Switch to py4j

.gitignore

@@ -1 +1,2 @@
 __pycache__
+bin/Lomikel*.jar

apps/utils/client.py

@@ -14,10 +14,9 @@
 # limitations under the License.
 """Utilities to work with the Fink HBase client"""
 
+from py4j.java_gateway import JavaGateway
 import os
 
-import jpype
-import jpype.imports
 import numpy as np
 import yaml
 
@@ -26,24 +25,6 @@
 from line_profiler import profile
 
 
-@profile
-def initialise_jvm(path=None):
-    """Start a JVM
-
-    Parameters
-    ----------
-    path: str, optional
-        Path to the HBase client. Default is relative to apps/
-    """
-    if not jpype.isJVMStarted():
-        if path is None:
-            path = os.path.dirname(apps_loc) + "/../bin/FinkBrowser.exe.jar"
-        jarpath = f"-Djava.class.path={path}"
-        jpype.startJVM(jpype.getDefaultJVMPath(), "-ea", jarpath, convertStrings=True)
-
-    jpype.attachThreadToJVM()
-
-
 @profile
 def connect_to_hbase_table(
     tablename: str,
@@ -62,35 +43,24 @@ def connect_to_hbase_table(
         Name of the rowkey in the table containing the schema. Default is given by the config file.
     nlimit: int, optional
         Maximum number of objects to return. Default is 10000
-    setphysicalrepo: bool, optional
-        If True, store cutouts queried on disk ("/tmp/Lomikel/HBaseClientBinaryDataRepository")
-        Needs client 02.01+. Default is False
     config_path: str, optional
         Path to the config file. Default is None (relative to the apps/ folder)
     """
-    initialise_jvm()
-
     if config_path is None:
         config_path = os.path.dirname(apps_loc) + "/../config.yml"
     args = yaml.load(
         open(config_path),
         yaml.Loader,
     )
 
-    import com.Lomikel.HBaser
-    from com.astrolabsoftware.FinkBrowser.Utils import Init
-
-    Init.init()
-
-    client = com.Lomikel.HBaser.HBaseClient(args["HBASEIP"], args["ZOOPORT"])
+    gateway = JavaGateway(auto_convert=True)
+    client = gateway.jvm.com.Lomikel.HBaser.HBaseClient(
+        args["HBASEIP"], args["ZOOPORT"]
+    )
 
     if schema_name is None:
         schema_name = args["SCHEMAVER"]
     client.connect(tablename, schema_name)
-    if setphysicalrepo:
-        import com.Lomikel.HBaser.FilesBinaryDataRepository
-
-        client.setRepository(com.Lomikel.HBaser.FilesBinaryDataRepository())
     client.setLimit(args["NLIMIT"])
 
     return client
@@ -131,21 +101,17 @@ def create_or_update_hbase_table(
     if len(np.unique(families)) != 1:
         raise NotImplementedError("`create_hbase_table` only accepts one family name")
 
-    initialise_jvm()
-
     if config_path is None:
         config_path = os.path.dirname(apps_loc) + "/../config.yml"
     args = yaml.load(
         open(config_path),
         yaml.Loader,
     )
 
-    import com.Lomikel.HBaser
-    from com.astrolabsoftware.FinkBrowser.Utils import Init
-
-    Init.init()
-
-    client = com.Lomikel.HBaser.HBaseClient(args["HBASEIP"], args["ZOOPORT"])
+    gateway = JavaGateway(auto_convert=True)
+    client = gateway.jvm.com.Lomikel.HBaser.HBaseClient(
+        args["HBASEIP"], args["ZOOPORT"]
+    )
 
     if create:
         # Create the table and connect without schema

apps/utils/decoding.py

@@ -14,6 +14,8 @@
 # limitations under the License.
 """Utilities to decode data from the HBase client"""
 
+from py4j.java_gateway import JavaGateway
+import json
 import pandas as pd
 import numpy as np
 
@@ -144,17 +146,12 @@ def format_hbase_output(
 @profile
 def hbase_to_dict(hbase_output):
     """Optimize hbase output TreeMap for faster conversion to DataFrame"""
-    # Naive Python implementation
-    # optimized = {i: dict(j) for i, j in hbase_output.items()}
-
-    # Here we assume JPype is already initialized
-    import json
-
-    from org.json import JSONObject
+    gateway = JavaGateway(auto_convert=True)
+    JSONObject = gateway.jvm.org.json.JSONObject
 
     # We do bulk export to JSON on Java side to avoid overheads of iterative access
     # and then parse it back to Dict in Python
-    optimized = json.loads(JSONObject(hbase_output).toString())
+    optimized = json.loads(JSONObject(str(hbase_output)).toString())
 
     return optimized
 

bin/download_client.sh

@@ -0,0 +1,25 @@
+#!/bin/bash
+# Copyright 2024 AstroLab Software
+# Author: Julien Peloton
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+BASEURL=https://hrivnac.web.cern.ch/Activities/Packages/Lomikel
+CLIENTVERSION=03.04.00x
+
+
+files="Lomikel-$CLIENTVERSION-ext.jar Lomikel-$CLIENTVERSION.exe.jar Lomikel-$CLIENTVERSION-HBase.jar Lomikel-$CLIENTVERSION-HBase.exe.jar Lomikel-$CLIENTVERSION.jar"
+
+for file in $files; do
+        echo $file
+        wget --directory-prefix=. $BASEURL/$file
+done

config.yml

@@ -8,6 +8,7 @@ CUTOUTAPIURL: http://localhost
 # HBase configuration
 HBASEIP: localhost
 ZOOPORT: 2183
+CLIENTVERSION: 03.04.00x
 
 # Table schema (schema_{fink_broker}_{fink_science})
 SCHEMAVER: schema_3.1_5.21.14

install/README.md

@@ -14,14 +14,58 @@ pip install -r requirements.txt
 
 Follow instructions in the [fink-cutout-api](https://github.com/astrolabsoftware/fink-cutout-api/blob/main/install/README.md).
 
+## Client installation and Fink gateway
+
+To access HBase tables, we use a client based on [Lomikel](https://github.com/hrivnac/Lomikel). To download the latest version of the client, go to `bin` and execute:
+
+```bash
+cd bin
+./download_client.sh
+```
+
+Do not forget to update the version in the `config.yml` file. Then install (as sudo) a new unit for systemd under `/etc/systemd/system/fink_gateway.service` (check the correct version numbers for JARs):
+
+```bash
+[Unit]
+Description=Start a JVM with Fink Java objects
+After=network.target
+
+[Service]
+User=almalinux
+Group=almalinux
+WorkingDirectory=/home/almalinux/fink-object-api/bin
+
+ExecStart=/bin/sh -c 'source /home/almalinux/.bashrc; exec java -cp "Lomikel-03.04.00x-HBase.exe.jar:py4j0.10.9.7.jar" com.Lomikel.Py4J.LomikelGatewayServer 2>&1 >> /tmp/fink_gateway.out'
+
+[Install]
+WantedBy=multi-user.target
+```
+
+Reload daemon and start the service:
+
+```bash
+sudo systemctl daemon-reload
+sudo systemctl start fink_gateway
+```
+
+Check carefuly the status:
+
+```bash
+sudo systemctl start fink_gateway
+```
+
+Note that having a JVM open all the time can lead to a memory leak, so it is probably wise to restart the service from time to time.
+
+
 ## Systemctl and gunicorn
 
-Install a new unit for systemd under `/etc/systemd/system/fink_object_api.service`:
+Install a new unit (as sudo) for systemd under `/etc/systemd/system/fink_object_api.service`:
 
 ```bash
 [Unit]
 Description=gunicorn daemon for fink_object_api
-After=network.target
+After=network.target fink_gateway.service fink_cutout_api.service
+Requires=fink_gateway.service fink_cutout_api.service
 
 [Service]
 User=almalinux
@@ -41,5 +85,4 @@ sudo systemctl daemon-reload
 sudo systemctl start fink_object_api
 ```
 
-
-You are ready to use the API!
+Note that this will automatically starts `fink_gateway.service` and `fink_cutout_api.service` if they were not started. You are ready to use the API!

requirements.txt

@@ -8,6 +8,6 @@ line_profiler
 requests
 pyarrow
 matplotlib
-JPype1
+py4j
 PyYAML
 pyspark