fix(ingest-xmlupload): apply mapping.csv even if extension has wrong …

…casing (DEV-3197) (#749)
dasch-swiss · Jan 19, 2024 · bac7c79 · bac7c79
1 parent bd915ef
commit bac7c79
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 10 deletions.
diff --git a/src/dsp_tools/commands/ingest_xmlupload/apply_ingest_id.py b/src/dsp_tools/commands/ingest_xmlupload/apply_ingest_id.py
@@ -56,13 +56,21 @@ def replace_filepath_with_sipi_id(
     used_media_file_paths = []
     new_tree = deepcopy(xml_tree)
     for elem in new_tree.iter():
-        if etree.QName(elem).localname.endswith("bitstream") and elem.text:
-            img_path = Path(elem.text)
-            img_path_str = str(img_path.relative_to(Path.cwd())) if img_path.is_absolute() else str(img_path)
-            if img_path_str in orig_path_2_id_filename:
-                elem.text = orig_path_2_id_filename[img_path_str]
-                used_media_file_paths.append(img_path_str)
-            else:
-                no_id_found.append((cast("etree._Element", elem.getparent()).attrib["id"], str(elem.text)))
+        if not etree.QName(elem).localname.endswith("bitstream") or not elem.text:
+            continue
+        img_path = Path(elem.text)
+        img_path = img_path.relative_to(Path.cwd()) if img_path.is_absolute() else img_path
+        img_path_str = str(img_path)
+        if img_path_str not in orig_path_2_id_filename:
+            img_path_str = str(img_path.with_suffix(img_path.suffix.lower()))
+        if img_path_str not in orig_path_2_id_filename:
+            img_path_str = str(img_path.with_suffix(img_path.suffix.upper()))
+
+        if img_path_str in orig_path_2_id_filename:
+            elem.text = orig_path_2_id_filename[img_path_str]
+            used_media_file_paths.append(img_path_str)
+        else:
+            no_id_found.append((cast("etree._Element", elem.getparent()).attrib["id"], str(elem.text)))
+
     unused_media_paths = [x for x in orig_path_2_id_filename if x not in used_media_file_paths]
     return new_tree, IngestInformation(unused_mediafiles=unused_media_paths, mediafiles_no_id=no_id_found)
diff --git a/src/dsp_tools/commands/ingest_xmlupload/user_information.py b/src/dsp_tools/commands/ingest_xmlupload/user_information.py
@@ -33,7 +33,7 @@ def ok_msg(self) -> str | None:
         if not self.unused_mediafiles and not self.mediafiles_no_id:
             return (
                 "All multimedia files referenced in the XML file were uploaded through dsp-ingest.\n"
-                "No multimedia files were uploaded through dsp-ingest that were not referenced in the XML file."
+                "All multimedia files uploaded through dsp-ingest were referenced in the XML file."
             )
         return None
 

diff --git a/test/unittests/commands/ingest_xmlupload/test_user_information.py b/test/unittests/commands/ingest_xmlupload/test_user_information.py
@@ -10,7 +10,7 @@ class TestIngestInformation:
     def test_no_problems(self) -> None:
         expected = (
             "All multimedia files referenced in the XML file were uploaded through dsp-ingest.\n"
-            "No multimedia files were uploaded through dsp-ingest that were not referenced in the XML file."
+            "All multimedia files uploaded through dsp-ingest were referenced in the XML file."
         )
         assert IngestInformation([], []).ok_msg() == expected