[css2xpath] placeholders for Rules

src/ecmascript/spidermonkey/css2xpath.c

@@ -131,6 +131,154 @@ dollar_equal_rule_apply(std::string &selector)
 	return preg_replace_callback(pattern, dollar_equal_rule_callback, selector);
 }
 
+class Translator;
+
+class Rule
+{
+	public:
+		std::string apply(std::string &selector)
+		{
+			return selector;
+		}
+	Rule()
+	{
+	}
+};
+
+class RegexRule : public Rule
+{
+	private:
+		const char *pattern, *replacement;
+	public:
+		RegexRule(const char *pat, const char *repl) : pattern(pat), replacement(repl)
+		{
+		}
+};
+
+class NotRule : public Rule
+{
+	public:
+	NotRule(Translator *t)
+	{
+	}
+};
+
+class NthChildRule : public Rule
+{
+	public:
+	NthChildRule()
+	{
+	}
+};
+
+class DollarEqualRule : public Rule
+{
+	public:
+	DollarEqualRule()
+	{
+	}
+};
+
+class Translator
+{
+	public:
+		std::string translate(std::string & selector)
+		{
+			for (auto r : rules)
+			{
+				selector = r->apply(selector);
+			}
+
+			return selector == "/" ? "/" : ("//" + selector);
+		}
+
+	Translator()
+	{
+	}
+
+private:
+	Rule *rules[33] = {
+		// prefix|name
+		new RegexRule("([a-zA-Z0-9\\_\\-\\*]+)\\|([a-zA-Z0-9\\_\\-\\*]+)", "$1:$2"),
+
+		// add @ for attribs
+		new RegexRule("\\[([^G\\]~\\$\\*\\^\\|\\!]+)(=[^\\]]+)?\\]", "[@$1$2]"),
+
+		// multiple queries
+		new RegexRule("\\s*,\\s*", "|"),
+		// , + ~ >
+		new RegexRule("\\s*([\\+~>])\\s*", "$1"),
+
+		//* ~ + >
+		new RegexRule("([a-zA-Z0-9\\_\\-\\*])~([a-zA-Z0-9\\_\\-\\*])", "$1/following-sibling::$2"),
+		new RegexRule("([a-zA-Z0-9\\_\\-\\*])\\+([a-zA-Z0-9\\_\\-\\*])", "$1/following-sibling::*[1]/self::$2"),
+		new RegexRule("([a-zA-Z0-9\\_\\-\\*])>([a-zA-Z0-9\\_\\-\\*])", "$1/$2"),
+		// all unescaped stuff escaped
+		new RegexRule("\\[([^=]+)=([^'|][^\\]]*)\\]", "[$1=\"$2\"]"),
+		// all descendant or self to //
+		new RegexRule("(^|[^a-zA-Z0-9\\_\\-\\*])([#\\.])([a-zA-Z0-9\\_\\-]+)", "$1*$2$3"),
+		new RegexRule("([\\>\\+\\|\\~\\,\\s])([a-zA-Z\\*]+)", "$1//$2"),
+		new RegexRule("\\s+\\/\\//", "//"),
+
+		// :first-child
+		new RegexRule("([a-zA-Z0-9\\_\\-\\*]+):first-child", "*[1]/self::$1"),
+
+		// :last-child
+		new RegexRule("([a-zA-Z0-9\\_\\-\\*]+)?:last-child", "$1[not(following-sibling::*)]"),
+
+		// :only-child
+		new RegexRule("([a-zA-Z0-9\\_\\-\\*]+):only-child", "*[last()=1]/self::$1"),
+
+		// :empty
+		new RegexRule("([a-zA-Z0-9\\_\\-\\*]+)?:empty", "$1[not(*) and not(normalize-space())]"),
+
+		// :not
+		new NotRule(this),
+
+		// :nth-child
+		new NthChildRule(),
+
+		// :contains(selectors)
+		new RegexRule(":contains\\(([^\\)]*)\\)", "[contains(string(.),\"$1\")]"),
+
+		// |= attrib
+		new RegexRule("\\[([a-zA-Z0-9\\_\\-]+)\\|=([^\\]]+)\\]", "[@$1=$2 or starts-with(@$1,concat($2,\"-\"))]"),
+
+		// *= attrib
+		new RegexRule("\\[([a-zA-Z0-9\\_\\-]+)\\*=([^\\]]+)\\]", "[contains(@$1,$2)]"),
+
+		// ~= attrib
+		new RegexRule("\\[([a-zA-Z0-9\\_\\-]+)~=([^\\]]+)\\]", "[contains(concat(\" \",normalize-space(@$1),\" \"),concat(\" \",$2,\" \"))]"),
+
+		// ^= attrib
+		new RegexRule("\\[([a-zA-Z0-9\\_\\-]+)\\^=([^\\]]+)\\]", "[starts-with(@$1,$2)]"),
+
+		// $= attrib
+		new DollarEqualRule(),
+
+		// != attrib
+		new RegexRule("\\[([a-zA-Z0-9\\_\\-]+)\\!=[\\\"']+?([^\\\"\\]]+)[\\\"']+?\\]", "[not(@$1) or @$1!=\"$2\"]"),
+
+		// ids
+		new RegexRule("#([a-zA-Z0-9\\_\\-]+)", "[@id=\"$1\"]"),
+
+		// classes
+		new RegexRule("\\.([a-zA-Z0-9_-]+)(?![^[]*])", "[contains(concat(\" \",normalize-space(@class),\" \"),\" $1 \")]"),
+
+		// normalize multiple filters
+		new RegexRule("\\]\\[([^\\]]+)", " and ($1)"),
+
+		// tag:pseudo selectors
+		new RegexRule("(:enabled)", "[not(@disabled)]"),
+		new RegexRule("(:checked)", "[@checked=\"checked\"]"),
+		new RegexRule(":(disabled)", "[@$1]"),
+		new RegexRule(":root", "/"),
+
+		// use * when tag was omitted
+		new RegexRule("^\[", "*["),
+		new RegexRule("\\|\\[", "|*[")
+	};
+};
 
 #if 0