def __init__(self, lib_name, astnode_names, astnode_kinds, prefix):
"""
:param str lib_name: Lower-case name for the generated library.
:param list[str] astnode_names: List of camel-with-mixed-case names for
all node types.
:param dict[int, str] astnode_kinds: Mapping of kinds ('Enum_Rep) to
camel-with-mixed-case node names.
:param str prefix: Prefix to use for command names.
"""
self.lib_name = lib_name
self.astnode_names = [Name(name) for name in astnode_names]
self.astnode_kinds = {
kind: Name(name)
for kind, name in astnode_kinds.items()
}
self.prefix = prefix
self.node_record = ('{}__implementation__root_node_record'.format(
self.lib_name))
"""
Name of the record type used to represent node data.
"""
self.entity_struct_names = self._entity_struct_names()
self.reparse_debug_info()
def construct_operand(op):
from langkit.expressions import Cast, New
expr = construct(op)
check_source_language(
expr.type == LogicVarType
or expr.type.matches(T.root_node)
or expr.type.matches(T.root_node.env_el()),
"Operands to a logic bind operator should be either "
"a logic variable or an ASTNode, got {}".format(expr.type)
)
if expr.type.matches(T.root_node.env_el()):
if expr.type is not T.root_node.env_el():
expr = Cast.Expr(expr, T.root_node.env_el())
elif expr.type.matches(T.root_node):
# Cast the ast node type if necessary
if expr.type is not T.root_node:
expr = Cast.Expr(expr, T.root_node)
# If the expression is a root node, implicitly construct an
# env_element from it.
expr = New.StructExpr(T.root_node.env_el(), {
Name('El'): expr,
Name('MD'): LiteralExpr('<>', None),
Name('Parents_Bindings'): LiteralExpr('null', None)
})
return expr
def token_base_name(self, token):
"""
Helper function to get the name of a token.
:param TokenAction|Enum|Name|str token: Input token. It can be either a
TokenAction subclass (i.e. a Lexer subclass attribute), an enum
value from "self.tokens", the token Name or a string (case
insensitive token name).
:rtype: Name
"""
if isinstance(token, TokenAction):
return token.name
elif isinstance(token, Name):
assert token in self.tokens_set
return token
else:
assert isinstance(
token, str), ("Bad type for {}, supposed to be str|{}".format(
token, self.tokens.__name__))
name = Name.from_lower(token.lower())
if name in self.tokens_set:
return name
elif token in self.literals_map:
return self.literals_map[token].name
else:
check_source_language(
False,
"{} token literal is not part of the valid tokens for "
"this grammar".format(token))
def token_base_name(self, token):
"""
Helper function to get the name of a token.
:param TokenAction|Enum|Name|str token: Input token. It can be either a
TokenAction subclass (i.e. a Lexer subclass attribute), an enum
value from "self.tokens_class", the token Name or a string (case
insensitive token name).
:rtype: Name
"""
if isinstance(token, TokenAction):
return token.name
elif isinstance(token, Name):
assert token in self.tokens_set
return token
else:
assert isinstance(token, str), (
"Bad type for {}, supposed to be str|{}".format(
token, self.tokens_class.__name__
)
)
name = Name.from_lower(token.lower())
if name in self.tokens_set:
return name
elif token in self.literals_map:
return self.literals_map[token].name
else:
raise Exception(
"{} token literal is not part of the valid tokens for "
"this grammar".format(token)
)
def __getattr__(self, attr: str) -> TokenAction:
"""
Shortcut to get a TokenAction stored in self.tokens.
"""
name = Name.from_camel(attr)
try:
return self.tokens.name_to_token[name]
except KeyError:
raise AttributeError(f"No such token: {attr}")
def adaify_name(context: Context, name: str) -> str:
"""
Turn a symbol name like a__b into an Ada-like name such as A.B.
Also strip the $LIB_NAME.Analysis prefix, if present.
"""
pfx = context.analysis_prefix
if name.startswith(pfx):
name = name[len(pfx):]
chunks = name.split('__')
return '.'.join(Name.from_lower(c).camel_with_underscores for c in chunks)
def _astnode_struct_names(self):
"""
Turn the set of ASTNode subclass names into a mapping from ASTNode
record names, as GDB will see them, to user-friendly ASTNode names.
"""
return {
'{}__implementation__bare_{}_type'.format(
self.lib_name, name.lower()
): Name.from_camel_with_underscores(name)
for name in self.astnode_names
}
def _entity_struct_names(self):
"""
Turn the set of AST node names into a set of encoded type names for the
corresponding entity records.
"""
return {
'{}__analysis__implementation__entity_{}'.format(
self.lib_name,
Name.from_camel_with_underscores(name).lower
) for name in self.astnode_names
} | {'{}__analysis__implementation__ast_envs__entity'
.format(self.lib_name)}
def __new__(mcs, name, bases, dct):
assert len(bases) == 1, (
"Multiple inheritance for LexerToken subclasses is not supported"
)
fields = []
for fld_name, fld_value in dct.items():
if isinstance(fld_value, TokenAction):
fld_value.name = Name.from_camel(fld_name)
fields.append(fld_value)
dct['fields'] = getattr(bases[0], 'fields', []) + fields
return type.__new__(mcs, name, bases, dct)
def add_tokens(self, cls):
for fld_name, fld_value in cls.__dict__.items():
if isinstance(fld_value, TokenAction):
dest_list = self.tokens
elif isinstance(fld_value, TokenFamily):
dest_list = self.token_families
else:
continue
# Several items here are shared: for example, the
# LexerToken.LexingFailure instance can be used in two different
# lexers, so we can't assume its name is always None. Just accept
# when it has already the expected name.
name = Name.from_camel(fld_name)
assert fld_value.name in (None, name)
fld_value.name = name
dest_list.append(fld_value)
def check_token_families(self, context):
"""
Pass that checks that either there are no defined token families, or
that they form a partition of existing tokens.
"""
def format_token_list(tokens):
return ', '.join(sorted(
t.dsl_name if isinstance(t, TokenAction) else str(t)
for t in tokens
))
# Sort token families by name to ensure legality checks and code
# generation determinism.
self.tokens.token_families.sort(key=lambda tf: tf.name)
all_tokens = set(self.tokens)
seen_tokens = set()
for family in self.tokens.token_families:
with family.diagnostic_context:
not_tokens = family.tokens - all_tokens
check_source_language(
not not_tokens,
'Invalid tokens: {}'.format(format_token_list(not_tokens))
)
already_seen_tokens = seen_tokens & family.tokens
check_source_language(
not already_seen_tokens,
'Tokens must belong to one family exclusively: {}'
.format(format_token_list(already_seen_tokens))
)
seen_tokens.update(family.tokens)
# Create a token family to host all tokens that are not associated with
# a specific token family.
default_family = TokenFamily(*list(all_tokens - seen_tokens))
default_family.name = Name('Default_Family')
self.tokens.token_families.append(default_family)
# Make it easy to get the family a token belongs to
for tf in self.tokens.token_families:
for t in tf.tokens:
self.tokens.token_to_family[t] = tf
def add_tokens(self, klass):
for fld_name, fld_value in klass.__dict__.items():
if isinstance(fld_value, TokenAction):
fld_value.name = Name.from_camel(fld_name)
self.fields.append(fld_value)
