Preserve node metadata when duplicating

ext/markly/cmark-gfm-extension_api.h

@@ -264,6 +264,10 @@ typedef void (*cmark_opaque_free_func) (cmark_syntax_extension *extension,
                                         cmark_mem *mem,
                                         cmark_node *node);
 
+typedef void (*cmark_opaque_copy_func)(cmark_syntax_extension *extension,
+                                       cmark_mem *mem, cmark_node *node,
+                                       cmark_node *source);
+
 /** Free a cmark_syntax_extension.
  */
 CMARK_GFM_EXPORT
@@ -406,6 +410,12 @@ CMARK_GFM_EXPORT
 void cmark_syntax_extension_set_opaque_free_func(cmark_syntax_extension *extension,
                                                  cmark_opaque_free_func func);
 
+/** See the documentation for 'cmark_syntax_extension'
+ */
+CMARK_GFM_EXPORT
+void cmark_syntax_extension_set_opaque_copy_func(
+    cmark_syntax_extension *extension, cmark_opaque_copy_func func);
+
 /** See the documentation for 'cmark_syntax_extension'
  */
 CMARK_GFM_EXPORT

ext/markly/cmark-gfm.h

@@ -196,6 +196,12 @@ CMARK_GFM_EXPORT cmark_node *cmark_node_new_with_mem_and_ext(cmark_node_type typ
                                                 cmark_mem *mem,
                                                 cmark_syntax_extension *extension);
 
+/** Creates an independent deep copy of 'node' and all its children.
+ * User data is not copied. Returns NULL when extension-specific node data
+ * cannot be copied.
+ */
+CMARK_GFM_EXPORT cmark_node *cmark_node_clone(cmark_node *node);
+
 /** Frees the memory allocated for a node and any children.
  */
 CMARK_GFM_EXPORT void cmark_node_free(cmark_node *node);

ext/markly/extensions/table.c

@@ -842,6 +842,29 @@ static void opaque_free(cmark_syntax_extension *self, cmark_mem *mem, cmark_node
   }
 }
 
+static void opaque_copy(cmark_syntax_extension *self, cmark_mem *mem,
+                        cmark_node *node, cmark_node *source) {
+  if (node->type == CMARK_NODE_TABLE) {
+    node_table *table = (node_table *)node->as.opaque;
+    node_table *source_table = (node_table *)source->as.opaque;
+
+    table->n_columns = source_table->n_columns;
+    table->n_rows = source_table->n_rows;
+    table->n_nonempty_cells = source_table->n_nonempty_cells;
+
+    if (source_table->alignments) {
+      table->alignments = mem->calloc(source_table->n_columns, sizeof(uint8_t));
+      memcpy(table->alignments, source_table->alignments,
+             source_table->n_columns * sizeof(uint8_t));
+    }
+  } else if (node->type == CMARK_NODE_TABLE_ROW) {
+    *(node_table_row *)node->as.opaque = *(node_table_row *)source->as.opaque;
+  } else if (node->type == CMARK_NODE_TABLE_CELL) {
+    mem->free(node->as.opaque);
+    node->as.cell_index = source->as.cell_index;
+  }
+}
+
 static int escape(cmark_syntax_extension *self, cmark_node *node, int c) {
   return
     node->type != CMARK_NODE_TABLE &&
@@ -867,6 +890,7 @@ cmark_syntax_extension *create_table_extension(void) {
   cmark_syntax_extension_set_html_render_func(self, html_render);
   cmark_syntax_extension_set_opaque_alloc_func(self, opaque_alloc);
   cmark_syntax_extension_set_opaque_free_func(self, opaque_free);
+  cmark_syntax_extension_set_opaque_copy_func(self, opaque_copy);
   cmark_syntax_extension_set_commonmark_escape_func(self, escape);
   CMARK_NODE_TABLE = cmark_syntax_extension_add_node(0);
   CMARK_NODE_TABLE_ROW = cmark_syntax_extension_add_node(0);

ext/markly/html.c

@@ -303,7 +303,7 @@ static int S_render_node(cmark_html_renderer *renderer, cmark_node *node,
 
   case CMARK_NODE_PARAGRAPH:
     parent = cmark_node_parent(node);
-    grandparent = cmark_node_parent(parent);
+    grandparent = parent ? cmark_node_parent(parent) : NULL;
     if (grandparent != NULL && grandparent->type == CMARK_NODE_LIST) {
       tight = grandparent->as.list.tight;
     } else {
@@ -316,7 +316,8 @@ static int S_render_node(cmark_html_renderer *renderer, cmark_node *node,
         cmark_html_render_sourcepos(node, html, options);
         cmark_strbuf_putc(html, '>');
       } else {
-        if (parent->type == CMARK_NODE_FOOTNOTE_DEFINITION && node->next == NULL) {
+        if (parent && parent->type == CMARK_NODE_FOOTNOTE_DEFINITION &&
+            node->next == NULL) {
           cmark_strbuf_putc(html, ' ');
           S_put_footnote_backref(renderer, html, parent);
         }

ext/markly/markly.c

@@ -271,6 +271,21 @@ static VALUE Markly_Node_new(VALUE self, VALUE type) {
 	return Markly_Node_wrap(node);
 }
 
+/*
+ * Duplicate the current node and all its children.
+ */
+static VALUE Markly_Node_duplicate(VALUE self) {
+	cmark_node *node;
+	TypedData_Get_Struct(self, cmark_node, &Markly_Node_Type, node);
+
+	cmark_node *copy = cmark_node_clone(node);
+	if (copy == NULL) {
+		rb_raise(Markly_Error, "could not duplicate node");
+	}
+
+	return Markly_Node_wrap(copy);
+}
+
 static VALUE Markly_Node_replace(VALUE self, VALUE other) {
 	cmark_node *current_node = NULL, *replacement_node = NULL;
 
@@ -1273,6 +1288,7 @@ static void Init_Markly_Node(VALUE Markly) {
 	rb_undef_alloc_func(Markly_Node);
 	rb_define_singleton_method(Markly_Node, "new", Markly_Node_new, 1);
 	Markly_Node_Fence = rb_struct_define_under(Markly_Node, "Fence", "character", "length", "indent", NULL);
+	rb_define_method(Markly_Node, "_dup", Markly_Node_duplicate, 0);
 
 	rb_define_method(Markly_Node, "replace", Markly_Node_replace, 1);
 

ext/markly/node.c

@@ -145,6 +145,144 @@ cmark_node *cmark_node_new(cmark_node_type type) {
   return cmark_node_new_with_ext(type, NULL);
 }
 
+static cmark_chunk S_clone_chunk(cmark_mem *mem, const cmark_chunk *source) {
+  cmark_chunk clone = {NULL, source->len, 1};
+  clone.data = (unsigned char *)mem->calloc(source->len + 1, 1);
+  if (source->len > 0) {
+    memcpy(clone.data, source->data, source->len);
+  }
+  return clone;
+}
+
+static cmark_node *S_clone_node(cmark_node *node) {
+  cmark_mem *mem = NODE_MEM(node);
+  cmark_node *clone = cmark_node_new_with_mem_and_ext(
+      (cmark_node_type)node->type, mem, node->extension);
+
+  if (!clone) {
+    return NULL;
+  }
+
+  cmark_strbuf_set(&clone->content, node->content.ptr, node->content.size);
+  clone->start_line = node->start_line;
+  clone->start_column = node->start_column;
+  clone->end_line = node->end_line;
+  clone->end_column = node->end_column;
+  clone->internal_offset = node->internal_offset;
+  clone->flags = node->flags;
+  clone->footnote = node->footnote;
+
+  switch (node->type) {
+  case CMARK_NODE_HEADING:
+    clone->as.heading = node->as.heading;
+    break;
+  case CMARK_NODE_LIST:
+  case CMARK_NODE_ITEM:
+    clone->as.list = node->as.list;
+    break;
+  case CMARK_NODE_CODE_BLOCK:
+  case CMARK_NODE_FRONT_MATTER:
+  case CMARK_NODE_CODE:
+    clone->as.code = node->as.code;
+    clone->as.code.info = S_clone_chunk(mem, &node->as.code.info);
+    clone->as.code.literal = S_clone_chunk(mem, &node->as.code.literal);
+    break;
+  case CMARK_NODE_TEXT:
+  case CMARK_NODE_HTML_INLINE:
+  case CMARK_NODE_HTML_BLOCK:
+  case CMARK_NODE_FOOTNOTE_REFERENCE:
+  case CMARK_NODE_FOOTNOTE_DEFINITION:
+    clone->as.literal = S_clone_chunk(mem, &node->as.literal);
+    break;
+  case CMARK_NODE_LINK:
+  case CMARK_NODE_IMAGE:
+    clone->as.link.url = S_clone_chunk(mem, &node->as.link.url);
+    clone->as.link.title = S_clone_chunk(mem, &node->as.link.title);
+    break;
+  case CMARK_NODE_CUSTOM_BLOCK:
+  case CMARK_NODE_CUSTOM_INLINE:
+    clone->as.custom.on_enter = S_clone_chunk(mem, &node->as.custom.on_enter);
+    clone->as.custom.on_exit = S_clone_chunk(mem, &node->as.custom.on_exit);
+    break;
+  default:
+    break;
+  }
+
+  if (node->extension && node->extension->opaque_alloc_func) {
+    if (!node->extension->opaque_copy_func) {
+      cmark_node_free(clone);
+      return NULL;
+    }
+
+    node->extension->opaque_copy_func(node->extension, mem, clone, node);
+  }
+
+  for (cmark_node *child = node->first_child; child; child = child->next) {
+    cmark_node *child_clone = S_clone_node(child);
+    if (!child_clone) {
+      cmark_node_free(clone);
+      return NULL;
+    }
+
+    if (!cmark_node_append_child(clone, child_clone)) {
+      cmark_node_free(child_clone);
+      cmark_node_free(clone);
+      return NULL;
+    }
+  }
+
+  return clone;
+}
+
+static cmark_node *S_find_clone(cmark_node *source, cmark_node *clone,
+                                cmark_node *target) {
+  if (source == target) {
+    return clone;
+  }
+
+  cmark_node *source_child = source->first_child;
+  cmark_node *clone_child = clone->first_child;
+  while (source_child && clone_child) {
+    cmark_node *result = S_find_clone(source_child, clone_child, target);
+    if (result) {
+      return result;
+    }
+    source_child = source_child->next;
+    clone_child = clone_child->next;
+  }
+
+  return NULL;
+}
+
+static void S_clone_footnote_links(cmark_node *source, cmark_node *clone,
+                                   cmark_node *source_root,
+                                   cmark_node *clone_root) {
+  if (source->parent_footnote_def) {
+    clone->parent_footnote_def =
+        S_find_clone(source_root, clone_root, source->parent_footnote_def);
+  }
+
+  cmark_node *source_child = source->first_child;
+  cmark_node *clone_child = clone->first_child;
+  while (source_child && clone_child) {
+    S_clone_footnote_links(source_child, clone_child, source_root, clone_root);
+    source_child = source_child->next;
+    clone_child = clone_child->next;
+  }
+}
+
+cmark_node *cmark_node_clone(cmark_node *node) {
+  if (!node) {
+    return NULL;
+  }
+
+  cmark_node *clone = S_clone_node(node);
+  if (clone) {
+    S_clone_footnote_links(node, clone, node, clone);
+  }
+  return clone;
+}
+
 static void free_node_as(cmark_node *node) {
   switch (node->type) {
     case CMARK_NODE_CODE_BLOCK:

ext/markly/syntax_extension.c

@@ -143,6 +143,11 @@ void cmark_syntax_extension_set_opaque_free_func(cmark_syntax_extension *extensi
   extension->opaque_free_func = func;
 }
 
+void cmark_syntax_extension_set_opaque_copy_func(
+    cmark_syntax_extension *extension, cmark_opaque_copy_func func) {
+  extension->opaque_copy_func = func;
+}
+
 void cmark_syntax_extension_set_commonmark_escape_func(cmark_syntax_extension *extension,
                                                        cmark_commonmark_escape_func func) {
   extension->commonmark_escape_func = func;

ext/markly/syntax_extension.h

@@ -28,6 +28,7 @@ struct cmark_syntax_extension {
   cmark_postprocess_func          postprocess_func;
   cmark_opaque_alloc_func         opaque_alloc_func;
   cmark_opaque_free_func          opaque_free_func;
+  cmark_opaque_copy_func          opaque_copy_func;
   cmark_commonmark_escape_func    commonmark_escape_func;
 };
 

lib/markly/node.rb

@@ -20,15 +20,7 @@ class Node
 		#
 		# @returns [Markly::Node] The duplicated node tree.
 		def dup
-			# This is a bit crazy, but it's the best I can come up with right now:
-			node = Markly.parse(self.to_markdown)
-
-			# If we aren't duplicating a document, we return `first_child` as the root will be a document node:
-			if self.type == :document
-				return node
-			else
-				return node.first_child
-			end
+			_dup
 		end
 
 		# Walk the node tree recursively.

releases.md

@@ -1,5 +1,9 @@
 # Releases
 
+## Unreleased
+
+  - Preserve complete node and extension metadata when duplicating node trees.
+
 ## v0.17.0
 
   - Add opt-in language prefixes for inline code spans with `Markly::INLINE_CODE_INFO`, expose code metadata through `Node#code_info`, and provide `Node#code_language` as a convenient language accessor.

test/markly/node.rb

@@ -25,6 +25,14 @@
 			expect(dup.to_html).to be == document.first_child.to_html
 			expect(dup).not.to be_equal(document.first_child)
 		end
+
+		it "preserves extension metadata" do
+			document = Markly.parse("| Left | Right |\n| :--- | ---: |\n| A | B |\n", extensions: [:table])
+			dup = document.dup
+
+			expect(dup.first_child.table_alignments).to be == [:left, :right]
+			expect(dup.to_html(extensions: [:table])).to be == document.to_html(extensions: [:table])
+		end
 	end
 
 	with "#type" do

test/markly/node/inline_code_info.rb

@@ -33,6 +33,14 @@
 			expect(document.to_commonmark).to be == "ruby:`Object.new`\n"
 		end
 
+		it "preserves the language when duplicated" do
+			copy = document.dup
+			code = copy.first_child.first_child
+
+			expect(code.code_info).to be == "ruby"
+			expect(copy.to_html).to be == document.to_html
+		end
+
 		it "can update and clear the language" do
 			code.code_info = "c++"
 			expect(code.code_info).to be == "c++"