| With the development of cloud services and Web 2.0,migrating traditional local spreadsheet editing systems to the cloud and adding multi-person real-time collaboration is of great importance to improve work efficiency.Although there are mature online collaborative applications in the industry,these applications are unable to fuse multiple parties’ wishes when faced with operational conflicts,but simply prompt users for conflicts or allow multiple conflicting versions to coexist,so it is a challenge in collaborative editing to preserve users’ wishes as much as possible while ensuring consistent results.Among the collaborative editing algorithms that support the fusion of operational wishes,the address space conversion algorithm has the following problems: on the one hand,current research is mainly designed for 1D document-based applications,and the collaborative editing support for 2D spreadsheet is poor.On the other hand,the algorithm suffers from excessive local storage and network transmission overhead in 2D spreadsheet scenarios.Second,selective undo is already an indispensable feature in collaborative editing systems,which allows users to recover from errors and improves the editing experience,but the string-oriented address space conversion algorithm does not support the selectable undo feature.In addition,current collaborative editing systems are usually built with C/S architecture,which leads to the problem of easy tampering of collaborative data.To solve these problems,this paper improves the address space conversion-based collaborative editing algorithm and designs a 2D table collaborative editing system based on Hyperledger Fabric.The main research contents of the paper are as follows:(1)Improved the string-wise address space transformation collaborative editing algorithm(ASTS).The main improvements are as follows: firstly,we summarize the 2D conflicts encountered in inserting/deleting cells in sheet co-operation and provide a conflict fusion scheme for ASTS,which takes into account the operation wishes of multiple users.Secondly,we summarize the possible conflicts encountered in cell text undo operations and provide a corresponding solution by adding Undo semantics to the algorithm so that it can support selective undo operations at the string level.The experiments show that the average operation time of Undo operation is 0.086 ms and 0.158 ms in character-level and string-level datasets,respectively,and the experimental results are basically consistent with the time complexity of the theoretical analysis,and the improved algorithm can meet the real-time requirements of the collaborative editing system.In addition,in order to optimize the redundant operations,save local storage space and reduce the network transmission overhead in 2D spreadsheet scenario,the cell text-based pre-fusion method ASTS-DIFF is designed.Compared with ASTS,ASTS-DIFF adds Move semantics to find the appropriate move position for redundant operations by computing the longest regular sequence.After datasets validation,it is found that the number of operations generated by ASTS-DIFFno Move is reduced by 38.5% on average when the character length N<600,and the number of operations generated by ASTS-DIFF-Move operation is reduced by 46.1% on average when N<1000.In terms of storage space consumption for operations,ASTS-DIFF reduces the number of operations by 45.3% on average compared to ASTS.(2)A 2D collaborative editing system based on Hyperledger Fabric federated chain is designed and implemented.The collaborative editing system is divided into a view layer,a collaborative data query middleware and a network layer.The 2D collaborative editing algorithm is deployed in the view layer and integrated into the view publish-subscribe module to realize user interaction and collaborative functions.The network layer defines the organization of each node on the Fabric and the definition of related rights and policies.In addition,the collaborative data query middleware is designed to solve the problems of low efficiency of submitting/querying collaborative information in Fabric network and not supporting block data rich query.The experimental results show that the configured Fabric network can achieve the best outgoing block efficiency when the Batch Timeout is 5s and Max Message Count is 720 for the setting of the sorting node outgoing block parameters.In addition,the QPS metrics of the collaborative data query middleware and Fabric＿network designed in this paper are compared on the basis of the optimal block-out parameters set for the sorting nodes.by a factor of 4.2,4.3,and 3.1,respectively,and on the Etherscan2021＿data dataset by a factor of 5.5,3.5,and 1.1,respectively.Finally,it is verified that the response time of the rich query statement meets the real-time requirement of the collaborative system. |
