def __init__(self, name, *args, **kws):
self.name = name
actual_default_pipeline = pipeline.ComposedPipelineStage(
default_pipeline_order)
self.pipelines = {
self.default_pipeline : actual_default_pipeline,
'type_infer' : pipeline.ComposedPipelineStage(
default_type_infer_pipeline_order),
'dummy_type_infer' : pipeline.ComposedPipelineStage(
default_dummy_type_infer_pipeline_order),
'compile' : pipeline.ComposedPipelineStage(
default_compile_pipeline_order),
'wrap_func' : pipeline.ComposedPipelineStage(
default_numba_wrapper_pipeline_order),
}
self.context = NumbaContext()
self.specializations = functions.FunctionCache(env=self)
self.exports = PyccEnvironment()
self.translation = self.TranslationEnvironment(self)
self.debug = logger.getEffectiveLevel() < logging.DEBUG
# FIXME: NumbaContext has up to now been used as a stand in
# for NumbaEnvironment, so the following member definitions
# should be moved into the environment, and the code that uses
# them should be updated.
context = self.context
context.env = self
context.numba_pipeline = actual_default_pipeline
context.function_cache = self.specializations
context.intrinsic_library = default_intrinsic_library(context)
context.external_library = default_external_library(context)
context.utility_library = default_utility_library(context)
self.llvm_context = translate.LLVMContextManager()
def __init__(self, name, *args, **kws):
self.name = name
actual_default_pipeline = pipeline.ComposedPipelineStage(
default_pipeline_order)
self.pipelines = {
self.default_pipeline:
actual_default_pipeline,
'normalize':
pipeline.ComposedPipelineStage(default_normalize_order),
'cf':
pipeline.ComposedPipelineStage(default_cf_pipeline_order),
'type_infer':
pipeline.ComposedPipelineStage(default_type_infer_pipeline_order),
'dummy_type_infer':
pipeline.ComposedPipelineStage(
default_dummy_type_infer_pipeline_order),
'compile':
pipeline.ComposedPipelineStage(default_compile_pipeline_order),
'wrap_func':
pipeline.ComposedPipelineStage(
default_numba_wrapper_pipeline_order),
'lower':
pipeline.ComposedPipelineStage(default_numba_lower_pipeline_order),
'late_translate':
pipeline.ComposedPipelineStage(
default_numba_late_translate_pipeline_order),
'codegen':
pipeline.ComposedPipelineStage(default_codegen_pipeline),
'post_codegen':
pipeline.ComposedPipelineStage(default_post_codegen_pipeline),
}
self.context = NumbaContext()
self.specializations = functions.FunctionCache(env=self)
self.exports = PyccEnvironment()
self.translation = self.TranslationEnvironment(self)
self.debug = logger.getEffectiveLevel() < logging.DEBUG
# FIXME: NumbaContext has up to now been used as a stand in
# for NumbaEnvironment, so the following member definitions
# should be moved into the environment, and the code that uses
# them should be updated.
context = self.context
context.env = self
context.numba_pipeline = actual_default_pipeline
context.function_cache = self.specializations
context.intrinsic_library = default_intrinsic_library(context)
context.external_library = default_external_library(context)
context.utility_library = default_utility_library(context)
self.llvm_context = translate.LLVMContextManager()
self.annotation_blocks = []
from numba import pipeline, extension_type_inference
from .minivect import minitypes
from numba.utils import debugout
from numba.intrinsic import default_intrinsic_library
from numba.external import default_external_library
from numba.external.utility import default_utility_library
from numba import double, int_
import llvm.core as _lc
context = utils.get_minivect_context()
context.llvm_context = translate.LLVMContextManager()
context.numba_pipeline = pipeline.Pipeline
function_cache = context.function_cache = functions.FunctionCache(context)
context.intrinsic_library = default_intrinsic_library(context)
context.external_library = default_external_library(context)
context.utility_library = default_utility_library(context)
pipeline_env = pipeline.PipelineEnvironment.init_env(
context, "Top-level compilation environment (in %s).\n" % __name__)
_EXPORTS_DOC = '''pycc environment
Includes flags, and modules for exported functions.
wrap - Boolean flag used to indicate that Python wrappers should be
generated for exported functions.
function_signature_map - Map from function names to type signatures
for the translated function (used for header generation).
function_module_map - Map from function names to LLVM modules that
define the translated function.