feat: started project and lexer

src/_include/mrk/lexer_internal.h

@@ -0,0 +1,40 @@
+#ifndef MRK_LEXER_INTERNAL
+#define MRK_LEXER_INTERNAL
+
+#include "mrk/lexer.h"
+
+struct mrk_lexer {
+  struct {
+    const char *s;
+    size_t len;
+  } buf;
+  struct {
+    size_t line;
+    size_t line_index;
+    size_t buf_index;
+  } pos;
+  struct {
+    size_t start;
+    size_t end;
+  } token;
+};
+
+/**
+ * Return the next character that would be consumed by a call to advance. At the
+ * end of the buffer, this value is nul.
+ */
+char mrk_lexer_peek(mrk_lexer *lexer);
+
+/**
+ * Advance the current position by one character, adding the new character to
+ * the curent token's context and returning it.
+ */
+char mrk_lexer_advance(mrk_lexer *lexer);
+
+/**
+ * Output the currently matched token to the token struct with the given type,
+ * and reset the lexer's tracked token.
+ */
+void mrk_lexer_emit(mrk_token *out, mrk_lexer *lexer, mrk_token_type type);
+
+#endif

src/lexer/lexer.c

@@ -0,0 +1,73 @@
+#include "mrk/lexer_internal.h"
+
+mrk_err mrk_lexer_init(mrk_lexer **out) {
+  MRK_CALLOC(out, 1, sizeof(mrk_lexer));
+
+  return mrk_err_ok;
+}
+
+void mrk_lexer_open(mrk_lexer *lexer, const char *buf, size_t len) {
+  lexer->buf.s = buf;
+  lexer->buf.len = len;
+  lexer->pos.line = 0;
+  lexer->pos.buf_index = 0;
+  lexer->token.start = 0;
+  lexer->token.end = 0;
+}
+
+bool mrk_lexer_at_end(const mrk_lexer *lexer) {
+  return (lexer->buf.len > 0 && lexer->pos.buf_index == lexer->buf.len) ||
+         (lexer->buf.s[lexer->pos.buf_index] == '\0');
+}
+
+char mrk_lexer_advance(mrk_lexer *lexer) {
+  if (mrk_lexer_at_end(lexer)) {
+    return '\0';
+  }
+
+  char c = lexer->buf.s[lexer->pos.buf_index];
+
+  // A newline is still part of the previous line, so if the last character was
+  // a newline, we now go to the next line
+  if (lexer->buf.s[lexer->pos.buf_index] == '\0') {
+    lexer->pos.line++;
+    lexer->pos.line_index = 0;
+  } else {
+    lexer->pos.line_index++;
+  }
+
+  lexer->pos.buf_index++;
+  lexer->token.end++;
+
+  return c;
+}
+
+char mrk_lexer_peek(mrk_lexer *lexer) {
+  if (mrk_lexer_at_end(lexer)) {
+    return '\0';
+  }
+
+  return lexer->buf.s[lexer->pos.buf_index];
+}
+
+void mrk_lexer_emit(mrk_token *out, mrk_lexer *lexer, mrk_token_type type) {
+  out->type = type;
+  out->start = lexer->token.start;
+  out->start = lexer->token.end;
+
+  lexer->token.start = lexer->token.end;
+}
+
+mrk_lexer_err mrk_lexer_next(mrk_token *out, mrk_lexer *lexer) {
+  if (mrk_lexer_at_end(lexer)) {
+    return mrk_lexer_err_done;
+  }
+
+  switch (mrk_lexer_advance(lexer)) {
+  case '#':
+    mrk_lexer_emit(out, lexer, mrk_token_type_pound);
+    break;
+  }
+
+  return mrk_lexer_err_ok;
+}