Cache common objects in decoder build

native_client/ctcdecode/Makefile

@@ -11,11 +11,14 @@ endif
 
 all: bindings
 
-clean:
+clean-keep-common:
 	rm -rf dist temp_build ds_ctcdecoder.egg-info
 	rm -f swigwrapper_wrap.cpp swigwrapper.py
 
-bindings: clean
+clean: clean-keep-common
+	rm -f common.a
+
+bindings: clean-keep-common
 	pip install --quiet $(PYTHON_PACKAGES) wheel==0.31.0 setuptools==39.1.0
 	AS=$(AS) CC=$(CC) CXX=$(CXX) LD=$(LD) CFLAGS="$(CFLAGS) $(CXXFLAGS)" LDFLAGS="$(LDFLAGS_NEEDED)" $(PYTHON_PATH) $(NUMPY_INCLUDE) python ./setup.py build_ext --num_processes $(NUM_PROCESSES) $(PYTHON_PLATFORM_NAME) $(SETUP_FLAGS)
 	find temp_build -type f -name "*.o" -delete

native_client/ctcdecode/build_common.py

@@ -0,0 +1,87 @@
+#!/usr/bin/env python
+from __future__ import absolute_import, division, print_function
+
+import glob
+import os
+import shlex
+import subprocess
+import sys
+
+from multiprocessing.dummy import Pool
+
+ARGS = ['-O3', '-DNDEBUG', '-DKENLM_MAX_ORDER=6', '-std=c++11',
+        '-Wno-unused-local-typedef', '-Wno-sign-compare']
+
+INCLUDES = [
+    '..',
+    '../kenlm',
+    'third_party/openfst-1.6.7/src/include',
+    'third_party/ThreadPool'
+]
+
+COMMON_FILES = (glob.glob('../kenlm/util/*.cc')
+                + glob.glob('../kenlm/lm/*.cc')
+                + glob.glob('../kenlm/util/double-conversion/*.cc'))
+
+COMMON_FILES += glob.glob('third_party/openfst-1.6.7/src/lib/*.cc')
+
+COMMON_FILES = [
+    fn for fn in COMMON_FILES
+    if not (fn.endswith('main.cc') or fn.endswith('test.cc') or fn.endswith(
+        'unittest.cc'))
+]
+
+COMMON_FILES += glob.glob('*.cpp')
+
+def build_common(out_name='common.a', build_dir='temp_build/temp_build', num_parallel=1):
+    compiler = os.environ.get('CXX', 'g++')
+    ar = os.environ.get('AR', 'ar')
+    libtool = os.environ.get('LIBTOOL', 'libtool')
+    cflags = os.environ.get('CFLAGS', '') + os.environ.get('CXXFLAGS', '')
+
+    for file in COMMON_FILES:
+        outfile = os.path.join(build_dir, os.path.splitext(file)[0] + '.o')
+        outdir = os.path.dirname(outfile)
+        if not os.path.exists(outdir):
+            print('mkdir', outdir)
+            os.makedirs(outdir)
+
+    def build_one(file):
+        outfile = os.path.join(build_dir, os.path.splitext(file)[0] + '.o')
+        if os.path.exists(outfile):
+            return
+
+        cmd = '{cc} -fPIC -c {cflags} {args} {includes} {infile} -o {outfile}'.format(
+            cc=compiler,
+            cflags=cflags,
+            args=' '.join(ARGS),
+            includes=' '.join('-I' + i for i in INCLUDES),
+            infile=file,
+            outfile=outfile,
+        )
+        print(cmd)
+        subprocess.check_call(shlex.split(cmd))
+        return outfile
+
+    pool = Pool(num_parallel)
+    obj_files = list(pool.imap_unordered(build_one, COMMON_FILES))
+
+    if sys.platform.startswith('darwin'):
+        cmd = '{libtool} -static -o {outfile} {infiles}'.format(
+            libtool=libtool,
+            outfile=out_name,
+            infiles=' '.join(obj_files),
+        )
+        print(cmd)
+        subprocess.check_call(shlex.split(cmd))
+    else:
+        cmd = '{ar} rcs {outfile} {infiles}'.format(
+            ar=ar,
+            outfile=out_name,
+            infiles=' '.join(obj_files)
+        )
+        print(cmd)
+        subprocess.check_call(shlex.split(cmd))
+
+if __name__ == '__main__':
+    build_common()

native_client/ctcdecode/setup.py

@@ -5,12 +5,13 @@ from distutils.command.build import build
 from setuptools import setup, Extension, distutils
 
 import argparse
-import glob
 import multiprocessing.pool
 import os
 import platform
 import sys
 
+from build_common import *
+
 try:
     import numpy
     try:
@@ -40,66 +41,25 @@ def read(fname):
 
 project_version = read('../../VERSION').strip()
 
-# monkey-patch for parallel compilation
-# See: https://stackoverflow.com/a/13176803
-def parallelCCompile(self,
-                     sources,
-                     output_dir=None,
-                     macros=None,
-                     include_dirs=None,
-                     debug=0,
-                     extra_preargs=None,
-                     extra_postargs=None,
-                     depends=None):
-    # those lines are copied from distutils.ccompiler.CCompiler directly
-    macros, objects, extra_postargs, pp_opts, build = self._setup_compile(
-        output_dir, macros, include_dirs, sources, depends, extra_postargs)
-    cc_args = self._get_cc_args(pp_opts, debug, extra_preargs)
+build_dir = 'temp_build/temp_build'
+common_build = 'common.a'
 
-    # parallel code
-    def _single_compile(obj):
-        try:
-            src, ext = build[obj]
-        except KeyError:
-            return
-        self._compile(obj, src, ext, cc_args, extra_postargs, pp_opts)
+if not os.path.exists(common_build):
+    if not os.path.exists(build_dir):
+        os.makedirs(build_dir)
 
-    # convert to list, imap is evaluated on-demand
-    thread_pool = multiprocessing.pool.ThreadPool(args[0].num_processes)
-    list(thread_pool.imap(_single_compile, objects))
-    return objects
-
-# hack compile to support parallel compiling
-distutils.ccompiler.CCompiler.compile = parallelCCompile
-
-FILES = glob.glob('../kenlm/util/*.cc') \
-        + glob.glob('../kenlm/lm/*.cc') \
-        + glob.glob('../kenlm/util/double-conversion/*.cc')
-
-FILES += glob.glob('third_party/openfst-1.6.7/src/lib/*.cc')
-
-FILES = [
-    fn for fn in FILES
-    if not (fn.endswith('main.cc') or fn.endswith('test.cc') or fn.endswith(
-        'unittest.cc'))
-]
-
-ARGS = ['-O3', '-DNDEBUG', '-DKENLM_MAX_ORDER=6', '-std=c++11',
-        '-Wno-unused-local-typedef', '-Wno-sign-compare']
+    build_common(out_name='common.a',
+                 build_dir=build_dir,
+                 num_parallel=args[0].num_processes)
 
 decoder_module = Extension(
     name='ds_ctcdecoder._swigwrapper',
-    sources=['swigwrapper.i'] + FILES + glob.glob('*.cpp'),
+    sources=['swigwrapper.i'],
     swig_opts=['-c++', '-extranative'],
     language='c++',
-    include_dirs=[
-        numpy_include,
-        '..',
-        '../kenlm',
-        'third_party/openfst-1.6.7/src/include',
-        'third_party/ThreadPool',
-    ],
-    extra_compile_args=ARGS
+    include_dirs=INCLUDES + [numpy_include],
+    extra_compile_args=ARGS,
+    extra_link_args=[common_build],
 )
 
 class BuildExtFirst(build):

taskcluster/host-build.sh

@@ -19,6 +19,11 @@ do_bazel_build
 
 do_deepspeech_binary_build
 
+# Decoder packages for Linux are built a separate task
+if [ "${OS}" = "Darwin" ]; then
+    do_deepspeech_decoder_build
+fi
+
 do_deepspeech_python_build
 
 do_deepspeech_nodejs_build

tc-tests-utils.sh

@@ -562,7 +562,7 @@ do_deepspeech_decoder_build()
 
     cp native_client/ctcdecode/dist/*.whl wheels
 
-    make -C native_client/ctcdecode clean
+    make -C native_client/ctcdecode clean-keep-common
 
     unset NUMPY_BUILD_VERSION
     unset NUMPY_DEP_VERSION

util/taskcluster.py

@@ -73,7 +73,7 @@ if __name__ == '__main__':
     parser.add_argument('--target', required=False,
                         help='Where to put the native client binary files')
     parser.add_argument('--arch', required=False,
-                        help='Which architecture to download binaries for. "arm" for ARM 7 (32-bit), "gpu" for CUDA enabled x86_64 binaries, "cpu" for CPU-only x86_64 binaries, "osx" for CPU-only x86_64 OSX binaries. Optional ("cpu" by default)')
+                        help='Which architecture to download binaries for. "arm" for ARM 7 (32-bit), "arm64" for ARM64, "gpu" for CUDA enabled x86_64 binaries, "cpu" for CPU-only x86_64 binaries, "osx" for CPU-only x86_64 OSX binaries. Optional ("cpu" by default)')
     parser.add_argument('--artifact', required=False,
                         default='native_client.tar.xz',
                         help='Name of the artifact to download. Defaults to "native_client.tar.xz"')