Merge branch 'master' of https://github.com/mideind/GreynirCorrect

Author: sveinbjornt

src/reynir_correct/classifier.py

@@ -42,14 +42,13 @@
 
 """
 
-from typing import List, overload
+from typing import List, Union, overload
 
 
 try:
     from datasets import load_dataset
     from transformers import pipeline  # type: ignore
 except:
-    import sys
     import warnings
 
     warningtext = (
@@ -82,19 +81,18 @@ def classify(self, text: str) -> bool:
     def classify(self, text: List[str]) -> List[bool]:
         ...
 
-    def classify(self, text):
+    def classify(self, text: Union[str, List[str]]) -> Union[List[bool], bool]:
         """Classify a sentence or sentences.
         For each sentence, return true iff the sentence probably contains an error."""
         if isinstance(text, str):
             text = [text]
 
-        result = self.pipe([self._domain_prefix + t for t in text])
-        result = [r["generated_text"] == self._true_label for r in result]
+        pipe_result = self.pipe([self._domain_prefix + t for t in text])
+        result: List[bool] = [
+            r["generated_text"] == self._true_label for r in pipe_result
+        ]
 
-        if len(result) == 1:
-            result = result[0]
-
-        return result
+        return result[0] if len(result) == 1 else result
 
 
 def _main() -> None:

src/reynir_correct/config/GreynirCorrect.conf

@@ -204,6 +204,8 @@
 "áratugs", "áratugar"
 "áratugsins", "áratugarins"
 "árð", "árið"
+"arfleið", "arfleifð"
+"arfleiðar", "arfleifðar"
 "Argentísk", "Argentínsk"
 "argentíska", "argentínska"
 "argentíski", "argentínski"

src/reynir_correct/pattern.py

@@ -369,7 +369,7 @@ def wrong_preposition_grin_af(self, match: SimpleTree) -> None:
         pp = match.first_match('P > { "af" }')
         if pp is None:
             pp = match.first_match('ADVP > { "af" }')
-        if np is None or pp is None:
+        if vp is None or np is None or pp is None:
             return
         pp_af = pp.first_match('"af"')
         if pp_af is None:

src/reynir_correct/wrappers.py

@@ -64,8 +64,12 @@
     Any,
     Union,
     cast,
+    TYPE_CHECKING,
 )
 
+if TYPE_CHECKING:
+    from .classifier import SentenceClassifier
+
 import sys
 import argparse
 import json