mcsheng_compile: factor out find_daddy_candidates

diff --git a/src/nfa/mcsheng_compile.cpp b/src/nfa/mcsheng_compile.cpp
index 2d9658f2775e1ea3bbf88b3f994d146568ef591d..2049fee03d9551c34e067f4380a8809d3e4908d9 100644 (file)
--- a/src/nfa/mcsheng_compile.cpp
+++ b/src/nfa/mcsheng_compile.cpp
@@ -617,7 +617,7 @@ void fill_in_succ_table_16(NFA *nfa, const dfa_info &info,
 #define MAX_SHERMAN_LIST_LEN 8
 
 static
-void addIfEarlier(set<dstate_id_t> &dest, dstate_id_t candidate,
+void addIfEarlier(flat_set<dstate_id_t> &dest, dstate_id_t candidate,
                   dstate_id_t max) {
     if (candidate < max) {
         dest.insert(candidate);
@@ -625,13 +625,35 @@ void addIfEarlier(set<dstate_id_t> &dest, dstate_id_t candidate,
 }
 
 static
-void addSuccessors(set<dstate_id_t> &dest, const dstate &source,
+void addSuccessors(flat_set<dstate_id_t> &dest, const dstate &source,
                    u16 alphasize, dstate_id_t curr_id) {
     for (symbol_t s = 0; s < alphasize; s++) {
         addIfEarlier(dest, source.next[s], curr_id);
     }
 }
 
+/* \brief Returns a set of states to search for a better daddy. */
+static
+flat_set<dstate_id_t> find_daddy_candidates(const dfa_info &info,
+                                            dstate_id_t curr_id) {
+    flat_set<dstate_id_t> hinted;
+
+    addIfEarlier(hinted, 0, curr_id);
+    addIfEarlier(hinted, info.raw.start_anchored, curr_id);
+    addIfEarlier(hinted, info.raw.start_floating, curr_id);
+
+    // Add existing daddy and his successors, then search back one generation.
+    const u16 alphasize = info.impl_alpha_size;
+    dstate_id_t daddy = info.states[curr_id].daddy;
+    for (u32 level = 0; daddy && level < 2; level++) {
+        addIfEarlier(hinted, daddy, curr_id);
+        addSuccessors(hinted, info.states[daddy], alphasize, curr_id);
+        daddy = info.states[daddy].daddy;
+    }
+
+    return hinted;
+}
+
 #define MAX_SHERMAN_SELF_LOOP 20
 
 static
@@ -671,22 +693,7 @@ void find_better_daddy(dfa_info &info, dstate_id_t curr_id,
     dstate_id_t best_daddy = 0;
     dstate &currState = info.states[curr_id];
 
-    set<dstate_id_t> hinted; /* set of states to search for a better daddy */
-    addIfEarlier(hinted, 0, curr_id);
-    addIfEarlier(hinted, info.raw.start_anchored, curr_id);
-    addIfEarlier(hinted, info.raw.start_floating, curr_id);
-
-    dstate_id_t mydaddy = currState.daddy;
-    if (mydaddy) {
-        addIfEarlier(hinted, mydaddy, curr_id);
-        addSuccessors(hinted, info.states[mydaddy], alphasize, curr_id);
-        dstate_id_t mygranddaddy = info.states[mydaddy].daddy;
-        if (mygranddaddy) {
-            addIfEarlier(hinted, mygranddaddy, curr_id);
-            addSuccessors(hinted, info.states[mygranddaddy], alphasize,
-                          curr_id);
-        }
-    }
+    flat_set<dstate_id_t> hinted = find_daddy_candidates(info, curr_id);
 
     for (const dstate_id_t &donor : hinted) {
         assert(donor < curr_id);