Deprecate sphinx.util.stemmer in favour of snowballstemmer

PorterStemmer, BaseStemmer, PyStemmer, StandardStemmer, and get_stemmer are deprecated
sphinx.util.stemmer.porter is removed

AUTHORS

@@ -96,5 +96,3 @@ authors and projects:
 
 * sphinx.util.jsdump uses the basestring encoding from simplejson,
   written by Bob Ippolito, released under the MIT license
-* sphinx.util.stemmer was written by Vivake Gupta, placed in the
-  Public Domain

sphinx/search/zh.py

@@ -4,8 +4,9 @@
 import re
 from typing import Dict, List
 
+import snowballstemmer
+
 from sphinx.search import SearchLanguage
-from sphinx.util.stemmer import get_stemmer
 
 try:
     import jieba
@@ -230,7 +231,7 @@ def init(self, options: Dict) -> None:
             if dict_path and os.path.isfile(dict_path):
                 jieba.load_userdict(dict_path)
 
-        self.stemmer = get_stemmer()
+        self.stemmer = snowballstemmer.stemmer('english')
 
     def split(self, input: str) -> List[str]:
         chinese: List[str] = []
@@ -252,8 +253,8 @@ def stem(self, word: str) -> str:
         should_not_be_stemmed = (
             word in self.latin_terms and
             len(word) >= 3 and
-            len(self.stemmer.stem(word.lower())) < 3
+            len(self.stemmer.stemWord(word.lower())) < 3
         )
         if should_not_be_stemmed:
             return word.lower()
-        return self.stemmer.stem(word.lower())
+        return self.stemmer.stemWord(word.lower())

sphinx/util/stemmer/__init__.py

@@ -1,16 +1,67 @@
 """Word stemming utilities for Sphinx."""
+import warnings
 
 import snowballstemmer
 
-from sphinx.util.stemmer.porter import PorterStemmer
+from sphinx.deprecation import RemovedInSphinx70Warning
+
+_ENGLISH_STEMMER_TYPE = type(snowballstemmer.stemmer('english'))
+
+
+class PorterStemmer(_ENGLISH_STEMMER_TYPE):
+    def __init__(self):
+        warnings.warn(f"{self.__class__.__name__} is deprecated, use "
+                      "snowballstemmer.stemmer('english') instead.",
+                      RemovedInSphinx70Warning, stacklevel=2)
+        super().__init__()
+
+    def stem(self, p: str, i: int, j: int) -> str:
+        warnings.warn(f"{self.__class__.__name__}.stem() is deprecated, use "
+                      "snowballstemmer.stemmer('english').stemWord() instead.",
+                      RemovedInSphinx70Warning, stacklevel=2)
+        return self.stemWord(p)
 
 
 class BaseStemmer:
+    def __init__(self):
+        warnings.warn(f"{self.__class__.__name__} is deprecated.",
+                      RemovedInSphinx70Warning, stacklevel=2)
+        super().__init__()
+
     def stem(self, word: str) -> str:
         raise NotImplementedError
 
 
-def get_stemmer() -> BaseStemmer:
-    stemmer = snowballstemmer.stemmer('english')
+class PyStemmer(BaseStemmer, _ENGLISH_STEMMER_TYPE):
+    def __init__(self):
+        warnings.warn(f"{self.__class__.__name__} is deprecated, use "
+                      "snowballstemmer.stemmer('english') instead.",
+                      RemovedInSphinx70Warning, stacklevel=2)
+        super().__init__()
 
+    def stem(self, word: str) -> str:
+        warnings.warn(f"{self.__class__.__name__}.stem() is deprecated, use "
+                      "snowballstemmer.stemmer('english').stemWord() instead.",
+                      RemovedInSphinx70Warning, stacklevel=2)
+        return self.stemWord(word)
 
+
+class StandardStemmer(BaseStemmer, PorterStemmer):
+    def __init__(self):
+        warnings.warn(f"{self.__class__.__name__} is deprecated, use "
+                      "snowballstemmer.stemmer('english') instead.",
+                      RemovedInSphinx70Warning, stacklevel=2)
+        super().__init__()
+
+    def stem(self, word: str) -> str:
+        warnings.warn(f"{self.__class__.__name__}.stem() is deprecated, use "
+                      "snowballstemmer.stemmer('english').stemWord() instead.",
+                      RemovedInSphinx70Warning, stacklevel=2)
+        return self.stemWord(word)
+
+
+def get_stemmer() -> BaseStemmer:
+    warnings.warn("get_stemmer() is deprecated, use "
+                  "snowballstemmer.stemmer('english') instead.",
+                  RemovedInSphinx70Warning, stacklevel=2)
+    return PyStemmer()