Merge pull request #5 from esgf2-us/lukasz/dev

Error event stream and JSON-PATCH HTTP

Author: sturoscy-personal

.pre-commit-config.yaml

@@ -4,7 +4,7 @@ repos:
     hooks:
       - id: black
         args: [--line-length=145]
-        language_version: python3.10
+        language_version: python3.13
 
   - repo: https://github.com/pycqa/flake8
     rev: 7.0.0

pyproject.toml

@@ -13,6 +13,7 @@ package-mode = false
 boto3 = "^1.35.49"
 confluent-kafka = "^2.6.0"
 globus-sdk = "^3.46.0"
+jsonpatch = "^1.33"
 python = "^3.10"
 python-dotenv = "^1.0.1"
 

requirements.txt

@@ -7,10 +7,13 @@ charset-normalizer==3.4.0
 click==8.1.8
 confluent-kafka==2.6.0
 cryptography==43.0.3
+dotenv==0.9.9
 globus-sdk==3.46.0
 idna==3.10
 isort==6.0.0
 jmespath==1.0.1
+jsonpatch==1.33.0
+jsonpointer==3.0.0
 mypy-extensions==1.0.0
 packaging==24.2
 pathspec==0.12.1

src/consumer.py

@@ -21,9 +21,7 @@ def process_messages(self, messages):
                 data = json.loads(msg.value())
                 messages_data.append(data)
             except json.JSONDecodeError as e:
-                logging.error(
-                    f"Data deserialization error at offset {msg.offset()}: {e}."
-                )
+                logging.error(f"Data deserialization error at offset {msg.offset()}: {e}.")
                 return False
         return messages_data
 
@@ -39,16 +37,14 @@ def start(self):
                 logging.info(f"Consumed {len(messages)} messages")
                 first_msg = messages[0]
                 offset = 0 if not first_msg.offset() else first_msg.offset()
-                self.seek_partition = TopicPartition(
-                    first_msg.topic(), first_msg.partition(), offset
-                )
+                self.seek_partition = TopicPartition(first_msg.topic(), first_msg.partition(), offset)
 
                 messages_data = self.process_messages(messages)
                 if not messages_data:
                     self.consumer.seek(self.seek_partition)
                     continue
 
-                if not self.message_processor.ingest(messages_data):
+                if not self.message_processor.process_messages(messages_data):
                     self.consumer.seek(self.seek_partition)
                     continue
 

src/globus.py

@@ -1,13 +1,13 @@
 import logging
 import time
+import jsonpatch
 
-from globus_sdk import (AccessTokenAuthorizer, ConfidentialAppAuthClient,
-                        SearchClient)
+from globus_sdk import AccessTokenAuthorizer, ConfidentialAppAuthClient, SearchClient
 from globus_sdk.scopes import SearchScopes
 
 
 class ConsumerSearchClient:
-    def __init__(self, credentials, search_index):
+    def __init__(self, credentials, search_index, error_producer):
         confidential_client = ConfidentialAppAuthClient(
             client_id=credentials.get("client_id"),
             client_secret=credentials.get("client_secret"),
@@ -19,10 +19,11 @@ def __init__(self, credentials, search_index):
         authorizer = AccessTokenAuthorizer(search_tokens.get("access_token"))
         self.search_client = SearchClient(authorizer=authorizer)
         self.esgf_index = search_index
+        self.error_producer = error_producer
 
-    def convert_assets(self, item):
+    def convert_assets(self, assets):
         converted_assets = []
-        for key, value in item.get("assets").items():
+        for key, value in assets.items():
             converted_assets.append({"name": key} | value)
         return converted_assets
 
@@ -45,12 +46,7 @@ def ingest(self, messages_data):
         gmeta = []
         for data in messages_data:
             item = data.get("data").get("payload").get("item")
-            assets = item.get("assets")
-            assets_list = []
-            for name, asset in assets.items():
-                asset["name"] = name
-                assets_list.append(asset)
-            item["assets"] = assets_list
+            item["assets"] = item.get("assets")
             gmeta.append(self.gmetaentry(item))
 
         gmetalist = {"ingest_type": "GMetaList", "ingest_data": {"gmeta": gmeta}}
@@ -70,5 +66,95 @@ def ingest(self, messages_data):
             time.sleep(1)
         return True
 
+    def post(self, message_data):
+        item = message_data.get("data").get("payload").get("item")
+        globus_response = self.search_client.get_subject(self.esgf_index, item.get("id"))
+        if globus_response.data:
+            logging.info(f"Item with ID {item.get('id')} already exists in the index.")
+            self.error_producer.produce(
+                topic="esgf-local.errors",
+                key=item.get("id"),
+                value=f"Item with ID {item.get('id')} already exists in the index.",
+            )
+            return None
+        item["assets"] = self.convert_assets(item.get("assets"))
+        gmeta_entry = self.gmetaentry(item)
+        return gmeta_entry
+
+    def json_patch(self, message_data):
+        payload = message_data.get("data").get("payload")
+        item_id = payload.get("item_id")
+        globus_response = self.search_client.get_subject(self.esgf_index, item_id)
+        if not globus_response.data:
+            logging.info(f"Item with ID {item_id} does not exist in the index.")
+            self.error_producer.produce(
+                topic="esgf-local.errors",
+                key=item_id,
+                value=f"Item with ID {item_id} does not exist in the index.",
+            )
+            return None
+        gmeta_entry = jsonpatch.apply_patch(
+            globus_response.data.get("content"), payload.get("patch")
+        ) 
+        return gmeta_entry
+
     def delete(self, subject):
-        self.search_client.delete_subject(self.esgf_index, subject)
+        globus_response = self.search_client.get_subject(self.esgf_index, subject)
+        if globus_response.data:
+            self.search_client.delete_subject(self.esgf_index, subject)
+            return True
+        logging.info(f"Item with ID {subject} does not exist in the index.")
+        self.error_producer.produce(
+            topic="esgf-local.errors",
+            key=subject,
+            value=f"Item with ID {subject} does not exist in the index.",
+        )
+        return None
+
+    def process_message(self, message_data):
+        try:
+            payload = message_data.get("data").get("payload")
+            method = payload.get("method")
+            if method == "POST":
+                return self.post(message_data)
+            if method == "PUT":
+                return self.put(message_data)
+            if method == "JSON_PATCH" or method == "PATCH":
+                return self.json_patch(message_data)
+            if method == "MERGE_PATCH":
+                return self.merge_patch(message_data)
+            return None
+        except Exception as e:
+            logging.error(f"Error processing message data: {e}")
+            self.error_producer.produce(
+                topic="esgf-local.errors",
+                key=message_data.get("data").get("payload").get("item").get("id"),
+                value=str(e),
+            )
+            return None
+
+    def process_messages(self, messages_data):
+        gmeta = []
+        for message_data in messages_data:
+            entry = self.process_message(message_data)
+            if entry:
+                gmeta.append(entry)
+        if not gmeta:
+            return False
+
+        gmetalist = {"ingest_type": "GMetaList", "ingest_data": {"gmeta": gmeta}}
+
+        r = self.search_client.ingest(self.esgf_index, gmetalist)
+        task_id = r.get("task_id")
+
+        while True:
+            r = self.search_client.get_task(task_id)
+            state = r.get("state")
+            if state == "SUCCESS":
+                return True
+            if state == "FAILED":
+                logging.error(f"Ingestion task {task_id} failed")
+                logging.error(r.text)
+                return False
+            time.sleep(1)
+        return True

src/main.py

@@ -2,17 +2,21 @@
 
 from consumer import KafkaConsumerService
 from globus import ConsumerSearchClient
+from producer import KafkaProducer
 from settings.consumer import (event_stream, globus_search,
                                globus_search_client_credentials)
+from settings.producer import error_event_stream
 
 logging.basicConfig(
     format="%(asctime)s - %(levelname)s - %(message)s", level=logging.INFO
 )
 
 
 if __name__ == "__main__":
+    error_producer = KafkaProducer(config=error_event_stream.get("config"))
+
     message_processor = ConsumerSearchClient(
-        globus_search_client_credentials, globus_search.get("index")
+        globus_search_client_credentials, globus_search.get("index"), error_producer
     )
 
     consumer_service = KafkaConsumerService(

src/producer.py

@@ -0,0 +1,44 @@
+import logging
+from abc import ABC, abstractmethod
+
+import attr
+from confluent_kafka import Producer
+
+# Setup logger
+logger = logging.getLogger(__name__)
+
+
+@attr.s
+class BaseProducer(ABC):
+    @abstractmethod
+    def produce(self, topic, message):
+        pass
+
+
+class StdoutProducer(BaseProducer):
+    def produce(self, topic, message):
+        logger.info(f"message: {message}")
+
+
+class KafkaProducer(BaseProducer):
+    def __init__(self, config):
+        self.producer = Producer(config)
+        logger.info("KafkaProducer initialized")
+
+    def produce(self, topic, key, value):
+        delivery_reports = []
+
+        def delivery_report(err, msg):
+            if err is not None:
+                logger.error(f"Delivery failed for message {msg.key()}: {err}")
+            else:
+                logger.info(
+                    f"Message {msg.key()} successfully delivered to {msg.topic()} [{msg.partition()}] at offset {msg.offset()}"
+                )
+            delivery_reports.append((err, msg))
+
+        self.producer.produce(
+            topic=topic, key=key, value=value, callback=delivery_report
+        )
+        self.producer.flush()
+        return delivery_reports

src/settings/consumer.py

@@ -18,7 +18,7 @@
             "enable.auto.commit": False,
             "group.id": "westconsumer",
         },
-        "topics": ["esgf-local"],
+        "topics": ["esgf-local.transactions"],
     }
 else:
     event_stream = {
@@ -45,6 +45,4 @@
 }
 
 # ESGF2 Globus Search
-globus_search = {
-    "index": os.environ.get("GLOBUS_SEARCH_INDEX")
-}
+globus_search = {"index": os.environ.get("GLOBUS_SEARCH_INDEX")}

src/settings/producer.py

@@ -0,0 +1,39 @@
+
+import logging
+import os
+import socket
+
+from dotenv import load_dotenv
+
+from src.utils import load_access_control_policy
+
+# Load the .env file
+load_dotenv()
+
+# Suppress some kafka message streams
+logger = logging.getLogger("kafka")
+logger.setLevel(logging.WARN)
+
+run_environment = os.environ.get("RUN_ENVIRONMENT", None)
+
+
+# Kafka connection details
+if run_environment == "local":
+    error_event_stream = {
+        "config": {
+            "bootstrap.servers": "host.docker.internal:9092",
+            "client.id": socket.gethostname(),
+        },
+        "topic": "esgf-local.errors",
+    }
+else:
+    error_event_stream = {
+        "config": {
+            "bootstrap.servers": os.environ.get("BOOTSTRAP_SERVERS"),
+            "security.protocol": "SASL_SSL",
+            "sasl.mechanisms": "PLAIN",
+            "sasl.username": os.environ.get("CONFLUENT_CLOUD_USERNAME"),
+            "sasl.password": os.environ.get("CONFLUENT_CLOUD_PASSWORD"),
+        },
+        "topic": os.environ.get("TOPIC", "esgf2.data-challenges.01.errors"),
+    }