From cc4aaded6f705cd3e22de48c0ca9865b05f955da Mon Sep 17 00:00:00 2001
From: Ben Webb <ben@salilab.org>
Date: Fri, 6 Dec 2024 17:11:09 -0800
Subject: [PATCH] Check new releases against archive structures

Using a representative subset of the PDB-IHM archive,
test reading in each structure and writing it out again.
Run this as part of the "new release" workflow. Closes #155.
---
 make-release.sh          |  4 +++-
 util/check-db-entries.py | 50 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 53 insertions(+), 1 deletion(-)
 create mode 100644 util/check-db-entries.py

diff --git a/make-release.sh b/make-release.sh
index 5dd40c1..2f0f678 100755
--- a/make-release.sh
+++ b/make-release.sh
@@ -5,7 +5,9 @@
 #   codespell . --skip '*.cif' -L assertIn
 # - Update AuditConformDumper to match latest IHM dictionary if necessary
 # - Run util/validate-outputs.py to make sure all example outputs validate
-#   (cd util; PYTHONPATH=.. python3 ./validate-outputs.py)
+#   (cd util; PYTHONPATH=.. python3 validate-outputs.py)
+# - Run util/check-db-entries.py to check against some real archive structures
+#   (cd util; PYTHONPATH=.. python3 check-db-entries.py)
 # - Make sure all python-modelcif tests work using new IHM version
 # - Update ChangeLog.rst, util/debian/changelog, and util/python-ihm.spec
 #   with the release number and date
diff --git a/util/check-db-entries.py b/util/check-db-entries.py
new file mode 100644
index 0000000..6515afc
--- /dev/null
+++ b/util/check-db-entries.py
@@ -0,0 +1,50 @@
+import unittest
+import ihm.reader
+import ihm.dumper
+import urllib.request
+import os
+
+
+class Tests(unittest.TestCase):
+    def _read_cif(self, pdb_id):
+        url = 'https://pdb-ihm.org/cif/%s.cif' % pdb_id
+        with urllib.request.urlopen(url) as fh:
+            s, = ihm.reader.read(fh)
+        return s
+
+    def _write_cif(self, s, check=True):
+        with open('test.cif', 'w') as fh:
+            ihm.dumper.write(fh, [s], check=check)
+        os.unlink('test.cif')
+
+    def test_9a0e(self):
+        """Test IMP structure with incorrect reference sequence (9a0e)"""
+        s = self._read_cif('9a0e')
+        self.assertRaises(ValueError, self._write_cif, s)
+        self._write_cif(s, check=False)
+
+    def test_8zzd(self):
+        """Test docking structure with incorrect assembly (8zzd)"""
+        s = self._read_cif('8zzd')
+        self.assertRaises(ValueError, self._write_cif, s)
+        self._write_cif(s, check=False)
+
+    def test_9a82(self):
+        """Test HADDOCK structure without errors (9a82)"""
+        s = self._read_cif('9a82')
+        self._write_cif(s)
+
+    def test_9a13(self):
+        """Test HADDOCK structure with incorrect null feature (9a13)"""
+        s = self._read_cif('9a13')
+        self.assertRaises(ValueError, self._write_cif, s)
+        self._write_cif(s, check=False)
+
+    def test_9a0t(self):
+        """Test IMP structure without errors (9a0t)"""
+        s = self._read_cif('9a0t')
+        self._write_cif(s)
+
+
+if __name__ == '__main__':
+    unittest.main()