efficient initialization of vectors-of-vectors-of-vectors

src/script_opt/CPP/InitsInfo.cc

@@ -7,6 +7,7 @@
 #include "zeek/ZeekString.h"
 #include "zeek/script_opt/CPP/Attrs.h"
 #include "zeek/script_opt/CPP/Compile.h"
+#include "zeek/script_opt/CPP/RuntimeInits.h"
 
 using namespace std;
 
@@ -85,25 +86,25 @@ void CPP_InitsInfo::BuildOffsetSet(CPPCompile* c) {
     offset_set = c->IndMgr().AddIndices(offsets_vec);
 }
 
-void CPP_InitsInfo::BuildCohort(CPPCompile* c, std::vector<std::shared_ptr<CPP_InitInfo>>& cohort) {
-    int n = 0;
-
-    for ( auto& co : cohort ) {
-        vector<string> ivs;
-        auto o = co->InitObj();
-        if ( o ) {
+static std::string describe_initializer(const Obj* o) {
     auto od = obj_desc(o);
 
     // Escape any embedded comment characters.
     od = regex_replace(od, std::regex("/\\*"), "<<SLASH-STAR>>");
     od = regex_replace(od, std::regex("\\*/"), "<<STAR-SLASH>>");
 
-            c->Emit("/* #%s: Initializing %s: */", Fmt(co->Offset()), od);
+    return od;
 }
 
+void CPP_InitsInfo::BuildCohort(CPPCompile* c, std::vector<std::shared_ptr<CPP_InitInfo>>& cohort) {
+    for ( auto& co : cohort ) {
+        vector<string> ivs;
+        auto o = co->InitObj();
+        if ( o )
+            c->Emit("/* #%s: Initializing %s: */", Fmt(co->Offset()), describe_initializer(o));
+
         co->InitializerVals(ivs);
         BuildCohortElement(c, co->InitializerType(), ivs);
-        ++n;
     }
 }
 
@@ -123,6 +124,7 @@ void CPP_InitsInfo::BuildCohortElement(CPPCompile* c, string init_type, vector<s
 }
 
 void CPP_CompoundInitsInfo::GenerateInitializers(CPPCompile* c) {
+    c->Emit("");
     c->Emit("static int %s_init[] = {", tag);
     int n = 0;
 
@@ -132,12 +134,25 @@ void CPP_CompoundInitsInfo::GenerateInitializers(CPPCompile* c) {
         if ( ++n > 1 )
             c->Emit("");
 
-        c->Emit(Fmt(int(cohort.size())) + ",");
-        BuildCohort(c, cohort);
-        c->Emit("-1,");
+        // Figure out the size of the cohort.
+        for ( auto& co : cohort ) {
+            auto o = co->InitObj();
+            if ( o )
+                c->Emit("/* #%s: Initializing %s: */", Fmt(co->Offset()), describe_initializer(o));
+
+            vector<string> ivs;
+            co->InitializerVals(ivs);
+            c->Emit(Fmt(int(ivs.size())) + ",");
+            BuildCohortElement(c, co->InitializerType(), ivs);
         }
 
-    c->Emit("-2,");
+        static const auto end_of_vv = Fmt(END_OF_VEC_VEC) + ",";
+        c->Emit(end_of_vv);
+    }
+
+    static const auto end_of_vvv = Fmt(END_OF_VEC_VEC_VEC) + ",";
+    c->Emit(end_of_vvv);
+
     c->IndentDown();
     c->Emit("};");
 
@@ -149,9 +164,9 @@ void CPP_CompoundInitsInfo::GenerateCohorts(CPPCompile* c) { c->Emit("%s_init",
 void CPP_CompoundInitsInfo::BuildCohortElement(CPPCompile* c, string init_type, vector<string>& ivs) {
     string init_line;
     for ( auto& iv : ivs )
-        init_line += iv;
+        init_line += iv + ",";
 
-    c->Emit("%s,", init_line);
+    c->Emit("%s", init_line);
 }
 
 void CPP_BasicConstInitsInfo::BuildCohortElement(CPPCompile* c, string init_type, vector<string>& ivs) {
@@ -581,7 +596,8 @@ void IndicesManager::Generate(CPPCompile* c) {
             c->Emit(line);
     }
 
-    c->Emit("-1");
+    static const auto end_of_vv = Fmt(END_OF_VEC_VEC);
+    c->Emit(end_of_vv);
     c->EndBlock(true);
 }
 

src/script_opt/CPP/RuntimeInits.cc

@@ -470,7 +470,7 @@ size_t generate_indices_set(int* inits, std::vector<std::vector<int>>& indices_s
     // can pre-allocate the outer vector.
     auto i_ptr = inits;
     int num_inits = 0;
-    while ( *i_ptr >= 0 ) {
+    while ( *i_ptr != END_OF_VEC_VEC && *i_ptr != END_OF_VEC_VEC_VEC ) {
         ++num_inits;
         int n = *i_ptr;
         i_ptr += n + 1; // skip over vector elements
@@ -479,7 +479,7 @@ size_t generate_indices_set(int* inits, std::vector<std::vector<int>>& indices_s
     indices_set.reserve(num_inits);
 
     i_ptr = inits;
-    while ( *i_ptr >= 0 ) {
+    while ( *i_ptr != END_OF_VEC_VEC ) {
         int n = *i_ptr;
         ++i_ptr;
         std::vector<int> indices;
@@ -495,19 +495,13 @@ size_t generate_indices_set(int* inits, std::vector<std::vector<int>>& indices_s
 }
 
 std::vector<std::vector<std::vector<int>>> generate_indices_set(int* inits) {
-    // Figure out how many vector-of-vectors there are.
-    int num_vv = 0;
-    for ( int i = 0; inits[i] >= -1; ++i )
-        if ( inits[i] == -1 )
-            ++num_vv;
+    std::vector<std::vector<std::vector<int>>> indices_set;
 
-    std::vector<std::vector<std::vector<int>>> indices_set(num_vv);
-
-    auto i_ptr = inits;
-    for ( int i = 0; i < num_vv; ++i )
-        i_ptr += generate_indices_set(i_ptr, indices_set[i]);
-
-    ASSERT(*i_ptr == -2);
+    while ( *inits != END_OF_VEC_VEC_VEC ) {
+        std::vector<std::vector<int>> cohort_inits;
+        inits += generate_indices_set(inits, cohort_inits);
+        indices_set.push_back(std::move(cohort_inits));
+    }
 
     return indices_set;
 }

src/script_opt/CPP/RuntimeInits.h

@@ -23,19 +23,24 @@ class InitsManager;
 // constructs the corresponding vector-of-vector-of-indices.  Each
 // vector-of-indices is represented first by an int specifying its
 // size, and then that many int's for its values.  We recognize the
-// end of the array upon encountering a "size" entry of -1.
+// end of the array upon encountering a "size" entry of END_OF_VEC_VEC.
 //
 // Returns how many elements were processed out of "inits", including its
 // terminator.
 extern size_t generate_indices_set(int* inits, std::vector<std::vector<int>>& indices_set);
 
 // The same but for one more level of vector construction. The source array
-// has sub-arrays terminated with -1 per the above, and the whole shebang
-// is terminated with -2.
+// has sub-arrays terminated with END_OF_VEC_VEC per the above, and the whole
+// shebang is terminated with END_OF_VEC_VEC_VEC.
 //
 // Returns the vector construction.
 extern std::vector<std::vector<std::vector<int>>> generate_indices_set(int* inits);
 
+// These need to be distinct from any values that can appear, which means
+// they should be negative, and not -1, which is used as a "N/A" value.
+#define END_OF_VEC_VEC -100
+#define END_OF_VEC_VEC_VEC -200
+
 // An abstract helper class used to access elements of an initialization vector.
 // We need the abstraction because InitsManager below needs to be able to refer
 // to any of a range of templated classes.