✨ Allow Minimal Sanity Check for XML/CSV Files via MAUS (#91)

src/maus/reader/flat_ahb_reader.py

@@ -34,6 +34,15 @@ def to_flat_ahb(self) -> FlatAnwendungshandbuch:
         raise NotImplementedError("The inheriting class has to implement this method")
 
 
+def check_file_can_be_parsed_as_ahb_csv(file_path: Path) -> None:
+    """
+    Returns nothing iff the given file is parsable as CSV and contains no obvious errors.
+    This is not a really sophisticated analysis but just a basic minimal sanity check.
+    In case of error an exception is raised.
+    """
+    _ = FlatAhbCsvReader(file_path)  # this may die with a meaningful exception
+
+
 class FlatAhbCsvReader(FlatAhbReader):
     """
     reads csv files and returns AHBs

src/maus/reader/mig_reader.py

@@ -9,6 +9,7 @@
 from lxml import etree  # type:ignore[import]
 
 from maus import SegmentGroupHierarchy
+from maus.edifact import EdifactFormat
 from maus.models._internal import EdifactStackSearchStrategy, MigFilterResult
 from maus.models.edifact_components import EdifactStack, EdifactStackLevel, EdifactStackQuery
 from maus.reader.etree_element_helpers import (
@@ -47,6 +48,16 @@ def get_edifact_stack(self, query: EdifactStackQuery) -> Optional[EdifactStack]:
 Result = TypeVar("Result")  #: is a type var to indicate an "arbitrary but same" type in a generic function
 
 
+def check_file_can_be_parsed_as_mig_xml(file_path: Path) -> None:
+    """
+    Returns nothing iff the given file is parsable as XML and contains no obvious errors.
+    This is not a really sophisticated analysis but just a basic minimal sanity check.
+    In case of error an exception is raised.
+    """
+    reader = MigXmlReader(file_path)
+    _ = EdifactFormat(reader.get_format_name())  # dies with an exception if the value is invalid
+
+
 # pylint:disable=c-extension-no-member
 class MigXmlReader(MigReader):
     """

tests/unit_tests/test_ahb_csv_reader.py

@@ -3,7 +3,7 @@
 
 import pytest  # type:ignore[import]
 
-from maus.reader.flat_ahb_reader import FlatAhbCsvReader
+from maus.reader.flat_ahb_reader import FlatAhbCsvReader, check_file_can_be_parsed_as_ahb_csv
 
 
 class TestAhbCsvReader:
@@ -235,3 +235,8 @@ def test_csv_file_reading_11042(self, datafiles):
     def test_merging_of_section_only_lines(self, input_lines: List[dict], expected_lines: List[dict]):
         actual = FlatAhbCsvReader.merge_section_only_lines(input_lines)
         assert actual == expected_lines
+
+    @pytest.mark.datafiles("./ahbs/FV2204/UTILMD/11042.csv")
+    def test_is_parsable(self, datafiles):
+        check_file_can_be_parsed_as_ahb_csv(Path(datafiles) / Path("11042.csv"))
+        # if no exception is thrown, the test is successful

tests/unit_tests/test_mig_xml_reader_real_data.py

@@ -5,7 +5,7 @@
 from lxml.etree import Element  # type:ignore[import]
 
 from maus.models.edifact_components import EdifactStackQuery
-from maus.reader.mig_reader import MigXmlReader
+from maus.reader.mig_reader import MigXmlReader, check_file_can_be_parsed_as_mig_xml
 
 ALL_MIG_XML_FILES = pytest.mark.datafiles(
     "./migs/FV2204/template_xmls/utilmd_1131.xml",
@@ -261,3 +261,8 @@ def test_simple_paths(
         actual_stack = reader.get_edifact_stack(query)
         assert actual_stack is not None
         assert actual_stack.to_json_path() == expected_path  # type:ignore[union-attr]
+
+    @ALL_MIG_XML_FILES
+    def test_is_parsable(self, datafiles):
+        check_file_can_be_parsed_as_mig_xml(Path(datafiles) / Path("utilmd_3225.xml"))
+        # if no exception is thrown, the test is successful