rcsb.workflow 0.62__tar.gz → 0.63__tar.gz

{rcsb_workflow-0.62 → rcsb_workflow-0.63}/HISTORY.txt

@@ -56,3 +56,5 @@
  4-Sep-2025 - V0.60 Update requirements and docker file
 10-Sep-2025 - V0.61 Return without error in bcif task functions, update content prefixes
 30-Sep-2025 - V0.62 Let jpg gen use model file over URL
+ 9-Sep-2025 - V0.63 Add ChemSearchIndex from weekly-update-workflow so it can be a cli call instead
+ 

{rcsb_workflow-0.62/rcsb.workflow.egg-info → rcsb_workflow-0.63}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rcsb.workflow
-Version: 0.62
+Version: 0.63
 Summary: RCSB Python data processing and ETL/ELT workflow entry points
 Home-page: https://github.com/rcsb/py-rcsb_workflow
 Author: John Westbrook

rcsb_workflow-0.63/rcsb/workflow/cli/ChemSearchIndexExec.py

@@ -0,0 +1,65 @@
+
+from rcsb.workflow.chem.ChemCompSearchIndexWorkflow import ChemCompSearchIndexWorkflow
+import argparse
+import logging
+
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s]-%(module)s.%(funcName)s: %(message)s")
+logger = logging.getLogger()
+
+
+def main():
+    """Build indices and configuration files required to support the chemical search service."""
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--cachePath", required=True, help="Cache path.")
+    parser.add_argument("--ccUrlTarget", required=True, help="Chem comp url target.")
+    parser.add_argument("--birdUrlTarget", required=True, help="Bird url target.")
+    parser.add_argument("--channel", required=True, help="Channel.")
+    parser.add_argument("--useCache", action="store_true", default=False, required=True, help="Should it use the cache.")
+    parser.add_argument("--ccFileNamePrefix", required=True, help="Chem comp file name prefix.")
+    parser.add_argument("--numProc", required=True, help="Number of procs for makeIndices.")
+    parser.add_argument("--blHostName", required=True, help="Build locker host name.")
+    parser.add_argument("--blUploadPathUnsynced", required=True, help="Build locker upload path unsynced.")
+    parser.add_argument("--sftpUserName", required=True, help="SFTP username.")
+    parser.add_argument("--sftpUserPassword", required=True, help="SFTP user password.")
+    args = parser.parse_args()
+
+    # Build chemical indices -
+    logger.info("Using cache path %r", args.cachePath)
+    logger.info("Using ccFileNamePrefix %s", args.ccFileNamePrefix)
+    ccidxWf = ChemCompSearchIndexWorkflow(cachePath=args.cachePath, ccFileNamePrefix=args.ccFileNamePrefix)
+    okI = ccidxWf.makeIndices(
+        args.ccUrlTarget, args.birdUrlTarget, numProc=int(args.numProc), useCache=args.useCache
+    )
+    logger.info("Chemical search index build completed with status %r", okI)
+    #
+    okS = False
+    if okI:
+        # Store chemical indices for future use -
+        sftpHost = args.blHostName
+        okS = ccidxWf.stashIndices(
+            "sftp://" + sftpHost,
+            args.blUploadPathUnsynced,
+            bundleLabel=args.channel.upper(),
+            userName=args.sftpUserName,
+            pw=args.sftpUserPassword,
+        )
+        logger.info("Chemical search index stashed with status %r", okS)
+    else:
+        msg = "Chemical search index completed with status %r" % okI
+        raise ValueError(msg)
+    #
+
+    if not okS:
+        error_msg = "Failed to stash chemical search index data"
+        raise ValueError(error_msg)
+
+    message = "Chemical Search Index completed with status %r stash status %r" % (okI, okS)
+    logger.info("Completion message is %s", message)
+
+
+if __name__ == "__main__":
+    try:
+        main()
+    except Exception as err:
+        fail_msg = f"Run failed {err}"
+        raise RuntimeError(fail_msg) from err

{rcsb_workflow-0.62 → rcsb_workflow-0.63}/rcsb/workflow/cli/__init__.py

@@ -2,4 +2,4 @@ __docformat__ = "google en"
 __author__ = "John Westbrook"
 __email__ = "john.westbrook@rcsb.org"
 __license__ = "Apache 2.0"
-__version__ = "0.62"
+__version__ = "0.63"

{rcsb_workflow-0.62 → rcsb_workflow-0.63/rcsb.workflow.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rcsb.workflow
-Version: 0.62
+Version: 0.63
 Summary: RCSB Python data processing and ETL/ELT workflow entry points
 Home-page: https://github.com/rcsb/py-rcsb_workflow
 Author: John Westbrook

{rcsb_workflow-0.62 → rcsb_workflow-0.63}/rcsb.workflow.egg-info/SOURCES.txt

@@ -21,6 +21,7 @@ rcsb/workflow/chem/ChemCompImageWorkflow.py
 rcsb/workflow/chem/ChemCompSearchIndexWorkflow.py
 rcsb/workflow/chem/__init__.py
 rcsb/workflow/cli/BcifExec.py
+rcsb/workflow/cli/ChemSearchIndexExec.py
 rcsb/workflow/cli/ExDbExec.py
 rcsb/workflow/cli/ImgExec.py
 rcsb/workflow/cli/__init__.py

{rcsb_workflow-0.62 → rcsb_workflow-0.63}/rcsb.workflow.egg-info/entry_points.txt

@@ -1,4 +1,5 @@
 [console_scripts]
 bcif_exec_cli = rcsb.workflow.cli.BcifExec:main
+chem_search_index_cli = rcsb.workflow.cli.ChemSearchIndexExec:main
 exdb_wf_cli = rcsb.workflow.cli.ExDbExec:main
 imgs_exec_cli = rcsb.workflow.cli.ImgExec:main

{rcsb_workflow-0.62 → rcsb_workflow-0.63}/setup.py

@@ -49,7 +49,14 @@ setup(
         "Programming Language :: Python :: 3.10",
     ],
     # entry_points={"console_scripts": ["cactvs_annotate_mol=rcsb.workflow.cactvsAnnotateMol:main"]},
-    entry_points={"console_scripts": ["exdb_wf_cli=rcsb.workflow.cli.ExDbExec:main", "imgs_exec_cli=rcsb.workflow.cli.ImgExec:main", "bcif_exec_cli=rcsb.workflow.cli.BcifExec:main"]},
+    entry_points={
+        "console_scripts": [
+            "exdb_wf_cli=rcsb.workflow.cli.ExDbExec:main",
+            "imgs_exec_cli=rcsb.workflow.cli.ImgExec:main",
+            "bcif_exec_cli=rcsb.workflow.cli.BcifExec:main",
+            "chem_search_index_cli=rcsb.workflow.cli.ChemSearchIndexExec:main"
+        ]
+    },
     #  The following is somewhat flakey --
     # dependency_links=[],
     install_requires=packagesRequired[1:],