css: speedup class matching

Class name matching is slow. It's extremely slow when there are many
class selectors. Mitigate this problem by:

1. when parsing css, store class name's hash value
2. during css check, split class names and calculate their hash.
3. for class selectors compare integers first to avoid expensive
   class name matching for most rules.

crengine/include/lvstring.h

@@ -820,7 +820,9 @@ class lString32
     value_type firstChar() { return empty() ? 0 : at(0); }
 
     /// calculates hash for string
+    static lUInt32 getHash(const lChar32 *begin, const lChar32 *end);
     lUInt32 getHash() const;
+
     /// returns character at specified position, with index bounds checking, fatal error if fails
     value_type & at( size_type pos ) { if ((unsigned)pos > (unsigned)pchunk->len) crFatalError(); return modify()[pos]; }
     /// returns character at specified position, without index bounds checking

crengine/include/lvstsheet.h

@@ -132,16 +132,23 @@ class LVCssSelectorRule
     lUInt16 _attrid;
     LVCssSelectorRule * _next;
     lString32 _value;
+    lUInt32 _valueHash = 0;
 public:
     LVCssSelectorRule(LVCssSelectorRuleType type)
     : _type(type), _id(0), _attrid(0), _next(NULL)
     { }
     LVCssSelectorRule( LVCssSelectorRule & v );
     void setId( lUInt16 id ) { _id = id; }
-    void setAttr( lUInt16 id, const lString32 value ) { _attrid = id; _value = value; }
+    void setAttr( lUInt16 id, const lString32 value ) {
+        _attrid = id;
+        _value = value;
+        _valueHash = _value.getHash();
+    }
     const LVCssSelectorRule * getNext() const { return _next; }
     void setNext(LVCssSelectorRule * next) { _next = next; }
     ~LVCssSelectorRule() { if (_next) delete _next; }
+    // A fail-fast check, returning false to rule out a match.
+    bool quickCheck(const lUInt32 *classHashes, size_t size) const;
     /// check condition for node
     bool check( const ldomNode * & node, bool allow_cache=true ) const;
     /// check next rules for node
@@ -178,6 +185,7 @@ class LVCssSelector {
     bool parse( const char * &str, lxmlDocBase * doc );
     lUInt16 getElementNameId() const { return _id; }
     bool check( const ldomNode * node, bool allow_cache=true ) const;
+    bool quickCheck(const lUInt32 *classHashes, size_t size) const;
     void applyToPseudoElement( const ldomNode * node, css_style_rec_t * style ) const;
     void apply( const ldomNode * node, css_style_rec_t * style ) const
     {

crengine/src/lvstring.cpp

@@ -1274,6 +1274,14 @@ bool lString32::atod( double &d, char dp ) const {
 }
 
 #define STRING_HASH_MULT 31
+
+lUInt32 lString32::getHash(const lChar32 *begin, const lChar32 *end) {
+    lUInt32 res = 0;
+    for (; begin < end; ++begin)
+        res = res * STRING_HASH_MULT + *begin;
+    return res;
+}
+
 lUInt32 lString32::getHash() const
 {
     lUInt32 res = 0;

crengine/src/lvstsheet.cpp

@@ -4922,6 +4922,16 @@ lUInt32 LVCssSelectorRule::getWeight() const {
     return 0;
 }
 
+bool LVCssSelectorRule::quickCheck(const lUInt32 *classHashes, size_t size) const {
+    if (_type != cssrt_class)
+        return true;
+    for (size_t i = 0; i < size; ++i) {
+        if (classHashes[i] == _valueHash)
+            return true;
+    }
+    return false;
+}
+
 bool LVCssSelectorRule::check( const ldomNode * & node, bool allow_cache ) const
 {
     if (!node || node->isNull() || node->isRoot())
@@ -5453,6 +5463,10 @@ bool LVCssSelector::check( const ldomNode * node, bool allow_cache ) const
     return true;
 }
 
+bool LVCssSelector::quickCheck(const lUInt32 *classHashes, size_t size) const {
+    return !_rules || _rules->quickCheck(classHashes, size);
+}
+
 bool parse_attr_value( const char * &str, char * buf, bool &parse_trailing_i, char stop_char=']' )
 {
     int pos = 0;
@@ -5953,6 +5967,7 @@ LVCssSelectorRule::LVCssSelectorRule( LVCssSelectorRule & v )
 : _type(v._type), _id(v._id), _attrid(v._attrid)
 , _next(NULL)
 , _value( v._value )
+, _valueHash(v._valueHash)
 {
     if ( v._next )
         _next = new LVCssSelectorRule( *v._next );
@@ -5988,6 +6003,21 @@ LVStyleSheet::LVStyleSheet( LVStyleSheet & sheet )
     _selector_count = sheet._selector_count;
 }
 
+template<typename F>
+static void for_each_split(const lChar32 *begin, F functor) {
+    const lChar32 *end = begin;
+    while (*end) {
+        if (*end == ' ') {
+            if (end > begin)
+                functor(begin, end);
+            begin = end + 1;
+        }
+        ++end;
+    }
+    if (end > begin)
+        functor(begin, end);
+}
+
 void LVStyleSheet::apply( const ldomNode * node, css_style_rec_t * style ) const
 {
     if (!_selectors.length())
@@ -6019,14 +6049,25 @@ void LVStyleSheet::apply( const ldomNode * node, css_style_rec_t * style ) const
     LVCssSelector * selector_0 = _selectors[0];
     LVCssSelector * selector_id = id>0 && id<_selectors.length() ? _selectors[id] : NULL;
 
+    LVArray<lUInt32> class_hash_array;
+    bool class_hash_inited = false;
+
     for (;;)
     {
         if (selector_0!=NULL)
         {
+            if (!class_hash_inited) {
+                class_hash_inited = true;
+                const lString32 &v = node->getAttributeValue(attr_class);
+                for_each_split(v.c_str(), [&](const lChar32 *begin, const lChar32 *end) {
+                    class_hash_array.add(lString32::getHash(begin, end));
+                });
+            }
             if (selector_id==NULL || selector_0->getSpecificity() < selector_id->getSpecificity() )
             {
                 // step by sel_0
-                selector_0->apply( node, style );
+                if (selector_0->quickCheck(class_hash_array.ptr(), class_hash_array.length()))
+                    selector_0->apply( node, style );
                 selector_0 = selector_0->getNext();
             }
             else