Introduce the notion of ns and filename in the reader and location

dev.org

@@ -68,6 +68,9 @@ Then here is the list or parsers that we have considered
 - [[https://www.hboehm.info/gc/][Boehm GC]]
 ** JIT
 - [[https://asmjit.com/][Machine code generator for C++]]
+** Optimizations
+- [[https://sunfishcode.github.io/blog/2018/10/22/Canonicalization.html][Canonicalization]]
+
 ** Compiler
 - [[https://eli.thegreenplace.net/2011/09/06/stack-frame-layout-on-x86-64][Stack frame layout on x86-64]]
 *** Branch instructions

include/serene/context.h

@@ -70,8 +70,11 @@ public:
   // --------------------------------------------------------------------------
   llvm::LLVMContext llvmContext;
   mlir::MLIRContext mlirContext;
+
   mlir::PassManager pm;
 
+  mlir::DiagnosticEngine &diagEngine;
+
   /// The source manager is responsible for loading namespaces and practically
   /// managing the source code in form of memory buffers.
   SourceMgr sourceManager;
@@ -96,7 +99,8 @@ public:
   std::shared_ptr<Namespace> getNS(llvm::StringRef ns_name);
 
   SereneContext()
-      : pm(&mlirContext), targetPhase(CompilationPhase::NoOptimization) {
+      : pm(&mlirContext), diagEngine(mlirContext.getDiagEngine()),
+        targetPhase(CompilationPhase::NoOptimization) {
     mlirContext.getOrLoadDialect<serene::slir::SereneDialect>();
     mlirContext.getOrLoadDialect<mlir::StandardOpsDialect>();
     // TODO: Get the crash report path dynamically from the cli

include/serene/reader/location.h

@@ -25,24 +25,22 @@
 #ifndef LOCATION_H
 #define LOCATION_H
 
-#include "mlir/IR/Location.h"
-#include "mlir/IR/MLIRContext.h"
-#include "mlir/IR/OpDefinition.h"
+#include "mlir/IR/Diagnostics.h"
 
+#include <mlir/IR/Location.h>
 #include <string>
 
 namespace serene {
+class SereneContext;
+
 namespace reader {
 
 /// It represents a location in the input string to the parser via `line`,
 struct Location {
-
+  llvm::StringRef ns;
   /// A pointer to the character that this location is pointing to
   /// it the input buffer
-  const char *c;
-
-  /// The id of the buffer that this location belongs too.
-  unsigned bufferId;
+  const char *c = nullptr;
 
   /// At this stage we only support 65535 lines of code in each file
   unsigned short int line;
@@ -52,6 +50,8 @@ struct Location {
   ::std::string toString() const;
   Location() = default;
   Location clone();
+
+  mlir::Location toMLIRLocation(SereneContext &ctx, llvm::StringRef ns);
 };
 
 class LocationRange {

include/serene/reader/reader.h

@@ -55,6 +55,9 @@ class Reader {
 private:
   SereneContext &ctx;
 
+  llvm::StringRef ns;
+  llvm::Optional<llvm::StringRef> filename;
+
   const char *current_char = NULL;
 
   llvm::StringRef buf;
@@ -91,8 +94,10 @@ private:
   bool isEndOfBuffer(const char *);
 
 public:
-  Reader(SereneContext &ctx, llvm::StringRef buf);
-  Reader(SereneContext &ctx, llvm::MemoryBufferRef buf);
+  Reader(SereneContext &ctx, llvm::StringRef buf, llvm::StringRef ns,
+         llvm::Optional<llvm::StringRef> filename);
+  Reader(SereneContext &ctx, llvm::MemoryBufferRef buf, llvm::StringRef ns,
+         llvm::Optional<llvm::StringRef> filename);
 
   // void setInput(const llvm::StringRef string);
 
@@ -105,7 +110,11 @@ public:
 
 /// Parses the given `input` string and returns a `Result<ast>`
 /// which may contains an AST or an `llvm::Error`
-Result<exprs::Ast> read(SereneContext &ctx, const llvm::StringRef input);
-Result<exprs::Ast> read(SereneContext &ctx, const llvm::MemoryBufferRef but);
+Result<exprs::Ast> read(SereneContext &ctx, const llvm::StringRef input,
+                        llvm::StringRef ns,
+                        llvm::Optional<llvm::StringRef> filename);
+Result<exprs::Ast> read(SereneContext &ctx, const llvm::MemoryBufferRef but,
+                        llvm::StringRef ns,
+                        llvm::Optional<llvm::StringRef> filename);
 } // namespace serene::reader
 #endif

include/serene/source_mgr.h

@@ -101,6 +101,10 @@ private:
   /// This is all of the buffers that we are reading from.
   std::vector<SrcBuffer> buffers;
 
+  /// A mapping from the ns name to buffer id. The ns name is a reference to
+  /// the actual name that is stored in the Namespace instance.
+  llvm::DenseMap<llvm::StringRef, unsigned> nsToBufId;
+
   // This is the list of directories we should search for include files in.
   std::vector<std::string> loadPaths;
 

src/serene/reader/location.cpp

@@ -24,6 +24,8 @@
 
 #include "serene/reader/location.h"
 
+#include "serene/context.h"
+
 #include <llvm/Support/FormatVariadic.h>
 #include <mlir/IR/Identifier.h>
 
@@ -40,8 +42,14 @@ std::string Location::toString() const {
   return llvm::formatv("{0}:{1}", line, col);
 };
 
-Location Location::clone() { return Location{c, bufferId, line, col}; }
+Location Location::clone() { return Location{ns, c, line, col}; }
 
+mlir::Location Location::toMLIRLocation(SereneContext &ctx,
+                                        llvm::StringRef ns) {
+  // TODO: Create a new Location attribute that is namespace base
+
+  return mlir::FileLineColLoc::get(&ctx.mlirContext, ns, line, col);
+}
 /// Increase the given location by one and set the line/col value in respect to
 /// the `newline` in place.
 /// \param loc The `Location` data

src/serene/reader/reader.cpp

@@ -24,11 +24,16 @@
 
 #include "serene/reader/reader.h"
 
+#include "mlir/IR/Diagnostics.h"
+#include "mlir/IR/Location.h"
+#include "mlir/Support/LogicalResult.h"
 #include "serene/exprs/list.h"
 #include "serene/exprs/number.h"
 #include "serene/exprs/symbol.h"
 #include "serene/namespace.h"
 
+#include "clang/AST/Stmt.h"
+#include "llvm/ADT/StringRef.h"
 #include "llvm/Support/ErrorHandling.h"
 
 #include <assert.h>
@@ -46,11 +51,17 @@ namespace serene {
 
 namespace reader {
 
-Reader::Reader(SereneContext &ctx, llvm::StringRef buffer)
-    : ctx(ctx), buf(buffer){};
+Reader::Reader(SereneContext &ctx, llvm::StringRef buffer, llvm::StringRef ns,
+               llvm::Optional<llvm::StringRef> filename)
+    : ctx(ctx), ns(ns), filename(filename), buf(buffer) {
+  current_location.ns = ns;
+};
 
-Reader::Reader(SereneContext &ctx, llvm::MemoryBufferRef buffer)
-    : ctx(ctx), buf(buffer.getBuffer()){};
+Reader::Reader(SereneContext &ctx, llvm::MemoryBufferRef buffer,
+               llvm::StringRef ns, llvm::Optional<llvm::StringRef> filename)
+    : ctx(ctx), ns(ns), filename(filename), buf(buffer.getBuffer()) {
+  current_location.ns = ns;
+};
 
 Reader::~Reader() { READER_LOG("Destroying the reader"); }
 
@@ -136,7 +147,7 @@ exprs::Node Reader::readNumber(bool neg) {
   loc.start = getCurrentLocation();
 
   while (!isEndOfBuffer(c) &&
-         ((!(isspace(*c)) && ((isdigit(*c)) | (*c == '.'))))) {
+         ((!(isspace(*c)) && (isdigit(*c) || *c == '.')))) {
 
     if (*c == '.' && floatNum == true) {
       ctx.sourceManager.PrintMessage(
@@ -155,6 +166,14 @@ exprs::Node Reader::readNumber(bool neg) {
     empty = false;
   }
 
+  if (std::isalpha(*c)) {
+    ctx.sourceManager.PrintMessage(
+        llvm::errs(), llvm::SMLoc::getFromPointer(c),
+        ctx.sourceManager.DK_Error,
+        llvm::formatv("Invalid digit for a number. Are you drunk?\n", c));
+    exit(1);
+  }
+
   if (!empty) {
     ungetChar();
     loc.end = getCurrentLocation();
@@ -260,7 +279,7 @@ exprs::Node Reader::readList() {
 
 /// Reads an expression by dispatching to the proper reader function.
 exprs::Node Reader::readExpr() {
-  auto c = getChar(true);
+  auto c = getChar(false);
   READER_LOG("Read char at `readExpr`: " << *c << " << " << current_pos << "|"
                                          << buf.size() << " BB "
                                          << isEndOfBuffer(c));
@@ -274,6 +293,7 @@ exprs::Node Reader::readExpr() {
   case '(': {
     return readList();
   }
+
   default:
     return readSymbol();
   }
@@ -283,11 +303,16 @@ exprs::Node Reader::readExpr() {
 /// Each expression type (from the reader perspective) has a
 /// reader function.
 Result<exprs::Ast> Reader::read() {
-  // auto c = getChar(true);
 
   // while (!isEndOfBuffer(c)) {
   for (size_t current_pos = 0; current_pos < buf.size();) {
-    // ungetChar();
+    auto c = getChar(true);
+    if (isEndOfBuffer(c)) {
+      break;
+    }
+
+    ungetChar();
+
     auto tmp{readExpr()};
 
     if (tmp) {
@@ -301,14 +326,19 @@ Result<exprs::Ast> Reader::read() {
   return Result<exprs::Ast>::success(std::move(this->ast));
 };
 
-Result<exprs::Ast> read(SereneContext &ctx, const llvm::StringRef input) {
-  reader::Reader r(ctx, input);
+Result<exprs::Ast> read(SereneContext &ctx, const llvm::StringRef input,
+                        llvm::StringRef ns,
+                        llvm::Optional<llvm::StringRef> filename) {
+  reader::Reader r(ctx, input, ns, filename);
   auto ast = r.read();
   return ast;
 }
 
-Result<exprs::Ast> read(SereneContext &ctx, const llvm::MemoryBufferRef input) {
-  reader::Reader r(ctx, input);
+Result<exprs::Ast> read(SereneContext &ctx, const llvm::MemoryBufferRef input,
+                        llvm::StringRef ns,
+                        llvm::Optional<llvm::StringRef> filename) {
+  reader::Reader r(ctx, input, ns, filename);
+
   auto ast = r.read();
   return ast;
 }

src/serene/source_mgr.cpp

@@ -87,7 +87,8 @@ NSPtr SourceMgr::readNamespace(SereneContext &ctx, std::string name,
   auto *buf = getMemoryBuffer(bufferId);
 
   // Read the content of the buffer by passing it the reader
-  auto maybeAst = reader::read(ctx, buf->getBuffer());
+  auto maybeAst = reader::read(ctx, buf->getBuffer(), name,
+                               llvm::Optional(llvm::StringRef(includedFile)));
 
   if (!maybeAst) {
     SMGR_LOG("Couldn't Read namespace: " + name)
@@ -97,6 +98,7 @@ NSPtr SourceMgr::readNamespace(SereneContext &ctx, std::string name,
   // Create the NS and set the AST
   auto ns =
       makeNamespace(ctx, name, llvm::Optional(llvm::StringRef(includedFile)));
+
   if (mlir::failed(ns->setTree(maybeAst.getValue()))) {
     SMGR_LOG("Couldn't set the AST for namespace: " + name)
     return nullptr;