more bug fixes

version0/crdt/text_document.hpp

@@ -147,8 +147,13 @@ struct TextDocument {
 		std::vector<Op> ops;
 
 		for (size_t i = first_idx; i < last_idx; i++) {
+			if (!state.list.at(i).value.has_value()) {
+				// allready deleted
+				continue;
+			}
+
 			ops.emplace_back(typename ListType::OpDel{
-				state.list[i].id
+				state.list.at(i).id
 			});
 
 			// TODO: do delets get a seq?????
@@ -163,7 +168,7 @@ struct TextDocument {
 	// note: rn it only creates 1 diff patch
 	std::vector<Op> merge(std::string_view text) {
 		if (text.empty()) {
-			if (state.list.empty()) {
+			if (state.list.empty() || state.doc_size == 0) {
 				// no op
 				return {};
 			} else {
@@ -185,6 +190,7 @@ struct TextDocument {
 		// find start and end of changes
 		// start
 		size_t list_start = 0;
+		size_t list_start_counted = 0;
 		size_t text_start = 0;
 		bool differ = false;
 		for (; list_start < state.list.size() && text_start < text.size();) {
@@ -201,6 +207,7 @@ struct TextDocument {
 
 			list_start++;
 			text_start++;
+			list_start_counted++;
 		}
 
 		// doc and text dont differ
@@ -214,7 +221,10 @@ struct TextDocument {
 		// +1 so i can have unsigned
 		size_t list_end = state.list.size();
 		size_t text_end = text.size();
-		for (; list_end > 0 && text_end > 0 && list_end >= list_start && text_end >= text_start;) {
+		//for (; list_end > 0 && text_end > 0 && list_end >= list_start && text_end >= text_start;) {
+		//while (list_end >= list_start && text_end >= text_start) {
+		size_t list_end_counted = 0;
+		while (list_start_counted - list_end_counted > state.doc_size && text_end >= text_start) {
 			// jump over tombstones
 			if (!state.list[list_end-1].value.has_value()) {
 				list_end--;
@@ -227,6 +237,7 @@ struct TextDocument {
 
 			list_end--;
 			text_end--;
+			list_end_counted++;
 		}
 
 		//std::cout << "list_end: " << list_end << " text_end: " << text_end << "\n";
@@ -237,7 +248,7 @@ struct TextDocument {
 		if (list_start <= list_end && list_start < state.list.size()) {
 			ops = delRange(
 				state.list[list_start].id,
-				list_end < state.list.size() ? std::make_optional(state.list[list_end].id) : std::nullopt
+				(list_start == list_end ? list_end+1 : list_end) < state.list.size() ? std::make_optional(state.list[list_end].id) : std::nullopt
 			);
 			//std::cout << "deleted: " << ops.size() << "\n";
 		}
@@ -245,11 +256,11 @@ struct TextDocument {
 		//std::cout << "text between: " << getText() << "\n";
 
 		// 2. add range (add all text_start - text_end)
-		if (text_start < text_end) {
+		if (state.doc_size < text.size()) {
 			auto tmp_add_ops = addText(
 				list_start == 0 ? std::nullopt : std::make_optional(state.list[list_start-1].id),
 				list_start == state.list.size() ? std::nullopt :std::make_optional(state.list.at(list_start).id),
-				text.substr(text_start, text_end-text_start)
+				text.substr(text_start, text.size() - state.doc_size)
 			);
 			//std::cout << "added: " << tmp_add_ops.size() << "\n";
 			ops.insert(ops.end(), tmp_add_ops.begin(), tmp_add_ops.end());

version0/test2.cpp

@@ -5,6 +5,7 @@
 #include <random>
 #include <iostream>
 #include <cassert>
+#include <variant>
 
 // single letter agent, for testing only
 using Agent = char;
@@ -398,8 +399,106 @@ void testChange1(size_t seed) {
 	assert(doc.getText() == otherdoc.getText());
 }
 
+void testBugSame(void) {
+	Doc doc;
+	doc.local_agent = 'A';
+
+	std::string_view new_text1{"a"};
+	doc.merge(new_text1);
+	assert(doc.getText() == new_text1);
+
+	std::string_view new_text2{"aa"};
+	doc.merge(new_text2);
+	assert(doc.getText() == new_text2);
+}
+
+void testBugDoubleDel(void) {
+	Doc doc;
+	doc.local_agent = 'A';
+
+	{
+		std::string_view new_text{"a"};
+		doc.merge(new_text);
+		assert(doc.getText() == new_text);
+	}
+
+	{
+		std::string_view new_text{""};
+		const auto ops = doc.merge(new_text);
+		assert(doc.getText() == new_text);
+		assert(ops.size() == 1);
+		assert(std::holds_alternative<ListType::OpDel>(ops.front()));
+		assert(std::get<ListType::OpDel>(ops.front()).id.seq == 0);
+	}
+
+	{
+		std::string_view new_text{""};
+		const auto ops = doc.merge(new_text);
+		assert(doc.getText() == new_text);
+		assert(ops.size() == 0);
+	}
+}
+
+void testBugSameDel(void) {
+	Doc doc;
+	doc.local_agent = 'A';
+
+	{
+		std::string_view new_text{"a"};
+		doc.merge(new_text);
+		assert(doc.getText() == new_text);
+	}
+
+	{
+		std::string_view new_text{"aa"};
+		const auto ops = doc.merge(new_text);
+		assert(doc.getText() == new_text);
+	}
+
+	{
+		std::string_view new_text{"a"};
+		const auto ops = doc.merge(new_text);
+		assert(doc.getText() == new_text);
+	}
+}
+
+void testBugSameDel2(void) {
+	Doc doc;
+	doc.local_agent = 'A';
+
+	{
+		std::string_view new_text{"a"};
+		doc.merge(new_text);
+		assert(doc.getText() == new_text);
+	}
+
+	{
+		std::string_view new_text{"aa"};
+		const auto ops = doc.merge(new_text);
+		assert(doc.getText() == new_text);
+	}
+
+	{
+		std::string_view new_text{"aaa"};
+		const auto ops = doc.merge(new_text);
+		assert(doc.getText() == new_text);
+	}
+
+	{
+		std::string_view new_text{"aa"};
+		const auto ops = doc.merge(new_text);
+		assert(doc.getText() == new_text);
+	}
+
+	{
+		std::string_view new_text{"a"};
+		const auto ops = doc.merge(new_text);
+		assert(doc.getText() == new_text);
+	}
+}
+
 int main(void) {
-	const size_t loops = 10'000;
+	const size_t loops = 1'000;
 	{
 		std::cout << "testEmptyDocAdds:\n";
 		for (size_t i = 0; i < loops; i++) {
@@ -453,6 +552,34 @@ int main(void) {
 		}
 	}
 
+	std::cout << std::string(40, '=') << "\n";
+
+	{
+		std::cout << "testBugSame:\n";
+		testBugSame();
+	}
+
+	std::cout << std::string(40, '=') << "\n";
+
+	{
+		std::cout << "testBugDoubleDel:\n";
+		testBugDoubleDel();
+	}
+
+	std::cout << std::string(40, '=') << "\n";
+
+	{
+		std::cout << "testBugSameDel:\n";
+		testBugSameDel();
+	}
+
+	std::cout << std::string(40, '=') << "\n";
+
+	{
+		std::cout << "testBugSameDel2:\n";
+		testBugSameDel2();
+	}
+
 	return 0;
 }
 

vim_research/test1.cpp

@@ -1,18 +1,83 @@
 #include <crdt/text_document.hpp>
 #include <nlohmann/json.hpp>
 
+extern "C" {
+#include <zed_net.h>
+}
+
 #include <memory>
 #include <string_view>
-#include <zed_net.h>
+#include <variant>
 
 #include <iostream>
 #include <cassert>
 
 // single letter agent, for testing only
-using Agent = char;
+//using Agent = char;
+using Agent = uint16_t; // tmp local port
 using Doc = GreenCRDT::TextDocument<Agent>;
 using ListType = Doc::ListType;
 
+std::ostream& operator<<(std::ostream& out, const std::optional<ListType::ListID>& id) {
+	if (id.has_value()) {
+		out << id.value().id << "-" << id.value().seq;
+	} else {
+		out << "null";
+	}
+	return out;
+}
+
+std::ostream& operator<<(std::ostream& out, const ListType::OpAdd& op) {
+	out
+		<< "Add{ id:" << op.id.id
+		<< "-" << op.id.seq
+		<< ", v:" << op.value
+		<< ", l:" << op.parent_left
+		<< ", r:" << op.parent_right
+		<< " }"
+	;
+	return out;
+}
+
+std::ostream& operator<<(std::ostream& out, const ListType::OpDel& op) {
+	out
+		<< "Del{ id:" << op.id.id
+		<< "-" << op.id.seq
+		<< " }"
+	;
+	return out;
+}
+
+std::ostream& operator<<(std::ostream& out, const Doc::Op& op) {
+	if (std::holds_alternative<ListType::OpAdd>(op)) {
+		out << std::get<ListType::OpAdd>(op);
+	} else if (std::holds_alternative<ListType::OpDel>(op)) {
+		out << std::get<ListType::OpDel>(op);
+	}
+	return out;
+}
+
+std::ostream& operator<<(std::ostream& out, const std::optional<char>& id) {
+	if (id.has_value()) {
+		out << id.value();
+	} else {
+		out << "null";
+	}
+	return out;
+}
+
+std::ostream& operator<<(std::ostream& out, const ListType::Entry& e) {
+	out
+		<< "{ id:" << e.id.id
+		<< "-" << e.id.seq
+		<< ", v:" << e.value
+		<< ", l:" << e.parent_left
+		<< ", r:" << e.parent_right
+		<< " }"
+	;
+	return out;
+}
+
 static bool send_command(zed_net_socket_t* remote_socket, const std::string_view mode, const std::string_view command) {
 	auto j = nlohmann::json::array();
 
@@ -106,7 +171,7 @@ int main(void) {
 
 	std::cout << "initialized zed_net\n";
 
-	const unsigned int port {1337};
+	const uint16_t port {1337};
 	zed_net_socket_t listen_socket;
 	if (zed_net_tcp_socket_open(
 		&listen_socket,
@@ -142,9 +207,11 @@ int main(void) {
 	// send doauto text changed for inital buffer
 
 	Doc doc;
+	doc.local_agent = remote_address.port; // tmp: use local port as id
+
 	while (true) {
-		// 10MiB
-		auto buffer = std::make_unique<std::array<uint8_t, 1024*1024*10>>();
+		// 100MiB
+		auto buffer = std::make_unique<std::array<uint8_t, 1024*1024*100>>();
 
 		int64_t bytes_received {0};
 		bytes_received = zed_net_tcp_socket_receive(&remote_socket, buffer->data(), buffer->size());
@@ -227,15 +294,34 @@ int main(void) {
 				}
 
 				std::string new_text;
-				for (const auto& line : j_lines) {
-					new_text += line;
-					new_text += '\n';
+				for (size_t i = 0; i < j_lines.size(); i++) {
+					if (!j_lines.at(i).empty()) {
+						new_text += static_cast<std::string>(j_lines.at(i));
+					}
+
+					if (i+1 < j_lines.size()) {
+						new_text += "\n";
+					}
 				}
 
+				//std::cout << "new_text:\n" << new_text << "\n";
+				//std::cout << "old_text:\n" << doc.getText() << "\n";
+				std::cout << "doc state: ";
+					for (const auto& e : doc.state.list) {
+						std::cout << e << " ";
+					}
+					std::cout << "\n";
+
 				const auto ops = doc.merge(new_text);
 				if (!ops.empty()) {
 					std::cout << "ops.size: " << ops.size() << "\n";
+					std::cout << "ops: ";
+					for (const auto& op : ops) {
+						std::cout << op << " ";
+					}
+					std::cout << "\n";
 				}
+				assert(doc.getText() == new_text);
 			} else {
 				std::cout << "unknown command '" << command << "'\n";
 			}