closes #173

CHANGELOG.md

@@ -8,6 +8,7 @@ The `pycldf` package adheres to [Semantic Versioning](http://semver.org/spec/v2.
   by `json.dumps`.
 - Fixed bug whereby SQLite conversion would fail when the name of a renamed column clashed with an
   existing column name in the table.
+- Emit warning when encountering invalid main part for mediaType property.
 
 
 ## [1.37.1] - 2024-03-18

src/pycldf/validators.py

@@ -1,4 +1,5 @@
 import re
+import warnings
 import functools
 
 
@@ -37,6 +38,12 @@ def valid_grammaticalityJudgement(dataset, table, column, row):
             raise ValueError('Glottolog language linked from ungrammatical example')
 
 
+def valid_mediaType(dataset, table, column, row):
+    main, _, sub = row[column.name].partition('/')
+    if not re.fullmatch('[a-z]+', main):
+        warnings.warn('Invalid main part in media type: {}'.format(main))
+
+
 VALIDATORS = [
     (
         None,
@@ -58,4 +65,8 @@ def valid_grammaticalityJudgement(dataset, table, column, row):
         None,
         'http://cldf.clld.org/v1.0/terms.rdf#grammaticalityJudgement',
         valid_grammaticalityJudgement),
+    (
+        None,
+        'http://cldf.clld.org/v1.0/terms.rdf#mediaType',
+        valid_mediaType),
 ]

tests/test_dataset.py

@@ -2,6 +2,7 @@
 import logging
 import zipfile
 import warnings
+import mimetypes
 import contextlib
 
 import pytest
@@ -177,6 +178,16 @@ def test_example_validators(ds):
         ds.validate()
 
 
+def test_invalid_mimetype(ds, recwarn):
+    ds.add_component('MediaTable')
+    ds.write(MediaTable=[{
+        'ID': '1',
+        'Media_Type': mimetypes.guess_type('f.png'),
+        'Download_URL': 'http://example.org'}])
+    ds.validate()
+    assert 'Invalid main part' in str(recwarn.pop(UserWarning).message)
+
+
 def test_regex_validator_for_listvalued_column(ds):
     ds.add_table(
         'test',