def transform(self, ast, env):
logger.debug('Running composed stages: %s', self.stages)
for stage in self.stages:
if env.debug:
stage_tuple = (stage, utils.ast2tree(ast))
logger.debug(pprint.pformat(stage_tuple))
ast = stage(ast, env)
return ast
def run_pipeline(self):
ast = self.ast
for method_name in self.order:
if __debug__ and logger.getEffectiveLevel() < logging.DEBUG:
stage_tuple = (method_name, utils.ast2tree(ast))
logger.debug(pprint.pformat(stage_tuple))
ast = getattr(self, method_name)(ast)
return self.func_signature, self.symtab, ast
def run_pipeline(self):
ast = self.ast
for method_name in self.order:
if __debug__ and logger.getEffectiveLevel() < logging.DEBUG:
stage_tuple = (method_name, utils.ast2tree(ast))
logger.debug(pprint.pformat(stage_tuple))
self._current_pipeline_stage = method_name
ast = getattr(self, method_name)(ast)
return self.func_signature, self.symtab, ast
def run_pipeline(self):
# Uses a special logger for logging profiling information.
logger = logging.getLogger("numba.pipeline.profiler")
ast = self.ast
talpha = _timer() # for profiling complete pipeline
for method_name in self.order:
ts = _timer() # for profiling individual stage
if __debug__ and logger.getEffectiveLevel() < logging.DEBUG:
stage_tuple = (method_name, utils.ast2tree(ast))
logger.debug(pprint.pformat(stage_tuple))
self._current_pipeline_stage = method_name
ast = getattr(self, method_name)(ast)
te = _timer() # for profiling individual stage
logger.info("%X pipeline stage %30s:\t%.3fms",
id(self), method_name, (te - ts) * 1000)
tomega = _timer() # for profiling complete pipeline
logger.info("%X pipeline entire:\t\t\t\t\t%.3fms",
id(self), (tomega - talpha) * 1000)
return self.func_signature, self.symtab, ast