From 1ab6c90257a18992957466728b6b29b025793628 Mon Sep 17 00:00:00 2001
From: Greg Burd <greg@burd.me>
Date: Mon, 29 Apr 2024 17:01:45 -0400
Subject: [PATCH] WIP

---
 src/sparsemap.c | 66 +++++++++++++++++++++++++++----------------------
 tests/test.c    | 18 ++++++++++++++
 2 files changed, 54 insertions(+), 30 deletions(-)

diff --git a/src/sparsemap.c b/src/sparsemap.c
index 726a047..874c7cf 100644
--- a/src/sparsemap.c
+++ b/src/sparsemap.c
@@ -52,7 +52,7 @@ void __attribute__((format(printf, 4, 5))) __sm_diag_(const char *file, int line
 
 #define __sm_assert(expr) \
   if (!(expr))            \
-  fprintf(stderr, "%s:%d:%s(): assertion failed! %s", __FILE__, __LINE__, __func__, #expr)
+  fprintf(stderr, "%s:%d:%s(): assertion failed! %s\n", __FILE__, __LINE__, __func__, #expr)
 
 #define __sm_when_diag(expr) \
   if (1)                     \
@@ -1260,45 +1260,51 @@ __sm_chunk_map_merge(sparsemap_t *map, sparsemap_idx_t offset, __sm_chunk_t src)
 void
 sparsemap_merge(sparsemap_t *map, sparsemap_t *other)
 {
-  uint8_t *dst = __sm_get_chunk_map_data(map, 0),
-          *dst_tail = __sm_get_chunk_map_end(map),
-          *src = __sm_get_chunk_map_data(other, 0);
+  uint8_t *src, *dst, *dst_tail;
+  size_t src_count = __sm_get_chunk_map_count(other),
+         dst_count = __sm_get_chunk_map_count(map);
 
-  size_t i, src_count = __sm_get_chunk_map_count(other), dst_count = __sm_get_chunk_map_count(map);
-  for (i = 0; i < src_count + dst_count;) {
-    __sm_chunk_t src_chunk, dst_chunk;
-    sm_idx_t src_start = *(sm_idx_t *)src;
+  dst = __sm_get_chunk_map_data(map, 0);
+  src = __sm_get_chunk_map_data(other, 0);
+  /* Foreach destination chunk, is there a source chunk with the same starting
+     offset? If so, if so let's merge them. */
+  for (size_t i = 0; i < dst_count + src_count; i++) {
+    __sm_chunk_t dst_chunk;
     sm_idx_t dst_start = *(sm_idx_t *)dst;
-    __sm_chunk_map_init(&src_chunk, src + sizeof(sm_idx_t));
     __sm_chunk_map_init(&dst_chunk, dst + sizeof(sm_idx_t));
-
-    if (src_start == dst_start) {
-      /* Chunks overlap, merge them. */
-      __sm_chunk_map_merge(map, src_start, src_chunk);
-      src = SM_NEXT_CHUNK_ADDR(src);
-      dst = SM_NEXT_CHUNK_ADDR(dst);
-      i += 2;
-    } else if (src_start < dst_start || i > dst_count) {
-      /* Copy the src chunk into the dest map. */
-      size_t s = __sm_chunk_map_get_size(&src_chunk);
-
-      /* Copy the index (sm_idx_t) */
-      *(sm_idx_t *)dst_tail = src_start;
-      dst_tail += sizeof(sm_idx_t);
+    size_t j = 0;
+    if (i < dst_count) {
+      while (j < src_count) {
+        __sm_chunk_t src_chunk;
+        sm_idx_t src_start = *(sm_idx_t *)src;
+        __sm_chunk_map_init(&src_chunk, src + sizeof(sm_idx_t));
+        if (src_start == dst_start) {
+          /* Chunks overlap, merge them. */
+          __sm_chunk_map_merge(map, src_start, src_chunk);
+          dst = SM_NEXT_CHUNK_ADDR(dst);
+          src = SM_NEXT_CHUNK_ADDR(src);
+          j = SIZE_MAX;
+          break;
+        }
+        dst = SM_NEXT_CHUNK_ADDR(dst);
+        j++;
+      }
+    }
+    if (j != SIZE_MAX) {
+      __sm_chunk_t src_chunk;
+      sm_idx_t src_start = *(sm_idx_t *)src;
+      __sm_chunk_map_init(&src_chunk, src + sizeof(sm_idx_t));
 
       /* Copy the chunk data to the buffer. */
-      memcpy(dst_tail, src, s);
+      dst_tail = __sm_get_chunk_map_end(map);
+      *(sm_idx_t *)dst_tail = src_start;
+      size_t size = __sm_chunk_map_get_size(&src_chunk);
+      memcpy(dst_tail + sizeof(sm_idx_t), src + sizeof(sm_idx_t), size);
 
       /* Update the chunk count and data_used. */
       map->m_data_used = 0;
       __sm_set_chunk_map_count(map, __sm_get_chunk_map_count(map) + 1);
-
       src = SM_NEXT_CHUNK_ADDR(src);
-      i += 1;
-      if (i > dst_count) {
-        dst = SM_NEXT_CHUNK_ADDR(dst);
-        i += 1;
-      }
     }
   }
 }
diff --git a/tests/test.c b/tests/test.c
index ac226da..6494d31 100644
--- a/tests/test.c
+++ b/tests/test.c
@@ -631,11 +631,29 @@ test_api_merge(const MunitParameter params[], void *data)
   assert_ptr_not_null(map);
   assert_ptr_not_null(other);
 
+  sparsemap_set(other, 0, true);
+  sparsemap_merge(map, other);
+
+  assert_true(sparsemap_is_set(other, 0));
+  assert_true(sparsemap_is_set(map, 0));
+
+  sparsemap_clear(map);
+  sparsemap_clear(other);
+
+  sparsemap_set(other, 2049, true);
+  sparsemap_merge(map, other);
+
+  assert_true(sparsemap_is_set(map, 2049));
+
+  sparsemap_clear(map);
+  sparsemap_clear(other);
+
   sparsemap_set(map, 0, true);
   sparsemap_set(other, 1, true);
   sparsemap_set(other, 2049, true);
   sparsemap_set(other, 8193, true);
   sparsemap_set(other, 8194, true);
+
   sparsemap_merge(map,other);
 
   assert_true(sparsemap_is_set(map, 0));