Minor path handling cleanups. NFC. (#15018)

Use shared utilities where possible.  This makes this code more explicit
and also paves the way for more use of `pathlib` over raw strings in the
future.

Author: sbc100

emcc.py

@@ -55,7 +55,7 @@
 
 logger = logging.getLogger('emcc')
 
-# endings = dot + a suffix, safe to test by  filename.endswith(endings)
+# endings = dot + a suffix, compare against result of shared.suffix()
 C_ENDINGS = ('.c', '.i')
 CXX_ENDINGS = ('.cpp', '.cxx', '.cc', '.c++', '.CPP', '.CXX', '.C', '.CC', '.C++', '.ii')
 OBJC_ENDINGS = ('.m', '.mi')
@@ -753,7 +753,7 @@ def emsdk_ldflags(user_args):
   library_paths = [
      shared.Cache.get_lib_dir(absolute=True)
   ]
-  ldflags = ['-L' + l for l in library_paths]
+  ldflags = [f'-L{l}' for l in library_paths]
 
   if '-nostdlib' in user_args:
     return ldflags
@@ -1775,7 +1775,7 @@ def default_setting(name, new_default):
     state.forced_stdlibs.append('libfetch')
     settings.JS_LIBRARIES.append((0, 'library_fetch.js'))
     if settings.USE_PTHREADS:
-      settings.FETCH_WORKER_FILE = unsuffixed(os.path.basename(target)) + '.fetch.js'
+      settings.FETCH_WORKER_FILE = unsuffixed_basename(target) + '.fetch.js'
 
   if settings.DEMANGLE_SUPPORT:
     settings.EXPORTED_FUNCTIONS += ['___cxa_demangle']
@@ -1894,7 +1894,7 @@ def default_setting(name, new_default):
     building.user_requested_exports.add('_emscripten_current_thread_process_queued_calls')
 
     # set location of worker.js
-    settings.PTHREAD_WORKER_FILE = unsuffixed(os.path.basename(target)) + '.worker.js'
+    settings.PTHREAD_WORKER_FILE = unsuffixed_basename(target) + '.worker.js'
   else:
     settings.JS_LIBRARIES.append((0, 'library_pthread_stub.js'))
 
@@ -2090,7 +2090,7 @@ def check_memory_setting(setting):
     settings.MEM_INIT_IN_WASM = True
 
   # wasm side modules have suffix .wasm
-  if settings.SIDE_MODULE and target.endswith('.js'):
+  if settings.SIDE_MODULE and shared.suffix(target) == '.js':
     diagnostics.warning('emcc', 'output suffix .js requested, but wasm side modules are just wasm files; emitting only a .wasm, no .js')
 
   sanitize = set()
@@ -2350,10 +2350,11 @@ def get_language_mode(args):
   def use_cxx(src):
     if 'c++' in language_mode or run_via_emxx:
       return True
+    suffix = shared.suffix(src)
     # Next consider the filename
-    if src.endswith(C_ENDINGS + OBJC_ENDINGS):
+    if suffix in C_ENDINGS + OBJC_ENDINGS:
       return False
-    if src.endswith(CXX_ENDINGS):
+    if suffix in CXX_ENDINGS:
       return True
     # Finally fall back to the default
     if settings.DEFAULT_TO_CXX:
@@ -2393,7 +2394,7 @@ def get_clang_command_asm(src_file):
   if state.mode == Mode.PCH:
     headers = [header for _, header in input_files]
     for header in headers:
-      if not header.endswith(HEADER_ENDINGS):
+      if not shared.suffix(header) in HEADER_ENDINGS:
         exit_with_error(f'cannot mix precompiled headers with non-header inputs: {headers} : {header}')
       cmd = get_clang_command(header)
       if options.output_file:
@@ -2423,7 +2424,7 @@ def get_object_filename(input_file):
       return in_temp(unsuffixed(uniquename(input_file)) + options.default_object_extension)
 
   def compile_source_file(i, input_file):
-    logger.debug('compiling source file: ' + input_file)
+    logger.debug(f'compiling source file: {input_file}')
     output_file = get_object_filename(input_file)
     if state.mode not in (Mode.COMPILE_ONLY, Mode.PREPROCESS_ONLY):
       linker_inputs.append((i, output_file))
@@ -2452,10 +2453,10 @@ def compile_source_file(i, input_file):
     if file_suffix in SOURCE_ENDINGS + ASSEMBLY_ENDINGS or (state.has_dash_c and file_suffix == '.bc'):
       compile_source_file(i, input_file)
     elif file_suffix in DYNAMICLIB_ENDINGS:
-      logger.debug('using shared library: ' + input_file)
+      logger.debug(f'using shared library: {input_file}')
       linker_inputs.append((i, input_file))
     elif building.is_ar(input_file):
-      logger.debug('using static library: ' + input_file)
+      logger.debug(f'using static library: {input_file}')
       ensure_archive_index(input_file)
       linker_inputs.append((i, input_file))
     elif language_mode:
@@ -2464,7 +2465,7 @@ def compile_source_file(i, input_file):
       exit_with_error('-E or -x required when input is from standard input')
     else:
       # Default to assuming the inputs are object files and pass them to the linker
-      logger.debug('using object file: ' + input_file)
+      logger.debug(f'using object file: {input_file}')
       linker_inputs.append((i, input_file))
 
   return linker_inputs
@@ -2586,7 +2587,7 @@ def phase_source_transforms(options, target):
       file_args.append('--use-preload-plugins')
     if not settings.ENVIRONMENT_MAY_BE_NODE:
       file_args.append('--no-node')
-    file_code = shared.check_call([shared.FILE_PACKAGER, unsuffixed(target) + '.data'] + file_args, stdout=PIPE).stdout
+    file_code = shared.check_call([shared.FILE_PACKAGER, shared.replace_suffix(target, '.data')] + file_args, stdout=PIPE).stdout
     options.pre_js = js_manipulation.add_files_pre_js(options.pre_js, file_code)
 
   # Apply pre and postjs files
@@ -3529,7 +3530,7 @@ def generate_worker_js(target, js_target, target_basename):
 
   # compiler output goes in .worker.js file
   else:
-    move_file(js_target, unsuffixed(js_target) + '.worker.js')
+    move_file(js_target, shared.replace_suffix(js_target, '.worker.js'))
     worker_target_basename = target_basename + '.worker'
     proxy_worker_filename = (settings.PROXY_TO_WORKER_FILENAME or worker_target_basename) + '.js'
 

tests/common.py

@@ -1048,8 +1048,8 @@ def _build_and_run(self, filename, expected_output, args=[], output_nicerizer=No
       engines += self.wasm_engines
       if self.get_setting('WASM2C') and not EMTEST_LACKS_NATIVE_CLANG:
         # compile the c file to a native executable.
-        c = shared.unsuffixed(js_file) + '.wasm.c'
-        executable = shared.unsuffixed(js_file) + '.exe'
+        c = shared.replace_suffix(js_file, '.wasm.c')
+        executable = shared.replace_suffix(js_file, '.exe')
         cmd = [shared.CLANG_CC, c, '-o', executable] + clang_native.get_clang_native_args()
         self.run_process(cmd, env=clang_native.get_clang_native_env())
         # we can now run the executable directly, without an engine, which

tests/jsrun.py

@@ -18,7 +18,7 @@ def make_command(filename, engine, args=[]):
   # if no engine is needed, indicated by None, then there is a native executable
   # provided which we can just run
   if engine[0] is None:
-    executable = shared.unsuffixed(os.path.abspath(filename)) + '.exe'
+    executable = shared.replace_suffix(os.path.abspath(filename), '.exe')
     return [executable] + args
   if type(engine) is not list:
     engine = [engine]
@@ -44,7 +44,7 @@ def make_command(filename, engine, args=[]):
     command_flags += ['run']
   if is_wasmer or is_wasmtime:
     # in a wasm runtime, run the wasm, not the js
-    filename = shared.unsuffixed(filename) + '.wasm'
+    filename = shared.replace_suffix(filename, '.wasm')
   # Separates engine flags from script flags
   flag_separator = ['--'] if is_d8 or is_jsc else []
   return engine + command_flags + [filename] + shell_option_flags + flag_separator + args

tests/test_core.py

@@ -1735,7 +1735,7 @@ def test_emscripten_get_now(self):
 
   def test_emscripten_get_compiler_setting(self):
     src = test_file('core/emscripten_get_compiler_setting.c')
-    output = shared.unsuffixed(src) + '.out'
+    output = shared.replace_suffix(src, '.out')
     # with assertions, a nice message is shown
     self.set_setting('ASSERTIONS')
     self.do_runf(src, 'You must build with -s RETAIN_COMPILER_SETTINGS=1', assert_returncode=NON_ZERO)
@@ -3532,7 +3532,7 @@ def dylink_testf(self, main, side=None, expected=None, force_c=False, main_emcc_
     # Same as dylink_test but takes source code as filenames on disc.
     old_args = self.emcc_args.copy()
     if not expected:
-      outfile = shared.unsuffixed(main) + '.out'
+      outfile = shared.replace_suffix(main, '.out')
       expected = read_file(outfile)
     if not side:
       side, ext = os.path.splitext(main)

tools/building.py

@@ -530,7 +530,7 @@ def scan_archive_group(group):
 def get_command_with_possible_response_file(cmd):
   # 8k is a bit of an arbitrary limit, but a reasonable one
   # for max command line size before we use a response file
-  if len(' '.join(cmd)) <= 8192:
+  if len(shared.shlex_join(cmd)) <= 8192:
     return cmd
 
   logger.debug('using response file for %s' % cmd[0])
@@ -813,7 +813,7 @@ def move_to_safe_7bit_ascii_filename(filename):
   # Specify input file relative to the temp directory to avoid specifying non-7-bit-ASCII path names.
   args += ['--js', move_to_safe_7bit_ascii_filename(filename)]
   cmd = closure_cmd + args + user_args
-  logger.debug('closure compiler: ' + ' '.join(cmd))
+  logger.debug(f'closure compiler: {shared.shlex_join(cmd)}')
 
   # Closure compiler does not work if any of the input files contain characters outside the
   # 7-bit ASCII range. Therefore make sure the command line we pass does not contain any such
@@ -847,7 +847,7 @@ def move_to_safe_7bit_ascii_filename(filename):
     logger.error(proc.stderr) # print list of errors (possibly long wall of text if input was minified)
 
     # Exit and print final hint to get clearer output
-    msg = 'closure compiler failed (rc: %d): %s' % (proc.returncode, shared.shlex_join(cmd))
+    msg = f'closure compiler failed (rc: {proc.returncode}): {shared.shlex_join(cmd)}'
     if not pretty:
       msg += ' the error message may be clearer with -g1 and EMCC_DEBUG=2 set'
     exit_with_error(msg)