From 10f52346ca4d0e5749a844a73942545fe3695381 Mon Sep 17 00:00:00 2001
From: Justin Viiret <justin.viiret@intel.com>
Date: Tue, 11 Apr 2017 13:56:51 +1000
Subject: [PATCH] smallwrite: bfs ordering, refine daddy selection

---
 src/nfa/mcclellancompile.cpp        | 14 +++++++++++---
 src/smallwrite/smallwrite_build.cpp |  7 +++++++
 2 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/src/nfa/mcclellancompile.cpp b/src/nfa/mcclellancompile.cpp
index e2466000..e875477b 100644
--- a/src/nfa/mcclellancompile.cpp
+++ b/src/nfa/mcclellancompile.cpp
@@ -841,9 +841,17 @@ void find_better_daddy(dfa_info &info, dstate_id_t curr_id, bool using8bit,
 
     flat_set<dstate_id_t> hinted;
     if (trust_daddy_states) {
-        hinted.insert(currState.daddy);
-        addIfEarlier(hinted, info.raw.start_floating, curr_id);
-        addIfEarlier(hinted, info.raw.start_anchored, curr_id);
+        // Use the daddy already set for this state so long as it isn't already
+        // a Sherman state.
+        if (!info.is_sherman(currState.daddy)) {
+            hinted.insert(currState.daddy);
+        } else {
+            // Fall back to granddaddy, which has already been processed (due
+            // to BFS ordering) and cannot be a Sherman state.
+            dstate_id_t granddaddy = info.states[currState.daddy].daddy;
+            assert(!info.is_sherman(granddaddy));
+            hinted.insert(granddaddy);
+        }
     } else {
         hinted = find_daddy_candidates(info, curr_id);
     }
diff --git a/src/smallwrite/smallwrite_build.cpp b/src/smallwrite/smallwrite_build.cpp
index fac8d012..ce3315e8 100644
--- a/src/smallwrite/smallwrite_build.cpp
+++ b/src/smallwrite/smallwrite_build.cpp
@@ -452,6 +452,13 @@ void buildAutomaton(LitTrie &trie,
     ACVisitor ac_vis(trie, failure_map, ordering);
     boost::breadth_first_search(trie, trie.root, visitor(ac_vis));
 
+    // Renumber with BFS ordering, which is assumed by other DFA construction
+    // code (i.e. Sherman state computation).
+    size_t idx = 0;
+    for (auto v : ordering) {
+        trie[v].index = idx++;
+    }
+
     // Compute missing edges from failure map.
     for (auto v : ordering) {
         DEBUG_PRINTF("vertex %zu\n", trie[v].index);