ng_literal_decorated: pre-check for narrow reach

src/nfagraph/ng_literal_decorated.cpp

@@ -210,6 +210,11 @@ bool handleDecoratedLiterals(RoseBuild &rose, const NGHolder &g,
         return false;
     }
 
+    if (!hasNarrowReachVertex(g)) {
+        DEBUG_PRINTF("no narrow reach vertices\n");
+        return false;
+    }
+
     if (hasLargeDegreeVertex(g)) {
         DEBUG_PRINTF("large degree\n");
         return false;

src/nfagraph/ng_util.cpp

@@ -257,6 +257,12 @@ bool hasBigCycles(const NGHolder &g) {
     return false;
 }
 
+bool hasNarrowReachVertex(const NGHolder &g, size_t max_reach_count) {
+    return any_of_in(vertices_range(g), [&](NFAVertex v) {
+        return !is_special(v, g) && g[v].char_reach.count() < max_reach_count;
+    });
+}
+
 bool can_never_match(const NGHolder &g) {
     assert(edge(g.accept, g.acceptEod, g).second);
     if (in_degree(g.accept, g) == 0 && in_degree(g.acceptEod, g) == 1) {

src/nfagraph/ng_util.h

@@ -233,6 +233,12 @@ bool hasReachableCycle(const NGHolder &g, NFAVertex src);
 /** True if g has any cycles which are not self-loops. */
 bool hasBigCycles(const NGHolder &g);
 
+/**
+ * \brief True if g has at least one non-special vertex with reach smaller than
+ * max_reach_count. The default of 200 is pretty conservative.
+ */
+bool hasNarrowReachVertex(const NGHolder &g, size_t max_reach_count = 200);
+
 /** Returns the set of all vertices that appear in any of the graph's cycles. */
 std::set<NFAVertex> findVerticesInCycles(const NGHolder &g);
 

src/nfagraph/ng_violet.cpp

@@ -2954,9 +2954,7 @@ RoseInGraph doInitialVioletTransform(const NGHolder &h, bool last_chance,
 
     /* Avoid running the Violet analysis at all on graphs with no vertices with
      * small reach, since we will not be able to extract any literals. */
-    if (all_of_in(vertices_range(h), [&](NFAVertex v) {
+    if (!hasNarrowReachVertex(h)) {
-            return is_special(v, h) || h[v].char_reach.count() >= 200;
-        })) {
         DEBUG_PRINTF("fail, no vertices with small reach\n");
         return vg;
     }