1#!/usr/bin/env python 2# 3#===- exploded-graph-rewriter.py - ExplodedGraph dump tool -----*- python -*--# 4# 5# Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 6# See https://llvm.org/LICENSE.txt for license information. 7# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 8# 9#===-----------------------------------------------------------------------===# 10 11 12from __future__ import print_function 13 14import argparse 15import collections 16import difflib 17import json 18import logging 19import os 20import re 21 22 23#===-----------------------------------------------------------------------===# 24# These data structures represent a deserialized ExplodedGraph. 25#===-----------------------------------------------------------------------===# 26 27 28# A helper function for finding the difference between two dictionaries. 29def diff_dicts(curr, prev): 30 removed = [k for k in prev if k not in curr or curr[k] != prev[k]] 31 added = [k for k in curr if k not in prev or curr[k] != prev[k]] 32 return (removed, added) 33 34 35# Represents any program state trait that is a dictionary of key-value pairs. 36class GenericMap(object): 37 def __init__(self, items): 38 self.generic_map = collections.OrderedDict(items) 39 40 def diff(self, prev): 41 return diff_dicts(self.generic_map, prev.generic_map) 42 43 def is_different(self, prev): 44 removed, added = self.diff(prev) 45 return len(removed) != 0 or len(added) != 0 46 47 48# A deserialized source location. 49class SourceLocation(object): 50 def __init__(self, json_loc): 51 super(SourceLocation, self).__init__() 52 logging.debug('json: %s' % json_loc) 53 self.line = json_loc['line'] 54 self.col = json_loc['column'] 55 self.filename = os.path.basename(json_loc['file']) \ 56 if 'file' in json_loc else '(main file)' 57 self.spelling = SourceLocation(json_loc['spelling']) \ 58 if 'spelling' in json_loc else None 59 60 def is_macro(self): 61 return self.spelling is not None 62 63 64# A deserialized program point. 65class ProgramPoint(object): 66 def __init__(self, json_pp): 67 super(ProgramPoint, self).__init__() 68 self.kind = json_pp['kind'] 69 self.tag = json_pp['tag'] 70 self.node_id = json_pp['node_id'] 71 self.is_sink = bool(json_pp['is_sink']) 72 self.has_report = bool(json_pp['has_report']) 73 if self.kind == 'Edge': 74 self.src_id = json_pp['src_id'] 75 self.dst_id = json_pp['dst_id'] 76 elif self.kind == 'Statement': 77 logging.debug(json_pp) 78 self.stmt_kind = json_pp['stmt_kind'] 79 self.cast_kind = json_pp['cast_kind'] \ 80 if 'cast_kind' in json_pp else None 81 self.stmt_point_kind = json_pp['stmt_point_kind'] 82 self.stmt_id = json_pp['stmt_id'] 83 self.pointer = json_pp['pointer'] 84 self.pretty = json_pp['pretty'] 85 self.loc = SourceLocation(json_pp['location']) \ 86 if json_pp['location'] is not None else None 87 elif self.kind == 'BlockEntrance': 88 self.block_id = json_pp['block_id'] 89 90 91# A single expression acting as a key in a deserialized Environment. 92class EnvironmentBindingKey(object): 93 def __init__(self, json_ek): 94 super(EnvironmentBindingKey, self).__init__() 95 # CXXCtorInitializer is not a Stmt! 96 self.stmt_id = json_ek['stmt_id'] if 'stmt_id' in json_ek \ 97 else json_ek['init_id'] 98 self.pretty = json_ek['pretty'] 99 self.kind = json_ek['kind'] if 'kind' in json_ek else None 100 101 def _key(self): 102 return self.stmt_id 103 104 def __eq__(self, other): 105 return self._key() == other._key() 106 107 def __hash__(self): 108 return hash(self._key()) 109 110 111# Deserialized description of a location context. 112class LocationContext(object): 113 def __init__(self, json_frame): 114 super(LocationContext, self).__init__() 115 self.lctx_id = json_frame['lctx_id'] 116 self.caption = json_frame['location_context'] 117 self.decl = json_frame['calling'] 118 self.loc = SourceLocation(json_frame['location']) \ 119 if json_frame['location'] is not None else None 120 121 def _key(self): 122 return self.lctx_id 123 124 def __eq__(self, other): 125 return self._key() == other._key() 126 127 def __hash__(self): 128 return hash(self._key()) 129 130 131# A group of deserialized Environment bindings that correspond to a specific 132# location context. 133class EnvironmentFrame(object): 134 def __init__(self, json_frame): 135 super(EnvironmentFrame, self).__init__() 136 self.location_context = LocationContext(json_frame) 137 self.bindings = collections.OrderedDict( 138 [(EnvironmentBindingKey(b), 139 b['value']) for b in json_frame['items']] 140 if json_frame['items'] is not None else []) 141 142 def diff_bindings(self, prev): 143 return diff_dicts(self.bindings, prev.bindings) 144 145 def is_different(self, prev): 146 removed, added = self.diff_bindings(prev) 147 return len(removed) != 0 or len(added) != 0 148 149 150# A deserialized Environment. This class can also hold other entities that 151# are similar to Environment, such as Objects Under Construction. 152class GenericEnvironment(object): 153 def __init__(self, json_e): 154 super(GenericEnvironment, self).__init__() 155 self.frames = [EnvironmentFrame(f) for f in json_e] 156 157 def diff_frames(self, prev): 158 # TODO: It's difficult to display a good diff when frame numbers shift. 159 if len(self.frames) != len(prev.frames): 160 return None 161 162 updated = [] 163 for i in range(len(self.frames)): 164 f = self.frames[i] 165 prev_f = prev.frames[i] 166 if f.location_context == prev_f.location_context: 167 if f.is_different(prev_f): 168 updated.append(i) 169 else: 170 # We have the whole frame replaced with another frame. 171 # TODO: Produce a nice diff. 172 return None 173 174 # TODO: Add support for added/removed. 175 return updated 176 177 def is_different(self, prev): 178 updated = self.diff_frames(prev) 179 return updated is None or len(updated) > 0 180 181 182# A single binding key in a deserialized RegionStore cluster. 183class StoreBindingKey(object): 184 def __init__(self, json_sk): 185 super(StoreBindingKey, self).__init__() 186 self.kind = json_sk['kind'] 187 self.offset = json_sk['offset'] 188 189 def _key(self): 190 return (self.kind, self.offset) 191 192 def __eq__(self, other): 193 return self._key() == other._key() 194 195 def __hash__(self): 196 return hash(self._key()) 197 198 199# A single cluster of the deserialized RegionStore. 200class StoreCluster(object): 201 def __init__(self, json_sc): 202 super(StoreCluster, self).__init__() 203 self.base_region = json_sc['cluster'] 204 self.bindings = collections.OrderedDict( 205 [(StoreBindingKey(b), b['value']) for b in json_sc['items']]) 206 207 def diff_bindings(self, prev): 208 return diff_dicts(self.bindings, prev.bindings) 209 210 def is_different(self, prev): 211 removed, added = self.diff_bindings(prev) 212 return len(removed) != 0 or len(added) != 0 213 214 215# A deserialized RegionStore. 216class Store(object): 217 def __init__(self, json_s): 218 super(Store, self).__init__() 219 self.ptr = json_s['pointer'] 220 self.clusters = collections.OrderedDict( 221 [(c['pointer'], StoreCluster(c)) for c in json_s['items']]) 222 223 def diff_clusters(self, prev): 224 removed = [k for k in prev.clusters if k not in self.clusters] 225 added = [k for k in self.clusters if k not in prev.clusters] 226 updated = [k for k in prev.clusters if k in self.clusters 227 and prev.clusters[k].is_different(self.clusters[k])] 228 return (removed, added, updated) 229 230 def is_different(self, prev): 231 removed, added, updated = self.diff_clusters(prev) 232 return len(removed) != 0 or len(added) != 0 or len(updated) != 0 233 234 235# Deserialized messages from a single checker in a single program state. 236# Basically a list of raw strings. 237class CheckerLines(object): 238 def __init__(self, json_lines): 239 super(CheckerLines, self).__init__() 240 self.lines = json_lines 241 242 def diff_lines(self, prev): 243 lines = difflib.ndiff(prev.lines, self.lines) 244 return [l.strip() for l in lines 245 if l.startswith('+') or l.startswith('-')] 246 247 def is_different(self, prev): 248 return len(self.diff_lines(prev)) > 0 249 250 251# Deserialized messages of all checkers, separated by checker. 252class CheckerMessages(object): 253 def __init__(self, json_m): 254 super(CheckerMessages, self).__init__() 255 self.items = collections.OrderedDict( 256 [(m['checker'], CheckerLines(m['messages'])) for m in json_m]) 257 258 def diff_messages(self, prev): 259 removed = [k for k in prev.items if k not in self.items] 260 added = [k for k in self.items if k not in prev.items] 261 updated = [k for k in prev.items if k in self.items 262 and prev.items[k].is_different(self.items[k])] 263 return (removed, added, updated) 264 265 def is_different(self, prev): 266 removed, added, updated = self.diff_messages(prev) 267 return len(removed) != 0 or len(added) != 0 or len(updated) != 0 268 269 270# A deserialized program state. 271class ProgramState(object): 272 def __init__(self, state_id, json_ps): 273 super(ProgramState, self).__init__() 274 logging.debug('Adding ProgramState ' + str(state_id)) 275 276 self.state_id = state_id 277 278 self.store = Store(json_ps['store']) \ 279 if json_ps['store'] is not None else None 280 281 self.environment = \ 282 GenericEnvironment(json_ps['environment']['items']) \ 283 if json_ps['environment'] is not None else None 284 285 self.constraints = GenericMap([ 286 (c['symbol'], c['range']) for c in json_ps['constraints'] 287 ]) if json_ps['constraints'] is not None else None 288 289 self.dynamic_types = GenericMap([ 290 (t['region'], '%s%s' % (t['dyn_type'], 291 ' (or a sub-class)' 292 if t['sub_classable'] else '')) 293 for t in json_ps['dynamic_types']]) \ 294 if json_ps['dynamic_types'] is not None else None 295 296 self.constructing_objects = \ 297 GenericEnvironment(json_ps['constructing_objects']) \ 298 if json_ps['constructing_objects'] is not None else None 299 300 self.checker_messages = CheckerMessages(json_ps['checker_messages']) \ 301 if json_ps['checker_messages'] is not None else None 302 303 304# A deserialized exploded graph node. Has a default constructor because it 305# may be referenced as part of an edge before its contents are deserialized, 306# and in this moment we already need a room for predecessors and successors. 307class ExplodedNode(object): 308 def __init__(self): 309 super(ExplodedNode, self).__init__() 310 self.predecessors = [] 311 self.successors = [] 312 313 def construct(self, node_id, json_node): 314 logging.debug('Adding ' + node_id) 315 self.ptr = node_id[4:] 316 self.points = [ProgramPoint(p) for p in json_node['program_points']] 317 self.node_id = self.points[-1].node_id 318 self.state = ProgramState(json_node['state_id'], 319 json_node['program_state']) \ 320 if json_node['program_state'] is not None else None 321 322 assert self.node_name() == node_id 323 324 def node_name(self): 325 return 'Node' + self.ptr 326 327 328# A deserialized ExplodedGraph. Constructed by consuming a .dot file 329# line-by-line. 330class ExplodedGraph(object): 331 # Parse .dot files with regular expressions. 332 node_re = re.compile( 333 '^(Node0x[0-9a-f]*) \\[shape=record,.*label="{(.*)\\\\l}"\\];$') 334 edge_re = re.compile( 335 '^(Node0x[0-9a-f]*) -> (Node0x[0-9a-f]*);$') 336 337 def __init__(self): 338 super(ExplodedGraph, self).__init__() 339 self.nodes = collections.defaultdict(ExplodedNode) 340 self.root_id = None 341 self.incomplete_line = '' 342 343 def add_raw_line(self, raw_line): 344 if raw_line.startswith('//'): 345 return 346 347 # Allow line breaks by waiting for ';'. This is not valid in 348 # a .dot file, but it is useful for writing tests. 349 if len(raw_line) > 0 and raw_line[-1] != ';': 350 self.incomplete_line += raw_line 351 return 352 raw_line = self.incomplete_line + raw_line 353 self.incomplete_line = '' 354 355 # Apply regexps one by one to see if it's a node or an edge 356 # and extract contents if necessary. 357 logging.debug('Line: ' + raw_line) 358 result = self.edge_re.match(raw_line) 359 if result is not None: 360 logging.debug('Classified as edge line.') 361 pred = result.group(1) 362 succ = result.group(2) 363 self.nodes[pred].successors.append(succ) 364 self.nodes[succ].predecessors.append(pred) 365 return 366 result = self.node_re.match(raw_line) 367 if result is not None: 368 logging.debug('Classified as node line.') 369 node_id = result.group(1) 370 if len(self.nodes) == 0: 371 self.root_id = node_id 372 # Note: when writing tests you don't need to escape everything, 373 # even though in a valid dot file everything is escaped. 374 node_label = result.group(2).replace('\\l', '') \ 375 .replace(' ', '') \ 376 .replace('\\"', '"') \ 377 .replace('\\{', '{') \ 378 .replace('\\}', '}') \ 379 .replace('\\\\', '\\') \ 380 .replace('\\|', '|') \ 381 .replace('\\<', '\\\\<') \ 382 .replace('\\>', '\\\\>') \ 383 .rstrip(',') 384 logging.debug(node_label) 385 json_node = json.loads(node_label) 386 self.nodes[node_id].construct(node_id, json_node) 387 return 388 logging.debug('Skipping.') 389 390 391#===-----------------------------------------------------------------------===# 392# Visitors traverse a deserialized ExplodedGraph and do different things 393# with every node and edge. 394#===-----------------------------------------------------------------------===# 395 396 397# A visitor that dumps the ExplodedGraph into a DOT file with fancy HTML-based 398# syntax highlighing. 399class DotDumpVisitor(object): 400 def __init__(self, do_diffs, dark_mode, gray_mode, 401 topo_mode, dump_dot_only): 402 super(DotDumpVisitor, self).__init__() 403 self._do_diffs = do_diffs 404 self._dark_mode = dark_mode 405 self._gray_mode = gray_mode 406 self._topo_mode = topo_mode 407 self._dump_dot_only = dump_dot_only 408 self._output = [] 409 410 def _dump_raw(self, s): 411 if self._dump_dot_only: 412 print(s, end='') 413 else: 414 self._output.append(s) 415 416 def output(self): 417 assert not self._dump_dot_only 418 return ''.join(self._output) 419 420 def _dump(self, s): 421 s = s.replace('&', '&') \ 422 .replace('{', '\\{') \ 423 .replace('}', '\\}') \ 424 .replace('\\<', '<') \ 425 .replace('\\>', '>') \ 426 .replace('\\l', '<br />') \ 427 .replace('|', '\\|') 428 if self._gray_mode: 429 s = re.sub(r'<font color="[a-z0-9]*">', '', s) 430 s = re.sub(r'</font>', '', s) 431 self._dump_raw(s) 432 433 @staticmethod 434 def _diff_plus_minus(is_added): 435 if is_added is None: 436 return '' 437 if is_added: 438 return '<font color="forestgreen">+</font>' 439 return '<font color="red">-</font>' 440 441 @staticmethod 442 def _short_pretty(s): 443 if s is None: 444 return None 445 if len(s) < 20: 446 return s 447 left = s.find('{') 448 right = s.rfind('}') 449 if left == -1 or right == -1 or left >= right: 450 return s 451 candidate = s[0:left + 1] + ' ... ' + s[right:] 452 if len(candidate) >= len(s): 453 return s 454 return candidate 455 456 @staticmethod 457 def _make_sloc(loc): 458 if loc is None: 459 return '<i>Invalid Source Location</i>' 460 461 def make_plain_loc(loc): 462 return '%s:<b>%s</b>:<b>%s</b>' \ 463 % (loc.filename, loc.line, loc.col) 464 465 if loc.is_macro(): 466 return '%s <font color="royalblue1">' \ 467 '(<i>spelling at </i> %s)</font>' \ 468 % (make_plain_loc(loc), make_plain_loc(loc.spelling)) 469 470 return make_plain_loc(loc) 471 472 def visit_begin_graph(self, graph): 473 self._graph = graph 474 self._dump_raw('digraph "ExplodedGraph" {\n') 475 if self._dark_mode: 476 self._dump_raw('bgcolor="gray10";\n') 477 self._dump_raw('label="";\n') 478 479 def visit_program_point(self, p): 480 if p.kind in ['Edge', 'BlockEntrance', 'BlockExit']: 481 color = 'gold3' 482 elif p.kind in ['PreStmtPurgeDeadSymbols', 483 'PostStmtPurgeDeadSymbols']: 484 color = 'red' 485 elif p.kind in ['CallEnter', 'CallExitBegin', 'CallExitEnd']: 486 color = 'dodgerblue' if self._dark_mode else 'blue' 487 elif p.kind in ['Statement']: 488 color = 'cyan4' 489 else: 490 color = 'forestgreen' 491 492 self._dump('<tr><td align="left">%s.</td>' % p.node_id) 493 494 if p.kind == 'Statement': 495 # This avoids pretty-printing huge statements such as CompoundStmt. 496 # Such statements show up only at [Pre|Post]StmtPurgeDeadSymbols 497 skip_pretty = 'PurgeDeadSymbols' in p.stmt_point_kind 498 stmt_color = 'cyan3' 499 self._dump('<td align="left" width="0">%s:</td>' 500 '<td align="left" width="0"><font color="%s">' 501 '%s</font> </td>' 502 '<td align="left"><i>S%s</i></td>' 503 '<td align="left"><font color="%s">%s</font></td>' 504 '<td align="left">%s</td></tr>' 505 % (self._make_sloc(p.loc), color, 506 '%s (%s)' % (p.stmt_kind, p.cast_kind) 507 if p.cast_kind is not None else p.stmt_kind, 508 p.stmt_id, stmt_color, p.stmt_point_kind, 509 self._short_pretty(p.pretty) 510 if not skip_pretty else '')) 511 elif p.kind == 'Edge': 512 self._dump('<td width="0"></td>' 513 '<td align="left" width="0">' 514 '<font color="%s">%s</font></td><td align="left">' 515 '[B%d] -\\> [B%d]</td></tr>' 516 % (color, 'BlockEdge', p.src_id, p.dst_id)) 517 elif p.kind == 'BlockEntrance': 518 self._dump('<td width="0"></td>' 519 '<td align="left" width="0">' 520 '<font color="%s">%s</font></td>' 521 '<td align="left">[B%d]</td></tr>' 522 % (color, p.kind, p.block_id)) 523 else: 524 # TODO: Print more stuff for other kinds of points. 525 self._dump('<td width="0"></td>' 526 '<td align="left" width="0" colspan="2">' 527 '<font color="%s">%s</font></td></tr>' 528 % (color, p.kind)) 529 530 if p.tag is not None: 531 self._dump('<tr><td width="0"></td><td width="0"></td>' 532 '<td colspan="3" align="left">' 533 '<b>Tag: </b> <font color="crimson">' 534 '%s</font></td></tr>' % p.tag) 535 536 if p.has_report: 537 self._dump('<tr><td width="0"></td><td width="0"></td>' 538 '<td colspan="3" align="left">' 539 '<font color="red"><b>Bug Report Attached' 540 '</b></font></td></tr>') 541 if p.is_sink: 542 self._dump('<tr><td width="0"></td><td width="0"></td>' 543 '<td colspan="3" align="left">' 544 '<font color="cornflowerblue"><b>Sink Node' 545 '</b></font></td></tr>') 546 547 def visit_environment(self, e, prev_e=None): 548 self._dump('<table border="0">') 549 550 def dump_location_context(lc, is_added=None): 551 self._dump('<tr><td>%s</td>' 552 '<td align="left"><b>%s</b></td>' 553 '<td align="left" colspan="2">' 554 '<font color="gray60">%s </font>' 555 '%s</td></tr>' 556 % (self._diff_plus_minus(is_added), 557 lc.caption, lc.decl, 558 ('(%s)' % self._make_sloc(lc.loc)) 559 if lc.loc is not None else '')) 560 561 def dump_binding(f, b, is_added=None): 562 self._dump('<tr><td>%s</td>' 563 '<td align="left"><i>S%s</i></td>' 564 '%s' 565 '<td align="left">%s</td>' 566 '<td align="left">%s</td></tr>' 567 % (self._diff_plus_minus(is_added), 568 b.stmt_id, 569 '<td align="left"><font color="%s"><i>' 570 '%s</i></font></td>' % ( 571 'lavender' if self._dark_mode else 'darkgreen', 572 ('(%s)' % b.kind) if b.kind is not None else ' ' 573 ), 574 self._short_pretty(b.pretty), f.bindings[b])) 575 576 frames_updated = e.diff_frames(prev_e) if prev_e is not None else None 577 if frames_updated: 578 for i in frames_updated: 579 f = e.frames[i] 580 prev_f = prev_e.frames[i] 581 dump_location_context(f.location_context) 582 bindings_removed, bindings_added = f.diff_bindings(prev_f) 583 for b in bindings_removed: 584 dump_binding(prev_f, b, False) 585 for b in bindings_added: 586 dump_binding(f, b, True) 587 else: 588 for f in e.frames: 589 dump_location_context(f.location_context) 590 for b in f.bindings: 591 dump_binding(f, b) 592 593 self._dump('</table>') 594 595 def visit_environment_in_state(self, selector, title, s, prev_s=None): 596 e = getattr(s, selector) 597 prev_e = getattr(prev_s, selector) if prev_s is not None else None 598 if e is None and prev_e is None: 599 return 600 601 self._dump('<hr /><tr><td align="left"><b>%s: </b>' % title) 602 if e is None: 603 self._dump('<i> Nothing!</i>') 604 else: 605 if prev_e is not None: 606 if e.is_different(prev_e): 607 self._dump('</td></tr><tr><td align="left">') 608 self.visit_environment(e, prev_e) 609 else: 610 self._dump('<i> No changes!</i>') 611 else: 612 self._dump('</td></tr><tr><td align="left">') 613 self.visit_environment(e) 614 615 self._dump('</td></tr>') 616 617 def visit_store(self, s, prev_s=None): 618 self._dump('<table border="0">') 619 620 def dump_binding(s, c, b, is_added=None): 621 self._dump('<tr><td>%s</td>' 622 '<td align="left">%s</td>' 623 '<td align="left">%s</td>' 624 '<td align="left">%s</td>' 625 '<td align="left">%s</td></tr>' 626 % (self._diff_plus_minus(is_added), 627 s.clusters[c].base_region, b.offset, 628 '(<i>Default</i>)' if b.kind == 'Default' 629 else '', 630 s.clusters[c].bindings[b])) 631 632 if prev_s is not None: 633 clusters_removed, clusters_added, clusters_updated = \ 634 s.diff_clusters(prev_s) 635 for c in clusters_removed: 636 for b in prev_s.clusters[c].bindings: 637 dump_binding(prev_s, c, b, False) 638 for c in clusters_updated: 639 bindings_removed, bindings_added = \ 640 s.clusters[c].diff_bindings(prev_s.clusters[c]) 641 for b in bindings_removed: 642 dump_binding(prev_s, c, b, False) 643 for b in bindings_added: 644 dump_binding(s, c, b, True) 645 for c in clusters_added: 646 for b in s.clusters[c].bindings: 647 dump_binding(s, c, b, True) 648 else: 649 for c in s.clusters: 650 for b in s.clusters[c].bindings: 651 dump_binding(s, c, b) 652 653 self._dump('</table>') 654 655 def visit_store_in_state(self, s, prev_s=None): 656 st = s.store 657 prev_st = prev_s.store if prev_s is not None else None 658 if st is None and prev_st is None: 659 return 660 661 self._dump('<hr /><tr><td align="left"><b>Store: </b>') 662 if st is None: 663 self._dump('<i> Nothing!</i>') 664 else: 665 if self._dark_mode: 666 self._dump(' <font color="gray30">(%s)</font>' % st.ptr) 667 else: 668 self._dump(' <font color="gray">(%s)</font>' % st.ptr) 669 if prev_st is not None: 670 if s.store.is_different(prev_st): 671 self._dump('</td></tr><tr><td align="left">') 672 self.visit_store(st, prev_st) 673 else: 674 self._dump('<i> No changes!</i>') 675 else: 676 self._dump('</td></tr><tr><td align="left">') 677 self.visit_store(st) 678 self._dump('</td></tr>') 679 680 def visit_generic_map(self, m, prev_m=None): 681 self._dump('<table border="0">') 682 683 def dump_pair(m, k, is_added=None): 684 self._dump('<tr><td>%s</td>' 685 '<td align="left">%s</td>' 686 '<td align="left">%s</td></tr>' 687 % (self._diff_plus_minus(is_added), 688 k, m.generic_map[k])) 689 690 if prev_m is not None: 691 removed, added = m.diff(prev_m) 692 for k in removed: 693 dump_pair(prev_m, k, False) 694 for k in added: 695 dump_pair(m, k, True) 696 else: 697 for k in m.generic_map: 698 dump_pair(m, k, None) 699 700 self._dump('</table>') 701 702 def visit_generic_map_in_state(self, selector, title, s, prev_s=None): 703 m = getattr(s, selector) 704 prev_m = getattr(prev_s, selector) if prev_s is not None else None 705 if m is None and prev_m is None: 706 return 707 708 self._dump('<hr />') 709 self._dump('<tr><td align="left">' 710 '<b>%s: </b>' % title) 711 if m is None: 712 self._dump('<i> Nothing!</i>') 713 else: 714 if prev_m is not None: 715 if m.is_different(prev_m): 716 self._dump('</td></tr><tr><td align="left">') 717 self.visit_generic_map(m, prev_m) 718 else: 719 self._dump('<i> No changes!</i>') 720 else: 721 self._dump('</td></tr><tr><td align="left">') 722 self.visit_generic_map(m) 723 724 self._dump('</td></tr>') 725 726 def visit_checker_messages(self, m, prev_m=None): 727 self._dump('<table border="0">') 728 729 def dump_line(l, is_added=None): 730 self._dump('<tr><td>%s</td>' 731 '<td align="left">%s</td></tr>' 732 % (self._diff_plus_minus(is_added), l)) 733 734 def dump_chk(chk, is_added=None): 735 dump_line('<i>%s</i>:' % chk, is_added) 736 737 if prev_m is not None: 738 removed, added, updated = m.diff_messages(prev_m) 739 for chk in removed: 740 dump_chk(chk, False) 741 for l in prev_m.items[chk].lines: 742 dump_line(l, False) 743 for chk in updated: 744 dump_chk(chk) 745 for l in m.items[chk].diff_lines(prev_m.items[chk]): 746 dump_line(l[1:], l.startswith('+')) 747 for chk in added: 748 dump_chk(chk, True) 749 for l in m.items[chk].lines: 750 dump_line(l, True) 751 else: 752 for chk in m.items: 753 dump_chk(chk) 754 for l in m.items[chk].lines: 755 dump_line(l) 756 757 self._dump('</table>') 758 759 def visit_checker_messages_in_state(self, s, prev_s=None): 760 m = s.checker_messages 761 prev_m = prev_s.checker_messages if prev_s is not None else None 762 if m is None and prev_m is None: 763 return 764 765 self._dump('<hr />') 766 self._dump('<tr><td align="left">' 767 '<b>Checker State: </b>') 768 if m is None: 769 self._dump('<i> Nothing!</i>') 770 else: 771 if prev_m is not None: 772 if m.is_different(prev_m): 773 self._dump('</td></tr><tr><td align="left">') 774 self.visit_checker_messages(m, prev_m) 775 else: 776 self._dump('<i> No changes!</i>') 777 else: 778 self._dump('</td></tr><tr><td align="left">') 779 self.visit_checker_messages(m) 780 781 self._dump('</td></tr>') 782 783 def visit_state(self, s, prev_s): 784 self.visit_store_in_state(s, prev_s) 785 self.visit_environment_in_state('environment', 'Environment', 786 s, prev_s) 787 self.visit_generic_map_in_state('constraints', 'Ranges', 788 s, prev_s) 789 self.visit_generic_map_in_state('dynamic_types', 'Dynamic Types', 790 s, prev_s) 791 self.visit_environment_in_state('constructing_objects', 792 'Objects Under Construction', 793 s, prev_s) 794 self.visit_checker_messages_in_state(s, prev_s) 795 796 def visit_node(self, node): 797 self._dump('%s [shape=record,' 798 % (node.node_name())) 799 if self._dark_mode: 800 self._dump('color="white",fontcolor="gray80",') 801 self._dump('label=<<table border="0">') 802 803 self._dump('<tr><td bgcolor="%s"><b>State %s</b></td></tr>' 804 % ("gray20" if self._dark_mode else "gray70", 805 node.state.state_id 806 if node.state is not None else 'Unspecified')) 807 if not self._topo_mode: 808 self._dump('<tr><td align="left" width="0">') 809 if len(node.points) > 1: 810 self._dump('<b>Program points:</b></td></tr>') 811 else: 812 self._dump('<b>Program point:</b></td></tr>') 813 self._dump('<tr><td align="left" width="0">' 814 '<table border="0" align="left" width="0">') 815 for p in node.points: 816 self.visit_program_point(p) 817 self._dump('</table></td></tr>') 818 819 if node.state is not None and not self._topo_mode: 820 prev_s = None 821 # Do diffs only when we have a unique predecessor. 822 # Don't do diffs on the leaf nodes because they're 823 # the important ones. 824 if self._do_diffs and len(node.predecessors) == 1 \ 825 and len(node.successors) > 0: 826 prev_s = self._graph.nodes[node.predecessors[0]].state 827 self.visit_state(node.state, prev_s) 828 self._dump_raw('</table>>];\n') 829 830 def visit_edge(self, pred, succ): 831 self._dump_raw('%s -> %s%s;\n' % ( 832 pred.node_name(), succ.node_name(), 833 ' [color="white"]' if self._dark_mode else '' 834 )) 835 836 def visit_end_of_graph(self): 837 self._dump_raw('}\n') 838 839 if not self._dump_dot_only: 840 import sys 841 import tempfile 842 843 def write_temp_file(suffix, data): 844 fd, filename = tempfile.mkstemp(suffix=suffix) 845 print('Writing "%s"...' % filename) 846 with os.fdopen(fd, 'w') as fp: 847 fp.write(data) 848 print('Done! Please remember to remove the file.') 849 return filename 850 851 try: 852 import graphviz 853 except ImportError: 854 # The fallback behavior if graphviz is not installed! 855 print('Python graphviz not found. Please invoke') 856 print(' $ pip install graphviz') 857 print('in order to enable automatic conversion to HTML.') 858 print() 859 print('You may also convert DOT to SVG manually via') 860 print(' $ dot -Tsvg input.dot -o output.svg') 861 print() 862 write_temp_file('.dot', self.output()) 863 return 864 865 svg = graphviz.pipe('dot', 'svg', self.output()) 866 867 filename = write_temp_file( 868 '.html', '<html><body bgcolor="%s">%s</body></html>' % ( 869 '#1a1a1a' if self._dark_mode else 'white', svg)) 870 if sys.platform == 'win32': 871 os.startfile(filename) 872 elif sys.platform == 'darwin': 873 os.system('open "%s"' % filename) 874 else: 875 os.system('xdg-open "%s"' % filename) 876 877 878#===-----------------------------------------------------------------------===# 879# Explorers know how to traverse the ExplodedGraph in a certain order. 880# They would invoke a Visitor on every node or edge they encounter. 881#===-----------------------------------------------------------------------===# 882 883 884# BasicExplorer explores the whole graph in no particular order. 885class BasicExplorer(object): 886 def __init__(self): 887 super(BasicExplorer, self).__init__() 888 889 def explore(self, graph, visitor): 890 visitor.visit_begin_graph(graph) 891 for node in sorted(graph.nodes): 892 logging.debug('Visiting ' + node) 893 visitor.visit_node(graph.nodes[node]) 894 for succ in sorted(graph.nodes[node].successors): 895 logging.debug('Visiting edge: %s -> %s ' % (node, succ)) 896 visitor.visit_edge(graph.nodes[node], graph.nodes[succ]) 897 visitor.visit_end_of_graph() 898 899 900#===-----------------------------------------------------------------------===# 901# Trimmers cut out parts of the ExplodedGraph so that to focus on other parts. 902# Trimmers can be combined together by applying them sequentially. 903#===-----------------------------------------------------------------------===# 904 905 906# SinglePathTrimmer keeps only a single path - the leftmost path from the root. 907# Useful when the trimmed graph is still too large. 908class SinglePathTrimmer(object): 909 def __init__(self): 910 super(SinglePathTrimmer, self).__init__() 911 912 def trim(self, graph): 913 visited_nodes = set() 914 node_id = graph.root_id 915 while True: 916 visited_nodes.add(node_id) 917 node = graph.nodes[node_id] 918 if len(node.successors) > 0: 919 succ_id = node.successors[0] 920 succ = graph.nodes[succ_id] 921 node.successors = [succ_id] 922 succ.predecessors = [node_id] 923 if succ_id in visited_nodes: 924 break 925 node_id = succ_id 926 else: 927 break 928 graph.nodes = {node_id: graph.nodes[node_id] 929 for node_id in visited_nodes} 930 931 932# TargetedTrimmer keeps paths that lead to specific nodes and discards all 933# other paths. Useful when you cannot use -trim-egraph (e.g. when debugging 934# a crash). 935class TargetedTrimmer(object): 936 def __init__(self, target_nodes): 937 super(TargetedTrimmer, self).__init__() 938 self._target_nodes = target_nodes 939 940 @staticmethod 941 def parse_target_node(node, graph): 942 if node.startswith('0x'): 943 ret = 'Node' + node 944 assert ret in graph.nodes 945 return ret 946 else: 947 for other_id in graph.nodes: 948 other = graph.nodes[other_id] 949 if other.node_id == int(node): 950 return other_id 951 952 @staticmethod 953 def parse_target_nodes(target_nodes, graph): 954 return [TargetedTrimmer.parse_target_node(node, graph) 955 for node in target_nodes.split(',')] 956 957 def trim(self, graph): 958 queue = self._target_nodes 959 visited_nodes = set() 960 961 while len(queue) > 0: 962 node_id = queue.pop() 963 visited_nodes.add(node_id) 964 node = graph.nodes[node_id] 965 for pred_id in node.predecessors: 966 if pred_id not in visited_nodes: 967 queue.append(pred_id) 968 graph.nodes = {node_id: graph.nodes[node_id] 969 for node_id in visited_nodes} 970 for node_id in graph.nodes: 971 node = graph.nodes[node_id] 972 node.successors = [succ_id for succ_id in node.successors 973 if succ_id in visited_nodes] 974 node.predecessors = [succ_id for succ_id in node.predecessors 975 if succ_id in visited_nodes] 976 977 978#===-----------------------------------------------------------------------===# 979# The entry point to the script. 980#===-----------------------------------------------------------------------===# 981 982 983def main(): 984 parser = argparse.ArgumentParser( 985 description='Display and manipulate Exploded Graph dumps.') 986 parser.add_argument('filename', type=str, 987 help='the .dot file produced by the Static Analyzer') 988 parser.add_argument('-v', '--verbose', action='store_const', 989 dest='loglevel', const=logging.DEBUG, 990 default=logging.WARNING, 991 help='enable info prints') 992 parser.add_argument('-d', '--diff', action='store_const', dest='diff', 993 const=True, default=False, 994 help='display differences between states') 995 parser.add_argument('-t', '--topology', action='store_const', 996 dest='topology', const=True, default=False, 997 help='only display program points, omit states') 998 parser.add_argument('-s', '--single-path', action='store_const', 999 dest='single_path', const=True, default=False, 1000 help='only display the leftmost path in the graph ' 1001 '(useful for trimmed graphs that still ' 1002 'branch too much)') 1003 parser.add_argument('--to', type=str, default=None, 1004 help='only display execution paths from the root ' 1005 'to the given comma-separated list of nodes ' 1006 'identified by a pointer or a stable ID; ' 1007 'compatible with --single-path') 1008 parser.add_argument('--dark', action='store_const', dest='dark', 1009 const=True, default=False, 1010 help='dark mode') 1011 parser.add_argument('--gray', action='store_const', dest='gray', 1012 const=True, default=False, 1013 help='black-and-white mode') 1014 parser.add_argument('--dump-dot-only', action='store_const', 1015 dest='dump_dot_only', const=True, default=False, 1016 help='instead of writing an HTML file and immediately ' 1017 'displaying it, dump the rewritten dot file ' 1018 'to stdout') 1019 args = parser.parse_args() 1020 logging.basicConfig(level=args.loglevel) 1021 1022 graph = ExplodedGraph() 1023 with open(args.filename) as fd: 1024 for raw_line in fd: 1025 raw_line = raw_line.strip() 1026 graph.add_raw_line(raw_line) 1027 1028 trimmers = [] 1029 if args.to is not None: 1030 trimmers.append(TargetedTrimmer( 1031 TargetedTrimmer.parse_target_nodes(args.to, graph))) 1032 if args.single_path: 1033 trimmers.append(SinglePathTrimmer()) 1034 1035 explorer = BasicExplorer() 1036 1037 visitor = DotDumpVisitor(args.diff, args.dark, args.gray, args.topology, 1038 args.dump_dot_only) 1039 1040 for trimmer in trimmers: 1041 trimmer.trim(graph) 1042 1043 explorer.explore(graph, visitor) 1044 1045 1046if __name__ == '__main__': 1047 main() 1048