Use enabled senses and update spaCy 3.8.2

also update test files because the new spaCy models
xxyzz · Oct 6, 2024 · 2a8d8bd · 2a8d8bd
1 parent 3e54a5b
commit 2a8d8bd
Show file tree

Hide file tree

Showing 8 changed files with 141 additions and 144 deletions.
diff --git 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
@@ -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
@@ -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
@@ -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
@@ -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
@@ -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,