Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 1 | # Copyright 2014 The Swarming Authors. All rights reserved. |
| 2 | # Use of this source code is governed under the Apache License, Version 2.0 that |
| 3 | # can be found in the LICENSE file. |
| 4 | |
| 5 | """Contains logic to parse .isolate files. |
| 6 | |
| 7 | This module doesn't touch the file system. It's the job of the client code to do |
| 8 | I/O on behalf of this module. |
| 9 | |
| 10 | See more information at |
| 11 | https://code.google.com/p/swarming/wiki/IsolateDesign |
| 12 | https://code.google.com/p/swarming/wiki/IsolateUserGuide |
| 13 | """ |
| 14 | |
| 15 | import ast |
| 16 | import copy |
| 17 | import itertools |
| 18 | import logging |
| 19 | import os |
| 20 | import re |
| 21 | |
| 22 | import isolateserver |
| 23 | |
| 24 | from utils import short_expression_finder |
| 25 | |
| 26 | # Files that should be 0-length when mapped. |
| 27 | KEY_TOUCHED = 'isolate_dependency_touched' |
| 28 | # Files that should be tracked by the build tool. |
| 29 | KEY_TRACKED = 'isolate_dependency_tracked' |
| 30 | # Files that should not be tracked by the build tool. |
| 31 | KEY_UNTRACKED = 'isolate_dependency_untracked' |
| 32 | |
| 33 | # Valid variable name. |
| 34 | VALID_VARIABLE = '[A-Za-z_][A-Za-z_0-9]*' |
| 35 | |
| 36 | |
| 37 | def determine_root_dir(relative_root, infiles): |
| 38 | """For a list of infiles, determines the deepest root directory that is |
| 39 | referenced indirectly. |
| 40 | |
| 41 | All arguments must be using os.path.sep. |
| 42 | """ |
| 43 | # The trick used to determine the root directory is to look at "how far" back |
| 44 | # up it is looking up. |
| 45 | deepest_root = relative_root |
| 46 | for i in infiles: |
| 47 | x = relative_root |
| 48 | while i.startswith('..' + os.path.sep): |
| 49 | i = i[3:] |
| 50 | assert not i.startswith(os.path.sep) |
| 51 | x = os.path.dirname(x) |
| 52 | if deepest_root.startswith(x): |
| 53 | deepest_root = x |
| 54 | logging.debug( |
| 55 | 'determine_root_dir(%s, %d files) -> %s' % ( |
| 56 | relative_root, len(infiles), deepest_root)) |
| 57 | return deepest_root |
| 58 | |
| 59 | |
| 60 | def replace_variable(part, variables): |
| 61 | m = re.match(r'<\((' + VALID_VARIABLE + ')\)', part) |
| 62 | if m: |
| 63 | if m.group(1) not in variables: |
| 64 | raise isolateserver.ConfigError( |
| 65 | 'Variable "%s" was not found in %s.\nDid you forget to specify ' |
| 66 | '--path-variable?' % (m.group(1), variables)) |
| 67 | return variables[m.group(1)] |
| 68 | return part |
| 69 | |
| 70 | |
| 71 | def eval_variables(item, variables): |
| 72 | """Replaces the .isolate variables in a string item. |
| 73 | |
| 74 | Note that the .isolate format is a subset of the .gyp dialect. |
| 75 | """ |
| 76 | return ''.join( |
| 77 | replace_variable(p, variables) |
| 78 | for p in re.split(r'(<\(' + VALID_VARIABLE + '\))', item)) |
| 79 | |
| 80 | |
| 81 | def split_touched(files): |
| 82 | """Splits files that are touched vs files that are read.""" |
| 83 | tracked = [] |
| 84 | touched = [] |
| 85 | for f in files: |
| 86 | if f.size: |
| 87 | tracked.append(f) |
| 88 | else: |
| 89 | touched.append(f) |
| 90 | return tracked, touched |
| 91 | |
| 92 | |
| 93 | def pretty_print(variables, stdout): |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 94 | """Outputs a .isolate file from the decoded variables. |
| 95 | |
| 96 | The .isolate format is GYP compatible. |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 97 | |
| 98 | Similar to pprint.print() but with NIH syndrome. |
| 99 | """ |
| 100 | # Order the dictionary keys by these keys in priority. |
| 101 | ORDER = ( |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 102 | 'variables', 'condition', 'command', 'read_only', |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 103 | KEY_TRACKED, KEY_UNTRACKED) |
| 104 | |
| 105 | def sorting_key(x): |
| 106 | """Gives priority to 'most important' keys before the others.""" |
| 107 | if x in ORDER: |
| 108 | return str(ORDER.index(x)) |
| 109 | return x |
| 110 | |
| 111 | def loop_list(indent, items): |
| 112 | for item in items: |
| 113 | if isinstance(item, basestring): |
| 114 | stdout.write('%s\'%s\',\n' % (indent, item)) |
| 115 | elif isinstance(item, dict): |
| 116 | stdout.write('%s{\n' % indent) |
| 117 | loop_dict(indent + ' ', item) |
| 118 | stdout.write('%s},\n' % indent) |
| 119 | elif isinstance(item, list): |
| 120 | # A list inside a list will write the first item embedded. |
| 121 | stdout.write('%s[' % indent) |
| 122 | for index, i in enumerate(item): |
| 123 | if isinstance(i, basestring): |
| 124 | stdout.write( |
| 125 | '\'%s\', ' % i.replace('\\', '\\\\').replace('\'', '\\\'')) |
| 126 | elif isinstance(i, dict): |
| 127 | stdout.write('{\n') |
| 128 | loop_dict(indent + ' ', i) |
| 129 | if index != len(item) - 1: |
| 130 | x = ', ' |
| 131 | else: |
| 132 | x = '' |
| 133 | stdout.write('%s}%s' % (indent, x)) |
| 134 | else: |
| 135 | assert False |
| 136 | stdout.write('],\n') |
| 137 | else: |
| 138 | assert False |
| 139 | |
| 140 | def loop_dict(indent, items): |
| 141 | for key in sorted(items, key=sorting_key): |
| 142 | item = items[key] |
| 143 | stdout.write("%s'%s': " % (indent, key)) |
| 144 | if isinstance(item, dict): |
| 145 | stdout.write('{\n') |
| 146 | loop_dict(indent + ' ', item) |
| 147 | stdout.write(indent + '},\n') |
| 148 | elif isinstance(item, list): |
| 149 | stdout.write('[\n') |
| 150 | loop_list(indent + ' ', item) |
| 151 | stdout.write(indent + '],\n') |
| 152 | elif isinstance(item, basestring): |
| 153 | stdout.write( |
| 154 | '\'%s\',\n' % item.replace('\\', '\\\\').replace('\'', '\\\'')) |
Marc-Antoine Ruel | 7124e39 | 2014-01-09 11:49:21 -0500 | [diff] [blame] | 155 | elif isinstance(item, (int, bool)) or item is None: |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 156 | stdout.write('%s\n' % item) |
| 157 | else: |
| 158 | assert False, item |
| 159 | |
| 160 | stdout.write('{\n') |
| 161 | loop_dict(' ', variables) |
| 162 | stdout.write('}\n') |
| 163 | |
| 164 | |
| 165 | def print_all(comment, data, stream): |
| 166 | """Prints a complete .isolate file and its top-level file comment into a |
| 167 | stream. |
| 168 | """ |
| 169 | if comment: |
| 170 | stream.write(comment) |
| 171 | pretty_print(data, stream) |
| 172 | |
| 173 | |
| 174 | def union(lhs, rhs): |
| 175 | """Merges two compatible datastructures composed of dict/list/set.""" |
| 176 | assert lhs is not None or rhs is not None |
| 177 | if lhs is None: |
| 178 | return copy.deepcopy(rhs) |
| 179 | if rhs is None: |
| 180 | return copy.deepcopy(lhs) |
| 181 | assert type(lhs) == type(rhs), (lhs, rhs) |
| 182 | if hasattr(lhs, 'union'): |
| 183 | # Includes set, ConfigSettings and Configs. |
| 184 | return lhs.union(rhs) |
| 185 | if isinstance(lhs, dict): |
| 186 | return dict((k, union(lhs.get(k), rhs.get(k))) for k in set(lhs).union(rhs)) |
| 187 | elif isinstance(lhs, list): |
| 188 | # Do not go inside the list. |
| 189 | return lhs + rhs |
| 190 | assert False, type(lhs) |
| 191 | |
| 192 | |
| 193 | def extract_comment(content): |
| 194 | """Extracts file level comment.""" |
| 195 | out = [] |
| 196 | for line in content.splitlines(True): |
| 197 | if line.startswith('#'): |
| 198 | out.append(line) |
| 199 | else: |
| 200 | break |
| 201 | return ''.join(out) |
| 202 | |
| 203 | |
| 204 | def eval_content(content): |
| 205 | """Evaluates a python file and return the value defined in it. |
| 206 | |
| 207 | Used in practice for .isolate files. |
| 208 | """ |
| 209 | globs = {'__builtins__': None} |
| 210 | locs = {} |
| 211 | try: |
| 212 | value = eval(content, globs, locs) |
| 213 | except TypeError as e: |
| 214 | e.args = list(e.args) + [content] |
| 215 | raise |
| 216 | assert locs == {}, locs |
| 217 | assert globs == {'__builtins__': None}, globs |
| 218 | return value |
| 219 | |
| 220 | |
| 221 | def match_configs(expr, config_variables, all_configs): |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 222 | """Returns the list of values from |values| that match the condition |expr|. |
| 223 | |
| 224 | Arguments: |
| 225 | expr: string that is evaluatable with eval(). It is a GYP condition. |
| 226 | config_variables: list of the name of the variables. |
| 227 | all_configs: list of the list of possible values. |
| 228 | |
| 229 | If a variable is not referenced at all, it is marked as unbounded (free) with |
| 230 | a value set to None. |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 231 | """ |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 232 | # It is more than just eval'ing the variable, it needs to be double checked to |
| 233 | # see if the variable is referenced at all. If not, the variable is free |
| 234 | # (unbounded). |
| 235 | # TODO(maruel): Use the intelligent way by inspecting expr instead of doing |
| 236 | # trial and error to figure out which variable is bound. |
| 237 | combinations = [] |
| 238 | for bound_variables in itertools.product( |
| 239 | (True, False), repeat=len(config_variables)): |
| 240 | # Add the combination of variables bound. |
| 241 | combinations.append( |
| 242 | ( |
| 243 | [c for c, b in zip(config_variables, bound_variables) if b], |
| 244 | set( |
| 245 | tuple(v if b else None for v, b in zip(line, bound_variables)) |
| 246 | for line in all_configs) |
| 247 | )) |
| 248 | |
| 249 | out = [] |
| 250 | for variables, configs in combinations: |
| 251 | # Strip variables and see if expr can still be evaluated. |
| 252 | for values in configs: |
| 253 | globs = {'__builtins__': None} |
| 254 | globs.update(zip(variables, (v for v in values if v is not None))) |
| 255 | try: |
| 256 | assertion = eval(expr, globs, {}) |
| 257 | except NameError: |
| 258 | continue |
| 259 | if not isinstance(assertion, bool): |
| 260 | raise isolateserver.ConfigError('Invalid condition') |
| 261 | if assertion: |
| 262 | out.append(values) |
| 263 | return out |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 264 | |
| 265 | |
| 266 | def verify_variables(variables): |
| 267 | """Verifies the |variables| dictionary is in the expected format.""" |
| 268 | VALID_VARIABLES = [ |
| 269 | KEY_TOUCHED, |
| 270 | KEY_TRACKED, |
| 271 | KEY_UNTRACKED, |
| 272 | 'command', |
| 273 | 'read_only', |
| 274 | ] |
| 275 | assert isinstance(variables, dict), variables |
| 276 | assert set(VALID_VARIABLES).issuperset(set(variables)), variables.keys() |
| 277 | for name, value in variables.iteritems(): |
| 278 | if name == 'read_only': |
Marc-Antoine Ruel | 7124e39 | 2014-01-09 11:49:21 -0500 | [diff] [blame] | 279 | assert value in (0, 1, 2, None), value |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 280 | else: |
| 281 | assert isinstance(value, list), value |
| 282 | assert all(isinstance(i, basestring) for i in value), value |
| 283 | |
| 284 | |
| 285 | def verify_ast(expr, variables_and_values): |
| 286 | """Verifies that |expr| is of the form |
| 287 | expr ::= expr ( "or" | "and" ) expr |
| 288 | | identifier "==" ( string | int ) |
| 289 | Also collects the variable identifiers and string/int values in the dict |
| 290 | |variables_and_values|, in the form {'var': set([val1, val2, ...]), ...}. |
| 291 | """ |
| 292 | assert isinstance(expr, (ast.BoolOp, ast.Compare)) |
| 293 | if isinstance(expr, ast.BoolOp): |
| 294 | assert isinstance(expr.op, (ast.And, ast.Or)) |
| 295 | for subexpr in expr.values: |
| 296 | verify_ast(subexpr, variables_and_values) |
| 297 | else: |
| 298 | assert isinstance(expr.left.ctx, ast.Load) |
| 299 | assert len(expr.ops) == 1 |
| 300 | assert isinstance(expr.ops[0], ast.Eq) |
| 301 | var_values = variables_and_values.setdefault(expr.left.id, set()) |
| 302 | rhs = expr.comparators[0] |
| 303 | assert isinstance(rhs, (ast.Str, ast.Num)) |
| 304 | var_values.add(rhs.n if isinstance(rhs, ast.Num) else rhs.s) |
| 305 | |
| 306 | |
| 307 | def verify_condition(condition, variables_and_values): |
| 308 | """Verifies the |condition| dictionary is in the expected format. |
| 309 | See verify_ast() for the meaning of |variables_and_values|. |
| 310 | """ |
| 311 | VALID_INSIDE_CONDITION = ['variables'] |
| 312 | assert isinstance(condition, list), condition |
| 313 | assert len(condition) == 2, condition |
| 314 | expr, then = condition |
| 315 | |
| 316 | test_ast = compile(expr, '<condition>', 'eval', ast.PyCF_ONLY_AST) |
| 317 | verify_ast(test_ast.body, variables_and_values) |
| 318 | |
| 319 | assert isinstance(then, dict), then |
| 320 | assert set(VALID_INSIDE_CONDITION).issuperset(set(then)), then.keys() |
| 321 | if not 'variables' in then: |
| 322 | raise isolateserver.ConfigError('Missing \'variables\' in condition %s' % |
| 323 | condition) |
| 324 | verify_variables(then['variables']) |
| 325 | |
| 326 | |
| 327 | def verify_root(value, variables_and_values): |
| 328 | """Verifies that |value| is the parsed form of a valid .isolate file. |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 329 | |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 330 | See verify_ast() for the meaning of |variables_and_values|. |
| 331 | """ |
Marc-Antoine Ruel | d7c032b | 2014-03-13 15:32:16 -0400 | [diff] [blame] | 332 | VALID_ROOTS = ['includes', 'conditions', 'variables'] |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 333 | assert isinstance(value, dict), value |
| 334 | assert set(VALID_ROOTS).issuperset(set(value)), value.keys() |
| 335 | |
| 336 | includes = value.get('includes', []) |
| 337 | assert isinstance(includes, list), includes |
| 338 | for include in includes: |
| 339 | assert isinstance(include, basestring), include |
| 340 | |
| 341 | conditions = value.get('conditions', []) |
| 342 | assert isinstance(conditions, list), conditions |
| 343 | for condition in conditions: |
| 344 | verify_condition(condition, variables_and_values) |
| 345 | |
Marc-Antoine Ruel | d7c032b | 2014-03-13 15:32:16 -0400 | [diff] [blame] | 346 | variables = value.get('variables', {}) |
| 347 | verify_variables(variables) |
| 348 | |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 349 | |
| 350 | def remove_weak_dependencies(values, key, item, item_configs): |
| 351 | """Removes any configs from this key if the item is already under a |
| 352 | strong key. |
| 353 | """ |
| 354 | if key == KEY_TOUCHED: |
| 355 | item_configs = set(item_configs) |
| 356 | for stronger_key in (KEY_TRACKED, KEY_UNTRACKED): |
| 357 | try: |
| 358 | item_configs -= values[stronger_key][item] |
| 359 | except KeyError: |
| 360 | pass |
| 361 | |
| 362 | return item_configs |
| 363 | |
| 364 | |
| 365 | def remove_repeated_dependencies(folders, key, item, item_configs): |
| 366 | """Removes any configs from this key if the item is in a folder that is |
| 367 | already included.""" |
| 368 | |
| 369 | if key in (KEY_UNTRACKED, KEY_TRACKED, KEY_TOUCHED): |
| 370 | item_configs = set(item_configs) |
| 371 | for (folder, configs) in folders.iteritems(): |
| 372 | if folder != item and item.startswith(folder): |
| 373 | item_configs -= configs |
| 374 | |
| 375 | return item_configs |
| 376 | |
| 377 | |
| 378 | def get_folders(values_dict): |
| 379 | """Returns a dict of all the folders in the given value_dict.""" |
| 380 | return dict( |
| 381 | (item, configs) for (item, configs) in values_dict.iteritems() |
| 382 | if item.endswith('/') |
| 383 | ) |
| 384 | |
| 385 | |
| 386 | def invert_map(variables): |
| 387 | """Converts {config: {deptype: list(depvals)}} to |
| 388 | {deptype: {depval: set(configs)}}. |
| 389 | """ |
| 390 | KEYS = ( |
| 391 | KEY_TOUCHED, |
| 392 | KEY_TRACKED, |
| 393 | KEY_UNTRACKED, |
| 394 | 'command', |
| 395 | 'read_only', |
| 396 | ) |
| 397 | out = dict((key, {}) for key in KEYS) |
| 398 | for config, values in variables.iteritems(): |
| 399 | for key in KEYS: |
| 400 | if key == 'command': |
| 401 | items = [tuple(values[key])] if key in values else [] |
| 402 | elif key == 'read_only': |
| 403 | items = [values[key]] if key in values else [] |
| 404 | else: |
| 405 | assert key in (KEY_TOUCHED, KEY_TRACKED, KEY_UNTRACKED) |
| 406 | items = values.get(key, []) |
| 407 | for item in items: |
| 408 | out[key].setdefault(item, set()).add(config) |
| 409 | return out |
| 410 | |
| 411 | |
| 412 | def reduce_inputs(values): |
| 413 | """Reduces the output of invert_map() to the strictest minimum list. |
| 414 | |
| 415 | Looks at each individual file and directory, maps where they are used and |
| 416 | reconstructs the inverse dictionary. |
| 417 | |
| 418 | Returns the minimized dictionary. |
| 419 | """ |
| 420 | KEYS = ( |
| 421 | KEY_TOUCHED, |
| 422 | KEY_TRACKED, |
| 423 | KEY_UNTRACKED, |
| 424 | 'command', |
| 425 | 'read_only', |
| 426 | ) |
| 427 | |
| 428 | # Folders can only live in KEY_UNTRACKED. |
| 429 | folders = get_folders(values.get(KEY_UNTRACKED, {})) |
| 430 | |
| 431 | out = dict((key, {}) for key in KEYS) |
| 432 | for key in KEYS: |
| 433 | for item, item_configs in values.get(key, {}).iteritems(): |
| 434 | item_configs = remove_weak_dependencies(values, key, item, item_configs) |
| 435 | item_configs = remove_repeated_dependencies( |
| 436 | folders, key, item, item_configs) |
| 437 | if item_configs: |
| 438 | out[key][item] = item_configs |
| 439 | return out |
| 440 | |
| 441 | |
| 442 | def convert_map_to_isolate_dict(values, config_variables): |
| 443 | """Regenerates back a .isolate configuration dict from files and dirs |
| 444 | mappings generated from reduce_inputs(). |
| 445 | """ |
| 446 | # Gather a list of configurations for set inversion later. |
| 447 | all_mentioned_configs = set() |
| 448 | for configs_by_item in values.itervalues(): |
| 449 | for configs in configs_by_item.itervalues(): |
| 450 | all_mentioned_configs.update(configs) |
| 451 | |
| 452 | # Invert the mapping to make it dict first. |
| 453 | conditions = {} |
| 454 | for key in values: |
| 455 | for item, configs in values[key].iteritems(): |
| 456 | then = conditions.setdefault(frozenset(configs), {}) |
| 457 | variables = then.setdefault('variables', {}) |
| 458 | |
Marc-Antoine Ruel | bb20b6d | 2014-01-10 18:47:47 -0500 | [diff] [blame] | 459 | if key == 'read_only': |
| 460 | if not isinstance(item, int): |
| 461 | raise isolateserver.ConfigError( |
| 462 | 'Unexpected entry type %r for key %s' % (item, key)) |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 463 | variables[key] = item |
Marc-Antoine Ruel | bb20b6d | 2014-01-10 18:47:47 -0500 | [diff] [blame] | 464 | elif key == 'command': |
| 465 | if not isinstance(item, tuple): |
| 466 | raise isolateserver.ConfigError( |
| 467 | 'Unexpected entry type %r for key %s' % (item, key)) |
| 468 | if key in variables: |
| 469 | raise isolateserver.ConfigError('Unexpected duplicate key %s' % key) |
| 470 | if not item: |
| 471 | raise isolateserver.ConfigError( |
| 472 | 'Expected non empty entry in %s' % key) |
| 473 | variables[key] = list(item) |
| 474 | elif key in (KEY_TOUCHED, KEY_TRACKED, KEY_UNTRACKED): |
| 475 | if not isinstance(item, basestring): |
| 476 | raise isolateserver.ConfigError('Unexpected entry type %r' % item) |
| 477 | if not item: |
| 478 | raise isolateserver.ConfigError( |
| 479 | 'Expected non empty entry in %s' % key) |
| 480 | # The list of items (files or dirs). Append the new item and keep |
| 481 | # the list sorted. |
| 482 | l = variables.setdefault(key, []) |
| 483 | l.append(item) |
| 484 | l.sort() |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 485 | else: |
Marc-Antoine Ruel | bb20b6d | 2014-01-10 18:47:47 -0500 | [diff] [blame] | 486 | raise isolateserver.ConfigError('Unexpected key %s' % key) |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 487 | |
| 488 | if all_mentioned_configs: |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 489 | # Change [(1, 2), (3, 4)] to [set(1, 3), set(2, 4)] |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 490 | config_values = map(set, zip(*all_mentioned_configs)) |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 491 | for i in config_values: |
| 492 | i.discard(None) |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 493 | sef = short_expression_finder.ShortExpressionFinder( |
| 494 | zip(config_variables, config_values)) |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 495 | conditions = sorted([sef.get_expr(c), v] for c, v in conditions.iteritems()) |
| 496 | else: |
| 497 | conditions = [] |
Marc-Antoine Ruel | d27c663 | 2014-03-13 15:29:36 -0400 | [diff] [blame] | 498 | out = {'conditions': conditions} |
| 499 | for c in conditions: |
| 500 | if c[0] == '': |
| 501 | # Extract the global. |
| 502 | out.update(c[1]) |
| 503 | conditions.remove(c) |
| 504 | break |
| 505 | return out |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 506 | |
| 507 | |
| 508 | class ConfigSettings(object): |
| 509 | """Represents the dependency variables for a single build configuration. |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 510 | |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 511 | The structure is immutable. |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 512 | |
| 513 | .touch, .tracked and .untracked are the list of dependencies. The items in |
| 514 | these lists use '/' as a path separator. |
| 515 | .command and .isolate_dir describe how to run the command. .isolate_dir uses |
| 516 | the OS' native path separator. It must be an absolute path, it's the path |
| 517 | where to start the command from. |
| 518 | .read_only describe how to map the files. |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 519 | """ |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 520 | def __init__(self, values, isolate_dir): |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 521 | verify_variables(values) |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 522 | assert isolate_dir is None or os.path.isabs(isolate_dir), isolate_dir |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 523 | self.touched = sorted(values.get(KEY_TOUCHED, [])) |
| 524 | self.tracked = sorted(values.get(KEY_TRACKED, [])) |
| 525 | self.untracked = sorted(values.get(KEY_UNTRACKED, [])) |
| 526 | self.command = values.get('command', [])[:] |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 527 | self.isolate_dir = isolate_dir |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 528 | self.read_only = values.get('read_only') |
| 529 | |
| 530 | def union(self, rhs): |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 531 | """Merges two config settings together into a new instance. |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 532 | |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 533 | self has priority over rhs for .command. Use the same |
| 534 | .isolate_dir as the one having a .command. Preferring self over rhs. |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 535 | """ |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 536 | var = { |
| 537 | KEY_TOUCHED: sorted(self.touched + rhs.touched), |
| 538 | KEY_TRACKED: sorted(self.tracked + rhs.tracked), |
| 539 | KEY_UNTRACKED: sorted(self.untracked + rhs.untracked), |
| 540 | 'command': self.command or rhs.command, |
| 541 | 'read_only': rhs.read_only if self.read_only is None else self.read_only, |
| 542 | } |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 543 | isolate_dir = self.isolate_dir if self.command else rhs.isolate_dir |
| 544 | return ConfigSettings(var, isolate_dir) |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 545 | |
| 546 | def flatten(self): |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 547 | """Converts the object into a dict.""" |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 548 | out = {} |
| 549 | if self.command: |
| 550 | out['command'] = self.command |
| 551 | if self.touched: |
| 552 | out[KEY_TOUCHED] = self.touched |
| 553 | if self.tracked: |
| 554 | out[KEY_TRACKED] = self.tracked |
| 555 | if self.untracked: |
| 556 | out[KEY_UNTRACKED] = self.untracked |
| 557 | if self.read_only is not None: |
| 558 | out['read_only'] = self.read_only |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 559 | # TODO(maruel): Probably better to not output it if command is None? |
| 560 | if self.isolate_dir is not None: |
| 561 | out['isolate_dir'] = self.isolate_dir |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 562 | return out |
| 563 | |
| 564 | |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 565 | def _safe_index(l, k): |
| 566 | try: |
| 567 | return l.index(k) |
| 568 | except ValueError: |
| 569 | return None |
| 570 | |
| 571 | |
| 572 | def _get_map_keys(dest_keys, in_keys): |
| 573 | """Returns a tuple of the indexes of each item in in_keys found in dest_keys. |
| 574 | |
| 575 | For example, if in_keys is ('A', 'C') and dest_keys is ('A', 'B', 'C'), the |
| 576 | return value will be (0, None, 1). |
| 577 | """ |
| 578 | return tuple(_safe_index(in_keys, k) for k in dest_keys) |
| 579 | |
| 580 | |
| 581 | def _map_keys(mapping, items): |
| 582 | """Returns a tuple with items placed at mapping index. |
| 583 | |
| 584 | For example, if mapping is (1, None, 0) and items is ('a', 'b'), it will |
| 585 | return ('b', None, 'c'). |
| 586 | """ |
| 587 | return tuple(items[i] if i != None else None for i in mapping) |
| 588 | |
| 589 | |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 590 | class Configs(object): |
| 591 | """Represents a processed .isolate file. |
| 592 | |
| 593 | Stores the file in a processed way, split by configuration. |
Marc-Antoine Ruel | 9ac1b91 | 2014-01-10 09:08:42 -0500 | [diff] [blame] | 594 | |
| 595 | At this point, we don't know all the possibilities. So mount a partial view |
| 596 | that we have. |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 597 | |
| 598 | This class doesn't hold isolate_dir, since it is dependent on the final |
| 599 | configuration selected. It is implicitly dependent on which .isolate defines |
| 600 | the 'command' that will take effect. |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 601 | """ |
Marc-Antoine Ruel | 67d3c0a | 2014-01-10 09:12:39 -0500 | [diff] [blame] | 602 | def __init__(self, file_comment, config_variables): |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 603 | self.file_comment = file_comment |
Marc-Antoine Ruel | 9ac1b91 | 2014-01-10 09:08:42 -0500 | [diff] [blame] | 604 | # Contains the names of the config variables seen while processing |
| 605 | # .isolate file(s). The order is important since the same order is used for |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 606 | # keys in self._by_config. |
Marc-Antoine Ruel | 67d3c0a | 2014-01-10 09:12:39 -0500 | [diff] [blame] | 607 | assert isinstance(config_variables, tuple) |
Marc-Antoine Ruel | 8170f49 | 2014-03-13 15:26:56 -0400 | [diff] [blame] | 608 | assert all(isinstance(c, basestring) for c in config_variables), ( |
| 609 | config_variables) |
Marc-Antoine Ruel | f0d0787 | 2014-03-27 16:59:03 -0400 | [diff] [blame] | 610 | config_variables = tuple(config_variables) |
| 611 | assert tuple(sorted(config_variables)) == config_variables, config_variables |
Marc-Antoine Ruel | 67d3c0a | 2014-01-10 09:12:39 -0500 | [diff] [blame] | 612 | self._config_variables = config_variables |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 613 | # The keys of _by_config are tuples of values for each of the items in |
Marc-Antoine Ruel | 67d3c0a | 2014-01-10 09:12:39 -0500 | [diff] [blame] | 614 | # self._config_variables. A None item in the list of the key means the value |
Marc-Antoine Ruel | 9ac1b91 | 2014-01-10 09:08:42 -0500 | [diff] [blame] | 615 | # is unbounded. |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 616 | self._by_config = {} |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 617 | |
Marc-Antoine Ruel | 67d3c0a | 2014-01-10 09:12:39 -0500 | [diff] [blame] | 618 | @property |
| 619 | def config_variables(self): |
| 620 | return self._config_variables |
| 621 | |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 622 | def get_config(self, config): |
| 623 | """Returns all configs that matches this config as a single ConfigSettings. |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 624 | """ |
Marc-Antoine Ruel | f0d0787 | 2014-03-27 16:59:03 -0400 | [diff] [blame] | 625 | # TODO(maruel): Fix ordering based on the bounded values. The keys are not |
| 626 | # necessarily sorted in the way that makes sense, they are alphabetically |
| 627 | # sorted. It is important because the left-most takes predescence. |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 628 | out = ConfigSettings({}, None) |
Marc-Antoine Ruel | f0d0787 | 2014-03-27 16:59:03 -0400 | [diff] [blame] | 629 | for k, v in sorted(self._by_config.iteritems()): |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 630 | if all(i == j or j is None for i, j in zip(config, k)): |
Marc-Antoine Ruel | 8170f49 | 2014-03-13 15:26:56 -0400 | [diff] [blame] | 631 | out = out.union(v) |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 632 | return out |
| 633 | |
Marc-Antoine Ruel | 8170f49 | 2014-03-13 15:26:56 -0400 | [diff] [blame] | 634 | def set_config(self, key, value): |
| 635 | """Sets the ConfigSettings for this key. |
| 636 | |
| 637 | The key is a tuple of bounded or unbounded variables. The global variable |
| 638 | is the key where all values are unbounded, e.g.: |
| 639 | (None,) * len(self._config_variables) |
| 640 | """ |
| 641 | assert key not in self._by_config, (key, self._by_config.keys()) |
| 642 | assert isinstance(key, tuple) |
| 643 | assert len(key) == len(self._config_variables), ( |
| 644 | key, self._config_variables) |
| 645 | assert isinstance(value, ConfigSettings) |
| 646 | self._by_config[key] = value |
| 647 | |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 648 | def union(self, rhs): |
Marc-Antoine Ruel | 8170f49 | 2014-03-13 15:26:56 -0400 | [diff] [blame] | 649 | """Returns a new Configs instance, the union of variables from self and rhs. |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 650 | |
Marc-Antoine Ruel | 8170f49 | 2014-03-13 15:26:56 -0400 | [diff] [blame] | 651 | Uses self.file_comment if available, otherwise rhs.file_comment. |
Marc-Antoine Ruel | f0d0787 | 2014-03-27 16:59:03 -0400 | [diff] [blame] | 652 | It keeps config_variables sorted in the output. |
Marc-Antoine Ruel | 8170f49 | 2014-03-13 15:26:56 -0400 | [diff] [blame] | 653 | """ |
| 654 | # Merge the keys of config_variables for each Configs instances. All the new |
| 655 | # variables will become unbounded. This requires realigning the keys. |
| 656 | config_variables = tuple(sorted( |
| 657 | set(self.config_variables) | set(rhs.config_variables))) |
| 658 | out = Configs(self.file_comment or rhs.file_comment, config_variables) |
| 659 | mapping_lhs = _get_map_keys(out.config_variables, self.config_variables) |
| 660 | mapping_rhs = _get_map_keys(out.config_variables, rhs.config_variables) |
| 661 | lhs_config = dict( |
| 662 | (_map_keys(mapping_lhs, k), v) for k, v in self._by_config.iteritems()) |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 663 | # pylint: disable=W0212 |
Marc-Antoine Ruel | 8170f49 | 2014-03-13 15:26:56 -0400 | [diff] [blame] | 664 | rhs_config = dict( |
| 665 | (_map_keys(mapping_rhs, k), v) for k, v in rhs._by_config.iteritems()) |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 666 | |
| 667 | for key in set(lhs_config) | set(rhs_config): |
Marc-Antoine Ruel | 8170f49 | 2014-03-13 15:26:56 -0400 | [diff] [blame] | 668 | out.set_config(key, union(lhs_config.get(key), rhs_config.get(key))) |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 669 | return out |
| 670 | |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 671 | def flatten(self): |
| 672 | """Returns a flat dictionary representation of the configuration. |
| 673 | """ |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 674 | return dict((k, v.flatten()) for k, v in self._by_config.iteritems()) |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 675 | |
| 676 | def make_isolate_file(self): |
| 677 | """Returns a dictionary suitable for writing to a .isolate file. |
| 678 | """ |
| 679 | dependencies_by_config = self.flatten() |
| 680 | configs_by_dependency = reduce_inputs(invert_map(dependencies_by_config)) |
| 681 | return convert_map_to_isolate_dict(configs_by_dependency, |
| 682 | self.config_variables) |
| 683 | |
| 684 | |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 685 | def load_isolate_as_config(isolate_dir, value, file_comment): |
| 686 | """Parses one .isolate file and returns a Configs() instance. |
| 687 | |
| 688 | Arguments: |
| 689 | isolate_dir: only used to load relative includes so it doesn't depend on |
| 690 | cwd. |
| 691 | value: is the loaded dictionary that was defined in the gyp file. |
| 692 | file_comment: comments found at the top of the file so it can be preserved. |
| 693 | |
| 694 | The expected format is strict, anything diverting from the format below will |
| 695 | throw an assert: |
| 696 | { |
| 697 | 'includes': [ |
| 698 | 'foo.isolate', |
| 699 | ], |
| 700 | 'conditions': [ |
| 701 | ['OS=="vms" and foo=42', { |
| 702 | 'variables': { |
| 703 | 'command': [ |
| 704 | ... |
| 705 | ], |
| 706 | 'isolate_dependency_tracked': [ |
| 707 | ... |
| 708 | ], |
| 709 | 'isolate_dependency_untracked': [ |
| 710 | ... |
| 711 | ], |
Marc-Antoine Ruel | 7124e39 | 2014-01-09 11:49:21 -0500 | [diff] [blame] | 712 | 'read_only': 0, |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 713 | }, |
| 714 | }], |
| 715 | ... |
| 716 | ], |
Marc-Antoine Ruel | d7c032b | 2014-03-13 15:32:16 -0400 | [diff] [blame] | 717 | 'variables': { |
| 718 | ... |
| 719 | }, |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 720 | } |
| 721 | """ |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 722 | assert os.path.isabs(isolate_dir), isolate_dir |
Marc-Antoine Ruel | d7c032b | 2014-03-13 15:32:16 -0400 | [diff] [blame] | 723 | if any(len(cond) == 3 for cond in value.get('conditions', [])): |
| 724 | raise isolateserver.ConfigError('Using \'else\' is not supported anymore.') |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 725 | variables_and_values = {} |
| 726 | verify_root(value, variables_and_values) |
| 727 | if variables_and_values: |
| 728 | config_variables, config_values = zip( |
| 729 | *sorted(variables_and_values.iteritems())) |
| 730 | all_configs = list(itertools.product(*config_values)) |
| 731 | else: |
Marc-Antoine Ruel | 9ac1b91 | 2014-01-10 09:08:42 -0500 | [diff] [blame] | 732 | config_variables = () |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 733 | all_configs = [] |
| 734 | |
Marc-Antoine Ruel | 67d3c0a | 2014-01-10 09:12:39 -0500 | [diff] [blame] | 735 | isolate = Configs(file_comment, config_variables) |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 736 | |
Marc-Antoine Ruel | d7c032b | 2014-03-13 15:32:16 -0400 | [diff] [blame] | 737 | # Add global variables. The global variables are on the empty tuple key. |
| 738 | isolate.set_config( |
| 739 | (None,) * len(config_variables), |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 740 | ConfigSettings(value.get('variables', {}), isolate_dir)) |
Marc-Antoine Ruel | d7c032b | 2014-03-13 15:32:16 -0400 | [diff] [blame] | 741 | |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 742 | # Add configuration-specific variables. |
| 743 | for expr, then in value.get('conditions', []): |
| 744 | configs = match_configs(expr, config_variables, all_configs) |
Marc-Antoine Ruel | 67d3c0a | 2014-01-10 09:12:39 -0500 | [diff] [blame] | 745 | new = Configs(None, config_variables) |
Marc-Antoine Ruel | 9ac1b91 | 2014-01-10 09:08:42 -0500 | [diff] [blame] | 746 | for config in configs: |
Marc-Antoine Ruel | b53d0c1 | 2014-03-28 13:46:27 -0400 | [diff] [blame^] | 747 | new.set_config(config, ConfigSettings(then['variables'], isolate_dir)) |
Marc-Antoine Ruel | 9ac1b91 | 2014-01-10 09:08:42 -0500 | [diff] [blame] | 748 | isolate = isolate.union(new) |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 749 | |
| 750 | # Load the includes. Process them in reverse so the last one take precedence. |
| 751 | for include in reversed(value.get('includes', [])): |
| 752 | if os.path.isabs(include): |
| 753 | raise isolateserver.ConfigError( |
| 754 | 'Failed to load configuration; absolute include path \'%s\'' % |
| 755 | include) |
| 756 | included_isolate = os.path.normpath(os.path.join(isolate_dir, include)) |
| 757 | with open(included_isolate, 'r') as f: |
| 758 | included_isolate = load_isolate_as_config( |
| 759 | os.path.dirname(included_isolate), |
| 760 | eval_content(f.read()), |
| 761 | None) |
| 762 | isolate = union(isolate, included_isolate) |
| 763 | |
| 764 | return isolate |
| 765 | |
| 766 | |
| 767 | def load_isolate_for_config(isolate_dir, content, config_variables): |
| 768 | """Loads the .isolate file and returns the information unprocessed but |
| 769 | filtered for the specific OS. |
| 770 | |
| 771 | Returns the command, dependencies and read_only flag. The dependencies are |
| 772 | fixed to use os.path.sep. |
| 773 | """ |
| 774 | # Load the .isolate file, process its conditions, retrieve the command and |
| 775 | # dependencies. |
| 776 | isolate = load_isolate_as_config(isolate_dir, eval_content(content), None) |
| 777 | try: |
| 778 | config_name = tuple( |
| 779 | config_variables[var] for var in isolate.config_variables) |
| 780 | except KeyError: |
| 781 | raise isolateserver.ConfigError( |
| 782 | 'These configuration variables were missing from the command line: %s' % |
| 783 | ', '.join( |
| 784 | sorted(set(isolate.config_variables) - set(config_variables)))) |
Marc-Antoine Ruel | 3ae9e6e | 2014-01-13 15:42:16 -0500 | [diff] [blame] | 785 | |
| 786 | # A configuration is to be created with all the combinations of free |
| 787 | # variables. |
| 788 | config = isolate.get_config(config_name) |
Marc-Antoine Ruel | a5a3622 | 2014-01-09 10:35:45 -0500 | [diff] [blame] | 789 | # Merge tracked and untracked variables, isolate.py doesn't care about the |
| 790 | # trackability of the variables, only the build tool does. |
| 791 | dependencies = [ |
| 792 | f.replace('/', os.path.sep) for f in config.tracked + config.untracked |
| 793 | ] |
| 794 | touched = [f.replace('/', os.path.sep) for f in config.touched] |
| 795 | return config.command, dependencies, touched, config.read_only |