Initial commit, copied from wwwhtml-parser. Works. Need to add unescape of html sequences

1 files changed, 54 insertions, 0 deletions

diff --git a/include/HtmlParser.h b/include/HtmlParser.h
new file mode 100644
index 0000000..72de123
--- /dev/null
+++ b/include/HtmlParser.h
@@ -0,0 +1,54 @@
+#ifndef HTML_PARSER_H
+#define HTML_PARSER_H
+
+/*
+    A small html parser with no dependencies and doesn't dynamically allocate any memory
+    and can parse and repair broken html (just like web browsers)
+*/
+
+#include <stddef.h>
+
+typedef struct {
+    const char *data;
+    size_t size;
+} StringView;
+
+typedef struct HtmlParser HtmlParser;
+
+typedef enum{
+    HTML_PARSE_TAG_START,
+    HTML_PARSE_TAG_END,
+    HTML_PARSE_ATTRIBUTE,
+    HTML_PARSE_TEXT,
+    HTML_PARSE_JAVASCRIPT_CODE
+} HtmlParseType;
+
+typedef void (*HtmlParseCallback)(HtmlParser *html_parser, HtmlParseType parse_type, void *userdata);
+
+#define UNCLOSED_TAGS_SIZE 2048
+
+struct HtmlParser {
+    const char *source;
+    size_t source_len;
+    size_t offset;
+    HtmlParseCallback parse_callback;
+    void *callback_userdata;
+
+    StringView tag_name;
+    StringView attribute_key;
+    StringView attribute_value;
+    StringView text;
+
+    int is_tag_void;
+    int inside_script_tag;
+
+    size_t unclosed_tags_offset;
+    StringView unclosed_tags[UNCLOSED_TAGS_SIZE];
+};
+
+void html_parser_init(HtmlParser *self, const char *html_source, size_t len, HtmlParseCallback parse_callback, void *userdata);
+void html_parser_deinit(HtmlParser *self);
+
+void html_parser_parse(HtmlParser *self);
+
+#endif /* HTML_PARSER_H */
\ No newline at end of file