From b25beb13773072c3b143842b45a7c32a1108f347 Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Mon, 15 Apr 2024 14:00:27 -0400
Subject: Updated FSTrie to use const char * instead of std::string

Note: this requires the caller to manage the memory of the strings
---
 include/framework/interface/Record.h | 17 +++++++++++++++++
 include/shard/FSTrie.h               | 18 +++++-------------
 2 files changed, 22 insertions(+), 13 deletions(-)

(limited to 'include')

diff --git a/include/framework/interface/Record.h b/include/framework/interface/Record.h
index d380f9b..19ccadd 100644
--- a/include/framework/interface/Record.h
+++ b/include/framework/interface/Record.h
@@ -132,6 +132,23 @@ struct Record {
     }
 };
 
+template<typename V>
+struct Record<const char*, V> {
+    const char* key;
+    V value;
+    size_t len;
+
+    inline bool operator<(const Record& other) const {
+        size_t n = std::min(len, other.len) + 1;
+        return strncmp(key, other.key, n) < 0;
+    }
+
+    inline bool operator==(const Record& other) const {
+        size_t n = std::min(len, other.len) + 1;
+        return strncmp(key, other.key, n) == 0;
+    }
+};
+
 template <typename K, typename V, typename W>
 struct WeightedRecord {
     K key;
diff --git a/include/shard/FSTrie.h b/include/shard/FSTrie.h
index 95f396f..be678ff 100644
--- a/include/shard/FSTrie.h
+++ b/include/shard/FSTrie.h
@@ -30,7 +30,7 @@ private:
 
     typedef decltype(R::key) K;
     typedef decltype(R::value) V;
-    static_assert(std::is_same_v<K, std::string>, "FST requires std::string keys.");
+    static_assert(std::is_same_v<K, const char*>, "FST requires const char* keys.");
 
 public:
     FSTrie(BufferView<R> buffer)
@@ -42,7 +42,7 @@ public:
         m_alloc_size = sizeof(Wrapped<R>) * buffer.get_record_count();
 
         size_t cnt = 0;
-        std::vector<K> keys;
+        std::vector<std::string> keys;
         keys.reserve(buffer.get_record_count());
 
         /*
@@ -68,14 +68,10 @@ public:
             m_data[cnt] = temp_buffer[i];
             m_data[cnt].clear_timestamp();
 
-            keys.push_back(m_data[cnt].rec.key);
+            keys.push_back(std::string(m_data[cnt].rec.key));
             cnt++;
         }
 
-        for (size_t i=0; i<keys.size() - 1; i++) {
-            assert(keys[i] <= keys[i+1]);
-        }
-
         m_reccnt = cnt;
         if (m_reccnt > 0) {
             m_fst = new fst::Trie(keys);
@@ -96,7 +92,7 @@ public:
         m_data = new Wrapped<R>[attemp_reccnt]();
         m_alloc_size = attemp_reccnt * sizeof(Wrapped<R>);
 
-        std::vector<K> keys;
+        std::vector<std::string> keys;
         keys.reserve(attemp_reccnt);
 
         // FIXME: For smaller cursor arrays, it may be more efficient to skip
@@ -128,7 +124,7 @@ public:
                 /* skip over records that have been deleted via tagging */
                 if (!cursor.ptr->is_deleted() && cursor.ptr->rec.key != "") {
                     m_data[m_reccnt] = *cursor.ptr;
-                    keys.push_back(m_data[m_reccnt].rec.key);
+                    keys.push_back(std::string(m_data[m_reccnt].rec.key));
 
                     m_reccnt++;
                 }
@@ -138,10 +134,6 @@ public:
             }
         }
 
-        for (size_t i=0; i<keys.size() - 1; i++) {
-            assert(keys[i] <= keys[i+1]);
-        }
-
         if (m_reccnt > 0) {
             m_fst = new fst::Trie(keys);
         }
-- 
cgit v1.2.3