diff --git a/data/deps.json b/data/deps.json index 49a7cc1..90bd8f2 100644 --- a/data/deps.json +++ b/data/deps.json @@ -1,8 +1,7 @@ { "lxml": "5.3.0", "rapidfuzz": "3.10.0", - "spacy": "3.7.6", - "spacy_cpu_model": "3.7.0", - "en_spacy_cpu_model": "3.7.1", - "thinc-apple-ops": "0.1.5" + "spacy": "3.8.2", + "spacy_cpu_model": "3.8.0", + "thinc-apple-ops": "1.0.0" } diff --git a/deps.py b/deps.py index f4dfb45..53f2aba 100644 --- a/deps.py +++ b/deps.py @@ -39,10 +39,8 @@ def install_deps(pkg: str, notif: Any) -> None: else: # Install X-Ray dependencies pip_install("rapidfuzz", dep_versions["rapidfuzz"], notif=notif) - - if pkg == "": - pip_install("spacy", dep_versions["spacy"], notif=notif) - else: + pip_install("spacy", dep_versions["spacy"], notif=notif) + if pkg != "": model_version = get_spacy_model_version(pkg, dep_versions) url = ( "https://github.com/explosion/spacy-models/releases/download/" diff --git a/epub.py b/epub.py index cb767d6..d9e221b 100644 --- a/epub.py +++ b/epub.py @@ -563,7 +563,7 @@ def find_sense_ids_with_pos( """ SELECT DISTINCT s.id FROM senses s JOIN lemmas l ON s.lemma_id = l.id - WHERE lemma = ? AND pos = ? + WHERE lemma = ? AND pos = ? AND enabled = 1 """, (lemma, pos), ): @@ -580,7 +580,7 @@ def find_sense_ids_without_pos(self, word: str) -> tuple[int, ...]: """ SELECT DISTINCT s.id FROM senses s JOIN lemmas l ON s.lemma_id = l.id - WHERE lemma = ? + WHERE lemma = ? AND enabled = 1 """, (word,), ): @@ -591,7 +591,7 @@ def find_sense_ids_without_pos(self, word: str) -> tuple[int, ...]: """ SELECT DISTINCT s.id FROM senses s JOIN forms f ON s.lemma_id = f.lemma_id AND s.pos = f.pos - WHERE form = ? + WHERE form = ? AND enabled = 1 """, (word,), ): diff --git a/parse_job.py b/parse_job.py index 0b0367b..31fd03d 100644 --- a/parse_job.py +++ b/parse_job.py @@ -535,7 +535,7 @@ def get_kindle_lemma_with_pos( """ SELECT difficulty, senses.id FROM senses JOIN lemmas ON senses.lemma_id = lemmas.id - WHERE lemma = ? AND pos = ? LIMIT 1 + WHERE lemma = ? AND pos = ? AND enabled = 1 LIMIT 1 """, (lemma, pos), ): diff --git a/tests/LanguageLayer.en.AZW3.json b/tests/LanguageLayer.en.AZW3.json index 976967c..08c4059 100644 --- a/tests/LanguageLayer.en.AZW3.json +++ b/tests/LanguageLayer.en.AZW3.json @@ -33,7 +33,7 @@ [ 1797, 5, - 32801 + 32803 ], [ 1866, @@ -48,7 +48,7 @@ [ 2113, 5, - 11140 + 11141 ], [ 2130, @@ -93,12 +93,12 @@ [ 2684, 2, - 66918 + 66938 ], [ 2712, 4, - 17656 + 2700 ], [ 2776, @@ -108,12 +108,12 @@ [ 3434, 5, - 70600 + 75440 ], [ 3443, 5, - 36194 + 36195 ], [ 3494, @@ -133,12 +133,12 @@ [ 3852, 5, - 2366 + 35605 ], [ 4229, 5, - 47529 + 47532 ], [ 4262, @@ -148,17 +148,17 @@ [ 4332, 5, - 2366 + 35605 ], [ 4957, 4, - 17784 + 9243 ], [ 4989, 5, - 48552 + 48553 ], [ 5034, @@ -183,7 +183,7 @@ [ 5292, 4, - 60243 + 66676 ], [ 5366, @@ -193,22 +193,22 @@ [ 5495, 5, - 54321 + 54323 ], [ 5563, - 5, - 40142 + 4, + 78727 ], [ 5578, 5, - 22640 + 26331 ], [ 5651, 4, - 58060 + 58061 ], [ 5671, diff --git a/tests/LanguageLayer.en.KFX.json b/tests/LanguageLayer.en.KFX.json index 4b46ff0..7bd7de3 100644 --- a/tests/LanguageLayer.en.KFX.json +++ b/tests/LanguageLayer.en.KFX.json @@ -33,7 +33,7 @@ [ 341, 5, - 32801 + 32803 ], [ 410, @@ -48,7 +48,7 @@ [ 657, 5, - 11140 + 11141 ], [ 674, @@ -93,12 +93,12 @@ [ 1018, 2, - 66918 + 66938 ], [ 1046, 4, - 17656 + 2700 ], [ 1110, @@ -108,12 +108,12 @@ [ 1213, 5, - 70600 + 75440 ], [ 1222, 5, - 36194 + 36195 ], [ 1273, @@ -138,7 +138,7 @@ [ 1634, 5, - 47529 + 47532 ], [ 1647, @@ -148,17 +148,17 @@ [ 1695, 5, - 2366 + 35605 ], [ 1735, 4, - 17784 + 9243 ], [ 1767, 5, - 48552 + 48553 ], [ 1775, @@ -183,7 +183,7 @@ [ 1922, 4, - 60243 + 66676 ], [ 1959, @@ -193,22 +193,22 @@ [ 2014, 5, - 54321 + 54323 ], [ 2045, - 5, - 40142 + 4, + 78727 ], [ 2060, 5, - 22640 + 26331 ], [ 2096, 4, - 58060 + 58061 ], [ 2116, diff --git a/tests/XRAY.entities.AZW3.json b/tests/XRAY.entities.AZW3.json index 6f5093e..481c75c 100644 --- a/tests/XRAY.entities.AZW3.json +++ b/tests/XRAY.entities.AZW3.json @@ -6,8 +6,8 @@ 1, 0, 0, - 199, - 223, + 192, + 209, 3, "0,1,2" ] @@ -41,28 +41,33 @@ "occurrence": [ [ 1, + 1280, + 15 + ], + [ + 2, 1845, 13 ], [ - 1, + 2, 1945, 14 ], [ - 1, + 2, 2042, 13 ], [ - 2, - 2568, - 15 + 3, + 2330, + 42 ], [ - 3, - 2917, - 18 + 1, + 2568, + 15 ], [ 4, @@ -81,16 +86,11 @@ ], [ 6, - 3844, - 6 + 3838, + 12 ], [ 7, - 3852, - 5 - ], - [ - 8, 4108, 22 ], @@ -100,14 +100,14 @@ 12 ], [ - 7, - 4332, - 5 + 8, + 6010, + 6 ], [ 9, - 6010, - 6 + 6533, + 7 ], [ 4, @@ -147,7 +147,7 @@ [ 14, 9633, - 6 + 14 ], [ 15, @@ -157,30 +157,25 @@ [ 16, 9701, - 15 + 9 ], [ 17, - 9720, - 7 - ], - [ - 18, 9741, 6 ], [ - 19, + 18, 9810, 6 ], [ - 20, + 19, 9821, - 5 + 15 ], [ - 21, + 20, 11351, 13 ], @@ -190,17 +185,17 @@ 8 ], [ - 22, + 21, 11453, 9 ], [ - 23, + 22, 11474, 7 ], [ - 24, + 23, 11485, 18 ], @@ -210,7 +205,7 @@ 15 ], [ - 25, + 24, 11742, 5 ], @@ -220,42 +215,47 @@ 8 ], [ - 26, + 25, 11839, 4 ], [ - 27, + 26, 11855, 11 ], [ - 28, + 27, 11928, 11 ], [ - 29, + 28, 12340, 8 ], [ - 30, + 29, 12349, 13 ], [ - 31, + 30, 12572, 10 + ], + [ + 31, + 12583, + 18 ] ], "type": [ [ - "218,154,95,326,241,130,101,103,96,15" + "216,4,148,93,312,100,185,245,15,247" ], [ - "4,32,22,187,104,245,189,42,186,5" + "21,31,159,124,183,99,101,243,41,64" ] ] } \ No newline at end of file diff --git a/tests/XRAY.entities.KFX.json b/tests/XRAY.entities.KFX.json index 18c39b7..2dabbc7 100644 --- a/tests/XRAY.entities.KFX.json +++ b/tests/XRAY.entities.KFX.json @@ -6,8 +6,8 @@ 1, 0, 0, - 205, - 222, + 195, + 207, 4, "0,1,2,3" ] @@ -84,186 +84,186 @@ ], [ 5, - 1184, - 19 - ], - [ - 6, 1240, 15 ], [ - 7, + 6, 1291, 9 ], [ - 8, + 7, 1346, 10 ], [ - 9, - 1542, - 6 + 7, + 1530, + 10 ], [ - 10, + 8, 1560, 22 ], [ - 8, + 7, 1683, 10 ], [ 9, - 1695, - 5 - ], - [ - 11, 2269, 6 ], [ - 6, + 10, + 2285, + 7 + ], + [ + 5, 2682, 8 ], [ - 6, + 5, 3096, 8 ], [ - 12, + 11, 3209, 11 ], [ - 13, + 12, 3452, 10 ], [ - 6, + 5, 3701, 16 ], [ - 14, + 13, 3913, 12 ], [ - 15, + 14, 3925, 10 ], [ - 16, + 15, 4039, - 6 + 14 ], [ - 17, + 16, 4069, 12 ], [ - 18, + 17, 4101, - 15 - ], - [ - 19, - 4118, - 7 + 9 ], [ - 20, + 18, 4139, 6 ], [ - 21, + 19, 4198, 6 ], [ - 22, + 20, 4209, - 5 + 13 ], [ - 23, + 21, 5531, 13 ], [ - 6, + 5, 5586, 8 ], [ - 24, + 22, 5633, 9 ], [ - 25, + 23, 5654, 7 ], [ - 26, + 24, 5665, 18 ], [ - 16, + 15, 5704, 15 ], [ - 27, + 25, 5884, 17 ], [ - 28, + 26, 5902, 4 ], [ - 29, + 27, 5911, 11 ], [ - 30, + 28, 5984, 11 ], [ - 31, + 29, 6356, 8 ], [ - 32, + 30, 6365, 13 + ], + [ + 31, + 6545, + 10 + ], + [ + 32, + 6556, + 18 ] ], "type": [ [ - "223,155,96,334,246,132,102,104,165,17" + "217,5,148,93,311,159,100,185,242,16" ], [ - "6,34,24,189,250,105,191,44,7,67" + "22,32,124,183,99,101,240,42,65,6" ] ] } \ No newline at end of file