[willow]: initial frontend work, unit tests (#8)dev/stefan

1 files changed, 72 insertions, 0 deletions

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