[empath-split] Tidy up error checking (#25380)

This puts all parser-related error checking routines in `parse_args` and
all other error checking routines, such as checking whether files exist,
in `check_errors`. The reason parser-related errors better be in
`parse_args` is we can use `parser.error` there, which prints the usage.

Also this checks for one more thing: whether `-o` or `--output` is
provided. This is forwarded to `wasm-split` and it is a mandatory option
for `wasm-split --multi-split`, but because `-o is required` error was
printed by not this script but `wasm-split`, it was printed later and
the message could be buried in warnings, which was confusing. We check
it here directly and error out when one is not provided.

Author: aheejin

tools/empath-split.py

@@ -68,13 +68,50 @@ def parse_args():
   parser.add_argument('--preserve-manifest', action='store_true',
                       help='Preserve generated manifest file. This sets --verbose too.')
   args, forwarded_args = parser.parse_known_args()
-  if '--manifest' in forwarded_args:
-    exit_with_error('manifest file will be generated by this script and should not be given')
+
   if args.preserve_manifest:
     args.verbose = True
+  if not args.wasm_split:
+    args.wasm_split = os.path.join(building.get_binaryen_bin(), shared.exe_suffix('wasm-split'))
+
+  if '--manifest' in forwarded_args:
+    parser.error('manifest file will be generated by this script and should not be given')
+  if '-o' not in forwarded_args and '--output' not in forwarded_args:
+    parser.error('-o (--output) is required')
   return args, forwarded_args
 
 
+def check_errors(args):
+  if args.wasm and not os.path.isfile(args.wasm):
+    exit_with_error(f"'{args.wasm}' was not found or not a file")
+  if args.paths_file and not os.path.isfile(args.paths_file):
+    exit_with_error(f"'{args.paths_file}' was not found or not a file")
+
+  if args.sourcemap:
+    if not os.path.isfile(args.sourcemap):
+      exit_with_error(f"'{args.sourcemap}' was not found or not a file")
+
+  if args.wasm:
+    with webassembly.Module(args.wasm) as module:
+      if not args.sourcemap and not emsymbolizer.get_sourceMappingURL_section(module):
+        exit_with_error('sourceMappingURL section does not exist')
+      sourcemap = module.get_sourceMappingURL()
+      if not os.path.isfile(sourcemap):
+        exit_with_error(f"'{sourcemap}' was not found or not a file")
+      if not module.has_name_section():
+        exit_with_error('Name section does not eixst')
+
+  if not os.path.isfile(args.wasm_split):
+    exit_with_error(f"'{args.wasm_split}' was not found or not a file")
+
+
+def get_sourceMappingURL(wasm, arg_sourcemap):
+  if arg_sourcemap:
+    return arg_sourcemap
+  with webassembly.Module(wasm) as module:
+    return module.get_sourceMappingURL()
+
+
 def get_path_to_functions_map(wasm, sourcemap, paths):
   def is_synthesized_func(func):
     # TODO There can be more
@@ -98,21 +135,13 @@ def is_synthesized_func(func):
   # {src file: list of functions} map, and construct {path: list of functions}
   # map from it
   with webassembly.Module(wasm) as module:
-    if not module.has_name_section():
-      exit_with_error('Name section does not eixst')
-    if not sourcemap:
-      if not emsymbolizer.get_sourceMappingURL_section(module):
-        exit_with_error('sourceMappingURL section does not exist')
-
     funcs = module.get_functions()
     func_names = module.get_function_names()
     assert len(funcs) == len(func_names)
 
     func_to_src = {}
     src_to_funcs = {}
 
-    if not sourcemap:
-      sourcemap = module.get_sourceMappingURL()
     sm = emsymbolizer.WasmSourceMap()
     sm.parse(sourcemap)
 
@@ -170,20 +199,9 @@ def is_synthesized_func(func):
 
 def main():
   args, forwarded_args = parse_args()
-  if args.wasm_split:
-    wasm_split = args.wasm_split
-  else:
-    wasm_split = os.path.join(building.get_binaryen_bin(), shared.exe_suffix('wasm-split'))
+  check_errors(args)
 
-  if not os.path.isfile(args.wasm):
-    exit_with_error(f"'{args.wasm}' was not found or not a file")
-  if not os.path.isfile(args.paths_file):
-    exit_with_error(f"'{args.paths_file}' was not found or not a file")
-  if args.sourcemap:
-    if not os.path.isfile(args.sourcemap):
-      exit_with_error(f"'{args.sourcemap}' was not found or not a file")
-  if not os.path.isfile(wasm_split):
-    exit_with_error(f"'{wasm_split}' was not found or not a file")
+  sourcemap = get_sourceMappingURL(args.wasm, args.sourcemap)
 
   paths = utils.read_file(args.paths_file).splitlines()
   paths = [utils.normalize_path(path.strip()) for path in paths if path.strip()]
@@ -193,7 +211,7 @@ def main():
   paths = list(dict.fromkeys(paths))
 
   # Compute {path: list of functions} map
-  path_to_funcs = get_path_to_functions_map(args.wasm, args.sourcemap, paths)
+  path_to_funcs = get_path_to_functions_map(args.wasm, sourcemap, paths)
 
   # Write .manifest file
   with tempfile.NamedTemporaryFile(suffix=".manifest", mode='w+', delete=args.preserve_manifest) as f:
@@ -213,7 +231,7 @@ def main():
         f.write('\n')
     f.flush()
 
-    cmd = [wasm_split, '--multi-split', args.wasm, '--manifest', manifest]
+    cmd = [args.wasm_split, '--multi-split', args.wasm, '--manifest', manifest]
     if args.verbose:
       # This option is used both in this script and wasm-split
       cmd.append('-v')