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 #define INCLUDE_VECTOR
  24 #include "system.h"
  25 #include "coretypes.h"
  26 #include "diagnostic.h"
  27 #include "diagnostic-metadata.h"
  28 #include "diagnostic-path.h"
  29 #include "json.h"
  30 #include "cpplib.h"
  31 #include "logical-location.h"
  32 #include "diagnostic-client-data-hooks.h"
  33 #include "diagnostic-diagram.h"
  34 #include "text-art/canvas.h"
  35 #include "diagnostic-format-sarif.h"
  36
  37 class sarif_builder;
  38
  39 /* Subclass of json::object for SARIF invocation objects
  40    (SARIF v2.1.0 section 3.20).  */
  41
  42 class sarif_invocation : public sarif_object
  43 {
  44 public:
  45   sarif_invocation ()
  46   : m_notifications_arr (new json::array ()),
  47     m_success (true)
  48   {}
  49
  50   void add_notification_for_ice (diagnostic_context *context,
  51                                  diagnostic_info *diagnostic,
  52                                  sarif_builder *builder);
  53   void prepare_to_flush (diagnostic_context *context);
  54
  55 private:
  56   json::array *m_notifications_arr;
  57   bool m_success;
  58 };
  59
  60 /* Subclass of sarif_object for SARIF result objects
  61    (SARIF v2.1.0 section 3.27).  */
  62
  63 class sarif_result : public sarif_object
  64 {
  65 public:
  66   sarif_result () : m_related_locations_arr (NULL) {}
  67
  68   void
  69   on_nested_diagnostic (diagnostic_context *context,
  70                         diagnostic_info *diagnostic,
  71                         diagnostic_t orig_diag_kind,
  72                         sarif_builder *builder);
  73   void on_diagram (diagnostic_context *context,
  74                    const diagnostic_diagram &diagram,
  75                    sarif_builder *builder);
  76
  77 private:
  78   void add_related_location (json::object *location_obj);
  79
  80   json::array *m_related_locations_arr;
  81 };
  82
  83 /* Subclass of sarif_object for SARIF notification objects
  84    (SARIF v2.1.0 section 3.58).
  85
  86    This subclass is specifically for notifying when an
  87    internal compiler error occurs.  */
  88
  89 class sarif_ice_notification : public sarif_object
  90 {
  91 public:
  92   sarif_ice_notification (diagnostic_context *context,
  93                           diagnostic_info *diagnostic,
  94                           sarif_builder *builder);
  95 };
  96
  97 /* Subclass of sarif_object for SARIF threadFlow objects
  98    (SARIF v2.1.0 section 3.37) for PATH.  */
  99
 100 class sarif_thread_flow : public sarif_object
 101 {
 102 public:
 103   sarif_thread_flow (const diagnostic_thread &thread);
 104
 105   void add_location (json::object *thread_flow_loc_obj)
 106   {
 107     m_locations_arr->append (thread_flow_loc_obj);
 108   }
 109
 110 private:
 111   json::array *m_locations_arr;
 112 };
 113
 114 /* A class for managing SARIF output (for -fdiagnostics-format=sarif-stderr
 115    and -fdiagnostics-format=sarif-file).
 116
 117    As diagnostics occur, we build "result" JSON objects, and
 118    accumulate state:
 119    - which source files are referenced
 120    - which warnings are emitted
 121    - which CWEs are used
 122
 123    At the end of the compile, we use the above to build the full SARIF
 124    object tree, adding the result objects to the correct place, and
 125    creating objects for the various source files, warnings and CWEs
 126    referenced.
 127
 128    Implemented:
 129    - fix-it hints
 130    - CWE metadata
 131    - diagnostic groups (see limitations below)
 132    - logical locations (e.g. cfun)
 133
 134    Known limitations:
 135    - GCC supports one-deep nesting of diagnostics (via auto_diagnostic_group),
 136      but we only capture location and message information from such nested
 137      diagnostics (e.g. we ignore fix-it hints on them)
 138    - doesn't yet capture command-line arguments: would be run.invocations
 139      property (SARIF v2.1.0 section 3.14.11), as invocation objects
 140      (SARIF v2.1.0 section 3.20), but we'd want to capture the arguments to
 141      toplev::main, and the response files.
 142    - doesn't capture escape_on_output_p
 143    - doesn't capture secondary locations within a rich_location
 144      (perhaps we should use the "relatedLocations" property: SARIF v2.1.0
 145      section 3.27.22)
 146    - doesn't capture "artifact.encoding" property
 147      (SARIF v2.1.0 section 3.24.9).
 148    - doesn't capture hashes of the source files
 149      ("artifact.hashes" property (SARIF v2.1.0 section 3.24.11).
 150    - doesn't capture the "analysisTarget" property
 151      (SARIF v2.1.0 section 3.27.13).
 152    - doesn't capture labelled ranges
 153    - doesn't capture -Werror cleanly
 154    - doesn't capture inlining information (can SARIF handle this?)
 155    - doesn't capture macro expansion information (can SARIF handle this?).  */
 156
 157 class sarif_builder
 158 {
 159 public:
 160   sarif_builder (diagnostic_context *context);
 161
 162   void end_diagnostic (diagnostic_context *context, diagnostic_info *diagnostic,
 163                        diagnostic_t orig_diag_kind);
 164   void emit_diagram (diagnostic_context *context,
 165                      const diagnostic_diagram &diagram);
 166   void end_group ();
 167
 168   void flush_to_file (FILE *outf);
 169
 170   json::array *make_locations_arr (diagnostic_info *diagnostic);
 171   json::object *make_location_object (const rich_location &rich_loc,
 172                                       const logical_location *logical_loc);
 173   json::object *make_message_object (const char *msg) const;
 174   json::object *
 175   make_message_object_for_diagram (diagnostic_context *context,
 176                                    const diagnostic_diagram &diagram);
 177
 178 private:
 179   sarif_result *make_result_object (diagnostic_context *context,
 180                                     diagnostic_info *diagnostic,
 181                                     diagnostic_t orig_diag_kind);
 182   void set_any_logical_locs_arr (json::object *location_obj,
 183                                  const logical_location *logical_loc);
 184   json::object *make_location_object (const diagnostic_event &event);
 185   json::object *
 186   make_logical_location_object (const logical_location &logical_loc) const;
 187   json::object *make_code_flow_object (const diagnostic_path &path);
 188   json::object *
 189   make_thread_flow_location_object (const diagnostic_event &event,
 190                                     int path_event_idx);
 191   json::array *maybe_make_kinds_array (diagnostic_event::meaning m) const;
 192   json::object *maybe_make_physical_location_object (location_t loc);
 193   json::object *make_artifact_location_object (location_t loc);
 194   json::object *make_artifact_location_object (const char *filename);
 195   json::object *make_artifact_location_object_for_pwd () const;
 196   json::object *maybe_make_region_object (location_t loc) const;
 197   json::object *maybe_make_region_object_for_context (location_t loc) const;
 198   json::object *make_region_object_for_hint (const fixit_hint &hint) const;
 199   json::object *make_multiformat_message_string (const char *msg) const;
 200   json::object *make_top_level_object (sarif_invocation *invocation_obj,
 201                                        json::array *results);
 202   json::object *make_run_object (sarif_invocation *invocation_obj,
 203                                  json::array *results);
 204   json::object *make_tool_object () const;
 205   json::object *make_driver_tool_component_object () const;
 206   json::array *maybe_make_taxonomies_array () const;
 207   json::object *maybe_make_cwe_taxonomy_object () const;
 208   json::object *make_tool_component_reference_object_for_cwe () const;
 209   json::object *
 210   make_reporting_descriptor_object_for_warning (diagnostic_context *context,
 211                                                 diagnostic_info *diagnostic,
 212                                                 diagnostic_t orig_diag_kind,
 213                                                 const char *option_text);
 214   json::object *make_reporting_descriptor_object_for_cwe_id (int cwe_id) const;
 215   json::object *
 216   make_reporting_descriptor_reference_object_for_cwe_id (int cwe_id);
 217   json::object *make_artifact_object (const char *filename);
 218   char *get_source_lines (const char *filename,
 219                           int start_line,
 220                           int end_line) const;
 221   json::object *maybe_make_artifact_content_object (const char *filename) const;
 222   json::object *maybe_make_artifact_content_object (const char *filename,
 223                                                     int start_line,
 224                                                     int end_line) const;
 225   json::object *make_fix_object (const rich_location &rich_loc);
 226   json::object *make_artifact_change_object (const rich_location &richloc);
 227   json::object *make_replacement_object (const fixit_hint &hint) const;
 228   json::object *make_artifact_content_object (const char *text) const;
 229   int get_sarif_column (expanded_location exploc) const;
 230
 231   diagnostic_context *m_context;
 232
 233   /* The JSON object for the invocation object.  */
 234   sarif_invocation *m_invocation_obj;
 235
 236   /* The JSON array of pending diagnostics.  */
 237   json::array *m_results_array;
 238
 239   /* The JSON object for the result object (if any) in the current
 240      diagnostic group.  */
 241   sarif_result *m_cur_group_result;
 242
 243   hash_set <const char *> m_filenames;
 244   bool m_seen_any_relative_paths;
 245   hash_set <free_string_hash> m_rule_id_set;
 246   json::array *m_rules_arr;
 247
 248   /* The set of all CWE IDs we've seen, if any.  */
 249   hash_set <int_hash <int, 0, 1> > m_cwe_id_set;
 250
 251   int m_tabstop;
 252 };
 253
 254 /* class sarif_object : public json::object.  */
 255
 256 sarif_property_bag &
 257 sarif_object::get_or_create_properties ()
 258 {
 259   json::value *properties_val = get ("properties");
 260   if (properties_val)
 261     {
 262       if (properties_val->get_kind () == json::JSON_OBJECT)
 263         return *static_cast <sarif_property_bag *> (properties_val);
 264     }
 265
 266   sarif_property_bag *bag = new sarif_property_bag ();
 267   set ("properties", bag);
 268   return *bag;
 269 }
 270
 271 /* class sarif_invocation : public sarif_object.  */
 272
 273 /* Handle an internal compiler error DIAGNOSTIC occurring on CONTEXT.
 274    Add an object representing the ICE to the notifications array.  */
 275
 276 void
 277 sarif_invocation::add_notification_for_ice (diagnostic_context *context,
 278                                             diagnostic_info *diagnostic,
 279                                             sarif_builder *builder)
 280 {
 281   m_success = false;
 282
 283   sarif_ice_notification *notification_obj
 284     = new sarif_ice_notification (context, diagnostic, builder);
 285   m_notifications_arr->append (notification_obj);
 286 }
 287
 288 void
 289 sarif_invocation::prepare_to_flush (diagnostic_context *context)
 290 {
 291   /* "executionSuccessful" property (SARIF v2.1.0 section 3.20.14).  */
 292   set_bool ("executionSuccessful", m_success);
 293
 294   /* "toolExecutionNotifications" property (SARIF v2.1.0 section 3.20.21).  */
 295   set ("toolExecutionNotifications", m_notifications_arr);
 296
 297   /* Call client hook, allowing it to create a custom property bag for
 298      this object (SARIF v2.1.0 section 3.8) e.g. for recording time vars.  */
 299   if (auto client_data_hooks = context->get_client_data_hooks ())
 300     client_data_hooks->add_sarif_invocation_properties (*this);
 301 }
 302
 303 /* class sarif_result : public sarif_object.  */
 304
 305 /* Handle secondary diagnostics that occur within a diagnostic group.
 306    The closest SARIF seems to have to nested diagnostics is the
 307    "relatedLocations" property of result objects (SARIF v2.1.0 section 3.27.22),
 308    so we lazily set this property and populate the array if and when
 309    secondary diagnostics occur (such as notes to a warning).  */
 310
 311 void
 312 sarif_result::on_nested_diagnostic (diagnostic_context *context,
 313                                     diagnostic_info *diagnostic,
 314                                     diagnostic_t /*orig_diag_kind*/,
 315                                     sarif_builder *builder)
 316 {
 317   /* We don't yet generate meaningful logical locations for notes;
 318      sometimes these will related to current_function_decl, but
 319      often they won't.  */
 320   json::object *location_obj
 321     = builder->make_location_object (*diagnostic->richloc, NULL);
 322   json::object *message_obj
 323     = builder->make_message_object (pp_formatted_text (context->printer));
 324   pp_clear_output_area (context->printer);
 325   location_obj->set ("message", message_obj);
 326
 327   add_related_location (location_obj);
 328 }
 329
 330 /* Handle diagrams that occur within a diagnostic group.
 331    The closest thing in SARIF seems to be to add a location to the
 332    "releatedLocations" property  (SARIF v2.1.0 section 3.27.22),
 333    and to put the diagram into the "message" property of that location
 334    (SARIF v2.1.0 section 3.28.5).  */
 335
 336 void
 337 sarif_result::on_diagram (diagnostic_context *context,
 338                           const diagnostic_diagram &diagram,
 339                           sarif_builder *builder)
 340 {
 341   json::object *location_obj = new json::object ();
 342   json::object *message_obj
 343     = builder->make_message_object_for_diagram (context, diagram);
 344   location_obj->set ("message", message_obj);
 345
 346   add_related_location (location_obj);
 347 }
 348
 349 /* Add LOCATION_OBJ to this result's "relatedLocations" array,
 350    creating it if it doesn't yet exist.  */
 351
 352 void
 353 sarif_result::add_related_location (json::object *location_obj)
 354 {
 355   if (!m_related_locations_arr)
 356     {
 357       m_related_locations_arr = new json::array ();
 358       set ("relatedLocations", m_related_locations_arr);
 359     }
 360   m_related_locations_arr->append (location_obj);
 361 }
 362
 363 /* class sarif_ice_notification : public sarif_object.  */
 364
 365 /* sarif_ice_notification's ctor.
 366    DIAGNOSTIC is an internal compiler error.  */
 367
 368 sarif_ice_notification::sarif_ice_notification (diagnostic_context *context,
 369                                                 diagnostic_info *diagnostic,
 370                                                 sarif_builder *builder)
 371 {
 372   /* "locations" property (SARIF v2.1.0 section 3.58.4).  */
 373   json::array *locations_arr = builder->make_locations_arr (diagnostic);
 374   set ("locations", locations_arr);
 375
 376   /* "message" property (SARIF v2.1.0 section 3.85.5).  */
 377   json::object *message_obj
 378     = builder->make_message_object (pp_formatted_text (context->printer));
 379   pp_clear_output_area (context->printer);
 380   set ("message", message_obj);
 381
 382   /* "level" property (SARIF v2.1.0 section 3.58.6).  */
 383   set_string ("level", "error");
 384 }
 385
 386 /* class sarif_thread_flow : public sarif_object.  */
 387
 388 sarif_thread_flow::sarif_thread_flow (const diagnostic_thread &thread)
 389 {
 390   /* "id" property (SARIF v2.1.0 section 3.37.2).  */
 391   label_text name (thread.get_name (false));
 392   set_string ("id", name.get ());
 393
 394   /* "locations" property (SARIF v2.1.0 section 3.37.6).  */
 395   m_locations_arr = new json::array ();
 396   set ("locations", m_locations_arr);
 397 }
 398
 399 /* class sarif_builder.  */
 400
 401 /* sarif_builder's ctor.  */
 402
 403 sarif_builder::sarif_builder (diagnostic_context *context)
 404 : m_context (context),
 405   m_invocation_obj (new sarif_invocation ()),
 406   m_results_array (new json::array ()),
 407   m_cur_group_result (NULL),
 408   m_seen_any_relative_paths (false),
 409   m_rule_id_set (),
 410   m_rules_arr (new json::array ()),
 411   m_tabstop (context->m_tabstop)
 412 {
 413 }
 414
 415 /* Implementation of "end_diagnostic" for SARIF output.  */
 416
 417 void
 418 sarif_builder::end_diagnostic (diagnostic_context *context,
 419                                diagnostic_info *diagnostic,
 420                                diagnostic_t orig_diag_kind)
 421 {
 422   if (diagnostic->kind == DK_ICE || diagnostic->kind == DK_ICE_NOBT)
 423     {
 424       m_invocation_obj->add_notification_for_ice (context, diagnostic, this);
 425       return;
 426     }
 427
 428   if (m_cur_group_result)
 429     /* Nested diagnostic.  */
 430     m_cur_group_result->on_nested_diagnostic (context,
 431                                               diagnostic,
 432                                               orig_diag_kind,
 433                                               this);
 434   else
 435     {
 436       /* Top-level diagnostic.  */
 437       sarif_result *result_obj
 438         = make_result_object (context, diagnostic, orig_diag_kind);
 439       m_results_array->append (result_obj);
 440       m_cur_group_result = result_obj;
 441     }
 442 }
 443
 444 /* Implementation of diagnostic_context::m_diagrams.m_emission_cb
 445    for SARIF output.  */
 446
 447 void
 448 sarif_builder::emit_diagram (diagnostic_context *context,
 449                              const diagnostic_diagram &diagram)
 450 {
 451   /* We must be within the emission of a top-level diagnostic.  */
 452   gcc_assert (m_cur_group_result);
 453   m_cur_group_result->on_diagram (context, diagram, this);
 454 }
 455
 456 /* Implementation of "end_group_cb" for SARIF output.  */
 457
 458 void
 459 sarif_builder::end_group ()
 460 {
 461   m_cur_group_result = NULL;
 462 }
 463
 464 /* Create a top-level object, and add it to all the results
 465    (and other entities) we've seen so far.
 466
 467    Flush it all to OUTF.  */
 468
 469 void
 470 sarif_builder::flush_to_file (FILE *outf)
 471 {
 472   m_invocation_obj->prepare_to_flush (m_context);
 473   json::object *top = make_top_level_object (m_invocation_obj, m_results_array);
 474   top->dump (outf);
 475   m_invocation_obj = NULL;
 476   m_results_array = NULL;
 477   fprintf (outf, "\n");
 478   delete top;
 479 }
 480
 481 /* Attempt to convert DIAG_KIND to a suitable value for the "level"
 482    property (SARIF v2.1.0 section 3.27.10).
 483
 484    Return NULL if there isn't one.  */
 485
 486 static const char *
 487 maybe_get_sarif_level (diagnostic_t diag_kind)
 488 {
 489   switch (diag_kind)
 490     {
 491     case DK_WARNING:
 492       return "warning";
 493     case DK_ERROR:
 494       return "error";
 495     case DK_NOTE:
 496     case DK_ANACHRONISM:
 497       return "note";
 498     default:
 499       return NULL;
 500     }
 501 }
 502
 503 /* Make a string for DIAG_KIND suitable for use a ruleId
 504    (SARIF v2.1.0 section 3.27.5) as a fallback for when we don't
 505    have anything better to use.  */
 506
 507 static char *
 508 make_rule_id_for_diagnostic_kind (diagnostic_t diag_kind)
 509 {
 510   static const char *const diagnostic_kind_text[] = {
 511 #define DEFINE_DIAGNOSTIC_KIND(K, T, C) (T),
 512 #include "diagnostic.def"
 513 #undef DEFINE_DIAGNOSTIC_KIND
 514     "must-not-happen"
 515   };
 516   /* Lose the trailing ": ".  */
 517   const char *kind_text = diagnostic_kind_text[diag_kind];
 518   size_t len = strlen (kind_text);
 519   gcc_assert (len > 2);
 520   gcc_assert (kind_text[len - 2] == ':');
 521   gcc_assert (kind_text[len - 1] == ' ');
 522   char *rstrip = xstrdup (kind_text);
 523   rstrip[len - 2] = '\0';
 524   return rstrip;
 525 }
 526
 527 /* Make a result object (SARIF v2.1.0 section 3.27) for DIAGNOSTIC.  */
 528
 529 sarif_result *
 530 sarif_builder::make_result_object (diagnostic_context *context,
 531                                    diagnostic_info *diagnostic,
 532                                    diagnostic_t orig_diag_kind)
 533 {
 534   sarif_result *result_obj = new sarif_result ();
 535
 536   /* "ruleId" property (SARIF v2.1.0 section 3.27.5).  */
 537   /* Ideally we'd have an option_name for these.  */
 538   if (char *option_text
 539         = context->make_option_name (diagnostic->option_index,
 540                                      orig_diag_kind, diagnostic->kind))
 541     {
 542       /* Lazily create reportingDescriptor objects for and add to m_rules_arr.
 543          Set ruleId referencing them.  */
 544       result_obj->set_string ("ruleId", option_text);
 545       if (m_rule_id_set.contains (option_text))
 546         free (option_text);
 547       else
 548         {
 549           /* This is the first time we've seen this ruleId.  */
 550           /* Add to set, taking ownership.  */
 551           m_rule_id_set.add (option_text);
 552
 553           json::object *reporting_desc_obj
 554             = make_reporting_descriptor_object_for_warning (context,
 555                                                             diagnostic,
 556                                                             orig_diag_kind,
 557                                                             option_text);
 558           m_rules_arr->append (reporting_desc_obj);
 559         }
 560     }
 561   else
 562     {
 563       /* Otherwise, we have an "error" or a stray "note"; use the
 564          diagnostic kind as the ruleId, so that the result object at least
 565          has a ruleId.
 566          We don't bother creating reportingDescriptor objects for these.  */
 567       char *rule_id = make_rule_id_for_diagnostic_kind (orig_diag_kind);
 568       result_obj->set_string ("ruleId", rule_id);
 569       free (rule_id);
 570     }
 571
 572   /* "taxa" property (SARIF v2.1.0 section 3.27.8).  */
 573   if (diagnostic->metadata)
 574     if (int cwe_id = diagnostic->metadata->get_cwe ())
 575       {
 576         json::array *taxa_arr = new json::array ();
 577         json::object *cwe_id_obj
 578           = make_reporting_descriptor_reference_object_for_cwe_id (cwe_id);
 579         taxa_arr->append (cwe_id_obj);
 580         result_obj->set ("taxa", taxa_arr);
 581       }
 582
 583   /* "level" property (SARIF v2.1.0 section 3.27.10).  */
 584   if (const char *sarif_level = maybe_get_sarif_level (diagnostic->kind))
 585     result_obj->set_string ("level", sarif_level);
 586
 587   /* "message" property (SARIF v2.1.0 section 3.27.11).  */
 588   json::object *message_obj
 589     = make_message_object (pp_formatted_text (context->printer));
 590   pp_clear_output_area (context->printer);
 591   result_obj->set ("message", message_obj);
 592
 593   /* "locations" property (SARIF v2.1.0 section 3.27.12).  */
 594   json::array *locations_arr = make_locations_arr (diagnostic);
 595   result_obj->set ("locations", locations_arr);
 596
 597   /* "codeFlows" property (SARIF v2.1.0 section 3.27.18).  */
 598   if (const diagnostic_path *path = diagnostic->richloc->get_path ())
 599     {
 600       json::array *code_flows_arr = new json::array ();
 601       json::object *code_flow_obj = make_code_flow_object (*path);
 602       code_flows_arr->append (code_flow_obj);
 603       result_obj->set ("codeFlows", code_flows_arr);
 604     }
 605
 606   /* The "relatedLocations" property (SARIF v2.1.0 section 3.27.22) is
 607      set up later, if any nested diagnostics occur within this diagnostic
 608      group.  */
 609
 610   /* "fixes" property (SARIF v2.1.0 section 3.27.30).  */
 611   const rich_location *richloc = diagnostic->richloc;
 612   if (richloc->get_num_fixit_hints ())
 613     {
 614       json::array *fix_arr = new json::array ();
 615       json::object *fix_obj = make_fix_object (*richloc);
 616       fix_arr->append (fix_obj);
 617       result_obj->set ("fixes", fix_arr);
 618     }
 619
 620   return result_obj;
 621 }
 622
 623 /* Make a reportingDescriptor object (SARIF v2.1.0 section 3.49)
 624    for a GCC warning.  */
 625
 626 json::object *
 627 sarif_builder::
 628 make_reporting_descriptor_object_for_warning (diagnostic_context *context,
 629                                               diagnostic_info *diagnostic,
 630                                               diagnostic_t /*orig_diag_kind*/,
 631                                               const char *option_text)
 632 {
 633   json::object *reporting_desc = new json::object ();
 634
 635   /* "id" property (SARIF v2.1.0 section 3.49.3).  */
 636   reporting_desc->set_string ("id", option_text);
 637
 638   /* We don't implement "name" property (SARIF v2.1.0 section 3.49.7), since
 639      it seems redundant compared to "id".  */
 640
 641   /* "helpUri" property (SARIF v2.1.0 section 3.49.12).  */
 642   if (char *option_url = context->make_option_url (diagnostic->option_index))
 643     {
 644       reporting_desc->set_string ("helpUri", option_url);
 645       free (option_url);
 646     }
 647
 648   return reporting_desc;
 649 }
 650
 651 /* Make a reportingDescriptor object (SARIF v2.1.0 section 3.49)
 652    for CWE_ID, for use within the CWE taxa array.  */
 653
 654 json::object *
 655 sarif_builder::make_reporting_descriptor_object_for_cwe_id (int cwe_id) const
 656 {
 657   json::object *reporting_desc = new json::object ();
 658
 659   /* "id" property (SARIF v2.1.0 section 3.49.3).  */
 660   {
 661     pretty_printer pp;
 662     pp_printf (&pp, "%i", cwe_id);
 663     reporting_desc->set_string ("id", pp_formatted_text (&pp));
 664   }
 665
 666   /* "helpUri" property (SARIF v2.1.0 section 3.49.12).  */
 667   {
 668     char *url = get_cwe_url (cwe_id);
 669     reporting_desc->set_string ("helpUri", url);
 670     free (url);
 671   }
 672
 673   return reporting_desc;
 674 }
 675
 676 /* Make a reportingDescriptorReference object (SARIF v2.1.0 section 3.52)
 677    referencing CWE_ID, for use within a result object.
 678    Also, add CWE_ID to m_cwe_id_set.  */
 679
 680 json::object *
 681 sarif_builder::
 682 make_reporting_descriptor_reference_object_for_cwe_id (int cwe_id)
 683 {
 684   json::object *desc_ref_obj = new json::object ();
 685
 686   /* "id" property (SARIF v2.1.0 section 3.52.4).  */
 687   {
 688     pretty_printer pp;
 689     pp_printf (&pp, "%i", cwe_id);
 690     desc_ref_obj->set_string ("id", pp_formatted_text (&pp));
 691   }
 692
 693   /* "toolComponent" property (SARIF v2.1.0 section 3.52.7).  */
 694   json::object *comp_ref_obj = make_tool_component_reference_object_for_cwe ();
 695   desc_ref_obj->set ("toolComponent", comp_ref_obj);
 696
 697   /* Add CWE_ID to our set.  */
 698   gcc_assert (cwe_id > 0);
 699   m_cwe_id_set.add (cwe_id);
 700
 701   return desc_ref_obj;
 702 }
 703
 704 /* Make a toolComponentReference object (SARIF v2.1.0 section 3.54) that
 705    references the CWE taxonomy.  */
 706
 707 json::object *
 708 sarif_builder::
 709 make_tool_component_reference_object_for_cwe () const
 710 {
 711   json::object *comp_ref_obj = new json::object ();
 712
 713   /* "name" property  (SARIF v2.1.0 section 3.54.3).  */
 714   comp_ref_obj->set_string ("name", "cwe");
 715
 716   return comp_ref_obj;
 717 }
 718
 719 /* Make an array suitable for use as the "locations" property of:
 720    - a "result" object (SARIF v2.1.0 section 3.27.12), or
 721    - a "notification" object (SARIF v2.1.0 section 3.58.4).  */
 722
 723 json::array *
 724 sarif_builder::make_locations_arr (diagnostic_info *diagnostic)
 725 {
 726   json::array *locations_arr = new json::array ();
 727   const logical_location *logical_loc = NULL;
 728   if (auto client_data_hooks = m_context->get_client_data_hooks ())
 729     logical_loc = client_data_hooks->get_current_logical_location ();
 730
 731   json::object *location_obj
 732     = make_location_object (*diagnostic->richloc, logical_loc);
 733   locations_arr->append (location_obj);
 734   return locations_arr;
 735 }
 736
 737 /* If LOGICAL_LOC is non-NULL, use it to create a "logicalLocations" property
 738    within LOCATION_OBJ (SARIF v2.1.0 section 3.28.4).  */
 739
 740 void
 741 sarif_builder::
 742 set_any_logical_locs_arr (json::object *location_obj,
 743                           const logical_location *logical_loc)
 744 {
 745   if (!logical_loc)
 746     return;
 747   json::object *logical_loc_obj = make_logical_location_object (*logical_loc);
 748   json::array *location_locs_arr = new json::array ();
 749   location_locs_arr->append (logical_loc_obj);
 750   location_obj->set ("logicalLocations", location_locs_arr);
 751 }
 752
 753 /* Make a location object (SARIF v2.1.0 section 3.28) for RICH_LOC
 754    and LOGICAL_LOC.  */
 755
 756 json::object *
 757 sarif_builder::make_location_object (const rich_location &rich_loc,
 758                                      const logical_location *logical_loc)
 759 {
 760   json::object *location_obj = new json::object ();
 761
 762   /* Get primary loc from RICH_LOC.  */
 763   location_t loc = rich_loc.get_loc ();
 764
 765   /* "physicalLocation" property (SARIF v2.1.0 section 3.28.3).  */
 766   if (json::object *phs_loc_obj = maybe_make_physical_location_object (loc))
 767     location_obj->set ("physicalLocation", phs_loc_obj);
 768
 769   /* "logicalLocations" property (SARIF v2.1.0 section 3.28.4).  */
 770   set_any_logical_locs_arr (location_obj, logical_loc);
 771
 772   return location_obj;
 773 }
 774
 775 /* Make a location object (SARIF v2.1.0 section 3.28) for EVENT
 776    within a diagnostic_path.  */
 777
 778 json::object *
 779 sarif_builder::make_location_object (const diagnostic_event &event)
 780 {
 781   json::object *location_obj = new json::object ();
 782
 783   /* "physicalLocation" property (SARIF v2.1.0 section 3.28.3).  */
 784   location_t loc = event.get_location ();
 785   if (json::object *phs_loc_obj = maybe_make_physical_location_object (loc))
 786     location_obj->set ("physicalLocation", phs_loc_obj);
 787
 788   /* "logicalLocations" property (SARIF v2.1.0 section 3.28.4).  */
 789   const logical_location *logical_loc = event.get_logical_location ();
 790   set_any_logical_locs_arr (location_obj, logical_loc);
 791
 792   /* "message" property (SARIF v2.1.0 section 3.28.5).  */
 793   label_text ev_desc = event.get_desc (false);
 794   json::object *message_obj = make_message_object (ev_desc.get ());
 795   location_obj->set ("message", message_obj);
 796
 797   return location_obj;
 798 }
 799
 800 /* Make a physicalLocation object (SARIF v2.1.0 section 3.29) for LOC,
 801    or return NULL;
 802    Add any filename to the m_artifacts.  */
 803
 804 json::object *
 805 sarif_builder::maybe_make_physical_location_object (location_t loc)
 806 {
 807   if (loc <= BUILTINS_LOCATION || LOCATION_FILE (loc) == NULL)
 808     return NULL;
 809
 810   json::object *phys_loc_obj = new json::object ();
 811
 812   /* "artifactLocation" property (SARIF v2.1.0 section 3.29.3).  */
 813   json::object *artifact_loc_obj = make_artifact_location_object (loc);
 814   phys_loc_obj->set ("artifactLocation", artifact_loc_obj);
 815   m_filenames.add (LOCATION_FILE (loc));
 816
 817   /* "region" property (SARIF v2.1.0 section 3.29.4).  */
 818   if (json::object *region_obj = maybe_make_region_object (loc))
 819     phys_loc_obj->set ("region", region_obj);
 820
 821   /* "contextRegion" property (SARIF v2.1.0 section 3.29.5).  */
 822   if (json::object *context_region_obj
 823         = maybe_make_region_object_for_context (loc))
 824     phys_loc_obj->set ("contextRegion", context_region_obj);
 825
 826   /* Instead, we add artifacts to the run as a whole,
 827      with artifact.contents.
 828      Could do both, though.  */
 829
 830   return phys_loc_obj;
 831 }
 832
 833 /* Make an artifactLocation object (SARIF v2.1.0 section 3.4) for LOC,
 834    or return NULL.  */
 835
 836 json::object *
 837 sarif_builder::make_artifact_location_object (location_t loc)
 838 {
 839   return make_artifact_location_object (LOCATION_FILE (loc));
 840 }
 841
 842 /* The ID value for use in "uriBaseId" properties (SARIF v2.1.0 section 3.4.4)
 843    for when we need to express paths relative to PWD.  */
 844
 845 #define PWD_PROPERTY_NAME ("PWD")
 846
 847 /* Make an artifactLocation object (SARIF v2.1.0 section 3.4) for FILENAME,
 848    or return NULL.  */
 849
 850 json::object *
 851 sarif_builder::make_artifact_location_object (const char *filename)
 852 {
 853   json::object *artifact_loc_obj = new json::object ();
 854
 855   /* "uri" property (SARIF v2.1.0 section 3.4.3).  */
 856   artifact_loc_obj->set_string ("uri", filename);
 857
 858   if (filename[0] != '/')
 859     {
 860       /* If we have a relative path, set the "uriBaseId" property
 861          (SARIF v2.1.0 section 3.4.4).  */
 862       artifact_loc_obj->set_string ("uriBaseId", PWD_PROPERTY_NAME);
 863       m_seen_any_relative_paths = true;
 864     }
 865
 866   return artifact_loc_obj;
 867 }
 868
 869 /* Get the PWD, or NULL, as an absolute file-based URI,
 870    adding a trailing forward slash (as required by SARIF v2.1.0
 871    section 3.14.14).  */
 872
 873 static char *
 874 make_pwd_uri_str ()
 875 {
 876   /* The prefix of a file-based URI, up to, but not including the path. */
 877 #define FILE_PREFIX ("file://")
 878
 879   const char *pwd = getpwd ();
 880   if (!pwd)
 881     return NULL;
 882   size_t len = strlen (pwd);
 883   if (len == 0 || pwd[len - 1] != '/')
 884     return concat (FILE_PREFIX, pwd, "/", NULL);
 885   else
 886     {
 887       gcc_assert (pwd[len - 1] == '/');
 888       return concat (FILE_PREFIX, pwd, NULL);
 889     }
 890 }
 891
 892 /* Make an artifactLocation object (SARIF v2.1.0 section 3.4) for the pwd,
 893    for use in the "run.originalUriBaseIds" property (SARIF v2.1.0
 894    section 3.14.14) when we have any relative paths.  */
 895
 896 json::object *
 897 sarif_builder::make_artifact_location_object_for_pwd () const
 898 {
 899   json::object *artifact_loc_obj = new json::object ();
 900
 901   /* "uri" property (SARIF v2.1.0 section 3.4.3).  */
 902   if (char *pwd = make_pwd_uri_str ())
 903     {
 904       gcc_assert (strlen (pwd) > 0);
 905       gcc_assert (pwd[strlen (pwd) - 1] == '/');
 906       artifact_loc_obj->set_string ("uri", pwd);
 907       free (pwd);
 908     }
 909
 910   return artifact_loc_obj;
 911 }
 912
 913 /* Get the column number within EXPLOC.  */
 914
 915 int
 916 sarif_builder::get_sarif_column (expanded_location exploc) const
 917 {
 918   cpp_char_column_policy policy (m_tabstop, cpp_wcwidth);
 919   return location_compute_display_column (m_context->get_file_cache (),
 920                                           exploc, policy);
 921 }
 922
 923 /* Make a region object (SARIF v2.1.0 section 3.30) for LOC,
 924    or return NULL.  */
 925
 926 json::object *
 927 sarif_builder::maybe_make_region_object (location_t loc) const
 928 {
 929   location_t caret_loc = get_pure_location (loc);
 930
 931   if (caret_loc <= BUILTINS_LOCATION)
 932     return NULL;
 933
 934   location_t start_loc = get_start (loc);
 935   location_t finish_loc = get_finish (loc);
 936
 937   expanded_location exploc_caret = expand_location (caret_loc);
 938   expanded_location exploc_start = expand_location (start_loc);
 939   expanded_location exploc_finish = expand_location (finish_loc);
 940
 941   if (exploc_start.file !=exploc_caret.file)
 942     return NULL;
 943   if (exploc_finish.file !=exploc_caret.file)
 944     return NULL;
 945
 946   json::object *region_obj = new json::object ();
 947
 948   /* "startLine" property (SARIF v2.1.0 section 3.30.5) */
 949   region_obj->set_integer ("startLine", exploc_start.line);
 950
 951   /* "startColumn" property (SARIF v2.1.0 section 3.30.6) */
 952   region_obj->set_integer ("startColumn", get_sarif_column (exploc_start));
 953
 954   /* "endLine" property (SARIF v2.1.0 section 3.30.7) */
 955   if (exploc_finish.line != exploc_start.line)
 956     region_obj->set_integer ("endLine", exploc_finish.line);
 957
 958   /* "endColumn" property (SARIF v2.1.0 section 3.30.8).
 959      This expresses the column immediately beyond the range.  */
 960   {
 961     int next_column = sarif_builder::get_sarif_column (exploc_finish) + 1;
 962     region_obj->set_integer ("endColumn", next_column);
 963   }
 964
 965   return region_obj;
 966 }
 967
 968 /* Make a region object (SARIF v2.1.0 section 3.30) for the "contextRegion"
 969    property (SARIF v2.1.0 section 3.29.5) of a physicalLocation.
 970
 971    This is similar to maybe_make_region_object, but ignores column numbers,
 972    covering the line(s) as a whole, and including a "snippet" property
 973    embedding those source lines, making it easier for consumers to show
 974    the pertinent source.  */
 975
 976 json::object *
 977 sarif_builder::maybe_make_region_object_for_context (location_t loc) const
 978 {
 979   location_t caret_loc = get_pure_location (loc);
 980
 981   if (caret_loc <= BUILTINS_LOCATION)
 982     return NULL;
 983
 984   location_t start_loc = get_start (loc);
 985   location_t finish_loc = get_finish (loc);
 986
 987   expanded_location exploc_caret = expand_location (caret_loc);
 988   expanded_location exploc_start = expand_location (start_loc);
 989   expanded_location exploc_finish = expand_location (finish_loc);
 990
 991   if (exploc_start.file !=exploc_caret.file)
 992     return NULL;
 993   if (exploc_finish.file !=exploc_caret.file)
 994     return NULL;
 995
 996   json::object *region_obj = new json::object ();
 997
 998   /* "startLine" property (SARIF v2.1.0 section 3.30.5) */
 999   region_obj->set_integer ("startLine", exploc_start.line);
1000
1001   /* "endLine" property (SARIF v2.1.0 section 3.30.7) */
1002   if (exploc_finish.line != exploc_start.line)
1003     region_obj->set_integer ("endLine", exploc_finish.line);
1004
1005   /* "snippet" property (SARIF v2.1.0 section 3.30.13).  */
1006   if (json::object *artifact_content_obj
1007          = maybe_make_artifact_content_object (exploc_start.file,
1008                                                exploc_start.line,
1009                                                exploc_finish.line))
1010     region_obj->set ("snippet", artifact_content_obj);
1011
1012   return region_obj;
1013 }
1014
1015 /* Make a region object (SARIF v2.1.0 section 3.30) for the deletion region
1016    of HINT (as per SARIF v2.1.0 section 3.57.3).  */
1017
1018 json::object *
1019 sarif_builder::make_region_object_for_hint (const fixit_hint &hint) const
1020 {
1021   location_t start_loc = hint.get_start_loc ();
1022   location_t next_loc = hint.get_next_loc ();
1023
1024   expanded_location exploc_start = expand_location (start_loc);
1025   expanded_location exploc_next = expand_location (next_loc);
1026
1027   json::object *region_obj = new json::object ();
1028
1029   /* "startLine" property (SARIF v2.1.0 section 3.30.5) */
1030   region_obj->set_integer ("startLine", exploc_start.line);
1031
1032   /* "startColumn" property (SARIF v2.1.0 section 3.30.6) */
1033   int start_col = get_sarif_column (exploc_start);
1034   region_obj->set_integer ("startColumn", start_col);
1035
1036   /* "endLine" property (SARIF v2.1.0 section 3.30.7) */
1037   if (exploc_next.line != exploc_start.line)
1038     region_obj->set_integer ("endLine", exploc_next.line);
1039
1040   /* "endColumn" property (SARIF v2.1.0 section 3.30.8).
1041      This expresses the column immediately beyond the range.  */
1042   int next_col =  get_sarif_column (exploc_next);
1043   region_obj->set_integer ("endColumn", next_col);
1044
1045   return region_obj;
1046 }
1047
1048 /* Attempt to get a string for a logicalLocation's "kind" property
1049    (SARIF v2.1.0 section 3.33.7).
1050    Return NULL if unknown.  */
1051
1052 static const char *
1053 maybe_get_sarif_kind (enum logical_location_kind kind)
1054 {
1055   switch (kind)
1056     {
1057     default:
1058       gcc_unreachable ();
1059     case LOGICAL_LOCATION_KIND_UNKNOWN:
1060       return NULL;
1061
1062     case LOGICAL_LOCATION_KIND_FUNCTION:
1063       return "function";
1064     case LOGICAL_LOCATION_KIND_MEMBER:
1065       return "member";
1066     case LOGICAL_LOCATION_KIND_MODULE:
1067       return "module";
1068     case LOGICAL_LOCATION_KIND_NAMESPACE:
1069       return "namespace";
1070     case LOGICAL_LOCATION_KIND_TYPE:
1071       return "type";
1072     case LOGICAL_LOCATION_KIND_RETURN_TYPE:
1073       return "returnType";
1074     case LOGICAL_LOCATION_KIND_PARAMETER:
1075       return "parameter";
1076     case LOGICAL_LOCATION_KIND_VARIABLE:
1077       return "variable";
1078     }
1079 }
1080
1081 /* Make a logicalLocation object (SARIF v2.1.0 section 3.33) for LOGICAL_LOC,
1082    or return NULL.  */
1083
1084 json::object *
1085 sarif_builder::
1086 make_logical_location_object (const logical_location &logical_loc) const
1087 {
1088   json::object *logical_loc_obj = new json::object ();
1089
1090   /* "name" property (SARIF v2.1.0 section 3.33.4).  */
1091   if (const char *short_name = logical_loc.get_short_name ())
1092     logical_loc_obj->set_string ("name", short_name);
1093
1094   /* "fullyQualifiedName" property (SARIF v2.1.0 section 3.33.5).  */
1095   if (const char *name_with_scope = logical_loc.get_name_with_scope ())
1096     logical_loc_obj->set_string ("fullyQualifiedName", name_with_scope);
1097
1098   /* "decoratedName" property (SARIF v2.1.0 section 3.33.6).  */
1099   if (const char *internal_name = logical_loc.get_internal_name ())
1100     logical_loc_obj->set_string ("decoratedName", internal_name);
1101
1102   /* "kind" property (SARIF v2.1.0 section 3.33.7).  */
1103   enum logical_location_kind kind = logical_loc.get_kind ();
1104   if (const char *sarif_kind_str = maybe_get_sarif_kind (kind))
1105     logical_loc_obj->set_string ("kind", sarif_kind_str);
1106
1107   return logical_loc_obj;
1108 }
1109
1110 /* Make a codeFlow object (SARIF v2.1.0 section 3.36) for PATH.  */
1111
1112 json::object *
1113 sarif_builder::make_code_flow_object (const diagnostic_path &path)
1114 {
1115   json::object *code_flow_obj = new json::object ();
1116
1117   /* "threadFlows" property (SARIF v2.1.0 section 3.36.3).  */
1118   json::array *thread_flows_arr = new json::array ();
1119
1120   /* Walk the events, consolidating into per-thread threadFlow objects,
1121      using the index with PATH as the overall executionOrder.  */
1122   hash_map<int_hash<diagnostic_thread_id_t, -1, -2>,
1123            sarif_thread_flow *> thread_id_map;
1124   for (unsigned i = 0; i < path.num_events (); i++)
1125     {
1126       const diagnostic_event &event = path.get_event (i);
1127       const diagnostic_thread_id_t thread_id = event.get_thread_id ();
1128       sarif_thread_flow *thread_flow_obj;
1129
1130       if (sarif_thread_flow **slot = thread_id_map.get (thread_id))
1131         thread_flow_obj = *slot;
1132       else
1133         {
1134           const diagnostic_thread &thread = path.get_thread (thread_id);
1135           thread_flow_obj = new sarif_thread_flow (thread);
1136           thread_flows_arr->append (thread_flow_obj);
1137           thread_id_map.put (thread_id, thread_flow_obj);
1138         }
1139
1140       /* Add event to thread's threadFlow object.  */
1141       json::object *thread_flow_loc_obj
1142         = make_thread_flow_location_object (event, i);
1143       thread_flow_obj->add_location (thread_flow_loc_obj);
1144     }
1145   code_flow_obj->set ("threadFlows", thread_flows_arr);
1146
1147   return code_flow_obj;
1148 }
1149
1150 /* Make a threadFlowLocation object (SARIF v2.1.0 section 3.38) for EVENT.  */
1151
1152 json::object *
1153 sarif_builder::make_thread_flow_location_object (const diagnostic_event &ev,
1154                                                  int path_event_idx)
1155 {
1156   json::object *thread_flow_loc_obj = new json::object ();
1157
1158   /* "location" property (SARIF v2.1.0 section 3.38.3).  */
1159   json::object *location_obj = make_location_object (ev);
1160   thread_flow_loc_obj->set ("location", location_obj);
1161
1162   /* "kinds" property (SARIF v2.1.0 section 3.38.8).  */
1163   diagnostic_event::meaning m = ev.get_meaning ();
1164   if (json::array *kinds_arr = maybe_make_kinds_array (m))
1165     thread_flow_loc_obj->set ("kinds", kinds_arr);
1166
1167   /* "nestingLevel" property (SARIF v2.1.0 section 3.38.10).  */
1168   thread_flow_loc_obj->set_integer ("nestingLevel", ev.get_stack_depth ());
1169
1170   /* "executionOrder" property (SARIF v2.1.0 3.38.11).
1171      Offset by 1 to match the human-readable values emitted by %@.  */
1172   thread_flow_loc_obj->set_integer ("executionOrder", path_event_idx + 1);
1173
1174   /* It might be nice to eventually implement the following for -fanalyzer:
1175      - the "stack" property (SARIF v2.1.0 section 3.38.5)
1176      - the "state" property (SARIF v2.1.0 section 3.38.9)
1177      - the "importance" property (SARIF v2.1.0 section 3.38.13).  */
1178
1179   return thread_flow_loc_obj;
1180 }
1181
1182 /* If M has any known meaning, make a json array suitable for the "kinds"
1183    property of a threadFlowLocation object (SARIF v2.1.0 section 3.38.8).
1184
1185    Otherwise, return NULL.  */
1186
1187 json::array *
1188 sarif_builder::maybe_make_kinds_array (diagnostic_event::meaning m) const
1189 {
1190   if (m.m_verb == diagnostic_event::VERB_unknown
1191       && m.m_noun == diagnostic_event::NOUN_unknown
1192       && m.m_property == diagnostic_event::PROPERTY_unknown)
1193     return NULL;
1194
1195   json::array *kinds_arr = new json::array ();
1196   if (const char *verb_str
1197         = diagnostic_event::meaning::maybe_get_verb_str (m.m_verb))
1198     kinds_arr->append (new json::string (verb_str));
1199   if (const char *noun_str
1200         = diagnostic_event::meaning::maybe_get_noun_str (m.m_noun))
1201     kinds_arr->append (new json::string (noun_str));
1202   if (const char *property_str
1203         = diagnostic_event::meaning::maybe_get_property_str (m.m_property))
1204     kinds_arr->append (new json::string (property_str));
1205   return kinds_arr;
1206 }
1207
1208 /* Make a message object (SARIF v2.1.0 section 3.11) for MSG.  */
1209
1210 json::object *
1211 sarif_builder::make_message_object (const char *msg) const
1212 {
1213   json::object *message_obj = new json::object ();
1214
1215   /* "text" property (SARIF v2.1.0 section 3.11.8).  */
1216   message_obj->set_string ("text", msg);
1217
1218   return message_obj;
1219 }
1220
1221 /* Make a message object (SARIF v2.1.0 section 3.11) for DIAGRAM.
1222    We emit the diagram as a code block within the Markdown part
1223    of the message.  */
1224
1225 json::object *
1226 sarif_builder::make_message_object_for_diagram (diagnostic_context *context,
1227                                                 const diagnostic_diagram &diagram)
1228 {
1229   json::object *message_obj = new json::object ();
1230
1231   /* "text" property (SARIF v2.1.0 section 3.11.8).  */
1232   message_obj->set_string ("text", diagram.get_alt_text ());
1233
1234   char *saved_prefix = pp_take_prefix (context->printer);
1235   pp_set_prefix (context->printer, NULL);
1236
1237   /* "To produce a code block in Markdown, simply indent every line of
1238      the block by at least 4 spaces or 1 tab."
1239      Here we use 4 spaces.  */
1240   diagram.get_canvas ().print_to_pp (context->printer, "    ");
1241   pp_set_prefix (context->printer, saved_prefix);
1242
1243   /* "markdown" property (SARIF v2.1.0 section 3.11.9).  */
1244   message_obj->set_string ("markdown", pp_formatted_text (context->printer));
1245
1246   pp_clear_output_area (context->printer);
1247
1248   return message_obj;
1249 }
1250
1251 /* Make a multiformatMessageString object (SARIF v2.1.0 section 3.12)
1252    for MSG.  */
1253
1254 json::object *
1255 sarif_builder::make_multiformat_message_string (const char *msg) const
1256 {
1257   json::object *message_obj = new json::object ();
1258
1259   /* "text" property (SARIF v2.1.0 section 3.12.3).  */
1260   message_obj->set_string ("text", msg);
1261
1262   return message_obj;
1263 }
1264
1265 #define SARIF_SCHEMA "https://raw.githubusercontent.com/oasis-tcs/sarif-spec/master/Schemata/sarif-schema-2.1.0.json"
1266 #define SARIF_VERSION "2.1.0"
1267
1268 /* Make a top-level sarifLog object (SARIF v2.1.0 section 3.13).
1269    Take ownership of INVOCATION_OBJ and RESULTS.  */
1270
1271 json::object *
1272 sarif_builder::make_top_level_object (sarif_invocation *invocation_obj,
1273                                       json::array *results)
1274 {
1275   json::object *log_obj = new json::object ();
1276
1277   /* "$schema" property (SARIF v2.1.0 section 3.13.3) .  */
1278   log_obj->set_string ("$schema", SARIF_SCHEMA);
1279
1280   /* "version" property (SARIF v2.1.0 section 3.13.2).  */
1281   log_obj->set_string ("version", SARIF_VERSION);
1282
1283   /* "runs" property (SARIF v2.1.0 section 3.13.4).  */
1284   json::array *run_arr = new json::array ();
1285   json::object *run_obj = make_run_object (invocation_obj, results);
1286   run_arr->append (run_obj);
1287   log_obj->set ("runs", run_arr);
1288
1289   return log_obj;
1290 }
1291
1292 /* Make a run object (SARIF v2.1.0 section 3.14).
1293    Take ownership of INVOCATION_OBJ and RESULTS.  */
1294
1295 json::object *
1296 sarif_builder::make_run_object (sarif_invocation *invocation_obj,
1297                                 json::array *results)
1298 {
1299   json::object *run_obj = new json::object ();
1300
1301   /* "tool" property (SARIF v2.1.0 section 3.14.6).  */
1302   json::object *tool_obj = make_tool_object ();
1303   run_obj->set ("tool", tool_obj);
1304
1305   /* "taxonomies" property (SARIF v2.1.0 section 3.14.8).  */
1306   if (json::array *taxonomies_arr = maybe_make_taxonomies_array ())
1307     run_obj->set ("taxonomies", taxonomies_arr);
1308
1309   /* "invocations" property (SARIF v2.1.0 section 3.14.11).  */
1310   {
1311     json::array *invocations_arr = new json::array ();
1312     invocations_arr->append (invocation_obj);
1313     run_obj->set ("invocations", invocations_arr);
1314   }
1315
1316   /* "originalUriBaseIds (SARIF v2.1.0 section 3.14.14).  */
1317   if (m_seen_any_relative_paths)
1318     {
1319       json::object *orig_uri_base_ids = new json::object ();
1320       run_obj->set ("originalUriBaseIds", orig_uri_base_ids);
1321       json::object *pwd_art_loc_obj = make_artifact_location_object_for_pwd ();
1322       orig_uri_base_ids->set (PWD_PROPERTY_NAME, pwd_art_loc_obj);
1323     }
1324
1325   /* "artifacts" property (SARIF v2.1.0 section 3.14.15).  */
1326   json::array *artifacts_arr = new json::array ();
1327   for (auto iter : m_filenames)
1328     {
1329       json::object *artifact_obj = make_artifact_object (iter);
1330       artifacts_arr->append (artifact_obj);
1331     }
1332   run_obj->set ("artifacts", artifacts_arr);
1333
1334   /* "results" property (SARIF v2.1.0 section 3.14.23).  */
1335   run_obj->set ("results", results);
1336
1337   return run_obj;
1338 }
1339
1340 /* Make a tool object (SARIF v2.1.0 section 3.18).  */
1341
1342 json::object *
1343 sarif_builder::make_tool_object () const
1344 {
1345   json::object *tool_obj = new json::object ();
1346
1347   /* "driver" property (SARIF v2.1.0 section 3.18.2).  */
1348   json::object *driver_obj = make_driver_tool_component_object ();
1349   tool_obj->set ("driver", driver_obj);
1350
1351   /* Report plugins via the "extensions" property
1352      (SARIF v2.1.0 section 3.18.3).  */
1353   if (auto client_data_hooks = m_context->get_client_data_hooks ())
1354     if (const client_version_info *vinfo
1355           = client_data_hooks->get_any_version_info ())
1356       {
1357         class my_plugin_visitor : public client_version_info :: plugin_visitor
1358         {
1359         public:
1360           void on_plugin (const diagnostic_client_plugin_info &p) final override
1361           {
1362             /* Create a toolComponent object (SARIF v2.1.0 section 3.19)
1363                for the plugin.  */
1364             json::object *plugin_obj = new json::object ();
1365             m_plugin_objs.safe_push (plugin_obj);
1366
1367             /* "name" property (SARIF v2.1.0 section 3.19.8).  */
1368             if (const char *short_name = p.get_short_name ())
1369               plugin_obj->set_string ("name", short_name);
1370
1371             /* "fullName" property (SARIF v2.1.0 section 3.19.9).  */
1372             if (const char *full_name = p.get_full_name ())
1373               plugin_obj->set_string ("fullName", full_name);
1374
1375             /* "version" property (SARIF v2.1.0 section 3.19.13).  */
1376             if (const char *version = p.get_version ())
1377               plugin_obj->set_string ("version", version);
1378           }
1379           auto_vec <json::object *> m_plugin_objs;
1380         };
1381         my_plugin_visitor v;
1382         vinfo->for_each_plugin (v);
1383         if (v.m_plugin_objs.length () > 0)
1384           {
1385             json::array *extensions_arr = new json::array ();
1386             tool_obj->set ("extensions", extensions_arr);
1387             for (auto iter : v.m_plugin_objs)
1388               extensions_arr->append (iter);
1389           }
1390       }
1391
1392   /* Perhaps we could also show GMP, MPFR, MPC, isl versions as other
1393      "extensions" (see toplev.cc: print_version).  */
1394
1395   return tool_obj;
1396 }
1397
1398 /* Make a toolComponent object (SARIF v2.1.0 section 3.19) for what SARIF
1399    calls the "driver" (see SARIF v2.1.0 section 3.18.1).  */
1400
1401 json::object *
1402 sarif_builder::make_driver_tool_component_object () const
1403 {
1404   json::object *driver_obj = new json::object ();
1405
1406   if (auto client_data_hooks = m_context->get_client_data_hooks ())
1407     if (const client_version_info *vinfo
1408           = client_data_hooks->get_any_version_info ())
1409       {
1410         /* "name" property (SARIF v2.1.0 section 3.19.8).  */
1411         if (const char *name = vinfo->get_tool_name ())
1412           driver_obj->set_string ("name", name);
1413
1414         /* "fullName" property (SARIF v2.1.0 section 3.19.9).  */
1415         if (char *full_name = vinfo->maybe_make_full_name ())
1416           {
1417             driver_obj->set_string ("fullName", full_name);
1418             free (full_name);
1419           }
1420
1421         /* "version" property (SARIF v2.1.0 section 3.19.13).  */
1422         if (const char *version = vinfo->get_version_string ())
1423           driver_obj->set_string ("version", version);
1424
1425         /* "informationUri" property (SARIF v2.1.0 section 3.19.17).  */
1426         if (char *version_url =  vinfo->maybe_make_version_url ())
1427           {
1428             driver_obj->set_string ("informationUri", version_url);
1429             free (version_url);
1430           }
1431       }
1432
1433   /* "rules" property (SARIF v2.1.0 section 3.19.23).  */
1434   driver_obj->set ("rules", m_rules_arr);
1435
1436   return driver_obj;
1437 }
1438
1439 /* If we've seen any CWE IDs, make an array for the "taxonomies" property
1440    (SARIF v2.1.0 section 3.14.8) of a run object, containting a singl
1441    toolComponent (3.19) as per 3.19.3, representing the CWE.
1442
1443    Otherwise return NULL.  */
1444
1445 json::array *
1446 sarif_builder::maybe_make_taxonomies_array () const
1447 {
1448   json::object *cwe_obj = maybe_make_cwe_taxonomy_object ();
1449   if (!cwe_obj)
1450     return NULL;
1451
1452   /* "taxonomies" property (SARIF v2.1.0 section 3.14.8).  */
1453   json::array *taxonomies_arr = new json::array ();
1454   taxonomies_arr->append (cwe_obj);
1455   return taxonomies_arr;
1456 }
1457
1458 /* If we've seen any CWE IDs, make a toolComponent object
1459    (SARIF v2.1.0 section 3.19) representing the CWE taxonomy, as per 3.19.3.
1460    Populate the "taxa" property with all of the CWE IDs in m_cwe_id_set.
1461
1462    Otherwise return NULL.  */
1463
1464 json::object *
1465 sarif_builder::maybe_make_cwe_taxonomy_object () const
1466 {
1467   if (m_cwe_id_set.is_empty ())
1468     return NULL;
1469
1470   json::object *taxonomy_obj = new json::object ();
1471
1472   /* "name" property (SARIF v2.1.0 section 3.19.8).  */
1473   taxonomy_obj->set_string ("name", "CWE");
1474
1475   /* "version" property (SARIF v2.1.0 section 3.19.13).  */
1476   taxonomy_obj->set_string ("version", "4.7");
1477
1478   /* "organization" property (SARIF v2.1.0 section 3.19.18).  */
1479   taxonomy_obj->set_string ("organization", "MITRE");
1480
1481   /* "shortDescription" property (SARIF v2.1.0 section 3.19.19).  */
1482   json::object *short_desc
1483     = make_multiformat_message_string ("The MITRE"
1484                                        " Common Weakness Enumeration");
1485   taxonomy_obj->set ("shortDescription", short_desc);
1486
1487   /* "taxa" property (SARIF v2.1.0 3.section 3.19.25).  */
1488   json::array *taxa_arr = new json::array ();
1489   for (auto cwe_id : m_cwe_id_set)
1490     {
1491       json::object *cwe_taxon
1492         = make_reporting_descriptor_object_for_cwe_id (cwe_id);
1493       taxa_arr->append (cwe_taxon);
1494     }
1495   taxonomy_obj->set ("taxa", taxa_arr);
1496
1497   return taxonomy_obj;
1498 }
1499
1500 /* Make an artifact object (SARIF v2.1.0 section 3.24).  */
1501
1502 json::object *
1503 sarif_builder::make_artifact_object (const char *filename)
1504 {
1505   json::object *artifact_obj = new json::object ();
1506
1507   /* "location" property (SARIF v2.1.0 section 3.24.2).  */
1508   json::object *artifact_loc_obj = make_artifact_location_object (filename);
1509   artifact_obj->set ("location", artifact_loc_obj);
1510
1511   /* "contents" property (SARIF v2.1.0 section 3.24.8).  */
1512   if (json::object *artifact_content_obj
1513         = maybe_make_artifact_content_object (filename))
1514     artifact_obj->set ("contents", artifact_content_obj);
1515
1516   /* "sourceLanguage" property (SARIF v2.1.0 section 3.24.10).  */
1517   if (auto client_data_hooks = m_context->get_client_data_hooks ())
1518     if (const char *source_lang
1519         = client_data_hooks->maybe_get_sarif_source_language (filename))
1520       artifact_obj->set_string ("sourceLanguage", source_lang);
1521
1522   return artifact_obj;
1523 }
1524
1525 /* Make an artifactContent object (SARIF v2.1.0 section 3.3) for the
1526    full contents of FILENAME.  */
1527
1528 json::object *
1529 sarif_builder::maybe_make_artifact_content_object (const char *filename) const
1530 {
1531   /* Let input.cc handle any charset conversion.  */
1532   char_span utf8_content
1533     = m_context->get_file_cache ().get_source_file_content (filename);
1534   if (!utf8_content)
1535     return NULL;
1536
1537   /* Don't add it if it's not valid UTF-8.  */
1538   if (!cpp_valid_utf8_p(utf8_content.get_buffer (), utf8_content.length ()))
1539     return NULL;
1540
1541   json::object *artifact_content_obj = new json::object ();
1542   artifact_content_obj->set ("text",
1543                              new json::string (utf8_content.get_buffer (),
1544                                                utf8_content.length ()));
1545   return artifact_content_obj;
1546 }
1547
1548 /* Attempt to read the given range of lines from FILENAME; return
1549    a freshly-allocated 0-terminated buffer containing them, or NULL.  */
1550
1551 char *
1552 sarif_builder::get_source_lines (const char *filename,
1553                                  int start_line,
1554                                  int end_line) const
1555 {
1556   auto_vec<char> result;
1557
1558   for (int line = start_line; line <= end_line; line++)
1559     {
1560       char_span line_content
1561         = m_context->get_file_cache ().get_source_line (filename, line);
1562       if (!line_content.get_buffer ())
1563         return NULL;
1564       result.reserve (line_content.length () + 1);
1565       for (size_t i = 0; i < line_content.length (); i++)
1566         result.quick_push (line_content[i]);
1567       result.quick_push ('\n');
1568     }
1569   result.safe_push ('\0');
1570
1571   return xstrdup (result.address ());
1572 }
1573
1574 /* Make an artifactContent object (SARIF v2.1.0 section 3.3) for the given
1575    run of lines within FILENAME (including the endpoints).  */
1576
1577 json::object *
1578 sarif_builder::maybe_make_artifact_content_object (const char *filename,
1579                                                    int start_line,
1580                                                    int end_line) const
1581 {
1582   char *text_utf8 = get_source_lines (filename, start_line, end_line);
1583
1584   if (!text_utf8)
1585     return NULL;
1586
1587   /* Don't add it if it's not valid UTF-8.  */
1588   if (!cpp_valid_utf8_p(text_utf8, strlen(text_utf8)))
1589     {
1590       free (text_utf8);
1591       return NULL;
1592     }
1593
1594   json::object *artifact_content_obj = new json::object ();
1595   artifact_content_obj->set_string ("text", text_utf8);
1596   free (text_utf8);
1597
1598   return artifact_content_obj;
1599 }
1600
1601 /* Make a fix object (SARIF v2.1.0 section 3.55) for RICHLOC.  */
1602
1603 json::object *
1604 sarif_builder::make_fix_object (const rich_location &richloc)
1605 {
1606   json::object *fix_obj = new json::object ();
1607
1608   /* "artifactChanges" property (SARIF v2.1.0 section 3.55.3).  */
1609   /* We assume that all fix-it hints in RICHLOC affect the same file.  */
1610   json::array *artifact_change_arr = new json::array ();
1611   json::object *artifact_change_obj = make_artifact_change_object (richloc);
1612   artifact_change_arr->append (artifact_change_obj);
1613   fix_obj->set ("artifactChanges", artifact_change_arr);
1614
1615   return fix_obj;
1616 }
1617
1618 /* Make an artifactChange object (SARIF v2.1.0 section 3.56) for RICHLOC.  */
1619
1620 json::object *
1621 sarif_builder::make_artifact_change_object (const rich_location &richloc)
1622 {
1623   json::object *artifact_change_obj = new json::object ();
1624
1625   /* "artifactLocation" property (SARIF v2.1.0 section 3.56.2).  */
1626   json::object *artifact_location_obj
1627     = make_artifact_location_object (richloc.get_loc ());
1628   artifact_change_obj->set ("artifactLocation", artifact_location_obj);
1629
1630   /* "replacements" property (SARIF v2.1.0 section 3.56.3).  */
1631   json::array *replacement_arr = new json::array ();
1632   for (unsigned int i = 0; i < richloc.get_num_fixit_hints (); i++)
1633     {
1634       const fixit_hint *hint = richloc.get_fixit_hint (i);
1635       json::object *replacement_obj = make_replacement_object (*hint);
1636       replacement_arr->append (replacement_obj);
1637     }
1638   artifact_change_obj->set ("replacements", replacement_arr);
1639
1640   return artifact_change_obj;
1641 }
1642
1643 /* Make a replacement object (SARIF v2.1.0 section 3.57) for HINT.  */
1644
1645 json::object *
1646 sarif_builder::make_replacement_object (const fixit_hint &hint) const
1647 {
1648   json::object *replacement_obj = new json::object ();
1649
1650   /* "deletedRegion" property (SARIF v2.1.0 section 3.57.3).  */
1651   json::object *region_obj = make_region_object_for_hint (hint);
1652   replacement_obj->set ("deletedRegion", region_obj);
1653
1654   /* "insertedContent" property (SARIF v2.1.0 section 3.57.4).  */
1655   json::object *content_obj = make_artifact_content_object (hint.get_string ());
1656   replacement_obj->set ("insertedContent", content_obj);
1657
1658   return replacement_obj;
1659 }
1660
1661 /* Make an artifactContent object (SARIF v2.1.0 section 3.3) for TEXT.  */
1662
1663 json::object *
1664 sarif_builder::make_artifact_content_object (const char *text) const
1665 {
1666   json::object *content_obj = new json::object ();
1667
1668   /* "text" property (SARIF v2.1.0 section 3.3.2).  */
1669   content_obj->set_string ("text", text);
1670
1671   return content_obj;
1672 }
1673
1674 /* Callback for diagnostic_context::ice_handler_cb for when an ICE
1675    occurs.  */
1676
1677 static void
1678 sarif_ice_handler (diagnostic_context *context)
1679 {
1680   /* Attempt to ensure that a .sarif file is written out.  */
1681   diagnostic_finish (context);
1682
1683   /* Print a header for the remaining output to stderr, and
1684      return, attempting to print the usual ICE messages to
1685      stderr.  Hopefully this will be helpful to the user in
1686      indicating what's gone wrong (also for DejaGnu, for pruning
1687      those messages).   */
1688   fnotice (stderr, "Internal compiler error:\n");
1689 }
1690
1691 class sarif_output_format : public diagnostic_output_format
1692 {
1693 public:
1694   void on_begin_group () final override
1695   {
1696     /* No-op,  */
1697   }
1698   void on_end_group () final override
1699   {
1700     m_builder.end_group ();
1701   }
1702   void
1703   on_begin_diagnostic (diagnostic_info *) final override
1704   {
1705     /* No-op,  */
1706   }
1707   void
1708   on_end_diagnostic (diagnostic_info *diagnostic,
1709                      diagnostic_t orig_diag_kind) final override
1710   {
1711     m_builder.end_diagnostic (&m_context, diagnostic, orig_diag_kind);
1712   }
1713   void on_diagram (const diagnostic_diagram &diagram) final override
1714   {
1715     m_builder.emit_diagram (&m_context, diagram);
1716   }
1717
1718 protected:
1719   sarif_output_format (diagnostic_context &context)
1720   : diagnostic_output_format (context),
1721     m_builder (&context)
1722   {}
1723
1724   sarif_builder m_builder;
1725 };
1726
1727 class sarif_stream_output_format : public sarif_output_format
1728 {
1729 public:
1730   sarif_stream_output_format (diagnostic_context &context, FILE *stream)
1731   : sarif_output_format (context),
1732     m_stream (stream)
1733   {
1734   }
1735   ~sarif_stream_output_format ()
1736   {
1737     m_builder.flush_to_file (m_stream);
1738   }
1739 private:
1740   FILE *m_stream;
1741 };
1742
1743 class sarif_file_output_format : public sarif_output_format
1744 {
1745 public:
1746   sarif_file_output_format (diagnostic_context &context,
1747                            const char *base_file_name)
1748   : sarif_output_format (context),
1749     m_base_file_name (xstrdup (base_file_name))
1750   {
1751   }
1752   ~sarif_file_output_format ()
1753   {
1754     char *filename = concat (m_base_file_name, ".sarif", NULL);
1755     free (m_base_file_name);
1756     m_base_file_name = nullptr;
1757     FILE *outf = fopen (filename, "w");
1758     if (!outf)
1759       {
1760         const char *errstr = xstrerror (errno);
1761         fnotice (stderr, "error: unable to open '%s' for writing: %s\n",
1762                  filename, errstr);
1763         free (filename);
1764         return;
1765       }
1766     m_builder.flush_to_file (outf);
1767     fclose (outf);
1768     free (filename);
1769   }
1770
1771 private:
1772   char *m_base_file_name;
1773 };
1774
1775 /* Populate CONTEXT in preparation for SARIF output (either to stderr, or
1776    to a file).  */
1777
1778 static void
1779 diagnostic_output_format_init_sarif (diagnostic_context *context)
1780 {
1781   /* Override callbacks.  */
1782   context->m_print_path = nullptr; /* handled in sarif_end_diagnostic.  */
1783   context->set_ice_handler_callback (sarif_ice_handler);
1784
1785   /* The metadata is handled in SARIF format, rather than as text.  */
1786   context->set_show_cwe (false);
1787   context->set_show_rules (false);
1788
1789   /* The option is handled in SARIF format, rather than as text.  */
1790   context->set_show_option_requested (false);
1791
1792   /* Don't colorize the text.  */
1793   pp_show_color (context->printer) = false;
1794 }
1795
1796 /* Populate CONTEXT in preparation for SARIF output to stderr.  */
1797
1798 void
1799 diagnostic_output_format_init_sarif_stderr (diagnostic_context *context)
1800 {
1801   diagnostic_output_format_init_sarif (context);
1802   context->set_output_format (new sarif_stream_output_format (*context,
1803                                                               stderr));
1804 }
1805
1806 /* Populate CONTEXT in preparation for SARIF output to a file named
1807    BASE_FILE_NAME.sarif.  */
1808
1809 void
1810 diagnostic_output_format_init_sarif_file (diagnostic_context *context,
1811                                           const char *base_file_name)
1812 {
1813   diagnostic_output_format_init_sarif (context);
1814   context->set_output_format (new sarif_file_output_format (*context,
1815                                                             base_file_name));
1816 }
1817
1818 /* Populate CONTEXT in preparation for SARIF output to STREAM.  */
1819
1820 void
1821 diagnostic_output_format_init_sarif_stream (diagnostic_context *context,
1822                                             FILE *stream)
1823 {
1824   diagnostic_output_format_init_sarif (context);
1825   context->set_output_format (new sarif_stream_output_format (*context,
1826                                                               stream));
1827 }