| paul@0 | 1 | #!/usr/bin/env python |
| paul@0 | 2 | |
| paul@0 | 3 | """ |
| paul@0 | 4 | Encoder functions, producing representations of program objects. |
| paul@0 | 5 | |
| paul@498 | 6 | Copyright (C) 2016, 2017 Paul Boddie <paul@boddie.org.uk> |
| paul@0 | 7 | |
| paul@0 | 8 | This program is free software; you can redistribute it and/or modify it under |
| paul@0 | 9 | the terms of the GNU General Public License as published by the Free Software |
| paul@0 | 10 | Foundation; either version 3 of the License, or (at your option) any later |
| paul@0 | 11 | version. |
| paul@0 | 12 | |
| paul@0 | 13 | This program is distributed in the hope that it will be useful, but WITHOUT |
| paul@0 | 14 | ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS |
| paul@0 | 15 | FOR A PARTICULAR PURPOSE. See the GNU General Public License for more |
| paul@0 | 16 | details. |
| paul@0 | 17 | |
| paul@0 | 18 | You should have received a copy of the GNU General Public License along with |
| paul@0 | 19 | this program. If not, see <http://www.gnu.org/licenses/>. |
| paul@0 | 20 | """ |
| paul@0 | 21 | |
| paul@498 | 22 | from common import first, InstructionSequence |
| paul@56 | 23 | |
| paul@609 | 24 | |
| paul@609 | 25 | |
| paul@609 | 26 | # Value digest computation. |
| paul@609 | 27 | |
| paul@609 | 28 | from base64 import b64encode |
| paul@609 | 29 | from hashlib import sha1 |
| paul@609 | 30 | |
| paul@609 | 31 | def digest(values): |
| paul@609 | 32 | m = sha1() |
| paul@609 | 33 | for value in values: |
| paul@609 | 34 | m.update(str(value)) |
| paul@609 | 35 | return b64encode(m.digest()).replace("+", "__").replace("/", "_").rstrip("=") |
| paul@609 | 36 | |
| paul@609 | 37 | |
| paul@609 | 38 | |
| paul@0 | 39 | # Output encoding and decoding for the summary files. |
| paul@0 | 40 | |
| paul@0 | 41 | def encode_attrnames(attrnames): |
| paul@0 | 42 | |
| paul@0 | 43 | "Encode the 'attrnames' representing usage." |
| paul@0 | 44 | |
| paul@0 | 45 | return ", ".join(attrnames) or "{}" |
| paul@0 | 46 | |
| paul@0 | 47 | def encode_constrained(constrained): |
| paul@0 | 48 | |
| paul@0 | 49 | "Encode the 'constrained' status for program summaries." |
| paul@0 | 50 | |
| paul@0 | 51 | return constrained and "constrained" or "deduced" |
| paul@0 | 52 | |
| paul@0 | 53 | def encode_usage(usage): |
| paul@0 | 54 | |
| paul@0 | 55 | "Encode attribute details from 'usage'." |
| paul@0 | 56 | |
| paul@0 | 57 | all_attrnames = [] |
| paul@0 | 58 | for t in usage: |
| paul@107 | 59 | attrname, invocation, assignment = t |
| paul@107 | 60 | all_attrnames.append("%s%s" % (attrname, invocation and "!" or assignment and "=" or "")) |
| paul@0 | 61 | return ", ".join(all_attrnames) or "{}" |
| paul@0 | 62 | |
| paul@88 | 63 | def decode_usage(s): |
| paul@88 | 64 | |
| paul@88 | 65 | "Decode attribute details from 's'." |
| paul@88 | 66 | |
| paul@88 | 67 | all_attrnames = set() |
| paul@88 | 68 | for attrname_str in s.split(", "): |
| paul@107 | 69 | all_attrnames.add((attrname_str.rstrip("!="), attrname_str.endswith("!"), attrname_str.endswith("="))) |
| paul@88 | 70 | |
| paul@88 | 71 | all_attrnames = list(all_attrnames) |
| paul@88 | 72 | all_attrnames.sort() |
| paul@88 | 73 | return tuple(all_attrnames) |
| paul@88 | 74 | |
| paul@0 | 75 | def encode_access_location(t): |
| paul@0 | 76 | |
| paul@0 | 77 | "Encode the access location 't'." |
| paul@0 | 78 | |
| paul@0 | 79 | path, name, attrname, version = t |
| paul@0 | 80 | return "%s %s %s:%d" % (path, name or "{}", attrname, version) |
| paul@0 | 81 | |
| paul@0 | 82 | def encode_location(t): |
| paul@0 | 83 | |
| paul@0 | 84 | "Encode the general location 't' in a concise form." |
| paul@0 | 85 | |
| paul@0 | 86 | path, name, attrname, version = t |
| paul@0 | 87 | if name is not None and version is not None: |
| paul@0 | 88 | return "%s %s:%d" % (path, name, version) |
| paul@0 | 89 | elif name is not None: |
| paul@0 | 90 | return "%s %s" % (path, name) |
| paul@0 | 91 | else: |
| paul@0 | 92 | return "%s :%s" % (path, attrname) |
| paul@0 | 93 | |
| paul@0 | 94 | def encode_modifiers(modifiers): |
| paul@0 | 95 | |
| paul@553 | 96 | "Encode assignment and invocation details from 'modifiers'." |
| paul@0 | 97 | |
| paul@0 | 98 | all_modifiers = [] |
| paul@0 | 99 | for t in modifiers: |
| paul@0 | 100 | all_modifiers.append(encode_modifier_term(t)) |
| paul@0 | 101 | return "".join(all_modifiers) |
| paul@0 | 102 | |
| paul@0 | 103 | def encode_modifier_term(t): |
| paul@0 | 104 | |
| paul@553 | 105 | "Encode modifier 't' representing an assignment or an invocation." |
| paul@0 | 106 | |
| paul@117 | 107 | assignment, invocation = t |
| paul@553 | 108 | if assignment: |
| paul@553 | 109 | return "=" |
| paul@553 | 110 | elif invocation is not None: |
| paul@557 | 111 | arguments, keywords = invocation |
| paul@557 | 112 | return "(%d;%s)" % (arguments, ",".join(keywords)) |
| paul@553 | 113 | else: |
| paul@553 | 114 | return "_" |
| paul@0 | 115 | |
| paul@553 | 116 | def decode_modifiers(s): |
| paul@553 | 117 | |
| paul@553 | 118 | "Decode 's' containing modifiers." |
| paul@553 | 119 | |
| paul@553 | 120 | i = 0 |
| paul@553 | 121 | end = len(s) |
| paul@0 | 122 | |
| paul@553 | 123 | modifiers = [] |
| paul@0 | 124 | |
| paul@553 | 125 | while i < end: |
| paul@553 | 126 | if s[i] == "=": |
| paul@553 | 127 | modifiers.append((True, None)) |
| paul@553 | 128 | i += 1 |
| paul@553 | 129 | elif s[i] == "(": |
| paul@557 | 130 | j = s.index(";", i) |
| paul@557 | 131 | arguments = int(s[i+1:j]) |
| paul@557 | 132 | i = j |
| paul@553 | 133 | j = s.index(")", i) |
| paul@557 | 134 | keywords = s[i+1:j] |
| paul@557 | 135 | keywords = keywords and keywords.split(",") or [] |
| paul@557 | 136 | modifiers.append((False, (arguments, keywords))) |
| paul@553 | 137 | i = j + 1 |
| paul@553 | 138 | else: |
| paul@553 | 139 | modifiers.append((False, None)) |
| paul@553 | 140 | i += 1 |
| paul@553 | 141 | |
| paul@553 | 142 | return modifiers |
| paul@0 | 143 | |
| paul@56 | 144 | |
| paul@56 | 145 | |
| paul@56 | 146 | # Test generation functions. |
| paul@56 | 147 | |
| paul@56 | 148 | def get_kinds(all_types): |
| paul@56 | 149 | |
| paul@56 | 150 | """ |
| paul@56 | 151 | Return object kind details for 'all_types', being a collection of |
| paul@56 | 152 | references for program types. |
| paul@56 | 153 | """ |
| paul@56 | 154 | |
| paul@56 | 155 | return map(lambda ref: ref.get_kind(), all_types) |
| paul@56 | 156 | |
| paul@237 | 157 | def test_label_for_kind(kind): |
| paul@56 | 158 | |
| paul@237 | 159 | "Return the label used for 'kind' in test details." |
| paul@56 | 160 | |
| paul@237 | 161 | return kind == "<instance>" and "instance" or "type" |
| paul@56 | 162 | |
| paul@237 | 163 | def test_label_for_type(ref): |
| paul@56 | 164 | |
| paul@237 | 165 | "Return the label used for 'ref' in test details." |
| paul@56 | 166 | |
| paul@237 | 167 | return test_label_for_kind(ref.get_kind()) |
| paul@56 | 168 | |
| paul@56 | 169 | |
| paul@56 | 170 | |
| paul@94 | 171 | # Instruction representation encoding. |
| paul@94 | 172 | |
| paul@94 | 173 | def encode_instruction(instruction): |
| paul@94 | 174 | |
| paul@94 | 175 | """ |
| paul@94 | 176 | Encode the 'instruction' - a sequence starting with an operation and |
| paul@94 | 177 | followed by arguments, each of which may be an instruction sequence or a |
| paul@94 | 178 | plain value - to produce a function call string representation. |
| paul@94 | 179 | """ |
| paul@94 | 180 | |
| paul@94 | 181 | op = instruction[0] |
| paul@94 | 182 | args = instruction[1:] |
| paul@94 | 183 | |
| paul@94 | 184 | if args: |
| paul@94 | 185 | a = [] |
| paul@113 | 186 | for arg in args: |
| paul@113 | 187 | if isinstance(arg, tuple): |
| paul@113 | 188 | a.append(encode_instruction(arg)) |
| paul@94 | 189 | else: |
| paul@113 | 190 | a.append(arg or "{}") |
| paul@94 | 191 | argstr = "(%s)" % ", ".join(a) |
| paul@94 | 192 | return "%s%s" % (op, argstr) |
| paul@94 | 193 | else: |
| paul@94 | 194 | return op |
| paul@94 | 195 | |
| paul@94 | 196 | |
| paul@94 | 197 | |
| paul@0 | 198 | # Output program encoding. |
| paul@0 | 199 | |
| paul@153 | 200 | attribute_loading_ops = ( |
| paul@153 | 201 | "__load_via_class", "__load_via_object", "__get_class_and_load", |
| paul@153 | 202 | ) |
| paul@153 | 203 | |
| paul@153 | 204 | attribute_ops = attribute_loading_ops + ( |
| paul@113 | 205 | "__store_via_object", |
| paul@113 | 206 | ) |
| paul@113 | 207 | |
| paul@153 | 208 | checked_loading_ops = ( |
| paul@113 | 209 | "__check_and_load_via_class", "__check_and_load_via_object", "__check_and_load_via_any", |
| paul@153 | 210 | ) |
| paul@153 | 211 | |
| paul@153 | 212 | checked_ops = checked_loading_ops + ( |
| paul@113 | 213 | "__check_and_store_via_class", "__check_and_store_via_object", "__check_and_store_via_any", |
| paul@113 | 214 | ) |
| paul@113 | 215 | |
| paul@113 | 216 | typename_ops = ( |
| paul@144 | 217 | "__test_common_instance", "__test_common_object", "__test_common_type", |
| paul@113 | 218 | ) |
| paul@113 | 219 | |
| paul@385 | 220 | type_ops = ( |
| paul@385 | 221 | "__test_specific_instance", "__test_specific_object", "__test_specific_type", |
| paul@385 | 222 | ) |
| paul@385 | 223 | |
| paul@141 | 224 | static_ops = ( |
| paul@595 | 225 | "__load_static_ignore", "__load_static_replace", "__load_static_test", "<test_context_static>", |
| paul@141 | 226 | ) |
| paul@141 | 227 | |
| paul@591 | 228 | context_values = ( |
| paul@591 | 229 | "<context>", |
| paul@591 | 230 | ) |
| paul@591 | 231 | |
| paul@591 | 232 | context_ops = ( |
| paul@601 | 233 | "<context>", "<set_context>", "<test_context_revert>", "<test_context_static>", |
| paul@591 | 234 | ) |
| paul@591 | 235 | |
| paul@602 | 236 | context_op_functions = ( |
| paul@602 | 237 | "<test_context_revert>", "<test_context_static>", |
| paul@602 | 238 | ) |
| paul@602 | 239 | |
| paul@153 | 240 | reference_acting_ops = attribute_ops + checked_ops + typename_ops |
| paul@153 | 241 | attribute_producing_ops = attribute_loading_ops + checked_loading_ops |
| paul@153 | 242 | |
| paul@591 | 243 | def encode_access_instruction(instruction, subs, context_index): |
| paul@113 | 244 | |
| paul@113 | 245 | """ |
| paul@113 | 246 | Encode the 'instruction' - a sequence starting with an operation and |
| paul@113 | 247 | followed by arguments, each of which may be an instruction sequence or a |
| paul@113 | 248 | plain value - to produce a function call string representation. |
| paul@113 | 249 | |
| paul@113 | 250 | The 'subs' parameter defines a mapping of substitutions for special values |
| paul@113 | 251 | used in instructions. |
| paul@482 | 252 | |
| paul@591 | 253 | The 'context_index' parameter defines the position in local context storage |
| paul@591 | 254 | for the referenced context or affected by a context operation. |
| paul@591 | 255 | |
| paul@482 | 256 | Return both the encoded instruction and a collection of substituted names. |
| paul@113 | 257 | """ |
| paul@113 | 258 | |
| paul@113 | 259 | op = instruction[0] |
| paul@113 | 260 | args = instruction[1:] |
| paul@482 | 261 | substituted = set() |
| paul@113 | 262 | |
| paul@591 | 263 | # Encode the arguments. |
| paul@113 | 264 | |
| paul@591 | 265 | a = [] |
| paul@591 | 266 | if args: |
| paul@153 | 267 | converting_op = op |
| paul@113 | 268 | for arg in args: |
| paul@591 | 269 | s, _substituted = encode_access_instruction_arg(arg, subs, converting_op, context_index) |
| paul@482 | 270 | substituted.update(_substituted) |
| paul@482 | 271 | a.append(s) |
| paul@153 | 272 | converting_op = None |
| paul@113 | 273 | |
| paul@591 | 274 | # Modify certain arguments. |
| paul@113 | 275 | |
| paul@591 | 276 | # Convert attribute name arguments to position symbols. |
| paul@113 | 277 | |
| paul@591 | 278 | if op in attribute_ops: |
| paul@591 | 279 | arg = a[1] |
| paul@591 | 280 | a[1] = encode_symbol("pos", arg) |
| paul@591 | 281 | |
| paul@591 | 282 | # Convert attribute name arguments to position and code symbols. |
| paul@113 | 283 | |
| paul@591 | 284 | elif op in checked_ops: |
| paul@591 | 285 | arg = a[1] |
| paul@591 | 286 | a[1] = encode_symbol("pos", arg) |
| paul@591 | 287 | a.insert(2, encode_symbol("code", arg)) |
| paul@113 | 288 | |
| paul@591 | 289 | # Convert type name arguments to position and code symbols. |
| paul@113 | 290 | |
| paul@591 | 291 | elif op in typename_ops: |
| paul@591 | 292 | arg = encode_type_attribute(args[1]) |
| paul@591 | 293 | a[1] = encode_symbol("pos", arg) |
| paul@591 | 294 | a.insert(2, encode_symbol("code", arg)) |
| paul@113 | 295 | |
| paul@591 | 296 | # Obtain addresses of type arguments. |
| paul@591 | 297 | |
| paul@591 | 298 | elif op in type_ops: |
| paul@591 | 299 | a[1] = "&%s" % a[1] |
| paul@113 | 300 | |
| paul@591 | 301 | # Obtain addresses of static objects. |
| paul@591 | 302 | |
| paul@591 | 303 | elif op in static_ops: |
| paul@591 | 304 | a[-1] = "&%s" % a[-1] |
| paul@385 | 305 | |
| paul@591 | 306 | # Add context storage information to certain operations. |
| paul@385 | 307 | |
| paul@595 | 308 | if op in context_ops: |
| paul@591 | 309 | a.insert(0, context_index) |
| paul@141 | 310 | |
| paul@602 | 311 | # Add the local context array to certain operations. |
| paul@602 | 312 | |
| paul@602 | 313 | if op in context_op_functions: |
| paul@602 | 314 | a.append("__tmp_contexts") |
| paul@602 | 315 | |
| paul@591 | 316 | # Define any argument string. |
| paul@141 | 317 | |
| paul@591 | 318 | if a: |
| paul@491 | 319 | argstr = "(%s)" % ", ".join(map(str, a)) |
| paul@591 | 320 | else: |
| paul@591 | 321 | argstr = "" |
| paul@113 | 322 | |
| paul@113 | 323 | # Substitute the first element of the instruction, which may not be an |
| paul@113 | 324 | # operation at all. |
| paul@113 | 325 | |
| paul@144 | 326 | if subs.has_key(op): |
| paul@482 | 327 | substituted.add(op) |
| paul@498 | 328 | |
| paul@498 | 329 | # Break accessor initialisation into initialisation and value-yielding |
| paul@498 | 330 | # parts: |
| paul@498 | 331 | |
| paul@498 | 332 | if op == "<set_accessor>" and isinstance(a[0], InstructionSequence): |
| paul@498 | 333 | ops = [] |
| paul@498 | 334 | ops += a[0].get_init_instructions() |
| paul@498 | 335 | ops.append("%s(%s)" % (subs[op], a[0].get_value_instruction())) |
| paul@498 | 336 | return ", ".join(map(str, ops)), substituted |
| paul@498 | 337 | |
| paul@144 | 338 | op = subs[op] |
| paul@498 | 339 | |
| paul@144 | 340 | elif not args: |
| paul@144 | 341 | op = "&%s" % encode_path(op) |
| paul@144 | 342 | |
| paul@482 | 343 | return "%s%s" % (op, argstr), substituted |
| paul@113 | 344 | |
| paul@591 | 345 | def encode_access_instruction_arg(arg, subs, op, context_index): |
| paul@113 | 346 | |
| paul@482 | 347 | """ |
| paul@591 | 348 | Encode 'arg' using 'subs' to define substitutions, 'op' to indicate the |
| paul@591 | 349 | operation to which the argument belongs, and 'context_index' to indicate any |
| paul@591 | 350 | affected context storage. |
| paul@591 | 351 | |
| paul@591 | 352 | Return a tuple containing the encoded form of 'arg' along with a collection |
| paul@591 | 353 | of any substituted values. |
| paul@482 | 354 | """ |
| paul@113 | 355 | |
| paul@113 | 356 | if isinstance(arg, tuple): |
| paul@591 | 357 | encoded, substituted = encode_access_instruction(arg, subs, context_index) |
| paul@153 | 358 | |
| paul@153 | 359 | # Convert attribute results to references where required. |
| paul@153 | 360 | |
| paul@153 | 361 | if op and op in reference_acting_ops and arg[0] in attribute_producing_ops: |
| paul@482 | 362 | return "%s.value" % encoded, substituted |
| paul@153 | 363 | else: |
| paul@482 | 364 | return encoded, substituted |
| paul@113 | 365 | |
| paul@113 | 366 | # Special values only need replacing, not encoding. |
| paul@113 | 367 | |
| paul@113 | 368 | elif subs.has_key(arg): |
| paul@591 | 369 | |
| paul@591 | 370 | # Handle values modified by storage details. |
| paul@591 | 371 | |
| paul@591 | 372 | if arg in context_values: |
| paul@591 | 373 | return "%s(%s)" % (subs.get(arg), context_index), set([arg]) |
| paul@591 | 374 | else: |
| paul@591 | 375 | return subs.get(arg), set([arg]) |
| paul@113 | 376 | |
| paul@258 | 377 | # Convert static references to the appropriate type. |
| paul@258 | 378 | |
| paul@258 | 379 | elif op and op in reference_acting_ops and arg != "<accessor>": |
| paul@482 | 380 | return "&%s" % encode_path(arg), set() |
| paul@258 | 381 | |
| paul@113 | 382 | # Other values may need encoding. |
| paul@113 | 383 | |
| paul@113 | 384 | else: |
| paul@482 | 385 | return encode_path(arg), set() |
| paul@113 | 386 | |
| paul@0 | 387 | def encode_function_pointer(path): |
| paul@0 | 388 | |
| paul@0 | 389 | "Encode 'path' as a reference to an output program function." |
| paul@0 | 390 | |
| paul@0 | 391 | return "__fn_%s" % encode_path(path) |
| paul@0 | 392 | |
| paul@0 | 393 | def encode_instantiator_pointer(path): |
| paul@0 | 394 | |
| paul@0 | 395 | "Encode 'path' as a reference to an output program instantiator." |
| paul@0 | 396 | |
| paul@0 | 397 | return "__new_%s" % encode_path(path) |
| paul@0 | 398 | |
| paul@491 | 399 | def encode_instructions(instructions): |
| paul@491 | 400 | |
| paul@491 | 401 | "Encode 'instructions' as a sequence." |
| paul@491 | 402 | |
| paul@491 | 403 | if len(instructions) == 1: |
| paul@491 | 404 | return instructions[0] |
| paul@491 | 405 | else: |
| paul@491 | 406 | return "(\n%s\n)" % ",\n".join(instructions) |
| paul@491 | 407 | |
| paul@136 | 408 | def encode_literal_constant(n): |
| paul@136 | 409 | |
| paul@136 | 410 | "Encode a name for the literal constant with the number 'n'." |
| paul@136 | 411 | |
| paul@609 | 412 | return "__const%s" % n |
| paul@136 | 413 | |
| paul@378 | 414 | def encode_literal_constant_size(value): |
| paul@378 | 415 | |
| paul@378 | 416 | "Encode a size for the literal constant with the given 'value'." |
| paul@378 | 417 | |
| paul@378 | 418 | if isinstance(value, basestring): |
| paul@378 | 419 | return len(value) |
| paul@378 | 420 | else: |
| paul@378 | 421 | return 0 |
| paul@378 | 422 | |
| paul@136 | 423 | def encode_literal_constant_member(value): |
| paul@136 | 424 | |
| paul@136 | 425 | "Encode the member name for the 'value' in the final program." |
| paul@136 | 426 | |
| paul@136 | 427 | return "%svalue" % value.__class__.__name__ |
| paul@136 | 428 | |
| paul@136 | 429 | def encode_literal_constant_value(value): |
| paul@136 | 430 | |
| paul@136 | 431 | "Encode the given 'value' in the final program." |
| paul@136 | 432 | |
| paul@136 | 433 | if isinstance(value, (int, float)): |
| paul@136 | 434 | return str(value) |
| paul@136 | 435 | else: |
| paul@451 | 436 | l = [] |
| paul@451 | 437 | |
| paul@451 | 438 | # Encode characters including non-ASCII ones. |
| paul@451 | 439 | |
| paul@451 | 440 | for c in str(value): |
| paul@451 | 441 | if c == '"': l.append('\\"') |
| paul@451 | 442 | elif c == '\n': l.append('\\n') |
| paul@451 | 443 | elif c == '\t': l.append('\\t') |
| paul@451 | 444 | elif c == '\r': l.append('\\r') |
| paul@512 | 445 | elif c == '\\': l.append('\\\\') |
| paul@451 | 446 | elif 0x20 <= ord(c) < 0x80: l.append(c) |
| paul@451 | 447 | else: l.append("\\x%02x" % ord(c)) |
| paul@451 | 448 | |
| paul@451 | 449 | return '"%s"' % "".join(l) |
| paul@136 | 450 | |
| paul@283 | 451 | def encode_literal_data_initialiser(style): |
| paul@283 | 452 | |
| paul@283 | 453 | """ |
| paul@283 | 454 | Encode a reference to a function populating the data for a literal having |
| paul@283 | 455 | the given 'style' ("mapping" or "sequence"). |
| paul@283 | 456 | """ |
| paul@283 | 457 | |
| paul@283 | 458 | return "__newdata_%s" % style |
| paul@283 | 459 | |
| paul@159 | 460 | def encode_literal_instantiator(path): |
| paul@159 | 461 | |
| paul@159 | 462 | """ |
| paul@159 | 463 | Encode a reference to an instantiator for a literal having the given 'path'. |
| paul@159 | 464 | """ |
| paul@159 | 465 | |
| paul@159 | 466 | return "__newliteral_%s" % encode_path(path) |
| paul@159 | 467 | |
| paul@136 | 468 | def encode_literal_reference(n): |
| paul@136 | 469 | |
| paul@136 | 470 | "Encode a reference to a literal constant with the number 'n'." |
| paul@136 | 471 | |
| paul@609 | 472 | return "__constvalue%s" % n |
| paul@136 | 473 | |
| paul@512 | 474 | |
| paul@512 | 475 | |
| paul@340 | 476 | # Track all encoded paths, detecting and avoiding conflicts. |
| paul@340 | 477 | |
| paul@340 | 478 | all_encoded_paths = {} |
| paul@340 | 479 | |
| paul@0 | 480 | def encode_path(path): |
| paul@0 | 481 | |
| paul@0 | 482 | "Encode 'path' as an output program object, translating special symbols." |
| paul@0 | 483 | |
| paul@0 | 484 | if path in reserved_words: |
| paul@0 | 485 | return "__%s" % path |
| paul@0 | 486 | else: |
| paul@340 | 487 | part_encoded = path.replace("#", "__").replace("$", "__") |
| paul@349 | 488 | |
| paul@349 | 489 | if "." not in path: |
| paul@349 | 490 | return part_encoded |
| paul@349 | 491 | |
| paul@340 | 492 | encoded = part_encoded.replace(".", "_") |
| paul@340 | 493 | |
| paul@340 | 494 | # Test for a conflict with the encoding of a different path, re-encoding |
| paul@340 | 495 | # if necessary. |
| paul@340 | 496 | |
| paul@340 | 497 | previous = all_encoded_paths.get(encoded) |
| paul@340 | 498 | replacement = "_" |
| paul@340 | 499 | |
| paul@340 | 500 | while previous: |
| paul@340 | 501 | if path == previous: |
| paul@340 | 502 | return encoded |
| paul@340 | 503 | replacement += "_" |
| paul@340 | 504 | encoded = part_encoded.replace(".", replacement) |
| paul@340 | 505 | previous = all_encoded_paths.get(encoded) |
| paul@340 | 506 | |
| paul@340 | 507 | # Store any new or re-encoded path. |
| paul@340 | 508 | |
| paul@340 | 509 | all_encoded_paths[encoded] = path |
| paul@340 | 510 | return encoded |
| paul@0 | 511 | |
| paul@136 | 512 | def encode_predefined_reference(path): |
| paul@136 | 513 | |
| paul@136 | 514 | "Encode a reference to a predefined constant value for 'path'." |
| paul@136 | 515 | |
| paul@136 | 516 | return "__predefined_%s" % encode_path(path) |
| paul@136 | 517 | |
| paul@150 | 518 | def encode_size(kind, path=None): |
| paul@150 | 519 | |
| paul@150 | 520 | """ |
| paul@150 | 521 | Encode a structure size reference for the given 'kind' of structure, with |
| paul@150 | 522 | 'path' indicating a specific structure name. |
| paul@150 | 523 | """ |
| paul@150 | 524 | |
| paul@150 | 525 | return "__%ssize%s" % (structure_size_prefixes.get(kind, kind), path and "_%s" % encode_path(path) or "") |
| paul@150 | 526 | |
| paul@0 | 527 | def encode_symbol(symbol_type, path=None): |
| paul@0 | 528 | |
| paul@0 | 529 | "Encode a symbol with the given 'symbol_type' and optional 'path'." |
| paul@0 | 530 | |
| paul@0 | 531 | return "__%s%s" % (symbol_type, path and "_%s" % encode_path(path) or "") |
| paul@0 | 532 | |
| paul@150 | 533 | def encode_tablename(kind, path): |
| paul@150 | 534 | |
| paul@150 | 535 | """ |
| paul@150 | 536 | Encode a table reference for the given 'kind' of table structure, indicating |
| paul@150 | 537 | a 'path' for the specific object concerned. |
| paul@150 | 538 | """ |
| paul@150 | 539 | |
| paul@150 | 540 | return "__%sTable_%s" % (table_name_prefixes[kind], encode_path(path)) |
| paul@150 | 541 | |
| paul@131 | 542 | def encode_type_attribute(path): |
| paul@131 | 543 | |
| paul@131 | 544 | "Encode the special type attribute for 'path'." |
| paul@131 | 545 | |
| paul@131 | 546 | return "#%s" % path |
| paul@131 | 547 | |
| paul@318 | 548 | def decode_type_attribute(s): |
| paul@318 | 549 | |
| paul@318 | 550 | "Decode the special type attribute 's'." |
| paul@318 | 551 | |
| paul@318 | 552 | return s[1:] |
| paul@318 | 553 | |
| paul@318 | 554 | def is_type_attribute(s): |
| paul@318 | 555 | |
| paul@318 | 556 | "Return whether 's' is a type attribute name." |
| paul@318 | 557 | |
| paul@318 | 558 | return s.startswith("#") |
| paul@318 | 559 | |
| paul@56 | 560 | |
| paul@56 | 561 | |
| paul@150 | 562 | # A mapping from kinds to structure size reference prefixes. |
| paul@150 | 563 | |
| paul@150 | 564 | structure_size_prefixes = { |
| paul@150 | 565 | "<class>" : "c", |
| paul@150 | 566 | "<module>" : "m", |
| paul@150 | 567 | "<instance>" : "i" |
| paul@150 | 568 | } |
| paul@150 | 569 | |
| paul@150 | 570 | # A mapping from kinds to table name prefixes. |
| paul@150 | 571 | |
| paul@150 | 572 | table_name_prefixes = { |
| paul@150 | 573 | "<class>" : "Class", |
| paul@150 | 574 | "<function>" : "Function", |
| paul@150 | 575 | "<module>" : "Module", |
| paul@150 | 576 | "<instance>" : "Instance" |
| paul@150 | 577 | } |
| paul@150 | 578 | |
| paul@150 | 579 | |
| paul@150 | 580 | |
| paul@0 | 581 | # Output language reserved words. |
| paul@0 | 582 | |
| paul@0 | 583 | reserved_words = [ |
| paul@0 | 584 | "break", "char", "const", "continue", |
| paul@0 | 585 | "default", "double", "else", |
| paul@0 | 586 | "float", "for", |
| paul@0 | 587 | "if", "int", "long", |
| paul@0 | 588 | "NULL", |
| paul@0 | 589 | "return", "struct", |
| paul@0 | 590 | "typedef", |
| paul@0 | 591 | "void", "while", |
| paul@0 | 592 | ] |
| paul@0 | 593 | |
| paul@0 | 594 | # vim: tabstop=4 expandtab shiftwidth=4 |
