pulumi/sdk/python/cmd/pulumi-language-python-exec

207 lines
9.8 KiB
Plaintext
Raw Permalink Normal View History

Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
#!/usr/bin/env python
# Copyright 2016-2018, Pulumi Corporation. All rights reserved.
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
import argparse
import asyncio
from typing import Optional
import logging
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
import os
import sys
import traceback
import runpy
from concurrent.futures import ThreadPoolExecutor
# The user might not have installed Pulumi yet in their environment - provide a high-quality error message in that case.
try:
import pulumi
import pulumi.runtime
except ImportError:
# For whatever reason, sys.stderr.write is not picked up by the engine as a message, but 'print' is. The Python
# langhost automatically flushes stdout and stderr on shutdown, so we don't need to do it here - just trust that
# Python does the sane thing when printing to stderr.
print(traceback.format_exc(), file=sys.stderr)
print("""
It looks like the Pulumi SDK has not been installed. Have you run pip install?
If you are running in a virtualenv, you must run pip install -r requirements.txt from inside the virtualenv.""", file=sys.stderr)
sys.exit(1)
# use exit code 32 to signal to the language host that an error message was displayed to the user
PYTHON_PROCESS_EXITED_AFTER_SHOWING_USER_ACTIONABLE_MESSAGE_CODE = 32
def get_abs_module_path(mod_path):
path, ext = os.path.splitext(mod_path)
if not ext:
path = os.path.join(path, '__main__')
return os.path.abspath(path)
def _get_user_stacktrace(user_program_abspath: str) -> str:
'''grabs the current stacktrace and truncates it to show the only stacks pertaining to a user's program'''
tb = traceback.extract_tb(sys.exc_info()[2])
for frame_index, frame in enumerate(tb):
# loop over stack frames until we reach the main program
# then return the traceback truncated to the user's code
cur_module = frame[0]
if get_abs_module_path(user_program_abspath) == get_abs_module_path(cur_module):
# we have detected the start of a user's stack trace
remaining_frames = len(tb)-frame_index
# include remaining frames from the bottom by negating
return traceback.format_exc(limit=-remaining_frames)
# we did not detect a __main__ program, return normal traceback
return traceback.format_exc()
def _set_default_executor(loop, parallelism: Optional[int]):
'''configure this event loop to respect the settings provided.'''
if parallelism is None:
return
2022-10-22 22:54:31 +00:00
parallelism = max(parallelism, 1)
exec = ThreadPoolExecutor(max_workers=parallelism)
loop.set_default_executor(exec)
return exec
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
if __name__ == "__main__":
# Parse the arguments, program name, and optional arguments.
ap = argparse.ArgumentParser(description='Execute a Pulumi Python program')
ap.add_argument('--project', help='Set the project name')
ap.add_argument('--stack', help='Set the stack name')
ap.add_argument('--parallel', help='Run P resource operations in parallel (default=none)')
ap.add_argument('--dry_run', help='Simulate resource changes, but without making them')
ap.add_argument('--pwd', help='Change the working directory before running the program')
ap.add_argument('--monitor', help='An RPC address for the resource monitor to connect to')
ap.add_argument('--engine', help='An RPC address for the engine to connect to')
ap.add_argument('--tracing', help='A Zipkin-compatible endpoint to send tracing data to')
ap.add_argument('--organization', help='Set the organization name')
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
ap.add_argument('PROGRAM', help='The Python program to run')
ap.add_argument('ARGS', help='Arguments to pass to the program', nargs='*')
args = ap.parse_args()
# If any config variables are present, parse and set them, so subsequent accesses are fast.
config_env = pulumi.runtime.get_config_env()
if hasattr(pulumi.runtime, "get_config_secret_keys_env") and hasattr(pulumi.runtime, "set_all_config"):
# If the pulumi SDK has `get_config_secret_keys_env` and `set_all_config`, use them
# to set the config and secret keys.
config_secret_keys_env = pulumi.runtime.get_config_secret_keys_env()
pulumi.runtime.set_all_config(config_env, config_secret_keys_env)
else:
# Otherwise, fallback to setting individual config values.
for k, v in config_env.items():
pulumi.runtime.set_config(k, v)
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
# Configure the runtime so that the user program hooks up to Pulumi as appropriate.
# New versions of pulumi python support setting organization, old versions do not
try:
settings = pulumi.runtime.Settings(
monitor=args.monitor,
engine=args.engine,
project=args.project,
stack=args.stack,
parallel=int(args.parallel),
dry_run=args.dry_run == "true",
organization=args.organization,
)
except TypeError:
settings = pulumi.runtime.Settings(
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
monitor=args.monitor,
engine=args.engine,
project=args.project,
stack=args.stack,
parallel=int(args.parallel),
dry_run=args.dry_run == "true"
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
)
pulumi.runtime.configure(settings)
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
# Finally, swap in the args, chdir if needed, and run the program as if it had been executed directly.
sys.argv = [args.PROGRAM] + args.ARGS
if args.pwd is not None:
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
os.chdir(args.pwd)
successful = False
Clean up legacy Py 3.7 support (#15276) <!--- Thanks so much for your contribution! If this is your first time contributing, please ensure that you have read the [CONTRIBUTING](https://github.com/pulumi/pulumi/blob/master/CONTRIBUTING.md) documentation. --> # Description <!--- Please include a summary of the change and which issue is fixed. Please also include relevant motivation and context. --> We've bumped the minimum version on the package and haven't officially supported versions before 3.8 for a while now. This cleans up a load of code we had just for compat with 3.7 and lower. ## Checklist - [x] I have run `make tidy` to update any new dependencies - [x] I have run `make lint` to verify my code passes the lint check - [x] I have formatted my code using `gofumpt` <!--- Please provide details if the checkbox below is to be left unchecked. --> - [ ] I have added tests that prove my fix is effective or that my feature works <!--- User-facing changes require a CHANGELOG entry. --> - [x] I have run `make changelog` and committed the `changelog/pending/<file>` documenting my change <!-- If the change(s) in this PR is a modification of an existing call to the Pulumi Cloud, then the service should honor older versions of the CLI where this change would not exist. You must then bump the API version in /pkg/backend/httpstate/client/api.go, as well as add it to the service. --> - [ ] Yes, there are changes in this PR that warrants bumping the Pulumi Cloud API version <!-- @Pulumi employees: If yes, you must submit corresponding changes in the service repo. -->
2024-01-30 10:23:57 +00:00
try:
# The docs for get_running_loop are somewhat misleading because they state:
# This function can only be called from a coroutine or a callback. However, if the function is
# called from outside a coroutine or callback (the standard case when running `pulumi up`), the function
# raises a RuntimeError as expected and falls through to the exception clause below.
loop = asyncio.get_running_loop()
except RuntimeError:
loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop)
# Configure the event loop to respect the parallelism value provided as input.
executor = _set_default_executor(loop, settings.parallel)
# We are (unfortunately) suppressing the log output of asyncio to avoid showing to users some of the bad things we
# do in our programming model.
#
# Fundamentally, Pulumi is a way for users to build asynchronous dataflow graphs that, as their deployments
# progress, resolve naturally and eventually result in the complete resolution of the graph. If one node in the
# graph fails (i.e. a resource fails to create, there's an exception in an apply, etc.), part of the graph remains
# unevaluated at the time that we exit.
#
# asyncio abhors this. It gets very upset if the process terminates without having observed every future that we
# have resolved. If we are terminating abnormally, it is highly likely that we are not going to observe every single
# future that we have created. Furthermore, it's *harmless* to do this - asyncio logs errors because it thinks it
# needs to tell users that they're doing bad things (which, to their credit, they are), but we are doing this
# deliberately.
#
# In order to paper over this for our users, we simply turn off the logger for asyncio. Users won't see any asyncio
# error messages, but if they stick to the Pulumi programming model, they wouldn't be seeing any anyway.
logging.getLogger("asyncio").setLevel(logging.CRITICAL)
exit_code = 1
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
try:
# record the location of the user's program to return user tracebacks
user_program_abspath = os.path.abspath(args.PROGRAM)
def run():
try:
Support async inline programs (#15278) <!--- Thanks so much for your contribution! If this is your first time contributing, please ensure that you have read the [CONTRIBUTING](https://github.com/pulumi/pulumi/blob/master/CONTRIBUTING.md) documentation. --> # Description <!--- Please include a summary of the change and which issue is fixed. Please also include relevant motivation and context. --> This updates the typing annotations and semantics of Python Automation API inline programs. It's now defined to be a callable that returns an optional awaitable of `None`. This supports both existing synchronous functions (should just return `None`) and now also supports async functions that return an `Awaitable` that eventually resolves to `None`. ## Checklist - [x] I have run `make tidy` to update any new dependencies - [x] I have run `make lint` to verify my code passes the lint check - [x] I have formatted my code using `gofumpt` <!--- Please provide details if the checkbox below is to be left unchecked. --> - [x] I have added tests that prove my fix is effective or that my feature works <!--- User-facing changes require a CHANGELOG entry. --> - [x] I have run `make changelog` and committed the `changelog/pending/<file>` documenting my change <!-- If the change(s) in this PR is a modification of an existing call to the Pulumi Cloud, then the service should honor older versions of the CLI where this change would not exist. You must then bump the API version in /pkg/backend/httpstate/client/api.go, as well as add it to the service. --> - [ ] Yes, there are changes in this PR that warrants bumping the Pulumi Cloud API version <!-- @Pulumi employees: If yes, you must submit corresponding changes in the service repo. --> --------- Co-authored-by: Julien P <julien@caffeine.lu>
2024-01-29 16:10:13 +00:00
runpy.run_path(args.PROGRAM, run_name='__main__')
except ImportError as e:
def fix_module_file(m: str) -> str:
# Work around python 11 reporting "<frozen runpy>" rather
# than runpy.__file__ in the traceback.
return runpy.__file__ if m == "<frozen runpy>" else m
# detect if the main pulumi python program does not exist
stack_modules = [fix_module_file(f.filename) for f in traceback.extract_tb(e.__traceback__)]
unique_modules = set(module for module in stack_modules)
last_module_name = stack_modules[-1]
# we identify a missing program error if
# 1. the only modules in the stack trace are
# - `pulumi-language-python-exec`
# - `runpy`
# 2. the last function in the stack trace is in the `runpy` module
if unique_modules == {
__file__, # the language runtime itself
runpy.__file__,
} and last_module_name == runpy.__file__ :
# this error will only be hit when the user provides a directory
# the engine has a check to determine if the `main` file exists and will fail early
# if a language runtime receives a directory, it's the language's responsibility to determine
# whether the provided directory has a pulumi program
pulumi.log.error(f"unable to find main python program `__main__.py` in `{user_program_abspath}`")
sys.exit(PYTHON_PROCESS_EXITED_AFTER_SHOWING_USER_ACTIONABLE_MESSAGE_CODE)
else:
raise e
coro = pulumi.runtime.run_in_stack(run)
loop.run_until_complete(coro)
exit_code = 0
Get the empty Python program working This change gets enough of the Python SDK up and running that the empty Python program will work. Mostly just scaffolding, but the basic structure is now in place. The primary remaining work is to wire up resource creation to the gRPC interfaces. In summary: * The basic structure is as follows: - Everything goes into sdk/python/. - sdk/python/cmd/pulumi-langhost-python is a Go language host that simply knows how to spawn Python processes to run out entrypoint in response to requests by the engine. - sdk/python/cmd/pulumi-langhost-python-exec is a little Python shim that is invoked by the language host to run Python programs, and is responsible for setting up the minimal goo before we can do so (RPC connections and the like). - sdk/python/lib/ contains a Python Pip package suitable for PyPi. - In there, we have two packages: the root pulumi package that contains all of the basic Pulumi programming model abstractions, and pulumi.runtime, which contains the implementation of resource registration, RPC interfacing with the engine, and so on. * Add logic in our test framework to conditionalize on the language type and react accordingly. This will allow us to skip Yarn for Python projects and eventually run Pip if there's a requirements.txt. * Created the basic project structure, including all of the usual Make targets for installing into the proper places. * Building also runs Pylint and we are clean. There are a few other minor things in here: * Add an "empty" test for both Node.js and Python. These pass. * Fix an existing bug in plugin shutdown logic. At some point, we started waiting for stderr/stdout to flush before shutting down the plugin; but if certain failures happen "early" during the plugin launch process, these channels will never get initialized and so waiting for them deadlocks. * Recently we seem to have added logic to delete test temp directories if a failure happened during initialization of said temp directories. This is unfortunate, because you often need to look at the temp directory to see what failed. We already clean them up elsewhere after the full test completes successfully, so I don't think we need to be doing this, and I've removed it. Still many loose ends (config, resources, etc), but it's a start!
2018-01-13 18:29:34 +00:00
except pulumi.RunError as e:
pulumi.log.error(str(e))
except Exception:
error_msg = "Program failed with an unhandled exception:\n" + _get_user_stacktrace(user_program_abspath)
pulumi.log.error(error_msg)
exit_code = PYTHON_PROCESS_EXITED_AFTER_SHOWING_USER_ACTIONABLE_MESSAGE_CODE
finally:
executor.shutdown(wait=True)
loop.close()
sys.stdout.flush()
sys.stderr.flush()
sys.exit(exit_code)