rcsb.workflow 0.69__tar.gz → 0.71__tar.gz

{rcsb_workflow-0.69 → rcsb_workflow-0.71}/HISTORY.txt

@@ -64,3 +64,5 @@
  7-May-2026 - V0.67 Configure logging at start of CLI executable scripts
 11-May-2026 - V0.68 Temporarily skip DrugBank data processing until dataset becomes available
 17-Jul-2026 - V0.69 Add shortlink support for bcif generation
+21-Jul-2026 - V0.70 Bcif hash support for extended ids
+19-Aug-2026 - V0.71 Support s3 stash and s3/minio for chem search

{rcsb_workflow-0.69 → rcsb_workflow-0.71}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.4
+Metadata-Version: 2.5
 Name: rcsb.workflow
-Version: 0.69
+Version: 0.71
 Summary: RCSB Python data processing and ETL/ELT workflow entry points
 Project-URL: Homepage, https://github.com/rcsb/py-rcsb_workflow
 Author-email: John Westbrook <john.westbrook@rcsb.org>
@@ -20,9 +20,9 @@ Requires-Dist: openeye-toolkits>=2024.1.1
 Requires-Dist: pymongo>=4.8.0
 Requires-Dist: rcsb-db>=1.814
 Requires-Dist: rcsb-exdb>=1.35
-Requires-Dist: rcsb-utils-chem>=0.84
+Requires-Dist: rcsb-utils-chem>=0.87
 Requires-Dist: rcsb-utils-chemref>=0.98
-Requires-Dist: rcsb-utils-io>=1.49
+Requires-Dist: rcsb-utils-io>=1.54
 Requires-Dist: rcsb-utils-seq>=0.84
 Requires-Dist: rcsb-utils-seqalign>=0.32
 Requires-Dist: rcsb-utils-targets>=0.83

{rcsb_workflow-0.69 → rcsb_workflow-0.71}/pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "hatchling.build"
 [project]
 name = "rcsb.workflow"
 description = "RCSB Python data processing and ETL/ELT workflow entry points"
-version = "0.69"
+version = "0.71"
 readme = "README.md"
 authors = [
     { name="John Westbrook", email="john.westbrook@rcsb.org" }
@@ -37,9 +37,9 @@ dependencies = [
     "pymongo >= 4.8.0",
     #
     "mmcif >= 1.0.0",
-    "rcsb.utils.io >= 1.49",
+    "rcsb.utils.io >= 1.54",
     "rcsb.db >= 1.814",
-    "rcsb.utils.chem >= 0.84",
+    "rcsb.utils.chem >= 0.87",
     "rcsb.utils.seq >= 0.84",
     "rcsb.utils.targets >= 0.83",
     "rcsb.utils.chemref >= 0.98",
@@ -61,6 +61,12 @@ chem_search_index_cli = "rcsb.workflow.cli.ChemSearchIndexExec:main"
 
 
 # ----------------  hatch configuration ----------------
+[tool.hatch.metadata]
+# How to use: In the "dependencies" list use
+#    "rcsb.utils.io @ git+https://github.com/rcsb/py-rcsb_utils_io.git@branch_name"
+# to install a specific branch of a package from GitHub
+allow-direct-references = true
+
 [tool.hatch.build.targets.sdist]
 include = [
     "/rcsb/**",

{rcsb_workflow-0.69 → rcsb_workflow-0.71}/rcsb/workflow/bcif/task_functions.py

@@ -243,7 +243,7 @@ def getOutputContentType(contentType: str) -> str:
 
 def getHash(pdbId: str, contentType: str) -> str:
     """get upper/lowercase pdb id prior to getting hash"""
-    result = pdbId[1:3]
+    result = pdbId[-3:-1]
     if contentType == ContentTypeEnum.COMPUTATIONAL.value:
         result = os.path.join(pdbId[0:2], pdbId[-6:-4], pdbId[-4:-2])
     return result

{rcsb_workflow-0.69 → rcsb_workflow-0.71}/rcsb/workflow/cli/ChemSearchIndexExec.py

@@ -35,9 +35,12 @@ def main():
     okS = False
     if okI:
         # Store chemical indices for future use -
-        sftpHost = args.blHostName
+        if args.blHostName.startswith(("s3://", "minio://")):
+            remoteHost = args.blHostName
+        else:
+            remoteHost = "sftp://" + args.blHostName
         okS = ccidxWf.stashIndices(
-            "sftp://" + sftpHost,
+            remoteHost,
             args.blUploadPathUnsynced,
             bundleLabel=args.channel.upper(),
             userName=args.sftpUserName,