1 #!/usr/bin/env python 2 3 """ 4 Encoder functions, producing representations of program objects. 5 6 Copyright (C) 2016 Paul Boddie <paul@boddie.org.uk> 7 8 This program is free software; you can redistribute it and/or modify it under 9 the terms of the GNU General Public License as published by the Free Software 10 Foundation; either version 3 of the License, or (at your option) any later 11 version. 12 13 This program is distributed in the hope that it will be useful, but WITHOUT 14 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS 15 FOR A PARTICULAR PURPOSE. See the GNU General Public License for more 16 details. 17 18 You should have received a copy of the GNU General Public License along with 19 this program. If not, see <http://www.gnu.org/licenses/>. 20 """ 21 22 from common import first 23 24 # Output encoding and decoding for the summary files. 25 26 def encode_attrnames(attrnames): 27 28 "Encode the 'attrnames' representing usage." 29 30 return ", ".join(attrnames) or "{}" 31 32 def encode_constrained(constrained): 33 34 "Encode the 'constrained' status for program summaries." 35 36 return constrained and "constrained" or "deduced" 37 38 def encode_usage(usage): 39 40 "Encode attribute details from 'usage'." 41 42 all_attrnames = [] 43 for t in usage: 44 attrname, invocation, assignment = t 45 all_attrnames.append("%s%s" % (attrname, invocation and "!" or assignment and "=" or "")) 46 return ", ".join(all_attrnames) or "{}" 47 48 def decode_usage(s): 49 50 "Decode attribute details from 's'." 51 52 all_attrnames = set() 53 for attrname_str in s.split(", "): 54 all_attrnames.add((attrname_str.rstrip("!="), attrname_str.endswith("!"), attrname_str.endswith("="))) 55 56 all_attrnames = list(all_attrnames) 57 all_attrnames.sort() 58 return tuple(all_attrnames) 59 60 def encode_access_location(t): 61 62 "Encode the access location 't'." 63 64 path, name, attrname, version = t 65 return "%s %s %s:%d" % (path, name or "{}", attrname, version) 66 67 def encode_location(t): 68 69 "Encode the general location 't' in a concise form." 70 71 path, name, attrname, version = t 72 if name is not None and version is not None: 73 return "%s %s:%d" % (path, name, version) 74 elif name is not None: 75 return "%s %s" % (path, name) 76 else: 77 return "%s :%s" % (path, attrname) 78 79 def encode_modifiers(modifiers): 80 81 "Encode assignment details from 'modifiers'." 82 83 all_modifiers = [] 84 for t in modifiers: 85 all_modifiers.append(encode_modifier_term(t)) 86 return "".join(all_modifiers) 87 88 def encode_modifier_term(t): 89 90 "Encode modifier 't' representing assignment status." 91 92 assignment, invocation = t 93 return assignment and "=" or invocation and "!" or "_" 94 95 def decode_modifier_term(s): 96 97 "Decode modifier term 's' representing assignment status." 98 99 return (s == "=", s == "!") 100 101 102 103 # Test generation functions. 104 105 def get_kinds(all_types): 106 107 """ 108 Return object kind details for 'all_types', being a collection of 109 references for program types. 110 """ 111 112 return map(lambda ref: ref.get_kind(), all_types) 113 114 def test_label_for_kind(kind): 115 116 "Return the label used for 'kind' in test details." 117 118 return kind == "<instance>" and "instance" or "type" 119 120 def test_label_for_type(ref): 121 122 "Return the label used for 'ref' in test details." 123 124 return test_label_for_kind(ref.get_kind()) 125 126 127 128 # Instruction representation encoding. 129 130 def encode_instruction(instruction): 131 132 """ 133 Encode the 'instruction' - a sequence starting with an operation and 134 followed by arguments, each of which may be an instruction sequence or a 135 plain value - to produce a function call string representation. 136 """ 137 138 op = instruction[0] 139 args = instruction[1:] 140 141 if args: 142 a = [] 143 for arg in args: 144 if isinstance(arg, tuple): 145 a.append(encode_instruction(arg)) 146 else: 147 a.append(arg or "{}") 148 argstr = "(%s)" % ", ".join(a) 149 return "%s%s" % (op, argstr) 150 else: 151 return op 152 153 154 155 # Output program encoding. 156 157 attribute_loading_ops = ( 158 "__load_via_class", "__load_via_object", "__get_class_and_load", 159 ) 160 161 attribute_ops = attribute_loading_ops + ( 162 "__store_via_object", 163 ) 164 165 checked_loading_ops = ( 166 "__check_and_load_via_class", "__check_and_load_via_object", "__check_and_load_via_any", 167 ) 168 169 checked_ops = checked_loading_ops + ( 170 "__check_and_store_via_class", "__check_and_store_via_object", "__check_and_store_via_any", 171 ) 172 173 typename_ops = ( 174 "__test_common_instance", "__test_common_object", "__test_common_type", 175 ) 176 177 type_ops = ( 178 "__test_specific_instance", "__test_specific_object", "__test_specific_type", 179 ) 180 181 static_ops = ( 182 "__load_static", 183 ) 184 185 reference_acting_ops = attribute_ops + checked_ops + typename_ops 186 attribute_producing_ops = attribute_loading_ops + checked_loading_ops 187 188 def encode_access_instruction(instruction, subs): 189 190 """ 191 Encode the 'instruction' - a sequence starting with an operation and 192 followed by arguments, each of which may be an instruction sequence or a 193 plain value - to produce a function call string representation. 194 195 The 'subs' parameter defines a mapping of substitutions for special values 196 used in instructions. 197 """ 198 199 op = instruction[0] 200 args = instruction[1:] 201 202 if not args: 203 argstr = "" 204 205 else: 206 # Encode the arguments. 207 208 a = [] 209 converting_op = op 210 for arg in args: 211 a.append(encode_access_instruction_arg(arg, subs, converting_op)) 212 converting_op = None 213 214 # Modify certain arguments. 215 216 # Convert attribute name arguments to position symbols. 217 218 if op in attribute_ops: 219 arg = a[1] 220 a[1] = encode_symbol("pos", arg) 221 222 # Convert attribute name arguments to position and code symbols. 223 224 elif op in checked_ops: 225 arg = a[1] 226 a[1] = encode_symbol("pos", arg) 227 a.insert(2, encode_symbol("code", arg)) 228 229 # Convert type name arguments to position and code symbols. 230 231 elif op in typename_ops: 232 arg = encode_type_attribute(args[1]) 233 a[1] = encode_symbol("pos", arg) 234 a.insert(2, encode_symbol("code", arg)) 235 236 # Obtain addresses of type arguments. 237 238 elif op in type_ops: 239 a[1] = "&%s" % a[1] 240 241 # Obtain addresses of static objects. 242 243 elif op in static_ops: 244 a[0] = "&%s" % a[0] 245 a[1] = "&%s" % a[1] 246 247 argstr = "(%s)" % ", ".join(a) 248 249 # Substitute the first element of the instruction, which may not be an 250 # operation at all. 251 252 if subs.has_key(op): 253 op = subs[op] 254 elif not args: 255 op = "&%s" % encode_path(op) 256 257 return "%s%s" % (op, argstr) 258 259 def encode_access_instruction_arg(arg, subs, op): 260 261 "Encode 'arg' using 'subs' to define substitutions." 262 263 if isinstance(arg, tuple): 264 encoded = encode_access_instruction(arg, subs) 265 266 # Convert attribute results to references where required. 267 268 if op and op in reference_acting_ops and arg[0] in attribute_producing_ops: 269 return "%s.value" % encoded 270 else: 271 return encoded 272 273 # Special values only need replacing, not encoding. 274 275 elif subs.has_key(arg): 276 return subs.get(arg) 277 278 # Convert static references to the appropriate type. 279 280 elif op and op in reference_acting_ops and arg != "<accessor>": 281 return "&%s" % encode_path(arg) 282 283 # Other values may need encoding. 284 285 else: 286 return encode_path(arg) 287 288 def encode_bound_reference(path): 289 290 "Encode 'path' as a bound method name." 291 292 return "__bound_%s" % encode_path(path) 293 294 def encode_function_pointer(path): 295 296 "Encode 'path' as a reference to an output program function." 297 298 return "__fn_%s" % encode_path(path) 299 300 def encode_initialiser_pointer(path): 301 302 "Encode 'path' as a reference to an initialiser function structure." 303 304 return encode_path("%s.__init__" % path) 305 306 def encode_instantiator_pointer(path): 307 308 "Encode 'path' as a reference to an output program instantiator." 309 310 return "__new_%s" % encode_path(path) 311 312 def encode_literal_constant(n): 313 314 "Encode a name for the literal constant with the number 'n'." 315 316 return "__const%d" % n 317 318 def encode_literal_constant_size(value): 319 320 "Encode a size for the literal constant with the given 'value'." 321 322 if isinstance(value, basestring): 323 return len(value) 324 else: 325 return 0 326 327 def encode_literal_constant_member(value): 328 329 "Encode the member name for the 'value' in the final program." 330 331 return "%svalue" % value.__class__.__name__ 332 333 def encode_literal_constant_value(value): 334 335 "Encode the given 'value' in the final program." 336 337 if isinstance(value, (int, float)): 338 return str(value) 339 else: 340 return '"%s"' % str(value).replace('"', '\\"').replace("\n", "\\n").replace("\t", "\\t").replace("\r", "\\r") 341 342 def encode_literal_data_initialiser(style): 343 344 """ 345 Encode a reference to a function populating the data for a literal having 346 the given 'style' ("mapping" or "sequence"). 347 """ 348 349 return "__newdata_%s" % style 350 351 def encode_literal_instantiator(path): 352 353 """ 354 Encode a reference to an instantiator for a literal having the given 'path'. 355 """ 356 357 return "__newliteral_%s" % encode_path(path) 358 359 def encode_literal_reference(n): 360 361 "Encode a reference to a literal constant with the number 'n'." 362 363 return "__constvalue%d" % n 364 365 # Track all encoded paths, detecting and avoiding conflicts. 366 367 all_encoded_paths = {} 368 369 def encode_path(path): 370 371 "Encode 'path' as an output program object, translating special symbols." 372 373 if path in reserved_words: 374 return "__%s" % path 375 else: 376 part_encoded = path.replace("#", "__").replace("$", "__") 377 378 if "." not in path: 379 return part_encoded 380 381 encoded = part_encoded.replace(".", "_") 382 383 # Test for a conflict with the encoding of a different path, re-encoding 384 # if necessary. 385 386 previous = all_encoded_paths.get(encoded) 387 replacement = "_" 388 389 while previous: 390 if path == previous: 391 return encoded 392 replacement += "_" 393 encoded = part_encoded.replace(".", replacement) 394 previous = all_encoded_paths.get(encoded) 395 396 # Store any new or re-encoded path. 397 398 all_encoded_paths[encoded] = path 399 return encoded 400 401 def encode_predefined_reference(path): 402 403 "Encode a reference to a predefined constant value for 'path'." 404 405 return "__predefined_%s" % encode_path(path) 406 407 def encode_size(kind, path=None): 408 409 """ 410 Encode a structure size reference for the given 'kind' of structure, with 411 'path' indicating a specific structure name. 412 """ 413 414 return "__%ssize%s" % (structure_size_prefixes.get(kind, kind), path and "_%s" % encode_path(path) or "") 415 416 def encode_symbol(symbol_type, path=None): 417 418 "Encode a symbol with the given 'symbol_type' and optional 'path'." 419 420 return "__%s%s" % (symbol_type, path and "_%s" % encode_path(path) or "") 421 422 def encode_tablename(kind, path): 423 424 """ 425 Encode a table reference for the given 'kind' of table structure, indicating 426 a 'path' for the specific object concerned. 427 """ 428 429 return "__%sTable_%s" % (table_name_prefixes[kind], encode_path(path)) 430 431 def encode_type_attribute(path): 432 433 "Encode the special type attribute for 'path'." 434 435 return "#%s" % path 436 437 def decode_type_attribute(s): 438 439 "Decode the special type attribute 's'." 440 441 return s[1:] 442 443 def is_type_attribute(s): 444 445 "Return whether 's' is a type attribute name." 446 447 return s.startswith("#") 448 449 450 451 # A mapping from kinds to structure size reference prefixes. 452 453 structure_size_prefixes = { 454 "<class>" : "c", 455 "<module>" : "m", 456 "<instance>" : "i" 457 } 458 459 # A mapping from kinds to table name prefixes. 460 461 table_name_prefixes = { 462 "<class>" : "Class", 463 "<function>" : "Function", 464 "<module>" : "Module", 465 "<instance>" : "Instance" 466 } 467 468 469 470 # Output language reserved words. 471 472 reserved_words = [ 473 "break", "char", "const", "continue", 474 "default", "double", "else", 475 "float", "for", 476 "if", "int", "long", 477 "NULL", 478 "return", "struct", 479 "typedef", 480 "void", "while", 481 ] 482 483 # vim: tabstop=4 expandtab shiftwidth=4