From a837d5f4ceddee27006de97bcbaa498756032c81 Mon Sep 17 00:00:00 2001
From: Max Schaefer <max@semmle.com>
Date: Thu, 9 Apr 2020 15:06:59 +0100
Subject: [PATCH] Add support for extracting Go frontend errors.

---
 extractor/dbscheme/tables.go | 28 +++++++++++++++++++
 extractor/extractor.go       | 52 +++++++++++++++++++++++++++++++-----
 ql/src/go.dbscheme           | 10 +++++++
 3 files changed, 83 insertions(+), 7 deletions(-)

diff --git a/extractor/dbscheme/tables.go b/extractor/dbscheme/tables.go
index c82fa0459b1..df146c8e4e5 100644
--- a/extractor/dbscheme/tables.go
+++ b/extractor/dbscheme/tables.go
@@ -4,6 +4,7 @@ import (
 	"go/ast"
 	"go/token"
 	gotypes "go/types"
+	"golang.org/x/tools/go/packages"
 )
 
 var defaultSnippet = AddDefaultSnippet(`
@@ -642,6 +643,20 @@ var ModLParenType = ModExprKind.NewBranch("@modlparen")
 // ModRParenType is the type of go.mod line block end AST nodes
 var ModRParenType = ModExprKind.NewBranch("@modrparen")
 
+// ErrorType is the type of frontend errors
+var ErrorType = NewPrimaryKeyType("@error")
+
+// ErrorKind is a case type for distinguishing different kinds of frontend errors
+var ErrorKind = NewCaseType(ErrorType, "kind")
+
+// ErrorTypes is a map from error kinds to the corresponding type
+var ErrorTypes = map[packages.ErrorKind]*BranchType{
+	packages.UnknownError: ErrorKind.NewBranch("@unknownerror"),
+	packages.ListError:    ErrorKind.NewBranch("@listerror"),
+	packages.ParseError:   ErrorKind.NewBranch("@parseerror"),
+	packages.TypeError:    ErrorKind.NewBranch("@typeerror"),
+}
+
 // LocationsDefaultTable is the table defining location objects
 var LocationsDefaultTable = NewTable("locations_default",
 	EntityColumn(LocationDefaultType, "id").Key(),
@@ -915,3 +930,16 @@ var ModTokensTable = NewTable("modtokens",
 	EntityColumn(ModExprType, "parent"),
 	IntColumn("idx"),
 ).KeySet("parent", "idx")
+
+// ErrorsTable is the table describing frontend errors
+var ErrorsTable = NewTable("errors",
+	EntityColumn(ErrorType, "id").Key(),
+	IntColumn("kind"),
+	StringColumn("msg"),
+	StringColumn("rawpos"),
+	StringColumn("file"),
+	IntColumn("line"),
+	IntColumn("col"),
+	EntityColumn(PackageType, "package"),
+	IntColumn("idx"),
+).KeySet("package", "idx")
diff --git a/extractor/extractor.go b/extractor/extractor.go
index f59ee9c228c..8fe96147808 100644
--- a/extractor/extractor.go
+++ b/extractor/extractor.go
@@ -57,13 +57,6 @@ func ExtractWithFlags(buildFlags []string, patterns []string) error {
 	packages.Visit(pkgs, func(pkg *packages.Package) bool {
 		return true
 	}, func(pkg *packages.Package) {
-		if len(pkg.Errors) != 0 {
-			log.Printf("Warning: encountered errors extracting package `%s`:", pkg.PkgPath)
-			for _, err := range pkg.Errors {
-				log.Printf("  %s", err.Error())
-			}
-		}
-
 		tw, err := trap.NewWriter(pkg.PkgPath, pkg)
 		if err != nil {
 			log.Fatal(err)
@@ -74,6 +67,14 @@ func ExtractWithFlags(buildFlags []string, patterns []string) error {
 		tw.ForEachObject(extractObjectType)
 		lbl := tw.Labeler.GlobalID(pkg.PkgPath + ";pkg")
 		dbscheme.PackagesTable.Emit(tw, lbl, pkg.Name, pkg.PkgPath, scope)
+
+		if len(pkg.Errors) != 0 {
+			log.Printf("Warning: encountered errors extracting package `%s`:", pkg.PkgPath)
+			for i, err := range pkg.Errors {
+				log.Printf("  %s", err.Error())
+				extractError(tw, err, lbl, i)
+			}
+		}
 	})
 
 	// this sets the number of threads that the Go runtime will spawn; this is separate
@@ -253,6 +254,43 @@ func extractObjectType(tw *trap.Writer, obj types.Object, lbl trap.Label) {
 	}
 }
 
+// extractError extracts the message and location of a frontend error
+func extractError(tw *trap.Writer, err packages.Error, pkglbl trap.Label, idx int) {
+	var (
+		lbl           = tw.Labeler.FreshID()
+		kind          = dbscheme.ErrorTypes[err.Kind].Index()
+		pos           = err.Pos
+		posComponents = strings.Split(err.Pos, ":")
+		file          = ""
+		line          = 0
+		col           = 0
+		e             error
+	)
+	switch len(posComponents) {
+	case 3:
+		// "file:line:col"
+		col, e = strconv.Atoi(posComponents[2])
+		if e != nil {
+			log.Printf("Warning: malformed column number `%s`: %v", posComponents[2], e)
+		}
+		fallthrough
+	case 2:
+		// "file:line"
+		file = posComponents[0]
+		line, e = strconv.Atoi(posComponents[1])
+		if e != nil {
+			log.Printf("Warning: malformed line number `%s`: %v", posComponents[1], e)
+		}
+	default:
+		// "", "-"
+		if pos != "" && pos != "-" {
+			log.Printf("Warning: malformed error position `%s`", pos)
+		}
+	}
+	file = filepath.ToSlash(srcarchive.TransformPath(file))
+	dbscheme.ErrorsTable.Emit(tw, lbl, kind, err.Msg, pos, file, line, col, pkglbl, idx)
+}
+
 // extractPackage extracts AST information for all files in the given package
 func extractPackage(pkg *packages.Package, wg *sync.WaitGroup,
 	goroutineSem *semaphore, fdSem *semaphore) {
diff --git a/ql/src/go.dbscheme b/ql/src/go.dbscheme
index f7fb4ff6229..ee5c327face 100644
--- a/ql/src/go.dbscheme
+++ b/ql/src/go.dbscheme
@@ -124,6 +124,10 @@ modexprs(unique int id: @modexpr, int kind: int ref, int parent: @modexprparent
 #keyset[parent, idx]
 modtokens(string token: string ref, int parent: @modexpr ref, int idx: int ref);
 
+#keyset[package, idx]
+errors(unique int id: @error, int kind: int ref, string msg: string ref, string rawpos: string ref,
+       string file: string ref, int line: int ref, int col: int ref, int package: @package ref, int idx: int ref);
+
 @container = @file | @folder;
 
 @locatable = @node | @localscope;
@@ -418,3 +422,9 @@ case @modexpr.kind of
 | 3 = @modlparen
 | 4 = @modrparen;
 
+case @error.kind of
+  0 = @unknownerror
+| 1 = @listerror
+| 2 = @parseerror
+| 3 = @typeerror;
+