def extract_runs_from_xml(self, sourcefilesTagList, global_required_files_pattern): ''' This function builds a list of SourcefileSets (containing filename with options). The files and their options are taken from the list of sourcefilesTags. ''' base_dir = self.benchmark.base_dir # runs are structured as sourcefile sets, one set represents one sourcefiles tag blocks = [] for index, sourcefilesTag in enumerate(sourcefilesTagList): sourcefileSetName = sourcefilesTag.get("name") matchName = sourcefileSetName or str(index) if self.benchmark.config.selected_sourcefile_sets \ and not any(util.wildcard_match(matchName, sourcefile_set) for sourcefile_set in self.benchmark.config.selected_sourcefile_sets): continue required_files_pattern = global_required_files_pattern.union( set(tag.text for tag in sourcefilesTag.findall('requiredfiles'))) # get lists of filenames task_def_files = self.get_task_def_files_from_xml(sourcefilesTag, base_dir) # get file-specific options for filenames fileOptions = util.get_list_from_xml(sourcefilesTag) propertyfile = util.text_or_none(util.get_single_child_from_xml(sourcefilesTag, PROPERTY_TAG)) # some runs need more than one sourcefile, # the first sourcefile is a normal 'include'-file, we use its name as identifier # for logfile and result-category all other files are 'append'ed. appendFileTags = sourcefilesTag.findall("append") currentRuns = [] for identifier in task_def_files: if identifier.endswith('.yml'): if appendFileTags: raise BenchExecException( "Cannot combine <append> and task-definition files in the same <tasks> tag.") run = self.create_run_from_task_definition( identifier, fileOptions, propertyfile, required_files_pattern) else: run = self.create_run_for_input_file( identifier, fileOptions, propertyfile, required_files_pattern, appendFileTags) if run: currentRuns.append(run) # add runs for cases without source files for run in sourcefilesTag.findall("withoutfile"): currentRuns.append(Run(run.text, [], fileOptions, self, propertyfile, required_files_pattern)) blocks.append(SourcefileSet(sourcefileSetName, index, currentRuns)) if self.benchmark.config.selected_sourcefile_sets: for selected in self.benchmark.config.selected_sourcefile_sets: if not any(util.wildcard_match(sourcefile_set.real_name, selected) for sourcefile_set in blocks): logging.warning( 'The selected tasks "%s" are not present in the input file, ' 'skipping them.', selected) return blocks
def extract_runs_from_xml(self, sourcefilesTagList, global_required_files_pattern): ''' This function builds a list of SourcefileSets (containing filename with options). The files and their options are taken from the list of sourcefilesTags. ''' # runs are structured as sourcefile sets, one set represents one sourcefiles tag blocks = [] for index, sourcefilesTag in enumerate(sourcefilesTagList): sourcefileSetName = sourcefilesTag.get("name") matchName = sourcefileSetName or str(index) if self.benchmark.config.selected_sourcefile_sets \ and not any(util.wildcard_match(matchName, sourcefile_set) for sourcefile_set in self.benchmark.config.selected_sourcefile_sets): continue required_files_pattern = set( tag.text for tag in sourcefilesTag.findall('requiredfiles')) # get lists of filenames tasks = self.get_tasks_from_xml(sourcefilesTag, self.benchmark.base_dir) # get file-specific options for filenames fileOptions = util.get_list_from_xml(sourcefilesTag) propertyfile = util.text_or_none( util.get_single_child_from_xml(sourcefilesTag, PROPERTY_TAG)) currentRuns = [] for identifier, sourcefiles in tasks: currentRuns.append( Run( identifier, sourcefiles, fileOptions, self, propertyfile, global_required_files_pattern.union( required_files_pattern))) blocks.append(SourcefileSet(sourcefileSetName, index, currentRuns)) if self.benchmark.config.selected_sourcefile_sets: for selected in self.benchmark.config.selected_sourcefile_sets: if not any( util.wildcard_match(sourcefile_set.real_name, selected) for sourcefile_set in blocks): logging.warning( 'The selected tasks "%s" are not present in the input file, ' 'skipping them.', selected) return blocks
def extract_runs_from_xml(self, sourcefilesTagList, global_required_files_pattern): """ This function builds a list of SourcefileSets (containing filename with options). The files and their options are taken from the list of sourcefilesTags. """ # runs are structured as sourcefile sets, one set represents one sourcefiles tag blocks = [] for index, sourcefilesTag in enumerate(sourcefilesTagList): sourcefileSetName = sourcefilesTag.get("name") matchName = sourcefileSetName or str(index) if self.benchmark.config.selected_sourcefile_sets and not any( util.wildcard_match(matchName, sourcefile_set) for sourcefile_set in self.benchmark.config.selected_sourcefile_sets ): continue required_files_pattern = set(tag.text for tag in sourcefilesTag.findall("requiredfiles")) # get lists of filenames sourcefiles = self.get_sourcefiles_from_xml(sourcefilesTag, self.benchmark.base_dir) # get file-specific options for filenames fileOptions = util.get_list_from_xml(sourcefilesTag) propertyfile = util.text_or_none(util.get_single_child_from_xml(sourcefilesTag, PROPERTY_TAG)) currentRuns = [] for sourcefile in sourcefiles: currentRuns.append( Run( sourcefile, fileOptions, self, propertyfile, global_required_files_pattern.union(required_files_pattern), ) ) blocks.append(SourcefileSet(sourcefileSetName, index, currentRuns)) if self.benchmark.config.selected_sourcefile_sets: for selected in self.benchmark.config.selected_sourcefile_sets: if not any(util.wildcard_match(sourcefile_set.real_name, selected) for sourcefile_set in blocks): logging.warning( 'The selected tasks "%s" are not present in the input file, ' "skipping them.", selected ) return blocks
def should_be_executed(self): return not self.benchmark.config.selected_run_definitions \ or any(util.wildcard_match(self.real_name, run_definition) for run_definition in self.benchmark.config.selected_run_definitions)
def __init__(self, benchmark_file, config, start_time): """ The constructor of Benchmark reads the source files, options, columns and the tool from the XML in the benchmark_file.. """ logging.debug("I'm loading the benchmark %s.", benchmark_file) self.config = config self.benchmark_file = benchmark_file self.base_dir = os.path.dirname(self.benchmark_file) # get benchmark-name self.name = os.path.basename(benchmark_file)[:-4] # remove ending ".xml" if config.name: self.name += "."+config.name self.start_time = start_time self.instance = time.strftime("%Y-%m-%d_%H%M", self.start_time) self.output_base_name = config.output_path + self.name + "." + self.instance self.log_folder = self.output_base_name + ".logfiles" + os.path.sep self.log_zip = self.output_base_name + ".logfiles.zip" self.result_files_folder = self.output_base_name + ".files" # parse XML try: rootTag = ElementTree.ElementTree().parse(benchmark_file) except ElementTree.ParseError as e: sys.exit('Benchmark file {} is invalid: {}'.format(benchmark_file, e)) if 'benchmark' != rootTag.tag: sys.exit("Benchmark file {} is invalid: " "It's root element is not named 'benchmark'.".format(benchmark_file)) # get tool tool_name = rootTag.get('tool') if not tool_name: sys.exit('A tool needs to be specified in the benchmark definition file.') (self.tool_module, self.tool) = load_tool_info(tool_name) self.tool_name = self.tool.name() # will be set from the outside if necessary (may not be the case in SaaS environments) self.tool_version = None self.executable = None self.display_name = rootTag.get('displayName') logging.debug("The tool to be benchmarked is %s.", self.tool_name) def parse_memory_limit(value): try: value = int(value) logging.warning( 'Value "%s" for memory limit interpreted as MB for backwards compatibility, ' 'specify a unit to make this unambiguous.', value) return value * _BYTE_FACTOR * _BYTE_FACTOR except ValueError: return util.parse_memory_value(value) def handle_limit_value(name, key, cmdline_value, parse_fn): value = rootTag.get(key, None) # override limit from XML with values from command line if cmdline_value is not None: if cmdline_value.strip() == "-1": # infinity value = None else: value = cmdline_value if value is not None: try: self.rlimits[key] = parse_fn(value) except ValueError as e: sys.exit('Invalid value for {} limit: {}'.format(name.lower(), e)) if self.rlimits[key] <= 0: sys.exit('{} limit "{}" is invalid, it needs to be a positive number ' '(or -1 on the command line for disabling it).'.format(name, value)) self.rlimits = {} keys = list(rootTag.keys()) handle_limit_value("Time", TIMELIMIT, config.timelimit, util.parse_timespan_value) handle_limit_value("Hard time", HARDTIMELIMIT, config.timelimit, util.parse_timespan_value) handle_limit_value("Wall time", WALLTIMELIMIT, config.walltimelimit, util.parse_timespan_value) handle_limit_value("Memory", MEMLIMIT, config.memorylimit, parse_memory_limit) handle_limit_value("Core", CORELIMIT, config.corelimit, int) if HARDTIMELIMIT in self.rlimits: hardtimelimit = self.rlimits.pop(HARDTIMELIMIT) if TIMELIMIT in self.rlimits: if hardtimelimit < self.rlimits[TIMELIMIT]: logging.warning( 'Hard timelimit %d is smaller than timelimit %d, ignoring the former.', hardtimelimit, self.rlimits[TIMELIMIT]) elif hardtimelimit > self.rlimits[TIMELIMIT]: self.rlimits[SOFTTIMELIMIT] = self.rlimits[TIMELIMIT] self.rlimits[TIMELIMIT] = hardtimelimit else: self.rlimits[TIMELIMIT] = hardtimelimit # get number of threads, default value is 1 self.num_of_threads = int(rootTag.get("threads")) if ("threads" in keys) else 1 if config.num_of_threads != None: self.num_of_threads = config.num_of_threads if self.num_of_threads < 1: logging.error("At least ONE thread must be given!") sys.exit() # get global options and property file self.options = util.get_list_from_xml(rootTag) self.propertyfile = util.text_or_none(util.get_single_child_from_xml(rootTag, PROPERTY_TAG)) # get columns self.columns = Benchmark.load_columns(rootTag.find("columns")) # get global source files, they are used in all run sets globalSourcefilesTags = rootTag.findall("tasks") + rootTag.findall("sourcefiles") # get required files self._required_files = set() for required_files_tag in rootTag.findall('requiredfiles'): required_files = util.expand_filename_pattern(required_files_tag.text, self.base_dir) if not required_files: logging.warning('Pattern %s in requiredfiles tag did not match any file.', required_files_tag.text) self._required_files = self._required_files.union(required_files) # get requirements self.requirements = Requirements(rootTag.findall("require"), self.rlimits, config) result_files_tags = rootTag.findall("resultfiles") if result_files_tags: self.result_files_patterns = [ os.path.normpath(p.text) for p in result_files_tags if p.text] for pattern in self.result_files_patterns: if pattern.startswith(".."): sys.exit("Invalid relative result-files pattern '{}'.".format(pattern)) else: # default is "everything below current directory" self.result_files_patterns = ["."] # get benchmarks self.run_sets = [] for (i, rundefinitionTag) in enumerate(rootTag.findall("rundefinition")): self.run_sets.append(RunSet(rundefinitionTag, self, i+1, globalSourcefilesTags)) if not self.run_sets: for (i, rundefinitionTag) in enumerate(rootTag.findall("test")): self.run_sets.append(RunSet(rundefinitionTag, self, i+1, globalSourcefilesTags)) if self.run_sets: logging.warning("Benchmark file %s uses deprecated <test> tags. " "Please rename them to <rundefinition>.", benchmark_file) else: logging.warning("Benchmark file %s specifies no runs to execute " "(no <rundefinition> tags found).", benchmark_file) if not any(runSet.should_be_executed() for runSet in self.run_sets): logging.warning("No <rundefinition> tag selected, nothing will be executed.") if config.selected_run_definitions: logging.warning("The selection %s does not match any run definitions of %s.", config.selected_run_definitions, [runSet.real_name for runSet in self.run_sets]) elif config.selected_run_definitions: for selected in config.selected_run_definitions: if not any(util.wildcard_match(run_set.real_name, selected) for run_set in self.run_sets): logging.warning( 'The selected run definition "%s" is not present in the input file, ' 'skipping it.', selected)
def should_be_executed(self): return not self.benchmark.config.selected_run_definitions or any( util.wildcard_match(self.real_name, run_definition) for run_definition in self.benchmark.config.selected_run_definitions )
def __init__(self, benchmark_file, config, start_time): """ The constructor of Benchmark reads the source files, options, columns and the tool from the XML in the benchmark_file.. """ logging.debug("I'm loading the benchmark %s.", benchmark_file) self.config = config self.benchmark_file = benchmark_file self.base_dir = os.path.dirname(self.benchmark_file) # get benchmark-name self.name = os.path.basename(benchmark_file)[:-4] # remove ending ".xml" if config.name: self.name += "." + config.name self.start_time = start_time self.instance = time.strftime("%Y-%m-%d_%H%M", self.start_time) self.output_base_name = config.output_path + self.name + "." + self.instance self.log_folder = self.output_base_name + ".logfiles" + os.path.sep self.log_zip = self.output_base_name + ".logfiles.zip" self.result_files_folder = self.output_base_name + ".files" # parse XML try: rootTag = ElementTree.ElementTree().parse(benchmark_file) except ElementTree.ParseError as e: sys.exit("Benchmark file {} is invalid: {}".format(benchmark_file, e)) if "benchmark" != rootTag.tag: sys.exit( "Benchmark file {} is invalid: " "It's root element is not named 'benchmark'.".format(benchmark_file) ) # get tool tool_name = rootTag.get("tool") if not tool_name: sys.exit("A tool needs to be specified in the benchmark definition file.") (self.tool_module, self.tool) = load_tool_info(tool_name) self.tool_name = self.tool.name() # will be set from the outside if necessary (may not be the case in SaaS environments) self.tool_version = None self.executable = None logging.debug("The tool to be benchmarked is %s.", self.tool_name) def parse_memory_limit(value): try: value = int(value) logging.warning( 'Value "%s" for memory limit interpreted as MB for backwards compatibility, ' "specify a unit to make this unambiguous.", value, ) return value * _BYTE_FACTOR * _BYTE_FACTOR except ValueError: return util.parse_memory_value(value) def handle_limit_value(name, key, cmdline_value, parse_fn): value = rootTag.get(key, None) # override limit from XML with values from command line if cmdline_value is not None: if cmdline_value.strip() == "-1": # infinity value = None else: value = cmdline_value if value is not None: try: self.rlimits[key] = parse_fn(value) except ValueError as e: sys.exit("Invalid value for {} limit: {}".format(name.lower(), e)) if self.rlimits[key] <= 0: sys.exit( '{} limit "{}" is invalid, it needs to be a positive number ' "(or -1 on the command line for disabling it).".format(name, value) ) self.rlimits = {} keys = list(rootTag.keys()) handle_limit_value("Time", TIMELIMIT, config.timelimit, util.parse_timespan_value) handle_limit_value("Hard time", HARDTIMELIMIT, config.timelimit, util.parse_timespan_value) handle_limit_value("Memory", MEMLIMIT, config.memorylimit, parse_memory_limit) handle_limit_value("Core", CORELIMIT, config.corelimit, int) if HARDTIMELIMIT in self.rlimits: hardtimelimit = self.rlimits.pop(HARDTIMELIMIT) if TIMELIMIT in self.rlimits: if hardtimelimit < self.rlimits[TIMELIMIT]: logging.warning( "Hard timelimit %d is smaller than timelimit %d, ignoring the former.", hardtimelimit, self.rlimits[TIMELIMIT], ) elif hardtimelimit > self.rlimits[TIMELIMIT]: self.rlimits[SOFTTIMELIMIT] = self.rlimits[TIMELIMIT] self.rlimits[TIMELIMIT] = hardtimelimit else: self.rlimits[TIMELIMIT] = hardtimelimit # get number of threads, default value is 1 self.num_of_threads = int(rootTag.get("threads")) if ("threads" in keys) else 1 if config.num_of_threads != None: self.num_of_threads = config.num_of_threads if self.num_of_threads < 1: logging.error("At least ONE thread must be given!") sys.exit() # get global options and property file self.options = util.get_list_from_xml(rootTag) self.propertyfile = util.text_or_none(util.get_single_child_from_xml(rootTag, PROPERTY_TAG)) # get columns self.columns = Benchmark.load_columns(rootTag.find("columns")) # get global source files, they are used in all run sets globalSourcefilesTags = rootTag.findall("tasks") + rootTag.findall("sourcefiles") # get required files self._required_files = set() for required_files_tag in rootTag.findall("requiredfiles"): required_files = util.expand_filename_pattern(required_files_tag.text, self.base_dir) if not required_files: logging.warning("Pattern %s in requiredfiles tag did not match any file.", required_files_tag.text) self._required_files = self._required_files.union(required_files) # get requirements self.requirements = Requirements(rootTag.findall("require"), self.rlimits, config) result_files_tags = rootTag.findall("resultfiles") if result_files_tags: self.result_files_patterns = [os.path.normpath(p.text) for p in result_files_tags if p.text] for pattern in self.result_files_patterns: if pattern.startswith(".."): sys.exit("Invalid relative result-files pattern '{}'.".format(pattern)) else: # default is "everything below current directory" self.result_files_patterns = ["."] # get benchmarks self.run_sets = [] for (i, rundefinitionTag) in enumerate(rootTag.findall("rundefinition")): self.run_sets.append(RunSet(rundefinitionTag, self, i + 1, globalSourcefilesTags)) if not self.run_sets: for (i, rundefinitionTag) in enumerate(rootTag.findall("test")): self.run_sets.append(RunSet(rundefinitionTag, self, i + 1, globalSourcefilesTags)) if self.run_sets: logging.warning( "Benchmark file %s uses deprecated <test> tags. " "Please rename them to <rundefinition>.", benchmark_file, ) else: logging.warning( "Benchmark file %s specifies no runs to execute " "(no <rundefinition> tags found).", benchmark_file ) if not any(runSet.should_be_executed() for runSet in self.run_sets): logging.warning("No <rundefinition> tag selected, nothing will be executed.") if config.selected_run_definitions: logging.warning( "The selection %s does not match any run definitions of %s.", config.selected_run_definitions, [runSet.real_name for runSet in self.run_sets], ) elif config.selected_run_definitions: for selected in config.selected_run_definitions: if not any(util.wildcard_match(run_set.real_name, selected) for run_set in self.run_sets): logging.warning( 'The selected run definition "%s" is not present in the input file, ' "skipping it.", selected )
def __init__(self, benchmark_file, config, start_time): """ The constructor of Benchmark reads the source files, options, columns and the tool from the XML in the benchmark_file.. """ logging.debug("I'm loading the benchmark %s.", benchmark_file) self.config = config self.benchmark_file = benchmark_file self.base_dir = os.path.dirname(self.benchmark_file) # get benchmark-name self.name = os.path.basename( benchmark_file)[:-4] # remove ending ".xml" if config.name: self.name += "." + config.name self.description = None if config.description_file is not None: try: self.description = util.read_file(config.description_file) except (OSError, UnicodeDecodeError) as e: raise BenchExecException( "File '{}' given for description could not be read: {}". format(config.description_file, e)) self.start_time = start_time self.instance = start_time.strftime(util.TIMESTAMP_FILENAME_FORMAT) self.output_base_name = config.output_path + self.name + "." + self.instance self.log_folder = self.output_base_name + ".logfiles" + os.path.sep self.log_zip = self.output_base_name + ".logfiles.zip" self.result_files_folder = self.output_base_name + ".files" # parse XML try: rootTag = ElementTree.ElementTree().parse(benchmark_file) except ElementTree.ParseError as e: sys.exit("Benchmark file {} is invalid: {}".format( benchmark_file, e)) if "benchmark" != rootTag.tag: sys.exit("Benchmark file {} is invalid: " "It's root element is not named 'benchmark'.".format( benchmark_file)) # get tool tool_name = rootTag.get("tool") if not tool_name: sys.exit( "A tool needs to be specified in the benchmark definition file." ) (self.tool_module, self.tool) = load_tool_info(tool_name, config) self.tool_name = self.tool.name() # will be set from the outside if necessary (may not be the case in SaaS environments) self.tool_version = None self.executable = None self.display_name = rootTag.get("displayName") def parse_memory_limit(value): # In a future BenchExec version, we could treat unit-less limits as bytes try: value = int(value) except ValueError: return util.parse_memory_value(value) else: raise ValueError( "Memory limit must have a unit suffix, e.g., '{} MB'". format(value)) def handle_limit_value(name, key, cmdline_value, parse_fn): value = rootTag.get(key, None) # override limit from XML with values from command line if cmdline_value is not None: if cmdline_value.strip() == "-1": # infinity value = None else: value = cmdline_value if value is not None: try: self.rlimits[key] = parse_fn(value) except ValueError as e: sys.exit("Invalid value for {} limit: {}".format( name.lower(), e)) if self.rlimits[key] <= 0: sys.exit( '{} limit "{}" is invalid, it needs to be a positive number ' "(or -1 on the command line for disabling it).".format( name, value)) self.rlimits = {} handle_limit_value("Time", TIMELIMIT, config.timelimit, util.parse_timespan_value) handle_limit_value("Hard time", HARDTIMELIMIT, config.timelimit, util.parse_timespan_value) handle_limit_value("Wall time", WALLTIMELIMIT, config.walltimelimit, util.parse_timespan_value) handle_limit_value("Memory", MEMLIMIT, config.memorylimit, parse_memory_limit) handle_limit_value("Core", CORELIMIT, config.corelimit, int) if HARDTIMELIMIT in self.rlimits: hardtimelimit = self.rlimits.pop(HARDTIMELIMIT) if TIMELIMIT in self.rlimits: if hardtimelimit < self.rlimits[TIMELIMIT]: logging.warning( "Hard timelimit %d is smaller than timelimit %d, ignoring the former.", hardtimelimit, self.rlimits[TIMELIMIT], ) elif hardtimelimit > self.rlimits[TIMELIMIT]: self.rlimits[SOFTTIMELIMIT] = self.rlimits[TIMELIMIT] self.rlimits[TIMELIMIT] = hardtimelimit else: self.rlimits[TIMELIMIT] = hardtimelimit self.num_of_threads = int(rootTag.get("threads", 1)) if config.num_of_threads is not None: self.num_of_threads = config.num_of_threads if self.num_of_threads < 1: logging.error("At least ONE thread must be given!") sys.exit() # get global options and property file self.options = util.get_list_from_xml(rootTag) self.propertytag = get_propertytag(rootTag) # get columns self.columns = Benchmark.load_columns(rootTag.find("columns")) # get global source files, they are used in all run sets if rootTag.findall("sourcefiles"): sys.exit( "Benchmark file {} has unsupported old format. " "Rename <sourcefiles> tags to <tasks>.".format(benchmark_file)) globalSourcefilesTags = rootTag.findall("tasks") # get required files self._required_files = set() for required_files_tag in rootTag.findall("requiredfiles"): required_files = util.expand_filename_pattern( required_files_tag.text, self.base_dir) if not required_files: logging.warning( "Pattern %s in requiredfiles tag did not match any file.", required_files_tag.text, ) self._required_files = self._required_files.union(required_files) # get requirements self.requirements = Requirements(rootTag.findall("require"), self.rlimits, config) result_files_tags = rootTag.findall("resultfiles") if result_files_tags: self.result_files_patterns = [ os.path.normpath(p.text) for p in result_files_tags if p.text ] for pattern in self.result_files_patterns: if pattern.startswith(".."): sys.exit( "Invalid relative result-files pattern '{}'.".format( pattern)) else: # default is "everything below current directory" self.result_files_patterns = ["."] # get benchmarks self.run_sets = [] for (i, rundefinitionTag) in enumerate(rootTag.findall("rundefinition")): self.run_sets.append( RunSet(rundefinitionTag, self, i + 1, globalSourcefilesTags)) if not self.run_sets: logging.warning( "Benchmark file %s specifies no runs to execute " "(no <rundefinition> tags found).", benchmark_file, ) if not any(runSet.should_be_executed() for runSet in self.run_sets): logging.warning( "No <rundefinition> tag selected, nothing will be executed.") if config.selected_run_definitions: logging.warning( "The selection %s does not match any run definitions of %s.", config.selected_run_definitions, [runSet.real_name for runSet in self.run_sets], ) elif config.selected_run_definitions: for selected in config.selected_run_definitions: if not any( util.wildcard_match(run_set.real_name, selected) for run_set in self.run_sets): logging.warning( 'The selected run definition "%s" is not present in the input file, ' "skipping it.", selected, )