diff options
| -rw-r--r-- | Pipfile | 3 | ||||
| -rw-r--r-- | Pipfile.lock | 277 | ||||
| -rw-r--r-- | fuzzycat/build.py | 2 | ||||
| -rw-r--r-- | fuzzycat/cluster.py | 4 | ||||
| -rw-r--r-- | fuzzycat/main.py | 11 |
5 files changed, 148 insertions, 149 deletions
@@ -13,13 +13,10 @@ pytest = "*" mypy = "*" [packages] -elasticsearch = "*" fuzzy = "*" fuzzycat = {editable = true, path = "."} -orjson = "*" # cf. https://github.com/pypa/pipenv/issues/4476 importlib-metadata = "==1.7.0" -pydantic = "*" tokenizers = "*" spacy = "*" nltk = "*" diff --git a/Pipfile.lock b/Pipfile.lock index 58e4c19..58715c6 100644 --- a/Pipfile.lock +++ b/Pipfile.lock @@ -1,7 +1,7 @@ { "_meta": { "hash": { - "sha256": "7ce35e56370b80894729da758bbf6a95964dd8566796399a76c2361ed3408759" + "sha256": "f26f7cc33688c532206d12bd14b01223974161a529c7b62fd6b62614e691d548" }, "pipfile-spec": 6, "requires": { @@ -47,10 +47,10 @@ }, "certifi": { "hashes": [ - "sha256:5930595817496dd21bb8dc35dad090f1c2cd0adfaf21204bf6732ca5d8ee34d3", - "sha256:8fc0819f1f30ba15bdb34cceffb9ef04d99f420f68eb75d901e9560b8749fc41" + "sha256:1f422849db327d534e3d0c5f02a263458c3955ec0aae4ff09b95f195c59f4edd", + "sha256:f05def092c44fbf25834a51509ef6e631dc19765ab8a57b4e7ab85531f0a9cf4" ], - "version": "==2020.6.20" + "version": "==2020.11.8" }, "chardet": { "hashes": [ @@ -87,11 +87,11 @@ }, "elasticsearch": { "hashes": [ - "sha256:5e08776fbb30c6e92408c7fa8c37d939210d291475ae2f364f0497975918b6fe", - "sha256:8c7e2374f53ee1b891ff2804116e0c7fb517585d6d5788ba668686bbc9d82e2d" + "sha256:9053ca99bc9db84f5d80e124a79a32dfa0f7079b2112b546a03241c0dbeda36d", + "sha256:9a21bfa7dc6a0b0dc142088bd653d8ce5ab284b4f7a3ded716185adf5276a7fe" ], - "index": "pypi", - "version": "==7.9.1" + "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3' and python_version < '4'", + "version": "==7.10.0" }, "ftfy": { "hashes": [ @@ -204,29 +204,6 @@ "markers": "python_version >= '3.6'", "version": "==1.19.4" }, - "orjson": { - "hashes": [ - "sha256:10c8abeb66db256fe36c4e2d38184fa1b38886594a2632f10a57fe3a40f905ef", - "sha256:1228424850dc7b25d0b54daacd6d220576f042d7c69362505acdb57d3b5c3e22", - "sha256:1c8d666599ec58322d24fa994edf7359c571cdb19aab5893f52aef4bdcb6e0f7", - "sha256:23f26dbb8378740c8d91ddbfaab1cbeb134d7d8a787e2ea40800def30df81b27", - "sha256:32c275ef90397e798f8134fb7a9c1d1d03c8eabf74c98b2552abfc78522676c4", - "sha256:428be770ad5d307e01acf7f41eacb73b1498bc2e12803cea9414211835f7fa60", - "sha256:5db5cca6b8e698225b65ad659306775f4503cb335de62ff37dbc064db31b1b79", - "sha256:6e9b33d7c5baa69fd1c4bfa149382f7fb7a0ec3d8c3e49dc1710d56c47fb586c", - "sha256:8025789a4902770ad46837fee5ac962ec35d5a9b39a75d5bd112cfc50e9c2dfc", - "sha256:857577b617425b09a3adc110c596e6d8801d481b671e2a4224d669c585521169", - "sha256:8bb241a582d25e13294424f80396c25ecb8d459e9e60cf114297fd57924d0a7b", - "sha256:941c0a083aeec2a9ef37390c3f12d5867e93fd2742c7bc264a56222842340c6d", - "sha256:a6c5646338d823b96c30b75db40f2cd85e91f1f1f7669994802276af555f7d66", - "sha256:ae606d50d1c24cebb48059effa6198d3de73a2299b9a1d50cc06c7d29331e83a", - "sha256:bdfdc925a446ef3b7502429a458303a96adc02c9e47a27a133a68403052731bf", - "sha256:d0e13f05c62cddf7619318545a9366693c93166452f18b253209579b1981c4d8", - "sha256:d520312744c3d5c27ca34ee78277819ed2ec8a3d748ea81217341e5cd509212f" - ], - "index": "pypi", - "version": "==3.4.3" - }, "plac": { "hashes": [ "sha256:398cb947c60c4c25e275e1f1dadf027e7096858fb260b8ece3b33bcff90d985f", @@ -281,64 +258,62 @@ "sha256:f83f679e727742b0c465e7ef992d6da4a7e5268b8edd8fdaf5303276374bef52", "sha256:fc21a37ff3f545de80b166e1735c4172b41b017948a3fb2d5e2f03c219eac50a" ], - "index": "pypi", + "markers": "python_version >= '3.6'", "version": "==1.7.2" }, "regex": { "hashes": [ - "sha256:03855ee22980c3e4863dc84c42d6d2901133362db5daf4c36b710dd895d78f0a", - "sha256:06b52815d4ad38d6524666e0d50fe9173533c9cc145a5779b89733284e6f688f", - "sha256:11116d424734fe356d8777f89d625f0df783251ada95d6261b4c36ad27a394bb", - "sha256:119e0355dbdd4cf593b17f2fc5dbd4aec2b8899d0057e4957ba92f941f704bf5", - "sha256:127a9e0c0d91af572fbb9e56d00a504dbd4c65e574ddda3d45b55722462210de", - "sha256:1ec66700a10e3c75f1f92cbde36cca0d3aaee4c73dfa26699495a3a30b09093c", - "sha256:227a8d2e5282c2b8346e7f68aa759e0331a0b4a890b55a5cfbb28bd0261b84c0", - "sha256:2564def9ce0710d510b1fc7e5178ce2d20f75571f788b5197b3c8134c366f50c", - "sha256:297116e79074ec2a2f885d22db00ce6e88b15f75162c5e8b38f66ea734e73c64", - "sha256:2dc522e25e57e88b4980d2bdd334825dbf6fa55f28a922fc3bfa60cc09e5ef53", - "sha256:3a5f08039eee9ea195a89e180c5762bfb55258bfb9abb61a20d3abee3b37fd12", - "sha256:3dfca201fa6b326239e1bccb00b915e058707028809b8ecc0cf6819ad233a740", - "sha256:49461446b783945597c4076aea3f49aee4b4ce922bd241e4fcf62a3e7c61794c", - "sha256:4afa350f162551cf402bfa3cd8302165c8e03e689c897d185f16a167328cc6dd", - "sha256:4b5a9bcb56cc146c3932c648603b24514447eafa6ce9295234767bf92f69b504", - "sha256:52e83a5f28acd621ba8e71c2b816f6541af7144b69cc5859d17da76c436a5427", - "sha256:625116aca6c4b57c56ea3d70369cacc4d62fead4930f8329d242e4fe7a58ce4b", - "sha256:654c1635f2313d0843028487db2191530bca45af61ca85d0b16555c399625b0e", - "sha256:8092a5a06ad9a7a247f2a76ace121183dc4e1a84c259cf9c2ce3bbb69fac3582", - "sha256:832339223b9ce56b7b15168e691ae654d345ac1635eeb367ade9ecfe0e66bee0", - "sha256:8ca9dca965bd86ea3631b975d63b0693566d3cc347e55786d5514988b6f5b84c", - "sha256:96f99219dddb33e235a37283306834700b63170d7bb2a1ee17e41c6d589c8eb9", - "sha256:9b6305295b6591e45f069d3553c54d50cc47629eb5c218aac99e0f7fafbf90a1", - "sha256:a62162be05edf64f819925ea88d09d18b09bebf20971b363ce0c24e8b4aa14c0", - "sha256:aacc8623ffe7999a97935eeabbd24b1ae701d08ea8f874a6ff050e93c3e658cf", - "sha256:b45bab9f224de276b7bc916f6306b86283f6aa8afe7ed4133423efb42015a898", - "sha256:b88fa3b8a3469f22b4f13d045d9bd3eda797aa4e406fde0a2644bc92bbdd4bdd", - "sha256:b8a686a6c98872007aa41fdbb2e86dc03b287d951ff4a7f1da77fb7f14113e4d", - "sha256:bd904c0dec29bbd0769887a816657491721d5f545c29e30fd9d7a1a275dc80ab", - "sha256:bf4f896c42c63d1f22039ad57de2644c72587756c0cfb3cc3b7530cfe228277f", - "sha256:c13d311a4c4a8d671f5860317eb5f09591fbe8259676b86a85769423b544451e", - "sha256:c2c6c56ee97485a127555c9595c069201b5161de9d05495fbe2132b5ac104786", - "sha256:c32c91a0f1ac779cbd73e62430de3d3502bbc45ffe5bb6c376015acfa848144b", - "sha256:c3466a84fce42c2016113101018a9981804097bacbab029c2d5b4fcb224b89de", - "sha256:c454ad88e56e80e44f824ef8366bb7e4c3def12999151fd5c0ea76a18fe9aa3e", - "sha256:c8a2b7ccff330ae4c460aff36626f911f918555660cc28163417cb84ffb25789", - "sha256:cb905f3d2e290a8b8f1579d3984f2cfa7c3a29cc7cba608540ceeed18513f520", - "sha256:cfcf28ed4ce9ced47b9b9670a4f0d3d3c0e4d4779ad4dadb1ad468b097f808aa", - "sha256:dd3e6547ecf842a29cf25123fbf8d2461c53c8d37aa20d87ecee130c89b7079b", - "sha256:de7fd57765398d141949946c84f3590a68cf5887dac3fc52388df0639b01eda4", - "sha256:ea37320877d56a7f0a1e6a625d892cf963aa7f570013499f5b8d5ab8402b5625", - "sha256:f1fce1e4929157b2afeb4bb7069204d4370bab9f4fc03ca1fbec8bd601f8c87d", - "sha256:f43109822df2d3faac7aad79613f5f02e4eab0fc8ad7932d2e70e2a83bd49c26" - ], - "version": "==2020.10.28" + "sha256:064d2fc83ab4ee0055fcc1ef38ec60e505742850a40061f854ac64cb3d8d6dd3", + "sha256:0951c78fa4cb26d1278a4b3784fcf973fc97ec39c07483328a74b034b0cc569c", + "sha256:0a235841237d4487329bcabcb5b902858f7967f5e684e08e968367f25b2c3d37", + "sha256:11d9100bd874ce8b2a037db9150e732cd768359fc25fe5f77973208aa24eb13e", + "sha256:19ac2bf0048a2f4d460ee20647e84ca160512a7ee8af844dc9207720778470f1", + "sha256:267d1b13f863e664150948ce2a9ed4927bf4ac7a068780f1ee8af83352aa17a2", + "sha256:3002ee2d4e8bbe4656237627203d8290a562d1fc1962deee470905ab63570345", + "sha256:32f8714c4bcc4b0d2aa259b1647e3c5b6cfe2e923c6c124234a5e03408224227", + "sha256:394b5be4fa72354a78763b317f82997ad881896dd4a860e429a6fa74afaacb07", + "sha256:396411bb5a7849aeda9c49873b8295919fdc118c50b57122b09cb2097047c118", + "sha256:3b46a4c73ec1f25361147a7a0fd86084f3627dc78d09bcbe14e70db12683efec", + "sha256:412969d58ecd4f576510ec88bcb7602e9e582bbef78859ed8c9ca4de4f9e891c", + "sha256:4159ecf20dffea07f4a7241b2a236f90eb622c7e8caab9f43caba5f27ca37284", + "sha256:48e94218f06317b6d32feb4ecff8b6025695450009bcb3291fb23daf79689431", + "sha256:56d1e298bb6482d0466399a6383181bf2627c37ad414e205b3ce0f85aa140be7", + "sha256:68267a7a5fb0bd9676b86f967143b6a6ecefb3eed4042ecc9e7f0e014aef8f74", + "sha256:6d128368def4b0cd95c0fc9d99a89ae73c083b25e67f27a410830e30f9df0edc", + "sha256:6e50b3b417ab2fd67bfa6235f0df4782fe2ff8be83f0c4435e1dc43d25052ee8", + "sha256:787e44e5f4fd027dd90b5ee0240b05dc1752cb43c2903617f25baa495fe551e9", + "sha256:8060be04baec546fe3afa6975d2998e15d1b655d7255f0e6b0ed3f482cccc218", + "sha256:826d0119f14f9a9ce25999a13ed5922c785b50e469800f6e5a6721318650ef49", + "sha256:83a390a653c13be1ab26287240df1fd9324ca8a0d31b603fa57cd7d9520648fa", + "sha256:84ab584dcb5e81815040d86148805a808acb0bee303d19638fe2f9488d704bc1", + "sha256:86ad88c7c2512094a85b0a01ce053bab1e28eafb8f3868bb8c22f4903e33f147", + "sha256:8cc3717146ce4040419639cf45455663a002a554806ddac46304acc5bd41dae2", + "sha256:9e8b3187f6beea8e56cb4b33c35049cbe376cf69aefaee5bc035309d88c98ca5", + "sha256:a9f76d9122359b09e38f27cd9c41729169171cf0fd73ec5b22cc4628f9e486ca", + "sha256:bb17a7fe9c47167337009ce18cd6e6b3edf3ca0063bf6bed6ce02515129c016a", + "sha256:beae9db1545f8116cfc9301a9601e9c975bb56ca22a38ac0fe06a72c3460f31a", + "sha256:bf02ab95ff5261ba108725dbd795bf6395eaac1b8468b41472d82d35b12b0295", + "sha256:c67fd5f3ad81f8301184354014e8e7510ab77e0c7e450a427d77f28ae8effbef", + "sha256:c8b1ad791debd67221fb1266f8d09730ae927acacb32d0dad9fd07a7d341a28f", + "sha256:ccfea4911ac28a8f744096bce1559e0bd86b09a53c8a9d5856ca8e1f5f4de1f5", + "sha256:cdb98be55db1b94c950822cbc10d3d768f01e184365851ebb42cd377486ced7b", + "sha256:cefcdb2ac3b67fd9f7244820ce1965c8cf352366199cc1358d67c6cc3c5c8bbc", + "sha256:d1e57c16c4840f1c3543507742e99b8398609474a0e6a6925476914479de3488", + "sha256:dd7bee615680d940dd44ac0a479f2bc5f73d6ca63a5915cd8d30739c14ca522c", + "sha256:df50ba964812606663ca9d23d374036bc5ae3d71e86168409cdd84ca7948d8a3", + "sha256:e03867f3baf64ecab47dfc9ddb58afc67acb6a0f80f6cf8ff9fa82962ec4d1cd", + "sha256:e7cdd5ee8053c82607432b7ebad37e2ece54548fef2b254f7bce6f7831904586", + "sha256:e899b69dd5d26655cb454835ea2fceb18832c9ee9c4fb45dc4cf8a6089d35312" + ], + "version": "==2020.11.11" }, "requests": { "hashes": [ - "sha256:b3559a131db72c33ee969480840fff4bb6dd111de7dd27c8ee1f820f4f00231b", - "sha256:fe75cc94a9443b9246fc7049224f75604b113c36acb93f87b80ed42c44cbb898" + "sha256:7f1a0b932f4a60a1a65caa4263921bb7d9ee911957e0ae4a23a6dd08185ad5f8", + "sha256:e786fa28d8c9154e6a4de5d46a1d921b8749f8b74e28bde23768e5e16eece998" ], "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4'", - "version": "==2.24.0" + "version": "==2.25.0" }, "spacy": { "hashes": [ @@ -358,21 +333,21 @@ }, "srsly": { "hashes": [ - "sha256:0d74328b0c9d6c4694ee2d0ab35232d1d7996cf1c205b253a636a9e1a2686490", - "sha256:15d3edf65eb91281991f9c02405c7c0c00d6e92cf50fefbd896b96df21c325a4", - "sha256:228665f648e5f88a21790c044428bd05d05b71e4466a89bba0752e0da0b8c791", - "sha256:25bd33819a0e95dbe5d49072d07c361827658cf5f7d69d15e580c43d5adb873a", - "sha256:2e1359a6fe106904b129651fed3dd240df07fe0669aaa68755e0133a6f222e40", - "sha256:310568b273602bfa922866bd503fd1892aef6e62040af4a783bd5de24865b28e", - "sha256:3a80183e7f31a1c9d9e4777c34e62b08105131412b72498a7406702843cfe522", - "sha256:6186dff418f4240a274ede369e69c304d7486de110f911d1b748d26af8138413", - "sha256:924a73334503d69b17ad85af22378bffe55ffb7210a0966fac115831a3e1ddec", - "sha256:c23ff73ade789bc621fa1152d407622ba4bbe6b113fd84fcb65bdceb12997aab", - "sha256:c8bf0111ce888831e22d092a573135281b495dee9d9b3c18dddae0fcacd0e4ca", - "sha256:e68198bc7a27d5c485aefaeac4dbd8630f4b501a95b34b3d09993db00fa9bd45", - "sha256:fb575a6b288da19fed05ed5f97528e9242a276e9a109b18f94074117cc435431" - ], - "version": "==1.0.3" + "sha256:02acfa09769380023e9bbed5c63f5f41b04dbc13ff392a7284df685127d224ae", + "sha256:2ae6327e013934126f2b7082ce36284785eed951635ac99c73e39de75ecd1345", + "sha256:2b9e2b7c00e522d6e54952564d702e72bc4729cc828e27a01b4dac70b870d44a", + "sha256:39a8bd6b383b4ea5c3a6793d790e1356e236701f421badef278a0ea80bc95d30", + "sha256:3f46e5efbbb0e857f302394257539aae4487c201cdafc3519db4bde897341fee", + "sha256:4b8bd311fa20fa5d1a0deac0dd8f3c43c7ca23df2c483a7102d96acf2cddb506", + "sha256:660b80ffa3ea4ef9db254e5a22d2f52294f15342910c3274844a8f5a81369cbc", + "sha256:713766bf4289ebe6f21d1e5203b6b75f71fd4512e877dbe5a3f660b4a51d77ac", + "sha256:9ca5633a5303ce0d0b84d1bdb6d029f665ba2b7d320f5482525b125ddfb8a390", + "sha256:b4fb0ff57e25a99e9ac9a61b262478efe8493b3cff0f50812c09534c76688b0b", + "sha256:c1fa001bbd87a771998c8e2487991fc0dfe16edc09e3dec0d64f9472832f89bd", + "sha256:deee259837443ea8a4c194967f56eb62809c3cf3a11cd9d4f425de4aea5ecb81", + "sha256:f7aee40cd66f190e7694d7e7c83be018b1a7b9546a305db52c8793d93b1d0a7e" + ], + "version": "==1.0.4" }, "thinc": { "hashes": [ @@ -393,26 +368,50 @@ }, "tokenizers": { "hashes": [ - "sha256:0f30e6a9ea56cc8d999b17142f121ddb24b005b9fce936e7df47cdff1f55edcc", - "sha256:219dafb24d26a10bd4162372901ac886254e1c6bf6510dde85df282469231eaf", - "sha256:27ec3543e6e5701586804b1536cf687d44f8167871bd84ab01e5700cbfa969ec", - "sha256:2d875e019e39bce56c35a126ef4338345d97bf5e53730c68d0f7128a05c3d6c0", - "sha256:3100ed9051558a3b2a9c2b7dda6880328a7c4c0736be72d279b0029339d55789", - "sha256:3f199b1042d56837b5764c212874510f8eab7702d7486fdb626432465717e4f7", - "sha256:424d8ab61b0b0afadf4d88beded036cb801e3a4ddba6804621568cefd5ddb317", - "sha256:5b35be31ca999f9fb4732d947fa629104a29af3633e73cf8311cb69948f80165", - "sha256:a8c197a0d8f8b91cc220315e1617b942d992ae70bade504838b70e49bf5e4454", - "sha256:b9c5c304bfdba4e703f8828bf4c57d3d65b3c77d2732f5ec81044abe8287f2de", - "sha256:bacaeb3621055aba9b5158a21dc8b5d80f65724ff47f414f9d06f1e087668d61", - "sha256:be639a8933206dfd6e1cab26683ce020c03c6a3757d4a1dd5c8854d2bbed09ae", - "sha256:e2d709d028fe87dc305c767acca2da0425029fa0ae2135a2139ead7de62bc276", - "sha256:e76f84123695ccfaef734912369edbcb8ed71723196987ffd086e07d00c9d73a", - "sha256:ea1db6a8ebdf29f0487212c35357574c70f72c7c311406603fabeaa44cc94a12", - "sha256:ef6208dfdd2536566d0eee06a03da1a74b9c9246ef5c4a213d5afd00f51d0976", - "sha256:f71161ed25ab649e9291139f7ce6a28cd520de5de9246e2907a19a6f329b9105" + "sha256:06e1a1c50c7600d8162d8f0eeed460ad9e9234ffee7d5c7bcd1308024d781647", + "sha256:082de5272363aee13f36641065a3dd2d78f5b51486e3ab7d6d34138905a46303", + "sha256:1313d63ce286c6c9812a51ea39ae84cf1b8f2887c8ce8cc813459fdfbf526c9b", + "sha256:15440ba1db7c7b3eb7b5881b276555e25420ce14639926585837b7b60ddb55a8", + "sha256:1764a705be63fb61abcaa96637399f124528f9a01925c88efb438aefe315b61b", + "sha256:2479ef9a30fe8a961cb49c8bf6a5c5e2ce8e1b87849374c9756f41cf06189bdf", + "sha256:2dd1156815cf2ca2a0942c8efc72e0725b6cd4640a61e026c72bf5a330f4383a", + "sha256:31184c4691aed1e84088d7a18c1000bbc59f7bedeec95774ec4027129ea16272", + "sha256:3cf5b470b2e06aadee22771740d87a706216385f881308c70cb317476ec40904", + "sha256:3ea3038008f1f74c8a1e1e2e73728690eed2d7fa4db0a51bcea391e644672426", + "sha256:3ea6d65a32c8b3236553e489573f42855af484d24bf96ab32a5d6d1a2c4b0ed0", + "sha256:427257e78b71e9310d0c035df9b054525d1da91cc46efbae95fee2d523b88eb9", + "sha256:4a5ddd6689e18b6c5398b97134e79e948e1bbe7664f6962aa63f50fb05cae091", + "sha256:4fd1a765af0a7aff7dab58d7fcd63a2e4a860e829b931bdfd59e2c56ba1769b9", + "sha256:53395c4423e8309b208f1e973337c08a3cb68af5eb9dee8d8618428fd4579803", + "sha256:535cf3edfd0df2c1887ea388691dd8f614331f47b41cb40c0901a2ce070ff7e0", + "sha256:543dcb31b8534cf3ad66817f925f50f4ccd182ed1433fcd07adaed5d389f682b", + "sha256:58e1904c3e75e37be379ee4b29b21b05189d54bfab0260b334cff6e5a44a4f45", + "sha256:768f36e743604f567f4e4817a76738ed1bcdaecfef5ae8c74bdf2277a7a1902d", + "sha256:800917d7085245db0b55f88b2a12bd0ba4eb5966e8b88bd9f21aa46aadfa8204", + "sha256:807f321731a3466b9e0230cbc8e6d9c5581d5ac6536d96360b5fe1ec457d837f", + "sha256:89f816e5aa61c464e9d82025f2c4f1f66cd92f648ab9194a154ba2b0e180dc70", + "sha256:8d8ca7daa2f2274ec9327961ac828c20fcadd76e88d07f611742f240a6c73abe", + "sha256:96879e21be25b63fb99fa7d65b50b05c2a0333f104ca003917df7433d6eb073e", + "sha256:9de00f951fa8c1cf5c54a5a813447c9bf810759822de6ba6cfa42d7f503ff799", + "sha256:9f79b57a4d6a1aa8379a931e8ee54cb155cc3f5f1ba5172bcdea504dbd4cb746", + "sha256:a03c101d8058c851a7647cc74c68d4db511d7a3db8a73f7ec715e4fe14281ed7", + "sha256:a3180c8a1cb77eca8fe9c291e0f197aee202c93ffdea4f96d06ca154f319980c", + "sha256:abdbd169738c33e2e643e7701230f43c2f4e6e03d49283d4250f19159f6a6c71", + "sha256:ac4c0a2f052a83146c6475dc22f9eb740d352b29779ac6036459f00d897025b8", + "sha256:b49f17c2ac2bf88875a74d63e8070fd5a69e8c3b2874dee47649826b603a3af1", + "sha256:b57fc7f2003f1f7b873dcffd5d0ee7c71f01709c54c36f4d191e4a7911d49565", + "sha256:bce664d24c744387760beab14cc7bd4e405bbef93c333ba3ca4a93347949c3ba", + "sha256:bd46747f5c7d6e1721234d5ec1c0038bcfe0050c147c92171c3ef5b36d6fb2a9", + "sha256:c496748853c0300b8b7be916e130f0de8224575ee72e8889405477f120bfe575", + "sha256:c60b8ba2d8a948bb40c39223a4b2553c7c1df9f732b0077722b91df5d63c5e37", + "sha256:c83f7a26d6f0c765906440c7f2b726cbd18e5c7a63e0364095600c91e2905cc4", + "sha256:d2824dedd9f26e3757159d99c743b287ebf78775ccf4a36a3e0ec7058ee66303", + "sha256:d518ef8323690cd4d51979ff2f44edbac5862db8c8af125e815e41cf4517c638", + "sha256:da361a88b21cd141441fb139d1ee05c815103d49d10b49bfb4218a240d0d5a84", + "sha256:f3351eef9187ba7b9ceb04ff74fcda535f26c4146fe40155c6ed6087302944fd" ], "index": "pypi", - "version": "==0.9.3" + "version": "==0.9.4" }, "toml": { "hashes": [ @@ -439,11 +438,11 @@ }, "urllib3": { "hashes": [ - "sha256:8d7eaa5a82a1cac232164990f04874c594c9453ec55eef02eab885aa02fc17a2", - "sha256:f5321fbe4bf3fefa0efd0bfe7fb14e90909eb62a48ccda331726b4319897dd5e" + "sha256:097116a6f16f13482d2a2e56792088b9b2920f4eb6b4f84a2c90555fb673db74", + "sha256:61ad24434555a42c0439770462df38b47d05d9e8e353d93ec3742900975e3e65" ], "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4' and python_version < '4'", - "version": "==1.25.11" + "version": "==1.26.1" }, "wasabi": { "hashes": [ @@ -494,10 +493,10 @@ }, "certifi": { "hashes": [ - "sha256:5930595817496dd21bb8dc35dad090f1c2cd0adfaf21204bf6732ca5d8ee34d3", - "sha256:8fc0819f1f30ba15bdb34cceffb9ef04d99f420f68eb75d901e9560b8749fc41" + "sha256:1f422849db327d534e3d0c5f02a263458c3955ec0aae4ff09b95f195c59f4edd", + "sha256:f05def092c44fbf25834a51509ef6e631dc19765ab8a57b4e7ab85531f0a9cf4" ], - "version": "==2020.6.20" + "version": "==2020.11.8" }, "cffi": { "hashes": [ @@ -600,11 +599,11 @@ }, "elasticsearch": { "hashes": [ - "sha256:5e08776fbb30c6e92408c7fa8c37d939210d291475ae2f364f0497975918b6fe", - "sha256:8c7e2374f53ee1b891ff2804116e0c7fb517585d6d5788ba668686bbc9d82e2d" + "sha256:9053ca99bc9db84f5d80e124a79a32dfa0f7079b2112b546a03241c0dbeda36d", + "sha256:9a21bfa7dc6a0b0dc142088bd653d8ce5ab284b4f7a3ded716185adf5276a7fe" ], - "index": "pypi", - "version": "==7.9.1" + "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3' and python_version < '4'", + "version": "==7.10.0" }, "ftfy": { "hashes": [ @@ -680,19 +679,19 @@ }, "jeepney": { "hashes": [ - "sha256:3479b861cc2b6407de5188695fa1a8d57e5072d7059322469b62628869b8e36e", - "sha256:d6c6b49683446d2407d2fe3acb7a368a77ff063f9182fe427da15d622adc24cf" + "sha256:2531d17ccfb3485d4eaee03c1a19a75f28b3ac0fbb5a1b683b77b820e5b0f509", + "sha256:e0e057fe2069a54257de32eb26cf14aac5fa90f5836f49926009a5022fb1e31a" ], "markers": "sys_platform == 'linux'", - "version": "==0.4.3" + "version": "==0.5.0" }, "keyring": { "hashes": [ - "sha256:4e34ea2fdec90c1c43d6610b5a5fafa1b9097db1802948e90caf5763974b8f8d", - "sha256:9aeadd006a852b78f4b4ef7c7556c2774d2432bbef8ee538a3e9089ac8b11466" + "sha256:12de23258a95f3b13e5b167f7a641a878e91eab8ef16fafc077720a95e6115bb", + "sha256:207bd66f2a9881c835dad653da04e196c678bf104f8252141d2d3c4f31051579" ], "markers": "python_version >= '3.6'", - "version": "==21.4.0" + "version": "==21.5.0" }, "mypy": { "hashes": [ @@ -823,7 +822,7 @@ "sha256:f83f679e727742b0c465e7ef992d6da4a7e5268b8edd8fdaf5303276374bef52", "sha256:fc21a37ff3f545de80b166e1735c4172b41b017948a3fb2d5e2f03c219eac50a" ], - "index": "pypi", + "markers": "python_version >= '3.6'", "version": "==1.7.2" }, "pygments": { @@ -859,11 +858,11 @@ }, "requests": { "hashes": [ - "sha256:b3559a131db72c33ee969480840fff4bb6dd111de7dd27c8ee1f820f4f00231b", - "sha256:fe75cc94a9443b9246fc7049224f75604b113c36acb93f87b80ed42c44cbb898" + "sha256:7f1a0b932f4a60a1a65caa4263921bb7d9ee911957e0ae4a23a6dd08185ad5f8", + "sha256:e786fa28d8c9154e6a4de5d46a1d921b8749f8b74e28bde23768e5e16eece998" ], "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4'", - "version": "==2.24.0" + "version": "==2.25.0" }, "requests-toolbelt": { "hashes": [ @@ -881,11 +880,11 @@ }, "secretstorage": { "hashes": [ - "sha256:15da8a989b65498e29be338b3b279965f1b8f09b9668bd8010da183024c8bff6", - "sha256:b5ec909dde94d4ae2fa26af7c089036997030f0cf0a5cb372b4cccabd81c143b" + "sha256:46305c3847ee3f7252b284e0eee5590fa6341c891104a2fd2313f8798c615a82", + "sha256:ed5279d788af258e4676fa26b6efb6d335a31f1f9f529b6f1e200f388fac33e1" ], "markers": "sys_platform == 'linux'", - "version": "==3.1.2" + "version": "==3.2.0" }, "six": { "hashes": [ @@ -979,11 +978,11 @@ }, "urllib3": { "hashes": [ - "sha256:8d7eaa5a82a1cac232164990f04874c594c9453ec55eef02eab885aa02fc17a2", - "sha256:f5321fbe4bf3fefa0efd0bfe7fb14e90909eb62a48ccda331726b4319897dd5e" + "sha256:097116a6f16f13482d2a2e56792088b9b2920f4eb6b4f84a2c90555fb673db74", + "sha256:61ad24434555a42c0439770462df38b47d05d9e8e353d93ec3742900975e3e65" ], "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4' and python_version < '4'", - "version": "==1.25.11" + "version": "==1.26.1" }, "wcwidth": { "hashes": [ diff --git a/fuzzycat/build.py b/fuzzycat/build.py index e49e7d7..49a061f 100644 --- a/fuzzycat/build.py +++ b/fuzzycat/build.py @@ -3,13 +3,13 @@ WIP: Build auxiliary data structures for lookup. """ import fileinput +import json import operator import sqlite3 import string import sys import tempfile -import orjson as json from nltk import word_tokenize from nltk.corpus import stopwords diff --git a/fuzzycat/cluster.py b/fuzzycat/cluster.py index 2005c2f..dd55a24 100644 --- a/fuzzycat/cluster.py +++ b/fuzzycat/cluster.py @@ -69,10 +69,10 @@ import string import subprocess import sys import tempfile +from dataclasses import dataclass, field from typing import Any, Callable, Dict, Generator, List, Optional, Tuple import fuzzy -from dataclasses import dataclass, field __all__ = [ "release_key_title", @@ -95,6 +95,7 @@ class Contrib: surname: Optional[str] role: Optional[str] + @dataclass class KeyDoc: """ @@ -104,6 +105,7 @@ class KeyDoc: title: str contribs: List[Contrib] = field(default_factory=list) + @dataclass class ClusterResult: """ diff --git a/fuzzycat/main.py b/fuzzycat/main.py index 6086798..900d5c0 100644 --- a/fuzzycat/main.py +++ b/fuzzycat/main.py @@ -12,16 +12,18 @@ Run, e.g. fuzzycat cluster --help for more options. Example: import argparse import cProfile as profile +import fileinput +import json import io import logging import pstats import sys import tempfile -import fileinput from fuzzycat.build import NgramLookup, TitleTokenList -from fuzzycat.cluster import (Cluster, release_key_title, release_key_title_normalized, - release_key_title_nysiis, release_key_title_ngram) +from fuzzycat.cluster import (Cluster, release_key_title, release_key_title_ngram, + release_key_title_normalized, release_key_title_nysiis) + def run_cluster(args): logger = logging.getLogger('main.run_cluster') @@ -44,8 +46,7 @@ def run_verify(args): TODO. Ok, we should not fetch data we have on disk (at the clustering step). """ - for line in fileinput.input(files=args.files): - pass + pass def run_build(args): |