"""
BuilderBase class is an abstract class that defines common
functions for any types of builders. Each type schema (script, compiler, spack)
is implemented as separate Builder which extends BuilderBase class.
"""
import datetime
import getpass
import json
import logging
import os
import re
import shutil
import socket
import stat
import uuid
from abc import ABC, abstractmethod
from pathlib import Path
from buildtest.buildsystem.checks import (
assert_range_check,
comparison_check,
contains_check,
exists_check,
file_count_check,
file_linecount_check,
file_regex_check,
is_dir_check,
is_file_check,
is_symlink_check,
linecount_check,
regex_check,
returncode_check,
runtime_check,
)
from buildtest.cli.compilers import BuildtestCompilers
from buildtest.defaults import BUILDTEST_EXECUTOR_DIR, console
from buildtest.exceptions import BuildTestError
from buildtest.scheduler.job import Job
from buildtest.scheduler.lsf import LSFJob
from buildtest.scheduler.pbs import PBSJob
from buildtest.scheduler.slurm import SlurmJob
from buildtest.schemas.defaults import schema_table
from buildtest.utils.command import BuildTestCommand
from buildtest.utils.file import (
create_dir,
is_file,
read_file,
resolve_path,
write_file,
)
from buildtest.utils.shell import Shell, is_csh_shell
from buildtest.utils.timer import Timer
from buildtest.utils.tools import deep_get
[docs]class BuilderBase(ABC):
"""The BuilderBase is an abstract class that implements common functions used for building and running test.
buildtest will create a builder object which resembles a test. The builder will contains metadata that is unique
to each builder that is captured in the report file upon completion of test.
"""
def __init__(
self,
name,
recipe,
buildspec,
executor,
buildexecutor,
testdir,
numprocs=None,
numnodes=None,
compiler=None,
):
"""The BuilderBase provides common functions for any builder. The builder
is an instance of BuilderBase. The initializer method will setup the builder
attributes based on input test by ``name`` parameter.
Args:
name (str): Name of test in buildspec recipe
recipe (str): The loaded test section from the buildspec file
buildspec (str): Full path to buildspec file
buildexecutor (:obj:`buildtest.executors.setup.BuildExecutor`): An instance of BuildExecutor class used for accessing executors
testdir (str): Test directory where tests are written. Must be full path on filesystem.
"""
self.name = name
self.compiler = compiler
self.cc = None
self.fc = None
self.cxx = None
self.cflags = None
self.cxxflags = None
self.fflags = None
self.ldflags = None
self.cppflags = None
self.metadata = {}
self.duration = 0
self.numprocs = numprocs
self.numnodes = numnodes
self._retry = None
self.timer = Timer()
# store state of builder which can be True/False. This value is changed by self.success(), self.failure()
self._buildstate = None
# The type must match the type of the builder
self.recipe = recipe
self.executor = executor
# For batch jobs this variable is an instance of Job class which would be one of the subclass
self.job = None
# Controls the state of the builder object, a complete job will set
# this value to True. A job cancellation or job failure in submission will set this to False
self._state = "PENDING"
# this value holds the 'status' property from the buildspec
self.status = None
# this value hosts the 'metrics' property from the buildspec
self.metrics = None
self.buildspec = buildspec
# used for storing output content
self._output = None
# used for storing error content
self._error = None
# strip .yml extension from file name
file_name = re.sub("[.](yml)", "", os.path.basename(buildspec))
self.testdir = os.path.join(testdir, self.executor, file_name, self.name)
self.logger = logging.getLogger(__name__)
self.logger.debug(f"Processing Buildspec File: {self.buildspec}")
self.logger.debug(f"Processing Test: {self.name}")
# get type attribute from Executor class (local, slurm, cobalt, pbs, lsf)
self.executor_type = buildexecutor.executors[self.executor].type
self.buildexecutor = buildexecutor
# Generate a unique id for the build based
self.testid = str(uuid.uuid4())
self._dependency = False
self._set_metadata_values()
self.shell_detection()
self.sched_init()
@property
def dependency(self):
return self._dependency
@dependency.setter
def dependency(self, state):
self._dependency = state
[docs] def shell_detection(self):
"""Detect shell and shebang used for test script"""
# if 'shell' property not defined in buildspec use this shell otherwise use the 'shell' property from the executor definition
self.shell = Shell(
self.recipe.get("shell")
or self.buildexecutor.executors[self.executor]._settings.get("shell")
or "bash"
)
# set shebang to value defined in Buildspec, if not defined then get one from Shell class
self.shebang = (
self.recipe.get("shebang") or f"{self.shell.shebang} {self.shell.opts}"
)
self.logger.debug("Using shell %s", self.shell.name)
self.logger.debug(f"Shebang used for test: {self.shebang}")
[docs] def get_test_extension(self):
"""Return the test extension, which depends on the shell type. By default we return `sh`
file extension for all shells except for `csh` which will return "csh" extension.
Returns
str: Returns test extension name for generated test.
"""
# for csh we return shell extension 'csh'
if is_csh_shell(self.shell.name):
return "csh"
# default file extension
return "sh"
[docs] def is_local_executor(self):
"""Return True if current builder executor type is LocalExecutor otherwise returns False.
Returns:
bool: returns True if builder is using executor type LocalExecutor otherwise returns False
"""
# import issue when putting this at top of file
# from buildtest.executors.local import LocalExecutor
# return isinstance(self.buildexecutor.executors[self.executor], LocalExecutor)
return self.buildexecutor.executors[self.executor].type == "local"
[docs] def is_container_executor(self):
# from buildtest.executors.container import ContainerExecutor
return self.buildexecutor.executors[self.executor].type == "container"
[docs] def is_slurm_executor(self):
"""Return True if current builder executor type is LocalExecutor otherwise returns False.
Returns:
bool: returns True if builder is using executor type LocalExecutor otherwise returns False
"""
# import issue when putting this at top of file
from buildtest.executors.slurm import SlurmExecutor
return isinstance(self.buildexecutor.executors[self.executor], SlurmExecutor)
[docs] def is_batch_job(self):
"""Return True/False if builder.job attribute is of type Job instance if not returns False.
This method indicates if builder has a job submitted to queue
"""
return isinstance(self.job, Job)
[docs] def start(self):
"""Keep internal timer for test using class :class:`buildtest.utils.timer.Timer`. This method will start the timer for builder which is invoked upon running test."""
# self.timer = Timer()
self.timer.start()
[docs] def stop(self):
"""Stop internal timer for builder."""
# self.duration += self.timer.stop()
self.timer.stop()
[docs] def retry(self, retry):
self._retry = retry
[docs] def build(self, modules=None, modulepurge=None, unload_modules=None):
"""This method is responsible for invoking setup, creating test
directory and writing test. This method is called from an instance
object of this class that does ``builder.build()``.
args:
modules (str, optional): Specify a list of modules to load in the build script
modulepurge (bool, optional): A boolean to control whether 'module purge' is run before running test
unload_modules (str, optional): Specify a list of modules to unload in the build script
"""
self._build_setup()
self._write_test()
self._write_build_script(modules, modulepurge, unload_modules)
[docs] def run(self, cmd, timeout=None):
"""Run the test and record the starttime and start timer. We also return the instance
object of type BuildTestCommand which is used by Executors for processing output and error
Returns:
If success, the return type is an object of type :class:`buildtest.utils.command.BuildTestCommand`
If their is a failure (non-zero) returncode we retry test and if it doesn't pass we
raise exception of :class:`buildtest.exceptions.RuntimeFailure`
"""
self.metadata["command"] = cmd
console.print(f"[blue]{self}[/]: Current Working Directory : {os.getcwd()}")
# capture output of 'env' and write to file 'build-env.sh' prior to running test
command = BuildTestCommand("env")
command.execute()
content = "".join(command.get_output())
self.metadata["buildenv"] = os.path.join(self.test_root, "build-env.txt")
write_file(self.metadata["buildenv"], content)
console.print(f"[blue]{self}[/]: Running Test via command: [cyan]{cmd}[/cyan]")
self.record_starttime()
self.running()
self.start()
command = BuildTestCommand(cmd)
command.execute(timeout=timeout)
self.logger.debug(f"Running Test via command: {cmd}")
ret = command.returncode()
err_msg = command.get_error()
# limit error messages to 60 lines
if len(err_msg) >= 60:
err_msg = err_msg[-60:]
if not self._retry or ret == 0:
return command
console.print(f"[red]{self}: failed to submit job with returncode: {ret}")
console.rule(f"[red]Error Message for {self}")
console.print(f"[red]{' '.join(err_msg)}")
########## Retry for failed tests ##########
console.print(
f"[red]{self}: Detected failure in running test, will attempt to retry test: {self._retry} times"
)
for run in range(1, self._retry + 1):
print(f"{self}: Run - {run}/{self._retry}")
command = BuildTestCommand(cmd)
console.print(
f"[blue]{self}[/]: Running Test via command: [cyan]{cmd}[/cyan]"
)
command.execute(timeout=timeout)
self.logger.debug(f"Running Test via command: {cmd}")
ret = command.returncode()
# if we recieve a returncode of 0 return immediately with the instance of command
if ret == 0:
return command
console.print(f"[red]{self}: failed to submit job with returncode: {ret}")
return command
[docs] def record_starttime(self):
"""This method will record the starttime when job starts execution by using
``datetime.datetime.now()``
"""
self._starttime = datetime.datetime.now()
# this is recorded in the report file
self.metadata["result"]["starttime"] = self._starttime.strftime("%Y/%m/%d %X")
[docs] def record_endtime(self):
"""This method is called upon termination of job, we get current time using
``datetime.datetime.now()`` and calculate runtime of job
"""
self._endtime = datetime.datetime.now()
# this is recorded in the report file
self.metadata["result"]["endtime"] = self._endtime.strftime("%Y/%m/%d %X")
self.runtime()
[docs] def runtime(self):
"""Calculate runtime of job by calculating delta between endtime and starttime. The unit of measure
is seconds.
"""
runtime = self._endtime - self._starttime
self._runtime = runtime.total_seconds()
# for batch jobs we use the elapsed time for job to calculate the runtime
if self.is_batch_job():
self._runtime = self.job.elapsedtime
self.metadata["result"]["runtime"] = self._runtime
[docs] def get_runtime(self):
"""Return runtime of test"""
return self._runtime
[docs] def state(self):
return self._state
[docs] def failed(self):
"""Mark test as failure by updating the ``self._state``. A fail test will not be reported in test report"""
self._state = "FAILED"
[docs] def complete(self):
"""Mark test as complete by updating the ``self._state``. A complete test assumes test ran to completion"""
self._state = "COMPLETE"
[docs] def running(self):
self._state = "RUNNING"
[docs] def is_pending(self):
return self._state == "PENDING"
[docs] def is_complete(self):
"""If builder completes execution of test this method will return ``True`` otherwise returns ``False``.
A builder could fail due to job cancellation, failure to submit job or raise exception during the run phase.
In those case, this method will return ``False``."""
return self._state == "COMPLETE"
[docs] def is_failed(self):
"""Return True if builder fails to run test."""
return self._state == "FAILED"
[docs] def is_running(self):
"""Return True if builder fails to run test."""
return self._state == "RUNNING"
[docs] def copy_stage_files(self):
"""Copy output and error file into test root directory."""
shutil.copy2(
os.path.join(self.stage_dir, os.path.basename(self.metadata["outfile"])),
os.path.join(self.test_root, os.path.basename(self.metadata["outfile"])),
)
shutil.copy2(
os.path.join(self.stage_dir, os.path.basename(self.metadata["errfile"])),
os.path.join(self.test_root, os.path.basename(self.metadata["errfile"])),
)
# update outfile and errfile metadata records which show up in report file
self.metadata["outfile"] = os.path.join(
self.test_root, os.path.basename(self.metadata["outfile"])
)
self.metadata["errfile"] = os.path.join(
self.test_root, os.path.basename(self.metadata["errfile"])
)
[docs] def _build_setup(self):
"""This method is the setup operation to get ready to build test which
includes the following:
1. Creating Test directory and stage directory
2. Resolve full path to generated test script and build script
3. Copy all files from buildspec directory to stage directory
"""
create_dir(self.testdir)
# num_content = len(os.listdir(self.testdir))
# the testid is incremented for every run, this can be done by getting
# length of all files in testdir and creating a directory. Subsequent
# runs will increment this counter
self.test_root = os.path.join(self.testdir, self.testid[:8])
create_dir(self.test_root)
msg = f"Creating test directory: {self.test_root}"
self.logger.debug(msg)
self.metadata["testroot"] = self.test_root
self.stage_dir = os.path.join(self.test_root, "stage")
# create stage and run directories
create_dir(self.stage_dir)
msg = f"Creating the stage directory: {self.stage_dir}"
self.logger.debug(msg)
self.metadata["stagedir"] = self.stage_dir
# Derive the path to the test script
self.testpath = "%s.%s" % (
os.path.join(self.stage_dir, self.name),
self.get_test_extension(),
)
self.testpath = os.path.expandvars(self.testpath)
self.metadata["testpath"] = os.path.join(
self.test_root, os.path.basename(self.testpath)
)
self.build_script = f"{os.path.join(self.stage_dir, self.name)}_build.sh"
# copy all files relative to buildspec file into stage directory
for fname in Path(os.path.dirname(self.buildspec)).glob("*"):
if fname.is_dir():
shutil.copytree(
fname, os.path.join(self.stage_dir, os.path.basename(fname))
)
elif fname.is_file():
shutil.copy2(fname, self.stage_dir)
console.print(f"[blue]{self}[/]: Creating Test Directory: {self.test_root}")
[docs] def _write_build_script(self, modules=None, modulepurge=None, unload_modules=None):
"""This method will write the content of build script that is run for when invoking
the builder run method. Upon creating file we set permission of builder script to 755
so test can be run.
"""
lines = ["#!/bin/bash"]
trap_msg = """
# Function to handle all signals and perform cleanup
function cleanup() {
echo "Signal trapped. Performing cleanup before exiting."
exitcode=$?
echo "buildtest: command \`$BASH_COMMAND' failed (exit code: $exitcode)"
exit $exitcode
}
# Trap all signals and call the cleanup function
trap cleanup SIGINT SIGTERM SIGHUP SIGQUIT SIGABRT SIGKILL SIGALRM SIGPIPE SIGTERM SIGTSTP SIGTTIN SIGTTOU
"""
lines.append(trap_msg)
lines += self._default_test_variables()
lines.append("# source executor startup script")
if modulepurge:
lines.append("module purge")
if unload_modules:
lines.append("# Specify list of modules to unload")
for module in unload_modules.split(","):
lines.append(f"module unload {module}")
if modules:
lines.append("# Specify list of modules to load")
for module in modules.split(","):
lines.append(f"module load {module}")
lines += [
f"source {os.path.join(BUILDTEST_EXECUTOR_DIR, self.executor, 'before_script.sh')}"
]
lines.append("# Run generated script")
# local executor
if self.is_local_executor():
lines += [" ".join(self._emit_command())]
elif self.is_container_executor():
lines += self.get_container_invocation()
# lines += ["docker run -it --rm -v $PWD:$PWD -w $PWD " + f"-v {self.stage_dir}:/buildtest" + f" ubuntu bash -c {self.testpath}"]
# batch executor
else:
launcher = self.buildexecutor.executors[self.executor].launcher_command(
numprocs=self.numprocs, numnodes=self.numnodes
)
lines += [" ".join(launcher) + " " + f"{self.testpath}"]
lines.append("# Get return code")
# get returncode of executed script which is retrieved by '$?'
lines.append("returncode=$?")
lines.append("# Exit with return code")
lines.append("exit $returncode")
lines = "\n".join(lines)
write_file(self.build_script, lines)
self.metadata["buildscript_content"] = lines
self.logger.debug(f"Writing build script: {self.build_script}")
self._set_execute_perm(self.build_script)
# copying build script into test_root directory since stage directory will be removed
dest = os.path.join(self.test_root, os.path.basename(self.build_script))
shutil.copy2(self.build_script, dest)
self.logger.debug(f"Copying build script to: {dest}")
self.build_script = dest
self.metadata["build_script"] = self.build_script
# console.print(f"[blue]{self}:[/] Writing build script: {self.build_script}")
[docs] def _write_test(self):
"""This method is responsible for invoking ``generate_script`` that
formulates content of testscript which is implemented in each subclass.
Next we write content to file and apply 755 permission on script so
it has executable permission.
"""
# Implementation to write file generate.sh
lines = []
lines += self.generate_script()
lines = "\n".join(lines)
self.logger.info(f"Opening Test File for Writing: {self.testpath}")
write_file(self.testpath, lines)
self.metadata["test_content"] = lines
self._set_execute_perm(self.testpath)
# copy testpath to run_dir
shutil.copy2(
self.testpath, os.path.join(self.test_root, os.path.basename(self.testpath))
)
[docs] def get_container_invocation(self):
"""This method returns a list of lines containing the container invocation"""
lines = []
platform = self.buildexecutor.executors[self.executor]._settings.get("platform")
image = self.buildexecutor.executors[self.executor]._settings.get("image")
options = self.buildexecutor.executors[self.executor]._settings.get("options")
mounts = self.buildexecutor.executors[self.executor]._settings.get("mounts")
if platform in ["docker", "podman"]:
lines += [
f"{platform} run -it --rm -v {self.stage_dir}:/buildtest -w /buildtest"
]
if mounts:
lines += [f"-v {mounts}"]
if options:
lines += [f"{options}"]
lines += [
f"{image} bash -c {os.path.join('/buildtest', os.path.basename(self.testpath))}"
]
elif platform == "singularity":
lines += [f"{platform} exec -B {self.stage_dir}/buildtest"]
if mounts:
lines += [f"-B {mounts}"]
if options:
lines += [f"{options}"]
lines += [f"{image} {self.testpath}"]
return [" ".join(lines)]
[docs] def _emit_command(self):
"""This method will return a shell command used to invoke the script that is used for tests that
use local executors
Returns:
list: a list to show generated command used to run test.
Test can be run without any argument with path to script: ``/path/to/script.sh``
Test can be run with shell name followed by path to script: ``bash /path/to/script.sh``
Test can be run with shell name, shell options and path to script: ``bash -x /path/to/script.sh``
"""
# if not self.recipe.get("shell") or self.recipe.get("shell") == "python":
if self.recipe.get("shell") == "python":
return [self.testpath]
# if not self.recipe.get("shell"):
# return [self.shell.name, self.shell.default_opts, self.testpath]
# if not self.shell.opts:
# return [self.shell.name, self.testpath]
return [self.shell.name, self.shell.opts, self.testpath]
[docs] def _emit_set_command(self):
"""This method will emit the set command for strict mode that will exit immediately. In bash, zsh the command is ``set -eo pipefail``. For csh, tcsh and sh there is
no such command so we return empty string."""
if self.shell.name == "bash" or self.shell.name == "zsh":
return "set -eo pipefail"
return ""
[docs] def _default_test_variables(self):
"""Return a list of lines inserted in build script that define buildtest specific variables
that can be referenced when writing tests. The buildtest variables all start with BUILDTEST_*
"""
lines = [
f"export BUILDTEST_TEST_NAME={self.name}",
f"export BUILDTEST_TEST_ROOT={self.test_root}",
f"export BUILDTEST_BUILDSPEC_DIR={os.path.dirname(self.buildspec)}",
f"export BUILDTEST_STAGE_DIR={self.stage_dir}",
]
if self.numnodes:
lines.append(f"export BUILDTEST_NUMNODES={self.numnodes}")
if self.numprocs:
lines.append(f"export BUILDTEST_NUMPROCS={self.numprocs}")
return lines
[docs] def sched_init(self):
"""This method will resolve scheduler fields: 'sbatch', 'pbs', 'bsub', 'cobalt'"""
self.sbatch = deep_get(
self.recipe, "executors", self.executor, "sbatch"
) or self.recipe.get("sbatch")
self.bsub = deep_get(
self.recipe, "executors", self.executor, "bsub"
) or self.recipe.get("bsub")
self.pbs = deep_get(
self.recipe, "executors", self.executor, "pbs"
) or self.recipe.get("pbs")
self.cobalt = deep_get(
self.recipe, "executors", self.executor, "cobalt"
) or self.recipe.get("cobalt")
self.burstbuffer = self.recipe.get("BB") or deep_get(
self.recipe, "executors", self.executor, "BB"
)
self.datawarp = self.recipe.get("DW") or deep_get(
self.recipe, "executors", self.executor, "DW"
)
[docs] def get_job_directives(self):
"""This method returns a list of lines containing the scheduler directives"""
lines = []
if self.sbatch:
for line in self.sbatch:
lines.append(f"#SBATCH {line}")
lines += [f"#SBATCH --job-name={self.name}"]
lines += [f"#SBATCH --output={self.name}.out"]
lines += [f"#SBATCH --error={self.name}.err"]
if self.bsub:
for line in self.bsub:
lines.append(f"#BSUB {line}")
lines += [f"#BSUB -J {self.name}"]
lines += [f"#BSUB -o {self.name}.out"]
lines += [f"#BSUB -e {self.name}.err"]
if self.pbs:
for line in self.pbs:
lines.append(f"#PBS {line}")
lines.append(f"#PBS -N {self.name}")
if self.cobalt:
for line in self.cobalt:
lines.append(f"#COBALT {line}")
lines.append(f"#COBALT --jobname={self.name}")
return lines
[docs] def _get_burst_buffer(self, burstbuffer):
"""Get Burst Buffer directives (**#BB**) lines specified by ``BB`` property
Args:
burstbuffer (str): Burst Buffer configuration specified by ``BB`` property in buildspec
Returns:
list: List of string values containing containing ``#BB`` directives written in test
"""
if not burstbuffer:
return
lines = []
lines.append("####### START OF BURST BUFFER DIRECTIVES #######")
for arg in burstbuffer:
lines += [f"#BB {arg} "]
lines.append("####### END OF BURST BUFFER DIRECTIVES #######")
return lines
[docs] def _get_data_warp(self, datawarp):
"""Get Cray Data Warp directives (**#DW**) lines specified by ``DW`` property.
Args:
datawarp (str): Data Warp configuration specified by ``DW`` property in buildspec
Returns:
list: List of string values containing containing ``#DW`` directives written in test
"""
if not datawarp:
return
lines = []
lines.append("####### START OF DATAWARP DIRECTIVES #######")
for arg in datawarp:
lines += [f"#DW {arg}"]
lines.append("####### END OF DATAWARP DIRECTIVES #######")
return lines
[docs] def _set_execute_perm(self, fname):
"""Set permission to 755 for a given file. The filepath must be an absolute path to file"""
# Change permission of the file to executable
os.chmod(
fname,
stat.S_IRWXU | stat.S_IRGRP | stat.S_IXGRP | stat.S_IROTH | stat.S_IXOTH,
)
self.logger.debug(f"Changing permission to 755 for script: {fname}")
[docs] def _get_environment(self, env):
"""Retrieve a list of environment variables defined in buildspec and
return them as list with the shell equivalent command
Args:
env (dict): list of environment variables defined by ``env`` property in buildspec
"""
lines = []
# if env not defined return immediately
if not env:
return
if not isinstance(env, dict):
raise BuildTestError(f"{env} must be a dict but got type: {type(env)}")
# bash, sh, zsh environment variable declaration is export KEY=VALUE
if re.fullmatch("(bash|sh|zsh|/bin/bash|/bin/sh|/bin/zsh)$", self.shell.name):
for k, v in env.items():
lines.append(f'export {k}="{v}"')
# tcsh, csh, environment variable declaration is setenv KEY VALUE
elif re.fullmatch("(tcsh|csh|/bin/tcsh|/bin/csh)$", self.shell.name):
for k, v in env.items():
lines.append(f'setenv {k} "{v}"')
else:
self.logger.warning(
f"{self.shell.name} is not supported, skipping environment variables."
)
return
return lines
[docs] def _get_variables(self, variables):
"""Retrieve a list of variables defined in buildspec and
return them as list with the shell equivalent command.
Args:
variables (dict): list of variable defined by ``vars`` property in buildspec
"""
lines = []
if not variables:
return
if not isinstance(variables, dict):
raise BuildTestError(
f"{variables} must be a dict but got type: {type(variables)}"
)
# bash, sh, zsh variable declaration is KEY=VALUE
if re.fullmatch("(bash|sh|zsh|/bin/bash|/bin/sh|/bin/zsh)$", self.shell.name):
for k, v in variables.items():
if v:
lines.append(f'{k}="{v}"')
else:
lines.append(f"{k}=")
# tcsh, csh variable declaration is set KEY=VALUE
elif re.fullmatch("(tcsh|csh|/bin/tcsh|/bin/csh)$", self.shell.name):
for k, v in variables.items():
if v:
lines.append(f'set {k}="{v}"')
else:
lines.append(f"set {k}=")
else:
self.logger.warning(
f"{self.shell.name} is not supported, skipping environment variables."
)
return
return lines
[docs] def add_metrics(self):
"""This method will update the metrics field stored in ``self.metadata['metrics']``. The ``metrics``
property can be defined in the buildspdec to assign value to a metrics name based on regular expression,
environment or variable assignment.
"""
if not self.metrics:
return
for key, metric in self.metrics.items():
# Default value of metric is an empty string
self.metadata["metrics"][key] = ""
regex = metric.get("regex")
file_regex = metric.get("file_regex")
self.metadata["metrics"][key] = ""
if regex:
stream = regex.get("stream")
content = self._output if stream == "stdout" else self._error
if regex.get("re") == "re.match":
match = re.match(regex["exp"], content, re.MULTILINE)
elif regex.get("re") == "re.fullmatch":
match = re.fullmatch(regex["exp"], content, re.MULTILINE)
else:
match = re.search(regex["exp"], content, re.MULTILINE)
if match:
try:
self.metadata["metrics"][key] = match.group(
regex.get("item", 0)
)
except IndexError:
self.logger.error(
f"Unable to fetch match group: {regex.get('item', 0)} for metric: {key}."
)
continue
elif file_regex:
fname = file_regex["file"]
if fname:
resolved_fname = resolve_path(fname)
if not is_file(resolved_fname):
msg = f"[blue]{self}[/]: Unable to resolve file path: {fname} for metric: {key}"
self.logger.error(msg)
console.print(msg, style="red")
continue
content = read_file(resolved_fname)
match = (
re.search(file_regex["exp"], content, re.MULTILINE)
if content
else None
)
if match:
try:
self.metadata["metrics"][key] = match.group(
file_regex.get("item", 0)
)
except IndexError:
self.logger.error(
f"Unable to fetch match group: {file_regex.get('item', 0)} for metric: {key}."
)
continue
self.metadata["metrics"][key] = str(self.metadata["metrics"][key])
[docs] def output(self):
"""Return output content"""
return self._output
[docs] def error(self):
"""Return error content"""
return self._error
[docs] @abstractmethod
def generate_script(self):
"""Build the testscript content implemented in each subclass"""
[docs] def post_run_steps(self):
"""This method is called after test is complete. This method will copy files from stage directory
such as output, error and test script. We will check state of test and mark job is complete.
"""
# ensure we are back in stage directory before processing. For batch jobs like Slurm the
# current working directory is changed to the submit line which can cause issues for file checks
os.chdir(self.stage_dir)
self._output = read_file(self.metadata["outfile"])
self._error = read_file(self.metadata["errfile"])
self.metadata["output"] = self._output
self.metadata["error"] = self._error
self.copy_stage_files()
# need these lines after self.copy_stage_files()
console.print(
f"[blue]{self}[/]: Test completed in {self.metadata['result']['runtime']} seconds with returncode: {self.metadata['result']['returncode']}"
)
console.print(
f"[blue]{self}[/]: Writing output file - [green1]{self.metadata['outfile']}"
)
console.print(
f"[blue]{self}[/]: Writing error file - [red3]{self.metadata['errfile']}"
)
self.add_metrics()
self.check_test_state()
# mark job is success if it finished all post run steps
self.complete()
[docs] def is_valid_metric(self, name):
if name not in list(self.metadata["metrics"].keys()):
return False
return True
[docs] def check_test_state(self):
"""This method is responsible for detecting state of test (PASS/FAIL) based on returncode or regular expression."""
self.metadata["result"]["state"] = "FAIL"
if self.metadata["result"]["returncode"] == 0:
self.metadata["result"]["state"] = "PASS"
# if status is defined in Buildspec, then check for returncode and regex
if self.status:
# if 'state' property is specified explicitly honor this value regardless of what is calculated
if self.status.get("state"):
self.metadata["result"]["state"] = self.status["state"]
return
if "returncode" in self.status:
self.metadata["check"]["returncode"] = returncode_check(self)
# check regex against output or error stream based on regular expression defined in status property. Return value is a boolean
if self.status.get("regex"):
self.metadata["check"]["regex"] = regex_check(self)
if self.status.get("runtime"):
self.metadata["check"]["runtime"] = runtime_check(self)
if self.status.get("file_regex"):
self.metadata["check"]["file_regex"] = file_regex_check(self)
if self.status.get("slurm_job_state") and isinstance(self.job, SlurmJob):
self.metadata["check"]["slurm_job_state"] = (
self.status["slurm_job_state"] == self.job.state()
)
if self.status.get("pbs_job_state") and isinstance(self.job, PBSJob):
self.metadata["check"]["pbs_job_state"] = (
self.status["pbs_job_state"] == self.job.state()
)
if self.status.get("lsf_job_state") and isinstance(self.job, LSFJob):
self.metadata["check"]["lsf_job_state"] = (
self.status["lsf_job_state"] == self.job.state()
)
if self.status.get("assert_ge"):
self.metadata["check"]["assert_ge"] = comparison_check(
builder=self, comparison_type="ge"
)
if self.status.get("assert_le"):
self.metadata["check"]["assert_le"] = comparison_check(
builder=self, comparison_type="le"
)
if self.status.get("assert_gt"):
self.metadata["check"]["assert_gt"] = comparison_check(
builder=self, comparison_type="gt"
)
if self.status.get("assert_lt"):
self.metadata["check"]["assert_lt"] = comparison_check(
builder=self, comparison_type="lt"
)
if self.status.get("assert_eq"):
self.metadata["check"]["assert_eq"] = comparison_check(
builder=self, comparison_type="eq"
)
if self.status.get("assert_ne"):
self.metadata["check"]["assert_ne"] = comparison_check(
builder=self, comparison_type="ne"
)
if self.status.get("assert_range"):
self.metadata["check"]["assert_range"] = assert_range_check(self)
if self.status.get("contains"):
self.metadata["check"]["contains"] = contains_check(
builder=self, comparison_type="contains"
)
if self.status.get("not_contains"):
self.metadata["check"]["not_contains"] = contains_check(
builder=self, comparison_type="not_contains"
)
if self.status.get("is_symlink"):
self.metadata["check"]["is_symlink"] = is_symlink_check(builder=self)
if self.status.get("exists"):
self.metadata["check"]["exists"] = exists_check(builder=self)
if self.status.get("is_dir"):
self.metadata["check"]["is_dir"] = is_dir_check(builder=self)
if self.status.get("is_file"):
self.metadata["check"]["is_file"] = is_file_check(builder=self)
if self.status.get("file_count"):
self.metadata["check"]["file_count"] = file_count_check(builder=self)
if self.status.get("linecount"):
self.metadata["check"]["linecount"] = linecount_check(builder=self)
if self.status.get("file_linecount"):
self.metadata["check"]["file_linecount"] = file_linecount_check(
builder=self
)
# filter out any None values from status check
status_checks = [
value for value in self.metadata["check"].values() if value is not None
]
state = (
all(status_checks)
if self.status.get("mode") in ["AND", "and"]
else any(status_checks)
)
self.metadata["result"]["state"] = "PASS" if state else "FAIL"
[docs] def _process_compiler_config(self):
"""This method is responsible for setting cc, fc, cxx class variables based
on compiler selection. The order of precedence is ``config``, ``default``,
then buildtest setting. Compiler settings in 'config' takes highest precedence,
this overrides any configuration in 'default'. Finally we resort to compiler
configuration in buildtest setting if none defined. This method is responsible
for setting cc, fc, cxx, cflags, cxxflags, fflags, ldflags, and cppflags.
"""
bc = BuildtestCompilers(configuration=self.configuration)
self.compiler_group = bc.compiler_name_to_group[self.compiler]
self.logger.debug(
f"[{self.name}]: compiler: {self.compiler} belongs to compiler group: {self.compiler_group}"
)
# compiler from buildtest settings
self.bc_compiler = self.configuration.target_config["compilers"]["compiler"][
self.compiler_group
][self.compiler]
self.logger.debug(self.bc_compiler)
# set compiler values based on 'default' property in buildspec. This can override
# compiler setting defined in configuration file. If default is not set we load from buildtest settings for appropriate compiler.
# set compiler variables to ones defined in buildtest configuration
self.cc = self.bc_compiler["cc"]
self.cxx = self.bc_compiler["cxx"]
self.fc = self.bc_compiler["fc"]
self.logger.debug(
f"[{self.name}]: Compiler setting for {self.compiler} from configuration file"
)
self.logger.debug(
f"[{self.name}]: {self.compiler}: {json.dumps(self.bc_compiler, indent=2)}"
)
# if default compiler setting provided in buildspec let's assign it.
if deep_get(self.compiler_section, "default", self.compiler_group):
self.cc = (
self.compiler_section["default"][self.compiler_group].get("cc")
or self.cc
)
self.fc = (
self.compiler_section["default"][self.compiler_group].get("fc")
or self.fc
)
self.cxx = (
self.compiler_section["default"][self.compiler_group].get("cxx")
or self.cxx
)
self.cflags = (
self.compiler_section["default"][self.compiler_group].get("cflags")
or self.cflags
)
self.cxxflags = (
self.compiler_section["default"][self.compiler_group].get("cxxflags")
or self.cxxflags
)
self.fflags = (
self.compiler_section["default"][self.compiler_group].get("fflags")
or self.fflags
)
self.ldflags = (
self.compiler_section["default"][self.compiler_group].get("ldflags")
or self.ldflags
)
self.cppflags = (
self.compiler_section["default"][self.compiler_group].get("cppflags")
or self.cppflags
)
# if compiler instance defined in config section read from buildspec. This overrides default section if specified
if deep_get(self.compiler_section, "config", self.compiler):
self.logger.debug(
f"[{self.name}]: Detected compiler: {self.compiler} in 'config' scope overriding default compiler group setting for: {self.compiler_group}"
)
self.cc = (
self.compiler_section["config"][self.compiler].get("cc") or self.cc
)
self.fc = (
self.compiler_section["config"][self.compiler].get("fc") or self.fc
)
self.cxx = (
self.compiler_section["config"][self.compiler].get("cxx") or self.cxx
)
self.cflags = (
self.compiler_section["config"][self.compiler].get("cflags")
or self.cflags
)
self.cxxflags = (
self.compiler_section["config"][self.compiler].get("cxxflags")
or self.cxxflags
)
self.fflags = (
self.compiler_section["config"][self.compiler].get("fflags")
or self.fflags
)
self.cppflags = (
self.compiler_section["config"][self.compiler].get("cppflags")
or self.cppflags
)
self.ldflags = (
self.compiler_section["config"][self.compiler].get("ldflags")
or self.ldflags
)
self.logger.debug(
f"cc: {self.cc}, cxx: {self.cxx} fc: {self.fc} cppflags: {self.cppflags} cflags: {self.cflags} fflags: {self.fflags} ldflags: {self.ldflags}"
)
# this condition is a safety check before compiling code to ensure if all C, C++, Fortran compiler not set we raise error
if not self.cc and not self.cxx and not self.fc:
raise BuildTestError(
"Unable to set C, C++, and Fortran compiler wrapper, please specify 'cc', 'cxx','fc' in your compiler settings in buildtest configuration or specify in buildspec file. "
)
[docs] def __str__(self):
return f"{self.name}/{self.metadata['id']}"
[docs] def __repr__(self):
return self.__str__()