Improve lisp compiler and interpreter.

i3toolwait

@@ -2,6 +2,7 @@
 # -*- coding: utf-8 -*-
 
 
+import string
 import typing
 import asyncio
 import signal
@@ -21,317 +22,391 @@ try:
 except ImportError:
     from yaml import SafeLoader
 
-class Expression:
-    def __init__(self):
-        pass
-    def reduce(self, ipc_data):
-        if self.should_call:
-            return self.call(ipc_data)
-        return functools.reduce(self.reduce_function(ipc_data), self.children)
-    @property
-    def should_call(self):
-        return False
-    @property
-    def children(self):
-        raise NotImplemented('TODO: implement in subclass')
-    def reduce_function(self, ipc_data):
-        raise NotImplemented('TODO: implement in subclass')
-    def call(self, ipc_data):
-        raise NotImplementedError('TODO: implement in subclass')
 
-class LiteralExpression(Expression):
+def lazy_fc_if(env, a, b, c):
-    def __init__(self, value):
+    a.reduce(env)
-        self._value = value
+    if a.reduced:
-    def __repr__(self) -> str:
+        b.reduce(env)
-        return f'"{self._value}"'
+        return b
-    @property
+    c.reduce(env)
-    def children(self):
-        return [self._value]
-    def reduce_function(self, ipc_data):
-        def reduce(a, b):
-            raise NotImplemented('I should never be called')
 
-class IntLiteralExpression(LiteralExpression):
+def lazy_fc_nif(env, a, b, c):
-    def __repr__(self) -> str:
+    a.reduce(env)
-        return str(self._value)
+    if not a.reduced:
+        b.reduce(env)
+        return b
+    c.reduce(env)
 
-class BoolLiteralExpression(LiteralExpression):
+def fc_load(env, path):
-    def __repr__(self) -> str:
+    ipc_value = env.input
-        return str(self._value)
+    for k in path.strip('.').split('.'):
+        ipc_value = ipc_value[k]
+    return ipc_value
 
-class AndExpression(Expression):
+def fc_has_key(env, path):
-    def __init__(self, children, *args, **kwargs):
+    ipc_value = env.input
-        self._children = children
+    for k in path.strip('.').split('.'):
-        super().__init__(*args, **kwargs)
+        try:
-    def __repr__(self) -> str:
-        cs = ' '.join([repr(c) for c in self.children])
-        return f'(& {cs})'
-    @property
-    def children(self):
-        return self._children
-    def reduce_function(self, ipc_data):
-        return lambda a, b: a.reduce(ipc_data) and b.reduce(ipc_data)
-
-class OrExpression(Expression):
-    def __init__(self, children, *args, **kwargs):
-        self._children = children
-        super().__init__(*args, **kwargs)
-    def __repr__(self) -> str:
-        cs = ' '.join([repr(c) for c in self.children])
-        return f'(| {cs})'
-    @property
-    def children(self):
-        return self._children
-    def reduce_function(self, ipc_data):
-        return lambda a, b: a.reduce(ipc_data) or b.reduce(ipc_data)
-
-class IfExpression(Expression):
-    def __init__(self, children, *args, **kwargs):
-        self._children = children
-        super().__init__(*args, **kwargs)
-    def __repr__(self) -> str:
-        cs = ' '.join([repr(c) for c in self.children])
-        return f'(? {cs})'
-    @property
-    def should_call(self):
-        return True
-    @property
-    def children(self):
-        return self._children
-    def call(self, ipc_data):
-        if self._children[0].reduce(ipc_data):
-            i = 1
-        else:
-            i = 2
-        return self._children[i].reduce(ipc_data)
-
-class EqExpression(Expression):
-    def __init__(self, children, *args, **kwargs):
-        self._children = children
-        super().__init__(*args, **kwargs)
-    def __repr__(self) -> str:
-        cs = ' '.join([repr(c) for c in self.children])
-        return f'(= {cs})'
-    @property
-    def children(self):
-        return self._children
-    def reduce_function(self, ipc_data):
-        def reduce(v0, v1):
-            print(f'reducing: {repr(self)}')
-            return v0.reduce(ipc_data) == v1.reduce(ipc_data)
-        return reduce
-
-class NeqExpression(Expression):
-    def __init__(self, children, *args, **kwargs):
-        self._children = children
-        super().__init__(*args, **kwargs)
-    def __repr__(self) -> str:
-        cs = ' '.join([repr(c) for c in self.children])
-        return f'(!= {cs})'
-    @property
-    def children(self):
-        return self._children
-    def reduce_function(self, ipc_data):
-        def reduce(v0, v1):
-            return v0.reduce(ipc_data) != v1.reduce(ipc_data)
-        return reduce
-
-class GtExpression(Expression):
-    def __init__(self, children, *args, **kwargs):
-        self._children = children
-        super().__init__(*args, **kwargs)
-    def __repr__(self) -> str:
-        cs = ' '.join([repr(c) for c in self.children])
-        return f'(> {cs})'
-    @property
-    def children(self):
-        return self._children
-    def reduce_function(self, ipc_data):
-        def reduce(v0, v1):
-            return v0.reduce(ipc_data) > v1.reduce(ipc_data)
-        return reduce
-
-class LtExpression(Expression):
-    def __init__(self, children, *args, **kwargs):
-        self._children = children
-        super().__init__(*args, **kwargs)
-    def __repr__(self) -> str:
-        cs = ' '.join([repr(c) for c in self.children])
-        return f'(< {cs})'
-    @property
-    def children(self):
-        return self._children
-    def reduce_function(self, ipc_data):
-        def reduce(v0, v1):
-            return v0.reduce(ipc_data) < v1.reduce(ipc_data)
-        return reduce
-
-class LoadExpression(Expression):
-    def __init__(self, value, *args, **kwargs):
-        self._value = value
-        super().__init__(*args, **kwargs)
-    def __repr__(self) -> str:
-        return f'(load {self._value})'
-    @property
-    def should_call(self):
-        return True
-    @property
-    def children(self):
-        return [self._value]
-    def call(self, ipc_data):
-        ipc_value = ipc_data
-        for k in self._value[0].children[0].strip('.').split('.'):
             ipc_value = ipc_value[k]
-        return ipc_value
+        except KeyError:
+            return False
+    return True
+
+class Environment:
+
+    def __init__(self, input):
+        self._input = input
+        self._variables = {}
+        self._functions = {
+            '__last__': lambda *a: a[-1], # special function, if multiple expressions, execute all and return result of last one
+            'defvar': lambda env, n, v: env.set_variable(n, v),
+            'write': lambda _, a: print(a),
+            'load': fc_load,
+            'has-key': fc_has_key,
+            '=': lambda _, a, b: a == b,
+            '!=': lambda _, a, b: a != b,
+            '>': lambda _, a, b: a > b,
+            '<': lambda _, a, b: a < b,
+            '>=': lambda _, a, b: a >= b,
+            '<=': lambda _, a, b: a <= b,
+            '+': lambda _, *a: sum(a),
+            '-': lambda _, a, b: a - b,
+            '*': lambda _, *a: functools.reduce(lambda a, b: a * b, a),
+            '/': lambda _, a, b: a // b,
+        }
+        self._lazy_functions = {
+            '?': lazy_fc_if,
+            '!?': lazy_fc_nif,
+        }
+
+    @property
+    def input(self):
+        return self._input
+
+    def set_variable(self, name: str, value: object):
+        self._variables[name] = value
+
+    def get_variable(self, name: str):
+        return self._variables[name]
+
+    def get_function(self, name: str):
+        return self._functions[name]
+
+    def get_lazy_function(self, name: str):
+        return self._lazy_functions[name]
+
+class Expression:
+
+    STATE_CONSTRUCTED = 0
+    STATE_REDUCED = 1
+
+    def __init__(self):
+        self._state = Expression.STATE_CONSTRUCTED 
+        self._reduced = None
+
+    def _reduce(self, env: Environment, args: list[object]):
+        _ = env, args
+        raise NotImplementedError('Implement in subclass')
+
+    def reduce(self, env: Environment):
+        if self._state == Expression.STATE_REDUCED:
+            return
+        self._reduced = self._reduce(env, [])
+        self._state = Expression.STATE_REDUCED
     
-class HasKeyExpression(Expression):
+    @property
-    def __init__(self, value, *args, **kwargs):
+    def reduced(self) -> object:
+        if self._state != Expression.STATE_REDUCED:
+            raise RuntimeError('Tried to get the reduced value before reducing')
+        return self._reduced
+
+class Constant(Expression):
+
+    def __init__(self, value):
+        super().__init__()
         self._value = value
-        super().__init__(*args, **kwargs)
-    def __repr__(self) -> str:
-        return f'(has-key {self._value})'
-    @property
-    def should_call(self):
-        return True
-    @property
-    def children(self):
-        return [self._value]
-    def call(self, ipc_data):
-        ipc_value = ipc_data
-        for k in self._value[0].children[0].strip('.').split('.'):
-            try:
-                ipc_value = ipc_value[k]
-            except KeyError:
-                return False
-        return True
 
+    def __repr__(self):
+        return repr(self._value)
 
-expression_mapping = {
+    def _reduce(self, env: Environment, args: list[Expression]):
-    '&': AndExpression,
+        _ = env, args
-    '|': OrExpression,
+        return self._value
-    '?': IfExpression,
-    '=': EqExpression,
-    '!=': NeqExpression,
-    '>': GtExpression,
-    '<': LtExpression,
-    'load': LoadExpression,
-    'has-key': HasKeyExpression,
-}
 
-def group_tokens(tokens: list[str]) -> list[list[str]]:
+class VariableSet(Constant):
-    groups = []
+    pass
-    current_group = []
-    
-    brace_count = 0
-    for token in tokens:
-        if token == '(':
-            brace_count += 1
-        elif token == ')':
-            brace_count -= 1
-            if brace_count == 0:
-                groups += [current_group]
-                current_group = []
-        elif brace_count == 0:
-            groups += [[token]]
-        else:
-            current_group += [token]
-    return groups
 
-def build_expression(tokens: list[str]) -> Expression:
+class VariableGet(Constant):
-    if tokens[0] == '(' and tokens[-1] == ')':
-        tokens = tokens[1:-1]
-    token_groups = group_tokens(tokens)
 
-    expressions = [build_expression(ts) for ts in token_groups[1:]] if len(token_groups) > 1 else []
+    def _reduce(self, env: Environment, args: list[Expression]):
-    root_expression = None
+        _ = args
-    token = token_groups[0][0]
+        return env.get_variable(self._value)
-    if token in expression_mapping:
-        root_expression = expression_mapping[token](expressions)
-    elif token.startswith('"'):
-        root_expression = LiteralExpression(token[1:-1])
-    elif token.isnumeric():
-        root_expression = IntLiteralExpression(int(token))
-    elif token in ('True', 'False'):
-        root_expression = BoolLiteralExpression(token == 'True')
-    assert isinstance(root_expression, Expression)
-    return root_expression
-    
-def take_space(s: str) -> tuple:
-    if s[0] in ' \n':
-        return None, s[1:]
-    return None, s
 
-def take_operator(s: str) -> tuple:
+class Function(Expression):
-    token = ''
+
-    for c in s:
+    def __init__(self, fc, args: list[Expression]):
-        if c in ''.join(set(expression_mapping.keys())): 
+        super().__init__()
-            token += c
+        self._fc = fc
-        else:
+        self._args = args
-            break
+
-    if token == '':
+    def __repr__(self):
-        return None, s
+        return f'({self._fc} {self._args})'
+
+    def _reduce(self, env: Environment, args: list[Expression]):
+        try:
+            fc = env.get_function(self._fc)
+            [a.reduce(env) for a in args]
+            r = fc(env, *[a.reduced for a in args])
+        except KeyError as e:
+            fc = env.get_lazy_function(self._fc)
+            r = fc(env, *args)
+        return r
+
+    def reduce(self, env: Environment):
+        if self._state == Expression.STATE_REDUCED:
+            return
+        self._reduced = self._reduce(env, self._args)
+        self._state = Expression.STATE_REDUCED
+
+class Token:
+
+    CONSTANT_STRING = 0
+    CONSTANT_INTEGER = 10
+    CONSTANT_BOOLEAN = 20
+    KEYWORD = 30
+    VARIABLE_SET = 40
+    VARIABLE_GET = 50
+    FUNCTION = 60
+    GROUPING_OPEN = 70
+    GROUPING_CLOSE = 80
+    WHITESPACE = 90
+
+    def __init__(self, t, v):
+        self.t = t
+        self.v = v
+
+    def __repr__(self):
+        return f'{self.v}::{self.t}'
+
+    def to_expression(self):
+        if self.t == Token.CONSTANT_STRING:
+            return Constant(self.v[1:-1]) # slice away the quotes
+        if self.t == Token.CONSTANT_INTEGER:
+            return Constant(int(self.v, base=0))
+        if self.t == Token.CONSTANT_BOOLEAN:
+            return Constant(self.v == 'True')
+        if self.t == Token.KEYWORD:
+            raise RuntimeError(f'This is a meta token type and should be swallowed by the sanitizer: {self}')
+        if self.t == Token.VARIABLE_GET:
+            return VariableGet(self.v)
+        if self.t == Token.VARIABLE_SET:
+            return VariableSet(self.v)
+        if self.t == Token.FUNCTION:
+            raise RuntimeError('Cant construct function just from its token')
+        if self.t == Token.GROUPING_OPEN or self.t == Token.GROUPING_CLOSE:
+            raise RuntimeError('Groupings should never be constructed, this is a bug')
+        if self.t == Token.WHITESPACE:
+            raise RuntimeError('Whitespaces should not be present in this stage of the build')
+        raise RuntimeError(f'The token type {self.t} is not implemented')
+
+def token_extract_string(stream: str) -> tuple[Token, str]:
+    if stream[0] != '"':
+        raise ValueError('No such token in stream')
+    i = stream.find('"', 1)
+    return Token(Token.CONSTANT_STRING, stream[:i+1]), stream[i+1:]
+
+def token_extract_integer(stream: str) -> tuple[Token, str]:
+    i = 0
+    base = None
+    if stream[i] in '+-':
+        i += 1
+    if stream[i] in '0123456789':
+        i += 1
     else:
-        return token, s[len(token):]
+        raise ValueError('Malformed integer')
 
-def take_brace(s: str) -> tuple:
+    if stream[i] in 'xbo':
-    if s[0] in '()':
+        base = stream[i]
-        return s[0], s[1:]
+        i += 1
+    int_set = {None: '0123456789', 'x': '0123456789abcdefABCDEF', 'b': '01', 'o': '01234567'}[base]
+    while stream[i] in int_set:
+        i += 1
+    return Token(Token.CONSTANT_INTEGER, stream[:i]), stream[i:]
+
+def token_extract_boolean(stream: str) -> tuple[Token, str]:
+    if stream.startswith('True'):
+        return Token(Token.CONSTANT_BOOLEAN, stream[:4]), stream[4:]
+    elif stream.startswith('False'):
+        return Token(Token.CONSTANT_BOOLEAN, stream[:5]), stream[5:]
+    raise ValueError('No such token in stream')
+
+def token_extract_keyword(stream: str) -> tuple[Token, str]:
+    i = 0
+    if stream[i] in string.ascii_letters + '_-><=!+-*/?':
+        i += 1
     else:
-        return None, s
+        raise ValueError('No keyword in stream')
+    while stream[i] in string.ascii_letters + '_-><=!+-*/?':
+        i += 1
+    return Token(Token.KEYWORD, stream[:i]), stream[i:]
 
-def take_literal(s: str) -> tuple:
+def token_extract_grouping_open(stream: str) -> tuple[Token, str]:
-    token = '"'
+    if stream[0] == '(':
-    if s[0] != '"':
+        return Token(Token.GROUPING_OPEN, '('), stream[1:]
-        return None, s
+    raise ValueError('No such token in stream')
-    for c in s[1:]:
-        token += c
-        if c == '"':
-            break
-    if not token.endswith('"'):
-        raise ValueError('Missing closing quotes (`"`)')
-    return token, s[len(token):]
 
-def take_int_literal(s: str) -> tuple:
+def token_extract_grouping_close(stream: str) -> tuple[Token, str]:
-    token = ''
+    if stream[0] == ')':
-    for c in s:
+        return Token(Token.GROUPING_CLOSE, ')'), stream[1:]
-        if not c.isnumeric():
+    raise ValueError('No such token in stream')
-            break
-        token += c
-    if token == '':
-        return None, s
-    return token, s[len(token):]
 
-def take_bool_literal(s: str) -> tuple:
+def token_extract_space(stream: str) -> tuple[Token, str]:
-    if s.startswith('True'):
+    i = 0
-        return 'True', s[len('True'):]
+    while stream[i] in string.whitespace:
-    if s.startswith('False'):
+        i += 1
-        return 'False', s[len('False'):]
+    return Token(Token.WHITESPACE, stream[:i]), stream[i:]
-    return None, s
 
-def tokenize(s: str) -> list[str]:
+def tokenize(program: str) -> list[Token]:
-    operator_extractors = [
+    extractors = [
-        take_operator,
+        token_extract_boolean,
-        take_brace,
+        token_extract_integer,
-        take_literal,
+        token_extract_string,
-        take_int_literal,
+        token_extract_keyword,
-        take_bool_literal,
+        token_extract_grouping_open,
-        take_space,
+        token_extract_grouping_close,
+        token_extract_space,
     ]
+    p = program
     tokens = []
-    while s != '':
+    while p:
-        previous_len = len(s)
+        success = False
-        for operator_extractor in operator_extractors:
+        for e in extractors:
-            token, s = operator_extractor(s)
+            try:
-            if token is not None:
+                t, p = e(p)
-                tokens += [token]
+                tokens += [t]
+                success = True
                 break
-        if len(s) == previous_len:
+            except ValueError:
-            raise ValueError(f'Could not tokenize string {s}')
+                pass
+        if not success:
+            raise ValueError('Program is invalid')
+    return [t for t in tokens if t.t != Token.WHITESPACE]
+
+def tokenize_sanitize_function(token_before: Token | None, token: Token, token_after: Token | None) -> Token | None:
+    if token_before is None:
+        return
+    if token_before.t == Token.GROUPING_OPEN and token.t == Token.KEYWORD:
+        return Token(Token.FUNCTION, token.v)
+
+def tokenize_sanitize_setvar(token_before: Token | None, token: Token, token_after: Token | None) -> Token | None:
+    if token_before is None: 
+        return
+    if (token_before.t == Token.FUNCTION and token_before.v == 'defvar') and token.t == Token.KEYWORD:
+        return Token(Token.VARIABLE_SET, token.v)
+
+def tokenize_sanitize_getvar(token_before: Token | None, token: Token, token_after: Token | None) -> Token | None:
+    if token_before is None:
+        if token.t == Token.KEYWORD: 
+            return Token(Token.VARIABLE_GET, token.v)
+        return
+    if (token_before.t != Token.FUNCTION or token_before.v != 'defvar') and token.t == Token.KEYWORD:
+        return Token(Token.VARIABLE_GET, token.v)
+
+def _tokenize_sanitize(tokens: list[Token]) -> tuple[bool, list[Token]]:
+    sanitizers = [
+        tokenize_sanitize_function,
+        tokenize_sanitize_setvar,
+        tokenize_sanitize_getvar,
+    ]
+    new_tokens = []
+    changed = False
+    for i in range(len(tokens)):
+        for s in sanitizers:
+            p_token = new_tokens[i-1] if i > 0 else None
+            n_token = tokens[i+1] if i < (len(tokens)-1) else None
+            new_token = s(p_token, tokens[i], n_token)
+            if new_token is not None:
+                changed = True
+                new_tokens += [new_token]
+                break
+        else:
+            new_tokens += [tokens[i]]
+    return changed, new_tokens
+
+def tokenize_sanitize(tokens: list[Token]) -> list[Token]:
+    _, tokens = _tokenize_sanitize(tokens)
     return tokens
 
-def parse(s: str) -> Expression:
+def take_token_group(tokens: list[Token], n: int = 1) -> list[Token]:
-    tokens = tokenize(s)
+    i = 0
-    return build_expression(tokens)
+    start = i
+    group_count = 0
+    consider_groups = False
+    while n:
+        if tokens[i].t == Token.GROUPING_OPEN:
+            consider_groups = True
+            if group_count == 0:
+                start = i
+            group_count += 1
+        elif tokens[i].t == Token.GROUPING_CLOSE:
+            group_count -= 1
+            if group_count == 0:
+                consider_groups = False
+        else:
+            if not consider_groups:
+                start = i
+        if group_count == 0:
+            n -= 1
+        if group_count < 0:
+            raise ValueError('reached past end')
+        i += 1
+    return tokens[start:i]
+
+def unwrap_token_group(tokens: list[Token]) -> list[Token]:
+    if tokens[0].t != Token.GROUPING_OPEN:
+        return tokens
+
+    brace_count = 0
+    for i, t in enumerate(tokens):
+        brace_count += int(t.t == Token.GROUPING_OPEN)
+        brace_count -= int(t.t == Token.GROUPING_CLOSE)
+        if i == len(tokens) - 2:
+            if brace_count > 0:
+                tokens = tokens[1:-1]
+            break
+    return tokens
+
+def build(tokens: list[Token]) -> Expression:
+    tokens = unwrap_token_group(tokens)
+    token_groups: list[list[Token]] = []
+    i = 1
+    while True:
+        try:
+            token_groups += [take_token_group(tokens, n=i)]
+            i += 1
+        except IndexError:
+            break
+
+    # special function case
+    if len(token_groups[0]) == 1 and token_groups[0][0].t == Token.FUNCTION:
+        token_0 = token_groups[0][0]
+        args = [build(tg) for tg in token_groups[1:]]
+        return Function(token_0.v, args)
+
+    # combine to multiple statements
+    if len(token_groups) > 1:
+        return Function('__last__', [build(tg) for tg in token_groups])
+
+    # create a basic expression
+    if len(token_groups) == 1 and len(token_groups[0]) == 1:
+        return token_groups[0][0].to_expression()
+
+    raise RuntimeError(f'Did not handle token case in build function, token_groups: {token_groups}')
+
+def parse(program: str) -> Expression:
+    tokens = tokenize_sanitize(tokenize(program))
+    expression = build(tokens)
+    return expression
 
 class Filter(Expression):
 
@@ -453,7 +528,8 @@ def window_new(runtime_data: RuntimeData, *, debug):
             print(json.dumps(e.ipc_data))
         async with runtime_data.lock:
             for i, cfg in enumerate(runtime_data.programs):
-                if cfg.match.reduce(e.ipc_data):
+                cfg.match.reduce(Environment(e.ipc_data))
+                if cfg.match.reduced:
                     container_id = e.ipc_data['container']['id']
                     await ipc.command(f'for_window [con_id="{container_id}"] focus')
                     await ipc.command(f'move container to workspace {cfg.workspace}')