Merge pull request #30 from acdh-oeaw/29-harden-library

29 harden library

setup.py

@@ -6,7 +6,7 @@
 
 setup(
     name="acdh-transkribus-utils",
-    version="2.10",
+    version="2.11",
     description="""some utility function to interact with the Transkribus-API""",
     long_description=readme,
     long_description_content_type="text/markdown",

tests/test_package.py

@@ -1,6 +1,8 @@
 import os
+import shutil
 import unittest
 from pathlib import Path
+import pytest
 
 from acdh_xml_pyutils.xml import XMLReader
 
@@ -86,7 +88,51 @@ def test_011_user_id(self):
 
     def test_012_add_user(self):
         client = CLIENT
-        col_id = 190357
-        status = client.add_user_to_collection(client.user, col_id=col_id, send_mail=False)
+        status = client.add_user_to_collection(
+            client.user, col_id=COL_ID, send_mail=False
+        )
         self.assertTrue(f"{190357}" in status)
         self.assertTrue(f"{client.user}" in status)
+
+    def test_013_wrong_credentials(self):
+        with pytest.raises(Exception):
+            ACDHTranskribusUtils(user="[email protected]", password="this-wont work")
+
+    def test_014_doc_md(self):
+        client = CLIENT
+        doc_id = 1351422
+        doc_md = client.get_doc_md(doc_id, COL_ID)
+        self.assertTrue(doc_md["docId"], doc_id)
+
+    def test_014_get_fulldoc_md(self):
+        client = CLIENT
+        doc_id = 1351422
+        doc_md = client.get_fulldoc_md(doc_id, COL_ID)
+        self.assertTrue(doc_md["extra_info"]["nrOfPages"], 10)
+
+    def test_015_save_mets_to_file(self):
+        client = CLIENT
+        doc_id = 1351422
+        f_name = f"{doc_id}_mets.xml"
+        try:
+            os.remove(f_name)
+        except OSError:
+            pass
+        client.save_mets_to_file(doc_id, COL_ID)
+        my_file = Path(f_name)
+        self.assertTrue(my_file.is_file())
+        os.remove(f_name)
+
+    def test_016_list_documents(self):
+        client = CLIENT
+        result = client.list_documents(COL_ID)
+        self.assertTrue(len(result), 5)
+
+    def test_017_dl_collection(self):
+        shutil.rmtree(f"{COL_ID}", ignore_errors=True)
+        result = CLIENT.collection_to_mets(COL_ID)
+        doc_id = 1351422
+        self.assertTrue(doc_id in result)
+        my_file = Path(os.path.join(f"{COL_ID}", f"{doc_id}_mets.xml"))
+        self.assertTrue(my_file.is_file())
+        shutil.rmtree(f"{COL_ID}", ignore_errors=True)

transkribus_utils/transkribus_utils.py

@@ -152,7 +152,7 @@ def get_fulldoc_md(self, doc_id, col_id, page_id="1"):
             result["img_url"] = doc_xml.xpath("./url/text()")[0]
             result["img_url"] = doc_xml.xpath("./url/text()")[0]
             result["extra_info"] = self.get_doc_md(
-                doc_id, base_url=self.base_url, col_id=col_id
+                doc_id, col_id=col_id
             )
             return result
         else:
@@ -282,7 +282,12 @@ def collection_to_mets(self, col_id, file_path=".", filter_by_doc_ids=[]):
         print(f"{len(doc_ids)} to download")
         counter = 1
         for doc_id in doc_ids:
-            save_mets = self.save_mets_to_file(doc_id, col_id, file_path=col_dir)
+            try:
+                save_mets = self.save_mets_to_file(doc_id, col_id, file_path=col_dir)
+            except Exception as e:
+                print(f"failed to save mets for DOC-ID: {doc_id} in COLLECTION: {col_id} due to ERROR: {e}")
+                counter += 1
+                continue
             file_list = self.save_image_names_to_file(doc_id, col_id, file_path=col_dir)
             print(f"saving: {save_mets}")
             print(f"saving: {file_list}")