trustgraph-base 0.11.11__tar.gz

trustgraph-base-0.11.11/PKG-INFO

@@ -0,0 +1,17 @@
+Metadata-Version: 2.1
+Name: trustgraph-base
+Version: 0.11.11
+Summary: TrustGraph provides a means to run a pipeline of flexible AI processing components in a flexible means to achieve a processing pipeline.
+Home-page: https://github.com/trustgraph-ai/trustgraph
+Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.11.11.tar.gz
+Author: trustgraph.ai
+Author-email: security@trustgraph.ai
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: GNU General Public License v3 or later (GPLv3+)
+Classifier: Operating System :: OS Independent
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+Requires-Dist: pulsar-client
+Requires-Dist: prometheus-client
+
+See https://trustgraph.ai/

trustgraph-base-0.11.11/README.md

@@ -0,0 +1 @@
+See https://trustgraph.ai/

trustgraph-base-0.11.11/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

trustgraph-base-0.11.11/setup.py

@@ -0,0 +1,42 @@
+import setuptools
+import os
+import importlib
+
+with open("README.md", "r") as fh:
+    long_description = fh.read()
+
+# Load a version number module
+spec = importlib.util.spec_from_file_location(
+    'version', 'trustgraph/base_version.py'
+)
+version_module = importlib.util.module_from_spec(spec)
+spec.loader.exec_module(version_module)
+
+version = version_module.__version__
+
+setuptools.setup(
+    name="trustgraph-base",
+    version=version,
+    author="trustgraph.ai",
+    author_email="security@trustgraph.ai",
+    description="TrustGraph provides a means to run a pipeline of flexible AI processing components in a flexible means to achieve a processing pipeline.",
+    long_description=long_description,
+    long_description_content_type="text/markdown",
+    url="https://github.com/trustgraph-ai/trustgraph",
+    packages=setuptools.find_namespace_packages(
+        where='./',
+    ),
+    classifiers=[ 
+        "Programming Language :: Python :: 3",
+        "License :: OSI Approved :: GNU General Public License v3 or later (GPLv3+)",
+        "Operating System :: OS Independent",
+    ],
+    python_requires='>=3.8',
+    download_url = "https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v" + version + ".tar.gz",
+    install_requires=[
+        "pulsar-client",
+        "prometheus-client",
+    ],
+    scripts=[
+    ]
+)

trustgraph-base-0.11.11/trustgraph/base/__init__.py

@@ -0,0 +1,6 @@
+
+from . base_processor import BaseProcessor
+from . consumer import Consumer
+from . producer import Producer
+from . consumer_producer import ConsumerProducer
+

trustgraph-base-0.11.11/trustgraph/base/base_processor.py

@@ -0,0 +1,119 @@
+
+import os
+import argparse
+import pulsar
+import _pulsar
+import time
+from prometheus_client import start_http_server, Info
+
+from .. log_level import LogLevel
+
+class BaseProcessor:
+
+    default_pulsar_host = os.getenv("PULSAR_HOST", 'pulsar://pulsar:6650')
+
+    def __init__(self, **params):
+
+        self.client = None
+
+        if not hasattr(__class__, "params_metric"):
+            __class__.params_metric = Info(
+                'params', 'Parameters configuration'
+            )
+
+        # FIXME: Maybe outputs information it should not
+        __class__.params_metric.info({
+            k: str(params[k])
+            for k in params
+        })
+
+        pulsar_host = params.get("pulsar_host", self.default_pulsar_host)
+        log_level = params.get("log_level", LogLevel.INFO)
+
+        self.pulsar_host = pulsar_host
+
+        self.client = pulsar.Client(
+            pulsar_host,
+            logger=pulsar.ConsoleLogger(log_level.to_pulsar())
+        )
+
+    def __del__(self):
+
+        if self.client:
+            self.client.close()
+
+    @staticmethod
+    def add_args(parser):
+
+        parser.add_argument(
+            '-p', '--pulsar-host',
+            default=__class__.default_pulsar_host,
+            help=f'Pulsar host (default: {__class__.default_pulsar_host})',
+        )
+
+        parser.add_argument(
+            '-l', '--log-level',
+            type=LogLevel,
+            default=LogLevel.INFO,
+            choices=list(LogLevel),
+            help=f'Output queue (default: info)'
+        )
+
+        parser.add_argument(
+            '--metrics',
+            action=argparse.BooleanOptionalAction,
+            default=True,
+            help=f'Metrics enabled (default: true)',
+        )
+
+        parser.add_argument(
+            '-P', '--metrics-port',
+            type=int,
+            default=8000,
+            help=f'Pulsar host (default: 8000)',
+        )
+
+    def run(self):
+        raise RuntimeError("Something should have implemented the run method")
+
+    @classmethod
+    def start(cls, prog, doc):
+
+        parser = argparse.ArgumentParser(
+            prog=prog,
+            description=doc
+        )
+
+        cls.add_args(parser)
+
+        args = parser.parse_args()
+        args = vars(args)
+
+        print(args)
+
+        if args["metrics"]:
+            start_http_server(args["metrics_port"])
+
+        while True:
+
+            try:
+
+                p = cls(**args)
+                p.run()
+
+            except KeyboardInterrupt:
+                print("Keyboard interrupt.")
+                return
+
+            except _pulsar.Interrupted:
+                print("Pulsar Interrupted.")
+                return
+
+            except Exception as e:
+
+                print(type(e))
+
+                print("Exception:", e, flush=True)
+                print("Will retry...", flush=True)
+
+                time.sleep(4)

trustgraph-base-0.11.11/trustgraph/base/consumer.py

@@ -0,0 +1,107 @@
+
+from pulsar.schema import JsonSchema
+from prometheus_client import Histogram, Info, Counter, Enum
+import time
+
+from . base_processor import BaseProcessor
+from .. exceptions import TooManyRequests
+
+class Consumer(BaseProcessor):
+
+    def __init__(self, **params):
+
+        if not hasattr(__class__, "state_metric"):
+            __class__.state_metric = Enum(
+                'processor_state', 'Processor state',
+                states=['starting', 'running', 'stopped']
+            )
+            __class__.state_metric.state('starting')
+
+        __class__.state_metric.state('starting')
+
+        super(Consumer, self).__init__(**params)
+
+        input_queue = params.get("input_queue")
+        subscriber = params.get("subscriber")
+        input_schema = params.get("input_schema")
+
+        if input_schema == None:
+            raise RuntimeError("input_schema must be specified")
+
+        if not hasattr(__class__, "request_metric"):
+            __class__.request_metric = Histogram(
+                'request_latency', 'Request latency (seconds)'
+            )
+
+        if not hasattr(__class__, "pubsub_metric"):
+            __class__.pubsub_metric = Info(
+                'pubsub', 'Pub/sub configuration'
+            )
+
+        if not hasattr(__class__, "processing_metric"):
+            __class__.processing_metric = Counter(
+                'processing_count', 'Processing count', ["status"]
+            )
+
+        __class__.pubsub_metric.info({
+            "input_queue": input_queue,
+            "subscriber": subscriber,
+            "input_schema": input_schema.__name__,
+        })
+
+        self.consumer = self.client.subscribe(
+            input_queue, subscriber,
+            schema=JsonSchema(input_schema),
+        )
+
+    def run(self):
+
+        __class__.state_metric.state('running')
+
+        while True:
+
+            msg = self.consumer.receive()
+
+            try:
+
+                with __class__.request_metric.time():
+                    self.handle(msg)
+
+                # Acknowledge successful processing of the message
+                self.consumer.acknowledge(msg)
+
+                __class__.processing_metric.labels(status="success").inc()
+
+            except TooManyRequests:
+                self.consumer.negative_acknowledge(msg)
+                print("TooManyRequests: will retry")
+                __class__.processing_metric.labels(status="rate-limit").inc()
+                time.sleep(5)
+                continue
+                
+            except Exception as e:
+
+                print("Exception:", e, flush=True)
+
+                # Message failed to be processed
+                self.consumer.negative_acknowledge(msg)
+
+                __class__.processing_metric.labels(status="error").inc()
+
+    @staticmethod
+    def add_args(parser, default_input_queue, default_subscriber):
+
+        BaseProcessor.add_args(parser)
+
+        parser.add_argument(
+            '-i', '--input-queue',
+            default=default_input_queue,
+            help=f'Input queue (default: {default_input_queue})'
+        )
+
+        parser.add_argument(
+            '-s', '--subscriber',
+            default=default_subscriber,
+            help=f'Queue subscriber name (default: {default_subscriber})'
+        )
+

trustgraph-base-0.11.11/trustgraph/base/consumer_producer.py

@@ -0,0 +1,139 @@
+
+from pulsar.schema import JsonSchema
+from prometheus_client import Histogram, Info, Counter, Enum
+import time
+
+from . base_processor import BaseProcessor
+from .. exceptions import TooManyRequests
+
+# FIXME: Derive from consumer?  And producer?
+
+class ConsumerProducer(BaseProcessor):
+
+    def __init__(self, **params):
+
+        if not hasattr(__class__, "state_metric"):
+            __class__.state_metric = Enum(
+                'processor_state', 'Processor state',
+                states=['starting', 'running', 'stopped']
+            )
+            __class__.state_metric.state('starting')
+
+        __class__.state_metric.state('starting')
+
+        input_queue = params.get("input_queue")
+        output_queue = params.get("output_queue")
+        subscriber = params.get("subscriber")
+        input_schema = params.get("input_schema")
+        output_schema = params.get("output_schema")
+
+        if not hasattr(__class__, "request_metric"):
+            __class__.request_metric = Histogram(
+                'request_latency', 'Request latency (seconds)'
+            )
+
+        if not hasattr(__class__, "output_metric"):
+            __class__.output_metric = Counter(
+                'output_count', 'Output items created'
+            )
+
+        if not hasattr(__class__, "pubsub_metric"):
+            __class__.pubsub_metric = Info(
+                'pubsub', 'Pub/sub configuration'
+            )
+
+        if not hasattr(__class__, "processing_metric"):
+            __class__.processing_metric = Counter(
+                'processing_count', 'Processing count', ["status"]
+            )
+
+        __class__.pubsub_metric.info({
+            "input_queue": input_queue,
+            "output_queue": output_queue,
+            "subscriber": subscriber,
+            "input_schema": input_schema.__name__,
+            "output_schema": output_schema.__name__,
+        })
+
+        super(ConsumerProducer, self).__init__(**params)
+
+        if input_schema == None:
+            raise RuntimeError("input_schema must be specified")
+
+        if output_schema == None:
+            raise RuntimeError("output_schema must be specified")
+
+        self.producer = self.client.create_producer(
+            topic=output_queue,
+            schema=JsonSchema(output_schema),
+        )
+
+        self.consumer = self.client.subscribe(
+            input_queue, subscriber,
+            schema=JsonSchema(input_schema),
+        )
+
+    def run(self):
+
+        __class__.state_metric.state('running')
+
+        while True:
+
+            msg = self.consumer.receive()
+
+            try:
+
+                with __class__.request_metric.time():
+                    resp = self.handle(msg)
+
+                # Acknowledge successful processing of the message
+                self.consumer.acknowledge(msg)
+
+                __class__.processing_metric.labels(status="success").inc()
+
+            except TooManyRequests:
+                self.consumer.negative_acknowledge(msg)
+                print("TooManyRequests: will retry")
+                __class__.processing_metric.labels(status="rate-limit").inc()
+                time.sleep(5)
+                continue
+
+            except Exception as e:
+
+                print("Exception:", e, flush=True)
+
+                # Message failed to be processed
+                self.consumer.negative_acknowledge(msg)
+
+                __class__.processing_metric.labels(status="error").inc()
+
+    def send(self, msg, properties={}):
+        self.producer.send(msg, properties)
+        __class__.output_metric.inc()
+
+    @staticmethod
+    def add_args(
+            parser, default_input_queue, default_subscriber,
+            default_output_queue,
+    ):
+
+        BaseProcessor.add_args(parser)
+
+        parser.add_argument(
+            '-i', '--input-queue',
+            default=default_input_queue,
+            help=f'Input queue (default: {default_input_queue})'
+        )
+
+        parser.add_argument(
+            '-s', '--subscriber',
+            default=default_subscriber,
+            help=f'Queue subscriber name (default: {default_subscriber})'
+        )
+
+        parser.add_argument(
+            '-o', '--output-queue',
+            default=default_output_queue,
+            help=f'Output queue (default: {default_output_queue})'
+        )
+

trustgraph-base-0.11.11/trustgraph/base/producer.py

@@ -0,0 +1,55 @@
+
+from pulsar.schema import JsonSchema
+from prometheus_client import Info, Counter
+
+from . base_processor import BaseProcessor
+
+class Producer(BaseProcessor):
+
+    def __init__(self, **params):
+
+        output_queue = params.get("output_queue")
+        output_schema = params.get("output_schema")
+
+        if not hasattr(__class__, "output_metric"):
+            __class__.output_metric = Counter(
+                'output_count', 'Output items created'
+            )
+
+        if not hasattr(__class__, "pubsub_metric"):
+            __class__.pubsub_metric = Info(
+                'pubsub', 'Pub/sub configuration'
+            )
+
+        __class__.pubsub_metric.info({
+            "output_queue": output_queue,
+            "output_schema": output_schema.__name__,
+        })
+
+        super(Producer, self).__init__(**params)
+
+        if output_schema == None:
+            raise RuntimeError("output_schema must be specified")
+
+        self.producer = self.client.create_producer(
+            topic=output_queue,
+            schema=JsonSchema(output_schema),
+        )
+
+    def send(self, msg, properties={}):
+        self.producer.send(msg, properties)
+        __class__.output_metric.inc()
+
+    @staticmethod
+    def add_args(
+            parser, default_input_queue, default_subscriber,
+            default_output_queue,
+    ):
+
+        BaseProcessor.add_args(parser)
+
+        parser.add_argument(
+            '-o', '--output-queue',
+            default=default_output_queue,
+            help=f'Output queue (default: {default_output_queue})'
+        )

trustgraph-base-0.11.11/trustgraph/base_version.py

@@ -0,0 +1 @@
+__version__ = "0.11.11"

trustgraph-base-0.11.11/trustgraph/clients/base.py

@@ -0,0 +1,125 @@
+
+import pulsar
+import _pulsar
+import hashlib
+import uuid
+import time
+from pulsar.schema import JsonSchema
+
+from .. exceptions import *
+
+# Default timeout for a request/response.  In seconds.
+DEFAULT_TIMEOUT=300
+
+# Ugly
+ERROR=_pulsar.LoggerLevel.Error
+WARN=_pulsar.LoggerLevel.Warn
+INFO=_pulsar.LoggerLevel.Info
+DEBUG=_pulsar.LoggerLevel.Debug
+
+class BaseClient:
+
+    def __init__(
+            self, log_level=ERROR,
+            subscriber=None,
+            input_queue=None,
+            output_queue=None,
+            input_schema=None,
+            output_schema=None,
+            pulsar_host="pulsar://pulsar:6650",
+    ):
+
+        if input_queue == None: raise RuntimeError("Need input_queue")
+        if output_queue == None: raise RuntimeError("Need output_queue")
+        if input_schema == None: raise RuntimeError("Need input_schema")
+        if output_schema == None: raise RuntimeError("Need output_schema")
+
+        if subscriber == None:
+            subscriber = str(uuid.uuid4())
+
+        self.client = pulsar.Client(
+            pulsar_host,
+            logger=pulsar.ConsoleLogger(log_level),
+        )
+
+        self.producer = self.client.create_producer(
+            topic=input_queue,
+            schema=JsonSchema(input_schema),
+            chunking_enabled=True,
+        )
+
+        self.consumer = self.client.subscribe(
+            output_queue, subscriber,
+            schema=JsonSchema(output_schema),
+        )
+
+        self.input_schema = input_schema
+        self.output_schema = output_schema
+
+    def call(self, **args):
+
+        timeout = args.get("timeout", DEFAULT_TIMEOUT)
+
+        if "timeout" in args:
+            del args["timeout"]
+
+        id = str(uuid.uuid4())
+
+        r = self.input_schema(**args)
+
+        end_time = time.time() + timeout
+
+        self.producer.send(r, properties={ "id": id })
+
+        while time.time() < end_time:
+
+            try:
+                msg = self.consumer.receive(timeout_millis=2500)
+            except pulsar.exceptions.Timeout:
+                continue
+
+            mid = msg.properties()["id"]
+
+            if mid == id:
+
+                value = msg.value()
+
+                if value.error:
+
+                    self.consumer.acknowledge(msg)
+
+                    if value.error.type == "llm-error":
+                        raise LlmError(value.error.message)
+
+                    elif value.error.type == "too-many-requests":
+                        raise TooManyRequests(value.error.message)
+
+                    elif value.error.type == "ParseError":
+                        raise ParseError(value.error.message)
+
+                    else:
+
+                        raise RuntimeError(
+                            f"{value.error.type}: {value.error.message}"
+                        )
+
+                resp = msg.value()
+                self.consumer.acknowledge(msg)
+                return resp
+
+            # Ignore messages with wrong ID
+            self.consumer.acknowledge(msg)
+
+        raise TimeoutError("Timed out waiting for response")
+
+    def __del__(self):
+
+        if hasattr(self, "consumer"):
+            self.consumer.close()
+            
+        if hasattr(self, "producer"):
+            self.producer.flush()
+            self.producer.close()
+            
+        self.client.close()
+

trustgraph-base-0.11.11/trustgraph/clients/document_embeddings_client.py

@@ -0,0 +1,45 @@
+
+import _pulsar
+
+from .. schema import DocumentEmbeddingsRequest, DocumentEmbeddingsResponse
+from .. schema import document_embeddings_request_queue
+from .. schema import document_embeddings_response_queue
+from . base import BaseClient
+
+# Ugly
+ERROR=_pulsar.LoggerLevel.Error
+WARN=_pulsar.LoggerLevel.Warn
+INFO=_pulsar.LoggerLevel.Info
+DEBUG=_pulsar.LoggerLevel.Debug
+
+class DocumentEmbeddingsClient(BaseClient):
+
+    def __init__(
+            self, log_level=ERROR,
+            subscriber=None,
+            input_queue=None,
+            output_queue=None,
+            pulsar_host="pulsar://pulsar:6650",
+    ):
+
+        if input_queue == None:
+            input_queue = document_embeddings_request_queue
+
+        if output_queue == None:
+            output_queue = document_embeddings_response_queue
+            
+        super(DocumentEmbeddingsClient, self).__init__(
+            log_level=log_level,
+            subscriber=subscriber,
+            input_queue=input_queue,
+            output_queue=output_queue,
+            pulsar_host=pulsar_host,
+            input_schema=DocumentEmbeddingsRequest,
+            output_schema=DocumentEmbeddingsResponse,
+        )
+
+    def request(self, vectors, limit=10, timeout=300):
+        return self.call(
+            vectors=vectors, limit=limit, timeout=timeout
+        ).documents
+