docs: state merge_patch complexity and RFC 7396 limitations; test non-object, discarded, binary cases

Signed-off-by: Niels Lohmann <mail@nlohmann.me>

docs/mkdocs/docs/api/basic_json/merge_patch.md

@@ -39,13 +39,33 @@ Basic guarantee: if an exception is thrown during the operation, the JSON value
 
 ## Complexity
 
-Linear in the lengths of `apply_patch`.
+Linear in the total number of values (including nested ones) in `apply_patch`, times the cost of a key-based object
+operation (looking up, adding, or erasing a member): logarithmic for [`json`](../json.md), and linear for
+[`ordered_json`](../ordered_json.md) (see [`ordered_map` complexity](../ordered_map.md#complexity)). For
+`ordered_json`, patching an object that has `n` members with a patch that has `m` members therefore takes O(n·m). The
+cost also includes one copy of `apply_patch` (see the notes below).
 
 ## Notes
 
-`apply_patch` may be `#!cpp *this` itself or refer to a value contained in `#!cpp *this` (for example, a subobject
-returned by `#!cpp (*this)[key]`); it is read as it was when `merge_patch()` was called, before any modification of
-`#!cpp *this`.
+- `apply_patch` may be `#!cpp *this` itself or refer to a value contained in `#!cpp *this` (for example, a subobject
+  returned by `#!cpp (*this)[key]`); it is read as it was when `merge_patch()` was called, before any modification of
+  `#!cpp *this`. To make this possible, the function copies `apply_patch` before it modifies `#!cpp *this`, so the cost
+  includes one copy of the patch.
+- A merge patch uses `#!json null` to remove a member, so it cannot set a member to `#!json null`: applying
+  `#!json {"a":null}` to `#!json {"a":1}` gives `#!json {}`.
+- Arrays are not merged: an array in the patch replaces the target value as a whole, and a `#!json null` inside an
+  array is kept as an element. `#!json null` members of the target that the patch does not mention are preserved.
+- A `#!json null` in the patch only removes a member; it does not prevent the creation of the enclosing objects.
+  Applying `#!json {"a":{"b":null}}` to `#!json {}` gives `#!json {"a":{}}` (see example 15 in Appendix A of the RFC).
+- If `apply_patch` is not an object, it replaces the target as a whole. In particular, a `#!json null` patch turns the
+  target into `#!json null`. If `apply_patch` is an object and the target is not, the target is first replaced by an
+  empty object, so applying `#!json {}` to `#!json 1` gives `#!json {}`.
+- A merge patch cannot express every change; for example, it cannot reorder array elements or remove a specific array
+  element. Use [JSON Patch](../../features/json_patch.md) ([RFC 6902](https://tools.ietf.org/html/rfc6902)) for that.
+  [`merge_diff`](merge_diff.md) has a matching limitation, see its notes.
+- A [discarded](is_discarded.md) value is treated like any other non-object value: a discarded
+  `apply_patch` turns the target into a discarded value, and an object patch replaces a discarded target by an object
+  as described above.
 
 ## Examples
 

docs/mkdocs/docs/features/merge_patch.md

@@ -2,7 +2,7 @@
 
 The library supports JSON Merge Patch ([RFC 7396](https://tools.ietf.org/html/rfc7396)) as a patch format.
 The merge patch format is primarily intended for use with the HTTP PATCH method as a means of describing a set of
-modifications to a target resource's content. This function applies a merge patch to the current JSON value.
+modifications to a target resource's content. Applying a merge patch changes the JSON value the patch is applied to.
 
 Instead of using [JSON Pointer](json_pointer.md) to specify values to be manipulated, it describes the changes using a
 syntax that closely mimics the document being modified. Unlike [JSON Patch](json_patch.md), a JSON Merge Patch cannot
@@ -19,6 +19,18 @@ another.
     [`update`](../api/basic_json/update.md) with `merge_objects` set to `#!cpp true`; see
     [Merging objects](modifying_values.md#merging-objects).
 
+!!! note "Limitations of RFC 7396"
+
+    - A patch cannot set a member to `#!json null`, because `#!json null` removes the member:
+      `#!json {"a":1}` patched with `#!json {"a":null}` gives `#!json {}`.
+    - Arrays are replaced as a whole, never merged, and a `#!json null` inside an array is kept.
+    - `#!json null` members of the target that the patch does not mention are preserved.
+    - A `#!json null` that would remove a member of a not yet existing object still creates that object:
+      `#!json {}` patched with `#!json {"a":{"b":null}}` gives `#!json {"a":{}}`.
+    - A patch that is not an object replaces the whole target (a `#!json null` patch makes the target
+      `#!json null`); an object patch applied to a non-object target starts from an empty object.
+    - [`merge_diff`](../api/basic_json/merge_diff.md#notes) has a matching limitation for `#!json null` members.
+
 ??? example
 
     The following code shows how a JSON Merge Patch is applied to a JSON document.

tests/src/unit-merge_patch.cpp

@@ -423,3 +423,86 @@ TEST_CASE("merge_patch() with an argument that aliases *this (#5641)")
         }
     }
 }
+
+TEST_CASE("JSON Merge Patch on non-object, discarded, and binary values")
+{
+    SECTION("an empty object patch turns a non-object target into an empty object")
+    {
+        json target = 1;
+        target.merge_patch(json::object());
+        CHECK(target == json::object());
+
+        json from_null;
+        from_null.merge_patch(json::object());
+        CHECK(from_null == json::object());
+    }
+
+    SECTION("a non-object patch replaces a deeply nested target")
+    {
+        // checked by type, since == recurses
+        SECTION("number")
+        {
+            json target = json::parse(nested_objects(1000, 0));
+            target.merge_patch(json(42));
+            CHECK(target.is_number_integer());
+            CHECK(target.get<int>() == 42);
+        }
+
+        SECTION("array")
+        {
+            json target = json::parse(nested_objects(1000, 0));
+            target.merge_patch(json({1, nullptr}));
+            CHECK(target.is_array());
+            CHECK(target.size() == 2);
+            CHECK(target[1].is_null());
+        }
+
+        SECTION("null")
+        {
+            json target = json::parse(nested_objects(1000, 0));
+            target.merge_patch(json(nullptr));
+            CHECK(target.is_null());
+        }
+    }
+
+    SECTION("discarded values")
+    {
+        const json discarded(json::value_t::discarded);
+
+        // a discarded patch is not an object and replaces the target
+        json target = {{"a", 1}};
+        target.merge_patch(discarded);
+        CHECK(target.is_discarded());
+
+        // a discarded target is replaced by an empty object first
+        json discarded_target(json::value_t::discarded);
+        discarded_target.merge_patch(json({{"a", 1}}));
+        CHECK(discarded_target == json({{"a", 1}}));
+
+        json both(json::value_t::discarded);
+        both.merge_patch(discarded);
+        CHECK(both.is_discarded());
+    }
+
+    SECTION("binary values")
+    {
+        const json bin = json::binary({1, 2, 3}, 7);
+
+        // a binary patch is not an object and replaces the target
+        json target = {{"a", 1}};
+        target.merge_patch(bin);
+        CHECK(target.is_binary());
+        CHECK(target.get_binary() == bin.get_binary());
+
+        // a binary target is replaced by an object
+        json binary_target = bin;
+        binary_target.merge_patch(json({{"a", 1}}));
+        CHECK(binary_target == json({{"a", 1}}));
+
+        // binary values nested in the patch are kept
+        json nested = json::object();
+        nested.merge_patch(json({{"a", bin}}));
+        CHECK(nested["a"].is_binary());
+        CHECK(nested["a"].get_binary() == bin.get_binary());
+    }
+}