Mercurial > lcfOS
annotate python/ppci/c3/codegenerator.py @ 354:5477e499b039
Added some sort of string functionality
| author | Windel Bouwman |
|---|---|
| date | Thu, 13 Mar 2014 18:59:06 +0100 |
| parents | b8ad45b3a573 |
| children | 42343d189e14 |
| rev | line source |
|---|---|
| 255 | 1 import logging |
| 354 | 2 import struct |
| 301 | 3 from .. import ir |
| 307 | 4 from .. import irutils |
| 308 | 5 from . import astnodes as ast |
| 230 | 6 |
| 228 | 7 |
| 307 | 8 class SemanticError(Exception): |
| 308 | 9 """ Error thrown when a semantic issue is observed """ |
| 307 | 10 def __init__(self, msg, loc): |
| 308 | 11 super().__init__() |
| 307 | 12 self.msg = msg |
| 13 self.loc = loc | |
| 14 | |
| 15 | |
| 16 class CodeGenerator(irutils.Builder): | |
| 288 | 17 """ |
| 274 | 18 Generates intermediate (IR) code from a package. The entry function is |
| 19 'genModule'. The main task of this part is to rewrite complex control | |
| 20 structures, such as while and for loops into simple conditional | |
| 21 jump statements. Also complex conditional statements are simplified. | |
| 22 Such as 'and' and 'or' statements are rewritten in conditional jumps. | |
| 23 And structured datatypes are rewritten. | |
| 307 | 24 |
| 25 Type checking is done in one run with code generation. | |
| 274 | 26 """ |
| 307 | 27 def __init__(self, diag): |
| 261 | 28 self.logger = logging.getLogger('c3cgen') |
| 307 | 29 self.diag = diag |
| 261 | 30 |
| 217 | 31 def gencode(self, pkg): |
| 308 | 32 """ Generate code for a single module """ |
| 268 | 33 self.prepare() |
| 308 | 34 assert type(pkg) is ast.Package |
| 307 | 35 self.pkg = pkg |
| 36 self.intType = pkg.scope['int'] | |
| 37 self.boolType = pkg.scope['bool'] | |
| 334 | 38 self.logger.debug('Generating ir-code for {}'.format(pkg.name), extra={'c3_ast':pkg}) |
| 288 | 39 self.varMap = {} # Maps variables to storage locations. |
| 217 | 40 self.funcMap = {} |
| 268 | 41 self.m = ir.Module(pkg.name) |
| 307 | 42 try: |
| 43 for s in pkg.innerScope.Functions: | |
| 44 f = self.newFunction(s.name) | |
| 45 self.funcMap[s] = f | |
| 46 for v in pkg.innerScope.Variables: | |
| 47 self.varMap[v] = self.newTemp() | |
| 48 for s in pkg.innerScope.Functions: | |
| 308 | 49 self.gen_function(s) |
| 307 | 50 except SemanticError as e: |
| 51 self.error(e.msg, e.loc) | |
| 313 | 52 if self.pkg.ok: |
| 53 return self.m | |
| 308 | 54 |
| 55 def error(self, msg, loc=None): | |
| 56 self.pkg.ok = False | |
| 57 self.diag.error(msg, loc) | |
| 307 | 58 |
| 308 | 59 def gen_function(self, fn): |
| 268 | 60 # TODO: handle arguments |
| 61 f = self.funcMap[fn] | |
| 272 | 62 f.return_value = self.newTemp() |
| 268 | 63 self.setFunction(f) |
| 269 | 64 l2 = self.newBlock() |
| 65 self.emit(ir.Jump(l2)) | |
| 66 self.setBlock(l2) | |
| 268 | 67 # generate room for locals: |
| 174 | 68 |
| 268 | 69 for sym in fn.innerScope: |
| 316 | 70 self.the_type(sym.typ) |
| 272 | 71 if sym.isParameter: |
| 316 | 72 p = ir.Parameter(sym.name) |
| 73 variable = ir.LocalVariable(sym.name + '_copy') | |
| 74 f.addParameter(p) | |
| 75 f.addLocal(variable) | |
| 76 # Move parameter into local copy: | |
| 77 self.emit(ir.Move(ir.Mem(variable), p)) | |
| 274 | 78 elif sym.isLocal: |
| 308 | 79 variable = ir.LocalVariable(sym.name) |
| 80 f.addLocal(variable) | |
| 81 elif isinstance(sym, ast.Variable): | |
| 82 variable = ir.LocalVariable(sym.name) | |
| 83 f.addLocal(variable) | |
| 274 | 84 else: |
| 275 | 85 raise NotImplementedError('{}'.format(sym)) |
| 308 | 86 self.varMap[sym] = variable |
| 268 | 87 |
| 88 self.genCode(fn.body) | |
| 272 | 89 self.emit(ir.Move(f.return_value, ir.Const(0))) |
| 269 | 90 self.emit(ir.Jump(f.epiloog)) |
| 268 | 91 self.setFunction(None) |
| 158 | 92 |
| 217 | 93 def genCode(self, code): |
| 308 | 94 """ Wrapper around gen_stmt to catch errors """ |
| 307 | 95 try: |
| 308 | 96 self.gen_stmt(code) |
| 307 | 97 except SemanticError as e: |
| 98 self.error(e.msg, e.loc) | |
| 99 | |
| 308 | 100 def gen_stmt(self, code): |
| 101 """ Generate code for a statement """ | |
| 102 assert isinstance(code, ast.Statement) | |
| 268 | 103 self.setLoc(code.loc) |
| 308 | 104 if type(code) is ast.Compound: |
| 221 | 105 for s in code.statements: |
| 106 self.genCode(s) | |
| 308 | 107 elif type(code) is ast.Empty: |
| 306 | 108 pass |
| 308 | 109 elif type(code) is ast.Assignment: |
| 268 | 110 lval = self.genExprCode(code.lval) |
| 307 | 111 rval = self.genExprCode(code.rval) |
| 112 if not self.equalTypes(code.lval.typ, code.rval.typ): | |
| 315 | 113 msg = 'Cannot assign {} to {}'.format(code.rval.typ, code.lval.typ) |
| 307 | 114 raise SemanticError(msg, code.loc) |
| 115 if not code.lval.lvalue: | |
| 116 raise SemanticError('No valid lvalue {}'.format(code.lval), code.lval.loc) | |
| 268 | 117 self.emit(ir.Move(lval, rval)) |
| 308 | 118 elif type(code) is ast.ExpressionStatement: |
| 275 | 119 self.emit(ir.Exp(self.genExprCode(code.ex))) |
| 308 | 120 elif type(code) is ast.If: |
| 268 | 121 bbtrue = self.newBlock() |
| 122 bbfalse = self.newBlock() | |
| 123 te = self.newBlock() | |
| 308 | 124 self.gen_cond_code(code.condition, bbtrue, bbfalse) |
| 268 | 125 self.setBlock(bbtrue) |
| 126 self.genCode(code.truestatement) | |
| 127 self.emit(ir.Jump(te)) | |
| 128 self.setBlock(bbfalse) | |
| 306 | 129 self.genCode(code.falsestatement) |
| 268 | 130 self.emit(ir.Jump(te)) |
| 131 self.setBlock(te) | |
| 308 | 132 elif type(code) is ast.Return: |
| 303 | 133 re = self.genExprCode(code.expr) |
| 134 self.emit(ir.Move(self.fn.return_value, re)) | |
| 135 self.emit(ir.Jump(self.fn.epiloog)) | |
| 136 b = self.newBlock() | |
| 137 self.setBlock(b) | |
| 308 | 138 elif type(code) is ast.While: |
| 268 | 139 bbdo = self.newBlock() |
| 140 bbtest = self.newBlock() | |
| 141 te = self.newBlock() | |
| 142 self.emit(ir.Jump(bbtest)) | |
| 143 self.setBlock(bbtest) | |
| 308 | 144 self.gen_cond_code(code.condition, bbdo, te) |
| 268 | 145 self.setBlock(bbdo) |
| 228 | 146 self.genCode(code.statement) |
| 268 | 147 self.emit(ir.Jump(bbtest)) |
| 148 self.setBlock(te) | |
| 315 | 149 elif type(code) is ast.For: |
| 150 bbdo = self.newBlock() | |
| 151 bbtest = self.newBlock() | |
| 152 te = self.newBlock() | |
| 153 self.genCode(code.init) | |
| 154 self.emit(ir.Jump(bbtest)) | |
| 155 self.setBlock(bbtest) | |
| 156 self.gen_cond_code(code.condition, bbdo, te) | |
| 157 self.setBlock(bbdo) | |
| 158 self.genCode(code.statement) | |
| 159 self.emit(ir.Jump(bbtest)) | |
| 160 self.setBlock(te) | |
| 222 | 161 else: |
| 268 | 162 raise NotImplementedError('Unknown stmt {}'.format(code)) |
| 230 | 163 |
| 308 | 164 def gen_cond_code(self, expr, bbtrue, bbfalse): |
| 165 """ Generate conditional logic. | |
| 166 Implement sequential logical operators. """ | |
| 167 if type(expr) is ast.Binop: | |
| 268 | 168 if expr.op == 'or': |
| 169 l2 = self.newBlock() | |
| 308 | 170 self.gen_cond_code(expr.a, bbtrue, l2) |
| 171 if not self.equalTypes(expr.a.typ, self.boolType): | |
| 307 | 172 raise SemanticError('Must be boolean', expr.a.loc) |
| 268 | 173 self.setBlock(l2) |
| 308 | 174 self.gen_cond_code(expr.b, bbtrue, bbfalse) |
| 175 if not self.equalTypes(expr.b.typ, self.boolType): | |
| 307 | 176 raise SemanticError('Must be boolean', expr.b.loc) |
| 268 | 177 elif expr.op == 'and': |
| 178 l2 = self.newBlock() | |
| 308 | 179 self.gen_cond_code(expr.a, l2, bbfalse) |
| 180 if not self.equalTypes(expr.a.typ, self.boolType): | |
| 307 | 181 self.error('Must be boolean', expr.a.loc) |
| 268 | 182 self.setBlock(l2) |
| 308 | 183 self.gen_cond_code(expr.b, bbtrue, bbfalse) |
| 184 if not self.equalTypes(expr.b.typ, self.boolType): | |
| 307 | 185 raise SemanticError('Must be boolean', expr.b.loc) |
| 305 | 186 elif expr.op in ['==', '>', '<', '!=', '<=', '>=']: |
| 228 | 187 ta = self.genExprCode(expr.a) |
| 188 tb = self.genExprCode(expr.b) | |
| 307 | 189 if not self.equalTypes(expr.a.typ, expr.b.typ): |
| 190 raise SemanticError('Types unequal {} != {}' | |
| 191 .format(expr.a.typ, expr.b.typ), expr.loc) | |
| 268 | 192 self.emit(ir.CJump(ta, expr.op, tb, bbtrue, bbfalse)) |
| 193 else: | |
| 311 | 194 raise SemanticError('non-bool: {}'.format(expr.op), expr.loc) |
| 307 | 195 expr.typ = self.boolType |
| 308 | 196 elif type(expr) is ast.Literal: |
| 197 self.genExprCode(expr) | |
| 288 | 198 if expr.val: |
| 268 | 199 self.emit(ir.Jump(bbtrue)) |
| 288 | 200 else: |
| 268 | 201 self.emit(ir.Jump(bbfalse)) |
| 228 | 202 else: |
| 288 | 203 raise NotImplementedError('Unknown cond {}'.format(expr)) |
| 354 | 204 |
| 205 # Check that the condition is a boolean value: | |
| 307 | 206 if not self.equalTypes(expr.typ, self.boolType): |
| 207 self.error('Condition must be boolean', expr.loc) | |
| 230 | 208 |
| 217 | 209 def genExprCode(self, expr): |
| 308 | 210 """ Generate code for an expression. Return the generated ir-value """ |
| 211 assert isinstance(expr, ast.Expression) | |
| 212 if type(expr) is ast.Binop: | |
| 307 | 213 expr.lvalue = False |
| 214 if expr.op in ['+', '-', '*', '/', '<<', '>>', '|', '&']: | |
| 215 ra = self.genExprCode(expr.a) | |
| 216 rb = self.genExprCode(expr.b) | |
| 217 if self.equalTypes(expr.a.typ, self.intType) and \ | |
| 218 self.equalTypes(expr.b.typ, self.intType): | |
| 219 expr.typ = expr.a.typ | |
| 220 else: | |
| 221 raise SemanticError('Can only add integers', expr.loc) | |
| 222 else: | |
| 223 raise NotImplementedError("Cannot use equality as expressions") | |
| 268 | 224 return ir.Binop(ra, expr.op, rb) |
| 308 | 225 elif type(expr) is ast.Unop: |
| 307 | 226 if expr.op == '&': |
| 227 ra = self.genExprCode(expr.a) | |
| 308 | 228 expr.typ = ast.PointerType(expr.a.typ) |
| 307 | 229 if not expr.a.lvalue: |
| 230 raise SemanticError('No valid lvalue', expr.a.loc) | |
| 231 expr.lvalue = False | |
| 232 assert type(ra) is ir.Mem | |
| 233 return ra.e | |
| 234 else: | |
| 235 raise NotImplementedError('Unknown unop {0}'.format(expr.op)) | |
| 308 | 236 elif type(expr) is ast.Identifier: |
| 307 | 237 # Generate code for this identifier. |
| 238 tg = self.resolveSymbol(expr) | |
| 239 expr.kind = type(tg) | |
| 240 expr.typ = tg.typ | |
| 279 | 241 # This returns the dereferenced variable. |
| 308 | 242 if isinstance(tg, ast.Variable): |
| 313 | 243 expr.lvalue = True |
| 307 | 244 return ir.Mem(self.varMap[tg]) |
| 313 | 245 elif isinstance(tg, ast.Constant): |
| 246 c_val = self.genExprCode(tg.value) | |
| 247 return self.evalConst(c_val) | |
|
280
02385f62f250
Rework from str interface to Instruction interface
Windel Bouwman
parents:
279
diff
changeset
|
248 else: |
| 308 | 249 raise NotImplementedError(str(tg)) |
| 250 elif type(expr) is ast.Deref: | |
| 222 | 251 # dereference pointer type: |
| 225 | 252 addr = self.genExprCode(expr.ptr) |
| 316 | 253 ptr_typ = self.the_type(expr.ptr.typ) |
| 307 | 254 expr.lvalue = True |
| 308 | 255 if type(ptr_typ) is ast.PointerType: |
| 307 | 256 expr.typ = ptr_typ.ptype |
| 257 return ir.Mem(addr) | |
| 258 else: | |
| 259 raise SemanticError('Cannot deref non-pointer', expr.loc) | |
| 308 | 260 elif type(expr) is ast.Member: |
| 279 | 261 base = self.genExprCode(expr.base) |
| 307 | 262 expr.lvalue = expr.base.lvalue |
| 316 | 263 basetype = self.the_type(expr.base.typ) |
| 308 | 264 if type(basetype) is ast.StructureType: |
| 307 | 265 if basetype.hasField(expr.field): |
| 266 expr.typ = basetype.fieldType(expr.field) | |
| 267 else: | |
| 268 raise SemanticError('{} does not contain field {}' | |
| 308 | 269 .format(basetype, expr.field), expr.loc) |
| 307 | 270 else: |
| 308 | 271 raise SemanticError('Cannot select {} of non-structure type {}' |
| 272 .format(expr.field, basetype), expr.loc) | |
| 307 | 273 |
| 279 | 274 assert type(base) is ir.Mem, type(base) |
| 316 | 275 bt = self.the_type(expr.base.typ) |
| 268 | 276 offset = ir.Const(bt.fieldOffset(expr.field)) |
| 308 | 277 return ir.Mem(ir.Add(base.e, offset)) |
| 354 | 278 elif type(expr) is ast.Index: |
| 279 """ Array indexing """ | |
| 280 base = self.genExprCode(expr.base) | |
| 281 idx = self.genExprCode(expr.i) | |
| 282 base_typ = self.the_type(expr.base.typ) | |
| 283 if not isinstance(base_typ, ast.ArrayType): | |
| 284 raise SemanticError('Cannot index non-array type {}'.format(base_typ), expr.base.loc) | |
| 285 idx_type = self.the_type(expr.i.typ) | |
| 286 if not self.equalTypes(idx_type, self.intType): | |
| 287 raise SemanticError('Index must be int not {}'.format(idx_type), expr.i.loc) | |
| 288 assert type(base) is ir.Mem | |
| 289 element_type = self.the_type(base_typ.element_type) | |
| 290 element_size = self.size_of(element_type) | |
| 291 expr.typ = base_typ.element_type | |
| 292 expr.lvalue = True | |
| 293 | |
| 294 return ir.Mem(ir.Add(base.e, ir.Mul(idx, ir.Const(element_size)))) | |
| 308 | 295 elif type(expr) is ast.Literal: |
| 307 | 296 expr.lvalue = False |
| 313 | 297 typemap = {int: 'int', float: 'double', bool: 'bool', str:'string'} |
| 307 | 298 if type(expr.val) in typemap: |
| 299 expr.typ = self.pkg.scope[typemap[type(expr.val)]] | |
| 300 else: | |
| 312 | 301 raise SemanticError('Unknown literal type {}'.format(expr.val), expr.loc) |
| 354 | 302 # Construct correct const value: |
| 303 if type(expr.val) is str: | |
| 304 cval = struct.pack('<I', len(expr.val)) + expr.val.encode('ascii') | |
| 305 return ir.Addr(ir.Const(cval)) | |
| 306 else: | |
| 307 return ir.Const(expr.val) | |
| 308 | 308 elif type(expr) is ast.TypeCast: |
| 309 return self.gen_type_cast(expr) | |
| 310 elif type(expr) is ast.FunctionCall: | |
| 311 return self.gen_function_call(expr) | |
| 225 | 312 else: |
| 259 | 313 raise NotImplementedError('Unknown expr {}'.format(expr)) |
| 307 | 314 |
| 308 | 315 def gen_type_cast(self, expr): |
| 316 """ Generate code for type casting """ | |
| 317 ar = self.genExprCode(expr.a) | |
| 316 | 318 from_type = self.the_type(expr.a.typ) |
| 319 to_type = self.the_type(expr.to_type) | |
| 308 | 320 if isinstance(from_type, ast.PointerType) and isinstance(to_type, ast.PointerType): |
| 321 expr.typ = expr.to_type | |
| 322 return ar | |
| 323 elif type(from_type) is ast.BaseType and from_type.name == 'int' and \ | |
| 324 isinstance(to_type, ast.PointerType): | |
| 325 expr.typ = expr.to_type | |
| 326 return ar | |
| 354 | 327 elif type(from_type) is ast.BaseType and from_type.name == 'byte' and \ |
| 328 type(to_type) is ast.BaseType and to_type.name == 'int': | |
| 329 expr.typ = expr.to_type | |
| 330 return ar | |
| 308 | 331 else: |
| 332 raise SemanticError('Cannot cast {} to {}' | |
| 333 .format(from_type, to_type), expr.loc) | |
| 353 | 334 |
| 308 | 335 def gen_function_call(self, expr): |
| 336 """ Generate code for a function call """ | |
| 337 # Evaluate the arguments: | |
| 338 args = [self.genExprCode(e) for e in expr.args] | |
| 339 # Check arguments: | |
| 340 tg = self.resolveSymbol(expr.proc) | |
| 341 if type(tg) is not ast.Function: | |
| 342 raise SemanticError('cannot call {}'.format(tg)) | |
| 343 ftyp = tg.typ | |
|
336
d1ecc493384e
Added spiffy armtoken class for bit fiddeling. Added cool test that checks for build repeatability
Windel Bouwman
parents:
334
diff
changeset
|
344 fname = tg.package.name + '_' + tg.name |
| 308 | 345 ptypes = ftyp.parametertypes |
| 346 if len(expr.args) != len(ptypes): | |
| 347 raise SemanticError('{} requires {} arguments, {} given' | |
| 348 .format(fname, len(ptypes), len(expr.args)), expr.loc) | |
| 349 for arg, at in zip(expr.args, ptypes): | |
| 350 if not self.equalTypes(arg.typ, at): | |
| 351 raise SemanticError('Got {}, expected {}' | |
| 352 .format(arg.typ, at), arg.loc) | |
| 353 # determine return type: | |
| 354 expr.typ = ftyp.returntype | |
| 355 return ir.Call(fname, args) | |
| 356 | |
| 313 | 357 def evalConst(self, c): |
| 358 if isinstance(c, ir.Const): | |
| 359 return c | |
| 360 else: | |
| 361 raise SemanticError('Cannot evaluate constant {}'.format(c)) | |
| 362 | |
| 307 | 363 def resolveSymbol(self, sym): |
| 308 | 364 if type(sym) is ast.Member: |
| 307 | 365 base = self.resolveSymbol(sym.base) |
| 308 | 366 if type(base) is not ast.Package: |
| 367 raise SemanticError('Base is not a package', sym.loc) | |
| 307 | 368 scope = base.innerScope |
| 369 name = sym.field | |
| 308 | 370 elif type(sym) is ast.Identifier: |
| 307 | 371 scope = sym.scope |
| 372 name = sym.target | |
| 373 else: | |
| 374 raise NotImplementedError(str(sym)) | |
| 375 if name in scope: | |
| 376 s = scope[name] | |
| 377 else: | |
| 378 raise SemanticError('{} undefined'.format(name), sym.loc) | |
| 308 | 379 assert isinstance(s, ast.Symbol) |
| 307 | 380 return s |
| 381 | |
| 316 | 382 def size_of(self, t): |
| 383 """ Determine the byte size of a type """ | |
| 384 t = self.the_type(t) | |
| 385 if type(t) is ast.BaseType: | |
| 386 return t.bytesize | |
| 387 elif type(t) is ast.StructureType: | |
| 388 return sum(self.size_of(mem.typ) for mem in t.mems) | |
| 354 | 389 elif type(t) is ast.ArrayType: |
| 390 return t.size * self.size_of(t.element_type) | |
| 316 | 391 else: |
| 392 raise NotImplementedError(str(t)) | |
| 393 | |
| 394 def the_type(self, t): | |
| 307 | 395 """ Recurse until a 'real' type is found """ |
| 308 | 396 if type(t) is ast.DefinedType: |
| 316 | 397 t = self.the_type(t.typ) |
| 308 | 398 elif type(t) in [ast.Identifier, ast.Member]: |
| 316 | 399 t = self.the_type(self.resolveSymbol(t)) |
| 400 elif type(t) is ast.StructureType: | |
| 401 # Setup offsets of fields. Is this the right place?: | |
| 402 offset = 0 | |
| 403 for mem in t.mems: | |
| 404 mem.offset = offset | |
| 405 offset = offset + self.size_of(mem.typ) | |
| 308 | 406 elif isinstance(t, ast.Type): |
| 307 | 407 pass |
| 408 else: | |
| 409 raise NotImplementedError(str(t)) | |
| 308 | 410 assert isinstance(t, ast.Type) |
| 307 | 411 return t |
| 412 | |
| 413 def equalTypes(self, a, b): | |
| 414 """ Compare types a and b for structural equavalence. """ | |
| 415 # Recurse into named types: | |
| 316 | 416 a = self.the_type(a) |
| 417 b = self.the_type(b) | |
| 307 | 418 |
| 419 if type(a) is type(b): | |
| 308 | 420 if type(a) is ast.BaseType: |
| 307 | 421 return a.name == b.name |
| 308 | 422 elif type(a) is ast.PointerType: |
| 307 | 423 return self.equalTypes(a.ptype, b.ptype) |
| 308 | 424 elif type(a) is ast.StructureType: |
| 307 | 425 if len(a.mems) != len(b.mems): |
| 426 return False | |
| 427 return all(self.equalTypes(am.typ, bm.typ) for am, bm in | |
| 428 zip(a.mems, b.mems)) | |
| 354 | 429 elif type(a) is ast.ArrayType: |
| 430 return self.equalTypes(a.element_type, b.element_type) | |
| 307 | 431 else: |
| 432 raise NotImplementedError('{} not implemented'.format(type(a))) | |
| 433 return False |