From add95b14f74e6dbe04a6efe98ff0f20424930b73 Mon Sep 17 00:00:00 2001
From: Stefan Weigl-Bosker <stefan@s00.xyz>
Date: Tue, 3 Feb 2026 14:59:53 -0500
Subject: [willow]: initial frontend work, unit tests (#8)

---
 willow/tools/willowc/include/ast.hpp           | 62 ++++++++++++++++++++++
 willow/tools/willowc/include/expr.hpp          |  4 ++
 willow/tools/willowc/include/parser.hpp        | 33 ++++++++++++
 willow/tools/willowc/include/sourcemanager.hpp | 26 ++++++++++
 willow/tools/willowc/include/tokenizer.hpp     | 72 ++++++++++++++++++++++++++
 5 files changed, 197 insertions(+)
 create mode 100644 willow/tools/willowc/include/ast.hpp
 create mode 100644 willow/tools/willowc/include/expr.hpp
 create mode 100644 willow/tools/willowc/include/parser.hpp
 create mode 100644 willow/tools/willowc/include/sourcemanager.hpp
 create mode 100644 willow/tools/willowc/include/tokenizer.hpp

(limited to 'willow/tools/willowc/include')

diff --git a/willow/tools/willowc/include/ast.hpp b/willow/tools/willowc/include/ast.hpp
new file mode 100644
index 0000000..8c59067
--- /dev/null
+++ b/willow/tools/willowc/include/ast.hpp
@@ -0,0 +1,62 @@
+#ifndef WILLOWC_INCLUDE_AST_HPP
+#define WILLOWC_INCLUDE_AST_HPP
+
+#include <cstdint>
+#include <memory>
+#include <vector>
+
+#include <willow/IR/Instructions.h>
+#include <willow/IR/Types.h>
+
+#include <tokenizer.hpp>
+
+namespace willowc {
+
+using Opcode = willow::Instruction::Opcode;
+using TokenIndex = std::size_t;
+
+// this is like willow::ValueKind, but treats groups all ssa values into 'Value'
+// (because they can't be differentiated by syntax alone)
+enum class ExprKind { Constant, BasicBlock, Function, Value };
+
+struct ExprAST {
+  ExprKind kind;
+
+  std::string name;
+  // token??
+};
+
+struct InstAST {
+  Opcode op;
+  std::string name;
+
+  std::vector<std::unique_ptr<ExprAST>> args;
+};
+
+struct BlockAST {
+  std::string label;
+  std::vector<std::unique_ptr<InstAST>> body;
+};
+
+struct ParameterAST {
+  std::string name;
+  willow::Type type;
+};
+
+struct FunctionDeclAST {
+  std::string name;
+  std::vector<std::unique_ptr<ParameterAST>> parameters;
+  std::string returntype;
+
+  std::vector<std::unique_ptr<BlockAST>> body;
+  // TODO: movable symbol table
+};
+
+struct ModuleAST {
+  std::vector<std::unique_ptr<FunctionDeclAST>> Functions;
+  // TODO: imports, symbol table
+};
+
+}; // namespace willowc
+
+#endif // WILLOWC_INCLUDE_AST_HPP
diff --git a/willow/tools/willowc/include/expr.hpp b/willow/tools/willowc/include/expr.hpp
new file mode 100644
index 0000000..15d2985
--- /dev/null
+++ b/willow/tools/willowc/include/expr.hpp
@@ -0,0 +1,4 @@
+#ifndef WILLOWC_INCLUDE_EXPR_HPP
+#define WILLOWC_INCLUDE_EXPR_HPP
+
+#endif // WILLOWC_INCLUDE_EXPR_HPP
diff --git a/willow/tools/willowc/include/parser.hpp b/willow/tools/willowc/include/parser.hpp
new file mode 100644
index 0000000..825dfdd
--- /dev/null
+++ b/willow/tools/willowc/include/parser.hpp
@@ -0,0 +1,33 @@
+#ifndef WILLOWC_INCLUDE_PARSER_HPP
+#define WILLOWC_INCLUDE_PARSER_HPP
+
+#include <tokenizer.hpp>
+#include <ast.hpp>
+
+#include <optional>
+#include <memory>
+#include <vector>
+
+namespace willowc {
+
+class Parser {
+  std::string_view buf;
+
+  std::vector<TokenKind> kinds;
+  std::vector<std::size_t> starts;
+  Tokenizer tokenizer;
+
+  std::size_t pos;
+
+public:
+  Parser(std::string_view buf) : buf(buf), tokenizer(buf) {}
+
+  std::optional<std::unique_ptr<ModuleAST>> parse();
+
+  TokenKind kind() const { return kinds[pos]; }
+  std::size_t start() const { return starts[pos]; }
+};
+
+} // namespace willowc
+
+#endif // WILLOWC_INCLUDE_PARSER_HPP
diff --git a/willow/tools/willowc/include/sourcemanager.hpp b/willow/tools/willowc/include/sourcemanager.hpp
new file mode 100644
index 0000000..a526e48
--- /dev/null
+++ b/willow/tools/willowc/include/sourcemanager.hpp
@@ -0,0 +1,26 @@
+#ifndef WILLOWC_INCLUDE_SOURCEMANAGER_HPP
+#define WILLOWC_INCLUDE_SOURCEMANAGER_HPP
+
+#include <filesystem>
+#include <memory>
+#include <string>
+#include <vector>
+
+namespace willowc {
+
+using FileID = std::uint32_t;
+
+class SourceManager {
+struct File {
+  std::string path;
+  std::unique_ptr<char[]> buf;
+};
+public:
+  std::optional<FileID> addFile(std::string_view path);
+private:
+  std::vector<File> file_table;
+};
+
+} // namespace willowc
+
+#endif // WILLOWC_INCLUDE_SOURCEMANAGER_HPP
diff --git a/willow/tools/willowc/include/tokenizer.hpp b/willow/tools/willowc/include/tokenizer.hpp
new file mode 100644
index 0000000..3de9d32
--- /dev/null
+++ b/willow/tools/willowc/include/tokenizer.hpp
@@ -0,0 +1,72 @@
+#ifndef WILLOWC_INCLUDE_TOKENIZER_HPP
+#define WILLOWC_INCLUDE_TOKENIZER_HPP
+
+#include <willow/IR/Location.h>
+
+namespace willowc {
+
+enum class TokenKind {
+  Function,
+  Variable,
+  Constant,
+  Type,
+  Label,
+  Inst,
+
+  Comma,
+  Semicolon,
+  LParen,
+  RParen,
+  LCurly,
+  RCurly,
+  Equals,
+  RArrow,
+  Comment,
+
+  FuncKW,
+  Eof,
+  Invalid,
+};
+
+struct Token {
+  std::size_t start, end;
+  TokenKind kind;
+};
+
+class Tokenizer {
+  std::string_view buf;
+  std::size_t offset;
+
+  void skip(std::size_t idx = 1) { offset += idx; }
+
+  char eat(std::size_t num = 1) {
+    if (offset >= buf.length())
+      return '\0';
+
+    char c = buf[offset];
+    offset += num;
+    return c;
+  }
+
+  char peek(std::size_t idx = 0) {
+    if (offset + idx >= buf.length())
+      return '\0';
+
+    return buf[offset + idx];
+  }
+
+  bool scan_id(bool accept_digits);
+  bool scan_dec();
+  bool scan_hex();
+  bool scan_constant();
+public:
+  explicit Tokenizer(std::string_view buf, std::size_t offset = 0)
+      : buf{buf}, offset{offset} {}
+
+  Token scan();
+  void seek(uint64_t offset);
+};
+
+} // namespace willowc
+
+#endif // WILLOWC_INCLUDE_TOKENIZER_HPP
-- 
cgit v1.2.3