commit 66f078e: [Project] Css: Start semantic parsing for rules

[Vsevolod Stakhov]

Author: Vsevolod Stakhov
Date: 2021-02-17 15:56:58 +0000
URL: https://github.com/rspamd/rspamd/commit/66f078e7f124c357eae417c01837e25761701128

[Project] Css: Start semantic parsing for rules

---
 src/libserver/css/css_parser.cxx   | 47 +++++++++++++++++++++++++++++++++++++-
 src/libserver/css/css_selector.hxx |  7 ++++--
 test/lua/unit/css.lua              |  6 ++---
 3 files changed, 54 insertions(+), 6 deletions(-)

diff --git a/src/libserver/css/css_parser.cxx b/src/libserver/css/css_parser.cxx
index 084d1b50a..a0a9d8847 100644
--- a/src/libserver/css/css_parser.cxx
+++ b/src/libserver/css/css_parser.cxx
@@ -22,6 +22,7 @@
 
 namespace rspamd::css {
 
+struct css_consumed_block;
 /*
  * Represents a consumed token by a parser
  */
@@ -33,7 +34,8 @@ struct css_consumed_block {
 		css_simple_block,
 		css_function,
 		css_function_arg,
-		css_component
+		css_component,
+		css_selector,
 	};
 
 	using consumed_block_ptr = std::unique_ptr<css_consumed_block>;
@@ -81,6 +83,17 @@ struct css_consumed_block {
 		content = std::move(tok);
 	}
 
+	/* Empty blocks used to avoid type checks in loops */
+	const inline static std::vector<consumed_block_ptr> empty_block_vec{};
+
+	auto get_blocks_or_empty() const -> const std::vector<consumed_block_ptr>& {
+		if (content.index() == 1) {
+			return std::get<std::vector<consumed_block_ptr>>(content);
+		}
+
+		return empty_block_vec;
+	}
+
 	auto token_type_str(void) const -> const char *
 	{
 		const auto *ret = "";
@@ -107,6 +120,9 @@ struct css_consumed_block {
 		case parser_tag_type::css_component:
 			ret = "component";
 			break;
+		case parser_tag_type::css_selector:
+			ret = "selector";
+			break;
 		}
 
 		return ret;
@@ -570,6 +586,35 @@ bool css_parser::consume_input(const std::string_view &sv)
 
 	}
 
+	const auto &rules = consumed_blocks->get_blocks_or_empty();
+
+	for (auto &&rule : rules) {
+		/*
+		 * For now, we do not need any of the at rules, so we can safely ignore them
+		 */
+		auto &&children = rule->get_blocks_or_empty();
+
+		if (children.size() > 1 &&
+			children[0]->tag == css_consumed_block::parser_tag_type::css_component) {
+			auto simple_block = std::find_if(children.begin(), children.end(),
+					[](auto &bl) {
+						return bl->tag == css_consumed_block::parser_tag_type::css_simple_block;
+					});
+
+			if (simple_block != children.end()) {
+				/*
+				 * We have a component and a simple block,
+				 * so we can parse a declaration
+				 */
+
+				/* First, tag all components as preamble */
+				for (auto it = children.begin(); it != simple_block; ++it) {
+					(*it)->tag = css_consumed_block::parser_tag_type::css_selector;
+				}
+			}
+		}
+	}
+
 	auto debug_str = consumed_blocks->debug_str();
 	msg_debug_css("consumed css: {%*s}", (int)debug_str.size(), debug_str.data());
 
diff --git a/src/libserver/css/css_selector.hxx b/src/libserver/css/css_selector.hxx
index c9f3046d5..8611630fd 100644
--- a/src/libserver/css/css_selector.hxx
+++ b/src/libserver/css/css_selector.hxx
@@ -24,6 +24,7 @@
 #include <optional>
 #include "contrib/expected/expected.hpp"
 #include "parse_error.hxx"
+#include "css_tokeniser.hxx"
 #include "html_tags.h"
 
 namespace rspamd::css {
@@ -41,14 +42,14 @@ struct css_selector {
 	selector_type type;
 	std::variant<tag_id_t, std::string> value;
 
-	constexpr std::optional<tag_id_t> to_tag (void) const {
+	 auto to_tag(void) const -> std::optional<tag_id_t> {
 		if (type == selector_type::SELECTOR_ELEMENT) {
 			return std::get<tag_id_t>(value);
 		}
 		return std::nullopt;
 	}
 
-	constexpr std::optional<const std::string_view> to_string (void) const {
+	auto to_string(void) const -> std::optional<const std::string_view> {
 		if (type == selector_type::SELECTOR_ELEMENT) {
 			return std::string_view(std::get<std::string>(value));
 		}
@@ -59,6 +60,8 @@ struct css_selector {
 															   size_t inlen);
 };
 
+
+
 }
 
 #endif //RSPAMD_CSS_SELECTOR_HXX
diff --git a/test/lua/unit/css.lua b/test/lua/unit/css.lua
index e040a6fd3..a5a8f533f 100644
--- a/test/lua/unit/css.lua
+++ b/test/lua/unit/css.lua
@@ -50,7 +50,7 @@ p {
 p:first-child {
   color: blue;
 }
-a[target=_blank] {
+a[target=_blank] #id{
   background-color: yellow;
 }
 * {
@@ -75,7 +75,7 @@ body {
 }
 
 /* Style the topnav links */
-.topnav a {
+.topnav a{
   float: left;
   display: block;
   color: #f2f2f2;
@@ -85,7 +85,7 @@ body {
 }
 
 /* Clear floats after the columns */
-.row:after {
+.row:after{
   content: "";
   display: table;
   clear: both;