Implemented tokenizer.HEADmaster

1 files changed, 54 insertions, 0 deletions

diff --git a/wv_parse.h b/wv_parse.h
new file mode 100644
index 0000000..ca3f8ac
--- /dev/null
+++ b/wv_parse.h
@@ -0,0 +1,54 @@
+#ifndef WV_PARSE_H
+#define WV_PARSE_H
+
+#include <stddef.h>
+
+#include "wv_mem.h"
+#include "wv_dom.h"
+#include "wv_vec.h"
+
+typedef enum {
+	WV_TOK_EOF = 0,
+	WV_TOK_TAG_OPEN,    /* E.g., "div" from <div */
+	WV_TOK_TAG_CLOSE,   /* E.g., "div" from </div> or just > */
+	WV_TOK_ATTR_KEY,
+	WV_TOK_ATTR_VAL,
+	WV_TOK_TEXT,
+	WV_TOK_ERROR
+} wv_token_type;
+
+struct wv_token {
+	wv_token_type type;
+	const char *start;
+	size_t len;
+};
+
+typedef enum {
+	WV_STATE_DATA,      /* Outside of any tags, looking for < */
+	WV_STATE_TAG,       /* In <...>, looking for tag name or attr keys */
+	WV_STATE_ATTR_VAL   /* Found a '=', looking for quoted value */
+} wv_state;
+
+struct wv_tokenizer {
+	const char *src;
+	size_t pos;
+	size_t len;
+	wv_state state;
+	char quote_char;    /* Keeps track of " vs ' for current value */
+};
+
+struct wv_parser {
+	struct wv_tokenizer tokenizer;
+	struct wv_arena *arena;
+	struct wv_vec stack;
+	wv_ref doc_ref;
+	wv_ref current_node;
+};
+
+void wv_tokenizer_init(struct wv_tokenizer *t, const char *src, 
+	size_t len);
+struct wv_token wv_tokenizer_next(struct wv_tokenizer *t);
+wv_ref wv_parse_document(struct wv_arena *arena, const char *src);
+
+#endif /* WV_PARSE_H */
+