emcc

#!/usr/bin/env python2
# -*- Mode: python -*-

'''
emcc - compiler helper script
=============================

emcc is a drop-in replacement for a compiler like gcc or clang.

Tell your build system to use this instead of the compiler, and similarly
use emar, emranlib etc. instead of the same command without 'em'.

Example uses:

 * For configure, instead of ./configure, cmake, etc., run emconfigure.py
   with that command as an argument, for example

    emconfigure.py ./configure [options]

   emconfigure.py is a tiny script that just sets some environment vars
   as a convenience. The command just shown is equivalent to

    EMMAKEN_JUST_CONFIGURE=1 RANLIB=PATH/emranlib AR=PATH/emar CXX=PATH/em++ CC=PATH/emcc ./configure [options]

   where PATH is the path to this file.

   EMMAKEN_JUST_CONFIGURE tells emcc that it is being run in ./configure,
   so it should relay everything to gcc/g++. You should not define that when
   running make, of course.

After setting that up, run your build system normally.

Note the appearance of em++ instead of emcc
for the C++ compiler. This is needed for cases where we get
a C++ file with a C extension, in which case CMake can be told
to run g++ on it despite the .c extension, see

  https://github.com/kripken/emscripten/issues/6

(If a similar situation occurs with ./configure, you can do the same there too.)

emcc can be influenced by a few environment variables:

  EMMAKEN_NO_SDK - Will tell emcc *not* to use the emscripten headers. Instead
                   your system headers will be used.

  EMMAKEN_COMPILER - The compiler to be used, if you don't want the default clang.
'''

import os, sys, shutil, tempfile, subprocess, shlex, time, re, logging
from subprocess import PIPE
from tools import shared, jsrun, system_libs
from tools.shared import Compression, execute, suffix, unsuffixed, unsuffixed_basename, WINDOWS, safe_move
from tools.response_file import read_response_file

# endings = dot + a suffix, safe to test by  filename.endswith(endings)
C_ENDINGS = ('.c', '.C', '.i')
CXX_ENDINGS = ('.cpp', '.cxx', '.cc', '.c++', '.CPP', '.CXX', '.CC', '.C++', '.ii')
OBJC_ENDINGS = ('.m', '.mi')
OBJCXX_ENDINGS = ('.mm', '.mii')
SOURCE_ENDINGS = C_ENDINGS + CXX_ENDINGS + OBJC_ENDINGS + OBJCXX_ENDINGS
BITCODE_ENDINGS = ('.bc', '.o', '.obj', '.lo')
DYNAMICLIB_ENDINGS = ('.dylib', '.so') # Windows .dll suffix is not included in this list, since those are never linked to directly on the command line.
STATICLIB_ENDINGS = ('.a',)
ASSEMBLY_ENDINGS = ('.ll',)
HEADER_ENDINGS = ('.h', '.hxx', '.hpp', '.hh', '.H', '.HXX', '.HPP', '.HH')
SUPPORTED_LINKER_FLAGS = ('--start-group', '-(', '--end-group', '-)')

LIB_PREFIXES = ('', 'lib')

JS_CONTAINING_SUFFIXES = ('js', 'html')

DEFERRED_REPONSE_FILES = ('EMTERPRETIFY_BLACKLIST', 'EMTERPRETIFY_WHITELIST')

# Mapping of emcc opt levels to llvm opt levels. We use llvm opt level 3 in emcc opt
# levels 2 and 3 (emcc 3 is unsafe opts, so unsuitable for the only level to get
# llvm opt level 3, and speed-wise emcc level 2 is already the slowest/most optimizing
# level)
LLVM_OPT_LEVEL = {
  0: 0,
  1: 1,
  2: 3,
  3: 3,
}

DEBUG = os.environ.get('EMCC_DEBUG')
if DEBUG == "0":
  DEBUG = None

TEMP_DIR = os.environ.get('EMCC_TEMP_DIR')
LEAVE_INPUTS_RAW = os.environ.get('EMCC_LEAVE_INPUTS_RAW') # Do not compile .ll files into .bc, just compile them with emscripten directly
                                                           # Not recommended, this is mainly for the test runner, or if you have some other
                                                           # specific need.
                                                           # One major limitation with this mode is that libc and libc++ cannot be
                                                           # added in. Also, LLVM optimizations will not be done, nor dead code elimination
# If emcc is running with LEAVE_INPUTS_RAW and then launches an emcc to build something like the struct info, then we don't want
# LEAVE_INPUTS_RAW to be active in that emcc subprocess.
if LEAVE_INPUTS_RAW:
  del os.environ['EMCC_LEAVE_INPUTS_RAW']
AUTODEBUG = os.environ.get('EMCC_AUTODEBUG') # If set to 1, we will run the autodebugger (the automatic debugging tool, see tools/autodebugger).
                                             # Note that this will disable inclusion of libraries. This is useful because including
                                             # dlmalloc makes it hard to compare native and js builds
EMCC_CFLAGS = os.environ.get('EMCC_CFLAGS') # Additional compiler flags that we treat as if they were passed to us on the commandline

if DEBUG: logging.warning('invocation: ' + ' '.join(sys.argv) + (' + ' + EMCC_CFLAGS if EMCC_CFLAGS else '') + '  (in ' + os.getcwd() + ')')
if EMCC_CFLAGS: sys.argv.append(EMCC_CFLAGS)

if DEBUG and LEAVE_INPUTS_RAW: logging.warning('leaving inputs raw')

stdout = PIPE if not DEBUG else None # suppress output of child processes
stderr = PIPE if not DEBUG else None # unless we are in DEBUG mode

EMCC_CXX = '--emscripten-cxx' in sys.argv
sys.argv = filter(lambda x: x != '--emscripten-cxx', sys.argv)

if len(sys.argv) <= 1 or ('--help' not in sys.argv and len(sys.argv) >= 2 and sys.argv[1] != '--version'):
  shared.check_sanity(force=DEBUG)

# Handle some global flags

if len(sys.argv) == 1:
  logging.warning('no input files')
  exit(1)

# read response files very early on
response_file = True
while response_file:
  response_file = None
  for index in range(1, len(sys.argv)):
    if sys.argv[index][0] == '@':
      # found one, loop again next time
      response_file = True
      extra_args = read_response_file(sys.argv[index])
      # slice in extra_args in place of the response file arg
      sys.argv[index:index+1] = extra_args
      break

if len(sys.argv) == 1 or '--help' in sys.argv:
  # Documentation for emcc and its options must be updated in: 
  #    site/source/docs/tools_reference/emcc.rst
  # A prebuilt local version of the documentation is available at: 
  #    site/build/text/docs/tools_reference/emcc.txt
  #    (it is read from there and printed out when --help is invoked)
  # You can also build docs locally as HTML or other formats in site/
  # An online HTML version (which may be of a different version of Emscripten)
  #    is up at http://kripken.github.io/emscripten-site/docs/tools_reference/emcc.html

  print '''%s

------------------------------------------------------------------

emcc: supported targets: llvm bitcode, javascript, NOT elf
(autoconf likes to see elf above to enable shared object support)
''' % (open(shared.path_from_root('site', 'build', 'text', 'docs', 'tools_reference', 'emcc.txt')).read())
  exit(0)

elif sys.argv[1] == '--version':
  revision = '(unknown revision)'
  here = os.getcwd()
  os.chdir(shared.path_from_root())
  try:
    revision = execute(['git', 'show'], stdout=PIPE, stderr=PIPE)[0].split('\n')[0]
  except:
    pass
  finally:
    os.chdir(here)
  print '''emcc (Emscripten gcc/clang-like replacement) %s (%s)
Copyright (C) 2014 the Emscripten authors (see AUTHORS.txt)
This is free and open source software under the MIT license.
There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
  ''' % (shared.EMSCRIPTEN_VERSION, revision)
  exit(0)

elif len(sys.argv) == 2 and sys.argv[1] == '-v': # -v with no inputs
  # autoconf likes to see 'GNU' in the output to enable shared object support
  print 'emcc (Emscripten gcc/clang-like replacement + linker emulating GNU ld) %s' % shared.EMSCRIPTEN_VERSION
  code = subprocess.call([shared.CLANG, '-v'])
  shared.check_sanity(force=True)
  exit(code)

elif '-dumpmachine' in sys.argv:
  print shared.get_llvm_target()
  exit(0)

def is_minus_s_for_emcc(newargs, i):
  assert newargs[i] == '-s'
  if i+1 < len(newargs) and '=' in newargs[i+1]: # -s OPT=VALUE is for us, -s by itself is a linker option
    return True
  else:
    logging.warning('treating -s as linker option and not as -s OPT=VALUE for js compilation')
    return False

# If this is a configure-type thing, do not compile to JavaScript, instead use clang
# to compile to a native binary (using our headers, so things make sense later)
CONFIGURE_CONFIG = (os.environ.get('EMMAKEN_JUST_CONFIGURE') or 'conftest.c' in sys.argv) and not os.environ.get('EMMAKEN_JUST_CONFIGURE_RECURSE')
CMAKE_CONFIG = 'CMakeFiles/cmTryCompileExec.dir' in ' '.join(sys.argv)# or 'CMakeCCompilerId' in ' '.join(sys.argv)
if CONFIGURE_CONFIG or CMAKE_CONFIG:
  debug_configure = 0 # XXX use this to debug configure stuff. ./configure's generally hide our normal output including stderr so we write to a file

  # Whether we fake configure tests using clang - the local, native compiler - or not. if not we generate JS and use node with a shebang
  # Beither approach is perfect, you can try both, but may need to edit configure scripts in some cases
  # By default we configure in js, which can break on local filesystem access, etc., but is otherwise accurate so we
  # disable this if we think we have to. A value of '2' here will force JS checks in all cases. In summary:
  # 0 - use native compilation for configure checks
  # 1 - use js when we think it will work
  # 2 - always use js for configure checks
  use_js = int(os.environ.get('EMCONFIGURE_JS') or 1)

  if debug_configure:
    tempout = '/tmp/emscripten_temp/out'
    if not os.path.exists(tempout):
      open(tempout, 'w').write('//\n')

  src = None
  for arg in sys.argv:
    if arg.endswith(SOURCE_ENDINGS):
      try:
        src = open(arg).read()
        if debug_configure: open(tempout, 'a').write('============= ' + arg + '\n' + src + '\n=============\n\n')
      except:
        pass
    elif arg.endswith('.s'):
      if debug_configure: open(tempout, 'a').write('(compiling .s assembly, must use clang\n')
      if use_js == 1: use_js = 0
    elif arg == '-E' or arg == '-M' or arg == '-MM':
      if use_js == 1: use_js = 0

  if src:
    if 'fopen' in src and '"w"' in src:
      if use_js == 1: use_js = 0 # we cannot write to files from js!
      if debug_configure: open(tempout, 'a').write('Forcing clang since uses fopen to write\n')

  compiler = os.environ.get('CONFIGURE_CC') or (shared.CLANG if not use_js else shared.EMCC) # if CONFIGURE_CC is defined, use that. let's you use local gcc etc. if you need that
  if not ('CXXCompiler' in ' '.join(sys.argv) or EMCC_CXX):
    compiler = shared.to_cc(compiler)

  def filter_emscripten_options(argv):
    idx = 0
    skip_next = False
    for el in argv:
      if skip_next:
        skip_next = False
        idx += 1
        continue
      if not use_js and el == '-s' and is_minus_s_for_emcc(argv, idx): # skip -s X=Y if not using js for configure
        skip_next = True
      else:
        yield el
      idx += 1

  if compiler == shared.EMCC: compiler = [shared.PYTHON, shared.EMCC]
  else: compiler = [compiler] 
  cmd = compiler + list(filter_emscripten_options(sys.argv[1:]))
  if not use_js: cmd += shared.EMSDK_OPTS + ['-D__EMSCRIPTEN__', '-DEMSCRIPTEN']
  if use_js: cmd += ['-s', 'ERROR_ON_UNDEFINED_SYMBOLS=1'] # configure tests should fail when an undefined symbol exists

  logging.debug('just configuring: ' + ' '.join(cmd))
  if debug_configure: open(tempout, 'a').write('emcc, just configuring: ' + ' '.join(cmd) + '\n\n')

  if not use_js:
    exit(subprocess.call(cmd))
  else:
    only_object = '-c' in cmd
    target = None
    for i in range(len(cmd)-1):
      if cmd[i] == '-o':
        if not only_object:
          cmd[i+1] += '.js'
        target = cmd[i+1]
        break
    if not target:
      target = 'a.out.js'
    os.environ['EMMAKEN_JUST_CONFIGURE_RECURSE'] = '1'
    ret = subprocess.call(cmd)
    os.environ['EMMAKEN_JUST_CONFIGURE_RECURSE'] = ''
    if not os.path.exists(target): exit(ret) # note that emcc -c will cause target to have the wrong value here;
                                             # but then, we don't care about bitcode outputs anyhow, below, so
                                             # skipping to exit(ret) is fine
    if target.endswith('.js'):
      shutil.copyfile(target, target[:-3])
      target = target[:-3]
    if not target.endswith(BITCODE_ENDINGS):
      src = open(target).read()
      full_node = ' '.join(shared.NODE_JS)
      if os.path.sep not in full_node:
        full_node = '/usr/bin/' + full_node # TODO: use whereis etc. And how about non-*NIX?
      open(target, 'w').write('#!' + full_node + '\n' + src) # add shebang
      import stat
      try:
        os.chmod(target, stat.S_IMODE(os.stat(target).st_mode) | stat.S_IXUSR) # make executable
      except:
        pass # can fail if e.g. writing the executable to /dev/null
    exit(ret)

if os.environ.get('EMMAKEN_COMPILER'):
  CXX = os.environ['EMMAKEN_COMPILER']
else:
  CXX = shared.CLANG

CC = shared.to_cc(CXX)

# If we got here from a redirection through emmakenxx.py, then force a C++ compiler here
if EMCC_CXX:
  CC = CXX

CC_ADDITIONAL_ARGS = shared.COMPILER_OPTS

EMMAKEN_CFLAGS = os.environ.get('EMMAKEN_CFLAGS')
if EMMAKEN_CFLAGS: sys.argv += shlex.split(EMMAKEN_CFLAGS)

# ---------------- Utilities ---------------

seen_names = {}
def uniquename(name):
  if name not in seen_names:
    seen_names[name] = str(len(seen_names))
  return unsuffixed(name) + '_' + seen_names[name] + (('.' + suffix(name)) if suffix(name) else '')

# ---------------- End configs -------------

if len(sys.argv) == 1 or sys.argv[1] in ['x', 't']:
  # noop ar
  logging.debug('just ar')
  sys.exit(0)

use_cxx = True

for i in range(1, len(sys.argv)):
  arg = sys.argv[i]
  if not arg.startswith('-'):
    if arg.endswith(C_ENDINGS + OBJC_ENDINGS):
      use_cxx = False

# Check if a target is specified
target = None
for i in range(len(sys.argv)-1):
  if sys.argv[i].startswith('-o='):
    raise Exception('Invalid syntax: do not use -o=X, use -o X')

  if sys.argv[i] == '-o':
    target = sys.argv[i+1]
    sys.argv = sys.argv[:i] + sys.argv[i+2:]
    break

specified_target = target
target = specified_target if specified_target is not None else 'a.out.js' # specified_target is the user-specified one, target is what we will generate
target_basename = unsuffixed_basename(target)

if '.' in target:
  final_suffix = target.split('.')[-1]
else:
  final_suffix = ''

if TEMP_DIR:
  temp_dir = TEMP_DIR
  if os.path.exists(temp_dir):
    shutil.rmtree(temp_dir) # clear it
  os.makedirs(temp_dir)
else:
  temp_root = shared.TEMP_DIR
  if not os.path.exists(temp_root):
    os.makedirs(temp_root)
  temp_dir = tempfile.mkdtemp(dir=temp_root)

def in_temp(name):
  return os.path.join(temp_dir, os.path.basename(name))

def in_directory(root, child):
  # make both path absolute
  root = os.path.realpath(root)
  child = os.path.realpath(child)

  # return true, if the common prefix of both is equal to directory
  # e.g. /a/b/c/d.rst and directory is /a/b, the common prefix is /a/b
  return os.path.commonprefix([root, child]) == root

# Parses the essential suffix of a filename, discarding Unix-style version numbers in the name. For example for 'libz.so.1.2.8' returns '.so'
def filename_type_suffix(filename):
  for i in reversed(filename.split('.')[1:]):
    if not i.isdigit():
      return i
  return ''

def filename_type_ending(filename):
  suffix = filename_type_suffix(filename)
  return '' if not suffix else ('.' + suffix)

# Log out times for emcc stages
log_time_last = time.time()
def log_time(name):
  global log_time_last
  now = time.time()
  logging.debug('emcc step "%s" took %.2f seconds', name, now - log_time_last)
  log_time_last = now

misc_temp_files = shared.configuration.get_temp_files()

try:
  call = CXX if use_cxx else CC

  ## Parse args

  newargs = sys.argv[1:]

  opt_level = 0
  debug_level = 0
  shrink_level = 0
  requested_debug = ''
  profiling = False
  profiling_funcs = False
  tracing = False
  emit_symbol_map = False
  js_opts = None
  llvm_opts = None
  llvm_lto = None
  closure = None
  js_transform = None
  pre_js = ''
  post_js = ''
  preload_files = []
  embed_files = []
  exclude_files = []
  compression = None
  ignore_dynamic_linking = False
  shell_path = shared.path_from_root('src', 'shell.html')
  js_libraries = []
  bind = False
  emrun = False
  save_bc = False
  memory_init_file = None
  use_preload_cache = False
  no_heap_copy = False
  proxy_to_worker = False
  default_object_extension = '.o'
  valid_abspaths = []

  def is_valid_abspath(path_name):
    # Any path that is underneath the emscripten repository root must be ok.
    if shared.path_from_root().replace('\\', '/') in path_name.replace('\\', '/'):
      return True

    for valid_abspath in valid_abspaths:
      if in_directory(valid_abspath, path_name):
        return True
    return False


  if use_cxx:
    default_cxx_std = '-std=c++03' # Enforce a consistent C++ standard when compiling .cpp files, if user does not specify one on the cmdline.
  else:
    default_cxx_std = '' # Compiling C code with .c files, don't enforce a default C++ std.

  def check_bad_eq(arg):
    assert '=' not in arg, 'Invalid parameter (do not use "=" with "--" options)'

  absolute_warning_shown = False

  # Scan for warning suppression message in advance from other cmdline flags, so that it works even if -I or -L directives are present before this.
  for i in range(len(newargs)):
    if newargs[i] == '-Wno-warn-absolute-paths':
      newargs[i] = ''
      absolute_warning_shown = True

  settings_changes = []

  def validate_arg_level(level_string, max_level, err_msg):
    try:
      level = int(level_string)
      assert 0 <= level <= max_level
    except:
      raise Exception(err_msg)
    return level

  should_exit = False

  for i in range(len(newargs)):
    newargs[i] = newargs[i].strip() # On Windows Vista (and possibly others), excessive spaces in the command line leak into the items in this array, so trim e.g. 'foo.cpp ' -> 'foo.cpp'
    if newargs[i].startswith('-O'):
      # Let -O default to -O2, which is what gcc does.
      requested_level = newargs[i][2:] or '2'
      if requested_level == 's':
        llvm_opts = ['-Os']
        requested_level = 2
        shrink_level = 1
        settings_changes.append('INLINING_LIMIT=50')
      elif requested_level == 'z':
        llvm_opts = ['-Oz']
        requested_level = 2
        shrink_level = 2
        settings_changes.append('INLINING_LIMIT=25')
      opt_level = validate_arg_level(requested_level, 3, 'Invalid optimization level: ' + newargs[i])
      # We leave the -O option in place so that the clang front-end runs in that
      # optimization mode, but we disable the actual optimization passes, as we'll
      # run them separately.
      newargs.append('-mllvm')
      newargs.append('-disable-llvm-optzns')
    elif newargs[i].startswith('--js-opts'):
      check_bad_eq(newargs[i])
      js_opts = eval(newargs[i+1])
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--llvm-opts'):
      check_bad_eq(newargs[i])
      llvm_opts = eval(newargs[i+1])
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--llvm-lto'):
      check_bad_eq(newargs[i])
      llvm_lto = eval(newargs[i+1])
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--closure'):
      check_bad_eq(newargs[i])
      closure = int(newargs[i+1])
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--js-transform'):
      check_bad_eq(newargs[i])
      js_transform = newargs[i+1]
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--pre-js'):
      check_bad_eq(newargs[i])
      pre_js += open(newargs[i+1]).read() + '\n'
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--post-js'):
      check_bad_eq(newargs[i])
      post_js += open(newargs[i+1]).read() + '\n'
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--minify'):
      check_bad_eq(newargs[i])
      assert newargs[i+1] == '0', '0 is the only supported option for --minify; 1 has been deprecated'
      debug_level = max(1, debug_level)
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('-g'):
      requested_level = newargs[i][2:] or '3'
      debug_level = validate_arg_level(requested_level, 4, 'Invalid debug level: ' + newargs[i])
      requested_debug = newargs[i]
      newargs[i] = ''
    elif newargs[i] == '-profiling' or newargs[i] == '--profiling':
      debug_level = 2
      profiling = True
      newargs[i] = ''
    elif newargs[i] == '-profiling-funcs' or newargs[i] == '--profiling-funcs':
      profiling_funcs = True
      newargs[i] = ''
    elif newargs[i] == '--tracing':
      tracing = True
      newargs[i] = ''
      newargs.append('-D__EMSCRIPTEN_TRACING__=1')
      settings_changes.append("EMSCRIPTEN_TRACING=1")
      js_libraries.append(shared.path_from_root('src', 'library_trace.js'))
    elif newargs[i] == '--emit-symbol-map':
      emit_symbol_map = True
      newargs[i] = ''
    elif newargs[i] == '--bind':
      bind = True
      newargs[i] = ''
      js_libraries.append(shared.path_from_root('src', 'embind', 'emval.js'))
      js_libraries.append(shared.path_from_root('src', 'embind', 'embind.js'))
      if default_cxx_std:
        default_cxx_std = '-std=c++11' # Force C++11 for embind code, but only if user has not explicitly overridden a standard.
    elif newargs[i].startswith('-std='):
      default_cxx_std = '' # User specified a standard to use, clear Emscripten from specifying it.
    elif newargs[i].startswith('--embed-file'):
      check_bad_eq(newargs[i])
      embed_files.append(newargs[i+1])
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--preload-file'):
      check_bad_eq(newargs[i])
      preload_files.append(newargs[i+1])
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--exclude-file'):
      check_bad_eq(newargs[i])
      exclude_files.append(newargs[i+1])
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--compression'):
      logging.warning('--compression is deprecated. Instead, it is recommended you use native gzip compression in your webserver')
      check_bad_eq(newargs[i])
      parts = newargs[i+1].split(',')
      assert len(parts) == 3, '--compression requires specifying    native_encoder,js_decoder,js_name  - see emcc --help. got: %s' % newargs[i+1]
      def locate(tool):
        if WINDOWS:
          if os.path.exists(tool+'.exe'):
            return tool+'.exe'
          if os.path.exists(tool+'.bat'):
            return tool+'.bat'
          if os.path.exists(tool+'.cmd'):
            return tool+'.cmd'
        return tool
      Compression.encoder = locate(parts[0])
      Compression.decoder = locate(parts[1])
      Compression.js_name = parts[2]
      assert os.path.exists(Compression.encoder), 'native encoder %s does not exist' % Compression.encoder
      assert os.path.exists(Compression.decoder), 'js decoder %s does not exist' % Compression.decoder
      Compression.on = True
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--use-preload-cache'):
      use_preload_cache = True
      newargs[i] = ''
    elif newargs[i].startswith('--no-heap-copy'):
      no_heap_copy = True
      newargs[i] = ''
    elif newargs[i] == '--ignore-dynamic-linking':
      ignore_dynamic_linking = True
      newargs[i] = ''
    elif newargs[i] == '-v':
      shared.COMPILER_OPTS += ['-v']
      #shared.LINK_CMD += ['-v'] # Enable verbose output from llvm-link, since it can take a long time and this gives some indication of progress.
      shared.check_sanity(force=True)
      newargs[i] = ''
    elif newargs[i].startswith('--shell-file'):
      check_bad_eq(newargs[i])
      shell_path = newargs[i+1]
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith('--js-library'):
      check_bad_eq(newargs[i])
      js_libraries.append(newargs[i+1])
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i] == '--remove-duplicates':
      logging.warning('--remove-duplicates is deprecated as it is no longer needed. If you cannot link without it, file a bug with a testcase')
      newargs[i] = ''
    elif newargs[i] == '--jcache':
      logging.error('jcache is no longer supported')
      newargs[i] = ''
    elif newargs[i] == '--cache':
      check_bad_eq(newargs[i])
      os.environ['EM_CACHE'] = newargs[i+1]
      shared.reconfigure_cache()
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i] == '--clear-cache':
      logging.warning('clearing cache')
      shared.Cache.erase()
      shared.check_sanity(force=True) # this is a good time for a sanity check
      should_exit = True
    elif newargs[i] == '--clear-ports':
      logging.warning('clearing ports')
      system_libs.Ports.erase()
      should_exit = True
    elif newargs[i] == '--show-ports':
      system_libs.show_ports()
      should_exit = True
    elif newargs[i] == '--save-bc':
      check_bad_eq(newargs[i])
      save_bc = newargs[i+1]
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i] == '--memory-init-file':
      check_bad_eq(newargs[i])
      memory_init_file = int(newargs[i+1])
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i] == '--proxy-to-worker':
      proxy_to_worker = True
      newargs[i] = ''
    elif newargs[i] == '--valid-abspath':
      valid_abspaths.append(newargs[i+1])
      newargs[i] = ''
      newargs[i+1] = ''
    elif newargs[i].startswith(('-I', '-L')):
      path_name = newargs[i][2:]
      if not absolute_warning_shown and os.path.isabs(path_name) and not is_valid_abspath(path_name):
        logging.warning('-I or -L of an absolute path "' + newargs[i] + '" encountered. If this is to a local system header/library, it may cause problems (local system files make sense for compiling natively on your system, but not necessarily to JavaScript). Pass \'-Wno-warn-absolute-paths\' to emcc to hide this warning.') # Of course an absolute path to a non-system-specific library or header is fine, and you can ignore this warning. The danger are system headers that are e.g. x86 specific and nonportable. The emscripten bundled headers are modified to be portable, local system ones are generally not
        absolute_warning_shown = True
    elif newargs[i] == '--emrun':
      emrun = True
      newargs[i] = ''
    elif newargs[i] == '--default-obj-ext':
      newargs[i] = ''
      default_object_extension = newargs[i+1]
      if not default_object_extension.startswith('.'):
        default_object_extension = '.' + default_object_extension
      newargs[i+1] = ''

  if should_exit:
    sys.exit(0)

  newargs = [arg for arg in newargs if arg is not '']

  # If user did not specify a default -std for C++ code, specify the emscripten default.
  if default_cxx_std:
    newargs = newargs + [default_cxx_std]

  if emrun:
    pre_js += open(shared.path_from_root('src', 'emrun_prejs.js')).read() + '\n'
    post_js += open(shared.path_from_root('src', 'emrun_postjs.js')).read() + '\n'

  if js_opts is None: js_opts = opt_level >= 2
  if llvm_opts is None: llvm_opts = LLVM_OPT_LEVEL[opt_level]
  if opt_level == 0: debug_level = max(3, debug_level)
  if memory_init_file is None: memory_init_file = opt_level >= 2

  # TODO: support source maps with js_transform
  if js_transform and debug_level >= 4:
    logging.warning('disabling source maps because a js transform is being done')
    debug_level = 3

  if DEBUG: start_time = time.time() # done after parsing arguments, which might affect debug state

  for i in range(len(newargs)):
    if newargs[i] == '-s':
      if is_minus_s_for_emcc(newargs, i):
        settings_changes.append(newargs[i+1])
        newargs[i] = newargs[i+1] = ''
  newargs = [arg for arg in newargs if arg is not '']

  # Find input files

  # These three arrays are used to store arguments of different types for
  # type-specific processing. In order to shuffle the arguments back together
  # after processing, all of these arrays hold tuples (original_index, value).
  # Note that the index part of the tuple can have a fractional part for input
  # arguments that expand into multiple processed arguments, as in -Wl,-f1,-f2.
  input_files = []
  libs = []
  link_flags = []

  # All of the above arg lists entries contain indexes into the full argument
  # list. In order to add extra implicit args (embind.cc, etc) below, we keep a
  # counter for the next index that should be used.
  next_arg_index = len(newargs)

  has_source_inputs = False
  has_header_inputs = False
  lib_dirs = [shared.path_from_root('system', 'local', 'lib'),
              shared.path_from_root('system', 'lib')]
  for i in range(len(newargs)): # find input files XXX this a simple heuristic. we should really analyze based on a full understanding of gcc params,
                                # right now we just assume that what is left contains no more |-x OPT| things
    arg = newargs[i]

    if i > 0:
      prev = newargs[i-1]
      if prev in ['-MT', '-MF', '-MQ', '-D', '-U', '-o', '-x', '-Xpreprocessor', '-include', '-imacros', '-idirafter', '-iprefix', '-iwithprefix', '-iwithprefixbefore', '-isysroot', '-imultilib', '-A', '-isystem', '-iquote', '-install_name', '-compatibility_version', '-current_version', '-I', '-L']: continue # ignore this gcc-style argument

    if os.path.islink(arg) and os.path.realpath(arg).endswith(SOURCE_ENDINGS + BITCODE_ENDINGS + DYNAMICLIB_ENDINGS + ASSEMBLY_ENDINGS + HEADER_ENDINGS):
      arg = os.path.realpath(arg)

    if not arg.startswith('-'):
      if not os.path.exists(arg):
        logging.error('%s: No such file or directory ("%s" was expected to be an input file, based on the commandline arguments provided)', arg, arg)
        exit(1)

      arg_ending = filename_type_ending(arg)
      if arg_ending.endswith(SOURCE_ENDINGS + BITCODE_ENDINGS + DYNAMICLIB_ENDINGS + ASSEMBLY_ENDINGS + HEADER_ENDINGS) or shared.Building.is_ar(arg): # we already removed -o <target>, so all these should be inputs
        newargs[i] = ''
        if arg_ending.endswith(SOURCE_ENDINGS):
          input_files.append((i, arg))
          has_source_inputs = True
        elif arg_ending.endswith(HEADER_ENDINGS):
          input_files.append((i, arg))
          has_header_inputs = True
        elif arg_ending.endswith(ASSEMBLY_ENDINGS) or shared.Building.is_bitcode(arg): # this should be bitcode, make sure it is valid
          input_files.append((i, arg))
        elif arg_ending.endswith(STATICLIB_ENDINGS + DYNAMICLIB_ENDINGS):
          # if it's not, and it's a library, just add it to libs to find later
          l = unsuffixed_basename(arg)
          for prefix in LIB_PREFIXES:
            if not prefix: continue
            if l.startswith(prefix):
              l = l[len(prefix):]
              break
          libs.append((i, l))
          newargs[i] = ''
        else:
          logging.warning(arg + ' is not valid LLVM bitcode')
      elif arg_ending.endswith(STATICLIB_ENDINGS):
        if not shared.Building.is_ar(arg):
          if shared.Building.is_bitcode(arg):
            logging.error(arg + ': File has a suffix of a static library ' + str(STATICLIB_ENDINGS) + ', but instead is an LLVM bitcode file! When linking LLVM bitcode files, use one of the suffixes ' + str(BITCODE_ENDINGS))
          else:
            logging.error(arg + ': Unknown format, not a static library!')
          exit(1)
      else:
        logging.error(arg + ": Input file has an unknown suffix, don't know what to do with it!")
        exit(1)
    elif arg.startswith('-L'):
      lib_dirs.append(arg[2:])
      newargs[i] = ''
    elif arg.startswith('-l'):
      libs.append((i, arg[2:]))
      newargs[i] = ''
    elif arg.startswith('-Wl,'):
      # Multiple comma separated link flags can be specified. Create fake
      # fractional indices for these: -Wl,a,b,c,d at index 4 becomes:
      # (4, a), (4.25, b), (4.5, c), (4.75, d)
      link_flags_to_add = arg.split(',')[1:]
      for flag_index, flag in enumerate(link_flags_to_add):
        # Only keep flags that shared.Building.link knows how to deal with.
        # We currently can't handle flags with options (like
        # -Wl,-rpath,/bin:/lib, where /bin:/lib is an option for the -rpath
        # flag).
        if flag in SUPPORTED_LINKER_FLAGS:
          link_flags.append((i + float(flag_index) / len(link_flags_to_add), flag))
      newargs[i] = ''

  original_input_files = input_files[:]

  newargs = [arg for arg in newargs if arg is not '']

  # -c means do not link in gcc, and for us, the parallel is to not go all the way to JS, but stop at bitcode
  has_dash_c = '-c' in newargs
  if has_dash_c:
    assert has_source_inputs or has_header_inputs, 'Must have source code or header inputs to use -c'
    target = target_basename + '.o'
    final_suffix = 'o'
  if '-E' in newargs:
    final_suffix = 'eout' # not bitcode, not js; but just result from preprocessing stage of the input file
  if '-M' in newargs or '-MM' in newargs:
    final_suffix = 'mout' # not bitcode, not js; but just dependency rule of the input file
  final_ending = ('.' + final_suffix) if len(final_suffix) > 0 else ''

  # Find library files
  for i, lib in libs:
    logging.debug('looking for library "%s"', lib)
    found = False
    for prefix in LIB_PREFIXES:
      for suff in STATICLIB_ENDINGS + DYNAMICLIB_ENDINGS:
        name = prefix + lib + suff
        for lib_dir in lib_dirs:
          path = os.path.join(lib_dir, name)
          if os.path.exists(path):
            logging.debug('found library "%s" at %s', lib, path)
            input_files.append((i, path))
            found = True
            break
        if found: break
      if found: break
    if not found and lib not in ['GL', 'GLU', 'glut', 'm', 'SDL']: # whitelist our default libraries
      logging.warning('emcc: cannot find library "%s"', lib)

  # If not compiling to JS, then we are compiling to an intermediate bitcode objects or library, so
  # ignore dynamic linking, since multiple dynamic linkings can interfere with each other
  if not filename_type_suffix(target) in JS_CONTAINING_SUFFIXES or ignore_dynamic_linking:
    def check(input_file):
      if filename_type_ending(input_file) in DYNAMICLIB_ENDINGS:
        if not ignore_dynamic_linking: logging.warning('ignoring dynamic library %s because not compiling to JS or HTML, remember to link it when compiling to JS or HTML at the end', os.path.basename(input_file))
        return False
      else:
        return True
    input_files = [(i, input_file) for (i, input_file) in input_files if check(input_file)]

  if len(input_files) == 0:
    logging.error('no input files\nnote that input files without a known suffix are ignored, make sure your input files end with one of: ' + str(SOURCE_ENDINGS + BITCODE_ENDINGS + DYNAMICLIB_ENDINGS + STATICLIB_ENDINGS + ASSEMBLY_ENDINGS + HEADER_ENDINGS))
    exit(0)

  newargs = CC_ADDITIONAL_ARGS + newargs

  assert not (Compression.on and final_suffix != 'html'), 'Compression only works when generating HTML'

  # If we are using embind and generating JS, now is the time to link in bind.cpp
  if bind and final_suffix in JS_CONTAINING_SUFFIXES:
    input_files.append((next_arg_index, shared.path_from_root('system', 'lib', 'embind', 'bind.cpp')))
    next_arg_index += 1

  # Apply optimization level settings
  shared.Settings.apply_opt_level(opt_level, noisy=True)

  if os.environ.get('EMCC_FAST_COMPILER') == '0':
    logging.critical('Non-fastcomp compiler is no longer available, please use fastcomp or an older version of emscripten')
    sys.exit(1)

  # Set ASM_JS default here so that we can override it from the command line.
  shared.Settings.ASM_JS = 1 if opt_level > 0 else 2

  pre_fastcomp_opts = []

  # Apply -s settings in newargs here (after optimization levels, so they can override them)
  for change in settings_changes:
    key, value = change.split('=')
    if value[0] == '@':
      if key not in DEFERRED_REPONSE_FILES:
        value = open(value[1:]).read()
      else:
        value = '"' + value + '"'
    else:
      value = value.replace('\\', '\\\\')
    exec('shared.Settings.' + key + ' = ' + value)
    if key == 'EXPORTED_FUNCTIONS':
      shared.Settings.ORIGINAL_EXPORTED_FUNCTIONS = shared.Settings.EXPORTED_FUNCTIONS[:] # used for warnings in emscripten.py

  try:
    assert shared.Settings.ASM_JS > 0, 'ASM_JS must be enabled in fastcomp'
    assert shared.Settings.SAFE_HEAP in [0, 1], 'safe heap must be 0 or 1 in fastcomp'
    assert shared.Settings.UNALIGNED_MEMORY == 0, 'forced unaligned memory not supported in fastcomp'
    assert shared.Settings.FORCE_ALIGNED_MEMORY == 0, 'forced aligned memory is not supported in fastcomp'
    assert shared.Settings.PGO == 0, 'pgo not supported in fastcomp'
    assert shared.Settings.USE_TYPED_ARRAYS == 2, 'altering USE_TYPED_ARRAYS is not supported'
    assert shared.Settings.QUANTUM_SIZE == 4, 'altering the QUANTUM_SIZE is not supported'
  except Exception, e:
    logging.error('Compiler settings are incompatible with fastcomp. You can fall back to the older compiler core, although that is not recommended, see http://kripken.github.io/emscripten-site/docs/building_from_source/LLVM-Backend.html')
    raise e

  fastcomp_opts = []
  if shared.Settings.NO_EXIT_RUNTIME:
    pre_fastcomp_opts += ['-emscripten-no-exit-runtime']
    if not llvm_lto: fastcomp_opts += ['-globalopt', '-globaldce']
  fastcomp_opts += ['-pnacl-abi-simplify-preopt', '-pnacl-abi-simplify-postopt']
  if shared.Settings.DISABLE_EXCEPTION_CATCHING != 1:
    fastcomp_opts += ['-enable-emscripten-cxx-exceptions']
    if shared.Settings.DISABLE_EXCEPTION_CATCHING == 2:
      fastcomp_opts += ['-emscripten-cxx-exceptions-whitelist=' + ','.join(shared.Settings.EXCEPTION_CATCHING_WHITELIST or ['fake'])]
  if shared.Settings.ASYNCIFY:
    fastcomp_opts += ['-emscripten-asyncify']
    fastcomp_opts += ['-emscripten-asyncify-functions=' + ','.join(shared.Settings.ASYNCIFY_FUNCTIONS)]
    fastcomp_opts += ['-emscripten-asyncify-whitelist=' + ','.join(shared.Settings.ASYNCIFY_WHITELIST)]

  assert not shared.Settings.PGO, 'cannot run PGO in ASM_JS mode'

  if shared.Settings.SAFE_HEAP and not js_opts:
    js_opts = True
    logging.debug('enabling js opts for SAFE_HEAP')

  if debug_level > 1 and closure:
    logging.warning('disabling closure because debug info was requested')
    closure = False

  assert not (shared.Settings.NO_DYNAMIC_EXECUTION and closure), 'cannot have both NO_DYNAMIC_EXECUTION and closure compiler enabled at the same time'

  if closure:
    shared.Settings.CLOSURE_COMPILER = closure
    assert os.path.exists(shared.CLOSURE_COMPILER), logging.error('fatal: Closure compiler (%s) does not exist', shared.CLOSURE_COMPILER)
    if closure == 2 and shared.Settings.ASM_JS == 1:
      logging.warning('not all asm.js optimizations are possible with --closure 2, disabling those - your code will be run more slowly')
      shared.Settings.ASM_JS = 2

  if shared.Settings.MAIN_MODULE:
    assert not shared.Settings.SIDE_MODULE
    shared.Settings.INCLUDE_FULL_LIBRARY = 1
  elif shared.Settings.SIDE_MODULE:
    assert not shared.Settings.MAIN_MODULE
    memory_init_file = False # memory init file is not supported with side modules, must be executable synchronously (for dlopen)

  if shared.Settings.MAIN_MODULE or shared.Settings.SIDE_MODULE:
    assert shared.Settings.ASM_JS, 'module linking requires asm.js output (-s ASM_JS=1)'
    shared.Settings.LINKABLE = 1 # TODO: add FORCE_DCE option for the brave people that do want to dce here and in side modules
    shared.Settings.RELOCATABLE = 1
    shared.Settings.PRECISE_I64_MATH = 1 # other might use precise math, we need to be able to print it
    assert not closure, 'cannot use closure compiler on shared modules'

  if shared.Settings.EMULATE_FUNCTION_POINTER_CASTS:
    shared.Settings.ALIASING_FUNCTION_POINTERS = 0

  if shared.Settings.STB_IMAGE and final_suffix in JS_CONTAINING_SUFFIXES:
    input_files.append((next_arg_index, shared.path_from_root('third_party', 'stb_image.c')))
    next_arg_index += 1
    shared.Settings.EXPORTED_FUNCTIONS += ['_stbi_load', '_stbi_load_from_memory', '_stbi_image_free']

  forced_stdlibs = []
  if shared.Settings.DEMANGLE_SUPPORT:
    shared.Settings.EXPORTED_FUNCTIONS += ['___cxa_demangle']
    forced_stdlibs += ['libcxxabi']

  if type(shared.Settings.EXPORTED_FUNCTIONS) in (list, tuple):
    # always need malloc and free to be kept alive and exported, for internal use and other modules
    for required_export in ['_malloc', '_free']:
      if required_export not in shared.Settings.EXPORTED_FUNCTIONS:
        shared.Settings.EXPORTED_FUNCTIONS.append(required_export)
  else:
    logging.debug('using response file for EXPORTED_FUNCTIONS, make sure it includes _malloc and _free')

  assert not (bind and shared.Settings.NO_DYNAMIC_EXECUTION), 'NO_DYNAMIC_EXECUTION disallows embind'

  if shared.Settings.RELOCATABLE:
    assert shared.Settings.GLOBAL_BASE < 1
    shared.Settings.EMULATED_FUNCTION_POINTERS = 1
    shared.Settings.ERROR_ON_UNDEFINED_SYMBOLS = shared.Settings.WARN_ON_UNDEFINED_SYMBOLS = 0
    if not shared.Settings.SIDE_MODULE:
      shared.Settings.EXPORT_ALL = 1

  if shared.Settings.EMTERPRETIFY:
    shared.Settings.FINALIZE_ASM_JS = 0
    #shared.Settings.GLOBAL_BASE = 8*256 # keep enough space at the bottom for a full stack frame, for z-interpreter
    shared.Settings.SIMPLIFY_IFS = 0 # this is just harmful for emterpreting
    shared.Settings.EXPORTED_FUNCTIONS += ['emterpret']
    if not js_opts:
      js_opts = True
      logging.debug('enabling js opts for EMTERPRETIFY')
    assert closure is not 2, 'EMTERPRETIFY requires valid asm.js, and is incompatible with closure 2 which disables that'

  if shared.Settings.DEAD_FUNCTIONS and not js_opts:
    js_opts = True
    logging.debug('enabling js opts for DEAD_FUNCTIONS')

  if proxy_to_worker:
    shared.Settings.PROXY_TO_WORKER = 1

  if js_opts:
    shared.Settings.RUNNING_JS_OPTS = 1

  if shared.Settings.USE_PTHREADS:
    if not any(s.startswith('PTHREAD_POOL_SIZE=') for s in settings_changes):
      settings_changes.append('PTHREAD_POOL_SIZE=0')
    js_libraries.append(shared.path_from_root('src', 'library_pthread.js'))
    newargs.append('-D__EMSCRIPTEN_PTHREADS__=1')
  else:
    js_libraries.append(shared.path_from_root('src', 'library_pthread_stub.js'))

  if shared.Settings.USE_PTHREADS:
    if shared.Settings.PROXY_TO_WORKER:
      logging.error('-s PROXY_TO_WORKER=1 is not yet supported with -s USE_PTHREADS=1!')
      exit(1)
    if shared.Settings.LINKABLE:
      logging.error('-s LINKABLE=1 is not supported with -s USE_PTHREADS=1!')
      exit(1)
    if shared.Settings.SIDE_MODULE:
      logging.error('-s SIDE_MODULE=1 is not supported with -s USE_PTHREADS=1!')
      exit(1)
    if shared.Settings.MAIN_MODULE:
      logging.error('-s MAIN_MODULE=1 is not supported with -s USE_PTHREADS=1!')
      exit(1)

  shared.Settings.EMSCRIPTEN_VERSION = shared.EMSCRIPTEN_VERSION
  shared.Settings.OPT_LEVEL = opt_level
  shared.Settings.DEBUG_LEVEL = debug_level

  ## Compile source code to bitcode

  logging.debug('compiling to bitcode')

  temp_files = []

  log_time('parse arguments and setup')

  # Precompiled headers support
  if has_header_inputs:
    headers = [header for _, header in input_files]
    for header in headers:
      assert header.endswith(HEADER_ENDINGS), 'if you have one header input, we assume you want to precompile headers, and cannot have source files or other inputs as well: ' + str(headers) + ' : ' + header
    args = newargs + shared.EMSDK_CXX_OPTS + headers
    if specified_target:
      args += ['-o', specified_target]
    args = system_libs.process_args(args, shared.Settings)
    logging.debug("running (for precompiled headers): " + call + ' ' + ' '.join(args))
    execute([call] + args) # let compiler frontend print directly, so colors are saved (PIPE kills that)
    sys.exit(0)

  def get_bitcode_file(input_file):
    if final_suffix not in JS_CONTAINING_SUFFIXES:
      # no need for a temp file, just emit to the right place
      if len(input_files) == 1:
        # can just emit directly to the target
        if specified_target:
          if specified_target.endswith('/') or specified_target.endswith('\\') or os.path.isdir(specified_target):
            return os.path.join(specified_target, os.path.basename(unsuffixed(input_file))) + default_object_extension
          return specified_target
        return unsuffixed(input_file) + final_ending
      else:
        if has_dash_c: return unsuffixed(input_file) + default_object_extension
    return in_temp(unsuffixed(uniquename(input_file)) + default_object_extension)

  # Request LLVM debug info if explicitly specified, or building bitcode with -g, or if building a source all the way to JS with -g
  if debug_level >= 4 or ((final_suffix not in JS_CONTAINING_SUFFIXES or (has_source_inputs and final_suffix in JS_CONTAINING_SUFFIXES)) and requested_debug == '-g'):
    if debug_level == 4 or not (final_suffix in JS_CONTAINING_SUFFIXES and js_opts): # do not save llvm debug info if js optimizer will wipe it out anyhow (but if source maps are used, keep it)
      newargs.append('-g') # preserve LLVM debug info
      debug_level = 4

  # Bitcode args generation code
  def get_bitcode_args(input_files):
    file_ending = filename_type_ending(input_files[0])
    args = [call] + newargs + input_files
    if file_ending.endswith(CXX_ENDINGS):
      args += shared.EMSDK_CXX_OPTS
    args = system_libs.process_args(args, shared.Settings)
    return args

  # -E preprocessor-only support
  if '-E' in newargs or '-M' in newargs or '-MM' in newargs:
    input_files = map(lambda x: x[1], input_files)
    cmd = get_bitcode_args(input_files)
    if specified_target:
      cmd += ['-o', specified_target]
    # Do not compile, but just output the result from preprocessing stage or output the dependency rule. Warning: clang and gcc behave differently with -MF! (clang seems to not recognize it)
    logging.debug(('just preprocessor ' if '-E' in newargs else 'just dependencies: ') + ' '.join(cmd))
    exit(subprocess.call(cmd))

  # First, generate LLVM bitcode. For each input file, we get base.o with bitcode
  for i, input_file in input_files:
    file_ending = filename_type_ending(input_file)
    if file_ending.endswith(SOURCE_ENDINGS):
      logging.debug('compiling source file: ' + input_file)
      output_file = get_bitcode_file(input_file)
      temp_files.append((i, output_file))
      args = get_bitcode_args([input_file]) + ['-emit-llvm', '-c', '-o', output_file]
      logging.debug("running: " + ' '.join(args))
      execute(args) # let compiler frontend print directly, so colors are saved (PIPE kills that)
      if not os.path.exists(output_file):
        logging.error('compiler frontend failed to generate LLVM bitcode, halting')
        sys.exit(1)
    else: # bitcode
      if file_ending.endswith(BITCODE_ENDINGS):
        logging.debug('using bitcode file: ' + input_file)
        temp_files.append((i, input_file))
      elif file_ending.endswith(DYNAMICLIB_ENDINGS) or shared.Building.is_ar(input_file):
        logging.debug('using library file: ' + input_file)
        temp_files.append((i, input_file))
      elif file_ending.endswith(ASSEMBLY_ENDINGS):
        if not LEAVE_INPUTS_RAW:
          logging.debug('assembling assembly file: ' + input_file)
          temp_file = in_temp(unsuffixed(uniquename(input_file)) + '.o')
          shared.Building.llvm_as(input_file, temp_file)
          temp_files.append((i, temp_file))
      else:
        logging.error(input_file + ': Unknown file suffix when compiling to LLVM bitcode!')
        sys.exit(1)

  log_time('bitcodeize inputs')

  if not LEAVE_INPUTS_RAW:
    assert len(temp_files) == len(input_files)

    # Optimize source files
    if llvm_opts > 0:
      for pos, (_, input_file) in enumerate(input_files):
        file_ending = filename_type_ending(input_file)
        if file_ending.endswith(SOURCE_ENDINGS):
          temp_file = temp_files[pos][1]
          logging.debug('optimizing %s', input_file)
          #if DEBUG: shutil.copyfile(temp_file, os.path.join(shared.configuration.CANONICAL_TEMP_DIR, 'to_opt.bc')) # useful when LLVM opt aborts
          new_temp_file = in_temp(unsuffixed(uniquename(temp_file)) + '.o')
          shared.Building.llvm_opt(temp_file, llvm_opts, new_temp_file)
          temp_files[pos] = (temp_files[pos][0], new_temp_file)

  # Decide what we will link
  linker_inputs = [val for _, val in sorted(temp_files + link_flags)]

  # If we were just asked to generate bitcode, stop there
  if final_suffix not in JS_CONTAINING_SUFFIXES:
    if not specified_target:
      assert len(temp_files) == len(input_files)
      for i in range(len(input_files)):
        safe_move(temp_files[i][1], unsuffixed_basename(input_files[i][1]) + final_ending)
    else:
      if len(input_files) == 1:
        _, input_file = input_files[0]
        _, temp_file = temp_files[0]
        bitcode_target = specified_target if specified_target else unsuffixed_basename(input_file) + final_ending
        if temp_file != input_file:
          safe_move(temp_file, bitcode_target)
        else:
          shutil.copyfile(temp_file, bitcode_target)
        temp_output_base = unsuffixed(temp_file)
        if os.path.exists(temp_output_base + '.d'):
          # There was a .d file generated, from -MD or -MMD and friends, save a copy of it to where the output resides,
          # adjusting the target name away from the temporary file name to the specified target.
          # It will be deleted with the rest of the temporary directory.
          deps = open(temp_output_base + '.d').read()
          deps = deps.replace(temp_output_base + default_object_extension, specified_target)
          with open(os.path.join(os.path.dirname(specified_target), os.path.basename(unsuffixed(input_file) + '.d')), "w") as out_dep:
            out_dep.write(deps)
      else:
        assert len(original_input_files) == 1 or not has_dash_c, 'fatal error: cannot specify -o with -c with multiple files' + str(sys.argv) + ':' + str(original_input_files)
        # We have a specified target (-o <target>), which is not JavaScript or HTML, and
        # we have multiple files: Link them
        logging.debug('link: ' + str(linker_inputs) + specified_target)
        # Sort arg tuples and pass the extracted values to link.
        shared.Building.link(linker_inputs, specified_target)
    logging.debug('stopping at bitcode')
    if final_suffix.lower() in ['so', 'dylib', 'dll']:
      logging.warning('Dynamic libraries (.so, .dylib, .dll) are currently not supported by Emscripten. For build system emulation purposes, Emscripten'
        + ' will now generate a static library file (.bc) with the suffix \'.' + final_suffix + '\'. For best practices,'
        + ' please adapt your build system to directly generate a static LLVM bitcode library by setting the output suffix to \'.bc.\')')
    exit(0)

  log_time('process inputs')

  ## Continue on to create JavaScript

  logging.debug('will generate JavaScript')

  extra_files_to_link = []

  if not LEAVE_INPUTS_RAW and \
     not shared.Settings.BUILD_AS_SHARED_LIB and \
     not shared.Settings.SIDE_MODULE: # shared libraries/side modules link no C libraries, need them in parent
    extra_files_to_link = system_libs.get_ports(shared.Settings)
    extra_files_to_link += system_libs.calculate([f for _, f in sorted(temp_files)] + extra_files_to_link, in_temp, stdout, stderr, forced=forced_stdlibs)

  log_time('calculate system libraries')

  # First, combine the bitcode files if there are several. We must also link if we have a singleton .a
  if len(input_files) + len(extra_files_to_link) > 1 or \
     (not LEAVE_INPUTS_RAW and not (suffix(temp_files[0][1]) in BITCODE_ENDINGS or suffix(temp_files[0][1]) in DYNAMICLIB_ENDINGS) and shared.Building.is_ar(temp_files[0][1])):
    linker_inputs += extra_files_to_link
    logging.debug('linking: ' + str(linker_inputs))
    # force archive contents to all be included, if just archives, or if linking shared modules
    force_archive_contents = len([temp for i, temp in temp_files if not temp.endswith(STATICLIB_ENDINGS)]) == 0 or not shared.Building.can_build_standalone()
    shared.Building.link(linker_inputs, in_temp(target_basename + '.bc'), force_archive_contents=force_archive_contents)
    final = in_temp(target_basename + '.bc')
  else:
    if not LEAVE_INPUTS_RAW:
      _, temp_file = temp_files[0]
      _, input_file = input_files[0]
      final = in_temp(target_basename + '.bc')
      if temp_file != input_file:
        shutil.move(temp_file, final)
      else:
        shutil.copyfile(temp_file, final)
    else:
      _, input_file = input_files[0]
      final = in_temp(input_file)
      shutil.copyfile(input_file, final)

  log_time('link')

  if DEBUG:
    emscripten_temp_dir = shared.get_emscripten_temp_dir()
    logging.debug('saving intermediate processing steps to %s', emscripten_temp_dir)

    intermediate_counter = 0
    def save_intermediate(name=None, suffix='js'):
      global intermediate_counter
      shutil.copyfile(final, os.path.join(emscripten_temp_dir, 'emcc-%d%s.%s' % (intermediate_counter, '' if name is None else '-' + name, suffix)))
      intermediate_counter += 1

    if not LEAVE_INPUTS_RAW: save_intermediate('basebc', 'bc')

  # Optimize, if asked to
  if not LEAVE_INPUTS_RAW:
    link_opts = [] if debug_level >= 4 else ['-strip-debug'] # remove LLVM debug if we are not asked for it

    if llvm_lto >= 2:
      logging.debug('running LLVM opts as pre-LTO')
      shared.Building.llvm_opt(in_temp(target_basename + '.bc'), llvm_opts)
      if DEBUG: save_intermediate('opt', 'bc')

    # If we can LTO, do it before dce, since it opens up dce opportunities
    if shared.Building.can_build_standalone() and llvm_lto and llvm_lto != 2:
      if not shared.Building.can_inline(): link_opts.append('-disable-inlining')
      # add a manual internalize with the proper things we need to be kept alive during lto
      link_opts += shared.Building.get_safe_internalize() + ['-std-link-opts']
      # execute it now, so it is done entirely before we get to the stage of legalization etc.
      shared.Building.llvm_opt(final, pre_fastcomp_opts + link_opts)
      if DEBUG: save_intermediate('lto', 'bc')
      pre_fastcomp_opts = []
      link_opts = []
    else:
      # At minimum remove dead functions etc., this potentially saves a lot in the size of the generated code (and the time to compile it)
      link_opts += shared.Building.get_safe_internalize() + ['-globaldce']

    if AUTODEBUG:
      # let llvm opt directly emit ll, to skip writing and reading all the bitcode
      link_opts += ['-S']
      shared.Building.llvm_opt(final, link_opts, final + '.link.ll')
      final = final + '.link.ll'
      if DEBUG: save_intermediate('linktime', 'll')
    else:
      if not save_bc:
        # Simplify LLVM bitcode for fastcomp
        link_opts = pre_fastcomp_opts + link_opts + fastcomp_opts
      shared.Building.llvm_opt(final, link_opts)
      if DEBUG: save_intermediate('linktime', 'bc')
      if save_bc:
        shutil.copyfile(final, save_bc)
        shared.Building.llvm_opt(final, fastcomp_opts, final + '.adsimp.bc')
        final += '.adsimp.bc'
        if DEBUG: save_intermediate('adsimp', 'bc')

  # Prepare .ll for Emscripten
  if LEAVE_INPUTS_RAW:
    assert len(input_files) == 1
  if DEBUG and save_bc: save_intermediate('ll', 'll')

  if AUTODEBUG:
    logging.debug('autodebug')
    execute([shared.PYTHON, shared.AUTODEBUGGER, final, final + '.ad.ll'])
    final += '.ad.ll'
    if DEBUG: save_intermediate('autodebug', 'll')

  # Simplify bitcode after autodebug
  if AUTODEBUG or LEAVE_INPUTS_RAW:
    shared.Building.llvm_opt(final, fastcomp_opts, final + '.adsimp.bc')
    final += '.adsimp.bc'
    if DEBUG: save_intermediate('adsimp', 'bc')

  log_time('post-link')

  # Emscripten
  logging.debug('LLVM => JS')
  extra_args = [] if not js_libraries else ['--libraries', ','.join(map(os.path.abspath, js_libraries))]
  if memory_init_file:
    shared.Settings.MEM_INIT_METHOD = 1
  else:
    assert shared.Settings.MEM_INIT_METHOD != 1
  final = shared.Building.emscripten(final, append_ext=False, extra_args=extra_args)
  if DEBUG: save_intermediate('original')

  log_time('emscript (llvm=>js)')

  # Embed and preload files
  if len(preload_files) + len(embed_files) > 0:
    logging.debug('setting up files')
    file_args = []
    if len(preload_files) > 0:
      file_args.append('--preload')
      file_args += preload_files
    if len(embed_files) > 0:
      file_args.append('--embed')
      file_args += embed_files
    if len(exclude_files) > 0:
      file_args.append('--exclude')
      file_args += exclude_files
    if Compression.on:
      file_args += ['--compress', Compression.encoder, Compression.decoder, Compression.js_name]
    if use_preload_cache:
      file_args.append('--use-preload-cache')
    if no_heap_copy:
      file_args.append('--no-heap-copy')
    if not closure:
      file_args.append('--no-closure')
    file_code = execute([shared.PYTHON, shared.FILE_PACKAGER, unsuffixed(target) + '.data'] + file_args, stdout=PIPE)[0]
    pre_js = file_code + pre_js

  # Apply pre and postjs files
  if pre_js or post_js:
    logging.debug('applying pre/postjses')
    src = open(final).read()
    final += '.pp.js'
    if WINDOWS: # Avoid duplicating \r\n to \r\r\n when writing out.
      if pre_js: pre_js = pre_js.replace('\r\n', '\n')
      if post_js: post_js = post_js.replace('\r\n', '\n')
    open(final, 'w').write(pre_js + src + post_js)
    if DEBUG: save_intermediate('pre-post')

  # Apply a source code transformation, if requested
  if js_transform:
    shutil.copyfile(final, final + '.tr.js')
    final += '.tr.js'
    posix = True if not shared.WINDOWS else False
    logging.debug('applying transform: %s', js_transform)
    subprocess.check_call(shlex.split(js_transform, posix=posix) + [os.path.abspath(final)])
    if DEBUG: save_intermediate('transformed')

  js_transform_tempfiles = [final]

  if shared.Settings.MEM_INIT_METHOD > 0:
    memfile = target + '.mem'
    shared.try_delete(memfile)
    def repl(m):
      # handle chunking of the memory initializer
      s = m.group(1)
      if len(s) == 0: return '' # don't emit 0-size ones
      membytes = [int(x or '0') for x in s.split(',')]
      while membytes and membytes[-1] == 0:
        membytes.pop()
      if not membytes: return ''
      if not memory_init_file:
        # memory initializer in a string literal
        return "memoryInitializer = '%s';" % shared.JS.generate_string_initializer(list(membytes))
      open(memfile, 'wb').write(''.join(map(chr, membytes)))
      if DEBUG:
        # Copy into temp dir as well, so can be run there too
        shared.safe_copy(memfile, os.path.join(shared.get_emscripten_temp_dir(), os.path.basename(memfile)))
      return 'memoryInitializer = "%s";' % os.path.basename(memfile)
    src = re.sub(shared.JS.memory_initializer_pattern, repl, open(final).read(), count=1)
    open(final + '.mem.js', 'w').write(src)
    final += '.mem.js'
    src = None
    js_transform_tempfiles[-1] = final # simple text substitution preserves comment line number mappings
    if DEBUG:
      if os.path.exists(memfile):
        save_intermediate('meminit')
        logging.debug('wrote memory initialization to %s', memfile)
      else:
        logging.debug('did not see memory initialization')
  elif not shared.Settings.MAIN_MODULE and not shared.Settings.SIDE_MODULE and debug_level < 4:
    # not writing a binary init, but we can at least optimize them by splitting them up
    src = open(final).read()
    src = shared.JS.optimize_initializer(src)
    if src is not None:
      logging.debug('optimizing memory initialization')
      open(final + '.mem.js', 'w').write(src)
      final += '.mem.js'
      src = None

  if shared.Settings.USE_PTHREADS:
    shutil.copyfile(shared.path_from_root('src', 'pthread-main.js'), os.path.join(os.path.dirname(os.path.abspath(target)), 'pthread-main.js'))

  log_time('source transforms')

  # It is useful to run several js optimizer passes together, to save on unneeded unparsing/reparsing
  js_optimizer_queue = []
  js_optimizer_extra_info = {}
  js_optimizer_queue_history = []
  js_optimizer_blacklist = (os.environ.get('EMCC_JSOPT_BLACKLIST') or '').split(',')
  minify_whitespace = False

  def flush_js_optimizer_queue(title='js_opts'):
    global final, js_optimizer_queue, js_optimizer_extra_info, js_optimizer_queue_history

    js_optimizer_queue = filter(lambda p: p not in js_optimizer_blacklist, js_optimizer_queue)

    if js_optimizer_extra_info is not None and len(js_optimizer_extra_info) == 0:
      js_optimizer_extra_info = None

    if len(js_optimizer_queue) > 0 and not(not shared.Settings.ASM_JS and len(js_optimizer_queue) == 1 and js_optimizer_queue[0] == 'last'):

      def run_passes(passes, title, just_split, just_concat):
        global final, minify_whitespace, emit_symbol_map, target
        passes = ['asm'] + passes
        if shared.Settings.PRECISE_F32:
          passes = ['asmPreciseF32'] + passes
        if emit_symbol_map and 'minifyNames' in passes:
          passes += ['symbolMap='+target+'.symbols']
        if profiling_funcs and 'minifyNames' in passes:
          passes += ['profilingFuncs']
        if minify_whitespace and 'last' in passes:
          passes += ['minifyWhitespace']
        logging.debug('applying js optimization passes: %s', ' '.join(passes))
        misc_temp_files.note(final)
        final = shared.Building.js_optimizer(final, passes, debug_level >= 4, js_optimizer_extra_info, just_split=just_split, just_concat=just_concat)
        misc_temp_files.note(final)
        js_transform_tempfiles.append(final)
        if DEBUG: save_intermediate(title, suffix='js' if 'emitJSON' not in passes else 'json')

      passes = js_optimizer_queue[:]

      if DEBUG != '2' or len(passes) < 2:
        # by assumption, our input is JS, and our output is JS. If a pass is going to run in the native optimizer in C++, then we
        # must give it JSON and receive from it JSON
        chunks = []
        curr = []
        for p in passes:
          if len(curr) == 0:
            curr.append(p)
          else:
            native = shared.js_optimizer.use_native(p, source_map=debug_level >= 4)
            last_native = shared.js_optimizer.use_native(curr[-1], source_map=debug_level >= 4)
            if native == last_native:
              curr.append(p)
            else:
              curr.append('emitJSON')
              chunks.append(curr)
              curr = ['receiveJSON', p]
        if len(curr) > 0:
          chunks.append(curr)
        if len(chunks) == 1:
          run_passes(chunks[0], title, just_split=False, just_concat=False)
        else:
          for i in range(len(chunks)):
            run_passes(chunks[i], 'js_opts_' + str(i), just_split='receiveJSON' in chunks[i], just_concat='emitJSON' in chunks[i])
      else:
        # DEBUG 2, run each pass separately
        for p in passes:
          js_optimizer_queue = [p]
          flush_js_optimizer_queue(p)
      js_optimizer_queue_history += js_optimizer_queue
      js_optimizer_queue = []
    js_optimizer_extra_info = {}

  if shared.Settings.DEAD_FUNCTIONS:
    js_optimizer_queue += ['eliminateDeadFuncs']
    js_optimizer_extra_info['dead_functions'] = shared.Settings.DEAD_FUNCTIONS

  if opt_level >= 1 and js_opts:
    logging.debug('running js post-opts')

    if DEBUG == '2':
      # Clean up the syntax a bit
      js_optimizer_queue += ['noop']

    def get_eliminate():
      if shared.Settings.ALLOW_MEMORY_GROWTH:
        return 'eliminateMemSafe'
      else:
        return 'eliminate'

    if opt_level >= 2:
      js_optimizer_queue += [get_eliminate()]

      if shared.Settings.AGGRESSIVE_VARIABLE_ELIMINATION:
        # note that this happens before registerize/minification, which can obfuscate the name of 'label', which is tricky
        js_optimizer_queue += ['aggressiveVariableElimination']

      js_optimizer_queue += ['simplifyExpressions']

      if shared.Settings.EMTERPRETIFY:
        # emterpreter code will not run through a JS optimizing JIT, do more work ourselves
        js_optimizer_queue += ['localCSE']
        # add explicit label setting, as we will run aggressiveVariableElimination late, *after* 'label' is no longer notable by name
        js_optimizer_queue += ['safeLabelSetting']

  if shared.Settings.EMTERPRETIFY:
    # add explicit label setting, as we will run aggressiveVariableElimination late, *after* 'label' is no longer notable by name
    js_optimizer_queue += ['safeLabelSetting']

  if opt_level >= 1 and js_opts:
    if opt_level >= 2:
      # simplify ifs if it is ok to make the code somewhat unreadable, and unless outlining (simplified ifs
      # with commaified code breaks late aggressive variable elimination)
      if shared.Settings.SIMPLIFY_IFS and (debug_level == 0 or profiling) and shared.Settings.OUTLINING_LIMIT == 0: js_optimizer_queue += ['simplifyIfs']

      if shared.Settings.PRECISE_F32: js_optimizer_queue += ['optimizeFrounds']

  def do_minify():
    global js_optimizer_queue

    if opt_level >= 2:
      if debug_level < 2 and not closure == 2:
        js_optimizer_queue += ['minifyNames']
      if debug_level == 0:
        global minify_whitespace
        minify_whitespace = True

    if closure == 1:
      js_optimizer_queue += ['closure']
    elif debug_level <= 2 and shared.Settings.FINALIZE_ASM_JS and not closure:
      js_optimizer_queue += ['cleanup']

  if js_opts:
    if shared.Settings.SAFE_HEAP: js_optimizer_queue += ['safeHeap']

    if shared.Settings.OUTLINING_LIMIT > 0:
      js_optimizer_queue += ['outline']
      js_optimizer_extra_info['sizeToOutline'] = shared.Settings.OUTLINING_LIMIT

    if opt_level >= 2 and debug_level < 3:
      if opt_level >= 3 or shrink_level > 0:
        js_optimizer_queue += ['registerizeHarder']
      else:
        js_optimizer_queue += ['registerize']

    if not shared.Settings.EMTERPRETIFY:
      do_minify()

    if opt_level >= 2:
      js_optimizer_queue += ['asmLastOpts']

    if shared.Settings.FINALIZE_ASM_JS: js_optimizer_queue += ['last']

    flush_js_optimizer_queue()

  if closure == 2:
    flush_js_optimizer_queue()

    logging.debug('running closure')
    # no need to add this to js_transform_tempfiles, because closure and
    # debug_level > 0 are never simultaneously true
    final = shared.Building.closure_compiler(final, pretty=debug_level >= 1)
    if DEBUG: save_intermediate('closure')

  log_time('js opts')

  if shared.Settings.MODULARIZE:
    logging.debug('Modularizing, assigning to var ' + shared.Settings.EXPORT_NAME)
    src = open(final).read()
    final = final + '.modular.js'
    f = open(final, 'w')
    f.write('var ' + shared.Settings.EXPORT_NAME + ' = function(Module) {\n')
    f.write('  Module = Module || {};\n')
    f.write('\n')
    f.write(src)
    f.write('\n')
    f.write('  return Module;\n')
    f.write('};\n')
    f.close()
    src = None

  js_target = unsuffixed(target) + '.js'

  if shared.Settings.EMTERPRETIFY:
    flush_js_optimizer_queue()
    logging.debug('emterpretifying')
    import json
    try:
      # move temp js to final position, alongside its mem init file
      shutil.move(final, js_target)
      args = [shared.PYTHON, shared.path_from_root('tools', 'emterpretify.py'), js_target, final + '.em.js', json.dumps(shared.Settings.EMTERPRETIFY_BLACKLIST), json.dumps(shared.Settings.EMTERPRETIFY_WHITELIST), '', str(shared.Settings.SWAPPABLE_ASM_MODULE)]
      if shared.Settings.EMTERPRETIFY_ASYNC:
        args += ['ASYNC=1']
      if shared.Settings.EMTERPRETIFY_ADVISE:
        args += ['ADVISE=1']
      if profiling or profiling_funcs:
        args += ['PROFILING=1']
      if shared.Settings.ASSERTIONS:
        args += ['ASSERTIONS=1']
      if shared.Settings.PRECISE_F32:
        args += ['FROUND=1']
      if shared.Settings.ALLOW_MEMORY_GROWTH:
        args += ['MEMORY_SAFE=1']
      execute(args)
      final = final + '.em.js'
    finally:
      shared.try_delete(js_target)

    if shared.Settings.EMTERPRETIFY_ADVISE:
      logging.warning('halting compilation due to EMTERPRETIFY_ADVISE')
      sys.exit(0)

    # minify (if requested) after emterpreter processing, and finalize output
    logging.debug('finalizing emterpreted code')
    shared.Settings.FINALIZE_ASM_JS = 1
    do_minify()
    js_optimizer_queue += ['last']
    flush_js_optimizer_queue()

    # finalize the original as well, if we will be swapping it in (TODO: add specific option for this)
    if shared.Settings.SWAPPABLE_ASM_MODULE:
      real = final
      original = js_target + '.orig.js' # the emterpretify tool saves the original here
      final = original
      logging.debug('finalizing original (non-emterpreted) code at ' + final)
      do_minify()
      js_optimizer_queue += ['last']
      flush_js_optimizer_queue()
      safe_move(final, original)
      final = real

  # Remove some trivial whitespace # TODO: do not run when compress has already been done on all parts of the code
  #src = open(final).read()
  #src = re.sub(r'\n+[ \n]*\n+', '\n', src)
  #open(final, 'w').write(src)

  def generate_source_map(map_file_base_name, offset=0):
    jsrun.run_js(shared.path_from_root('tools', 'source-maps', 'sourcemapper.js'),
      shared.NODE_JS, js_transform_tempfiles +
        ['--sourceRoot', os.getcwd(),
         '--mapFileBaseName', map_file_base_name,
         '--offset', str(offset)])

  # If we were asked to also generate HTML, do that
  if final_suffix == 'html':
    logging.debug('generating HTML')
    shell = open(shell_path).read()
    assert '{{{ SCRIPT }}}' in shell, 'HTML shell must contain  {{{ SCRIPT }}}  , see src/shell.html for an example'
    html = open(target, 'w')
    base_js_target = os.path.basename(js_target)
    if proxy_to_worker:
      child_js = shared.Settings.PROXY_TO_WORKER_FILENAME or target_basename
      html.write(shell.replace('{{{ SCRIPT }}}', '''
<script>
  if ((',' + window.location.search.substr(1) + ',').indexOf(',noProxy,') < 0) {
    console.log('running code in a web worker');
''' + open(shared.path_from_root('src', 'webGLClient.js')).read() + '\n' + shared.read_and_preprocess(shared.path_from_root('src', 'proxyClient.js')).replace('{{{ filename }}}', child_js).replace('{{{ IDBStore.js }}}', open(shared.path_from_root('src', 'IDBStore.js')).read()) + '''
  } else {
    // note: no support for code mods (PRECISE_F32==2)
    console.log('running code on the main thread');
    var script = document.createElement('script');
    script.src = "%s.js";
    document.body.appendChild(script);
  }
</script>''' % child_js))
      shutil.move(final, js_target)
    elif not Compression.on:
      # Normal code generation path
      if debug_level >= 4:
        generate_source_map(target)
      shutil.move(final, js_target)
      need_mods = shared.Settings.PRECISE_F32 == 2
      if not need_mods:
        # Non-modifiable code, just load the code directly
        script_tag = '''<script async type="text/javascript" src="%s"></script>''' % base_js_target
      else:
        # Potentially-modifiable code, load as text, modify, then execute. This lets you
        # patch the code on the client machine right before it is executed, perhaps based
        # on information about the client.
        checks = []
        mods = []
        if shared.Settings.PRECISE_F32 == 2:
          checks.append('!Math.fround')
          if 'minifyNames' not in js_optimizer_queue_history:
            # simple dumb replace
            mods.append('''
console.log('optimizing out Math.fround calls');
code = code.replace(/Math_fround\(/g, '(').replace("'use asm'", "'almost asm'")
''')
          else:
            # minified, not quite so simple - TODO
            mods.append('''
try {
  console.log('optimizing out Math.fround calls');
  var m = /var ([^=]+)=global\.Math\.fround;/.exec(code);
  var minified = m[1];
  if (!minified) throw 'fail';
  var startAsm = code.indexOf('// EMSCRIPTEN_START_FUNCS');
  var endAsm = code.indexOf('// EMSCRIPTEN_END_FUNCS');
  var asm = code.substring(startAsm, endAsm);
  do {
    var moar = false; // we need to re-do, as x(x( will not be fixed
    asm = asm.replace(new RegExp('[^a-zA-Z0-9\\\\$\\\\_]' + minified + '\\\\(', 'g'), function(s) { moar = true; return s[0] + '(' });
  } while (moar);
  code = code.substring(0, startAsm) + asm + code.substring(endAsm);
  code = code.replace("'use asm'", "'almost asm'");
} catch(e) { console.log('failed to optimize out Math.fround calls ' + e) }
''')

        fixes = ''
        for i in range(len(checks)):
          fixes += 'if (' + checks[i] + ') { ' + mods[i] + ' }\n'

        # if all the checks are negative, just emit a script tag normally, that's better.
        # otherwise, do an xhr to get the code as text, modify, and load asynchronously
        code = 'if (!(' + ' || '.join(checks) + ''')) {
  var script = document.createElement('script');
  script.src = "''' + base_js_target + '''";
  document.body.appendChild(script);
} else {
  var codeXHR = new XMLHttpRequest();
  codeXHR.open('GET', '%s', true);
  codeXHR.onload = function() {
    var code = codeXHR.responseText;
    %s
    var blob = new Blob([code], { type: 'text/javascript' });
    codeXHR = null;
    var src = URL.createObjectURL(blob);
    var script = document.createElement('script');
    script.src = URL.createObjectURL(blob);
    script.onload = function() {
      URL.revokeObjectURL(script.src);
    };
    document.body.appendChild(script);
  };
  codeXHR.send(null);
}
''' % (base_js_target, fixes)
        script_tag = '''<script>%s</script>''' % code
      html.write(shell.replace('{{{ SCRIPT }}}', script_tag))
    else:
      # Compress the main code
      shutil.move(final, js_target)
      Compression.compress(js_target)

      # Run the decompressor in a worker, and add code to
      #   1. download the compressed file
      #   2. decompress to a typed array
      #   3. convert to a string of source code
      #   4. insert a script element with that source code (more effective than eval)
      decoding = '''
        var decompressWorker = new Worker('decompress.js');
        var decompressCallbacks = [];
        var decompressions = 0;
        Module["decompress"] = function(data, callback) {
          var id = decompressCallbacks.length;
          decompressCallbacks.push(callback);
          decompressWorker.postMessage({ data: data, id: id });
          if (Module['setStatus']) {
            decompressions++;
            Module['setStatus']('Decompressing...');
          }
        };
        decompressWorker.onmessage = function(event) {
          decompressCallbacks[event.data.id](event.data.data);
          decompressCallbacks[event.data.id] = null;
          if (Module['setStatus']) {
            decompressions--;
            if (decompressions == 0) {
              Module['setStatus']('');
            }
          }
        };
        var compiledCodeXHR = new XMLHttpRequest();
        compiledCodeXHR.open('GET', '%s', true);
        compiledCodeXHR.responseType = 'arraybuffer';
        compiledCodeXHR.onload = function() {
          var arrayBuffer = compiledCodeXHR.response;
          if (!arrayBuffer) throw('Loading compressed code failed.');
          var byteArray = new Uint8Array(arrayBuffer);
          Module.decompress(byteArray, function(decompressed) {
            var source = Array.prototype.slice.apply(decompressed).map(function(x) { return String.fromCharCode(x) }).join(''); // createObjectURL instead?
            var scriptTag = document.createElement('script');
            scriptTag.setAttribute('type', 'text/javascript');
            scriptTag.innerHTML = source;
            document.body.appendChild(scriptTag);
          });
        };
        compiledCodeXHR.send(null);
''' % Compression.compressed_name(base_js_target)
      html.write(shell.replace('{{{ SCRIPT }}}', '<script>' + decoding + '</script>'))

      # Add decompressor with web worker glue code
      decompressor = open('decompress.js', 'w')
      decompressor.write(open(Compression.decoder).read())
      decompressor.write('''
        onmessage = function(event) {
          postMessage({ data: %s(event.data.data), id: event.data.id });
        };
''' % Compression.js_name)
      decompressor.close()

    html.close()
  else:
    if debug_level >= 4: generate_source_map(target)
    if proxy_to_worker:
      worker_target_basename = target_basename + '.worker'
      open(target, 'w').write(open(shared.path_from_root('src', 'webGLClient.js')).read() + '\n' + open(shared.path_from_root('src', 'proxyClient.js')).read().replace('{{{ filename }}}', shared.Settings.PROXY_TO_WORKER_FILENAME or worker_target_basename).replace('{{{ IDBStore.js }}}', open(shared.path_from_root('src', 'IDBStore.js')).read()))
      shutil.move(final, target[:-3] + '.worker.js')
    else:
      # copy final JS to output normally
      shutil.move(final, target)

  log_time('final emitting')

  if DEBUG: logging.debug('total time: %.2f seconds', (time.time() - start_time))

finally:
  if not TEMP_DIR:
    try:
      shutil.rmtree(temp_dir)
    except:
      pass
  else:
    logging.info('emcc saved files are in:' + temp_dir)
  misc_temp_files.clean()