ng_find_matches: limit how big we're willing to go

Tests which require tracking more than 15K states (including edit
distance states) are very, very slow.

diff --git a/unit/internal/nfagraph_find_matches.cpp b/unit/internal/nfagraph_find_matches.cpp
index 99fdb09ed78a78a571a3cd11c1bb3c80a8d35163..92c514d85f4e7768dfc93e39884fff140b48ad0d 100644 (file)
--- a/unit/internal/nfagraph_find_matches.cpp
+++ b/unit/internal/nfagraph_find_matches.cpp
@@ -212,7 +212,8 @@ TEST_P(MatchesTest, Check) {
     bool utf8 = (t.flags & HS_FLAG_UTF8) > 0;
 
     set<pair<size_t, size_t>> matches;
-    findMatches(*g, rm, t.input, matches, 0, t.notEod, utf8);
+    bool success = findMatches(*g, rm, t.input, matches, 0, t.notEod, utf8);
+    ASSERT_TRUE(success);
 
     set<pair<size_t, size_t>> expected(begin(t.matches), end(t.matches));
 

diff --git a/util/ng_find_matches.cpp b/util/ng_find_matches.cpp
index b3c81574262ce03f079bdb9e22d81aa9ef19ea81..0890319d55178d23a47f32cbc21132b99b72f0ea 100644 (file)
--- a/util/ng_find_matches.cpp
+++ b/util/ng_find_matches.cpp
@@ -52,6 +52,9 @@ using StateBitSet = boost::dynamic_bitset<>;
 
 namespace {
 
+/** \brief Max number of states (taking edit distance into account). */
+static constexpr size_t STATE_COUNT_MAX = 15000;
+
 // returns all successors up to a given depth in a vector of sets, indexed by
 // zero-based depth from source vertex
 static
@@ -1034,7 +1037,7 @@ void filterMatches(MatchSet &matches) {
  *
  *  Fills \a matches with offsets into the data stream where a match is found.
  */
-void findMatches(const NGHolder &g, const ReportManager &rm,
+bool findMatches(const NGHolder &g, const ReportManager &rm,
                  const string &input, MatchSet &matches,
                  const u32 edit_distance, const bool notEod, const bool utf8) {
     assert(hasCorrectlyNumberedVertices(g));
@@ -1042,7 +1045,12 @@ void findMatches(const NGHolder &g, const ReportManager &rm,
     // compile time, so make it an assert
     assert(!edit_distance || !utf8);
 
-    DEBUG_PRINTF("Finding matches\n");
+    const size_t total_states = num_vertices(g) * (3 * edit_distance + 1);
+    DEBUG_PRINTF("Finding matches (%zu total states)\n", total_states);
+    if (total_states > STATE_COUNT_MAX) {
+        DEBUG_PRINTF("too big\n");
+        return false;
+    }
 
     GraphCache gc(edit_distance, g);
 #ifdef DEBUG
@@ -1068,7 +1076,7 @@ void findMatches(const NGHolder &g, const ReportManager &rm,
                      state.next.count());
         if (state.next.empty()) {
             filterMatches(matches);
-            return;
+            return true;
         }
         state.states = state.next;
         state.prev = state.cur;
@@ -1086,4 +1094,5 @@ void findMatches(const NGHolder &g, const ReportManager &rm,
     getMatches(g, matches, state, !notEod);
 
     filterMatches(matches);
+    return true;
 }

diff --git a/util/ng_find_matches.h b/util/ng_find_matches.h
index 39ee3f68ded9e78e0dd07a5334efcf1065878d15..9860c202e2ed66aa7465016d87238fb82d555f7e 100644 (file)
--- a/util/ng_find_matches.h
+++ b/util/ng_find_matches.h
@@ -44,11 +44,15 @@ struct BoundaryReports;
 
 } // namespace ue2
 
-/** \brief Find all matches for a given graph when executed against \a input.
+/**
+ * \brief Find all matches for a given graph when executed against \a input.
  *
- *  Fills \a matches with offsets into the data stream where a match is found.
+ * Fills \a matches with offsets into the data stream where a match is found.
+ *
+ * Returns false if this graph is too large to find its matches in reasonable
+ * time.
  */
-void findMatches(const ue2::NGHolder &g, const ue2::ReportManager &rm,
+bool findMatches(const ue2::NGHolder &g, const ue2::ReportManager &rm,
                  const std::string &input,
                  std::set<std::pair<size_t, size_t>> &matches,
                  const unsigned int max_edit_distance, const bool notEod,