Merge branch 'main' into textExtract

mailcom/inout.py

@@ -2,15 +2,17 @@
 from email.parser import BytesParser
 import eml_parser
 from pathlib import Path
+import os
 
 
 def list_of_files(directory_name: str) -> list[Path]:
-    """Get all the eml files in the directory and put them in a list."""
+    if not os.path.exists(directory_name): # check if given dir exists raises error otherwise
+        raise OSError("Path {} does not exist".format(directory_name))
     mypath = Path(directory_name)
     pattern = [".eml", ".html"]  # we would not change the file type through user input
     email_list = [mp.resolve() for mp in mypath.glob("**/*") if mp.suffix in pattern]
     if len(email_list) == 0:
-        raise ValueError("The directory {} does not contain .eml or .html files".format(mypath))
+        raise ValueError("The directory {} does not contain .eml or .html files. Please check that the directory is containing the email data files".format(mypath))
     return email_list
 
 

mailcom/test/test_inout.py

@@ -10,6 +10,9 @@ def test_list_of_files_empty(tmp_path):
     with pytest.raises(ValueError):
         list_of_files(tmp_path)
 
+def test_list_of_files_dir_not_existing():
+    with pytest.raises(OSError):
+        list_of_files("nonexistingDir")
 
 def test_list_of_files_correct_format(tmp_path):
     p = tmp_path / "test.eml"