "cc-by-sa-40" + }, + { + "pname": "nl_core_news_md", + "version": "3.0.0", + "sha256": "0g9dkzwxndcf05bnhkd9fzqj7n614naspyhalg6h9h1kb7v3m2ak", + "license": "cc-by-sa-40" + }, + { + "pname": "nl_core_news_sm", + "version": "3.0.0", + "sha256": "1l4mk3gs15yc5kssy4x4lyab9kmg9y199h4hvizwh8y1ifqbqy03", + "license": "cc-by-sa-40" + }, + { + "pname": "pl_core_news_lg", + "version": "3.0.0", + "sha256": "14ldch7rps1qxh3ldczh7f29ra3dq2kxaxpfbx7r6f1xpmk5s1rv", + "license": "gpl3" + }, + { + "pname": "pl_core_news_md", + "version": "3.0.0", + "sha256": "0fx6ipd8ll2d0w8qwn9cjw0q7w0r3l40467d6mizi4mx93q7m7iw", + "license": "gpl3" + }, + { + "pname": "pl_core_news_sm", + "version": "3.0.0", + "sha256": "0p1gcniyrn9qya4wg1qd6ijfchc7lhk0dh4iba8y71mss3n162fs", + "license": "gpl3" + }, + { + "pname": "pt_core_news_lg", + "version": "3.0.0", + "sha256": "1vsw1ng364466jz6ffyj9dj3jh4s68gk7csxylc1fj7wac8jxrbj", + "license": "cc-by-sa-40" + }, + { + "pname": "pt_core_news_md", + "version": "3.0.0", + "sha256": "11laikyd6m9zprk7bnfn0v2lixvkcgvpv95dp0zpc0q2izmky6q8", + "license": "cc-by-sa-40" + }, + { + "pname": "pt_core_news_sm", + "version": "3.0.0", + "sha256": "12d59q1gvpx8dj48iq17sindd6wid09hnjb4fw0rb00bb28rmqk1", + "license": "cc-by-sa-40" + }, + { + "pname": "ro_core_news_lg", + "version": "3.0.0", + "sha256": "11mkip5piy6g7rg51ljqjn61s4ydlafl6qp3v29zmm3lghc66h8c", + "license": "cc-by-sa-40" + }, + { + "pname": "ro_core_news_md", + "version": "3.0.0", + "sha256": "1jyf3khki7xqbp6ir0p4y2p7rdhs36zq2i1960ik4kr2mhnzrblg", + "license": "cc-by-sa-40" + }, + { + "pname": "ro_core_news_sm", + "version": "3.0.0", + "sha256": "0gc61gdfgji92mjdpznkf30nl1fz7378h9bz8dlhvnc401hjzsky", + "license": "cc-by-sa-40" + }, + { + "pname": "ru_core_news_lg", + "version": "3.0.0", + "sha256": "1x1hxvhki62ypj3x0s4syfhz3znlflp36qkp4l2g2sbxjj2qw7n3", + "license": "mit" + }, + { + "pname": "ru_core_news_md", + "version": "3.0.0", + "sha256": "0ks0qdyq6627cbg8fbbhvr83d3m8njs2aj8pri540gz9nrbj5479", + "license": "mit" + }, + { + "pname": "ru_core_news_sm", + "version": "3.0.0", + "sha256": "1x3bmd7f0fqf03wds01imwpbv4xng1qq9iq61m8rbqvskm5jlzbb", + "license": "mit" + }, + { + "pname": "xx_ent_wiki_sm", + "version": "3.0.0", + "sha256": "1115vap4c6snvkwq8bmc8dld1mw5ky0w9n112hadp85pv012ccds", + "license": "mit" + }, + { + "pname": "xx_sent_ud_sm", + "version": "3.0.0", + "sha256": "062g3xfb3fp33b56wa4fj84smr5rlc0dbja102khxnqm2aakk99k", + "license": "cc-by-sa-30" + }, + { + "pname": "zh_core_web_lg", + "version": "3.0.0", + "sha256": "1ai34fc2wfmb35f1zissddf6jjqpg51wqiyqqq35h03jyf4731jr", + "license": "mit" + }, + { + "pname": "zh_core_web_md", + "version": "3.0.0", + "sha256": "10npzl8nvyj4jdn2f9iai9inq5c4x3hxdk0ycgg9wcgqaj09gnxa", + "license": "mit" + }, + { + "pname": "zh_core_web_sm", + "version": "3.0.0", + "sha256": "1f9x5lr8vnvb1n8hc59vm2xi6kv2rj78x1vm916z6ic3vg7vwl1h", + "license": "mit" + }, + { + "pname": "zh_core_web_trf", + "version": "3.0.0", + "sha256": "178w8dfcvx4aabasid6r0pnwqd5k02cvlq35siqjgfn7j3zb56z0", + "license": "mit" + } +] diff --git a/pkgs/development/python-modules/spacy/models.nix b/pkgs/development/python-modules/spacy/models.nix index 8c57d2e0d262..c34bbdfb83d8 100644 --- a/pkgs/development/python-modules/spacy/models.nix +++ b/pkgs/development/python-modules/spacy/models.nix @@ -1,4 +1,12 @@ -{ lib, buildPythonPackage, fetchurl, jieba, pkuseg, spacy }: +{ lib +, buildPythonPackage +, fetchurl +, jieba +, pymorphy2 +, sentencepiece +, spacy +, spacy-pkuseg +, spacy-transformers }: let buildModelPackage = { pname, version, sha256, license }: let @@ -12,7 +20,15 @@ let }; propagatedBuildInputs = [ spacy ] - ++ lib.optionals (lang == "zh") [ jieba pkuseg ]; + ++ lib.optionals (lang == "zh") [ jieba spacy-pkuseg ] + ++ lib.optionals (lib.hasSuffix "_trf" pname) [ spacy-transformers ] + ++ lib.optionals (lang == "ru") [ pymorphy2 ] + ++ lib.optionals (pname == "fr_dep_news_trf") [ sentencepiece ]; + + postPatch = lib.optionals (pname == "fr_dep_news_trf") '' + substituteInPlace meta.json \ + --replace "sentencepiece==0.1.91" "sentencepiece>=0.1.91" + ''; pythonImportsCheck = [ pname ];