hackc: subroutine (partial)

4 files changed, 248 insertions, 147 deletions

diff --git a/projects/hackc/classes.py b/projects/hackc/classes.py
new file mode 100644
index 0000000..1875ce7
--- /dev/null
+++ b/projects/hackc/classes.py
@@ -0,0 +1,245 @@
+from .tokens import Token
+from .utils import *
+
+SCOPES = ["static", "field", "var"]
+VAR_TYPES = ["int", "char", "boolean"]
+RETURN_TYPES = ["int", "char", "boolean", "void"]
+SUBROUTINE_CATS = ["constructor", "method", "function"]
+
+
+class Class:
+    def __init__(self, name: Token, variables: list, subroutines: list):
+        self.name = name
+        self.variables = variables
+        self.subroutines = subroutines
+
+    @classmethod
+    def from_tokens(cls, tokens: list):
+        """Construct a class from a list of tokens.
+
+        In standard Jack, one file is exactly one class.
+
+        Format:
+        'class' <name> '{'
+            <variable>*
+            <subroutine>*
+        '}'
+        """
+        tokens_total = len(tokens)
+        if tokens_total < 4:
+            return None
+        if tokens[0] != "class":
+            raise JackSyntaxError(
+                f"Expected `class`, got `{tokens[0]}` instead", tokens[0]
+            )
+
+        name = tokens[1]
+        if name.type != "identifier":
+            raise JackSyntaxError(f"You cannot name a class `{name}`", name)
+
+        if tokens[2] != LEFT_BRACE:
+            raise JackSyntaxError(
+                f"Expected `{LEFT_BRACE}`, got `{tokens[2]}` instead", tokens[2]
+            )
+
+        tokens_consumed = 3
+
+        while tokens_consumed < tokens_total:
+            variables, token_cnt = Variable.from_tokens(
+                tokens[tokens_consumed:], context="class"
+            )
+            if variables is None:
+                break
+            variables.print_verbose()
+            tokens_consumed += token_cnt
+
+        while tokens_consumed < tokens_total:
+            subroutine, token_cnt = Subroutine.from_tokens(tokens[tokens_consumed:])
+            if subroutine is None:
+                break
+            subroutine.print_verbose()
+            tokens_consumed += token_cnt
+
+        return Class(name, variables, [])
+
+
+class Variable:
+    def __init__(self, scope: Token, type: Token, names: list[Token]):
+        self.scope = scope
+        self.type = type
+        self.names = names
+
+    @classmethod
+    def from_tokens(cls, tokens: list, context: str) -> tuple:
+        """Construct variable declaration statement from a list of tokens.
+        Return a tuple of an instance of Variable and number of tokens consumed.
+        When `tokens` does not begin with a variable declaration, return (None, 0).
+
+        context -- "class" (<scope> = static | field) or "subroutine" (<scope> = var)
+
+        Format:
+        <scope> <type> <name> (, <name>)* ;
+
+        <scope> = static | field | var
+        <type> = int | char | boolean | <class>
+        """
+        if len(tokens) < 4 or tokens[0] not in SCOPES:
+            # not variable declaration
+            return (None, 0)
+
+        scope = tokens[0]
+        if scope in ["static", "field"] and context != "class":
+            raise JackSyntaxError(
+                f"You cannot declare a {scope} variable in a subroutine", scope
+            )
+        if scope == "var" and context != "subroutine":
+            raise JackSyntaxError(
+                f"You cannot declare a local variable outside of a subroutine",
+                scope,
+            )
+
+        type = tokens[1]
+        if type not in VAR_TYPES and type.type != "identifier":
+            raise JackSyntaxError(f"Expected datatype, got `{tokens[1]}` instead", type)
+
+        tokens_consumed = 2
+        names = []  # names of variables
+        expecting_identifier = True
+        found_semicolon = False
+
+        for token in tokens[2:]:
+            tokens_consumed += 1
+            if token.type == "identifier":
+                if expecting_identifier:
+                    names.append(token)
+                    expecting_identifier = False
+                else:
+                    raise JackSyntaxError(f"Expected `,`, got `{token}` instead", token)
+            elif token == ",":
+                if not expecting_identifier:
+                    expecting_identifier = True
+                else:
+                    raise JackSyntaxError(
+                        f"Expected variable name, got `,` instead", token
+                    )
+            elif token == ";":
+                if expecting_identifier:
+                    raise JackSyntaxError(
+                        f"Expected variable name, got `;` instead", token
+                    )
+                found_semicolon = True
+                break
+            else:
+                expected = "variable name" if expecting_identifier else "`,` or `;`"
+                raise JackSyntaxError(
+                    f"Expected {expected}, got `{token}` instead", token
+                )
+
+        if not found_semicolon:
+            # TODO: print caret at end of token
+            raise JackSyntaxError(f"Missing semicolon", token)
+
+        return (Variable(scope, type, names), tokens_consumed)
+
+    def print_verbose(self):
+        print(f"Declare {len(self.names)} variable(s):")
+        for name in self.names:
+            print(self.scope, self.type, name)
+
+
+class ParamList:
+    def __init__(self, params: list[tuple]):
+        self.params = params
+
+    @classmethod
+    def from_tokens(cls, tokens: list) -> tuple:
+        """Construct parameter list of subroutine from tokens.
+
+        Format:
+        '(' (<type> <name> (, <type> <name>)*)? ')'
+
+        <type> = int | char | boolean | <class>
+        """
+        if len(tokens) < 2 or tokens[0] != LEFT_PAREN:
+            return (None, 0)
+
+        if tokens[1] == RIGHT_PAREN:
+            # empty param list, i.e. '(' ')'
+            return (ParamList([]), 2)
+
+        tokens_consumed = 1
+        params = []
+        for type, name, delim in zip(tokens[1::3], tokens[2::3], tokens[3::3]):
+            tokens_consumed += 3
+            if type not in VAR_TYPES and type.type != "identifier":
+                raise JackSyntaxError(f"Expected datatype, got `{type}` instead", type)
+            if not name:
+                # TODO: print caret at end of type
+                raise JackSyntaxError("Expected variable name", type)
+            if name.type != "identifier":
+                raise JackSyntaxError(
+                    f"Expected variable name, got `{name}` instead", name
+                )
+            if not delim:
+                raise JackSyntaxError(f"Expected `,` or `{RIGHT_PAREN}`", name)
+            if delim == ",":
+                params.append((type, name))
+                continue
+            elif delim == RIGHT_PAREN:
+                params.append((type, name))
+                break
+            else:
+                raise JackSyntaxError(
+                    f"Expected `,` or `{RIGHT_PAREN}`, got `{delim}` instead", delim
+                )
+
+        return (ParamList(params), tokens_consumed)
+
+
+class Subroutine:
+    def __init__(
+        self, category: Token, type: Token, name: Token, params: ParamList, body
+    ):
+        self.category = category
+        self.type = type
+        self.name = name
+        self.params = params
+        self.body = body
+
+    @classmethod
+    def from_tokens(cls, tokens: list) -> tuple:
+        """Construct subroutine from tokens.
+
+        Format:
+        <category> <return type> <name> <paramlist> <body>
+
+        <category> = constructor | method | function
+        <return type> = int | char | boolean | void | <class>
+        """
+        if len(tokens) < 7 or tokens[0] not in SUBROUTINE_CATS:
+            # not a subroutine
+            return (None, 0)
+        category = tokens[0]
+
+        return_type = tokens[1]
+        if return_type not in RETURN_TYPES and return_type.type != "identifier":
+            raise JackSyntaxError(
+                f"Expected datatype, got `{return_type}` instead", return_type
+            )
+
+        name = tokens[2]
+        if name.type != "identifier":
+            raise JackSyntaxError(
+                f"Expected {category} name, got `{name}` instead", name
+            )
+
+        tokens_consumed = 3
+        params, token_cnt = ParamList.from_tokens(tokens[tokens_consumed:])
+        if params is None:
+            raise JackSyntaxError("Expected parameter list", tokens[tokens_consumed])
+        tokens_consumed += token_cnt
+
+        return (Subroutine(category, return_type, name, params, None), tokens_consumed)
+
+    def print_verbose(self):
+        print(f"Define {self.category} {self.name}, returns {self.type}")
diff --git a/projects/hackc/parser.py b/projects/hackc/parser.py
index 2c34d1b..2fc30ee 100644
--- a/projects/hackc/parser.py
+++ b/projects/hackc/parser.py
@@ -1,5 +1,5 @@
 from .tokens import Token
-from .syntax import Class
+from .classes import Class
 from .utils import *
 
 KEYWORDS = [
diff --git a/projects/hackc/syntax.py b/projects/hackc/syntax.py
deleted file mode 100644
index 281bd17..0000000
--- a/projects/hackc/syntax.py
+++ /dev/null
@@ -1,146 +0,0 @@
-from .tokens import Token
-from .utils import *
-
-SCOPES = ["static", "field", "var"]
-PRIMITIVE_TYPES = ["int", "char", "boolean"]
-
-
-class Class:
-    def __init__(self, name: Token, variables: list, subroutines: list):
-        self.name = name
-        self.variables = variables
-        self.subroutines = subroutines
-
-    @classmethod
-    def from_tokens(cls, tokens: list):
-        """Construct a class from a list of tokens.
-
-        In standard Jack, one file is exactly one class.
-
-        Format:
-        class <name> {
-            <zero or more class variable declarations>
-            <zero or more subroutines>
-        }
-        """
-        if len(tokens) < 4:
-            return None
-        if tokens[0] != "class":
-            raise JackSyntaxError(
-                f"Expected `class`, got `{tokens[0]}` instead", tokens[0]
-            )
-
-        name = tokens[1]
-        if name.type != "identifier":
-            raise JackSyntaxError(f"You cannot name a class `{name}`", name)
-
-        if tokens[2] != LEFT_BRACE:
-            raise JackSyntaxError(
-                f"Expected `{LEFT_BRACE}`, got `{tokens[2]}` instead", tokens[2]
-            )
-
-        tokens_consumed = 3
-
-        while True:
-            variables, token_cnt = Variable.from_tokens(
-                tokens[tokens_consumed:], context="class"
-            )
-            if variables is None:
-                break
-            variables.print_verbose()
-            tokens_consumed += token_cnt
-
-        return Class(name, variables, [])
-
-
-class Variable:
-    def __init__(self, scope: Token, type: Token, names: list[Token]):
-        self.scope = scope
-        self.type = type
-        self.names = names
-
-    @classmethod
-    def from_tokens(cls, tokens: list, context: str) -> tuple:
-        """Construct variable declaration statement from a list of tokens.
-        Return a tuple of an instance of Variable and number of tokens consumed.
-        When `tokens` does not begin with a variable declaration, return (None, 0).
-
-        context -- "class" (<scope> = static | field) or "subroutine" (<scope> = var)
-
-        Format:
-        <scope> <type> <one or more names, joined with a comma>;
-
-        <scope> = static | field | var
-        <type> = int | char | boolean | <class name>
-        """
-        if len(tokens) < 4 or tokens[0] not in SCOPES:
-            # not variable declaration
-            return (None, 0)
-
-        scope = tokens[0]
-        if scope in ["static", "field"] and context != "class":
-            raise JackSyntaxError(
-                f"You cannot declare a {scope} variable in a subroutine", scope
-            )
-        if scope == "var" and context != "subroutine":
-            raise JackSyntaxError(
-                f"You cannot declare a local variable outside of a subroutine",
-                scope,
-            )
-
-        type = tokens[1]
-        if type not in PRIMITIVE_TYPES and type.type != "identifier":
-            raise JackSyntaxError(f"Expected datatype, got `{tokens[1]}` instead", type)
-
-        tokens_consumed = 2
-        names = []  # names of variables
-        expecting_identifier = True
-
-        for token in tokens[2:]:
-            tokens_consumed += 1
-            if token.type == "identifier":
-                if expecting_identifier:
-                    names.append(token)
-                    expecting_identifier = False
-                else:
-                    raise JackSyntaxError(f"Expected `,`, got `{token}` instead", token)
-            elif token == ",":
-                if not expecting_identifier:
-                    expecting_identifier = True
-                else:
-                    raise JackSyntaxError(
-                        f"Expected variable name, got `,` instead", token
-                    )
-            elif token == ";":
-                if expecting_identifier:
-                    raise JackSyntaxError(
-                        f"Expected variable name, got `;` instead", token
-                    )
-                break
-            else:
-                expected = "variable name" if expecting_identifier else "`,` or `;`"
-                raise JackSyntaxError(
-                    f"Expected {expected}, got `{token}` instead", token
-                )
-
-        return (Variable(scope, type, names), tokens_consumed)
-
-    def print_verbose(self):
-        print(f"Declare {len(self.names)} variable(s):")
-        for name in self.names:
-            print(self.scope, self.type, name)
-
-
-class Subroutine:
-    def __init__(self):
-        pass
-
-
-class Statement:
-    def __init__(self):
-        super().__init__()
-
-
-class IfStatement(Statement):
-    def __init__(self):
-        super().__init__()
diff --git a/projects/hackc/utils.py b/projects/hackc/utils.py
index 4861088..4b9c9f0 100644
--- a/projects/hackc/utils.py
+++ b/projects/hackc/utils.py
@@ -7,6 +7,8 @@ EXIT_CODE_SYNTAX_ERROR = 4
 # vim autoindent misbehaves if I type these verbatim in strings
 LEFT_BRACE = "{"
 RIGHT_BRACE = "}"
+LEFT_PAREN = "("
+RIGHT_PAREN = ")"
 
 class JackSyntaxError(Exception):
     def __init__(self, msg, token):