Merge pull request #52 from UIUCLibrary/dev

Dev

Author: henryborchers

galatea/clean_tsv.py

@@ -204,6 +204,7 @@ def clean_tsv(
         modified_data = []
         dialect = get_tsv_dialect(tsv_file)
         field_names = galatea.tsv.get_field_names(source)
+        row: TableRow[Marc_Entry]
         for row in iter_tsv_fp(tsv_file, dialect):
             transformed_row = transform_row_and_merge(
                 row.entry,

galatea/cli.py

@@ -23,6 +23,7 @@
 
 logger = logging.getLogger(__name__)
 
+
 def get_versions_from_package() -> Optional[str]:
     """Get version information from the package metadata."""
     try:

galatea/modifiers.py

@@ -12,6 +12,7 @@
 if typing.TYPE_CHECKING:
     from galatea.marc import MarcEntryDataTypes
 
+
 def split_and_modify(
     entry: MarcEntryDataTypes,
     funcs: List[Callable[[MarcEntryDataTypes], MarcEntryDataTypes]],
@@ -117,6 +118,7 @@ def regex_transform(
         return None
     return re.sub(pattern, replacement, entry)
 
+
 @functools.cache
 def _get_relator_term_regex():
     terms = (
@@ -125,6 +127,7 @@ def _get_relator_term_regex():
     relator_terms_in_regex = "|".join(terms)
     return fr"({relator_terms_in_regex})\.?"
 
+
 def remove_relator_terms(entry: MarcEntryDataTypes):
     """Remove any relator terms from the string.
 

galatea/validate_authorized_terms.py

@@ -89,26 +89,6 @@ def check_terms(name: str, cache: CachedApiCheck) -> bool:
     return cache.get_data(name).status_code == 200
 
 
-# def _iter_things_to_check(
-#     source: pathlib.Path,
-# ) -> Iterator[Tuple[int, str, str]]:
-#     field_names = {
-#         "260$a",
-#         "264$a",
-#     }
-#     with open(source) as tsv_file:
-#         dialect = get_tsv_dialect(tsv_file)
-#     for row in iter_tsv_file(source, dialect):
-#         for field_name in field_names:
-#             field = row.entry[field_name]
-#             if not field:
-#                 continue
-#             field = field.strip()
-#             for name in field.split("||"):
-#                 cleaned_string = name.strip()
-#                 yield row.line_number, field_name, cleaned_string
-#
-
 def optional_rate_limited_iterator(
     iterable: Iterable[T],
     bypass_sleep_func: Callable[[T], bool] = lambda *_: False,
@@ -137,6 +117,7 @@ def optional_rate_limited_iterator(
                     start_time = time.time()
         yield results
 
+
 class IterTerms(collections.abc.Iterable):
     tsv_file_row_iterator = iter_tsv_file
 
@@ -145,10 +126,11 @@ def __init__(self, source):
         self.field_names = set()
 
     def iter_rows(self):
-        with open(self._source) as tsv_file:
+        with open(self._source, encoding="utf-8") as tsv_file:
             dialect = get_tsv_dialect(tsv_file)
         for row in IterTerms.tsv_file_row_iterator(self._source, dialect):
             yield row
+
     def __iter__(self):
         for row in self.iter_rows():
             for field_name in self.field_names:
@@ -160,6 +142,7 @@ def __iter__(self):
                     cleaned_string = name.strip()
                     yield row.line_number, field_name, cleaned_string
 
+
 def validate_authorized_terms(source: pathlib.Path) -> None:
     """Validate Authorized terms.
 

pyproject.toml

@@ -6,14 +6,14 @@ requires-python = ">= 3.10"
 authors = [
     {name="University Library at The University of Illinois at Urbana Champaign: Preservation Services", email="[email protected]"}
 ]
-license = { file="LICENSE"}
+license = "NCSA"
+license-files = ["LICENSE"]
 readme = { file= "README.md", content-type="text/markdown" }
 classifiers = [
     "Development Status :: 3 - Alpha",
     "Environment :: Console",
     "Natural Language :: English",
     "Intended Audience :: Science/Research",
-    "License :: OSI Approved :: University of Illinois/NCSA Open Source License",
     "Programming Language :: Python :: 3",
     "Topic :: System :: Archiving"
 
@@ -27,7 +27,7 @@ dependencies = [
 galatea = "galatea.cli:main"
 
 [build-system]
-requires = ["setuptools>=75.1.0"]
+requires = ["setuptools>=77.0.0"]
 build-backend = "setuptools.build_meta"
 
 [tool.setuptools]
@@ -59,4 +59,5 @@ ignore = ["D203"]
 convention = "google"
 
 [tool.ruff.lint.per-file-ignores]
-"tests/**" = ["D"]
+"tests/**" = ["D"]
+"docs/conf.py" = ["D"]

tests/test_clean_tsv.py

@@ -1,11 +1,10 @@
 import pathlib
-from unittest.mock import Mock, mock_open, patch, ANY, create_autospec
+from unittest.mock import Mock, mock_open, patch, ANY
 
 import pytest
 import galatea.tsv
 import galatea.clean_tsv
 from galatea import clean_tsv, modifiers
-import io
 
 def test_make_empty_strings_none_removes_empty_strings():
     record = {"1": "somedata", "50": ""}