Move some code out of tools/building.py. NFC (#21782)

Some of these function are only called in single file so
there is no need for them to be in this shared location.

Author: sbc100

test/test_other.py

@@ -58,6 +58,24 @@
 wasm_opt = Path(building.get_binaryen_bin(), 'wasm-opt')
 
 
+def is_bitcode(filename):
+  try:
+    # look for magic signature
+    b = open(filename, 'rb').read(4)
+    if b[:2] == b'BC':
+      return True
+    # on macOS, there is a 20-byte prefix which starts with little endian
+    # encoding of 0x0B17C0DE
+    elif b == b'\xDE\xC0\x17\x0B':
+      b = bytearray(open(filename, 'rb').read(22))
+      return b[20:] == b'BC'
+  except IndexError:
+    # not enough characters in the input
+    # note that logging will be done on the caller function
+    pass
+  return False
+
+
 def uses_canonical_tmp(func):
   """Decorator that signals the use of the canonical temp by a test method.
 
@@ -5048,7 +5066,7 @@ def test_bitcode_linking(self):
     self.run_process([EMCC, '-flto', '-c', test_file('hello_world.c')])
     self.assertExists('hello_world.o')
     self.run_process([EMCC, '-flto', '-r', 'hello_world.o', '-o', 'hello_world2.o'])
-    building.is_bitcode('hello_world.o')
+    is_bitcode('hello_world.o')
     building.is_wasm('hello_world2.o')
 
   @parameterized({
@@ -5588,7 +5606,7 @@ def test_emit_llvm_asm(self):
 
   def test_emit_llvm(self):
     self.run_process([EMCC, test_file('hello_world.c'), '-c', '-emit-llvm'])
-    self.assertTrue(building.is_bitcode('hello_world.bc'))
+    self.assertTrue(is_bitcode('hello_world.bc'))
 
   def test_compile_ll_file(self):
     self.run_process([EMCC, test_file('hello_world.c'), '-S', '-emit-llvm'])
@@ -8775,12 +8793,12 @@ def test_lto(self):
       print('wasm in object')
       self.run_process([EMXX, src] + args + ['-c', '-o', 'hello_obj.o'])
       self.assertTrue(building.is_wasm('hello_obj.o'))
-      self.assertFalse(building.is_bitcode('hello_obj.o'))
+      self.assertFalse(is_bitcode('hello_obj.o'))
 
       print('bitcode in object')
       self.run_process([EMXX, src] + args + ['-c', '-o', 'hello_bitcode.o', '-flto'])
       self.assertFalse(building.is_wasm('hello_bitcode.o'))
-      self.assertTrue(building.is_bitcode('hello_bitcode.o'))
+      self.assertTrue(is_bitcode('hello_bitcode.o'))
 
       print('use bitcode object (LTO)')
       self.run_process([EMXX, 'hello_bitcode.o'] + args + ['-flto'])
@@ -8813,7 +8831,7 @@ def test_lto_flags(self):
       (['-sWASM_OBJECT_FILES'], False),
     ]:
       self.run_process([EMCC, test_file('hello_world.c')] + flags + ['-c', '-o', 'a.o'])
-      seen_bitcode = building.is_bitcode('a.o')
+      seen_bitcode = is_bitcode('a.o')
       self.assertEqual(expect_bitcode, seen_bitcode, 'must emit LTO-capable bitcode when flags indicate so (%s)' % str(flags))
 
   # We have LTO tests covered in 'wasmltoN' targets in test_core.py, but they
@@ -10181,17 +10199,17 @@ def test_is_bitcode(self):
 
     with open(fname, 'wb') as f:
       f.write(b'foo')
-    self.assertFalse(building.is_bitcode(fname))
+    self.assertFalse(is_bitcode(fname))
 
     with open(fname, 'wb') as f:
       f.write(b'\xDE\xC0\x17\x0B')
       f.write(16 * b'\x00')
       f.write(b'BC')
-    self.assertTrue(building.is_bitcode(fname))
+    self.assertTrue(is_bitcode(fname))
 
     with open(fname, 'wb') as f:
       f.write(b'BC')
-    self.assertTrue(building.is_bitcode(fname))
+    self.assertTrue(is_bitcode(fname))
 
   def test_is_ar(self):
     fname = 'tmp.a'
@@ -11987,7 +12005,7 @@ def test_bitcode_input(self):
     # Verify that bitcode files are accepted as input
     create_file('main.c', 'void foo(); int main() { return 0; }')
     self.run_process([EMCC, '-emit-llvm', '-c', '-o', 'main.bc', 'main.c'])
-    self.assertTrue(building.is_bitcode('main.bc'))
+    self.assertTrue(is_bitcode('main.bc'))
     self.run_process([EMCC, '-c', '-o', 'main.o', 'main.bc'])
     self.assertTrue(building.is_wasm('main.o'))
 

tools/building.py

@@ -32,7 +32,7 @@
 from .shared import LLVM_DWARFDUMP, demangle_c_symbol_name
 from .shared import get_emscripten_temp_dir, exe_suffix, is_c_symbol
 from .utils import WINDOWS
-from .settings import settings, default_setting
+from .settings import settings
 from .feature_matrix import UNSUPPORTED
 
 logger = logging.getLogger('building')
@@ -46,20 +46,6 @@
 user_requested_exports: Set[str] = set()
 
 
-# .. but for Popen, we cannot have doublequotes, so provide functionality to
-# remove them when needed.
-def remove_quotes(arg):
-  if isinstance(arg, list):
-    return [remove_quotes(a) for a in arg]
-
-  if arg.startswith('"') and arg.endswith('"'):
-    return arg[1:-1].replace('\\"', '"')
-  elif arg.startswith("'") and arg.endswith("'"):
-    return arg[1:-1].replace("\\'", "'")
-  else:
-    return arg
-
-
 def get_building_env():
   cache.ensure()
   env = os.environ.copy()
@@ -1063,34 +1049,15 @@ def handle_final_wasm_symbols(wasm_file, symbols_file, debug_info):
 
 
 def is_ar(filename):
+  """Return True if a the given filename is an ar archive, False otherwise.
+  """
   try:
-    if _is_ar_cache.get(filename):
-      return _is_ar_cache[filename]
     header = open(filename, 'rb').read(8)
-    sigcheck = header in (b'!<arch>\n', b'!<thin>\n')
-    _is_ar_cache[filename] = sigcheck
-    return sigcheck
   except Exception as e:
     logger.debug('is_ar failed to test whether file \'%s\' is a llvm archive file! Failed on exception: %s' % (filename, e))
     return False
 
-
-def is_bitcode(filename):
-  try:
-    # look for magic signature
-    b = open(filename, 'rb').read(4)
-    if b[:2] == b'BC':
-      return True
-    # on macOS, there is a 20-byte prefix which starts with little endian
-    # encoding of 0x0B17C0DE
-    elif b == b'\xDE\xC0\x17\x0B':
-      b = bytearray(open(filename, 'rb').read(22))
-      return b[20:] == b'BC'
-  except IndexError:
-    # not enough characters in the input
-    # note that logging will be done on the caller function
-    pass
-  return False
+  return header in (b'!<arch>\n', b'!<thin>\n')
 
 
 def is_wasm(filename):
@@ -1113,85 +1080,6 @@ def is_wasm_dylib(filename):
   return False
 
 
-def map_to_js_libs(library_name):
-  """Given the name of a special Emscripten-implemented system library, returns an
-  pair containing
-  1. Array of absolute paths to JS library files, inside emscripten/src/ that corresponds to the
-     library name. `None` means there is no mapping and the library will be processed by the linker
-     as a require for normal native library.
-  2. Optional name of a corresponding native library to link in.
-  """
-  # Some native libraries are implemented in Emscripten as system side JS libraries
-  library_map = {
-    'embind': ['embind/embind.js', 'embind/emval.js'],
-    'EGL': ['library_egl.js'],
-    'GL': ['library_webgl.js', 'library_html5_webgl.js'],
-    'webgl.js': ['library_webgl.js', 'library_html5_webgl.js'],
-    'GLESv2': ['library_webgl.js'],
-    # N.b. there is no GLESv3 to link to (note [f] in https://www.khronos.org/registry/implementers_guide.html)
-    'GLEW': ['library_glew.js'],
-    'glfw': ['library_glfw.js'],
-    'glfw3': ['library_glfw.js'],
-    'GLU': [],
-    'glut': ['library_glut.js'],
-    'openal': ['library_openal.js'],
-    'X11': ['library_xlib.js'],
-    'SDL': ['library_sdl.js'],
-    'uuid': ['library_uuid.js'],
-    'fetch': ['library_fetch.js'],
-    'websocket': ['library_websocket.js'],
-    # These 4 libraries are separate under glibc but are all rolled into
-    # libc with musl.  For compatibility with glibc we just ignore them
-    # completely.
-    'dl': [],
-    'm': [],
-    'rt': [],
-    'pthread': [],
-    # This is the name of GNU's C++ standard library. We ignore it here
-    # for compatibility with GNU toolchains.
-    'stdc++': [],
-  }
-  settings_map = {
-    'glfw': {'USE_GLFW': 2},
-    'glfw3': {'USE_GLFW': 3},
-    'SDL': {'USE_SDL': 1},
-  }
-
-  if library_name in settings_map:
-    for key, value in settings_map[library_name].items():
-      default_setting(key, value)
-
-  if library_name in library_map:
-    libs = library_map[library_name]
-    logger.debug('Mapping library `%s` to JS libraries: %s' % (library_name, libs))
-    return libs
-
-  if library_name.endswith('.js') and os.path.isfile(path_from_root('src', f'library_{library_name}')):
-    return [f'library_{library_name}']
-
-  return None
-
-
-# Map a linker flag to a settings. This lets a user write -lSDL2 and it will
-# have the same effect as -sUSE_SDL=2.
-def map_and_apply_to_settings(library_name):
-  # most libraries just work, because the -l name matches the name of the
-  # library we build. however, if a library has variations, which cause us to
-  # build multiple versions with multiple names, then we need this mechanism.
-  library_map = {
-    # SDL2_mixer's built library name contains the specific codecs built in.
-    'SDL2_mixer': [('USE_SDL_MIXER', 2)],
-  }
-
-  if library_name in library_map:
-    for key, value in library_map[library_name]:
-      logger.debug('Mapping library `%s` to settings changes: %s = %s' % (library_name, key, value))
-      setattr(settings, key, value)
-    return True
-
-  return False
-
-
 def emit_wasm_source_map(wasm_file, map_file, final_wasm):
   # source file paths must be relative to the location of the map (which is
   # emitted alongside the wasm)

tools/link.py

@@ -1978,6 +1978,20 @@ def phase_embind_aot(wasm_target, js_syms):
   write_file(final_js, src)
 
 
+# for Popen, we cannot have doublequotes, so provide functionality to
+# remove them when needed.
+def remove_quotes(arg):
+  if isinstance(arg, list):
+    return [remove_quotes(a) for a in arg]
+
+  if arg.startswith('"') and arg.endswith('"'):
+    return arg[1:-1].replace('\\"', '"')
+  elif arg.startswith("'") and arg.endswith("'"):
+    return arg[1:-1].replace("\\'", "'")
+  else:
+    return arg
+
+
 @ToolchainProfiler.profile_block('source transforms')
 def phase_source_transforms(options):
   # Apply a source code transformation, if requested
@@ -1986,7 +2000,7 @@ def phase_source_transforms(options):
   final_js += '.tr.js'
   posix = not shared.WINDOWS
   logger.debug('applying transform: %s', options.js_transform)
-  shared.check_call(building.remove_quotes(shlex.split(options.js_transform, posix=posix) + [os.path.abspath(final_js)]))
+  shared.check_call(remove_quotes(shlex.split(options.js_transform, posix=posix) + [os.path.abspath(final_js)]))
   save_intermediate('transformed')
 
 
@@ -2637,6 +2651,85 @@ def find_library(lib, lib_dirs):
   return None
 
 
+def map_to_js_libs(library_name):
+  """Given the name of a special Emscripten-implemented system library, returns an
+  pair containing
+  1. Array of absolute paths to JS library files, inside emscripten/src/ that corresponds to the
+     library name. `None` means there is no mapping and the library will be processed by the linker
+     as a require for normal native library.
+  2. Optional name of a corresponding native library to link in.
+  """
+  # Some native libraries are implemented in Emscripten as system side JS libraries
+  library_map = {
+    'embind': ['embind/embind.js', 'embind/emval.js'],
+    'EGL': ['library_egl.js'],
+    'GL': ['library_webgl.js', 'library_html5_webgl.js'],
+    'webgl.js': ['library_webgl.js', 'library_html5_webgl.js'],
+    'GLESv2': ['library_webgl.js'],
+    # N.b. there is no GLESv3 to link to (note [f] in https://www.khronos.org/registry/implementers_guide.html)
+    'GLEW': ['library_glew.js'],
+    'glfw': ['library_glfw.js'],
+    'glfw3': ['library_glfw.js'],
+    'GLU': [],
+    'glut': ['library_glut.js'],
+    'openal': ['library_openal.js'],
+    'X11': ['library_xlib.js'],
+    'SDL': ['library_sdl.js'],
+    'uuid': ['library_uuid.js'],
+    'fetch': ['library_fetch.js'],
+    'websocket': ['library_websocket.js'],
+    # These 4 libraries are separate under glibc but are all rolled into
+    # libc with musl.  For compatibility with glibc we just ignore them
+    # completely.
+    'dl': [],
+    'm': [],
+    'rt': [],
+    'pthread': [],
+    # This is the name of GNU's C++ standard library. We ignore it here
+    # for compatibility with GNU toolchains.
+    'stdc++': [],
+  }
+  settings_map = {
+    'glfw': {'USE_GLFW': 2},
+    'glfw3': {'USE_GLFW': 3},
+    'SDL': {'USE_SDL': 1},
+  }
+
+  if library_name in settings_map:
+    for key, value in settings_map[library_name].items():
+      default_setting(key, value)
+
+  if library_name in library_map:
+    libs = library_map[library_name]
+    logger.debug('Mapping library `%s` to JS libraries: %s' % (library_name, libs))
+    return libs
+
+  if library_name.endswith('.js') and os.path.isfile(utils.path_from_root('src', f'library_{library_name}')):
+    return [f'library_{library_name}']
+
+  return None
+
+
+# Map a linker flag to a settings. This lets a user write -lSDL2 and it will
+# have the same effect as -sUSE_SDL=2.
+def map_and_apply_to_settings(library_name):
+  # most libraries just work, because the -l name matches the name of the
+  # library we build. however, if a library has variations, which cause us to
+  # build multiple versions with multiple names, then we need this mechanism.
+  library_map = {
+    # SDL2_mixer's built library name contains the specific codecs built in.
+    'SDL2_mixer': [('USE_SDL_MIXER', 2)],
+  }
+
+  if library_name in library_map:
+    for key, value in library_map[library_name]:
+      logger.debug('Mapping library `%s` to settings changes: %s = %s' % (library_name, key, value))
+      setattr(settings, key, value)
+    return True
+
+  return False
+
+
 def process_libraries(state, linker_inputs):
   new_flags = []
   libraries = []
@@ -2652,7 +2745,7 @@ def process_libraries(state, linker_inputs):
 
     logger.debug('looking for library "%s"', lib)
 
-    js_libs = building.map_to_js_libs(lib)
+    js_libs = map_to_js_libs(lib)
     if js_libs is not None:
       libraries += [(i, js_lib) for js_lib in js_libs]
 
@@ -2667,7 +2760,7 @@ def process_libraries(state, linker_inputs):
     if js_libs is not None:
       continue
 
-    if building.map_and_apply_to_settings(lib):
+    if map_and_apply_to_settings(lib):
       continue
 
     path = None