Add type registry dumper (#549)

.run/spice.run.xml

@@ -1,5 +1,5 @@
 <component name="ProjectRunConfigurationManager">
-  <configuration default="false" name="spice" type="CMakeRunConfiguration" factoryName="Application" PROGRAM_PARAMS="build -O0 -d -g ../../media/test-project/test.spice" REDIRECT_INPUT="false" ELEVATE="false" USE_EXTERNAL_CONSOLE="false" EMULATE_TERMINAL="false" PASS_PARENT_ENVS_2="true" PROJECT_NAME="Spice" TARGET_NAME="spice" CONFIG_NAME="Debug" RUN_TARGET_PROJECT_NAME="Spice" RUN_TARGET_NAME="spice">
+  <configuration default="false" name="spice" type="CMakeRunConfiguration" factoryName="Application" PROGRAM_PARAMS="build -O2 -d ../../media/test-project/test.spice" REDIRECT_INPUT="false" ELEVATE="false" USE_EXTERNAL_CONSOLE="false" EMULATE_TERMINAL="false" PASS_PARENT_ENVS_2="true" PROJECT_NAME="Spice" TARGET_NAME="spice" CONFIG_NAME="Debug" RUN_TARGET_PROJECT_NAME="Spice" RUN_TARGET_NAME="spice">
     <envs>
       <env name="LLVM_ADDITIONAL_FLAGS" value="-lole32 -lws2_32" />
       <env name="LLVM_BUILD_INCLUDE_DIR" value="$PROJECT_DIR$/../llvm-project-latest/build/include" />

docs/docs/cli/build.md

@@ -25,6 +25,7 @@ You can apply following options to the `build` subcommand:
 | `-cst`       | `--dump-cst`         | Dump CST as serialized string and SVG image                                                                     |
 | `-ast`       | `--dump-ast`         | Dump AST as serialized string and SVG image                                                                     |
 | `-symtab`    | `--dump-symtab`      | Dump serialized symbol tables                                                                                   |
+| `-types`     | `--dump-types`       | Dump all used types                                                                                             |
 | `-ir`        | `--dump-ir`          | Dump LLVM-IR                                                                                                    |
 | `-s`, `-asm` | `--dump-assembly`    | Dump Assembly code                                                                                              |
 | `-b`, `-obj` | `--dump-object-file` | Dump object files                                                                                               |

docs/docs/cli/install.md

@@ -25,6 +25,7 @@ You can apply following options to the `install` subcommand:
 | `-cst`       | `--dump-cst`         | Dump CST as serialized string and SVG image                                          |
 | `-ast`       | `--dump-ast`         | Dump AST as serialized string and SVG image                                          |
 | `-symtab`    | `--dump-symtab`      | Dump serialized symbol tables                                                        |
+| `-types`     | `--dump-types`       | Dump all used types                                                                  |
 | `-ir`        | `--dump-ir`          | Dump LLVM-IR                                                                         |
 | `-s`, `-asm` | `--dump-assembly`    | Dump Assembly code                                                                   |
 | `-b`, `-obj` | `--dump-object-file` | Dump object files                                                                    |

docs/docs/cli/run.md

@@ -25,6 +25,7 @@ You can apply following options to the `run` subcommand:
 | `-cst`       | `--dump-cst`         | Dump CST as serialized string and SVG image                                                 |
 | `-ast`       | `--dump-ast`         | Dump AST as serialized string and SVG image                                                 |
 | `-symtab`    | `--dump-symtab`      | Dump serialized symbol tables                                                               |
+| `-types`     | `--dump-types`       | Dump all used types                                                                         |
 | `-ir`        | `--dump-ir`          | Dump LLVM-IR                                                                                |
 | `-s`, `-asm` | `--dump-assembly`    | Dump Assembly code                                                                          |
 | `-b`, `-obj` | `--dump-object-file` | Dump object files                                                                           |

docs/docs/cli/test.md

@@ -27,6 +27,7 @@ You can apply following options to the `test` subcommand:
 | `-cst`       | `--dump-cst`         | Dump CST as serialized string and SVG image                                                 |
 | `-ast`       | `--dump-ast`         | Dump AST as serialized string and SVG image                                                 |
 | `-symtab`    | `--dump-symtab`      | Dump serialized symbol tables                                                               |
+| `-types`     | `--dump-types`       | Dump all used types                                                                         |
 | `-ir`        | `--dump-ir`          | Dump LLVM-IR                                                                                |
 | `-s`, `-asm` | `--dump-assembly`    | Dump Assembly code                                                                          |
 | `-j <n>`     | `--jobs <n>`         | Set number of jobs to parallelize compilation (default is auto)                             |

media/test-project/test.spice

@@ -1,32 +1,28 @@
-import "std/os/env";
-import "bootstrap/lexer/lexer";
+import "std/os/thread";
 
-f<int> main() {
-    String filePath = getEnv("SPICE_STD_DIR") + "/../test/test-files/bootstrap-compiler/standalone-lexer-test/test-file.spice";
-    Lexer lexer = Lexer(filePath.getRaw());
-    unsigned long tokenCount = 0l;
-    while (!lexer.isEOF()) {
-        Token token = lexer.getToken();
-        token.print();
-        lexer.advance();
-        tokenCount++;
-    }
-    printf("\nLexed tokens: %d\n", tokenCount);
+f<int> fib(int n) {
+    if n <= 2 { return 1; }
+    return fib(n - 1) + fib(n - 2);
 }
 
-/*import "std/iterator/number-iterator";
-
 f<int> main() {
-    NumberIterator<int> itInt = range(1, 10);
-    dyn idxAndValueInt = itInt.getIdx();
-    assert idxAndValueInt.getSecond() == 4;
-}*/
-
-/*import "std/data/hash-table";
-
-f<int> main() {
-    HashTable<int, string> map = HashTable<int, string>(3l);
-}*/
+    int threadCount = 8;
+    Thread[8] threads = [];
+    for unsigned int i = 0; i < threadCount; i++ {
+        threads[i] = Thread(p() {
+            int res = fib(30);
+            printf("Thread returned with result: %d\n", res);
+        });
+        Thread& thread = threads[i];
+        thread.run();
+    }
+    printf("Started all threads. Waiting for results ...\n");
+    for unsigned int i = 0; i < threadCount; i++ {
+        Thread& thread = threads[i];
+        thread.join();
+    }
+    printf("Program finished");
+}
 
 /*import "bootstrap/util/block-allocator";
 import "bootstrap/util/memory";

src/SourceFile.cpp

@@ -483,6 +483,14 @@ void SourceFile::concludeCompilation() {
   if (!cliOptions.ignoreCache)
     resourceManager.cacheManager.cacheSourceFile(this);
 
+  // Save type registry as string in the compiler output
+  if (mainFile && (cliOptions.dumpSettings.dumpTypes || cliOptions.testMode))
+    compilerOutput.typesString = TypeRegistry::dump();
+
+  // Dump type registry
+  if (mainFile && cliOptions.dumpSettings.dumpTypes)
+    dumpOutput(compilerOutput.typesString, "Type Registry", "type-registry.out");
+
   // Print warning if verifier is disabled
   if (parent == nullptr && cliOptions.disableVerifier) {
     const std::string warningMessage =

src/SourceFile.h

@@ -92,6 +92,7 @@ struct CompilerOutput {
   std::string irString;
   std::string irOptString;
   std::string asmString;
+  std::string typesString;
   std::vector<CompilerWarning> warnings;
   TimerOutput times;
 };

src/driver/Driver.cpp

@@ -337,6 +337,8 @@ void Driver::addCompileSubcommandOptions(CLI::App *subCmd) {
   subCmd->add_flag<bool>("--dump-ast,-ast", cliOptions.dumpSettings.dumpAST, "Dump AST as serialized string and SVG image");
   // --dump-symtab
   subCmd->add_flag<bool>("--dump-symtab,-symtab", cliOptions.dumpSettings.dumpSymbolTable, "Dump serialized symbol tables");
+  // --dump-types
+  subCmd->add_flag<bool>("--dump-types,-types", cliOptions.dumpSettings.dumpTypes, "Dump all used types");
   // --dump-ir
   subCmd->add_flag<bool>("--dump-ir,-ir", cliOptions.dumpSettings.dumpIR, "Dump LLVM-IR");
   // --dump-assembly

src/driver/Driver.h

@@ -55,6 +55,7 @@ struct CliOptions {
     bool dumpCST = false;
     bool dumpAST = false;
     bool dumpSymbolTable = false;
+    bool dumpTypes = false;
     bool dumpIR = false;
     bool dumpAssembly = false;
     bool dumpObjectFile = false;

src/global/TypeRegistry.cpp

@@ -2,6 +2,8 @@
 
 #include "TypeRegistry.h"
 
+#include <sstream>
+
 #include <util/CustomHashFunctions.h>
 
 namespace spice::compiler {
@@ -77,6 +79,26 @@ const Type *TypeRegistry::getOrInsert(const TypeChain &typeChain) { return getOr
  */
 size_t TypeRegistry::getTypeCount() { return types.size(); }
 
-void TypeRegistry::clear() { types.clear();}
+/**
+ * Dump all types in the type registry
+ */
+std::string TypeRegistry::dump() {
+  std::vector<std::string> typeStrings;
+  typeStrings.reserve(types.size());
+  for (const auto &type : types)
+    typeStrings.push_back(type.second->getName());
+  // Sort to ensure deterministic output
+  std::sort(typeStrings.begin(), typeStrings.end());
+  // Serialize type registry
+  std::stringstream typeRegistryString;
+  for (const std::string &typeString : typeStrings)
+    typeRegistryString << typeString << "\n";
+  return typeRegistryString.str();
+}
+
+/**
+ * Clear the type registry
+ */
+void TypeRegistry::clear() { types.clear(); }
 
 } // namespace spice::compiler

src/global/TypeRegistry.h

@@ -24,6 +24,7 @@ class TypeRegistry {
                                  const TypeChainElementData &data, const QualTypeList &templateTypes);
   static const Type *getOrInsert(const TypeChain& typeChain);
   static size_t getTypeCount();
+  static std::string dump();
   static void clear();
 
 private:

src/symboltablebuilder/TypeChain.cpp

@@ -102,7 +102,9 @@ void TypeChainElement::getName(std::stringstream &name, bool withSize) const {
     name << "f";
     if (data.hasCaptures)
       name << "[]";
-    name << "<" << paramTypes.front().getName(true) << ">(";
+    if (!paramTypes.empty())
+      name << "<" << paramTypes.front().getName(true) << ">";
+    name << "(";
     for (size_t i = 1; i < paramTypes.size(); i++) {
       if (i > 1)
         name << ",";

src/util/CustomHashFunctions.cpp

@@ -6,7 +6,7 @@ namespace std {
 
 size_t hash<spice::compiler::TypeChainElement>::operator()(const spice::compiler::TypeChainElement &tce) const {
   // Hasher for QualTypeList
-  const auto pred = [](size_t acc, const spice::compiler::QualType &val) {
+  constexpr auto pred = [](size_t acc, const spice::compiler::QualType &val) {
     // Combine the previous hash value with the current element's hash, adjusted by a prime number to reduce collisions
     return acc * 31 + std::hash<spice::compiler::QualType>{}(val);
   };

test/TestRunner.cpp

@@ -17,6 +17,7 @@
 #include <exception/ParserError.h>
 #include <exception/SemanticError.h>
 #include <global/GlobalResourceManager.h>
+#include <global/TypeRegistry.h>
 #include <symboltablebuilder/SymbolTable.h>
 #include <util/FileUtil.h>
 
@@ -55,6 +56,7 @@ void execTestCase(const TestCase &testCase) {
           /* dumpCST= */ false,
           /* dumpAST= */ false,
           /* dumpSymbolTables= */ false,
+          /* dumpTypes= */ false,
           /* dumpIR= */ false,
           /* dumpAssembly= */ false,
           /* dumpObjectFile= */ false,
@@ -71,6 +73,7 @@ void execTestCase(const TestCase &testCase) {
       /* disableVerifier= */ false,
       /* testMode= */ true,
   };
+  static_assert(sizeof(CliOptions::DumpSettings) == 9, "CliOptions::DumpSettings struct size changed");
   static_assert(sizeof(CliOptions) == 360, "CliOptions struct size changed");
 
   // Instantiate GlobalResourceManager
@@ -196,6 +199,9 @@ void execTestCase(const TestCase &testCase) {
       resourceManager.linker.link();
     }
 
+    // Check type registry output
+    TestUtil::checkRefMatch(testCase.testPath / REF_NAME_TYPE_REGISTRY, [&]() { return TypeRegistry::dump(); });
+
     // Check if the execution output matches the expected output
     TestUtil::checkRefMatch(testCase.testPath / REF_NAME_EXECUTION_OUTPUT, [&]() {
       const std::filesystem::path cliFlagsFile = testCase.testPath / INPUT_NAME_CLI_FLAGS;

test/test-files/benchmark/success-ackermann/type-registry.out

@@ -0,0 +1,7 @@
+bool
+dyn
+f()
+f<int>()
+f<int>(int,int)
+int
+invalid

test/test-files/benchmark/success-faculty/type-registry.out

@@ -0,0 +1,7 @@
+bool
+dyn
+f()
+f<int>()
+f<int>(int)
+int
+invalid

test/test-files/benchmark/success-fibonacci-threaded/type-registry.out

@@ -0,0 +1,37 @@
+*
+Error
+Error&
+Error*
+Pthread_attr_t
+Pthread_attr_t&
+Pthread_attr_t*
+Thread
+Thread
+Thread&
+Thread*
+bool
+byte
+byte*
+byte**
+char
+char*
+dyn
+dyn
+f()
+f<int>()
+f<int>(int)
+f<int>(long*,Pthread_attr_t*,p())
+f<int>(long,byte**)
+f<long>()
+import
+int
+invalid
+long
+long
+long*
+p()
+p()
+p(int,string)
+p(p())
+p(string)
+string

test/test-files/benchmark/success-fibonacci/type-registry.out

@@ -0,0 +1,7 @@
+bool
+dyn
+f()
+f<int>()
+f<int>(int)
+int
+invalid

test/test-files/benchmark/success-hello-world/type-registry.out

@@ -0,0 +1,5 @@
+bool
+f()
+f<int>()
+int
+invalid

test/test-files/benchmark/success-hello-world2/type-registry.out

@@ -0,0 +1,7 @@
+bool
+char
+f()
+f<int>()
+int
+invalid
+long

test/test-files/benchmark/success-pidigits/type-registry.out

@@ -0,0 +1,6 @@
+bool
+f()
+f<int>()
+int
+invalid
+long

test/util/TestUtil.h

@@ -30,6 +30,7 @@ const char *const REF_NAME_SOURCE = "source.spice";
 const char *const REF_NAME_PARSE_TREE = "parse-tree.dot";
 const char *const REF_NAME_SYNTAX_TREE = "syntax-tree.dot";
 const char *const REF_NAME_SYMBOL_TABLE = "symbol-table.json";
+const char *const REF_NAME_TYPE_REGISTRY = "type-registry.out";
 const char *const REF_NAME_IR = "ir-code.ll";
 const char *const REF_NAME_ASM = "assembly.asm";
 const char *const REF_NAME_OPT_IR[5] = {"ir-code-O1.ll", "ir-code-O2.ll", "ir-code-O3.ll", "ir-code-Os.ll", "ir-code-Oz.ll"};