From 4a703ed803595ef6f0a2f7b792b771d982866b65 Mon Sep 17 00:00:00 2001
From: Kalev Takkis <ktakkis@informaticsmatters.com>
Date: Mon, 12 Feb 2024 13:03:08 +0000
Subject: [PATCH] Some changes to cset_upload.py to allow site observation
 short codes (#527)

* stashing

* fix: cset_upload.py updated to allow new-style site observation codes

NB! this probably still won't work! I suspect the file I was given is
broken and I cannot test it further
---
 viewer/cset_upload.py | 25 ++++++++++++++++---------
 viewer/sdf_check.py   |  8 +++++---
 2 files changed, 21 insertions(+), 12 deletions(-)

diff --git a/viewer/cset_upload.py b/viewer/cset_upload.py
index f1e84b6f..9cbeed32 100644
--- a/viewer/cset_upload.py
+++ b/viewer/cset_upload.py
@@ -198,9 +198,12 @@ def get_site_observation(
                 zfile_hashvals=zfile_hashvals,
             )
         else:
-            name = f'{compound_set.target.title}-{pdb_fn}'
+            name = pdb_fn
             try:
-                site_obvs = SiteObservation.objects.get(code__contains=name)
+                site_obvs = SiteObservation.objects.get(
+                    code__contains=name,
+                    experiment__experiment_upload__target__title=target,
+                )
             except SiteObservation.DoesNotExist:
                 # Initial SiteObservation lookup failed.
                 logger.warning(
@@ -210,7 +213,10 @@ def get_site_observation(
                 )
                 # Try alternatives.
                 # If all else fails then the site_obvs will be 'None'
-                qs = SiteObservation.objects.filter(code__contains=name)
+                qs = SiteObservation.objects.filter(
+                    code__contains=name,
+                    experiment__experiment_upload__target__title=target,
+                )
                 if qs.exists():
                     logger.info(
                         'Found SiteObservation containing name=%s qs=%s',
@@ -219,7 +225,10 @@ def get_site_observation(
                     )
                 else:
                     alt_name = name.split(':')[0].split('_')[0]
-                    qs = SiteObservation.objects.filter(code__contains=alt_name)
+                    qs = SiteObservation.objects.filter(
+                        code__contains=alt_name,
+                        experiment__experiment_upload__target__title=target,
+                    )
                     if qs.exists():
                         logger.info(
                             'Found SiteObservation containing alternative name=%s qs=%s',
@@ -328,15 +337,13 @@ def set_mol(
             # try exact match first
             try:
                 site_obvs = SiteObservation.objects.get(
-                    code__contains=str(compound_set.target.title + '-' + i),
+                    code=str(i),
                     experiment__experiment_upload__target_id=compound_set.target,
                 )
                 ref = site_obvs
             except SiteObservation.DoesNotExist:
                 qs = SiteObservation.objects.filter(
-                    code__contains=str(
-                        compound_set.target.title + '-' + i.split(':')[0].split('_')[0]
-                    ),
+                    code=str(i.split(':')[0].split('_')[0]),
                     experiment__experiment_upload__target_id=compound_set.target,
                 )
                 if not qs.exists():
@@ -503,7 +510,7 @@ def set_descriptions(
         computed_set.save()
 
         description_dict = description_mol.GetPropsAsDict()
-        for key in list(description_dict.keys()):
+        for key in description_dict.keys():
             if key in descriptions_needed and key not in [
                 'ref_mols',
                 'ref_pdb',
diff --git a/viewer/sdf_check.py b/viewer/sdf_check.py
index 949f1001..411128e4 100755
--- a/viewer/sdf_check.py
+++ b/viewer/sdf_check.py
@@ -89,10 +89,12 @@ def check_refmol(mol, validate_dict, target=None):
 
         for ref_mol in ref_mols:
             ref_strip = ref_mol.strip()
-            query_string = f'{target}-' + ref_strip.split(':')[0].split('_')[0]
-            query = SiteObservation.objects.filter(code__contains=query_string)
+            query = SiteObservation.objects.filter(
+                code=ref_strip,
+                experiment__experiment_upload__target__title=target,
+            )
             if len(query) == 0:
-                msg = f"No SiteObservation code contains '{query_string}'"
+                msg = f"No SiteObservation code contains '{ref_strip}'"
                 validate_dict = add_warning(
                     molecule_name=mol.GetProp('_Name'),
                     field='ref_mol',