smarter initialization of optional analyzers; use importlib to test o…

…ptional feature availability
NatLibFi · Nov 21, 2024 · fd8b9a7 · fd8b9a7
1 parent 4809561
commit fd8b9a7
Show file tree

Hide file tree

Showing 4 changed files with 27 additions and 19 deletions.
diff --git a/annif/analyzer/__init__.py b/annif/analyzer/__init__.py
@@ -8,7 +8,7 @@
 import annif
 from annif.util import parse_args
 
-from . import simple, simplemma, snowball
+from . import simple, simplemma, snowball, spacy, voikko
 
 if TYPE_CHECKING:
     from annif.analyzer.analyzer import Analyzer
@@ -17,7 +17,10 @@
 
 
 def register_analyzer(analyzer):
-    _analyzers[analyzer.name] = analyzer
+    if analyzer.is_available():
+        _analyzers[analyzer.name] = analyzer
+    else:
+        annif.logger.debug(f"{analyzer.name} analyzer not available, not enabling it")
 
 
 def get_analyzer(analyzerspec: str) -> Analyzer:
@@ -37,18 +40,5 @@ def get_analyzer(analyzerspec: str) -> Analyzer:
 register_analyzer(simple.SimpleAnalyzer)
 register_analyzer(snowball.SnowballAnalyzer)
 register_analyzer(simplemma.SimplemmaAnalyzer)
-
-# Optional analyzers
-try:
-    from . import voikko
-
-    register_analyzer(voikko.VoikkoAnalyzer)
-except ImportError:
-    annif.logger.debug("voikko not available, not enabling voikko analyzer")
-
-try:
-    from . import spacy
-
-    register_analyzer(spacy.SpacyAnalyzer)
-except ImportError:
-    annif.logger.debug("spaCy not available, not enabling spacy analyzer")
+register_analyzer(voikko.VoikkoAnalyzer)
+register_analyzer(spacy.SpacyAnalyzer)
diff --git a/annif/analyzer/analyzer.py b/annif/analyzer/analyzer.py
@@ -22,6 +22,11 @@ class Analyzer(metaclass=abc.ABCMeta):
     name = None
     token_min_length = 3  # default value, can be overridden in instances
 
+    @staticmethod
+    def is_available() -> bool:
+        """Return True if the analyzer is available for use, False if not."""
+        return True  # can be overridden in implementations if necessary
+
     def __init__(self, **kwargs) -> None:
         if _KEY_TOKEN_MIN_LENGTH in kwargs:
             self.token_min_length = int(kwargs[_KEY_TOKEN_MIN_LENGTH])

diff --git a/annif/analyzer/spacy.py b/annif/analyzer/spacy.py
@@ -2,6 +2,8 @@
 
 from __future__ import annotations
 
+import importlib
+
 import annif.util
 from annif.exception import OperationFailedException
 
@@ -13,6 +15,11 @@
 class SpacyAnalyzer(analyzer.Analyzer):
     name = "spacy"
 
+    @staticmethod
+    def is_available() -> bool:
+        # return True iff spaCy is installed
+        return importlib.util.find_spec("spacy") is not None
+
     def __init__(self, param: str, **kwargs) -> None:
         import spacy
 

diff --git a/annif/analyzer/voikko.py b/annif/analyzer/voikko.py
@@ -3,15 +3,19 @@
 from __future__ import annotations
 
 import functools
-
-import voikko.libvoikko
+import importlib
 
 from . import analyzer
 
 
 class VoikkoAnalyzer(analyzer.Analyzer):
     name = "voikko"
 
+    @staticmethod
+    def is_available() -> bool:
+        # return True iff Voikko is installed
+        return importlib.util.find_spec("voikko") is not None
+
     def __init__(self, param: str, **kwargs) -> None:
         self.param = param
         self.voikko = None
@@ -26,6 +30,8 @@ def __getstate__(self) -> dict[str, str | None]:
 
     @functools.lru_cache(maxsize=500000)
     def _normalize_word(self, word: str) -> str:
+        import voikko.libvoikko
+
         if self.voikko is None:
             self.voikko = voikko.libvoikko.Voikko(self.param)
         result = self.voikko.analyze(word)