Mercurial > lcfOS
view python/c3/parser.py @ 219:1fa3e0050b49
Expanded ad hoc code generator
author | Windel Bouwman |
---|---|
date | Sat, 06 Jul 2013 12:38:09 +0200 |
parents | c1ccb1cb4cef |
children | 3f6c30a5d234 |
line wrap: on
line source
from . import astnodes, lexer from ppci import CompilerError # binop precedence for expressions: binopPrecs = {'or': 5, 'and': 10, \ '<': 20, '>': 20, '==': 20, '<=': 20, '>=': 20, '!=': 20, \ '+': 30, '-': 30, '*': 40, '/': 40 } class Parser: """ Parses sourcecode into an abstract syntax tree (AST) """ def __init__(self, diag): self.diag = diag def parseSource(self, source): self.initLex(source) try: self.parsePackage() return self.mod except CompilerError as e: self.diag.addDiag(e) def Error(self, msg): raise CompilerError(msg, self.token.loc) # Lexer helpers: def Consume(self, typ): if self.Peak == typ: return self.NextToken() else: self.Error('Excected: "{0}", got "{1}"'.format(typ, self.Peak)) @property def Peak(self): return self.token.typ @property def PeakPrec(self): if self.Peak in binopPrecs: return binopPrecs[self.Peak] return -1 def hasConsumed(self, typ): if self.Peak == typ: self.Consume(typ) return True return False def NextToken(self): t = self.token if t.typ != 'END': self.token = self.tokens.__next__() return t def initLex(self, source): self.tokens = lexer.tokenize(source) # Lexical stage self.token = self.tokens.__next__() def addDeclaration(self, decl): self.currentPart.declarations.append(decl) def parseUses(self): pass def parsePackage(self): self.Consume('package') name = self.Consume('ID') self.Consume(';') self.mod = astnodes.Package(name.val, name.loc) self.currentPart = self.mod self.parseUses() # TODO: parse uses while self.Peak != 'END': self.parseTopLevel() self.Consume('END') def parseTopLevel(self): if self.Peak == 'function': self.parseFunctionDef() elif self.Peak == 'var': self.parseVarDef() elif self.Peak == 'const': self.parseConstDef() elif self.Peak == 'type': self.parseTypeDef() else: self.Error('Expected function, var, const or type') def parseDesignator(self): """ A designator designates an object """ name = self.Consume('ID') d = astnodes.Designator(name.val, name.loc) return d # Type system def parseTypeSpec(self): # For now, do simple type spec, just parse an ID: #return self.parseDesignator() if self.Peak == 'struct': self.Consume('struct') self.Consume('{') mems = [] while self.Peak != '}': mem_t = self.parseTypeSpec() mem_n = self.Consume('ID') mems.append((mem_t, mem_n)) while self.hasConsumed(','): mem_n = self.Consume('ID') mems.append((mem_t, mem_n)) self.Consume(';') self.Consume('}') theT = astnodes.StructureType(mems) else: theT = self.parseDesignator() # Check for pointer suffix: while self.hasConsumed('*'): theT = astnodes.PointerType(theT) return theT def parseTypeDef(self): self.Consume('type') newtype = self.parseTypeSpec() typename = self.Consume('ID') # TODO: action here :) self.Consume(';') return astnodes.DefinedType(typename, newtype) # Variable declarations: def parseVarDef(self): self.Consume('var') t = self.parseTypeSpec() def parseVar(): name = self.Consume('ID') v = astnodes.Variable(name.val, t) v.loc = name.loc if self.hasConsumed('='): v.ival = self.parseExpression() self.addDeclaration(v) parseVar() while self.hasConsumed(','): parseVar() self.Consume(';') def parseConstDef(self): self.Consume('const') t = self.parseTypeSpec() def parseConst(): name = self.Consume('ID') self.Consume('=') val = self.parseExpression() c = astnodes.Constant(name.val, t, val) c.loc = name.loc parseConst() while self.hasConsumed(','): parseConst() self.Consume(';') # Procedures def parseFunctionDef(self): loc = self.Consume('function').loc returntype = self.parseTypeSpec() fname = self.Consume('ID').val f = astnodes.Function(fname, loc) self.addDeclaration(f) savePart = self.currentPart self.currentPart = f self.Consume('(') parameters = [] if not self.hasConsumed(')'): def parseParameter(): typ = self.parseTypeSpec() name = self.Consume('ID') param = astnodes.Variable(name.val, typ) param.loc = name.loc self.addDeclaration(param) parameters.append(param) parseParameter() while self.hasConsumed(','): parseParameter() self.Consume(')') paramtypes = [p.typ for p in parameters] f.typ = astnodes.FunctionType(paramtypes, returntype) f.body = self.parseCompoundStatement() self.currentPart = savePart # Statements: def parseAssignment(self, lval): lval = astnodes.VariableUse(lval, lval.loc) loc = self.Consume('=').loc rval = self.parseExpression() self.Consume(';') return astnodes.Assignment(lval, rval, loc) def parseCall(self, func): self.Consume('(') args = [] if not self.hasConsumed(')'): args.append(self.parseExpression()) while self.hasConsumed(','): args.append(self.parseExpression()) self.Consume(')') return astnodes.FunctionCall(func, args, func.loc) def parseIfStatement(self): loc = self.Consume('if').loc self.Consume('(') condition = self.parseExpression() self.Consume(')') yes = self.parseCompoundStatement() if self.hasConsumed('else'): no = self.parseCompoundStatement() else: no = astnodes.EmptyStatement() return astnodes.IfStatement(condition, yes, no, loc) def parseWhileStatement(self): loc = self.Consume('while').loc self.Consume('(') condition = self.parseExpression() self.Consume(')') statements = self.parseCompoundStatement() return astnodes.WhileStatement(condition, statements, loc) def parseReturnStatement(self): self.Consume('return') expr = self.parseExpression() self.Consume(';') return astnodes.ReturnStatement(expr) def parseCompoundStatement(self): self.Consume('{') statements = [] while not self.hasConsumed('}'): s = self.parseStatement() if not type(s) is astnodes.EmptyStatement: statements.append(s) return astnodes.CompoundStatement(statements) def parseStatement(self): # Determine statement type based on the pending token: if self.Peak == 'if': return self.parseIfStatement() elif self.Peak == 'while': return self.parseWhileStatement() elif self.Peak == '{': return self.parseCompoundStatement() elif self.hasConsumed(';'): return astnodes.EmptyStatement() elif self.Peak == 'var': self.parseVarDef() return astnodes.EmptyStatement() elif self.Peak == 'return': return self.parseReturnStatement() else: designator = self.parseDesignator() if self.Peak == '(': return self.parseCall(designator) elif self.Peak == '=': return self.parseAssignment(designator) else: self.Error('Unable to determine statement') # Parsing expressions: def parseExpression(self): return self.parseBinopRhs(self.parsePrimary(), 0) def parsePrimary(self): if self.hasConsumed('('): e = self.parseExpression() self.Consume(')') return e elif self.Peak == 'NUMBER': val = self.Consume('NUMBER') return astnodes.Literal(val.val, val.loc) elif self.Peak == 'REAL': val = self.Consume('REAL') return astnodes.Literal(val.val, val.loc) elif self.Peak == 'true': val = self.Consume('true') return astnodes.Literal(True, val.loc) elif self.Peak == 'false': val = self.Consume('false') return astnodes.Literal(False, val.loc) elif self.Peak == 'ID': d = self.parseDesignator() if self.Peak == '(': return self.parseCall(d) else: return astnodes.VariableUse(d, d.loc) self.Error('Expected NUM, ID or (expr), got {0}'.format(self.Peak)) def parseBinopRhs(self, lhs, min_prec): while self.PeakPrec >= min_prec: op_prec = self.PeakPrec op = self.Consume(self.Peak) rhs = self.parsePrimary() while self.PeakPrec > op_prec: rhs = self.parseBinopRhs(rhs, self.PeakPrec) lhs = astnodes.Binop(lhs, op.typ, rhs, op.loc) return lhs