add pytest-cov

3 files changed, 261 insertions, 97 deletions

diff --git a/Pipfile b/Pipfile
index 549138f..b7fb926 100644
--- a/Pipfile
+++ b/Pipfile
@@ -22,6 +22,7 @@ tokenizers = "*"
 spacy = "*"
 nltk = "*"
 regex = "*"
+pytest-cov = "*"
 
 [requires]
 python_version = "3.7"
diff --git a/Pipfile.lock b/Pipfile.lock
index 9c5ef07..f193858 100644
--- a/Pipfile.lock
+++ b/Pipfile.lock
@@ -1,7 +1,7 @@
 {
     "_meta": {
         "hash": {
-            "sha256": "34665b414f8131dd88f22fdf14394d2c9638fa85f3c9079dfbf416be1467e1c9"
+            "sha256": "28f8a7b7cea3c6f503ca9b2fe15865d5ee721d1a89da999ff1d9aecc27e492a7"
         },
         "pipfile-spec": 6,
         "requires": {
@@ -16,26 +16,32 @@
         ]
     },
     "default": {
+        "attrs": {
+            "hashes": [
+                "sha256:31b2eced602aa8423c2aea9c76a724617ed67cf9513173fd3a4f03e3a929c7e6",
+                "sha256:832aa3cde19744e49938b91fea06d69ecb9e649c93ba974535d08ad92164f700"
+            ],
+            "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'",
+            "version": "==20.3.0"
+        },
         "blis": {
             "hashes": [
-                "sha256:00473602629ba69fe6565108e21957e918cb48b59f5bf2f6bfb6e04de42500cb",
-                "sha256:03c368c9716ca814c436550a5f1e02ccf74850e613602519e3941d212e5aa177",
-                "sha256:135450caabc8aea9bb9250329ebdf7189982d9b57d5c92789b2ba2fe52c247a7",
-                "sha256:1402d9cbb0fbc21b749dd5b87d7ee14249e74a0ca38be6ecc56b3b356fca2f21",
-                "sha256:26b16d6005bb2671699831b5cc699905215d1abde1ec5c1d04de7dcd9eb29f75",
-                "sha256:3347a4b1b7d3ae14476aac9a6f7bf8ebf464863f4ebf4aea228874a7694ea240",
-                "sha256:38fe877a4b52e762f5e137a412e3c256545a696a12ae8c40d67b8815d2bb5097",
-                "sha256:4fb89c47ee06b58a4410a16fd5794847517262c9d2a342643475b477dfeff0a4",
-                "sha256:77a6486b9794af01bcdfd1bc6e067c93add4b93292e6f95bf6e5ce7f98bf0163",
-                "sha256:856142a11e37fd2c47c5006a3197e157bb8469a491a73d2d442223dd3279df84",
-                "sha256:8aeaf6954351593a1e412f80e398aa51df588d3c0de74b9f3323b694c603381b",
-                "sha256:9ede123065f3cacb109967755b3d83d4ca0de90643a9058129a6ab2d4051954f",
-                "sha256:d1d59faebc1c94f8f4f77154ef4b9d6d40364b111cf8fde48ee3b524c85f1075",
-                "sha256:d69257d317e86f34a7f230a2fd1f021fd2a1b944137f40d8cdbb23bd334cd0c4",
-                "sha256:ddd732c5274d1082fa92e2c42317587d5ebabce7741ca98120f69bd45d004b99",
-                "sha256:f0b0dad4d6268d9dba0a65a9db12dd7a2d8686b648399e4aa1aec7550697e99e"
-            ],
-            "version": "==0.4.1"
+                "sha256:19557b14763253ca3d4f6cfc9c9fe2eed3d65db14fa273ced8b0c17ce2bfda4a",
+                "sha256:221eef1b351063b0eb6682c7af2916884ce6c5015d95f6a28b7e45bea35e493c",
+                "sha256:3917491f366467a7339997dcfe501a282409cc50f461c6e1491d342a424442f1",
+                "sha256:55bacf2a1eb72ae728b665ac46433bcfee670de84f9bdf99987adf836fd92309",
+                "sha256:59cc549e904fd4169105f92d6fac94495eca228c357232f29444ae79981af974",
+                "sha256:6e65ef84205fadc1c41e59bd9d0d67fca8c41d463761b2dfe447003840172062",
+                "sha256:7f3e9564023e59a0abe9642a94408b3d2fe05ff205d800d512a67828b61819a8",
+                "sha256:9130e194f01124975a7bd76b8b65550f1f308610d596ec1992280b656e0c11be",
+                "sha256:ae53b880a1bca8115f1edccfa6e8800661c71347f9b75ed3f1af8549f080dd6b",
+                "sha256:b0fa63e4ca21a189698ca764367ae91ccba2274d885563c81205c1ac09f36fa6",
+                "sha256:c04165d196e16f652d21c2d1403a55fbae8721d74dbca1668fdfec6766687a12",
+                "sha256:ca0bccd31a216931ada487264ad07fa64a0b5a42b51c80cdebb1d3764b813d3b",
+                "sha256:e4e8b0dfd32e12f6d380b77a916ef15793756b856d8b00453a7d0ed4ba29f9ea"
+            ],
+            "markers": "python_version >= '3.6'",
+            "version": "==0.7.3"
         },
         "catalogue": {
             "hashes": [
@@ -67,6 +73,46 @@
             "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4'",
             "version": "==7.1.2"
         },
+        "coverage": {
+            "hashes": [
+                "sha256:0203acd33d2298e19b57451ebb0bed0ab0c602e5cf5a818591b4918b1f97d516",
+                "sha256:0f313707cdecd5cd3e217fc68c78a960b616604b559e9ea60cc16795c4304259",
+                "sha256:1c6703094c81fa55b816f5ae542c6ffc625fec769f22b053adb42ad712d086c9",
+                "sha256:1d44bb3a652fed01f1f2c10d5477956116e9b391320c94d36c6bf13b088a1097",
+                "sha256:280baa8ec489c4f542f8940f9c4c2181f0306a8ee1a54eceba071a449fb870a0",
+                "sha256:29a6272fec10623fcbe158fdf9abc7a5fa032048ac1d8631f14b50fbfc10d17f",
+                "sha256:2b31f46bf7b31e6aa690d4c7a3d51bb262438c6dcb0d528adde446531d0d3bb7",
+                "sha256:2d43af2be93ffbad25dd959899b5b809618a496926146ce98ee0b23683f8c51c",
+                "sha256:381ead10b9b9af5f64646cd27107fb27b614ee7040bb1226f9c07ba96625cbb5",
+                "sha256:47a11bdbd8ada9b7ee628596f9d97fbd3851bd9999d398e9436bd67376dbece7",
+                "sha256:4d6a42744139a7fa5b46a264874a781e8694bb32f1d76d8137b68138686f1729",
+                "sha256:50691e744714856f03a86df3e2bff847c2acede4c191f9a1da38f088df342978",
+                "sha256:530cc8aaf11cc2ac7430f3614b04645662ef20c348dce4167c22d99bec3480e9",
+                "sha256:582ddfbe712025448206a5bc45855d16c2e491c2dd102ee9a2841418ac1c629f",
+                "sha256:63808c30b41f3bbf65e29f7280bf793c79f54fb807057de7e5238ffc7cc4d7b9",
+                "sha256:71b69bd716698fa62cd97137d6f2fdf49f534decb23a2c6fc80813e8b7be6822",
+                "sha256:7858847f2d84bf6e64c7f66498e851c54de8ea06a6f96a32a1d192d846734418",
+                "sha256:78e93cc3571fd928a39c0b26767c986188a4118edc67bc0695bc7a284da22e82",
+                "sha256:7f43286f13d91a34fadf61ae252a51a130223c52bfefb50310d5b2deb062cf0f",
+                "sha256:86e9f8cd4b0cdd57b4ae71a9c186717daa4c5a99f3238a8723f416256e0b064d",
+                "sha256:8f264ba2701b8c9f815b272ad568d555ef98dfe1576802ab3149c3629a9f2221",
+                "sha256:9342dd70a1e151684727c9c91ea003b2fb33523bf19385d4554f7897ca0141d4",
+                "sha256:9361de40701666b034c59ad9e317bae95c973b9ff92513dd0eced11c6adf2e21",
+                "sha256:9669179786254a2e7e57f0ecf224e978471491d660aaca833f845b72a2df3709",
+                "sha256:aac1ba0a253e17889550ddb1b60a2063f7474155465577caa2a3b131224cfd54",
+                "sha256:aef72eae10b5e3116bac6957de1df4d75909fc76d1499a53fb6387434b6bcd8d",
+                "sha256:bd3166bb3b111e76a4f8e2980fa1addf2920a4ca9b2b8ca36a3bc3dedc618270",
+                "sha256:c1b78fb9700fc961f53386ad2fd86d87091e06ede5d118b8a50dea285a071c24",
+                "sha256:c3888a051226e676e383de03bf49eb633cd39fc829516e5334e69b8d81aae751",
+                "sha256:c5f17ad25d2c1286436761b462e22b5020d83316f8e8fcb5deb2b3151f8f1d3a",
+                "sha256:c851b35fc078389bc16b915a0a7c1d5923e12e2c5aeec58c52f4aa8085ac8237",
+                "sha256:cb7df71de0af56000115eafd000b867d1261f786b5eebd88a0ca6360cccfaca7",
+                "sha256:cedb2f9e1f990918ea061f28a0f0077a07702e3819602d3507e2ff98c8d20636",
+                "sha256:e8caf961e1b1a945db76f1b5fa9c91498d15f545ac0ababbe575cfab185d3bd8"
+            ],
+            "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": "==5.3"
+        },
         "cymem": {
             "hashes": [
                 "sha256:02c3994ee57003dd57d9a34417beb9597e51176e9378a757344347b72510ce8d",
@@ -127,6 +173,13 @@
             "index": "pypi",
             "version": "==1.7.0"
         },
+        "iniconfig": {
+            "hashes": [
+                "sha256:011e24c64b7f47f6ebd835bb12a743f2fbe9a26d4cecaa7f53bc4f35ee9da8b3",
+                "sha256:bc3af051d7d14b2ee5ef9969666def0cd1a000e121eaea580d4a313df4b37f32"
+            ],
+            "version": "==1.1.1"
+        },
         "joblib": {
             "hashes": [
                 "sha256:698c311779f347cf6b7e6b8a39bb682277b8ee4aba8cf9507bc0cf4cd4737b72",
@@ -204,6 +257,14 @@
             "markers": "python_version >= '3.6'",
             "version": "==1.19.4"
         },
+        "packaging": {
+            "hashes": [
+                "sha256:4357f74f47b9c12db93624a82154e9b120fa8293699949152b22065d556079f8",
+                "sha256:998416ba6962ae7fbd6596850b80e17859a5753ba17c32284f67bfff33784181"
+            ],
+            "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'",
+            "version": "==20.4"
+        },
         "plac": {
             "hashes": [
                 "sha256:398cb947c60c4c25e275e1f1dadf027e7096858fb260b8ece3b33bcff90d985f",
@@ -211,6 +272,14 @@
             ],
             "version": "==1.1.3"
         },
+        "pluggy": {
+            "hashes": [
+                "sha256:15b2acde666561e1298d71b523007ed7364de07029219b604cf808bfa1c765b0",
+                "sha256:966c145cd83c96502c3c3868f50408687b38434af77734af1e9ca461a4081d2d"
+            ],
+            "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'",
+            "version": "==0.13.1"
+        },
         "preshed": {
             "hashes": [
                 "sha256:13a779205d55ce323976ac06df597f9ec2d6f0563ebcf5652176cf4520c7d540",
@@ -233,6 +302,14 @@
             ],
             "version": "==3.0.4"
         },
+        "py": {
+            "hashes": [
+                "sha256:366389d1db726cd2fcfc79732e75410e5fe4d31db13692115529d34069a043c2",
+                "sha256:9ca6883ce56b4e8da7e79ac18787889fa5206c79dcc67fb065376cd2fe03f342"
+            ],
+            "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'",
+            "version": "==1.9.0"
+        },
         "pydantic": {
             "hashes": [
                 "sha256:01f0291f4951580f320f7ae3f2ecaf0044cdebcc9b45c5f882a7e84453362420",
@@ -261,6 +338,30 @@
             "markers": "python_version >= '3.6'",
             "version": "==1.7.2"
         },
+        "pyparsing": {
+            "hashes": [
+                "sha256:c203ec8783bf771a155b207279b9bccb8dea02d8f0c9e5f8ead507bc3246ecc1",
+                "sha256:ef9d7589ef3c200abe66653d3f1ab1033c3c419ae9b9bdb1240a85b024efc88b"
+            ],
+            "markers": "python_version >= '2.6' and python_version not in '3.0, 3.1, 3.2, 3.3'",
+            "version": "==2.4.7"
+        },
+        "pytest": {
+            "hashes": [
+                "sha256:4288fed0d9153d9646bfcdf0c0428197dba1ecb27a33bb6e031d002fa88653fe",
+                "sha256:c0a7e94a8cdbc5422a51ccdad8e6f1024795939cc89159a0ae7f0b316ad3823e"
+            ],
+            "markers": "python_version >= '3.5'",
+            "version": "==6.1.2"
+        },
+        "pytest-cov": {
+            "hashes": [
+                "sha256:45ec2d5182f89a81fc3eb29e3d1ed3113b9e9a873bcddb2a71faaab066110191",
+                "sha256:47bd0ce14056fdd79f93e1713f88fad7bdcc583dcd7783da86ef2f085a0bb88e"
+            ],
+            "index": "pypi",
+            "version": "==2.10.1"
+        },
         "regex": {
             "hashes": [
                 "sha256:02951b7dacb123d8ea6da44fe45ddd084aa6777d4b2454fa0da61d569c6fa538",
@@ -316,21 +417,32 @@
             "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4'",
             "version": "==2.25.0"
         },
+        "six": {
+            "hashes": [
+                "sha256:30639c035cdb23534cd4aa2dd52c3bf48f06e5f4a941509c8bafd8ce11080259",
+                "sha256:8b74bedcbbbaca38ff6d7491d76f2b06b3592611af620f8426e82dddb04a5ced"
+            ],
+            "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'",
+            "version": "==1.15.0"
+        },
         "spacy": {
             "hashes": [
-                "sha256:0f5d088c1d2a1fcf247090854927cd0ba4e28266323af112dead20ff020ded1c",
-                "sha256:11b9517cdcbea166a9461093821d12bf632aea7dd14b6e3c549871903bda41b8",
-                "sha256:1fcfb911b254af3144b3e65a2daf671cb26b6243ec431089ccb28cbe03d826de",
-                "sha256:366eaae9634c59f89015ad11db1d8559c327ab665a5f644c71155c76711ee50a",
-                "sha256:3bafcc134c340c5d7556612344d2844522d452b99a21f2b0a9b640f6c55f1110",
-                "sha256:4944a1118f6dbb49201749d72527b749f74032e1026ddf387bc3a7e172ff0300",
-                "sha256:7a6b7486f71930e7de7100feb72036e3ccb8c18509ff23e8453cff0b28470ea4",
-                "sha256:818de26e0e383f64ccbe3db185574920de05923d8deac8bbb12113b9e33cee1f",
-                "sha256:b7df3622e9a867294b913cd0a4fba99d47162af1cfd3a840c5943b25f390bb5c",
-                "sha256:f7b3a17730786979f964b16ee1e4a9146cd05016f100afb274dd66336dfc39eb"
+                "sha256:070ae51898fb1ab34b50dd37bf9767dcebb92662c523d5d5ea7a2246b752c9cf",
+                "sha256:0b432ef51b9e230015b735f1aac4233da36086de015e49fe87a29ce344f5d128",
+                "sha256:1b178f0c1ce42a92168c2eba16b1337ac54a709537eab3a4351698d0bfce4203",
+                "sha256:3d88f7504b9c8021796d30618eb9fa3a3185535af352ce982152e3aeb5ef73a3",
+                "sha256:45681c37579dae6040a485bf434e27398180fdfa2fa4ba21d25c6bf3337dba7f",
+                "sha256:799fa5fc172ff0a5bc8eb5dfcd1db200747c114320d2dc40060594a71efa3e53",
+                "sha256:7e7cba1077b021adb5d309cf77756e3421bf6964f652bdf89e86cf1b43b250b0",
+                "sha256:8cb47a526a7e26eda548d17ffc498e671bb6b3eb70a2de8f88b42a41eed81c8d",
+                "sha256:b2f60e406bbc086a6e8376bccd9ed4e217a06037a3b52ad5efd5a3deff164865",
+                "sha256:c4ff005113bffbf7b2841c0e09bd520688b858ea2010209c551d5726b1034b7e",
+                "sha256:e96fc3e27c42b3ce5e7b114ab060ade83fc3760a86b03d407bb999a6133fb2ae",
+                "sha256:f1b37866c6fff23ccf49e56448ad927c0e14070ce1c0f3f55f694c6d51b40e94",
+                "sha256:f4989a66f12e096aa7d3145010227e347dbd28c7233d9aa842e3450c9ca665ed"
             ],
             "index": "pypi",
-            "version": "==2.3.2"
+            "version": "==2.3.3"
         },
         "srsly": {
             "hashes": [
@@ -352,20 +464,21 @@
         },
         "thinc": {
             "hashes": [
-                "sha256:0139fa84dc9b8d88af15e648fc4ae13d899b8b5e49cb26a8f4a0604ee9ad8a9e",
-                "sha256:061633bf334e3728173d59d6001e8cdef3839166c71e23b3c5f74f5fae3c0d7c",
-                "sha256:0df8c5762359a3a4d8d494aa2eff11c4936c4f34559fe1b3ab1d13d24c76b509",
-                "sha256:33db4a9182c78c8f4823b1765274bbb0caa8f4269dbd102f2e6ab2f7f91a6084",
-                "sha256:36237f711f0b3da932bd28cc366a92f6f1b6d1f95ad6cbbc8166b94785b38e40",
-                "sha256:387d25e57e53eed86d24f2657ab9555703043de27211764835a38e2e31b3c8e9",
-                "sha256:55b9e02e4b8395cee0a8a810bd8af4d7600b04520bab60df1fc513d50a41eec5",
-                "sha256:5d633cc5c210a02ba706ed7e800f4dc906ba1e10b85e3ed40d77fdb7e7674a20",
-                "sha256:947806f4cbbcaf8dd046942acd5e52d55ac805303985a2e36de4734be5496bf1",
-                "sha256:d3ff8cfbf583ac788a85f5e0e3cf00edf2f6bc5ba2b2ca264771870c07cb5717",
-                "sha256:d70e71b0561bbf844bc9f737f60150b0f8f04dfd603151869d93a5735deb6219",
-                "sha256:e2ebeeafd79bb86697388fccc5996d6ea1e69106e2a7fc3a1092d626b522cc01"
-            ],
-            "version": "==7.4.1"
+                "sha256:16002dc0feb941d4fc985cee90a0af8081190d098247d4e7f6db7d3da1ae0222",
+                "sha256:187b8245124305b0a709a7f7c79775742c270d6c176a3879afecdaadcfb92fb4",
+                "sha256:1e516f78d63ab710fcea49f0eed1c8068f2ca29ce5c6323a2a23fb8140fceb45",
+                "sha256:241b8c9e9e068d1728e1fb97d95bfc4a956a8eebb40c63ecbe68aad416dac26c",
+                "sha256:2ea3fd503974078a24a9f3d07c49d2adf4f0252530780eda38de3bf6517237e2",
+                "sha256:312bbb18f668e190b14563c2d38928d187faaef69217ea5ff4d7e0d6e2e1f21c",
+                "sha256:39405be25a5d232e4fb6e1d4830568bd826867d43878d093497c9db277f3e90f",
+                "sha256:63471847f282ca2f2b2f88d3e6fb15240f4ea0ec6931549ddceb57967deeca18",
+                "sha256:862e846033a31644be600c7f9606a965e913bcc42a45cad64a0c77c81c677bdf",
+                "sha256:9e4772c1094b2997394fe95377eb76e15c2a2f50cdb46e2351442a8a9c370146",
+                "sha256:c98491b083165f48bda95f5533f7d9dbd3980d32ad621bfe579ff08ef625a4d3",
+                "sha256:d28d5b300f7e18fa143b3efba0c9321e886fa6ec36a553fac5b32f686df6584a",
+                "sha256:e609261414e6978692c17b9ae091925edf3361b47c8f8c086390684ce604805f"
+            ],
+            "version": "==7.4.3"
         },
         "tokenizers": {
             "hashes": [
@@ -424,11 +537,11 @@
         },
         "tqdm": {
             "hashes": [
-                "sha256:9ad44aaf0fc3697c06f6e05c7cf025dd66bc7bcb7613c66d85f4464c47ac8fad",
-                "sha256:ef54779f1c09f346b2b5a8e5c61f96fbcb639929e640e59f8cf810794f406432"
+                "sha256:3d3f1470d26642e88bd3f73353cb6ff4c51ef7d5d7efef763238f4bc1f7e4e81",
+                "sha256:5ff3f5232b19fa4c5531641e480b7fad4598819f708a32eb815e6ea41c5fa313"
             ],
-            "markers": "python_version >= '2.6' and python_version not in '3.0, 3.1, 3.2, 3.3'",
-            "version": "==4.51.0"
+            "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'",
+            "version": "==4.53.0"
         },
         "unidecode": {
             "hashes": [
@@ -501,44 +614,42 @@
         },
         "cffi": {
             "hashes": [
-                "sha256:005f2bfe11b6745d726dbb07ace4d53f057de66e336ff92d61b8c7e9c8f4777d",
-                "sha256:09e96138280241bd355cd585148dec04dbbedb4f46128f340d696eaafc82dd7b",
-                "sha256:0b1ad452cc824665ddc682400b62c9e4f5b64736a2ba99110712fdee5f2505c4",
-                "sha256:0ef488305fdce2580c8b2708f22d7785ae222d9825d3094ab073e22e93dfe51f",
-                "sha256:15f351bed09897fbda218e4db5a3d5c06328862f6198d4fb385f3e14e19decb3",
-                "sha256:22399ff4870fb4c7ef19fff6eeb20a8bbf15571913c181c78cb361024d574579",
-                "sha256:23e5d2040367322824605bc29ae8ee9175200b92cb5483ac7d466927a9b3d537",
-                "sha256:2791f68edc5749024b4722500e86303a10d342527e1e3bcac47f35fbd25b764e",
-                "sha256:2f9674623ca39c9ebe38afa3da402e9326c245f0f5ceff0623dccdac15023e05",
-                "sha256:3363e77a6176afb8823b6e06db78c46dbc4c7813b00a41300a4873b6ba63b171",
-                "sha256:33c6cdc071ba5cd6d96769c8969a0531be2d08c2628a0143a10a7dcffa9719ca",
-                "sha256:3b8eaf915ddc0709779889c472e553f0d3e8b7bdf62dab764c8921b09bf94522",
-                "sha256:3cb3e1b9ec43256c4e0f8d2837267a70b0e1ca8c4f456685508ae6106b1f504c",
-                "sha256:3eeeb0405fd145e714f7633a5173318bd88d8bbfc3dd0a5751f8c4f70ae629bc",
-                "sha256:44f60519595eaca110f248e5017363d751b12782a6f2bd6a7041cba275215f5d",
-                "sha256:4d7c26bfc1ea9f92084a1d75e11999e97b62d63128bcc90c3624d07813c52808",
-                "sha256:529c4ed2e10437c205f38f3691a68be66c39197d01062618c55f74294a4a4828",
-                "sha256:6642f15ad963b5092d65aed022d033c77763515fdc07095208f15d3563003869",
-                "sha256:85ba797e1de5b48aa5a8427b6ba62cf69607c18c5d4eb747604b7302f1ec382d",
-                "sha256:8f0f1e499e4000c4c347a124fa6a27d37608ced4fe9f7d45070563b7c4c370c9",
-                "sha256:a624fae282e81ad2e4871bdb767e2c914d0539708c0f078b5b355258293c98b0",
-                "sha256:b0358e6fefc74a16f745afa366acc89f979040e0cbc4eec55ab26ad1f6a9bfbc",
-                "sha256:bbd2f4dfee1079f76943767fce837ade3087b578aeb9f69aec7857d5bf25db15",
-                "sha256:bf39a9e19ce7298f1bd6a9758fa99707e9e5b1ebe5e90f2c3913a47bc548747c",
-                "sha256:c11579638288e53fc94ad60022ff1b67865363e730ee41ad5e6f0a17188b327a",
-                "sha256:c150eaa3dadbb2b5339675b88d4573c1be3cb6f2c33a6c83387e10cc0bf05bd3",
-                "sha256:c53af463f4a40de78c58b8b2710ade243c81cbca641e34debf3396a9640d6ec1",
-                "sha256:cb763ceceae04803adcc4e2d80d611ef201c73da32d8f2722e9d0ab0c7f10768",
-                "sha256:cc75f58cdaf043fe6a7a6c04b3b5a0e694c6a9e24050967747251fb80d7bce0d",
-                "sha256:d80998ed59176e8cba74028762fbd9b9153b9afc71ea118e63bbf5d4d0f9552b",
-                "sha256:de31b5164d44ef4943db155b3e8e17929707cac1e5bd2f363e67a56e3af4af6e",
-                "sha256:e66399cf0fc07de4dce4f588fc25bfe84a6d1285cc544e67987d22663393926d",
-                "sha256:f0620511387790860b249b9241c2f13c3a80e21a73e0b861a2df24e9d6f56730",
-                "sha256:f4eae045e6ab2bb54ca279733fe4eb85f1effda392666308250714e01907f394",
-                "sha256:f92cdecb618e5fa4658aeb97d5eb3d2f47aa94ac6477c6daf0f306c5a3b9e6b1",
-                "sha256:f92f789e4f9241cd262ad7a555ca2c648a98178a953af117ef7fad46aa1d5591"
-            ],
-            "version": "==1.14.3"
+                "sha256:00a1ba5e2e95684448de9b89888ccd02c98d512064b4cb987d48f4b40aa0421e",
+                "sha256:00e28066507bfc3fe865a31f325c8391a1ac2916219340f87dfad602c3e48e5d",
+                "sha256:045d792900a75e8b1e1b0ab6787dd733a8190ffcf80e8c8ceb2fb10a29ff238a",
+                "sha256:0638c3ae1a0edfb77c6765d487fee624d2b1ee1bdfeffc1f0b58c64d149e7eec",
+                "sha256:105abaf8a6075dc96c1fe5ae7aae073f4696f2905fde6aeada4c9d2926752362",
+                "sha256:155136b51fd733fa94e1c2ea5211dcd4c8879869008fc811648f16541bf99668",
+                "sha256:1a465cbe98a7fd391d47dce4b8f7e5b921e6cd805ef421d04f5f66ba8f06086c",
+                "sha256:1d2c4994f515e5b485fd6d3a73d05526aa0fcf248eb135996b088d25dfa1865b",
+                "sha256:2c24d61263f511551f740d1a065eb0212db1dbbbbd241db758f5244281590c06",
+                "sha256:51a8b381b16ddd370178a65360ebe15fbc1c71cf6f584613a7ea08bfad946698",
+                "sha256:594234691ac0e9b770aee9fcdb8fa02c22e43e5c619456efd0d6c2bf276f3eb2",
+                "sha256:5cf4be6c304ad0b6602f5c4e90e2f59b47653ac1ed9c662ed379fe48a8f26b0c",
+                "sha256:64081b3f8f6f3c3de6191ec89d7dc6c86a8a43911f7ecb422c60e90c70be41c7",
+                "sha256:6bc25fc545a6b3d57b5f8618e59fc13d3a3a68431e8ca5fd4c13241cd70d0009",
+                "sha256:798caa2a2384b1cbe8a2a139d80734c9db54f9cc155c99d7cc92441a23871c03",
+                "sha256:7c6b1dece89874d9541fc974917b631406233ea0440d0bdfbb8e03bf39a49b3b",
+                "sha256:840793c68105fe031f34d6a086eaea153a0cd5c491cde82a74b420edd0a2b909",
+                "sha256:8d6603078baf4e11edc4168a514c5ce5b3ba6e3e9c374298cb88437957960a53",
+                "sha256:9cc46bc107224ff5b6d04369e7c595acb700c3613ad7bcf2e2012f62ece80c35",
+                "sha256:9f7a31251289b2ab6d4012f6e83e58bc3b96bd151f5b5262467f4bb6b34a7c26",
+                "sha256:9ffb888f19d54a4d4dfd4b3f29bc2c16aa4972f1c2ab9c4ab09b8ab8685b9c2b",
+                "sha256:a7711edca4dcef1a75257b50a2fbfe92a65187c47dab5a0f1b9b332c5919a3fb",
+                "sha256:af5c59122a011049aad5dd87424b8e65a80e4a6477419c0c1015f73fb5ea0293",
+                "sha256:b18e0a9ef57d2b41f5c68beefa32317d286c3d6ac0484efd10d6e07491bb95dd",
+                "sha256:b4e248d1087abf9f4c10f3c398896c87ce82a9856494a7155823eb45a892395d",
+                "sha256:ba4e9e0ae13fc41c6b23299545e5ef73055213e466bd107953e4a013a5ddd7e3",
+                "sha256:c6332685306b6417a91b1ff9fae889b3ba65c2292d64bd9245c093b1b284809d",
+                "sha256:d9efd8b7a3ef378dd61a1e77367f1924375befc2eba06168b6ebfa903a5e59ca",
+                "sha256:df5169c4396adc04f9b0a05f13c074df878b6052430e03f50e68adf3a57aa28d",
+                "sha256:ebb253464a5d0482b191274f1c8bf00e33f7e0b9c66405fbffc61ed2c839c775",
+                "sha256:ec80dc47f54e6e9a78181ce05feb71a0353854cc26999db963695f950b5fb375",
+                "sha256:f032b34669220030f905152045dfa27741ce1a6db3324a5bc0b96b6c7420c87b",
+                "sha256:f60567825f791c6f8a592f3c6e3bd93dd2934e3f9dac189308426bd76b00ef3b",
+                "sha256:f803eaa94c2fcda012c047e62bc7a51b0bdabda1cad7a92a522694ea2d76e49f"
+            ],
+            "version": "==1.14.4"
         },
         "chardet": {
             "hashes": [
@@ -680,11 +791,11 @@
         },
         "jeepney": {
             "hashes": [
-                "sha256:2531d17ccfb3485d4eaee03c1a19a75f28b3ac0fbb5a1b683b77b820e5b0f509",
-                "sha256:e0e057fe2069a54257de32eb26cf14aac5fa90f5836f49926009a5022fb1e31a"
+                "sha256:7d59b6622675ca9e993a6bd38de845051d315f8b0c72cca3aef733a20b648657",
+                "sha256:aec56c0eb1691a841795111e184e13cad504f7703b9a64f63020816afa79a8ae"
             ],
             "markers": "sys_platform == 'linux'",
-            "version": "==0.5.0"
+            "version": "==0.6.0"
         },
         "keyring": {
             "hashes": [
@@ -847,7 +958,7 @@
                 "sha256:4288fed0d9153d9646bfcdf0c0428197dba1ecb27a33bb6e031d002fa88653fe",
                 "sha256:c0a7e94a8cdbc5422a51ccdad8e6f1024795939cc89159a0ae7f0b316ad3823e"
             ],
-            "index": "pypi",
+            "markers": "python_version >= '3.5'",
             "version": "==6.1.2"
         },
         "pyyaml": {
@@ -855,11 +966,13 @@
                 "sha256:06a0d7ba600ce0b2d2fe2e78453a470b5a6e000a985dd4a4e54e436cc36b0e97",
                 "sha256:240097ff019d7c70a4922b6869d8a86407758333f02203e0fc6ff79c5dcede76",
                 "sha256:4f4b913ca1a7319b33cfb1369e91e50354d6f07a135f3b901aca02aa95940bd2",
+                "sha256:6034f55dab5fea9e53f436aa68fa3ace2634918e8b5994d82f3621c04ff5ed2e",
                 "sha256:69f00dca373f240f842b2931fb2c7e14ddbacd1397d57157a9b005a6a9942648",
                 "sha256:73f099454b799e05e5ab51423c7bcf361c58d3206fa7b0d555426b1f4d9a3eaf",
                 "sha256:74809a57b329d6cc0fdccee6318f44b9b8649961fa73144a98735b0aaf029f1f",
                 "sha256:7739fc0fa8205b3ee8808aea45e968bc90082c10aef6ea95e855e10abf4a37b2",
                 "sha256:95f71d2af0ff4227885f7a6605c37fd53d3a106fcab511b8860ecca9fcf400ee",
+                "sha256:ad9c67312c84def58f3c04504727ca879cb0013b2517c85a9a253f0cb6380c0a",
                 "sha256:b8eac752c5e14d3eca0e6dd9199cd627518cb5ec06add0de9d32baeee6fe645d",
                 "sha256:cc8955cfbfc7a115fa81d85284ee61147059a753344bc51098f3ccd69b0d7e0c",
                 "sha256:d13155f591e6fcc1ec3b30685d50bf0711574e2c0dfffd7644babf8b5102ca1a"
@@ -898,11 +1011,11 @@
         },
         "secretstorage": {
             "hashes": [
-                "sha256:46305c3847ee3f7252b284e0eee5590fa6341c891104a2fd2313f8798c615a82",
-                "sha256:ed5279d788af258e4676fa26b6efb6d335a31f1f9f529b6f1e200f388fac33e1"
+                "sha256:30cfdef28829dad64d6ea1ed08f8eff6aa115a77068926bcc9f5225d5a3246aa",
+                "sha256:5c36f6537a523ec5f969ef9fad61c98eb9e017bc601d811e53aa25bece64892f"
             ],
             "markers": "sys_platform == 'linux'",
-            "version": "==3.2.0"
+            "version": "==3.3.0"
         },
         "six": {
             "hashes": [
@@ -922,11 +1035,11 @@
         },
         "tqdm": {
             "hashes": [
-                "sha256:9ad44aaf0fc3697c06f6e05c7cf025dd66bc7bcb7613c66d85f4464c47ac8fad",
-                "sha256:ef54779f1c09f346b2b5a8e5c61f96fbcb639929e640e59f8cf810794f406432"
+                "sha256:3d3f1470d26642e88bd3f73353cb6ff4c51ef7d5d7efef763238f4bc1f7e4e81",
+                "sha256:5ff3f5232b19fa4c5531641e480b7fad4598819f708a32eb815e6ea41c5fa313"
             ],
-            "markers": "python_version >= '2.6' and python_version not in '3.0, 3.1, 3.2, 3.3'",
-            "version": "==4.51.0"
+            "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'",
+            "version": "==4.53.0"
         },
         "traitlets": {
             "hashes": [
diff --git a/setup.py b/setup.py
new file mode 100644
index 0000000..7a39369
--- /dev/null
+++ b/setup.py
@@ -0,0 +1,50 @@
+import setuptools
+
+from fuzzycat import __version__
+
+with open("README.md", "r") as fh:
+    long_description = fh.read()
+
+    setuptools.setup(
+        name="fuzzycat",
+        version=__version__,
+        author="Martin Czygan",
+        author_email="martin@archive.org",
+        description="Fuzzy matching utilities for scholarly metadata",
+        long_description=long_description,
+        long_description_content_type="text/markdown",
+        url="https://github.com/miku/fuzzycat",
+        packages=setuptools.find_packages(),
+        classifiers=[
+            "Programming Language :: Python :: 3",
+            "License :: OSI Approved :: MIT License",
+            "Operating System :: OS Independent",
+        ],
+        python_requires=">=3.5",
+        zip_safe=False,
+        entry_points={"console_scripts": [
+            "fuzzycat=fuzzycat.main:main"
+        ]},
+        install_requires=[
+            "elasticsearch>=7",
+            "ftfy",
+            "fuzzy",
+            "pydantic",
+            "toml",
+            "unidecode>=0.10",
+            # "fatcat-openapi-client",
+            # "simhash",
+        ],
+        extras_require={"dev": [
+            "ipython",
+            "isort",
+            "jupyter",
+            "matplotlib",
+            "pandas",
+            "pylint",
+            "pytest",
+            "pytest-cov",
+            "twine",
+            "yapf",
+        ],},
+    )