[Dependency Scanning] Refactor the scanner to resolve unqualified module imports

This changes the scanner's behavior to "resolve" a discovered module's dependencies to a set of Module IDs: module name + module kind (swift textual, swift binary, clang, etc.).

The 'ModuleDependencyInfo' objects that are stored in the dependency scanner's cache now carry a set of kind-qualified ModuleIDs for their dependencies, in addition to unqualified imported module names of their dependencies.

Previously, the scanner's internal state would cache a module dependnecy as having its own set of dependencies which were stored as names of imported modules. This led to a design where any time we needed to process the dependency downstream from its discovery (e.g. cycle detection, graph construction), we had to query the ASTContext to resolve this dependency's imports, which shouldn't be necessary. Now, upon discovery, we "resolve" a discovered dependency by executing a lookup for each of its imported module names (this operation happens regardless of this patch) and store a fully-resolved set of dependencies in the dependency module info.

Moreover, looking up a given module dependency by name (via `ASTContext`'s `getModuleDependencies`) would result in iterating over the scanner's module "loaders" and querying each for the module name. The corresponding modules would then check the scanner's cache for a respective discovered module, and if no such module is found the "loader" would search the filesystem.

This meant that in practice, we searched the filesystem on many occasions where we actually had cached the required dependency, as follows:
Suppose we had previously discovered a Clang module "foo" and cached its dependency info.
-> ASTContext.getModuleDependencies("foo")
--> (1) Swift Module "Loader" checks caches for a Swift module "foo" and doesn't find one, so it searches the filesystem for "foo" and fails to find one.
--> (2) Clang Module "Loader" checks caches for a Clang module "foo", finds one and returns it to the client.

This means that we were always searching the filesystem in (1) even if we knew that to be futile.
With this change, queries to `ASTContext`'s `getModuleDependencies` will always check all the caches first, and only delegate to the scanner "loaders" if no cached dependency is found. The loaders are then no longer in the business of checking the cached contents.

To handle cases in the scanner where we must only lookup either a Swift-only module or a Clang-only module, this patch splits 'getModuleDependencies' into an alrady-existing 'getSwiftModuleDependencies' and a newly-added 'getClangModuleDependencies'.

Author: artemcm

include/swift/AST/ASTContext.h

@@ -991,16 +991,18 @@ class ASTContext final {
 
   /// Retrieve the module dependencies for the module with the given name.
   ///
-  /// \param isUnderlyingClangModule When true, only look for a Clang module
-  /// with the given name, ignoring any Swift modules.
   Optional<ModuleDependencyInfo> getModuleDependencies(
       StringRef moduleName,
-      bool isUnderlyingClangModule,
       ModuleDependenciesCache &cache,
       InterfaceSubContextDelegate &delegate,
-      bool cacheOnly = false,
       llvm::Optional<std::pair<std::string, swift::ModuleDependencyKind>> dependencyOf = None);
 
+  /// Retrieve the module dependencies for the Clang module with the given name.
+  Optional<ModuleDependencyInfo> getClangModuleDependencies(
+      StringRef moduleName,
+      ModuleDependenciesCache &cache,
+      InterfaceSubContextDelegate &delegate);
+
   /// Retrieve the module dependencies for the Swift module with the given name.
   Optional<ModuleDependencyInfo> getSwiftModuleDependencies(
       StringRef moduleName,

include/swift/AST/ModuleDependencies.h

@@ -74,29 +74,44 @@ enum class ModuleDependencyKind : int8_t {
   LastKind = SwiftPlaceholder + 1
 };
 
+using ModuleDependencyID = std::pair<std::string, ModuleDependencyKind>;
+using ModuleDependencyIDSetVector =
+    llvm::SetVector<ModuleDependencyID, std::vector<ModuleDependencyID>,
+                    std::set<ModuleDependencyID>>;
+
 struct ModuleDependencyKindHash {
   std::size_t operator()(ModuleDependencyKind k) const {
     using UnderlyingType = std::underlying_type<ModuleDependencyKind>::type;
     return std::hash<UnderlyingType>{}(static_cast<UnderlyingType>(k));
   }
 };
 
+namespace dependencies {
+  std::string createEncodedModuleKindAndName(ModuleDependencyID id);
+}
+
 /// Base class for the variant storage of ModuleDependencyInfo.
 ///
 /// This class is mostly an implementation detail for \c ModuleDependencyInfo.
 class ModuleDependencyInfoStorageBase {
 public:
   const ModuleDependencyKind dependencyKind;
 
-  ModuleDependencyInfoStorageBase(ModuleDependencyKind dependencyKind)
-      : dependencyKind(dependencyKind) { }
+  ModuleDependencyInfoStorageBase(ModuleDependencyKind dependencyKind,
+                                  bool resolved = false)
+      : dependencyKind(dependencyKind), resolved(resolved) { }
 
   virtual ModuleDependencyInfoStorageBase *clone() const = 0;
 
   virtual ~ModuleDependencyInfoStorageBase();
 
   /// The set of modules on which this module depends.
-  std::vector<std::string> moduleDependencies;
+  std::vector<std::string> moduleImports;
+
+  /// The set of modules on which this module depends, resolved
+  /// to Module IDs, qualified by module kind: Swift, Clang, etc.
+  ModuleDependencyIDSetVector resolvedModuleDependencies;
+  bool resolved;
 };
 
 struct CommonSwiftTextualModuleDependencyDetails {
@@ -400,9 +415,41 @@ class ModuleDependencyInfo {
           compiledModulePath, moduleDocPath, sourceInfoPath));
   }
 
-  /// Retrieve the module-level dependencies.
-  ArrayRef<std::string> getModuleDependencies() const {
-    return storage->moduleDependencies;
+  /// Retrieve the module-level imports.
+  ArrayRef<std::string> getModuleImports() const {
+    return storage->moduleImports;
+  }
+
+  /// Retreive the module-level dependencies.
+  const ModuleDependencyIDSetVector &getModuleDependencies() const {
+    assert(storage->resolved);
+    return storage->resolvedModuleDependencies;
+  }
+
+  /// Resolve a dependency's set of `imports` with qualified Module IDs
+  void resolveDependencies(ModuleDependencyIDSetVector dependencyIDs) {
+#ifndef NDEBUG
+    // Sanity check against the identified imports for this module
+    for (auto importedModuleName : storage->moduleImports) {
+      auto matchingModuleIDIt =
+          std::find_if(dependencyIDs.begin(), dependencyIDs.end(),
+                       [&](ModuleDependencyID dependencyID) {
+                         return dependencyID.first == importedModuleName;
+                       });
+      assert(matchingModuleIDIt != dependencyIDs.end() &&
+             "Imported module not found in resolved module dependency set");
+    }
+#endif
+    assert(!storage->resolved && "Resolving an already-resolved dependency");
+    storage->resolved = true;
+    storage->resolvedModuleDependencies = dependencyIDs;
+  }
+
+  bool isResolved() const {
+    return storage->resolved;
+  }
+  void setIsResolved(bool isResolved) {
+    storage->resolved = isResolved;
   }
 
   /// Whether the dependencies are for a Swift module: either Textual, Source, Binary, or Placeholder.
@@ -444,19 +491,22 @@ class ModuleDependencyInfo {
     getAsPlaceholderDependencyModule() const;
 
   /// Add a dependency on the given module, if it was not already in the set.
-  void addModuleDependency(StringRef module,
-                           llvm::StringSet<> *alreadyAddedModules = nullptr);
+  void addModuleImport(StringRef module,
+                       llvm::StringSet<> *alreadyAddedModules = nullptr);
 
   /// Add a dependency on the given module, if it was not already in the set.
-  void addModuleDependency(ImportPath::Module module,
-                           llvm::StringSet<> *alreadyAddedModules = nullptr) {
-    addModuleDependency(module.front().Item.str(), alreadyAddedModules);
+  void addModuleImport(ImportPath::Module module,
+                       llvm::StringSet<> *alreadyAddedModules = nullptr) {
+    addModuleImport(module.front().Item.str(), alreadyAddedModules);
   }
 
-  /// Add all of the module dependencies for the imports in the given source
-  /// file to the set of module dependencies.
-  void addModuleDependencies(const SourceFile &sf,
-                             llvm::StringSet<> &alreadyAddedModules);
+  /// Add all of the module imports in the given source
+  /// file to the set of module imports.
+  void addModuleImport(const SourceFile &sf,
+                       llvm::StringSet<> &alreadyAddedModules);
+  /// Add a kind-qualified module dependency ID to the set of
+  /// module dependencies.
+  void addModuleDependency(ModuleDependencyID dependencyID);
 
   /// Get the bridging header.
   Optional<std::string> getBridgingHeader() const;
@@ -480,8 +530,6 @@ class ModuleDependencyInfo {
   collectCrossImportOverlayNames(ASTContext &ctx, StringRef moduleName);
 };
 
-using ModuleDependencyID = std::pair<std::string, ModuleDependencyKind>;
-using ModuleDependenciesVector = llvm::SmallVector<ModuleDependencyInfo, 1>;
 using ModuleNameToDependencyMap = llvm::StringMap<ModuleDependencyInfo>;
 using ModuleDependenciesKindMap =
     std::unordered_map<ModuleDependencyKind,
@@ -587,8 +635,8 @@ class SwiftDependencyScanningService {
   }
 
   /// Whether we have cached dependency information for the given module.
-  bool hasDependencies(StringRef moduleName,
-                       Optional<ModuleDependencyKind> kind) const;
+  bool hasDependency(StringRef moduleName,
+                     Optional<ModuleDependencyKind> kind) const;
 
   /// Return a pointer to the context-specific cache state of the current
   /// scanning action.
@@ -606,15 +654,15 @@ class SwiftDependencyScanningService {
   ///
   /// \returns the cached result, or \c None if there is no cached entry.
   Optional<ModuleDependencyInfo>
-  findDependencies(StringRef moduleName,
-                   Optional<ModuleDependencyKind> kind) const;
+  findDependency(StringRef moduleName,
+                 Optional<ModuleDependencyKind> kind) const;
 
   /// Record dependencies for the given module.
-  const ModuleDependencyInfo *recordDependencies(StringRef moduleName,
+  const ModuleDependencyInfo *recordDependency(StringRef moduleName,
                                                ModuleDependencyInfo dependencies);
 
   /// Update stored dependencies for the given module.
-  const ModuleDependencyInfo *updateDependencies(ModuleDependencyID moduleID,
+  const ModuleDependencyInfo *updateDependency(ModuleDependencyID moduleID,
                                                ModuleDependencyInfo dependencies);
 
   /// Reference the list of all module dependencies that are not source-based
@@ -658,7 +706,7 @@ class ModuleDependenciesCache {
   /// The Clang dependency scanner tool
   clang::tooling::dependencies::DependencyScanningTool clangScanningTool;
   /// Discovered Clang modules are only cached locally.
-  llvm::StringMap<ModuleDependenciesVector> clangModuleDependencies;
+  llvm::StringMap<ModuleDependencyInfo> clangModuleDependencies;
 
   /// Retrieve the dependencies map that corresponds to the given dependency
   /// kind.
@@ -669,8 +717,8 @@ class ModuleDependenciesCache {
 
   /// Local cache results lookup, only for modules which were discovered during
   /// the current scanner invocation.
-  bool hasDependenciesLocalOnly(StringRef moduleName,
-                                Optional<ModuleDependencyKind> kind) const;
+  bool hasDependencyLocalOnly(StringRef moduleName,
+                              Optional<ModuleDependencyKind> kind) const;
 
   /// Local cache results lookup, only for modules which were discovered during
   /// the current scanner invocation.
@@ -687,8 +735,8 @@ class ModuleDependenciesCache {
 
 public:
   /// Whether we have cached dependency information for the given module.
-  bool hasDependencies(StringRef moduleName,
-                       Optional<ModuleDependencyKind> kind) const;
+  bool hasDependency(StringRef moduleName,
+                     Optional<ModuleDependencyKind> kind) const;
 
   /// Produce a reference to the Clang scanner tool associated with this cache
   clang::tooling::dependencies::DependencyScanningTool& getClangScannerTool() {
@@ -702,16 +750,21 @@ class ModuleDependenciesCache {
   ///
   /// \returns the cached result, or \c None if there is no cached entry.
   Optional<ModuleDependencyInfo>
-  findDependencies(StringRef moduleName,
-                   Optional<ModuleDependencyKind> kind) const;
+  findDependency(StringRef moduleName,
+                 Optional<ModuleDependencyKind> kind) const;
 
   /// Record dependencies for the given module.
-  void recordDependencies(StringRef moduleName,
-                          ModuleDependencyInfo dependencies);
+  void recordDependency(StringRef moduleName,
+                        ModuleDependencyInfo dependencies);
 
   /// Update stored dependencies for the given module.
-  void updateDependencies(ModuleDependencyID moduleID,
-                          ModuleDependencyInfo dependencies);
+  void updateDependency(ModuleDependencyID moduleID,
+                        ModuleDependencyInfo dependencies);
+
+  /// Resolve a dependency module's set of imports
+  /// to a kind-qualified set of module IDs.
+  void resolveDependencyImports(ModuleDependencyID moduleID,
+                                ModuleDependencyIDSetVector dependencyIDs);
 
   const std::vector<ModuleDependencyID> &getAllSourceModules() const {
     return globalScanningService.getAllSourceModules();

include/swift/DependencyScan/SerializedModuleDependencyCacheFormat.h

@@ -36,7 +36,7 @@ using llvm::BCVBR;
 
 /// Every .moddepcache file begins with these 4 bytes, for easy identification.
 const unsigned char MODULE_DEPENDENCY_CACHE_FORMAT_SIGNATURE[] = {'I', 'M', 'D','C'};
-const unsigned MODULE_DEPENDENCY_CACHE_FORMAT_VERSION_MAJOR = 3;
+const unsigned MODULE_DEPENDENCY_CACHE_FORMAT_VERSION_MAJOR = 4;
 /// Increment this on every change.
 const unsigned MODULE_DEPENDENCY_CACHE_FORMAT_VERSION_MINOR = 0;
 
@@ -45,19 +45,21 @@ const unsigned MODULE_DEPENDENCY_CACHE_FORMAT_VERSION_MINOR = 0;
 using IdentifierIDField = BCVBR<13>;
 using FileIDField = IdentifierIDField;
 using ModuleIDField = IdentifierIDField;
-using ContextHashField = IdentifierIDField;
+using ContextHashIDField = IdentifierIDField;
 
 /// A bit that indicates whether or not a module is a framework
 using IsFrameworkField = BCFixed<1>;
 
 /// Arrays of various identifiers, distinguished for readability
 using IdentifierIDArryField = llvm::BCArray<IdentifierIDField>;
+using ModuleIDArryField = llvm::BCArray<IdentifierIDField>;
 
 /// Identifiers used to refer to the above arrays
 using FileIDArrayIDField = IdentifierIDField;
 using ContextHashIDField = IdentifierIDField;
-using DependencyIDArrayIDField = IdentifierIDField;
+using ImportArrayIDField = IdentifierIDField;
 using FlagIDArrayIDField = IdentifierIDField;
+using DependencyIDArrayIDField = IdentifierIDField;
 
 /// The ID of the top-level block containing the dependency graph
 const unsigned GRAPH_BLOCK_ID = llvm::bitc::FIRST_APPLICATION_BLOCKID;
@@ -115,34 +117,35 @@ using IdentifierArrayLayout =
 // - SwiftPlaceholderModuleDetails
 // - ClangModuleDetails
 using ModuleInfoLayout =
-    BCRecordLayout<MODULE_NODE,             // ID
-                   IdentifierIDField,       // module name
-                   ContextHashIDField,      // 
-                   DependencyIDArrayIDField // directDependencies
+    BCRecordLayout<MODULE_NODE,                  // ID
+                   IdentifierIDField,            // moduleName
+                   ContextHashIDField,           // contextHash
+                   ImportArrayIDField,           // moduleImports
+                   DependencyIDArrayIDField      // resolvedModuleDependencies
                    >;
 
 using SwiftInterfaceModuleDetailsLayout =
     BCRecordLayout<SWIFT_INTERFACE_MODULE_DETAILS_NODE, // ID
-                   FileIDField,        // outputFilePath
-                   FileIDField,        // swiftInterfaceFile
-                   FileIDArrayIDField, // compiledModuleCandidates
-                   FlagIDArrayIDField, // buildCommandLine
-                   FlagIDArrayIDField, // extraPCMArgs
-                   ContextHashField,   // contextHash
-                   IsFrameworkField,   // isFramework
-                   FileIDField,        // bridgingHeaderFile
-                   FileIDArrayIDField, // sourceFiles
-                   FileIDArrayIDField, // bridgingSourceFiles
-                   FileIDArrayIDField   // bridgingModuleDependencies
+                   FileIDField,                         // outputFilePath
+                   FileIDField,                         // swiftInterfaceFile
+                   FileIDArrayIDField,                  // compiledModuleCandidates
+                   FlagIDArrayIDField,                  // buildCommandLine
+                   FlagIDArrayIDField,                  // extraPCMArgs
+                   ContextHashIDField,                  // contextHash
+                   IsFrameworkField,                    // isFramework
+                   FileIDField,                         // bridgingHeaderFile
+                   FileIDArrayIDField,                  // sourceFiles
+                   FileIDArrayIDField,                  // bridgingSourceFiles
+                   FileIDArrayIDField                   // bridgingModuleDependencies
                    >;
 
 using SwiftSourceModuleDetailsLayout =
     BCRecordLayout<SWIFT_SOURCE_MODULE_DETAILS_NODE, // ID
-                   FlagIDArrayIDField, // extraPCMArgs
-                   FileIDField,        // bridgingHeaderFile
-                   FileIDArrayIDField, // sourceFiles
-                   FileIDArrayIDField, // bridgingSourceFiles
-                   FileIDArrayIDField  // bridgingModuleDependencies
+                   FlagIDArrayIDField,               // extraPCMArgs
+                   FileIDField,                      // bridgingHeaderFile
+                   FileIDArrayIDField,               // sourceFiles
+                   FileIDArrayIDField,               // bridgingSourceFiles
+                   FileIDArrayIDField                // bridgingModuleDependencies
                    >;
 
 using SwiftBinaryModuleDetailsLayout =
@@ -157,14 +160,14 @@ using SwiftPlaceholderModuleDetailsLayout =
     BCRecordLayout<SWIFT_PLACEHOLDER_MODULE_DETAILS_NODE, // ID
                    FileIDField,                           // compiledModulePath
                    FileIDField,                           // moduleDocPath
-                   FileIDField // moduleSourceInfoPath
+                   FileIDField                            // moduleSourceInfoPath
                    >;
 
 using ClangModuleDetailsLayout =
     BCRecordLayout<CLANG_MODULE_DETAILS_NODE, // ID
                    FileIDField,               // pcmOutputPath
                    FileIDField,               // moduleMapPath
-                   ContextHashField,          // contextHash
+                   ContextHashIDField,        // contextHash
                    FlagIDArrayIDField,        // commandLine
                    FileIDArrayIDField,        // fileDependencies
                    FlagIDArrayIDField         // capturedPCMArgs