Merge pull request #7 from jftuga/feature-automate-download

advise user on model download
jftuga · Jan 3, 2025 · 26c0fa7 · 26c0fa7
2 parents afb0cd1 + 713437d
commit 26c0fa7
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 46 deletions.
diff --git a/Pipfile.lock b/Pipfile.lock
diff --git a/deidentification/deidentification.py b/deidentification/deidentification.py
@@ -89,7 +89,12 @@ def __init__(self, config: DeidentificationConfig = DeidentificationConfig()):
             torch.load = self.__safe_load
             spacy.prefer_gpu()
             if not Deidentification.nlp:
-                Deidentification.nlp = spacy.load(self.config.spacy_model)
+                try:
+                    Deidentification.nlp = spacy.load(self.config.spacy_model)
+                except OSError as err:
+                    self.model_not_found_error(str(err))
+                except Exception as err:
+                    raise err
 
     def __str__(self) -> str:
         """Return a string representation of the Deidentification instance.
@@ -107,6 +112,26 @@ def __str__(self) -> str:
         ]
         return "\n".join(program_info) + "\n" + str(self.config)
 
+    def model_not_found_error(self, err: str):
+        """Handles errors related to missing spaCy models and provides installation instructions.
+
+        This function processes spaCy model errors, specifically handling cases where
+        a required model cannot be found. If the error indicates a missing model,
+        it prints installation instructions to stderr and exits the program.
+
+        Args:
+            err: Error message string from the spaCy library.
+        """
+        print(file=sys.stderr)
+        print(str(err), file=sys.stderr)
+        if "Can't find model" in str(err):
+            print(file=sys.stderr)
+            print("Please manually run the following command one time to download the required model:", file=sys.stderr)
+            print(file=sys.stderr)
+            print(f"python -m spacy download {self.config.spacy_model}", file=sys.stderr)
+            print(file=sys.stderr)
+            sys.exit(1)
+
     def deidentify(self, text: str) -> str:
         """De-identify personal information in the input text.
 

diff --git a/deidentification/deidentification_constants.py b/deidentification/deidentification_constants.py
@@ -1,6 +1,6 @@
 pgmName = "deidentification"
 pgmUrl = "https://github.com/jftuga/deidentification"
-pgmVersion = "1.1.2"
+pgmVersion = "1.2.0"
 
 GENDER_PRONOUNS = {
     "he": "HE/SHE",
@@ -15,30 +15,6 @@
     "mrs.": "",
     "ms.": ""}
 
-# ALL_CONTRACTIONS = (
-#     "it",
-#     "that",
-#     "what",
-#     "there",
-#     "here",
-#     "let",
-#     "how",
-#     "where",
-#     "who",
-#     "when",
-#     "one",
-#     "somebody",
-#     "someone",
-#     "something",
-#     "nobody",
-#     "everyone",
-#     "anybody",
-#     "nothing",
-#     "why",
-#     "this",
-#     "which")
-
-
 HTML_BEGIN = """<!DOCTYPE html>
 <html>
 <head>
@@ -81,4 +57,3 @@ class bcolors:
     ENDC = '\033[0m'
     BOLD = '\033[1m'
     UNDERLINE = '\033[4m'
-