Use python context manager when reading webassembly modules (#17516)

Fixes: #17452

Author: sbc100

emscripten.py

@@ -306,11 +306,13 @@ def emscript(in_wasm, out_wasm, outfile_js, memfile):
   update_settings_glue(out_wasm, metadata)
 
   if not settings.WASM_BIGINT and metadata['emJsFuncs']:
-    module = webassembly.Module(in_wasm)
-    types = module.get_types()
     import_map = {}
-    for imp in module.get_imports():
-      import_map[imp.field] = imp
+
+    with webassembly.Module(in_wasm) as module:
+      types = module.get_types()
+      for imp in module.get_imports():
+        import_map[imp.field] = imp
+
     for em_js_func, raw in metadata.get('emJsFuncs', {}).items():
       c_sig = raw.split('<::>')[0].strip('()')
       if not c_sig or c_sig == 'void':

emsymbolizer.py

@@ -183,28 +183,28 @@ def symbolize_address_sourcemap(module, address, force_file):
 
 
 def main(args):
-  module = webassembly.Module(args.wasm_file)
-  base = 16 if args.address.lower().startswith('0x') else 10
-  address = int(args.address, base)
-  symbolized = 0
-
-  if args.addrtype == 'code':
-    address += get_codesec_offset(module)
-
-  if ((has_debug_line_section(module) and not args.source) or
-     'dwarf' in args.source):
-    symbolize_address_dwarf(module, address)
-    symbolized += 1
-
-  if ((get_sourceMappingURL_section(module) and not args.source) or
-     'sourcemap' in args.source):
-    symbolize_address_sourcemap(module, address, args.file)
-    symbolized += 1
-
-  if not symbolized:
-    raise Error('No .debug_line or sourceMappingURL section found in '
-                f'{module.filename}.'
-                " I don't know how to symbolize this file yet")
+  with webassembly.Module(args.wasm_file) as module:
+    base = 16 if args.address.lower().startswith('0x') else 10
+    address = int(args.address, base)
+    symbolized = 0
+
+    if args.addrtype == 'code':
+      address += get_codesec_offset(module)
+
+    if ((has_debug_line_section(module) and not args.source) or
+       'dwarf' in args.source):
+      symbolize_address_dwarf(module, address)
+      symbolized += 1
+
+    if ((get_sourceMappingURL_section(module) and not args.source) or
+       'sourcemap' in args.source):
+      symbolize_address_sourcemap(module, address, args.file)
+      symbolized += 1
+
+    if not symbolized:
+      raise Error('No .debug_line or sourceMappingURL section found in '
+                  f'{module.filename}.'
+                  " I don't know how to symbolize this file yet")
 
 
 def get_args():

tests/test_core.py

@@ -7229,7 +7229,8 @@ def test_demangle_stacks(self, extra_args):
     self.do_core_test('test_demangle_stacks.cpp', assert_returncode=NON_ZERO)
 
     # there should be a name section in the file
-    self.assertTrue(webassembly.Module('test_demangle_stacks.wasm').has_name_section())
+    with webassembly.Module('test_demangle_stacks.wasm') as m:
+      self.assertTrue(m.has_name_section())
 
     print('without assertions, the stack is not printed, but a message suggesting assertions is')
     self.set_setting('ASSERTIONS', 0)
@@ -8117,7 +8118,8 @@ def test_asyncify_lists(self, args, should_pass, response=None):
       filename = 'test_asyncify_lists.wasm'
       # there should be no name section. sanitizers, however, always enable that
       if not is_sanitizing(self.emcc_args) and '--profiling-funcs' not in self.emcc_args:
-        self.assertFalse(webassembly.Module(filename).has_name_section())
+        with webassembly.Module(filename) as m:
+          self.assertFalse(m.has_name_section())
       # in a fully-optimized build, imports and exports are minified too and we
       # can verify that our function names appear nowhere
       if '-O3' in self.emcc_args:

tests/test_other.py

@@ -8680,14 +8680,14 @@ def test_separate_dwarf(self):
     self.assertNotIn(bytes(os.path.join('subdir', 'output.wasm.debug.wasm'), 'ascii'), wasm)
 
     # Check that the dwarf file has only dwarf, name, and non-code sections
-    debug_wasm = webassembly.Module('subdir/output.wasm.debug.wasm')
-    if not debug_wasm.has_name_section():
-      self.fail('name section not found in separate dwarf file')
-    for sec in debug_wasm.sections():
-      if sec.type == webassembly.SecType.CODE:
-        self.fail(f'section of type "{sec.type}" found in separate dwarf file')
-      if sec.name and sec.name != 'name' and not sec.name.startswith('.debug'):
-        self.fail(f'non-debug section "{sec.name}" found in separate dwarf file')
+    with webassembly.Module('subdir/output.wasm.debug.wasm') as debug_wasm:
+      if not debug_wasm.has_name_section():
+        self.fail('name section not found in separate dwarf file')
+      for sec in debug_wasm.sections():
+        if sec.type == webassembly.SecType.CODE:
+          self.fail(f'section of type "{sec.type}" found in separate dwarf file')
+        if sec.name and sec.name != 'name' and not sec.name.startswith('.debug'):
+          self.fail(f'non-debug section "{sec.name}" found in separate dwarf file')
 
     # Check that dwarfdump can dump the debug info
     dwdump = self.run_process(

tools/building.py

@@ -1354,12 +1354,12 @@ def is_wasm_dylib(filename):
   """Detect wasm dynamic libraries by the presence of the "dylink" custom section."""
   if not is_wasm(filename):
     return False
-  module = webassembly.Module(filename)
-  section = next(module.sections())
-  if section.type == webassembly.SecType.CUSTOM:
-    module.seek(section.offset)
-    if module.read_string() in ('dylink', 'dylink.0'):
-      return True
+  with webassembly.Module(filename) as module:
+    section = next(module.sections())
+    if section.type == webassembly.SecType.CUSTOM:
+      module.seek(section.offset)
+      if module.read_string() in ('dylink', 'dylink.0'):
+        return True
   return False
 
 

tools/extract_metadata.py

@@ -227,15 +227,15 @@ def update_metadata(filename, metadata):
   declares = []
   invoke_funcs = []
   em_js_funcs = set(metadata['emJsFuncs'])
-  module = webassembly.Module(filename)
-  for i in module.get_imports():
-    if i.kind == webassembly.ExternType.FUNC:
-      if i.field.startswith('invoke_'):
-        invoke_funcs.append(i.field)
-      elif i.field not in em_js_funcs:
-        declares.append(i.field)
-
-  exports = [e.name for e in module.get_exports() if e.kind in [webassembly.ExternType.FUNC, webassembly.ExternType.TAG]]
+  with webassembly.Module(filename) as module:
+    for i in module.get_imports():
+      if i.kind == webassembly.ExternType.FUNC:
+        if i.field.startswith('invoke_'):
+          invoke_funcs.append(i.field)
+        elif i.field not in em_js_funcs:
+          declares.append(i.field)
+
+    exports = [e.name for e in module.get_exports() if e.kind in [webassembly.ExternType.FUNC, webassembly.ExternType.TAG]]
   metadata['declares'] = declares
   metadata['exports'] = exports
   metadata['invokeFuncs'] = invoke_funcs
@@ -249,53 +249,54 @@ def get_string_at(module, address):
 
 
 def extract_metadata(filename):
-  module = webassembly.Module(filename)
   export_names = []
   declares = []
   invoke_funcs = []
   global_imports = []
   em_js_funcs = {}
-  exports = module.get_exports()
-  imports = module.get_imports()
-
-  for i in imports:
-    if i.kind == webassembly.ExternType.GLOBAL:
-      global_imports.append(i.field)
-
-  export_map = {e.name: e for e in exports}
-  for e in exports:
-    if e.kind == webassembly.ExternType.GLOBAL and e.name.startswith('__em_js__'):
-      name = e.name[len('__em_js__'):]
-      globl = module.get_global(e.index)
-      string_address = get_global_value(globl)
-      em_js_funcs[name] = get_string_at(module, string_address)
-
-  for i in imports:
-    if i.kind == webassembly.ExternType.FUNC:
-      if i.field.startswith('invoke_'):
-        invoke_funcs.append(i.field)
-      elif i.field not in em_js_funcs:
-        declares.append(i.field)
-
-  export_names = [e.name for e in exports if e.kind in [webassembly.ExternType.FUNC, webassembly.ExternType.TAG]]
-
-  features = module.parse_features_section()
-  features = ['--enable-' + f[1] for f in features if f[0] == '+']
-  features = [f.replace('--enable-atomics', '--enable-threads') for f in features]
-  features = [f.replace('--enable-simd128', '--enable-simd') for f in features]
-  features = [f.replace('--enable-nontrapping-fptoint', '--enable-nontrapping-float-to-int') for f in features]
-
-  # If main does not read its parameters, it will just be a stub that
-  # calls __original_main (which has no parameters).
-  metadata = {}
-  metadata['asmConsts'] = get_asm_strings(module, export_map)
-  metadata['declares'] = declares
-  metadata['emJsFuncs'] = em_js_funcs
-  metadata['exports'] = export_names
-  metadata['features'] = features
-  metadata['globalImports'] = global_imports
-  metadata['invokeFuncs'] = invoke_funcs
-  metadata['mainReadsParams'] = get_main_reads_params(module, export_map)
-  metadata['namedGlobals'] = get_named_globals(module, exports)
-  # print("Metadata parsed: " + pprint.pformat(metadata))
-  return metadata
+
+  with webassembly.Module(filename) as module:
+    exports = module.get_exports()
+    imports = module.get_imports()
+
+    for i in imports:
+      if i.kind == webassembly.ExternType.GLOBAL:
+        global_imports.append(i.field)
+
+    export_map = {e.name: e for e in exports}
+    for e in exports:
+      if e.kind == webassembly.ExternType.GLOBAL and e.name.startswith('__em_js__'):
+        name = e.name[len('__em_js__'):]
+        globl = module.get_global(e.index)
+        string_address = get_global_value(globl)
+        em_js_funcs[name] = get_string_at(module, string_address)
+
+    for i in imports:
+      if i.kind == webassembly.ExternType.FUNC:
+        if i.field.startswith('invoke_'):
+          invoke_funcs.append(i.field)
+        elif i.field not in em_js_funcs:
+          declares.append(i.field)
+
+    export_names = [e.name for e in exports if e.kind in [webassembly.ExternType.FUNC, webassembly.ExternType.TAG]]
+
+    features = module.parse_features_section()
+    features = ['--enable-' + f[1] for f in features if f[0] == '+']
+    features = [f.replace('--enable-atomics', '--enable-threads') for f in features]
+    features = [f.replace('--enable-simd128', '--enable-simd') for f in features]
+    features = [f.replace('--enable-nontrapping-fptoint', '--enable-nontrapping-float-to-int') for f in features]
+
+    # If main does not read its parameters, it will just be a stub that
+    # calls __original_main (which has no parameters).
+    metadata = {}
+    metadata['asmConsts'] = get_asm_strings(module, export_map)
+    metadata['declares'] = declares
+    metadata['emJsFuncs'] = em_js_funcs
+    metadata['exports'] = export_names
+    metadata['features'] = features
+    metadata['globalImports'] = global_imports
+    metadata['invokeFuncs'] = invoke_funcs
+    metadata['mainReadsParams'] = get_main_reads_params(module, export_map)
+    metadata['namedGlobals'] = get_named_globals(module, exports)
+    # print("Metadata parsed: " + pprint.pformat(metadata))
+    return metadata

tools/webassembly.py

@@ -192,8 +192,15 @@ def __init__(self, filename):
     self._cache = {}
 
   def __del__(self):
+    assert not self.buf, '`__exit__` should have already been called, please use context manager'
+
+  def __enter__(self):
+    return self
+
+  def __exit__(self, exc_type, exc_val, exc_tb): # noqa
     if self.buf:
       self.buf.close()
+      self.buf = None
 
   def read_at(self, offset, count):
     self.buf.seek(offset)
@@ -532,15 +539,15 @@ def get_global(self, idx):
 
 
 def parse_dylink_section(wasm_file):
-  module = Module(wasm_file)
-  return module.parse_dylink_section()
+  with Module(wasm_file) as module:
+    return module.parse_dylink_section()
 
 
 def get_exports(wasm_file):
-  module = Module(wasm_file)
-  return module.get_exports()
+  with Module(wasm_file) as module:
+    return module.get_exports()
 
 
 def get_imports(wasm_file):
-  module = Module(wasm_file)
-  return module.get_imports()
+  with Module(wasm_file) as module:
+    return module.get_imports()