gcc/diagnostic-format-sarif.cc

   1 /* SARIF output for diagnostics
   2    Copyright (C) 2018-2023 Free Software Foundation, Inc.
   3    Contributed by David Malcolm <dmalcolm@redhat.com>.
   4
   5 This file is part of GCC.
   6
   7 GCC is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 3, or (at your option) any later
  10 version.
  11
  12 GCC is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with GCC; see the file COPYING3.  If not see
  19 <http://www.gnu.org/licenses/>.  */
  20
  21
  22 #include "config.h"
  23 #include "system.h"
  24 #include "coretypes.h"
  25 #include "diagnostic.h"
  26 #include "diagnostic-metadata.h"
  27 #include "diagnostic-path.h"
  28 #include "json.h"
  29 #include "cpplib.h"
  30 #include "logical-location.h"
  31 #include "diagnostic-client-data-hooks.h"
  32
  33 class sarif_builder;
  34
  35 /* Subclass of json::object for SARIF invocation objects
  36    (SARIF v2.1.0 section 3.20).  */
  37
  38 class sarif_invocation : public json::object
  39 {
  40 public:
  41   sarif_invocation ()
  42   : m_notifications_arr (new json::array ()),
  43     m_success (true)
  44   {}
  45
  46   void add_notification_for_ice (diagnostic_context *context,
  47                                  diagnostic_info *diagnostic,
  48                                  sarif_builder *builder);
  49   void prepare_to_flush ();
  50
  51 private:
  52   json::array *m_notifications_arr;
  53   bool m_success;
  54 };
  55
  56 /* Subclass of json::object for SARIF result objects
  57    (SARIF v2.1.0 section 3.27).  */
  58
  59 class sarif_result : public json::object
  60 {
  61 public:
  62   sarif_result () : m_related_locations_arr (NULL) {}
  63
  64   void
  65   on_nested_diagnostic (diagnostic_context *context,
  66                         diagnostic_info *diagnostic,
  67                         diagnostic_t orig_diag_kind,
  68                         sarif_builder *builder);
  69
  70 private:
  71   json::array *m_related_locations_arr;
  72 };
  73
  74 /* Subclass of json::object for SARIF notification objects
  75    (SARIF v2.1.0 section 3.58).
  76
  77    This subclass is specifically for notifying when an
  78    internal compiler error occurs.  */
  79
  80 class sarif_ice_notification : public json::object
  81 {
  82 public:
  83   sarif_ice_notification (diagnostic_context *context,
  84                           diagnostic_info *diagnostic,
  85                           sarif_builder *builder);
  86 };
  87
  88 /* A class for managing SARIF output (for -fdiagnostics-format=sarif-stderr
  89    and -fdiagnostics-format=sarif-file).
  90
  91    As diagnostics occur, we build "result" JSON objects, and
  92    accumulate state:
  93    - which source files are referenced
  94    - which warnings are emitted
  95    - which CWEs are used
  96
  97    At the end of the compile, we use the above to build the full SARIF
  98    object tree, adding the result objects to the correct place, and
  99    creating objects for the various source files, warnings and CWEs
 100    referenced.
 101
 102    Implemented:
 103    - fix-it hints
 104    - CWE metadata
 105    - diagnostic groups (see limitations below)
 106    - logical locations (e.g. cfun)
 107
 108    Known limitations:
 109    - GCC supports one-deep nesting of diagnostics (via auto_diagnostic_group),
 110      but we only capture location and message information from such nested
 111      diagnostics (e.g. we ignore fix-it hints on them)
 112    - doesn't yet capture command-line arguments: would be run.invocations
 113      property (SARIF v2.1.0 section 3.14.11), as invocation objects
 114      (SARIF v2.1.0 section 3.20), but we'd want to capture the arguments to
 115      toplev::main, and the response files.
 116    - doesn't capture escape_on_output_p
 117    - doesn't capture secondary locations within a rich_location
 118      (perhaps we should use the "relatedLocations" property: SARIF v2.1.0
 119      section 3.27.22)
 120    - doesn't capture "artifact.encoding" property
 121      (SARIF v2.1.0 section 3.24.9).
 122    - doesn't capture hashes of the source files
 123      ("artifact.hashes" property (SARIF v2.1.0 section 3.24.11).
 124    - doesn't capture the "analysisTarget" property
 125      (SARIF v2.1.0 section 3.27.13).
 126    - doesn't capture labelled ranges
 127    - doesn't capture -Werror cleanly
 128    - doesn't capture inlining information (can SARIF handle this?)
 129    - doesn't capture macro expansion information (can SARIF handle this?).  */
 130
 131 class sarif_builder
 132 {
 133 public:
 134   sarif_builder (diagnostic_context *context);
 135
 136   void end_diagnostic (diagnostic_context *context, diagnostic_info *diagnostic,
 137                        diagnostic_t orig_diag_kind);
 138
 139   void end_group ();
 140
 141   void flush_to_file (FILE *outf);
 142
 143   json::array *make_locations_arr (diagnostic_info *diagnostic);
 144   json::object *make_location_object (const rich_location &rich_loc,
 145                                       const logical_location *logical_loc);
 146   json::object *make_message_object (const char *msg) const;
 147
 148 private:
 149   sarif_result *make_result_object (diagnostic_context *context,
 150                                     diagnostic_info *diagnostic,
 151                                     diagnostic_t orig_diag_kind);
 152   void set_any_logical_locs_arr (json::object *location_obj,
 153                                  const logical_location *logical_loc);
 154   json::object *make_location_object (const diagnostic_event &event);
 155   json::object *
 156   make_logical_location_object (const logical_location &logical_loc) const;
 157   json::object *make_code_flow_object (const diagnostic_path &path);
 158   json::object *make_thread_flow_object (const diagnostic_path &path);
 159   json::object *
 160   make_thread_flow_location_object (const diagnostic_event &event);
 161   json::array *maybe_make_kinds_array (diagnostic_event::meaning m) const;
 162   json::object *maybe_make_physical_location_object (location_t loc);
 163   json::object *make_artifact_location_object (location_t loc);
 164   json::object *make_artifact_location_object (const char *filename);
 165   json::object *make_artifact_location_object_for_pwd () const;
 166   json::object *maybe_make_region_object (location_t loc) const;
 167   json::object *maybe_make_region_object_for_context (location_t loc) const;
 168   json::object *make_region_object_for_hint (const fixit_hint &hint) const;
 169   json::object *make_multiformat_message_string (const char *msg) const;
 170   json::object *make_top_level_object (sarif_invocation *invocation_obj,
 171                                        json::array *results);
 172   json::object *make_run_object (sarif_invocation *invocation_obj,
 173                                  json::array *results);
 174   json::object *make_tool_object () const;
 175   json::object *make_driver_tool_component_object () const;
 176   json::array *maybe_make_taxonomies_array () const;
 177   json::object *maybe_make_cwe_taxonomy_object () const;
 178   json::object *make_tool_component_reference_object_for_cwe () const;
 179   json::object *
 180   make_reporting_descriptor_object_for_warning (diagnostic_context *context,
 181                                                 diagnostic_info *diagnostic,
 182                                                 diagnostic_t orig_diag_kind,
 183                                                 const char *option_text);
 184   json::object *make_reporting_descriptor_object_for_cwe_id (int cwe_id) const;
 185   json::object *
 186   make_reporting_descriptor_reference_object_for_cwe_id (int cwe_id);
 187   json::object *make_artifact_object (const char *filename);
 188   json::object *maybe_make_artifact_content_object (const char *filename) const;
 189   json::object *maybe_make_artifact_content_object (const char *filename,
 190                                                     int start_line,
 191                                                     int end_line) const;
 192   json::object *make_fix_object (const rich_location &rich_loc);
 193   json::object *make_artifact_change_object (const rich_location &richloc);
 194   json::object *make_replacement_object (const fixit_hint &hint) const;
 195   json::object *make_artifact_content_object (const char *text) const;
 196   int get_sarif_column (expanded_location exploc) const;
 197
 198   diagnostic_context *m_context;
 199
 200   /* The JSON object for the invocation object.  */
 201   sarif_invocation *m_invocation_obj;
 202
 203   /* The JSON array of pending diagnostics.  */
 204   json::array *m_results_array;
 205
 206   /* The JSON object for the result object (if any) in the current
 207      diagnostic group.  */
 208   sarif_result *m_cur_group_result;
 209
 210   hash_set <const char *> m_filenames;
 211   bool m_seen_any_relative_paths;
 212   hash_set <free_string_hash> m_rule_id_set;
 213   json::array *m_rules_arr;
 214
 215   /* The set of all CWE IDs we've seen, if any.  */
 216   hash_set <int_hash <int, 0, 1> > m_cwe_id_set;
 217
 218   int m_tabstop;
 219 };
 220
 221 static sarif_builder *the_builder;
 222
 223 /* class sarif_invocation : public json::object.  */
 224
 225 /* Handle an internal compiler error DIAGNOSTIC occurring on CONTEXT.
 226    Add an object representing the ICE to the notifications array.  */
 227
 228 void
 229 sarif_invocation::add_notification_for_ice (diagnostic_context *context,
 230                                             diagnostic_info *diagnostic,
 231                                             sarif_builder *builder)
 232 {
 233   m_success = false;
 234
 235   sarif_ice_notification *notification_obj
 236     = new sarif_ice_notification (context, diagnostic, builder);
 237   m_notifications_arr->append (notification_obj);
 238 }
 239
 240 void
 241 sarif_invocation::prepare_to_flush ()
 242 {
 243   /* "executionSuccessful" property (SARIF v2.1.0 section 3.20.14).  */
 244   set ("executionSuccessful", new json::literal (m_success));
 245
 246   /* "toolExecutionNotifications" property (SARIF v2.1.0 section 3.20.21).  */
 247   set ("toolExecutionNotifications", m_notifications_arr);
 248 }
 249
 250 /* class sarif_result : public json::object.  */
 251
 252 /* Handle secondary diagnostics that occur within a diagnostic group.
 253    The closest SARIF seems to have to nested diagnostics is the
 254    "relatedLocations" property of result objects (SARIF v2.1.0 section 3.27.22),
 255    so we lazily set this property and populate the array if and when
 256    secondary diagnostics occur (such as notes to a warning).  */
 257
 258 void
 259 sarif_result::on_nested_diagnostic (diagnostic_context *context,
 260                                     diagnostic_info *diagnostic,
 261                                     diagnostic_t /*orig_diag_kind*/,
 262                                     sarif_builder *builder)
 263 {
 264   if (!m_related_locations_arr)
 265     {
 266       m_related_locations_arr = new json::array ();
 267       set ("relatedLocations", m_related_locations_arr);
 268     }
 269
 270   /* We don't yet generate meaningful logical locations for notes;
 271      sometimes these will related to current_function_decl, but
 272      often they won't.  */
 273   json::object *location_obj
 274     = builder->make_location_object (*diagnostic->richloc, NULL);
 275   json::object *message_obj
 276     = builder->make_message_object (pp_formatted_text (context->printer));
 277   pp_clear_output_area (context->printer);
 278   location_obj->set ("message", message_obj);
 279
 280   m_related_locations_arr->append (location_obj);
 281 }
 282
 283 /* class sarif_ice_notification : public json::object.  */
 284
 285 /* sarif_ice_notification's ctor.
 286    DIAGNOSTIC is an internal compiler error.  */
 287
 288 sarif_ice_notification::sarif_ice_notification (diagnostic_context *context,
 289                                                 diagnostic_info *diagnostic,
 290                                                 sarif_builder *builder)
 291 {
 292   /* "locations" property (SARIF v2.1.0 section 3.58.4).  */
 293   json::array *locations_arr = builder->make_locations_arr (diagnostic);
 294   set ("locations", locations_arr);
 295
 296   /* "message" property (SARIF v2.1.0 section 3.85.5).  */
 297   json::object *message_obj
 298     = builder->make_message_object (pp_formatted_text (context->printer));
 299   pp_clear_output_area (context->printer);
 300   set ("message", message_obj);
 301
 302   /* "level" property (SARIF v2.1.0 section 3.58.6).  */
 303   set ("level", new json::string ("error"));
 304 }
 305
 306 /* class sarif_builder.  */
 307
 308 /* sarif_builder's ctor.  */
 309
 310 sarif_builder::sarif_builder (diagnostic_context *context)
 311 : m_context (context),
 312   m_invocation_obj (new sarif_invocation ()),
 313   m_results_array (new json::array ()),
 314   m_cur_group_result (NULL),
 315   m_seen_any_relative_paths (false),
 316   m_rule_id_set (),
 317   m_rules_arr (new json::array ()),
 318   m_tabstop (context->tabstop)
 319 {
 320 }
 321
 322 /* Implementation of "end_diagnostic" for SARIF output.  */
 323
 324 void
 325 sarif_builder::end_diagnostic (diagnostic_context *context,
 326                                diagnostic_info *diagnostic,
 327                                diagnostic_t orig_diag_kind)
 328 {
 329   if (diagnostic->kind == DK_ICE || diagnostic->kind == DK_ICE_NOBT)
 330     {
 331       m_invocation_obj->add_notification_for_ice (context, diagnostic, this);
 332       return;
 333     }
 334
 335   if (m_cur_group_result)
 336     /* Nested diagnostic.  */
 337     m_cur_group_result->on_nested_diagnostic (context,
 338                                               diagnostic,
 339                                               orig_diag_kind,
 340                                               this);
 341   else
 342     {
 343       /* Top-level diagnostic.  */
 344       sarif_result *result_obj
 345         = make_result_object (context, diagnostic, orig_diag_kind);
 346       m_results_array->append (result_obj);
 347       m_cur_group_result = result_obj;
 348     }
 349 }
 350
 351 /* Implementation of "end_group_cb" for SARIF output.  */
 352
 353 void
 354 sarif_builder::end_group ()
 355 {
 356   m_cur_group_result = NULL;
 357 }
 358
 359 /* Create a top-level object, and add it to all the results
 360    (and other entities) we've seen so far.
 361
 362    Flush it all to OUTF.  */
 363
 364 void
 365 sarif_builder::flush_to_file (FILE *outf)
 366 {
 367   m_invocation_obj->prepare_to_flush ();
 368   json::object *top = make_top_level_object (m_invocation_obj, m_results_array);
 369   top->dump (outf);
 370   m_invocation_obj = NULL;
 371   m_results_array = NULL;
 372   fprintf (outf, "\n");
 373   delete top;
 374 }
 375
 376 /* Attempt to convert DIAG_KIND to a suitable value for the "level"
 377    property (SARIF v2.1.0 section 3.27.10).
 378
 379    Return NULL if there isn't one.  */
 380
 381 static const char *
 382 maybe_get_sarif_level (diagnostic_t diag_kind)
 383 {
 384   switch (diag_kind)
 385     {
 386     case DK_WARNING:
 387       return "warning";
 388     case DK_ERROR:
 389       return "error";
 390     case DK_NOTE:
 391     case DK_ANACHRONISM:
 392       return "note";
 393     default:
 394       return NULL;
 395     }
 396 }
 397
 398 /* Make a string for DIAG_KIND suitable for use a ruleId
 399    (SARIF v2.1.0 section 3.27.5) as a fallback for when we don't
 400    have anything better to use.  */
 401
 402 static char *
 403 make_rule_id_for_diagnostic_kind (diagnostic_t diag_kind)
 404 {
 405   static const char *const diagnostic_kind_text[] = {
 406 #define DEFINE_DIAGNOSTIC_KIND(K, T, C) (T),
 407 #include "diagnostic.def"
 408 #undef DEFINE_DIAGNOSTIC_KIND
 409     "must-not-happen"
 410   };
 411   /* Lose the trailing ": ".  */
 412   const char *kind_text = diagnostic_kind_text[diag_kind];
 413   size_t len = strlen (kind_text);
 414   gcc_assert (len > 2);
 415   gcc_assert (kind_text[len - 2] == ':');
 416   gcc_assert (kind_text[len - 1] == ' ');
 417   char *rstrip = xstrdup (kind_text);
 418   rstrip[len - 2] = '\0';
 419   return rstrip;
 420 }
 421
 422 /* Make a result object (SARIF v2.1.0 section 3.27) for DIAGNOSTIC.  */
 423
 424 sarif_result *
 425 sarif_builder::make_result_object (diagnostic_context *context,
 426                                    diagnostic_info *diagnostic,
 427                                    diagnostic_t orig_diag_kind)
 428 {
 429   sarif_result *result_obj = new sarif_result ();
 430
 431   /* "ruleId" property (SARIF v2.1.0 section 3.27.5).  */
 432   /* Ideally we'd have an option_name for these.  */
 433   if (char *option_text
 434         = context->option_name (context, diagnostic->option_index,
 435                                 orig_diag_kind, diagnostic->kind))
 436     {
 437       /* Lazily create reportingDescriptor objects for and add to m_rules_arr.
 438          Set ruleId referencing them.  */
 439       result_obj->set ("ruleId", new json::string (option_text));
 440       if (m_rule_id_set.contains (option_text))
 441         free (option_text);
 442       else
 443         {
 444           /* This is the first time we've seen this ruleId.  */
 445           /* Add to set, taking ownership.  */
 446           m_rule_id_set.add (option_text);
 447
 448           json::object *reporting_desc_obj
 449             = make_reporting_descriptor_object_for_warning (context,
 450                                                             diagnostic,
 451                                                             orig_diag_kind,
 452                                                             option_text);
 453           m_rules_arr->append (reporting_desc_obj);
 454         }
 455     }
 456   else
 457     {
 458       /* Otherwise, we have an "error" or a stray "note"; use the
 459          diagnostic kind as the ruleId, so that the result object at least
 460          has a ruleId.
 461          We don't bother creating reportingDescriptor objects for these.  */
 462       char *rule_id = make_rule_id_for_diagnostic_kind (orig_diag_kind);
 463       result_obj->set ("ruleId", new json::string (rule_id));
 464       free (rule_id);
 465     }
 466
 467   /* "taxa" property (SARIF v2.1.0 section 3.27.8).  */
 468   if (diagnostic->metadata)
 469     if (int cwe_id = diagnostic->metadata->get_cwe ())
 470       {
 471         json::array *taxa_arr = new json::array ();
 472         json::object *cwe_id_obj
 473           = make_reporting_descriptor_reference_object_for_cwe_id (cwe_id);
 474         taxa_arr->append (cwe_id_obj);
 475         result_obj->set ("taxa", taxa_arr);
 476       }
 477
 478   /* "level" property (SARIF v2.1.0 section 3.27.10).  */
 479   if (const char *sarif_level = maybe_get_sarif_level (diagnostic->kind))
 480     result_obj->set ("level", new json::string (sarif_level));
 481
 482   /* "message" property (SARIF v2.1.0 section 3.27.11).  */
 483   json::object *message_obj
 484     = make_message_object (pp_formatted_text (context->printer));
 485   pp_clear_output_area (context->printer);
 486   result_obj->set ("message", message_obj);
 487
 488   /* "locations" property (SARIF v2.1.0 section 3.27.12).  */
 489   json::array *locations_arr = make_locations_arr (diagnostic);
 490   result_obj->set ("locations", locations_arr);
 491
 492   /* "codeFlows" property (SARIF v2.1.0 section 3.27.18).  */
 493   if (const diagnostic_path *path = diagnostic->richloc->get_path ())
 494     {
 495       json::array *code_flows_arr = new json::array ();
 496       json::object *code_flow_obj = make_code_flow_object (*path);
 497       code_flows_arr->append (code_flow_obj);
 498       result_obj->set ("codeFlows", code_flows_arr);
 499     }
 500
 501   /* The "relatedLocations" property (SARIF v2.1.0 section 3.27.22) is
 502      set up later, if any nested diagnostics occur within this diagnostic
 503      group.  */
 504
 505   /* "fixes" property (SARIF v2.1.0 section 3.27.30).  */
 506   const rich_location *richloc = diagnostic->richloc;
 507   if (richloc->get_num_fixit_hints ())
 508     {
 509       json::array *fix_arr = new json::array ();
 510       json::object *fix_obj = make_fix_object (*richloc);
 511       fix_arr->append (fix_obj);
 512       result_obj->set ("fixes", fix_arr);
 513     }
 514
 515   return result_obj;
 516 }
 517
 518 /* Make a reportingDescriptor object (SARIF v2.1.0 section 3.49)
 519    for a GCC warning.  */
 520
 521 json::object *
 522 sarif_builder::
 523 make_reporting_descriptor_object_for_warning (diagnostic_context *context,
 524                                               diagnostic_info *diagnostic,
 525                                               diagnostic_t /*orig_diag_kind*/,
 526                                               const char *option_text)
 527 {
 528   json::object *reporting_desc = new json::object ();
 529
 530   /* "id" property (SARIF v2.1.0 section 3.49.3).  */
 531   reporting_desc->set ("id", new json::string (option_text));
 532
 533   /* We don't implement "name" property (SARIF v2.1.0 section 3.49.7), since
 534      it seems redundant compared to "id".  */
 535
 536   /* "helpUri" property (SARIF v2.1.0 section 3.49.12).  */
 537   if (context->get_option_url)
 538     {
 539       char *option_url
 540         = context->get_option_url (context, diagnostic->option_index);
 541       if (option_url)
 542         {
 543           reporting_desc->set ("helpUri", new json::string (option_url));
 544           free (option_url);
 545         }
 546     }
 547
 548   return reporting_desc;
 549 }
 550
 551 /* Make a reportingDescriptor object (SARIF v2.1.0 section 3.49)
 552    for CWE_ID, for use within the CWE taxa array.  */
 553
 554 json::object *
 555 sarif_builder::make_reporting_descriptor_object_for_cwe_id (int cwe_id) const
 556 {
 557   json::object *reporting_desc = new json::object ();
 558
 559   /* "id" property (SARIF v2.1.0 section 3.49.3).  */
 560   {
 561     pretty_printer pp;
 562     pp_printf (&pp, "%i", cwe_id);
 563     reporting_desc->set ("id", new json::string (pp_formatted_text (&pp)));
 564   }
 565
 566   /* "helpUri" property (SARIF v2.1.0 section 3.49.12).  */
 567   {
 568     char *url = get_cwe_url (cwe_id);
 569     reporting_desc->set ("helpUri", new json::string (url));
 570     free (url);
 571   }
 572
 573   return reporting_desc;
 574 }
 575
 576 /* Make a reportingDescriptorReference object (SARIF v2.1.0 section 3.52)
 577    referencing CWE_ID, for use within a result object.
 578    Also, add CWE_ID to m_cwe_id_set.  */
 579
 580 json::object *
 581 sarif_builder::
 582 make_reporting_descriptor_reference_object_for_cwe_id (int cwe_id)
 583 {
 584   json::object *desc_ref_obj = new json::object ();
 585
 586   /* "id" property (SARIF v2.1.0 section 3.52.4).  */
 587   {
 588     pretty_printer pp;
 589     pp_printf (&pp, "%i", cwe_id);
 590     desc_ref_obj->set ("id", new json::string (pp_formatted_text (&pp)));
 591   }
 592
 593   /* "toolComponent" property (SARIF v2.1.0 section 3.52.7).  */
 594   json::object *comp_ref_obj = make_tool_component_reference_object_for_cwe ();
 595   desc_ref_obj->set ("toolComponent", comp_ref_obj);
 596
 597   /* Add CWE_ID to our set.  */
 598   gcc_assert (cwe_id > 0);
 599   m_cwe_id_set.add (cwe_id);
 600
 601   return desc_ref_obj;
 602 }
 603
 604 /* Make a toolComponentReference object (SARIF v2.1.0 section 3.54) that
 605    references the CWE taxonomy.  */
 606
 607 json::object *
 608 sarif_builder::
 609 make_tool_component_reference_object_for_cwe () const
 610 {
 611   json::object *comp_ref_obj = new json::object ();
 612
 613   /* "name" property  (SARIF v2.1.0 section 3.54.3).  */
 614   comp_ref_obj->set ("name", new json::string ("cwe"));
 615
 616   return comp_ref_obj;
 617 }
 618
 619 /* Make an array suitable for use as the "locations" property of:
 620    - a "result" object (SARIF v2.1.0 section 3.27.12), or
 621    - a "notification" object (SARIF v2.1.0 section 3.58.4).  */
 622
 623 json::array *
 624 sarif_builder::make_locations_arr (diagnostic_info *diagnostic)
 625 {
 626   json::array *locations_arr = new json::array ();
 627   const logical_location *logical_loc = NULL;
 628   if (m_context->m_client_data_hooks)
 629     logical_loc
 630       = m_context->m_client_data_hooks->get_current_logical_location ();
 631
 632   json::object *location_obj
 633     = make_location_object (*diagnostic->richloc, logical_loc);
 634   locations_arr->append (location_obj);
 635   return locations_arr;
 636 }
 637
 638 /* If LOGICAL_LOC is non-NULL, use it to create a "logicalLocations" property
 639    within LOCATION_OBJ (SARIF v2.1.0 section 3.28.4).  */
 640
 641 void
 642 sarif_builder::
 643 set_any_logical_locs_arr (json::object *location_obj,
 644                           const logical_location *logical_loc)
 645 {
 646   if (!logical_loc)
 647     return;
 648   json::object *logical_loc_obj = make_logical_location_object (*logical_loc);
 649   json::array *location_locs_arr = new json::array ();
 650   location_locs_arr->append (logical_loc_obj);
 651   location_obj->set ("logicalLocations", location_locs_arr);
 652 }
 653
 654 /* Make a location object (SARIF v2.1.0 section 3.28) for RICH_LOC
 655    and LOGICAL_LOC.  */
 656
 657 json::object *
 658 sarif_builder::make_location_object (const rich_location &rich_loc,
 659                                      const logical_location *logical_loc)
 660 {
 661   json::object *location_obj = new json::object ();
 662
 663   /* Get primary loc from RICH_LOC.  */
 664   location_t loc = rich_loc.get_loc ();
 665
 666   /* "physicalLocation" property (SARIF v2.1.0 section 3.28.3).  */
 667   if (json::object *phs_loc_obj = maybe_make_physical_location_object (loc))
 668     location_obj->set ("physicalLocation", phs_loc_obj);
 669
 670   /* "logicalLocations" property (SARIF v2.1.0 section 3.28.4).  */
 671   set_any_logical_locs_arr (location_obj, logical_loc);
 672
 673   return location_obj;
 674 }
 675
 676 /* Make a location object (SARIF v2.1.0 section 3.28) for EVENT
 677    within a diagnostic_path.  */
 678
 679 json::object *
 680 sarif_builder::make_location_object (const diagnostic_event &event)
 681 {
 682   json::object *location_obj = new json::object ();
 683
 684   /* "physicalLocation" property (SARIF v2.1.0 section 3.28.3).  */
 685   location_t loc = event.get_location ();
 686   if (json::object *phs_loc_obj = maybe_make_physical_location_object (loc))
 687     location_obj->set ("physicalLocation", phs_loc_obj);
 688
 689   /* "logicalLocations" property (SARIF v2.1.0 section 3.28.4).  */
 690   const logical_location *logical_loc = event.get_logical_location ();
 691   set_any_logical_locs_arr (location_obj, logical_loc);
 692
 693   /* "message" property (SARIF v2.1.0 section 3.28.5).  */
 694   label_text ev_desc = event.get_desc (false);
 695   json::object *message_obj = make_message_object (ev_desc.get ());
 696   location_obj->set ("message", message_obj);
 697
 698   return location_obj;
 699 }
 700
 701 /* Make a physicalLocation object (SARIF v2.1.0 section 3.29) for LOC,
 702    or return NULL;
 703    Add any filename to the m_artifacts.  */
 704
 705 json::object *
 706 sarif_builder::maybe_make_physical_location_object (location_t loc)
 707 {
 708   if (loc <= BUILTINS_LOCATION || LOCATION_FILE (loc) == NULL)
 709     return NULL;
 710
 711   json::object *phys_loc_obj = new json::object ();
 712
 713   /* "artifactLocation" property (SARIF v2.1.0 section 3.29.3).  */
 714   json::object *artifact_loc_obj = make_artifact_location_object (loc);
 715   phys_loc_obj->set ("artifactLocation", artifact_loc_obj);
 716   m_filenames.add (LOCATION_FILE (loc));
 717
 718   /* "region" property (SARIF v2.1.0 section 3.29.4).  */
 719   if (json::object *region_obj = maybe_make_region_object (loc))
 720     phys_loc_obj->set ("region", region_obj);
 721
 722   /* "contextRegion" property (SARIF v2.1.0 section 3.29.5).  */
 723   if (json::object *context_region_obj
 724         = maybe_make_region_object_for_context (loc))
 725     phys_loc_obj->set ("contextRegion", context_region_obj);
 726
 727   /* Instead, we add artifacts to the run as a whole,
 728      with artifact.contents.
 729      Could do both, though.  */
 730
 731   return phys_loc_obj;
 732 }
 733
 734 /* Make an artifactLocation object (SARIF v2.1.0 section 3.4) for LOC,
 735    or return NULL.  */
 736
 737 json::object *
 738 sarif_builder::make_artifact_location_object (location_t loc)
 739 {
 740   return make_artifact_location_object (LOCATION_FILE (loc));
 741 }
 742
 743 /* The ID value for use in "uriBaseId" properties (SARIF v2.1.0 section 3.4.4)
 744    for when we need to express paths relative to PWD.  */
 745
 746 #define PWD_PROPERTY_NAME ("PWD")
 747
 748 /* Make an artifactLocation object (SARIF v2.1.0 section 3.4) for FILENAME,
 749    or return NULL.  */
 750
 751 json::object *
 752 sarif_builder::make_artifact_location_object (const char *filename)
 753 {
 754   json::object *artifact_loc_obj = new json::object ();
 755
 756   /* "uri" property (SARIF v2.1.0 section 3.4.3).  */
 757   artifact_loc_obj->set ("uri", new json::string (filename));
 758
 759   if (filename[0] != '/')
 760     {
 761       /* If we have a relative path, set the "uriBaseId" property
 762          (SARIF v2.1.0 section 3.4.4).  */
 763       artifact_loc_obj->set ("uriBaseId", new json::string (PWD_PROPERTY_NAME));
 764       m_seen_any_relative_paths = true;
 765     }
 766
 767   return artifact_loc_obj;
 768 }
 769
 770 /* Get the PWD, or NULL, as an absolute file-based URI,
 771    adding a trailing forward slash (as required by SARIF v2.1.0
 772    section 3.14.14).  */
 773
 774 static char *
 775 make_pwd_uri_str ()
 776 {
 777   /* The prefix of a file-based URI, up to, but not including the path. */
 778 #define FILE_PREFIX ("file://")
 779
 780   const char *pwd = getpwd ();
 781   if (!pwd)
 782     return NULL;
 783   size_t len = strlen (pwd);
 784   if (len == 0 || pwd[len - 1] != '/')
 785     return concat (FILE_PREFIX, pwd, "/", NULL);
 786   else
 787     {
 788       gcc_assert (pwd[len - 1] == '/');
 789       return concat (FILE_PREFIX, pwd, NULL);
 790     }
 791 }
 792
 793 /* Make an artifactLocation object (SARIF v2.1.0 section 3.4) for the pwd,
 794    for use in the "run.originalUriBaseIds" property (SARIF v2.1.0
 795    section 3.14.14) when we have any relative paths.  */
 796
 797 json::object *
 798 sarif_builder::make_artifact_location_object_for_pwd () const
 799 {
 800   json::object *artifact_loc_obj = new json::object ();
 801
 802   /* "uri" property (SARIF v2.1.0 section 3.4.3).  */
 803   if (char *pwd = make_pwd_uri_str ())
 804     {
 805       gcc_assert (strlen (pwd) > 0);
 806       gcc_assert (pwd[strlen (pwd) - 1] == '/');
 807       artifact_loc_obj->set ("uri", new json::string (pwd));
 808       free (pwd);
 809     }
 810
 811   return artifact_loc_obj;
 812 }
 813
 814 /* Get the column number within EXPLOC.  */
 815
 816 int
 817 sarif_builder::get_sarif_column (expanded_location exploc) const
 818 {
 819   cpp_char_column_policy policy (m_tabstop, cpp_wcwidth);
 820   return location_compute_display_column (exploc, policy);
 821 }
 822
 823 /* Make a region object (SARIF v2.1.0 section 3.30) for LOC,
 824    or return NULL.  */
 825
 826 json::object *
 827 sarif_builder::maybe_make_region_object (location_t loc) const
 828 {
 829   location_t caret_loc = get_pure_location (loc);
 830
 831   if (caret_loc <= BUILTINS_LOCATION)
 832     return NULL;
 833
 834   location_t start_loc = get_start (loc);
 835   location_t finish_loc = get_finish (loc);
 836
 837   expanded_location exploc_caret = expand_location (caret_loc);
 838   expanded_location exploc_start = expand_location (start_loc);
 839   expanded_location exploc_finish = expand_location (finish_loc);
 840
 841   if (exploc_start.file !=exploc_caret.file)
 842     return NULL;
 843   if (exploc_finish.file !=exploc_caret.file)
 844     return NULL;
 845
 846   json::object *region_obj = new json::object ();
 847
 848   /* "startLine" property (SARIF v2.1.0 section 3.30.5) */
 849   region_obj->set ("startLine", new json::integer_number (exploc_start.line));
 850
 851   /* "startColumn" property (SARIF v2.1.0 section 3.30.6) */
 852   region_obj->set ("startColumn",
 853                    new json::integer_number (get_sarif_column (exploc_start)));
 854
 855   /* "endLine" property (SARIF v2.1.0 section 3.30.7) */
 856   if (exploc_finish.line != exploc_start.line)
 857     region_obj->set ("endLine", new json::integer_number (exploc_finish.line));
 858
 859   /* "endColumn" property (SARIF v2.1.0 section 3.30.8).
 860      This expresses the column immediately beyond the range.  */
 861   {
 862     int next_column = sarif_builder::get_sarif_column (exploc_finish) + 1;
 863     region_obj->set ("endColumn", new json::integer_number (next_column));
 864   }
 865
 866   return region_obj;
 867 }
 868
 869 /* Make a region object (SARIF v2.1.0 section 3.30) for the "contextRegion"
 870    property (SARIF v2.1.0 section 3.29.5) of a physicalLocation.
 871
 872    This is similar to maybe_make_region_object, but ignores column numbers,
 873    covering the line(s) as a whole, and including a "snippet" property
 874    embedding those source lines, making it easier for consumers to show
 875    the pertinent source.  */
 876
 877 json::object *
 878 sarif_builder::maybe_make_region_object_for_context (location_t loc) const
 879 {
 880   location_t caret_loc = get_pure_location (loc);
 881
 882   if (caret_loc <= BUILTINS_LOCATION)
 883     return NULL;
 884
 885   location_t start_loc = get_start (loc);
 886   location_t finish_loc = get_finish (loc);
 887
 888   expanded_location exploc_caret = expand_location (caret_loc);
 889   expanded_location exploc_start = expand_location (start_loc);
 890   expanded_location exploc_finish = expand_location (finish_loc);
 891
 892   if (exploc_start.file !=exploc_caret.file)
 893     return NULL;
 894   if (exploc_finish.file !=exploc_caret.file)
 895     return NULL;
 896
 897   json::object *region_obj = new json::object ();
 898
 899   /* "startLine" property (SARIF v2.1.0 section 3.30.5) */
 900   region_obj->set ("startLine", new json::integer_number (exploc_start.line));
 901
 902   /* "endLine" property (SARIF v2.1.0 section 3.30.7) */
 903   if (exploc_finish.line != exploc_start.line)
 904     region_obj->set ("endLine", new json::integer_number (exploc_finish.line));
 905
 906   /* "snippet" property (SARIF v2.1.0 section 3.30.13).  */
 907   if (json::object *artifact_content_obj
 908          = maybe_make_artifact_content_object (exploc_start.file,
 909                                                exploc_start.line,
 910                                                exploc_finish.line))
 911     region_obj->set ("snippet", artifact_content_obj);
 912
 913   return region_obj;
 914 }
 915
 916 /* Make a region object (SARIF v2.1.0 section 3.30) for the deletion region
 917    of HINT (as per SARIF v2.1.0 section 3.57.3).  */
 918
 919 json::object *
 920 sarif_builder::make_region_object_for_hint (const fixit_hint &hint) const
 921 {
 922   location_t start_loc = hint.get_start_loc ();
 923   location_t next_loc = hint.get_next_loc ();
 924
 925   expanded_location exploc_start = expand_location (start_loc);
 926   expanded_location exploc_next = expand_location (next_loc);
 927
 928   json::object *region_obj = new json::object ();
 929
 930   /* "startLine" property (SARIF v2.1.0 section 3.30.5) */
 931   region_obj->set ("startLine", new json::integer_number (exploc_start.line));
 932
 933   /* "startColumn" property (SARIF v2.1.0 section 3.30.6) */
 934   int start_col = get_sarif_column (exploc_start);
 935   region_obj->set ("startColumn",
 936                    new json::integer_number (start_col));
 937
 938   /* "endLine" property (SARIF v2.1.0 section 3.30.7) */
 939   if (exploc_next.line != exploc_start.line)
 940     region_obj->set ("endLine", new json::integer_number (exploc_next.line));
 941
 942   /* "endColumn" property (SARIF v2.1.0 section 3.30.8).
 943      This expresses the column immediately beyond the range.  */
 944   int next_col =  get_sarif_column (exploc_next);
 945   region_obj->set ("endColumn", new json::integer_number (next_col));
 946
 947   return region_obj;
 948 }
 949
 950 /* Attempt to get a string for a logicalLocation's "kind" property
 951    (SARIF v2.1.0 section 3.33.7).
 952    Return NULL if unknown.  */
 953
 954 static const char *
 955 maybe_get_sarif_kind (enum logical_location_kind kind)
 956 {
 957   switch (kind)
 958     {
 959     default:
 960       gcc_unreachable ();
 961     case LOGICAL_LOCATION_KIND_UNKNOWN:
 962       return NULL;
 963
 964     case LOGICAL_LOCATION_KIND_FUNCTION:
 965       return "function";
 966     case LOGICAL_LOCATION_KIND_MEMBER:
 967       return "member";
 968     case LOGICAL_LOCATION_KIND_MODULE:
 969       return "module";
 970     case LOGICAL_LOCATION_KIND_NAMESPACE:
 971       return "namespace";
 972     case LOGICAL_LOCATION_KIND_TYPE:
 973       return "type";
 974     case LOGICAL_LOCATION_KIND_RETURN_TYPE:
 975       return "returnType";
 976     case LOGICAL_LOCATION_KIND_PARAMETER:
 977       return "parameter";
 978     case LOGICAL_LOCATION_KIND_VARIABLE:
 979       return "variable";
 980     }
 981 }
 982
 983 /* Make a logicalLocation object (SARIF v2.1.0 section 3.33) for LOGICAL_LOC,
 984    or return NULL.  */
 985
 986 json::object *
 987 sarif_builder::
 988 make_logical_location_object (const logical_location &logical_loc) const
 989 {
 990   json::object *logical_loc_obj = new json::object ();
 991
 992   /* "name" property (SARIF v2.1.0 section 3.33.4).  */
 993   if (const char *short_name = logical_loc.get_short_name ())
 994     logical_loc_obj->set ("name", new json::string (short_name));
 995
 996   /* "fullyQualifiedName" property (SARIF v2.1.0 section 3.33.5).  */
 997   if (const char *name_with_scope = logical_loc.get_name_with_scope ())
 998     logical_loc_obj->set ("fullyQualifiedName",
 999                           new json::string (name_with_scope));
1000
1001   /* "decoratedName" property (SARIF v2.1.0 section 3.33.6).  */
1002   if (const char *internal_name = logical_loc.get_internal_name ())
1003     logical_loc_obj->set ("decoratedName", new json::string (internal_name));
1004
1005   /* "kind" property (SARIF v2.1.0 section 3.33.7).  */
1006   enum logical_location_kind kind = logical_loc.get_kind ();
1007   if (const char *sarif_kind_str = maybe_get_sarif_kind (kind))
1008     logical_loc_obj->set ("kind", new json::string (sarif_kind_str));
1009
1010   return logical_loc_obj;
1011 }
1012
1013 /* Make a codeFlow object (SARIF v2.1.0 section 3.36) for PATH.  */
1014
1015 json::object *
1016 sarif_builder::make_code_flow_object (const diagnostic_path &path)
1017 {
1018   json::object *code_flow_obj = new json::object ();
1019
1020   /* "threadFlows" property (SARIF v2.1.0 section 3.36.3).
1021      Currently we only support one thread per result.  */
1022   json::array *thread_flows_arr = new json::array ();
1023   json::object *thread_flow_obj = make_thread_flow_object (path);
1024   thread_flows_arr->append (thread_flow_obj);
1025   code_flow_obj->set ("threadFlows", thread_flows_arr);
1026
1027   return code_flow_obj;
1028 }
1029
1030 /* Make a threadFlow object (SARIF v2.1.0 section 3.37) for PATH.  */
1031
1032 json::object *
1033 sarif_builder::make_thread_flow_object (const diagnostic_path &path)
1034 {
1035   json::object *thread_flow_obj = new json::object ();
1036
1037   /* "locations" property (SARIF v2.1.0 section 3.37.6).  */
1038   json::array *locations_arr = new json::array ();
1039   for (unsigned i = 0; i < path.num_events (); i++)
1040     {
1041       const diagnostic_event &event = path.get_event (i);
1042       json::object *thread_flow_loc_obj
1043         = make_thread_flow_location_object (event);
1044       locations_arr->append (thread_flow_loc_obj);
1045     }
1046   thread_flow_obj->set ("locations", locations_arr);
1047
1048   return thread_flow_obj;
1049 }
1050
1051 /* Make a threadFlowLocation object (SARIF v2.1.0 section 3.38) for EVENT.  */
1052
1053 json::object *
1054 sarif_builder::make_thread_flow_location_object (const diagnostic_event &ev)
1055 {
1056   json::object *thread_flow_loc_obj = new json::object ();
1057
1058   /* "location" property (SARIF v2.1.0 section 3.38.3).  */
1059   json::object *location_obj = make_location_object (ev);
1060   thread_flow_loc_obj->set ("location", location_obj);
1061
1062   /* "kinds" property (SARIF v2.1.0 section 3.38.8).  */
1063   diagnostic_event::meaning m = ev.get_meaning ();
1064   if (json::array *kinds_arr = maybe_make_kinds_array (m))
1065     thread_flow_loc_obj->set ("kinds", kinds_arr);
1066
1067   /* "nestingLevel" property (SARIF v2.1.0 section 3.38.10).  */
1068   thread_flow_loc_obj->set ("nestingLevel",
1069                             new json::integer_number (ev.get_stack_depth ()));
1070
1071   /* It might be nice to eventually implement the following for -fanalyzer:
1072      - the "stack" property (SARIF v2.1.0 section 3.38.5)
1073      - the "state" property (SARIF v2.1.0 section 3.38.9)
1074      - the "importance" property (SARIF v2.1.0 section 3.38.13).  */
1075
1076   return thread_flow_loc_obj;
1077 }
1078
1079 /* If M has any known meaning, make a json array suitable for the "kinds"
1080    property of a threadFlowLocation object (SARIF v2.1.0 section 3.38.8).
1081
1082    Otherwise, return NULL.  */
1083
1084 json::array *
1085 sarif_builder::maybe_make_kinds_array (diagnostic_event::meaning m) const
1086 {
1087   if (m.m_verb == diagnostic_event::VERB_unknown
1088       && m.m_noun == diagnostic_event::NOUN_unknown
1089       && m.m_property == diagnostic_event::PROPERTY_unknown)
1090     return NULL;
1091
1092   json::array *kinds_arr = new json::array ();
1093   if (const char *verb_str
1094         = diagnostic_event::meaning::maybe_get_verb_str (m.m_verb))
1095     kinds_arr->append (new json::string (verb_str));
1096   if (const char *noun_str
1097         = diagnostic_event::meaning::maybe_get_noun_str (m.m_noun))
1098     kinds_arr->append (new json::string (noun_str));
1099   if (const char *property_str
1100         = diagnostic_event::meaning::maybe_get_property_str (m.m_property))
1101     kinds_arr->append (new json::string (property_str));
1102   return kinds_arr;
1103 }
1104
1105 /* Make a message object (SARIF v2.1.0 section 3.11) for MSG.  */
1106
1107 json::object *
1108 sarif_builder::make_message_object (const char *msg) const
1109 {
1110   json::object *message_obj = new json::object ();
1111
1112   /* "text" property (SARIF v2.1.0 section 3.11.8).  */
1113   message_obj->set ("text", new json::string (msg));
1114
1115   return message_obj;
1116 }
1117
1118 /* Make a multiformatMessageString object (SARIF v2.1.0 section 3.12)
1119    for MSG.  */
1120
1121 json::object *
1122 sarif_builder::make_multiformat_message_string (const char *msg) const
1123 {
1124   json::object *message_obj = new json::object ();
1125
1126   /* "text" property (SARIF v2.1.0 section 3.12.3).  */
1127   message_obj->set ("text", new json::string (msg));
1128
1129   return message_obj;
1130 }
1131
1132 #define SARIF_SCHEMA "https://raw.githubusercontent.com/oasis-tcs/sarif-spec/master/Schemata/sarif-schema-2.1.0.json"
1133 #define SARIF_VERSION "2.1.0"
1134
1135 /* Make a top-level sarifLog object (SARIF v2.1.0 section 3.13).
1136    Take ownership of INVOCATION_OBJ and RESULTS.  */
1137
1138 json::object *
1139 sarif_builder::make_top_level_object (sarif_invocation *invocation_obj,
1140                                       json::array *results)
1141 {
1142   json::object *log_obj = new json::object ();
1143
1144   /* "$schema" property (SARIF v2.1.0 section 3.13.3) .  */
1145   log_obj->set ("$schema", new json::string (SARIF_SCHEMA));
1146
1147   /* "version" property (SARIF v2.1.0 section 3.13.2).  */
1148   log_obj->set ("version", new json::string (SARIF_VERSION));
1149
1150   /* "runs" property (SARIF v2.1.0 section 3.13.4).  */
1151   json::array *run_arr = new json::array ();
1152   json::object *run_obj = make_run_object (invocation_obj, results);
1153   run_arr->append (run_obj);
1154   log_obj->set ("runs", run_arr);
1155
1156   return log_obj;
1157 }
1158
1159 /* Make a run object (SARIF v2.1.0 section 3.14).
1160    Take ownership of INVOCATION_OBJ and RESULTS.  */
1161
1162 json::object *
1163 sarif_builder::make_run_object (sarif_invocation *invocation_obj,
1164                                 json::array *results)
1165 {
1166   json::object *run_obj = new json::object ();
1167
1168   /* "tool" property (SARIF v2.1.0 section 3.14.6).  */
1169   json::object *tool_obj = make_tool_object ();
1170   run_obj->set ("tool", tool_obj);
1171
1172   /* "taxonomies" property (SARIF v2.1.0 section 3.14.8).  */
1173   if (json::array *taxonomies_arr = maybe_make_taxonomies_array ())
1174     run_obj->set ("taxonomies", taxonomies_arr);
1175
1176   /* "invocations" property (SARIF v2.1.0 section 3.14.11).  */
1177   {
1178     json::array *invocations_arr = new json::array ();
1179     invocations_arr->append (invocation_obj);
1180     run_obj->set ("invocations", invocations_arr);
1181   }
1182
1183   /* "originalUriBaseIds (SARIF v2.1.0 section 3.14.14).  */
1184   if (m_seen_any_relative_paths)
1185     {
1186       json::object *orig_uri_base_ids = new json::object ();
1187       run_obj->set ("originalUriBaseIds", orig_uri_base_ids);
1188       json::object *pwd_art_loc_obj = make_artifact_location_object_for_pwd ();
1189       orig_uri_base_ids->set (PWD_PROPERTY_NAME, pwd_art_loc_obj);
1190     }
1191
1192   /* "artifacts" property (SARIF v2.1.0 section 3.14.15).  */
1193   json::array *artifacts_arr = new json::array ();
1194   for (auto iter : m_filenames)
1195     {
1196       json::object *artifact_obj = make_artifact_object (iter);
1197       artifacts_arr->append (artifact_obj);
1198     }
1199   run_obj->set ("artifacts", artifacts_arr);
1200
1201   /* "results" property (SARIF v2.1.0 section 3.14.23).  */
1202   run_obj->set ("results", results);
1203
1204   return run_obj;
1205 }
1206
1207 /* Make a tool object (SARIF v2.1.0 section 3.18).  */
1208
1209 json::object *
1210 sarif_builder::make_tool_object () const
1211 {
1212   json::object *tool_obj = new json::object ();
1213
1214   /* "driver" property (SARIF v2.1.0 section 3.18.2).  */
1215   json::object *driver_obj = make_driver_tool_component_object ();
1216   tool_obj->set ("driver", driver_obj);
1217
1218   /* Report plugins via the "extensions" property
1219      (SARIF v2.1.0 section 3.18.3).  */
1220   if (m_context->m_client_data_hooks)
1221     if (const client_version_info *vinfo
1222           = m_context->m_client_data_hooks->get_any_version_info ())
1223       {
1224         class my_plugin_visitor : public client_version_info :: plugin_visitor
1225         {
1226         public:
1227           void on_plugin (const diagnostic_client_plugin_info &p) final override
1228           {
1229             /* Create a toolComponent object (SARIF v2.1.0 section 3.19)
1230                for the plugin.  */
1231             json::object *plugin_obj = new json::object ();
1232             m_plugin_objs.safe_push (plugin_obj);
1233
1234             /* "name" property (SARIF v2.1.0 section 3.19.8).  */
1235             if (const char *short_name = p.get_short_name ())
1236               plugin_obj->set ("name", new json::string (short_name));
1237
1238             /* "fullName" property (SARIF v2.1.0 section 3.19.9).  */
1239             if (const char *full_name = p.get_full_name ())
1240               plugin_obj->set ("fullName", new json::string (full_name));
1241
1242             /* "version" property (SARIF v2.1.0 section 3.19.13).  */
1243             if (const char *version = p.get_version ())
1244               plugin_obj->set ("version", new json::string (version));
1245           }
1246           auto_vec <json::object *> m_plugin_objs;
1247         };
1248         my_plugin_visitor v;
1249         vinfo->for_each_plugin (v);
1250         if (v.m_plugin_objs.length () > 0)
1251           {
1252             json::array *extensions_arr = new json::array ();
1253             tool_obj->set ("extensions", extensions_arr);
1254             for (auto iter : v.m_plugin_objs)
1255               extensions_arr->append (iter);
1256           }
1257       }
1258
1259   /* Perhaps we could also show GMP, MPFR, MPC, isl versions as other
1260      "extensions" (see toplev.cc: print_version).  */
1261
1262   return tool_obj;
1263 }
1264
1265 /* Make a toolComponent object (SARIF v2.1.0 section 3.19) for what SARIF
1266    calls the "driver" (see SARIF v2.1.0 section 3.18.1).  */
1267
1268 json::object *
1269 sarif_builder::make_driver_tool_component_object () const
1270 {
1271   json::object *driver_obj = new json::object ();
1272
1273   if (m_context->m_client_data_hooks)
1274     if (const client_version_info *vinfo
1275           = m_context->m_client_data_hooks->get_any_version_info ())
1276       {
1277         /* "name" property (SARIF v2.1.0 section 3.19.8).  */
1278         if (const char *name = vinfo->get_tool_name ())
1279           driver_obj->set ("name", new json::string (name));
1280
1281         /* "fullName" property (SARIF v2.1.0 section 3.19.9).  */
1282         if (char *full_name = vinfo->maybe_make_full_name ())
1283           {
1284             driver_obj->set ("fullName", new json::string (full_name));
1285             free (full_name);
1286           }
1287
1288         /* "version" property (SARIF v2.1.0 section 3.19.13).  */
1289         if (const char *version = vinfo->get_version_string ())
1290           driver_obj->set ("version", new json::string (version));
1291
1292         /* "informationUri" property (SARIF v2.1.0 section 3.19.17).  */
1293         if (char *version_url =  vinfo->maybe_make_version_url ())
1294           {
1295             driver_obj->set ("informationUri", new json::string (version_url));
1296             free (version_url);
1297           }
1298       }
1299
1300   /* "rules" property (SARIF v2.1.0 section 3.19.23).  */
1301   driver_obj->set ("rules", m_rules_arr);
1302
1303   return driver_obj;
1304 }
1305
1306 /* If we've seen any CWE IDs, make an array for the "taxonomies" property
1307    (SARIF v2.1.0 section 3.14.8) of a run object, containting a singl
1308    toolComponent (3.19) as per 3.19.3, representing the CWE.
1309
1310    Otherwise return NULL.  */
1311
1312 json::array *
1313 sarif_builder::maybe_make_taxonomies_array () const
1314 {
1315   json::object *cwe_obj = maybe_make_cwe_taxonomy_object ();
1316   if (!cwe_obj)
1317     return NULL;
1318
1319   /* "taxonomies" property (SARIF v2.1.0 section 3.14.8).  */
1320   json::array *taxonomies_arr = new json::array ();
1321   taxonomies_arr->append (cwe_obj);
1322   return taxonomies_arr;
1323 }
1324
1325 /* If we've seen any CWE IDs, make a toolComponent object
1326    (SARIF v2.1.0 section 3.19) representing the CWE taxonomy, as per 3.19.3.
1327    Populate the "taxa" property with all of the CWE IDs in m_cwe_id_set.
1328
1329    Otherwise return NULL.  */
1330
1331 json::object *
1332 sarif_builder::maybe_make_cwe_taxonomy_object () const
1333 {
1334   if (m_cwe_id_set.is_empty ())
1335     return NULL;
1336
1337   json::object *taxonomy_obj = new json::object ();
1338
1339   /* "name" property (SARIF v2.1.0 section 3.19.8).  */
1340   taxonomy_obj->set ("name", new json::string ("CWE"));
1341
1342   /* "version" property (SARIF v2.1.0 section 3.19.13).  */
1343   taxonomy_obj->set ("version", new json::string ("4.7"));
1344
1345   /* "organization" property (SARIF v2.1.0 section 3.19.18).  */
1346   taxonomy_obj->set ("organization", new json::string ("MITRE"));
1347
1348   /* "shortDescription" property (SARIF v2.1.0 section 3.19.19).  */
1349   json::object *short_desc
1350     = make_multiformat_message_string ("The MITRE"
1351                                        " Common Weakness Enumeration");
1352   taxonomy_obj->set ("shortDescription", short_desc);
1353
1354   /* "taxa" property (SARIF v2.1.0 3.section 3.19.25).  */
1355   json::array *taxa_arr = new json::array ();
1356   for (auto cwe_id : m_cwe_id_set)
1357     {
1358       json::object *cwe_taxon
1359         = make_reporting_descriptor_object_for_cwe_id (cwe_id);
1360       taxa_arr->append (cwe_taxon);
1361     }
1362   taxonomy_obj->set ("taxa", taxa_arr);
1363
1364   return taxonomy_obj;
1365 }
1366
1367 /* Make an artifact object (SARIF v2.1.0 section 3.24).  */
1368
1369 json::object *
1370 sarif_builder::make_artifact_object (const char *filename)
1371 {
1372   json::object *artifact_obj = new json::object ();
1373
1374   /* "location" property (SARIF v2.1.0 section 3.24.2).  */
1375   json::object *artifact_loc_obj = make_artifact_location_object (filename);
1376   artifact_obj->set ("location", artifact_loc_obj);
1377
1378   /* "contents" property (SARIF v2.1.0 section 3.24.8).  */
1379   if (json::object *artifact_content_obj
1380         = maybe_make_artifact_content_object (filename))
1381     artifact_obj->set ("contents", artifact_content_obj);
1382
1383   /* "sourceLanguage" property (SARIF v2.1.0 section 3.24.10).  */
1384   if (m_context->m_client_data_hooks)
1385     if (const char *source_lang
1386         = m_context->m_client_data_hooks->maybe_get_sarif_source_language
1387             (filename))
1388       artifact_obj->set ("sourceLanguage", new json::string (source_lang));
1389
1390   return artifact_obj;
1391 }
1392
1393 /* Make an artifactContent object (SARIF v2.1.0 section 3.3) for the
1394    full contents of FILENAME.  */
1395
1396 json::object *
1397 sarif_builder::maybe_make_artifact_content_object (const char *filename) const
1398 {
1399   /* Let input.cc handle any charset conversion.  */
1400   char_span utf8_content = get_source_file_content (filename);
1401   if (!utf8_content)
1402     return NULL;
1403
1404   /* Don't add it if it's not valid UTF-8.  */
1405   if (!cpp_valid_utf8_p(utf8_content.get_buffer (), utf8_content.length ()))
1406     return NULL;
1407
1408   json::object *artifact_content_obj = new json::object ();
1409   artifact_content_obj->set ("text",
1410                              new json::string (utf8_content.get_buffer (),
1411                                                utf8_content.length ()));
1412   return artifact_content_obj;
1413 }
1414
1415 /* Attempt to read the given range of lines from FILENAME; return
1416    a freshly-allocated 0-terminated buffer containing them, or NULL.  */
1417
1418 static char *
1419 get_source_lines (const char *filename,
1420                   int start_line,
1421                   int end_line)
1422 {
1423   auto_vec<char> result;
1424
1425   for (int line = start_line; line <= end_line; line++)
1426     {
1427       char_span line_content = location_get_source_line (filename, line);
1428       if (!line_content.get_buffer ())
1429         return NULL;
1430       result.reserve (line_content.length () + 1);
1431       for (size_t i = 0; i < line_content.length (); i++)
1432         result.quick_push (line_content[i]);
1433       result.quick_push ('\n');
1434     }
1435   result.safe_push ('\0');
1436
1437   return xstrdup (result.address ());
1438 }
1439
1440 /* Make an artifactContent object (SARIF v2.1.0 section 3.3) for the given
1441    run of lines within FILENAME (including the endpoints).  */
1442
1443 json::object *
1444 sarif_builder::maybe_make_artifact_content_object (const char *filename,
1445                                                    int start_line,
1446                                                    int end_line) const
1447 {
1448   char *text_utf8 = get_source_lines (filename, start_line, end_line);
1449
1450   if (!text_utf8)
1451     return NULL;
1452
1453   /* Don't add it if it's not valid UTF-8.  */
1454   if (!cpp_valid_utf8_p(text_utf8, strlen(text_utf8)))
1455     {
1456       free (text_utf8);
1457       return NULL;
1458     }
1459
1460   json::object *artifact_content_obj = new json::object ();
1461   artifact_content_obj->set ("text", new json::string (text_utf8));
1462   free (text_utf8);
1463
1464   return artifact_content_obj;
1465 }
1466
1467 /* Make a fix object (SARIF v2.1.0 section 3.55) for RICHLOC.  */
1468
1469 json::object *
1470 sarif_builder::make_fix_object (const rich_location &richloc)
1471 {
1472   json::object *fix_obj = new json::object ();
1473
1474   /* "artifactChanges" property (SARIF v2.1.0 section 3.55.3).  */
1475   /* We assume that all fix-it hints in RICHLOC affect the same file.  */
1476   json::array *artifact_change_arr = new json::array ();
1477   json::object *artifact_change_obj = make_artifact_change_object (richloc);
1478   artifact_change_arr->append (artifact_change_obj);
1479   fix_obj->set ("artifactChanges", artifact_change_arr);
1480
1481   return fix_obj;
1482 }
1483
1484 /* Make an artifactChange object (SARIF v2.1.0 section 3.56) for RICHLOC.  */
1485
1486 json::object *
1487 sarif_builder::make_artifact_change_object (const rich_location &richloc)
1488 {
1489   json::object *artifact_change_obj = new json::object ();
1490
1491   /* "artifactLocation" property (SARIF v2.1.0 section 3.56.2).  */
1492   json::object *artifact_location_obj
1493     = make_artifact_location_object (richloc.get_loc ());
1494   artifact_change_obj->set ("artifactLocation", artifact_location_obj);
1495
1496   /* "replacements" property (SARIF v2.1.0 section 3.56.3).  */
1497   json::array *replacement_arr = new json::array ();
1498   for (unsigned int i = 0; i < richloc.get_num_fixit_hints (); i++)
1499     {
1500       const fixit_hint *hint = richloc.get_fixit_hint (i);
1501       json::object *replacement_obj = make_replacement_object (*hint);
1502       replacement_arr->append (replacement_obj);
1503     }
1504   artifact_change_obj->set ("replacements", replacement_arr);
1505
1506   return artifact_change_obj;
1507 }
1508
1509 /* Make a replacement object (SARIF v2.1.0 section 3.57) for HINT.  */
1510
1511 json::object *
1512 sarif_builder::make_replacement_object (const fixit_hint &hint) const
1513 {
1514   json::object *replacement_obj = new json::object ();
1515
1516   /* "deletedRegion" property (SARIF v2.1.0 section 3.57.3).  */
1517   json::object *region_obj = make_region_object_for_hint (hint);
1518   replacement_obj->set ("deletedRegion", region_obj);
1519
1520   /* "insertedContent" property (SARIF v2.1.0 section 3.57.4).  */
1521   json::object *content_obj = make_artifact_content_object (hint.get_string ());
1522   replacement_obj->set ("insertedContent", content_obj);
1523
1524   return replacement_obj;
1525 }
1526
1527 /* Make an artifactContent object (SARIF v2.1.0 section 3.3) for TEXT.  */
1528
1529 json::object *
1530 sarif_builder::make_artifact_content_object (const char *text) const
1531 {
1532   json::object *content_obj = new json::object ();
1533
1534   /* "text" property (SARIF v2.1.0 section 3.3.2).  */
1535   content_obj->set ("text", new json::string (text));
1536
1537   return content_obj;
1538 }
1539
1540 /* No-op implementation of "begin_diagnostic" for SARIF output.  */
1541
1542 static void
1543 sarif_begin_diagnostic (diagnostic_context *, diagnostic_info *)
1544 {
1545 }
1546
1547 /* Implementation of "end_diagnostic" for SARIF output.  */
1548
1549 static void
1550 sarif_end_diagnostic (diagnostic_context *context, diagnostic_info *diagnostic,
1551                       diagnostic_t orig_diag_kind)
1552 {
1553   gcc_assert (the_builder);
1554   the_builder->end_diagnostic (context, diagnostic, orig_diag_kind);
1555 }
1556
1557 /* No-op implementation of "begin_group_cb" for SARIF output.  */
1558
1559 static void
1560 sarif_begin_group (diagnostic_context *)
1561 {
1562 }
1563
1564 /* Implementation of "end_group_cb" for SARIF output.  */
1565
1566 static void
1567 sarif_end_group (diagnostic_context *)
1568 {
1569   gcc_assert (the_builder);
1570   the_builder->end_group ();
1571 }
1572
1573 /* Flush the top-level array to OUTF.  */
1574
1575 static void
1576 sarif_flush_to_file (FILE *outf)
1577 {
1578   gcc_assert (the_builder);
1579   the_builder->flush_to_file (outf);
1580   delete the_builder;
1581   the_builder = NULL;
1582 }
1583
1584 /* Callback for final cleanup for SARIF output to stderr.  */
1585
1586 static void
1587 sarif_stderr_final_cb (diagnostic_context *)
1588 {
1589   gcc_assert (the_builder);
1590   sarif_flush_to_file (stderr);
1591 }
1592
1593 static char *sarif_output_base_file_name;
1594
1595 /* Callback for final cleanup for SARIF output to a file.  */
1596
1597 static void
1598 sarif_file_final_cb (diagnostic_context *)
1599 {
1600   char *filename = concat (sarif_output_base_file_name, ".sarif", NULL);
1601   FILE *outf = fopen (filename, "w");
1602   if (!outf)
1603     {
1604       const char *errstr = xstrerror (errno);
1605       fnotice (stderr, "error: unable to open '%s' for writing: %s\n",
1606                filename, errstr);
1607       free (filename);
1608       return;
1609     }
1610   gcc_assert (the_builder);
1611   sarif_flush_to_file (outf);
1612   fclose (outf);
1613   free (filename);
1614 }
1615
1616 /* Callback for diagnostic_context::ice_handler_cb for when an ICE
1617    occurs.  */
1618
1619 static void
1620 sarif_ice_handler (diagnostic_context *context)
1621 {
1622   /* Attempt to ensure that a .sarif file is written out.  */
1623   diagnostic_finish (context);
1624
1625   /* Print a header for the remaining output to stderr, and
1626      return, attempting to print the usual ICE messages to
1627      stderr.  Hopefully this will be helpful to the user in
1628      indicating what's gone wrong (also for DejaGnu, for pruning
1629      those messages).   */
1630   fnotice (stderr, "Internal compiler error:\n");
1631 }
1632
1633 /* Populate CONTEXT in preparation for SARIF output (either to stderr, or
1634    to a file).  */
1635
1636 static void
1637 diagnostic_output_format_init_sarif (diagnostic_context *context)
1638 {
1639   the_builder = new sarif_builder (context);
1640
1641   /* Override callbacks.  */
1642   context->begin_diagnostic = sarif_begin_diagnostic;
1643   context->end_diagnostic = sarif_end_diagnostic;
1644   context->begin_group_cb = sarif_begin_group;
1645   context->end_group_cb =  sarif_end_group;
1646   context->print_path = NULL; /* handled in sarif_end_diagnostic.  */
1647   context->ice_handler_cb = sarif_ice_handler;
1648
1649   /* The metadata is handled in SARIF format, rather than as text.  */
1650   context->show_cwe = false;
1651   context->show_rules = false;
1652
1653   /* The option is handled in SARIF format, rather than as text.  */
1654   context->show_option_requested = false;
1655
1656   /* Don't colorize the text.  */
1657   pp_show_color (context->printer) = false;
1658 }
1659
1660 /* Populate CONTEXT in preparation for SARIF output to stderr.  */
1661
1662 void
1663 diagnostic_output_format_init_sarif_stderr (diagnostic_context *context)
1664 {
1665   diagnostic_output_format_init_sarif (context);
1666   context->final_cb = sarif_stderr_final_cb;
1667 }
1668
1669 /* Populate CONTEXT in preparation for SARIF output to a file named
1670    BASE_FILE_NAME.sarif.  */
1671
1672 void
1673 diagnostic_output_format_init_sarif_file (diagnostic_context *context,
1674                                          const char *base_file_name)
1675 {
1676   diagnostic_output_format_init_sarif (context);
1677   context->final_cb = sarif_file_final_cb;
1678   sarif_output_base_file_name = xstrdup (base_file_name);
1679 }