Initial import: grid-bot — grid trading bot for BTC-USDT on Cifra Markets

kronos-venv/lib/python3.12/site-packages/torch/distributed/elastic/metrics/__init__.py

@@ -0,0 +1,168 @@
+#!/usr/bin/env/python3
+# mypy: allow-untyped-defs
+
+# Copyright (c) Facebook, Inc. and its affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+"""Metrics API.
+
+**Overview**:
+
+The metrics API in torchelastic is used to publish telemetry metrics.
+It is designed to be used by torchelastic's internal modules to
+publish metrics for the end user with the goal of increasing visibility
+and helping with debugging. However you may use the same API in your
+jobs to publish metrics to the same metrics ``sink``.
+
+A ``metric`` can be thought of as timeseries data
+and is uniquely identified by the string-valued tuple
+``(metric_group, metric_name)``.
+
+torchelastic makes no assumptions about what a ``metric_group`` is
+and what relationship it has with ``metric_name``. It is totally up
+to the user to use these two fields to uniquely identify a metric.
+
+.. note:: The metric group ``torchelastic`` is reserved by torchelastic for
+          platform level metrics that it produces.
+          For instance torchelastic may output the latency (in milliseconds)
+          of a re-rendezvous operation from the agent as
+          ``(torchelastic, agent.rendezvous.duration.ms)``
+
+A sensible way to use metric groups is to map them to a stage or module
+in your job. You may also encode certain high level properties
+the job such as the region or stage (dev vs prod).
+
+**Publish Metrics**:
+
+Using torchelastic's metrics API is similar to using python's logging
+framework. You first have to configure a metrics handler before
+trying to add metric data.
+
+The example below measures the latency for the ``calculate()`` function.
+
+::
+
+  import time
+  import torch.distributed.elastic.metrics as metrics
+
+  # makes all metrics other than the one from "my_module" to go /dev/null
+  metrics.configure(metrics.NullMetricsHandler())
+  metrics.configure(metrics.ConsoleMetricsHandler(), "my_module")
+
+
+  def my_method():
+      start = time.time()
+      calculate()
+      end = time.time()
+      metrics.put_metric("calculate_latency", int(end - start), "my_module")
+
+You may also use the torch.distributed.elastic.metrics.prof` decorator
+to conveniently and succinctly profile functions
+
+::
+
+  # -- in module examples.foobar --
+
+  import torch.distributed.elastic.metrics as metrics
+
+  metrics.configure(metrics.ConsoleMetricsHandler(), "foobar")
+  metrics.configure(metrics.ConsoleMetricsHandler(), "Bar")
+
+
+  @metrics.prof
+  def foo():
+      pass
+
+
+  class Bar:
+      @metrics.prof
+      def baz():
+          pass
+
+``@metrics.prof`` will publish the following metrics
+::
+
+  <leaf_module or classname>.success - 1 if the function finished successfully
+  <leaf_module or classname>.failure - 1 if the function threw an exception
+  <leaf_module or classname>.duration.ms - function duration in milliseconds
+
+**Configuring Metrics Handler**:
+
+`torch.distributed.elastic.metrics.MetricHandler` is responsible for emitting
+the added metric values to a particular destination. Metric groups can be
+configured with different metric handlers.
+
+By default torchelastic emits all metrics to ``/dev/null``.
+By adding the following configuration metrics,
+``torchelastic`` and ``my_app`` metric groups will be printed out to
+console.
+
+::
+
+  import torch.distributed.elastic.metrics as metrics
+
+  metrics.configure(metrics.ConsoleMetricHandler(), group="torchelastic")
+  metrics.configure(metrics.ConsoleMetricHandler(), group="my_app")
+
+**Writing a Custom Metric Handler**:
+
+If you want your metrics to be emitted to a custom location, implement
+the `torch.distributed.elastic.metrics.MetricHandler` interface
+and configure your job to use your custom metric handler.
+
+Below is a toy example that prints the metrics to ``stdout``
+
+::
+
+  import torch.distributed.elastic.metrics as metrics
+
+
+  class StdoutMetricHandler(metrics.MetricHandler):
+      def emit(self, metric_data):
+          ts = metric_data.timestamp
+          group = metric_data.group_name
+          name = metric_data.name
+          value = metric_data.value
+          print(f"[{ts}][{group}]: {name}={value}")
+
+
+  metrics.configure(StdoutMetricHandler(), group="my_app")
+
+Now all metrics in the group ``my_app`` will be printed to stdout as:
+
+::
+
+  [1574213883.4182858][my_app]: my_metric=<value>
+  [1574213940.5237644][my_app]: my_metric=<value>
+
+"""
+
+from typing import Optional
+
+from .api import (  # noqa: F401
+    configure,
+    ConsoleMetricHandler,
+    get_elapsed_time_ms,
+    getStream,
+    MetricData,
+    MetricHandler,
+    MetricsConfig,
+    NullMetricHandler,
+    prof,
+    profile,
+    publish_metric,
+    put_metric,
+)
+
+
+def initialize_metrics(cfg: MetricsConfig | None = None):
+    pass
+
+
+try:
+    from torch.distributed.elastic.metrics.static_init import *  # type: ignore[import] # noqa: F401 F403
+except ModuleNotFoundError:
+    pass

kronos-venv/lib/python3.12/site-packages/torch/distributed/elastic/metrics/api.py

@@ -0,0 +1,216 @@
+#!/usr/bin/env python3
+# mypy: allow-untyped-defs
+
+# Copyright (c) Facebook, Inc. and its affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+import abc
+import time
+from collections import namedtuple
+from functools import wraps
+from typing_extensions import deprecated
+
+
+__all__ = [
+    "MetricsConfig",
+    "MetricHandler",
+    "ConsoleMetricHandler",
+    "NullMetricHandler",
+    "MetricStream",
+    "configure",
+    "getStream",
+    "prof",
+    "profile",
+    "put_metric",
+    "publish_metric",
+    "get_elapsed_time_ms",
+    "MetricData",
+]
+
+MetricData = namedtuple("MetricData", ["timestamp", "group_name", "name", "value"])
+
+
+class MetricsConfig:
+    __slots__ = ["params"]
+
+    def __init__(self, params: dict[str, str] | None = None):
+        self.params = params
+        if self.params is None:
+            self.params = {}
+
+
+class MetricHandler(abc.ABC):
+    @abc.abstractmethod
+    def emit(self, metric_data: MetricData):
+        pass
+
+
+class ConsoleMetricHandler(MetricHandler):
+    def emit(self, metric_data: MetricData):
+        print(
+            f"[{metric_data.timestamp}][{metric_data.group_name}]: {metric_data.name}={metric_data.value}"
+        )
+
+
+class NullMetricHandler(MetricHandler):
+    def emit(self, metric_data: MetricData):
+        pass
+
+
+class MetricStream:
+    def __init__(self, group_name: str, handler: MetricHandler):
+        self.group_name = group_name
+        self.handler = handler
+
+    def add_value(self, metric_name: str, metric_value: int):
+        self.handler.emit(
+            MetricData(time.time(), self.group_name, metric_name, metric_value)
+        )
+
+
+_metrics_map: dict[str, MetricHandler] = {}
+_default_metrics_handler: MetricHandler = NullMetricHandler()
+
+
+# pyre-fixme[9]: group has type `str`; used as `None`.
+def configure(handler: MetricHandler, group: str | None = None):
+    if group is None:
+        global _default_metrics_handler
+        # pyre-fixme[9]: _default_metrics_handler has type `NullMetricHandler`; used
+        #  as `MetricHandler`.
+        _default_metrics_handler = handler
+    else:
+        _metrics_map[group] = handler
+
+
+def getStream(group: str):
+    handler = _metrics_map.get(group, _default_metrics_handler)
+    return MetricStream(group, handler)
+
+
+def _get_metric_name(fn):
+    qualname = fn.__qualname__
+    split = qualname.split(".")
+    if len(split) == 1:
+        module = fn.__module__
+        if module:
+            return module.split(".")[-1] + "." + split[0]
+        else:
+            return split[0]
+    else:
+        return qualname
+
+
+def prof(fn=None, group: str = "torchelastic"):
+    r"""
+    @profile decorator publishes duration.ms, count, success, failure metrics for the function that it decorates.
+
+    The metric name defaults to the qualified name (``class_name.def_name``) of the function.
+    If the function does not belong to a class, it uses the leaf module name instead.
+
+    Usage
+
+    ::
+
+     @metrics.prof
+     def x():
+         pass
+
+
+     @metrics.prof(group="agent")
+     def y():
+         pass
+    """
+
+    def wrap(f):
+        @wraps(f)
+        def wrapper(*args, **kwargs):
+            key = _get_metric_name(f)
+            try:
+                start = time.time()
+                result = f(*args, **kwargs)
+                put_metric(f"{key}.success", 1, group)
+            except Exception:
+                put_metric(f"{key}.failure", 1, group)
+                raise
+            finally:
+                put_metric(f"{key}.duration.ms", get_elapsed_time_ms(start), group)  # type: ignore[possibly-undefined]
+            return result
+
+        return wrapper
+
+    if fn:
+        return wrap(fn)
+    else:
+        return wrap
+
+
+@deprecated("Deprecated, use `@prof` instead", category=FutureWarning)
+def profile(group=None):
+    """
+    @profile decorator adds latency and success/failure metrics to any given function.
+
+    Usage
+
+    ::
+
+     @metrics.profile("my_metric_group")
+     def some_function(<arguments>):
+    """
+
+    def wrap(func):
+        @wraps(func)
+        def wrapper(*args, **kwargs):
+            try:
+                start_time = time.time()
+                result = func(*args, **kwargs)
+                # pyrefly: ignore [bad-argument-type]
+                publish_metric(group, f"{func.__name__}.success", 1)
+            except Exception:
+                # pyrefly: ignore [bad-argument-type]
+                publish_metric(group, f"{func.__name__}.failure", 1)
+                raise
+            finally:
+                publish_metric(
+                    # pyrefly: ignore [bad-argument-type]
+                    group,
+                    f"{func.__name__}.duration.ms",
+                    get_elapsed_time_ms(start_time),  # type: ignore[possibly-undefined]
+                )
+            return result
+
+        return wrapper
+
+    return wrap
+
+
+def put_metric(metric_name: str, metric_value: int, metric_group: str = "torchelastic"):
+    """
+    Publish a metric data point.
+
+    Usage
+
+    ::
+
+     put_metric("metric_name", 1)
+     put_metric("metric_name", 1, "metric_group_name")
+    """
+    getStream(metric_group).add_value(metric_name, metric_value)
+
+
+@deprecated(
+    "Deprecated, use `put_metric(metric_group)(metric_name, metric_value)` instead",
+    category=FutureWarning,
+)
+def publish_metric(metric_group: str, metric_name: str, metric_value: int):
+    metric_stream = getStream(metric_group)
+    metric_stream.add_value(metric_name, metric_value)
+
+
+def get_elapsed_time_ms(start_time_in_seconds: float):
+    """Return the elapsed time in millis from the given start time."""
+    end_time = time.time()
+    return int((end_time - start_time_in_seconds) * 1000)