Search optimizations

lib/libZRCola/include/zrcola/character.h

@@ -26,6 +26,7 @@
 #include <map>
 #include <ostream>
 #include <vector>
+#include <set>
 #include <string>
 
 #pragma warning(push)
@@ -51,6 +52,12 @@ namespace ZRCola {
     };
 
 
+    ///
+    /// Blank character category
+    ///
+    const chrcatid_t chrcatid_t_blank = {};
+
+
     ///
     /// Compares two character category IDs
     ///
@@ -151,7 +158,32 @@ namespace ZRCola {
         ///
         inline character_db() : idxChr(data) {}
 
-        void search_by_desc(_In_z_ const wchar_t *str, _Inout_ std::map<wchar_t, unsigned long> &hits, _Inout_ std::map<wchar_t, unsigned long> &hits_sub) const;
+        ///
+        /// Search for characters by description in given categories
+        ///
+        /// \param[in   ] str       Search string
+        /// \param[in   ] cats      Set of categories, character must be a part of
+        /// \param[inout] hits      (character, count) map to append full-word hits to
+        /// \param[inout] hits_sub  (character, count) map to append partial-word hits to
+        ///
+        void Search(_In_z_ const wchar_t *str, _In_ const std::set<chrcatid_t> &cats, _Inout_ std::map<wchar_t, unsigned long> &hits, _Inout_ std::map<wchar_t, unsigned long> &hits_sub) const;
+
+        ///
+        /// Get character category
+        ///
+        /// \param[in] c  Character
+        ///
+        /// \returns
+        /// - Character category if character found
+        /// - `ZRCola::chrcatid_t_blank` otherwise
+        ///
+        inline chrcatid_t GetCharCat(wchar_t c) const
+        {
+            char _chr[sizeof(character)];
+            ((character *)_chr)->chr = c;
+            indexChar::size_type start;
+            return idxChr.find(*((character *)_chr), start) ? idxChr[start].cat : chrcatid_t_blank;
+        }
     };
 
 

lib/libZRCola/include/zrcola/common.h

@@ -248,6 +248,40 @@ namespace ZRCola {
             return false;
         }
 
+        ///
+        /// Search for the first element in the index
+        ///
+        /// \param[in]  el     Element we are looking for (needle)
+        /// \param[out] start  Index of the first matching element found
+        ///
+        /// \returns
+        /// - \c true if found
+        /// - \c false otherwise
+        ///
+        bool find(_In_ const T_data &el, _Out_ size_type &start) const
+        {
+            // Start with the full search area.
+            size_t end;
+            for (start = 0, end = size(); start < end; ) {
+                size_type m = (start + end) / 2;
+                int r = compare(el, at(m));
+                     if (r < 0) end   = m;
+                else if (r > 0) start = m + 1;
+                else {
+                    // Narrow the search area on the left to start at the first element in the run.
+                    for (size_type end2 = m; start < end2;) {
+                        size_type m = (start + end2) / 2;
+                        int r = compare(el, at(m));
+                        if (r <= 0) end2 = m; else start = m + 1;
+                    }
+
+                    return true;
+                }
+            }
+
+            return false;
+        }
+
     private:
         static int __cdecl compare_s(void *p, const void *a, const void *b)
         {