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