Update spaCy English models

data/deps.json

@@ -5,7 +5,9 @@
   "spacy": "3.7.5",
   "spacy_cpu_model": "3.7.0",
   "spacy_trf_model": "3.7.2",
+  "en_spacy_cpu_model": "3.7.1",
+  "en_spacy_trf_model": "3.7.3",
   "thinc-apple-ops": "0.1.5",
-  "torch": "2.3.1",
+  "torch": "2.4.0",
   "typing-extensions": "4.12.2"
 }

deps.py

@@ -11,6 +11,7 @@
     PROFICIENCY_RELEASE_URL,
     Prefs,
     get_plugin_path,
+    get_spacy_model_version,
     get_wiktionary_klld_path,
     kindle_db_path,
     load_plugin_json,
@@ -44,9 +45,7 @@ def install_deps(pkg: str, notif: Any) -> None:
         if pkg == "":
             pip_install("spacy", dep_versions["spacy"], notif=notif)
         else:
-            model_version = dep_versions[
-                "spacy_trf_model" if pkg.endswith("_trf") else "spacy_cpu_model"
-            ]
+            model_version = get_spacy_model_version(pkg, dep_versions)
             url = (
                 "https://github.com/explosion/spacy-models/releases/download/"
                 f"{pkg}-{model_version}/{pkg}-{model_version}-py3-none-any.whl"

dump_lemmas.py

@@ -6,6 +6,7 @@
     from .utils import (
         Prefs,
         custom_lemmas_folder,
+        get_spacy_model_version,
         insert_installed_libs,
         load_plugin_json,
         use_kindle_ww_db,
@@ -14,6 +15,7 @@
     from utils import (
         Prefs,
         custom_lemmas_folder,
+        get_spacy_model_version,
         insert_installed_libs,
         load_plugin_json,
         use_kindle_ww_db,
@@ -74,9 +76,7 @@ def dump_spacy_docs(
     save_spacy_docs(
         nlp,
         spacy_model,
-        pkg_versions[
-            "spacy_trf_model" if spacy_model.endswith("_trf") else "spacy_cpu_model"
-        ],
+        get_spacy_model_version(spacy_model, pkg_versions),
         lemma_lang,
         is_kindle,
         lemmas_conn,

parse_job.py

@@ -32,6 +32,7 @@
         Prefs,
         dump_prefs,
         get_plugin_path,
+        get_spacy_model_version,
         get_user_agent,
         get_wiktionary_klld_path,
         insert_installed_libs,
@@ -68,6 +69,7 @@
     from utils import (
         CJK_LANGS,
         Prefs,
+        get_spacy_model_version,
         insert_installed_libs,
         kindle_db_path,
         load_languages_data,
@@ -795,9 +797,7 @@ def create_spacy_matcher(
 
     disabled_pipes = list(set(["ner", "parser", "senter"]) & set(nlp.pipe_names))
     pkg_versions = load_plugin_json(plugin_path, "data/deps.json")
-    model_version = pkg_versions[
-        "spacy_trf_model" if model.endswith("_trf") else "spacy_cpu_model"
-    ]
+    model_version = get_spacy_model_version(model, pkg_versions)
     # Chinese words don't have inflection forms, only use phrase matcher
     use_lemma_matcher = prefs["use_pos"] and lemma_lang != "zh" and model != ""
     phrase_matcher = PhraseMatcher(nlp.vocab, attr="LOWER")

utils.py

@@ -185,3 +185,14 @@ def load_languages_data(
         supported_languages["zh_cn"] = supported_languages["zh"].copy()
         supported_languages["zh_cn"]["name"] = "Simplified Chinese"
     return supported_languages
+
+
+def get_spacy_model_version(
+    model_name: str, dependency_versions: dict[str, str]
+) -> str:
+    key = "spacy_trf_model" if model_name.endswith("_trf") else "spacy_cpu_model"
+    lang_code = model_name[:2]
+    lang_key = f"{lang_code}_{key}"
+    if lang_key in dependency_versions:
+        return dependency_versions[lang_key]
+    return dependency_versions.get(key, "")