feat(go_indexer): add support for using file as top-level caller (#5637)

kythe/go/indexer/BUILD

@@ -258,6 +258,13 @@ go_indexer_test(
     import_path = "test/pkginit",
 )
 
+go_indexer_test(
+    name = "packageinit_file_test",
+    srcs = ["testdata/basic/packageinit_file.go"],
+    import_path = "test/pkginit",
+    use_file_as_top_level_scope = True,
+)
+
 go_indexer_test(
     name = "types_test",
     srcs = ["testdata/types.go"],

kythe/go/indexer/cmd/go_indexer/go_indexer.go

@@ -52,6 +52,7 @@ var (
 	verbose                        = flag.Bool("verbose", false, "Emit verbose log information")
 	contOnErr                      = flag.Bool("continue", false, "Log errors encountered during analysis but do not exit unsuccessfully")
 	useCompilationCorpusForAll     = flag.Bool("use_compilation_corpus_for_all", false, "If enabled, all Entry VNames are given the corpus of the compilation unit being indexed. This includes items in the go std library and builtin types.")
+	useFileAsTopLevelScope         = flag.Bool("use_file_as_top_level_scope", false, "If enabled, use the file node for top-level callsite scopes")
 	overrideStdlibCorpus           = flag.String("override_stdlib_corpus", "", "If set, all stdlib nodes are assigned this corpus. Note that this takes precedence over --use_compilation_corpus_for_all")
 
 	writeEntry func(context.Context, *spb.Entry) error
@@ -161,6 +162,7 @@ func indexGo(ctx context.Context, unit *apb.CompilationUnit, f indexer.Fetcher)
 		DocBase:                        docURL,
 		OnlyEmitDocURIsForStandardLibs: *onlyEmitDocURIsForStandardLibs,
 		UseCompilationCorpusForAll:     *useCompilationCorpusForAll,
+		UseFileAsTopLevelScope:         *useFileAsTopLevelScope,
 		OverrideStdlibCorpus:           *overrideStdlibCorpus,
 	})
 }

kythe/go/indexer/emit.go

@@ -58,6 +58,9 @@ type EmitOptions struct {
 	// If true, emit childof edges for an anchor's semantic scope.
 	EmitAnchorScopes bool
 
+	// If true, use the enclosing file for top-level callsite scopes.
+	UseFileAsTopLevelScope bool
+
 	// If set, use this as the base URL for links to godoc.  The import path is
 	// appended to the path of this URL to obtain the target URL to link to.
 	DocBase *url.URL
@@ -88,6 +91,13 @@ func (e *EmitOptions) emitAnchorScopes() bool {
 	return e.EmitAnchorScopes
 }
 
+func (e *EmitOptions) useFileAsTopLevelScope() bool {
+	if e == nil {
+		return false
+	}
+	return e.UseFileAsTopLevelScope
+}
+
 // shouldEmit reports whether the indexer should emit a node for the given
 // vname.  Presently this is true if vname denotes a standard library and the
 // corresponding option is enabled.
@@ -1258,6 +1268,9 @@ func (e *emitter) callContext(stack stackFunc) *funcInfo {
 		case *ast.FuncDecl, *ast.FuncLit:
 			return e.pi.function[p]
 		case *ast.File:
+			if e.opts.useFileAsTopLevelScope() {
+				return &funcInfo{vname: e.pi.FileVName(p)}
+			}
 			fi := e.pi.packageInit[p]
 			if fi == nil {
 				// Lazily emit a virtual node to represent the static

kythe/go/indexer/testdata/basic/packageinit_file.go

@@ -0,0 +1,12 @@
+// Package pkginit tests callgraph relationships for calls occurring at the
+// top-level file.
+package pkginit
+
+import "fmt"
+
+// Ensure that callsites at the package level are blamed on the file.
+
+//- A=@"fmt.Sprint(27)" ref/call _FmtSprint
+//- A childof File
+//- File.node/kind file
+var p = fmt.Sprint(27)

kythe/go/indexer/testdata/go_indexer_test.bzl

@@ -172,6 +172,9 @@ def _go_entries(ctx):
     if ctx.attr.use_compilation_corpus_for_all:
         iargs.append("-use_compilation_corpus_for_all")
 
+    if ctx.attr.use_file_as_top_level_scope:
+        iargs.append("-use_file_as_top_level_scope")
+
     if ctx.attr.override_stdlib_corpus:
         iargs.append("-override_stdlib_corpus=%s" % ctx.attr.override_stdlib_corpus)
 
@@ -210,6 +213,7 @@ go_entries = rule(
         # The suffix used to recognize linkage metadata files, if non-empty.
         "metadata_suffix": attr.string(default = ""),
         "use_compilation_corpus_for_all": attr.bool(default = False),
+        "use_file_as_top_level_scope": attr.bool(default = False),
         "override_stdlib_corpus": attr.string(default = ""),
 
         # The location of the Go indexer binary.
@@ -259,6 +263,7 @@ def _go_indexer(
         emit_anchor_scopes = False,
         allow_duplicates = False,
         use_compilation_corpus_for_all = False,
+        use_file_as_top_level_scope = False,
         override_stdlib_corpus = "",
         metadata_suffix = "",
         extra_extractor_args = []):
@@ -284,6 +289,7 @@ def _go_indexer(
         has_marked_source = has_marked_source,
         emit_anchor_scopes = emit_anchor_scopes,
         use_compilation_corpus_for_all = use_compilation_corpus_for_all,
+        use_file_as_top_level_scope = use_file_as_top_level_scope,
         override_stdlib_corpus = override_stdlib_corpus,
         kzip = ":" + kzip,
         metadata_suffix = metadata_suffix,
@@ -305,6 +311,7 @@ def go_indexer_test(
         emit_anchor_scopes = False,
         allow_duplicates = False,
         use_compilation_corpus_for_all = False,
+        use_file_as_top_level_scope = False,
         override_stdlib_corpus = "",
         metadata_suffix = "",
         extra_extractor_args = []):
@@ -315,6 +322,7 @@ def go_indexer_test(
         has_marked_source = has_marked_source,
         emit_anchor_scopes = emit_anchor_scopes,
         use_compilation_corpus_for_all = use_compilation_corpus_for_all,
+        use_file_as_top_level_scope = use_file_as_top_level_scope,
         override_stdlib_corpus = override_stdlib_corpus,
         importpath = import_path,
         metadata_suffix = metadata_suffix,