handle clang module export statements in SymbolGraphGen

Author: QuietMisdreavus

lib/SymbolGraphGen/SymbolGraphGen.cpp

@@ -63,23 +63,43 @@ int symbolgraphgen::emitSymbolGraphForModule(
     ModuleDecl *M, const SymbolGraphOptions &Options) {
   ModuleDecl::ImportCollector importCollector(Options.MinimumAccessLevel);
 
-  bool IncludeClangSubmodules = false;
-  if (auto *ClangModule = M->findUnderlyingClangModule()) {
-    // If a Clang module has a definition that includes `export *; module * {
-    // export *; }` then we want to treat those inferred submodules as part of
-    // the parent module, to preserve historical behavior.
-    if (ClangModule->InferSubmodules && ClangModule->InferExportWildcard) {
-      IncludeClangSubmodules = true;
+  SmallPtrSet<const clang::Module *, 2> ExportedClangModules = {};
+  SmallPtrSet<const clang::Module *, 2> WildcardExportClangModules = {};
+  if (const auto *ClangModule = M->findUnderlyingClangModule()) {
+    // Scan through the Clang module's exports and collect them for later
+    // handling
+    for (auto ClangExport : ClangModule->Exports) {
+      if (ClangExport.getInt()) {
+        // Blanket exports are represented as a true boolean tag
+        if (const auto *ExportParent = ClangExport.getPointer()) {
+          // If a pointer is present, this is a scoped blanket export, like
+          // `export Submodule.*`
+          WildcardExportClangModules.insert(ExportParent);
+        } else {
+          // Otherwise it represents a full blanket `export *`
+          WildcardExportClangModules.insert(ClangModule);
+        }
+      } else if (!ClangExport.getInt() && ClangExport.getPointer()) {
+        // This is an explicit `export Submodule`
+        ExportedClangModules.insert(ClangExport.getPointer());
+      }
     }
   }
 
-  auto importFilter = [&Options, &IncludeClangSubmodules,
-                       &M](const ModuleDecl *module) {
+  auto importFilter = [&Options, &WildcardExportClangModules,
+                       &ExportedClangModules](const ModuleDecl *module) {
     if (!module)
       return false;
 
-    if (IncludeClangSubmodules && module->isSubmoduleOf(M)) {
-      return true;
+    if (const auto *ClangModule = module->findUnderlyingClangModule()) {
+      if (ExportedClangModules.contains(ClangModule)) {
+        return true;
+      }
+
+      for (const auto *ClangParent : WildcardExportClangModules) {
+        if (ClangModule->isSubModuleOf(ClangParent))
+          return true;
+      }
     }
 
     if (Options.AllowedReexportedModules.has_value())
@@ -90,7 +110,8 @@ int symbolgraphgen::emitSymbolGraphForModule(
     return false;
   };
 
-  if (Options.AllowedReexportedModules.has_value() || IncludeClangSubmodules)
+  if (Options.AllowedReexportedModules.has_value() ||
+      !WildcardExportClangModules.empty() || !ExportedClangModules.empty())
     importCollector.importFilter = std::move(importFilter);
 
   SmallVector<Decl *, 64> ModuleDecls;

test/SymbolGraph/ClangImporter/PartialSubmoduleExport.swift

@@ -0,0 +1,80 @@
+// RUN: %empty-directory(%t)
+
+// RUN: split-file %s %t
+
+// RUN: %target-swift-symbolgraph-extract -sdk %clang-importer-sdk -module-name PartialSubmoduleExport -I %t/PartialSubmoduleExport -output-dir %t -pretty-print -v
+// RUN: %FileCheck %s --input-file %t/PartialSubmoduleExport.symbols.json
+
+// REQUIRES: objc_interop
+
+// The PartialSubmoduleExport module below is structured like this:
+
+// PartialSubmoduleExport
+// - GroupA
+//   - GroupAOne
+//   - GroupATwo
+// - GroupB
+//   - GroupBOne
+//   - GroupBTwo
+
+// The module map then exports `GroupA.*` and `GroupB.GroupBOne` explicitly.
+// This test ensures that the expected symbols are in the module map,
+// and that the deliberately excluded `groupBTwo` symbol is left out.
+
+//--- PartialSubmoduleExport/module.modulemap
+module PartialSubmoduleExport {
+    header "PartialSubmoduleExport.h"
+
+    explicit module GroupA {
+        umbrella header "GroupA/GroupA.h"
+
+        module * { export * }
+    }
+
+    explicit module GroupB {
+        umbrella header "GroupB/GroupB.h"
+
+        module * { export * }
+    }
+
+    export GroupA.*
+    export GroupB.GroupBOne
+}
+
+//--- PartialSubmoduleExport/PartialSubmoduleExport.h
+#include "GroupA/GroupA.h"
+#include "GroupB/GroupB.h"
+
+// CHECK-DAG: "precise": "c:PartialSubmoduleExport.h@umbrellaVar"
+static int umbrellaVar = 0;
+
+//--- PartialSubmoduleExport/GroupA/GroupA.h
+#include "GroupAOne.h"
+#include "GroupATwo.h"
+
+// CHECK-DAG: "precise": "c:GroupA.h@groupAVar"
+static int groupAVar = 0;
+
+//--- PartialSubmoduleExport/GroupA/GroupAOne.h
+// CHECK-DAG: "precise": "c:GroupAOne.h@groupAOne"
+static int groupAOne = 1;
+
+//--- PartialSubmoduleExport/GroupA/GroupATwo.h
+// CHECK-DAG: "precise": "c:GroupATwo.h@groupATwo"
+static int groupATwo = 2;
+
+//--- PartialSubmoduleExport/GroupB/GroupB.h
+#include "GroupBOne.h"
+#include "GroupBTwo.h"
+
+static int groupBVar = 0;
+
+//--- PartialSubmoduleExport/GroupB/GroupBOne.h
+// CHECK-DAG: "precise": "c:GroupBOne.h@groupBOne"
+static int groupBOne = 1;
+
+//--- PartialSubmoduleExport/GroupB/GroupBTwo.h
+// Because GroupBTwo is not exported in the top-level module map,
+// this shouldn't be in the symbol graph
+// CHECK-NOT: "precise": "c:GroupBTwo.h@groupBTwo"
+static int groupBTwo = 2;

test/SymbolGraph/ClangImporter/UmbrellaSubmodules.swift

@@ -1,6 +1,5 @@
 // RUN: %empty-directory(%t)
 
-// RUN: %empty-directory(%t/SwiftAttr)
 // RUN: split-file %s %t
 
 // RUN: %target-swift-symbolgraph-extract -sdk %clang-importer-sdk -module-name UmbrellaSubmodules -I %t/UmbrellaSubmodules -output-dir %t -pretty-print -v