bhfutils 0.2.4__tar.gz → 0.2.7__tar.gz

{bhfutils-0.2.4 → bhfutils-0.2.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: bhfutils
-Version: 0.2.4
+Version: 0.2.7
 Summary: Utilities that are used by any spider of Behoof project
 Home-page: https://behoof.app/
 Author: Teplygin Vladimir

{bhfutils-0.2.4 → bhfutils-0.2.7}/bhfutils/crawler/pipelines.py

@@ -109,19 +109,51 @@ class KafkaPipeline(object):
                                          bytes=my_bytes,
                                          backups=my_backups)
 
+        # Security settings are optional: a spider running inside the VPC talks to the
+        # PLAINTEXT listener and defines none of them, while one running outside needs
+        # SASL_SSL. Defaults keep existing projects working without touching them.
+        producer_kwargs = dict(bootstrap_servers=settings['KAFKA_HOSTS'],
+                               retries=3,
+                               linger_ms=settings['KAFKA_PRODUCER_BATCH_LINGER_MS'],
+                               buffer_memory=settings['KAFKA_PRODUCER_BUFFER_BYTES'],
+                               value_serializer=lambda m: m.encode('utf-8'),
+                               max_request_size=settings['KAFKA_PRODUCER_MAX_REQUEST_SIZE'])
+
+        security_protocol = settings.get('KAFKA_SECURITY_PROTOCOL', 'PLAINTEXT')
+        if security_protocol != 'PLAINTEXT':
+            producer_kwargs['security_protocol'] = security_protocol
+        if 'SASL' in security_protocol:
+            producer_kwargs['sasl_mechanism'] = settings.get('KAFKA_SASL_MECHANISM', 'PLAIN')
+            producer_kwargs['sasl_plain_username'] = settings.get('KAFKA_SASL_USERNAME')
+            producer_kwargs['sasl_plain_password'] = settings.get('KAFKA_SASL_PASSWORD')
+        if 'SSL' in security_protocol:
+            # Left as None for a certificate issued by a public CA: kafka-python then
+            # falls back to the system trust store. Set it only for a private CA.
+            producer_kwargs['ssl_cafile'] = settings.get('KAFKA_SSL_CAFILE')
+
+        # Only passed when set explicitly, so the client keeps its own default: on for
+        # kafka-python 3.x, absent in 2.x. Verified working against the 3.4.1 broker
+        # both ways, so there is no reason to force either.
+        enable_idempotence = settings.get('KAFKA_PRODUCER_ENABLE_IDEMPOTENCE')
+        if enable_idempotence is not None and 'enable_idempotence' in KafkaProducer.DEFAULT_CONFIG:
+            producer_kwargs['enable_idempotence'] = enable_idempotence
+
+        # Tuning options come and go between client releases - kafka-python 3.x dropped
+        # buffer_memory, for one - and an unknown key makes the constructor fail outright.
+        # Drop what this client does not know about instead, and say so.
+        for key in [k for k in producer_kwargs if k not in KafkaProducer.DEFAULT_CONFIG]:
+            logger.warning("Kafka client does not support setting, ignoring it",
+                           extra={'setting': key})
+            producer_kwargs.pop(key)
+
         try:
-            producer = KafkaProducer(bootstrap_servers=settings['KAFKA_HOSTS'],
-                                     retries=3,
-                                     linger_ms=settings['KAFKA_PRODUCER_BATCH_LINGER_MS'],
-                                     buffer_memory=settings['KAFKA_PRODUCER_BUFFER_BYTES'],
-                                     value_serializer=lambda m: m.encode('utf-8'),
-                                     max_request_size=settings['KAFKA_PRODUCER_MAX_REQUEST_SIZE'])
+            producer = KafkaProducer(**producer_kwargs)
         except Exception as e:
-            logger.error("Unable to connect to Kafka in Pipeline"", raising exit flag.")
-            # this is critical so we choose to exit.
-            # exiting because this is a different thread from the crawlers
-            # and we want to ensure we can connect to Kafka when we boot
-            sys.exit(1)
+            # Raise rather than sys.exit(): this runs inside the Twisted reactor while
+            # Scrapy builds the pipelines, where SystemExit leaves the process hanging
+            # with nothing in the crawl log. An exception aborts the crawl visibly.
+            logger.error("Unable to connect to Kafka in Pipeline", extra={'error': str(e)})
+            raise
         topic_prefix = settings['KAFKA_TOPIC_PREFIX']
         use_base64 = settings['KAFKA_BASE_64_ENCODE']
 

{bhfutils-0.2.4 → bhfutils-0.2.7}/bhfutils/crawler/settings_template.py

@@ -22,6 +22,20 @@ KAFKA_PRODUCER_BATCH_LINGER_MS = 25  # 25 ms before flush
 KAFKA_PRODUCER_BUFFER_BYTES = 4 * 1024 * 1024  # 4MB before blocking
 KAFKA_PRODUCER_MAX_REQUEST_SIZE = 1024 * 1024 # 1MB
 
+# Security for the Kafka connection. PLAINTEXT is right for a spider running inside
+# the VPC, which reaches the broker's INTERNAL listener. A spider running anywhere
+# else must go through the EXTERNAL listener and set SASL_SSL plus credentials.
+KAFKA_SECURITY_PROTOCOL = 'PLAINTEXT'
+KAFKA_SASL_MECHANISM = 'PLAIN'
+KAFKA_SASL_USERNAME = None
+KAFKA_SASL_PASSWORD = None
+# Only needed for a certificate from a private CA; leave None for a public one.
+KAFKA_SSL_CAFILE = None
+
+# Leave None to keep whatever the installed kafka-python defaults to. Set True or False
+# only to override it deliberately.
+KAFKA_PRODUCER_ENABLE_IDEMPOTENCE = None
+
 ZOOKEEPER_ASSIGN_PATH = '/scrapy-cluster/crawler/'
 ZOOKEEPER_ID = 'all'
 ZOOKEEPER_HOSTS = 'localhost:2181'

{bhfutils-0.2.4 → bhfutils-0.2.7}/bhfutils.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: bhfutils
-Version: 0.2.4
+Version: 0.2.7
 Summary: Utilities that are used by any spider of Behoof project
 Home-page: https://behoof.app/
 Author: Teplygin Vladimir

{bhfutils-0.2.4 → bhfutils-0.2.7}/setup.py

@@ -43,7 +43,7 @@ if 'nosetests' in sys.argv[1:]:
 
 setup(
     name='bhfutils',
-    version='0.2.4',
+    version='0.2.7',
     description='Utilities that are used by any spider of Behoof project',
     long_description=readme(),
     long_description_content_type='text/x-rst',