feat(lsm): iterator over items in trie

lsm/src/_include/lsm/trie_internal.h

@@ -5,14 +5,11 @@
 #include "lsm/str_internal.h"
 #include "lsm/trie.h"
 
-/**
- * A node inside a trie structure
- */
-typedef struct lsm_trie_node {
+struct lsm_trie_node {
   lsm_bt bt;
   lsm_str skip;
   void *data;
-} lsm_trie_node;
+};
 
 /**
  * Allocate and initialize a new trie node
@@ -33,4 +30,13 @@ struct lsm_trie {
   uint64_t size;
 };
 
+struct lsm_trie_iterator {
+  lsm_trie_node *next;
+  struct {
+    lsm_bt_iterator *arr;
+    size_t len;
+    size_t cap;
+  } stack;
+};
+
 #endif

lsm/src/bt/lsm_bt.c

@@ -184,7 +184,7 @@ void lsm_bt_iter(lsm_bt_iterator *out, const lsm_bt *bt) {
   }
 }
 
-bool lsm_bt_iter_next(const void **out, char *key_out, lsm_bt_iterator *iter) {
+bool lsm_bt_iter_next(void **out, char *key_out, lsm_bt_iterator *iter) {
   if (iter->next == NULL) {
     return false;
   }

lsm/src/trie/lsm_trie_iter.c

@@ -0,0 +1,117 @@
+#include "lsm.h"
+#include "lsm/trie_internal.h"
+
+/**
+ * Advance the iterator so that `next` now points to a node containing data. If
+ * no more data nodes are found, `next` is set to NULL. The initial value of
+ * `next` is ignored.
+ */
+lsm_error lsm_trie_iter_next_data_node(lsm_trie_iterator *iter) {
+  do {
+    // Traverse back up the stack until we reach either the end of the subtree,
+    // or an iterator that hasn't been exhausted yet
+    while ((iter->stack.len > 0) &&
+           !lsm_bt_iter_next((void **)&iter->next, NULL,
+                             &iter->stack.arr[iter->stack.len - 1])) {
+      iter->stack.len--;
+    }
+
+    if (iter->stack.len == 0) {
+      iter->next = NULL;
+
+      return lsm_error_ok;
+    }
+
+    // Add node's subtree to the stack
+    if (iter->stack.len == iter->stack.cap) {
+      lsm_bt_iterator *arr = realloc(
+          iter->stack.arr, iter->stack.cap * 2 * sizeof(lsm_bt_iterator));
+
+      if (arr == NULL) {
+        return lsm_error_failed_alloc;
+      }
+
+      iter->stack.arr = arr;
+      iter->stack.cap *= 2;
+    }
+
+    lsm_bt_iter(&iter->stack.arr[iter->stack.len], &iter->next->bt);
+    iter->stack.len++;
+
+  } while (iter->next->data == NULL);
+
+  return lsm_error_ok;
+}
+
+lsm_error lsm_trie_iter(lsm_trie_iterator **out, const lsm_trie *trie,
+                        const lsm_str *prefix) {
+  uint64_t prefix_len = prefix == NULL ? 0 : lsm_str_len(prefix);
+
+  uint64_t index = 0;
+  lsm_trie_node *node = trie->root;
+
+  // Simplified traversal down the trie to find the root of the subtree
+  // contaiing all nodes with the given prefix
+  while ((index < prefix_len) && (node != NULL)) {
+    char c = lsm_str_char(prefix, index);
+    lsm_error res = lsm_bt_search((void **)&node, &node->bt, c);
+
+    switch (res) {
+    case lsm_error_ok:
+      index += 1 + lsm_str_len(&node->skip);
+      break;
+    case lsm_error_not_found:
+      node = NULL;
+      break;
+    default:
+      return res;
+    };
+  }
+
+  lsm_trie_iterator *iter = calloc(1, sizeof(lsm_trie_iterator));
+
+  if (iter == NULL) {
+    return lsm_error_failed_alloc;
+  }
+
+  iter->next = node;
+
+  if (node != NULL) {
+    lsm_bt_iterator *arr = malloc(sizeof(lsm_bt_iterator));
+
+    if (arr == NULL) {
+      free(iter);
+
+      return lsm_error_failed_alloc;
+    }
+
+    iter->stack.arr = arr;
+    iter->stack.cap = 1;
+    iter->stack.len = 1;
+
+    lsm_bt_iter(&iter->stack.arr[0], &node->bt);
+
+    if (node->data == NULL) {
+      LSM_RES(lsm_trie_iter_next_data_node(iter));
+    }
+  }
+
+  *out = iter;
+
+  return lsm_error_ok;
+}
+
+lsm_error lsm_trie_iter_next(void **out, lsm_trie_iterator *iter) {
+  if (iter->next == NULL) {
+    return lsm_error_done;
+  }
+
+  *out = iter->next->data;
+
+  return lsm_trie_iter_next_data_node(iter);
+}
+
+void lsm_trie_iter_free(lsm_trie_iterator *iter) {
+  free(iter->stack.arr);
+  free(iter);
+}