jit_utils.py

# Torch
from torch._C import _jit_python_print
from torch._six import PY2
from torch.autograd import Variable
from torch.autograd.function import _nested_map
from torch.jit.annotations import BroadcastingList2, BroadcastingList3  # noqa: F401
from torch.onnx import OperatorExportTypes
import torch
import torch.cuda
import torch.jit
import torch.jit._logging
import torch.jit.frontend
import torch.jit.quantized

# Testing utils
from common_utils import TestCase, IS_WINDOWS, \
    freeze_rng_state, TemporaryFileName

# Standard library
from contextlib import contextmanager
from functools import reduce
from itertools import chain
import inspect
import io
import math
import os
import pickle
import tempfile
import textwrap


def execWrapper(code, glob, loc):
    if PY2:
        exec(code) in glob, loc
    else:
        exec(code, glob, loc)


class JitTestCase(TestCase):
    _do_cuda_memory_leak_check = True
    _restored_warnings = False

    def setHooks(self):
        torch._C._jit_set_emit_hooks(self.emitModuleHook, self.emitFunctionHook)

    def clearHooks(self):
        torch._C._jit_set_emit_hooks(None, None)

    def setUp(self):
        super(JitTestCase, self).setUp()
        # unittest overrides all warning filters and forces all of them to show up
        # after we install our own to silence those coming from inside PyTorch.
        # This will ensure that our filter still takes precedence.
        if not JitTestCase._restored_warnings:
            torch.jit.TracerWarning.ignore_lib_warnings()
            JitTestCase._restored_warnings = True
        self.setHooks()

    def tearDown(self):
        super(JitTestCase, self).tearDown()
        # needs to be cleared because python might be unloaded before
        # the callback gets destucted
        self.clearHooks()
        torch._C._jit_clear_class_registry()

    def _isHookExceptionOk(self, e):
        se = str(e)
        allowed = ("Could not export Python function",
                   "closures are not exportable")
        for a in allowed:
            if a in se:
                return True
        return False

    def emitFunctionHook(self, func):
        # func has invalid names for export, skip the jitter check
        if func.name == "<lambda>" or "aten::" in func.name or not _inline_everything:
            return
        # disable the hook while we parse code, otherwise we will re-enter the hook
        with torch.jit._disable_emit_hooks():
            try:
                src, constants = _jit_python_print(func)
                cu = torch.jit.CompilationUnit()._import(src, constants)
                func2 = getattr(cu, func.name)
                src2, constants2 = _jit_python_print(func2)
                self.assertMultiLineEqual(src, src2)
            except RuntimeError as e:
                if not self._isHookExceptionOk(e):
                    raise

    def emitModuleHook(self, module):
        import zipfile

        def copy_structure_and_params(m):
            c = torch.jit.ScriptModule()
            for name, v in m._get_parameters():
                c._c._register_parameter(name, v, False)
            for name, the_type, v in m._get_attributes():
                c._c._register_attribute(name, the_type, v)
            for name, s in m._get_modules():
                c._c._register_module(name, copy_structure_and_params(s)._c)
            return c

        # disable the hook while we parse code, otherwise we will re-enter the hook
        with torch.jit._disable_emit_hooks():
            try:
                if len(module.code) == 0:
                    # short-circuit if this is an empty module
                    return
                # save the module to a buffer
                buffer = io.BytesIO()
                torch.jit.save(module, buffer)
                # copy the data in the buffer so we can restore it later. This
                # is because py2 and py3 have different semantics with zipfile
                # and it's easier to just work with a fresh copy each time.
                buffer_copy = buffer.getvalue()

                # crack open the zip format to get at the main module code
                archive = zipfile.ZipFile(buffer)
                # check that we have no duplicate names
                self.assertEqual(len(set(archive.namelist())), len(archive.namelist()))
                main_module = archive.open('archive/code/archive.py')
                main_module_code = "".join([line.decode() for line in main_module])
                main_module_debug_file = archive.open('archive/debug/archive.pkl')
                main_module_debug = pickle.load(main_module_debug_file)
            except RuntimeError as e:
                if not self._isHookExceptionOk(e):
                    raise
                else:
                    return

            # import the model again (from a the copy we made of the original)
            buffer2 = io.BytesIO(buffer_copy)
            imported = torch.jit.load(buffer2)

            # save it again
            saved_module_buffer_2 = io.BytesIO()
            torch.jit.save(imported, saved_module_buffer_2)

            saved_module_buffer_2.seek(0)
            archive2 = zipfile.ZipFile(saved_module_buffer_2)
            main_module_2 = archive2.open('archive/code/archive.py')
            main_module_2_code = "".join([line.decode() for line in main_module_2])
            main_module_2_debug_file = archive.open('archive/debug/archive.pkl')
            main_module_2_debug = pickle.load(main_module_2_debug_file)

            self.assertMultiLineEqual(main_module_code, main_module_2_code)
            self.assertEqual(main_module_debug, main_module_2_debug)

    def getExportImportCopy(self, m, also_test_file=True, map_location=None):
        if isinstance(m, torch._C.Function):
            src, constants = _jit_python_print(m)
            cu = torch.jit.CompilationUnit()._import(src, constants)
            return getattr(cu, m.name)

        buffer = io.BytesIO()
        torch.jit.save(m, buffer)
        buffer.seek(0)
        imported = torch.jit.load(buffer, map_location=map_location)

        if not also_test_file:
            return imported

        with TemporaryFileName() as fname:
            imported.save(fname)
            return torch.jit.load(fname, map_location=map_location)

    def getExportImportCopyWithPacking(self, m, also_test_file=True, map_location=None):
        buffer = io.BytesIO()
        m.apply(lambda s: s._pack() if s._c._has_method('_pack') else None)
        torch.jit.save(m, buffer)
        m.apply(lambda s: s._unpack() if s._c._has_method('_unpack') else None)
        buffer.seek(0)
        imported = torch.jit.load(buffer, map_location=map_location)
        imported.apply(lambda s: s._unpack() if s._c._has_method('_unpack') else None)

        if not also_test_file:
            return imported

        # Ideally we would like to not have to manually delete the file, but NamedTemporaryFile
        # opens the file, and it cannot be opened multiple times in Windows. To support Windows,
        # close the file after creation and try to remove it manually
        f = tempfile.NamedTemporaryFile(delete=False)
        try:
            f.close()
            imported.save(f.name)
            result = torch.jit.load(f.name, map_location=map_location)
        finally:
            os.unlink(f.name)

        result.apply(lambda s: s._unpack() if s._c._has_method('_unpack') else None)
        return result

    def assertGraphContains(self, graph, kind):
        self.assertTrue(any(n.kind() == kind for n in graph.nodes()))

    def assertGraphContainsExactly(self, graph, kind, num_kind_nodes, consider_subgraphs=False):
        def perform_assert(graph, kind, actual, expected, consider_subgraphs):
            if actual == expected:
                return
            subgraph = 'including' if consider_subgraphs else 'excluding'
            raise AssertionError(
                '{}\nError: graph contains {} {} nodes ({} subgraphs) but expected {}'.format(
                    graph, actual, kind, subgraph, expected))

        if consider_subgraphs:
            strgraph = str(graph)
            count = strgraph.count(kind) - strgraph.count('with {}'.format(kind))
            perform_assert(graph, kind, count, num_kind_nodes,
                           consider_subgraphs)
            return

        nodes = [node for node in graph.nodes()
                 if node.kind() == kind]
        perform_assert(graph, kind, len(nodes), num_kind_nodes,
                       consider_subgraphs)

    def assertExpectedONNXGraph(self, trace, *args, **kwargs):
        torch.onnx._optimize_trace(trace, operator_export_type=OperatorExportTypes.ONNX)
        self.assertExpectedGraph(trace, *args, **kwargs)

    def assertExpectedGraph(self, trace, *args, **kwargs):
        if isinstance(trace, torch._C.Graph):
            graph = trace
        else:
            graph = trace.graph()

        torch._C._jit_pass_lint(graph)
        torch._C._jit_pass_dce(graph)
        torch._C._jit_pass_lint(graph)
        graph = torch._C._jit_pass_canonicalize(graph)
        torch._C._jit_pass_lint(graph)
        self.assertExpected(str(graph), *args, **kwargs)

    def assertAutodiffNode(self, graph, should_autodiff_node, nonfusible_nodes, fusible_nodes):
        diff_nodes = graph.findAllNodes('prim::DifferentiableGraph')
        diff_subgraphs = [node.g('Subgraph') for node in diff_nodes]

        # For any non-fusible node, it must show up in one of the DifferentiableGraph.
        found_all_nonfusible_nodes = (len(diff_subgraphs) == 0 and len(nonfusible_nodes) == 0)\
            or all([any(g.findNode(n) is not None for g in diff_subgraphs) for n in nonfusible_nodes])

        # For any fusible node, it must show up in one of the FusionGroup in the DifferentiableGraph.
        fusion_nodes = list(chain.from_iterable([g.findAllNodes('prim::FusionGroup') for g in diff_subgraphs]))
        fusion_subgraphs = [node.g('Subgraph') for node in fusion_nodes]
        found_all_fusible_nodes = (len(fusion_nodes) == 0 and len(fusible_nodes) == 0)\
            or all([any(g.findNode(n) is not None for g in fusion_subgraphs) for n in fusible_nodes])

        self.assertEqual(should_autodiff_node, found_all_nonfusible_nodes and found_all_fusible_nodes)

    def run_pass(self, name, trace):
        if isinstance(trace, torch._C.Graph):
            graph = trace
            set_graph = False
        else:
            set_graph = True
            graph = trace.graph()

        torch._C._jit_pass_lint(graph)
        result = getattr(torch._C, '_jit_pass_' + name)(graph)
        if result is not None:
            graph = result
        torch._C._jit_pass_lint(graph)

        if set_graph:
            trace.set_graph(graph)
        return graph

    def get_frame_vars(self, frames_up):
        frame = inspect.currentframe()
        i = 0
        while i < frames_up + 1:
            frame = frame.f_back
            i += 1
        defined_vars = {}
        defined_vars.update(frame.f_locals)
        defined_vars.update(frame.f_globals)
        return defined_vars

    def checkScriptRaisesRegex(self, script, inputs, exception, regex,
                               optimize=True, outputs=None, capture_output=False):
        """
        Checks that a given function will throw the correct exception,
        when executed with normal python, the string frontend, and the AST frontend
        """
        # normal python
        with self.assertRaisesRegex(exception, regex):
            script(*inputs)
        # string frontend
        with self.assertRaisesRegex(exception, regex):
            source = textwrap.dedent(inspect.getsource(script))
            cu = torch.jit.CompilationUnit(source, optimize)
            ge = getattr(cu, script.__name__)
            ge(*inputs)
        # python AST frontend
        with self.assertRaisesRegex(exception, regex):
            ge = torch.jit.script(script, optimize)
            ge(*inputs)

    def checkScript(self,
                    script,
                    inputs,
                    optimize=True,
                    name='func',
                    capture_output=False,
                    frames_up=1):
        if isinstance(script, str):
            # Compile the string to a Script function
            cu = torch.jit.CompilationUnit(script, optimize, _frames_up=frames_up)

            # Execute the Python function so we can run it later and get its
            # outputs
            frame = self.get_frame_vars(frames_up)
            the_locals = {}
            execWrapper(script, glob=frame, loc=the_locals)
            frame.update(the_locals)

            python_fn = frame[name]
            scripted_fn = getattr(cu, name)
        else:

            # Check the string frontend first
            source = textwrap.dedent(inspect.getsource(script))
            self.checkScript(
                source,
                inputs,
                optimize,
                script.__name__,
                capture_output,
                frames_up=2)

            # Continue checking the Python frontend
            scripted_fn = torch.jit.script(script, optimize, _frames_up=1)
            python_fn = script

        if capture_output:
            with self.capture_stdout() as script_stdout:
                script_outputs = scripted_fn(*inputs)
            with self.capture_stdout() as _python_stdout:
                python_outputs = python_fn(*inputs)
            if not IS_WINDOWS:
                self.assertExpected(script_stdout[0], subname='stdout')
        else:
            script_outputs = scripted_fn(*inputs)
            python_outputs = python_fn(*inputs)
        self.assertEqual(python_outputs, script_outputs)

        return scripted_fn

    def checkTrace(self, func, reference_tensors, input_tensors=None,
                   optimize=True, drop=None, allow_unused=False, verbose=False,
                   inputs_require_grads=True, check_tolerance=1e-5, export_import=True,
                   _force_outplace=False):
        # TODO: check gradients for parameters, not just inputs
        def allSum(vs):
            # drop allows us to remove some values from ever being used
            # to test unused outputs
            if drop is not None:
                vs = vs[:-drop]
            # we don't want all the grad for all the outputs to be the same
            # so we multiply each by a constant
            return sum(math.log(i + 2) * v.sum() for i, v in enumerate(vs) if v is not None)
        if input_tensors is None:
            input_tensors = reference_tensors

        def do_input_map(fn, input):
            return _nested_map(lambda t: isinstance(t, torch.Tensor), fn)(input)

        def flatten_inputs(inputs):
            def input_reduce(input, fn, acc):
                if isinstance(input, torch.Tensor):
                    fn(input, acc)
                elif isinstance(input, dict):
                    reduce(lambda acc, key: input_reduce(input[key], fn, acc), input, acc)
                else:
                    reduce(lambda acc, val: input_reduce(val, fn, acc), input, acc)
                return acc
            return tuple(input_reduce(recording_inputs, lambda t, acc: acc.append(t), []))

        nograd_inputs = reference_tensors
        if inputs_require_grads:
            recording_inputs = do_input_map(lambda t: t.clone().requires_grad_(), reference_tensors)
            flattened_recording_inputs = flatten_inputs(recording_inputs)
        else:
            recording_inputs = reference_tensors

        ge = torch.jit.trace(func, input_tensors, optimize=optimize, check_tolerance=check_tolerance,
                             _force_outplace=_force_outplace)

        if export_import:
            ge = self.getExportImportCopy(ge)

        if verbose:
            print(ge.graph)

        # test no gradients case
        outputs = func(*nograd_inputs)
        outputs_ge = ge(*nograd_inputs)
        self.assertEqual(outputs, outputs_ge)

        # test single grad case
        outputs = func(*recording_inputs)
        if inputs_require_grads:
            grads = torch.autograd.grad(allSum(outputs), flattened_recording_inputs,
                                        allow_unused=allow_unused)

        outputs_ge = ge(*recording_inputs)
        if inputs_require_grads:
            grads_ge = torch.autograd.grad(allSum(outputs_ge), flattened_recording_inputs,
                                           allow_unused=allow_unused)
        self.assertEqual(outputs, outputs_ge)
        if inputs_require_grads:
            self.assertEqual(grads, grads_ge)

        # test the grad grad case

        outputs = func(*recording_inputs)
        l1 = allSum(outputs)
        if inputs_require_grads:
            grads = torch.autograd.grad(l1, flattened_recording_inputs, create_graph=True,
                                        allow_unused=allow_unused)
        if inputs_require_grads:
            l2 = (allSum(grads) * l1)
            grads2 = torch.autograd.grad(l2, flattened_recording_inputs, allow_unused=allow_unused)

        if inputs_require_grads:
            recording_inputs = do_input_map(lambda t: Variable(t, requires_grad=True), reference_tensors)
            flattened_recording_inputs = flatten_inputs(recording_inputs)

        outputs_ge = ge(*recording_inputs)
        l1_ge = allSum(outputs_ge)
        if inputs_require_grads:
            grads_ge = torch.autograd.grad(
                l1_ge, flattened_recording_inputs, create_graph=True, allow_unused=allow_unused)

        if inputs_require_grads:
            l2_ge = (allSum(grads_ge) * l1_ge)
            grads2_ge = torch.autograd.grad(l2_ge, flattened_recording_inputs, allow_unused=allow_unused)

        self.assertEqual(outputs, outputs_ge)
        if inputs_require_grads:
            self.assertEqual(grads, grads_ge)
            for g2, g2_ge in zip(grads2, grads2_ge):
                if g2 is None and g2_ge is None:
                    continue
                self.assertTrue(torch.allclose(g2, g2_ge, atol=8e-4, rtol=8e-4))

        return ge

    def createFunctionFromGraph(self, trace):
        graph = trace if isinstance(trace, torch._C.Graph) else trace.graph()
        return torch._C._create_function_from_graph("forward", graph)

    def assertExportImport(self, trace, inputs):
        m = self.createFunctionFromGraph(trace)
        self.assertExportImportModule(m, inputs)

    def assertExportImportModule(self, m, inputs):
        m_import = self.getExportImportCopy(m)
        self.assertEqual(self.runAndSaveRNG(m, inputs),
                         self.runAndSaveRNG(m_import, inputs))

    def runAndSaveRNG(self, func, inputs, kwargs=None):
        kwargs = kwargs if kwargs else {}
        with freeze_rng_state():
            results = func(*inputs, **kwargs)
        return results

@contextmanager
def enable_profiling_mode():
    torch._C._jit_set_profiling_mode(True)
    yield
    torch._C._jit_set_profiling_mode(False)

_inline_everything = True
@contextmanager
def disable_inline_everything_mode():
    global _inline_everything
    old = _inline_everything
    _inline_everything = False
    torch._C._jit_set_inline_everything_mode(False)
    yield
    _inline_everything = old
    torch._C._jit_set_inline_everything_mode(old)


# note: not re-entrant, use unnested only
@contextmanager
def disable_autodiff_subgraph_inlining(enabled=True):
    torch._C._debug_set_autodiff_subgraph_inlining(not enabled)
    yield
    torch._C._debug_set_autodiff_subgraph_inlining(True)


# make it easy to quicky define/trace a function for these tests
def _trace(*args, **kwargs):
    def wrapper(func):
        return torch.jit.trace(func, args, **kwargs)
    return wrapper


def enable_cpu_fuser(fn):
    def wrapper(*args, **kwargs):
        torch._C._jit_override_can_fuse_on_cpu(True)
        try:
            fn(*args, **kwargs)
        finally:
            torch._C._jit_override_can_fuse_on_cpu(False)
    return wrapper


def enable_cpu_fuser_if(cond):
    if cond:
        return enable_cpu_fuser
    else:
        def noop_fuser(fn):
            def wrapper(*args, **kwargs):
                return fn(*args, **kwargs)
            return wrapper
        return noop_fuser