Add support for sorting in LRU caches

Performs an in-place merge sort

src/fish_tests.cpp

@@ -1125,11 +1125,30 @@ static void test_escape_sequences(void) {
 
 class test_lru_t : public lru_cache_t<test_lru_t, int> {
    public:
-    test_lru_t() : lru_cache_t<test_lru_t, int>(16) {}
+    static constexpr size_t test_capacity = 16;
+    typedef std::pair<wcstring, int> value_type;
 
-    std::vector<std::pair<wcstring, int>> evicted;
+    test_lru_t() : lru_cache_t<test_lru_t, int>(test_capacity) {}
+
+    std::vector<value_type> evicted;
 
     void entry_was_evicted(wcstring key, int val) { evicted.push_back({key, val}); }
+
+    std::vector<value_type> values() const {
+        std::vector<value_type> result;
+        for (const auto &p : *this) {
+            result.push_back(p);
+        }
+        return result;
+    }
+
+    std::vector<int> ints() const {
+        std::vector<int> result;
+        for (const auto &p : *this) {
+            result.push_back(p.second);
+        }
+        return result;
+    }
 };
 
 static void test_lru(void) {
@@ -1137,15 +1156,48 @@ static void test_lru(void) {
 
     test_lru_t cache;
     std::vector<std::pair<wcstring, int>> expected_evicted;
+    std::vector<std::pair<wcstring, int>> expected_values;
     int total_nodes = 20;
     for (int i = 0; i < total_nodes; i++) {
         do_test(cache.size() == size_t(std::min(i, 16)));
+        do_test(cache.values() == expected_values);
         if (i < 4) expected_evicted.push_back({to_string(i), i});
         // Adding the node the first time should work, and subsequent times should fail.
         do_test(cache.insert(to_string(i), i));
         do_test(!cache.insert(to_string(i), i + 1));
+
+        expected_values.push_back({to_string(i), i});
+        while (expected_values.size() > test_lru_t::test_capacity) {
+            expected_values.erase(expected_values.begin());
+        }
     }
     do_test(cache.evicted == expected_evicted);
+    do_test(cache.values() == expected_values);
+
+    // Stable-sort ints in reverse order
+    // This a/2 check ensures that some different ints compare the same
+    // It also gives us a different order than we started with
+    auto comparer = [](int a, int b){
+        return a/2 > b/2;
+    };
+    std::vector<int> ints = cache.ints();
+    std::stable_sort(ints.begin(), ints.end(), comparer);
+
+    cache.stable_sort(comparer);
+    std::vector<int> new_ints = cache.ints();
+    if (new_ints != ints) {
+        auto commajoin = [](const std::vector<int> &vs) {
+            wcstring ret;
+            for (int v : vs) {
+                append_format(ret, L"%d,", v);
+            }
+            if (! ret.empty()) ret.pop_back();
+            return ret;
+        };
+        err(L"LRU stable sort failed. Expected %ls, got %ls\n", commajoin(new_ints).c_str(), commajoin(ints).c_str());
+    }
+
+
     cache.evict_all_nodes();
     do_test(cache.evicted.size() == size_t(total_nodes));
 }

src/lru.h

@@ -113,7 +113,72 @@ class lru_cache_t {
         USE(value);
     }
 
+    // Implementation of merge step for mergesort
+    // Given two singly linked lists left and right,
+    // and a binary func F implementing less-than, return
+    // the list in sorted order
+    template <typename F>
+    static lru_link_t *merge(lru_link_t *left, size_t left_len,
+                             lru_link_t *right, size_t right_len,
+                             const F &func) {
+        assert(left_len > 0 && right_len > 0);
+
+        auto popleft = [&](){
+            lru_link_t *ret = left;
+            left = left->next;
+            left_len--;
+            return ret;
+        };
+
+        auto popright = [&](){
+            lru_link_t *ret = right;
+            right = right->next;
+            right_len--;
+            return ret;
+        };
+
+        lru_link_t *head;
+        lru_link_t **cursor = &head;
+        while (left_len && right_len) {
+            bool goleft = ! func(static_cast<lru_node_t *>(left)->value,
+                                 static_cast<lru_node_t *>(right)->value);
+            *cursor = goleft ? popleft() : popright();
+            cursor = &(*cursor)->next;
+        }
+        while (left_len || right_len) {
+            *cursor = left_len ? popleft() : popright();
+            cursor = &(*cursor)->next;
+        }
+        return head;
+    }
+
+    // mergesort the given list of the given length
+    // This only sets the next pointers, not the prev ones
+    template<typename F>
+    static lru_link_t *mergesort(lru_link_t *node, size_t length, const F &func) {
+        if (length <= 1) {
+            return node;
+        }
+        // divide us into two lists, left and right
+        const size_t left_len = length / 2;
+        const size_t right_len = length - left_len;
+        lru_link_t *left = node;
+
+        lru_link_t *right = node;
+        for (size_t i=0; i < left_len; i++) {
+            right = right->next;
+        }
+
+        // Recursive sorting
+        left = mergesort(left, left_len, func);
+        right = mergesort(right, right_len, func);
+
+        // Merge them
+        return merge(left, left_len, right, right_len, func);
+    }
+
    public:
+
     // Constructor
     // Note our linked list is always circular!
     explicit lru_cache_t(size_t max_size = 1024) : max_node_count(max_size) {
@@ -159,7 +224,8 @@ class lru_cache_t {
         // Try inserting; return false if it was already in the set.
         auto iter_inserted = this->node_map.emplace(std::move(key), lru_node_t(std::move(value)));
         if (!iter_inserted.second) {
-            // already present
+            // already present - so promote it
+            promote_node(&iter_inserted.first->second);
             return false;
         }
 
@@ -178,6 +244,32 @@ class lru_cache_t {
     // Number of entries
     size_t size() { return this->node_map.size(); }
 
+    // Sorting support
+    // Given a binary function F implementing less-than on the contents, place the nodes in sorted order
+    template<typename F>
+    void stable_sort(const F &func) {
+        // Perform the sort. This sets forward pointers only
+        size_t length = this->size();
+        if (length <= 1) {
+            return;
+        }
+
+        lru_link_t *sorted = mergesort(this->mouth.next, length, func);
+        mouth.next = sorted;
+        // Go through and set back back pointers
+        lru_link_t *cursor = sorted;
+        lru_link_t *prev = &mouth;
+        for (size_t i=0; i < length; i++) {
+            cursor->prev = prev;
+            prev = cursor;
+            cursor = cursor->next;
+        }
+        // prev is now last element in list
+        // make the list circular
+        prev->next = &mouth;
+        mouth.prev = prev;
+    }
+
     void evict_all_nodes(void) {
         while (this->size() > 0) {
             evict_last_node();
@@ -186,12 +278,12 @@ class lru_cache_t {
 
     // Iterator for walking nodes, from least recently used to most.
     class iterator {
-        lru_link_t *node;
+        const lru_link_t *node;
 
        public:
         typedef std::pair<const wcstring &, const CONTENTS &> value_type;
 
-        explicit iterator(lru_link_t *val) : node(val) {}
+        explicit iterator(const lru_link_t *val) : node(val) {}
         void operator++() { node = node->prev; }
         bool operator==(const iterator &other) { return node == other.node; }
         bool operator!=(const iterator &other) { return !(*this == other); }
@@ -202,8 +294,8 @@ class lru_cache_t {
         }
     };
 
-    iterator begin() { return iterator(mouth.prev); }
+    iterator begin() const { return iterator(mouth.prev); };
-    iterator end() { return iterator(&mouth); }
+    iterator end() const { return iterator(&mouth); };
 };
 
 #endif