cc: emit diagnostics as SARIF report

docs/Tools/vast-front.md

@@ -54,6 +54,9 @@ Additional customization options include:
   - After each pass that was specified as an option store MLIR into a file (format is `src.pass_name`).
   - `"*"` stores snapshot after every conversion.
 
+- `-vast-output-sarif="report.sarif"`
+  - Outputs diagnostics as a SARIF report file.
+
 ## Pipelines
 
 WIP pipelines documentation

include/vast/Config/config.h.cmake

@@ -19,6 +19,8 @@ namespace vast {
 
     constexpr std::string_view version = "${VAST_VERSION}";
 
+    constexpr std::string_view homepage_url = "${PROJECT_HOMEPAGE_URL}";
+
     constexpr std::string_view bug_report_url = "${BUG_REPORT_URL}";
 
     constexpr std::string_view default_resource_dir = "${VAST_DEFAULT_RESOURCE_DIR}";

include/vast/Frontend/Options.hpp

@@ -100,6 +100,8 @@ namespace vast::cc
         constexpr option_t vast_verify_diags = "verify-diags";
         constexpr option_t disable_emit_cxx_default = "disable-emit-cxx-default";
 
+        constexpr option_t output_sarif = "output-sarif";
+
         bool emit_only_mlir(const vast_args &vargs);
         bool emit_only_llvm(const vast_args &vargs);
     } // namespace opt

lib/vast/Frontend/CMakeLists.txt

@@ -8,6 +8,7 @@ add_vast_library(Frontend
     Targets.cpp
 
     LINK_LIBS PUBLIC
+    gap::gap-sarif
     MLIRBytecodeWriter
     VASTCodeGen
 )

lib/vast/Frontend/Consumer.cpp

@@ -1,6 +1,8 @@
 // Copyright (c) 2023-present, Trail of Bits, Inc.
 
 #include "vast/Frontend/Consumer.hpp"
+#include "mlir/IR/Diagnostics.h"
+#include "mlir/IR/Location.h"
 
 VAST_RELAX_WARNINGS
 #include <llvm/Support/Signals.h>
@@ -17,6 +19,7 @@ VAST_RELAX_WARNINGS
 VAST_UNRELAX_WARNINGS
 
 #include <filesystem>
+#include <fstream>
 
 #include "vast/CodeGen/CodeGenDriver.hpp"
 
@@ -27,6 +30,10 @@ VAST_UNRELAX_WARNINGS
 
 #include "vast/Target/LLVMIR/Convert.hpp"
 
+#include "vast/Config/config.h"
+
+#include <gap/sarif/sarif.hpp>
+
 namespace vast::cc {
 
     [[nodiscard]] target_dialect parse_target_dialect(string_ref from);
@@ -99,9 +106,7 @@ namespace vast::cc {
 
     void vast_consumer::HandleVTable(clang::CXXRecordDecl * /* decl */) { VAST_UNIMPLEMENTED; }
 
-    owning_module_ref vast_consumer::result() {
-        return driver->freeze();
-    }
+    owning_module_ref vast_consumer::result() { return driver->freeze(); }
 
     //
     // vast stream consumer
@@ -144,34 +149,42 @@ namespace vast::cc {
         }
     }
 
-    void vast_stream_consumer::emit_backend_output(
-        backend backend_action, owning_module_ref mod
-    ) {
+    void
+    vast_stream_consumer::emit_backend_output(backend backend_action, owning_module_ref mod) {
         llvm::LLVMContext llvm_context;
 
         process_mlir_module(target_dialect::llvm, mod.get());
 
         auto llvm_mod = target::llvmir::translate(mod.get(), llvm_context);
-        auto dl  = driver->acontext().getTargetInfo().getDataLayoutString();
+        auto dl       = driver->acontext().getTargetInfo().getDataLayoutString();
 
         clang::EmitBackendOutput(
             opts.diags, opts.headers, opts.codegen, opts.target, opts.lang, dl, llvm_mod.get(),
             backend_action, &opts.vfs, std::move(output_stream)
         );
     }
 
-    void vast_stream_consumer::process_mlir_module(
-        target_dialect target, mlir::ModuleOp mod
-    ) {
+    static gap::sarif::physical_location get_physical_loc(mlir::FileLineColLoc loc) {
+        std::filesystem::path file_path{ loc.getFilename().str() };
+        auto abs_path = std::filesystem::absolute(file_path);
+        return {
+            .artifactLocation{ { .uri{ "file://" + abs_path.string() } } },
+            .region{ {
+                .startLine   = loc.getLine(),
+                .startColumn = loc.getColumn(),
+            } },
+        };
+    }
+
+    void vast_stream_consumer::process_mlir_module(target_dialect target, mlir::ModuleOp mod) {
         // Handle source manager properly given that lifetime analysis
         // might emit warnings and remarks.
         auto &src_mgr     = driver->acontext().getSourceManager();
         auto &mctx        = driver->mcontext();
         auto main_file_id = src_mgr.getMainFileID();
 
-        auto file_buff = llvm::MemoryBuffer::getMemBuffer(
-            src_mgr.getBufferOrFake(main_file_id)
-        );
+        auto file_buff =
+            llvm::MemoryBuffer::getMemBuffer(src_mgr.getBufferOrFake(main_file_id));
 
         llvm::SourceMgr mlir_src_mgr;
         mlir_src_mgr.AddNewSourceBuffer(std::move(file_buff), llvm::SMLoc());
@@ -186,16 +199,93 @@ namespace vast::cc {
             llvm::DebugFlag = true;
         }
 
+        gap::sarif::run sarif_run{
+            .tool{
+                  .driver{
+                  .name           = "vast-front",
+                  .organization   = "Trail of Bits, inc.",
+                  .product        = "VAST",
+                  .version        = std::string{ vast::version },
+                  .informationUri = std::string{ vast::homepage_url },
+                  }, },
+            .invocations{
+                  {
+                  .arguments{ vargs.args.begin(), vargs.args.end() },
+                  .executionSuccessful = true,
+                  }, },
+        };
+
         // Setup and execute vast pipeline
         auto file_entry = src_mgr.getFileEntryRefForID(main_file_id);
         VAST_CHECK(file_entry, "failed to recover file entry ref");
-        auto snapshot_prefix = std::filesystem::path(file_entry->getName().str()).stem().string();
+        auto snapshot_prefix =
+            std::filesystem::path(file_entry->getName().str()).stem().string();
 
-        auto pipeline = setup_pipeline(pipeline_source::ast, target, mctx, vargs, snapshot_prefix);
+        auto pipeline =
+            setup_pipeline(pipeline_source::ast, target, mctx, vargs, snapshot_prefix);
         VAST_CHECK(pipeline, "failed to setup pipeline");
 
+        mlir::ScopedDiagnosticHandler sarif_diag_handler{
+            &mctx,
+            [&](mlir::Diagnostic &diag) {
+                gap::sarif::result result{
+                    .ruleId{ "mlir-diag" },
+                    .message{ .text = diag.str() },
+                };
+
+                auto loc = diag.getLocation();
+                if (auto flc_loc = loc.dyn_cast< mlir::FileLineColLoc >()) {
+                    result.locations.push_back({
+                        .physicalLocation{ get_physical_loc(flc_loc) },
+                    });
+                } else if (auto name_loc = loc.dyn_cast< mlir::NameLoc >()) {
+                    if (auto flc_child_loc =
+                            name_loc.getChildLoc().dyn_cast< mlir::FileLineColLoc >())
+                    {
+                        result.locations.push_back(
+                            { .physicalLocation{ get_physical_loc(flc_child_loc) },
+                              .logicalLocations{ { .name = name_loc.getName().str() } } }
+                        );
+                    }
+                }
+
+                switch (diag.getSeverity()) {
+                    using enum gap::sarif::level;
+                    case mlir::DiagnosticSeverity::Note:
+                        result.level = kNote;
+                        break;
+                    case mlir::DiagnosticSeverity::Warning:
+                        result.level = kWarning;
+                        break;
+                    case mlir::DiagnosticSeverity::Error:
+                        result.level = kError;
+                        break;
+                    case mlir::DiagnosticSeverity::Remark:
+                        result.level = kNote;
+                        break;
+                }
+                sarif_run.results.push_back(result);
+            }
+        };
+
         auto result = pipeline->run(mod);
-        VAST_CHECK(mlir::succeeded(result), "MLIR pass manager failed when running vast passes");
+
+        sarif_run.invocations[0].executionSuccessful = mlir::succeeded(result);
+
+        gap::sarif::root sarif_report{
+            .version = gap::sarif::version::k2_1_0,
+            .runs{ sarif_run },
+        };
+
+        if (auto sarif_path = vargs.get_option(opt::output_sarif)) {
+            std::ofstream os{ sarif_path.value().str() };
+            nlohmann::json j = sarif_report;
+            os << j;
+        }
+
+        VAST_CHECK(
+            mlir::succeeded(result), "MLIR pass manager failed when running vast passes"
+        );
 
         // Verify the diagnostic handler to make sure that each of the
         // diagnostics matched.
@@ -210,9 +300,7 @@ namespace vast::cc {
         // }
     }
 
-    void vast_stream_consumer::emit_mlir_output(
-        target_dialect target, owning_module_ref mod
-    ) {
+    void vast_stream_consumer::emit_mlir_output(target_dialect target, owning_module_ref mod) {
         if (!output_stream || !mod) {
             return;
         }

vcpkg.json

@@ -1,6 +1,11 @@
 {
     "$schema": "https://raw.githubusercontent.com/microsoft/vcpkg-tool/main/docs/vcpkg.schema.json",
     "dependencies": [
-        "gap"
+        {
+            "name": "gap",
+            "features": [
+                "sarif"
+            ]
+        }
     ]
 }