CN101169713A - Model Transformation Framework and Method Supporting Separation of Transformation Engine and Mapping Rules - Google Patents

Abstract

The invention discloses a model conversion framework and method supporting the separation of conversion engine and mapping rules. The framework includes a model extractor, and further includes: a target platform manager, an agent rule manager, a mapping rule query engine, and a mapping rule library , mapping rule editor, mapping rule execution engine, script combiner, script interpreter and public data area; proxy rules pass model elements to the mapping rule query engine, and the mapping rule query engine queries the mapping rule base according to the model elements to obtain the model elements The corresponding mapping rules and proxy rules will then hand over the model elements and mapping rules to the mapping rule execution engine for processing; the mapping rule execution engine will send the processed target script file to the script combiner for merge processing to generate a target script file; script The interpreter interprets the target script file to generate the target model or code. The invention realizes the separation of the model transformation engine and the mapping rules.

Description

Model Transformation Framework and Method Supporting Separation of Transformation Engine and Mapping Rules

technical field

The invention relates to the technical field of computer software development, in particular to a model conversion framework and method based on a Model Driven Architecture (MDA) that supports the separation of a conversion engine and a mapping rule.

Background technique

Object Management Group (OMG) proposed a new system software development methodology - Model Driven Architecture (MDA) at the end of 2002. The core idea of this architecture is to separate the business function logic of the system from the implementation platform technology to avoid tight coupling between business logic and implementation technology. At the same time, MDA can map the platform-independent system function model to different implementations through model conversion technology. On the platform model, the conversion to specific code is finally realized. So that the portability, interoperability and reusability of the software system can be maintained for a long time, it can cope with future technological changes, and prolong the life cycle of the software. At present, MDA-based software development - MDD (Model Driven Development, Model Driven Development) has attracted great attention and attention from academia and industry.

Model conversion technology is one of the key technologies of MDA-based software development, and it is the link to maintain the connection between models of different abstract levels in MDA. Using the model conversion technology, it is possible to realize the conversion from the platform independent model (Platform Independent Model, PIM) to the platform specific model (Platform Specific Model, PSM), and the conversion from the platform dependent model PSM to the implementation code. In order to successfully promote MDA, OMG is trying to standardize model transformation. In March 2002, it released a draft called QVT (MOF Queries, Views and Transformations), where MOF (Meta Object Facility) is a meta object facility. In November 2005, the final version of the specification proposal was finalized and is currently in the final voting stage.

Since model conversion has not yet formed a set of standardized methods, there is a lack of specifications to support the standardization of model conversion, resulting in the technology of model conversion relying on the MDA tools of specific manufacturers. Most of the existing MDA tools do not support model transformation perfectly, and there are still many technical problems to be solved.

Most of the conversion engines provided by current MDA tools (implementing the function of model conversion) implement source model extractors to support scanning and accessing models; some general conversion rule mechanisms are also defined to support conversion between specific models. However, there is insufficient support or no support for the conversion of models related to specific domains. In a specific field, the semantics of UML (Unified Modeling Language, Unified Modeling Language) often cannot fully express the requirements, so UML needs to be extended. Generally, the Profile (extension set) mechanism of UML is used, and the stereotype (Stereotype), label Value (TaggedValue) and constraint (Constraint) to extend existing UML elements and introduce new model elements. Using UML's Profile extension mechanism, users can customize some domain-related modeling elements to express certain exact semantics in a specific domain. If the source model contains new extended modeling elements defined in the UML Profile, a corresponding transformation is required for each new model element. Under the conversion mechanism of the existing MDA tool, it is necessary to add new conversion rules or modify existing conversion rules to realize the conversion of new model elements. Therefore, for specific or domain-related software development activities, it is often necessary to extend the existing conversion engine and add specific conversion rules to complete the corresponding model conversion.

Current transformation engines provide support for extensions to varying degrees. However, the transformation engine and the transformation rules are usually bound together, and a transformation engine usually has only one same rule set, that is, all transformation rules deal with the same source metamodel and the same target metamodel. If the newly added rules and the rules in the conversion engine belong to the same rule set, the conversion engine can be extended more conveniently. If they do not belong to the same ruleset, a new transformation engine will often need to be developed. This mode not only increases the technical difficulty and cost of model transformation engine development, but also greatly reduces the reusability of most general functions in model transformation engines, hindering the application and promotion of MDA in specific or domain-related software development activities .

The relevant prior art includes: the U.S. patent "Extending Existing Model-To-Model Transformations" published on June 15, 2006 with the publication number US2006/0130011A1 (extending existing models to model transformations) and June 2006 The US patent "Model-To-Model Transformation By Kind" published on the 15th with the publication number US2006/0130008 A1 (model-to-model transformation based on the model element type), and the publication number US2006 published on June 15, 2006 /0130009 A1's US patent "Dynamically Configurable Model-To-Model Transformation Engine" (dynamically configurable model-to-model transformation engine).

No. 0130011, 0130009, and 0130008 U.S. patents have the following disadvantages: these three patents are all aimed at different technical aspects of the same conversion engine, and their main disadvantage is that the conversion engine and conversion rules are bound together. The conversion technology only realizes the reuse on the same rule set (that is, all the conversion rules in a conversion engine process the same source meta-model and the same target meta-model), and the source model based on the same meta-model should be converted into When the target model is based on a different meta-model, a new transformation engine and corresponding transformation rules must be redeveloped.

Contents of the invention

The technical problem to be solved by the present invention is to provide a model conversion framework and method that support the separation of the conversion engine and the mapping rules, so as to realize the separation of the model conversion engine and the mapping rules.

In order to achieve the above object, the present invention provides a model conversion framework that supports the separation of the conversion engine and the mapping rules, including a model extractor for extracting model elements in the source model, characterized in that it also includes: a target platform management device, a proxy rule manager, a mapping rule query engine, a mapping rule library, a mapping rule editor, a mapping rule execution engine, a script merger, a script interpreter and a public data area;

The target platform manager processes the relationship between the proxy rule and the mapping rule of the same target platform; the proxy rule manager searches for the proxy rule corresponding to the model element according to the type of the model element; the mapping rule editor The mapper is edited in a visual way to generate mapping rules, and stored in the mapping rule base; the public data area is used to store the data required for model conversion;

The proxy rule transmits the model element to the mapping rule query engine, and the mapping rule query engine queries the mapping rule library according to the model element to obtain a mapping rule corresponding to the model element, and the proxy rule Then hand over the model element and the mapping rule to the mapping rule execution engine for processing;

The mapping rule execution engine sends the processed target script file to the script combiner for merging processing to generate a target script file; the script interpreter interprets the target script file to generate a target model or code.

In the model conversion framework that supports the separation of the conversion engine and the mapping rules, the mapping rule base is further divided into: a version class package, an evaluation package, a statement package, and a rule package.

The model conversion framework that supports the separation of the conversion engine and the mapping rules, wherein the mapping rules further include: version class nodes, evaluation nodes, branches, and statement nodes; wherein the version class nodes are for A reference to a version class, the evaluation node is a reference to an evaluation, the statement node is a reference to a statement, and the branch is used to connect the evaluation node and the statement node.

The model conversion framework that supports the separation of the conversion engine and the mapping rules, wherein, in the mapping rules, the number of the version node and the evaluation node is one, and the branch node, the The number of statement nodes is one or more.

The model conversion framework that supports the separation of the conversion engine and the mapping rules, wherein the same type, evaluation or statement is included in multiple mapping rules, and one branch is only included in one in the mapping rules.

The model conversion framework that supports the separation of the conversion engine and the mapping rules, wherein the mapping rule query engine further includes:

A feature string query module, configured to match a feature string input by the proxy rule with another feature string in the mapping rule and return the mapping rule when the two are the same; or

The unary model query module is configured to match the unary model passed in by the proxy rule with another metamodel in the mapping rule and return the mapping rule when the two are the same.

In order to achieve the above object, the present invention also provides a model conversion method using the model conversion framework to separate the conversion engine from the mapping rules, which is characterized in that it includes:

Step 71, extract the model elements in the source model through the model extractor, and pass it to the proxy rule manager, and the proxy rule manager obtains the proxy rule corresponding to the model element according to the type of the model element ;

Step 72, the proxy rule transmits the model element to the mapping rule query engine, and the mapping rule query engine queries the mapping rule library according to the model element to obtain a mapping rule corresponding to the model element, so The proxy rules then pass the model elements and the mapping rules to the mapping rule execution engine for processing; and

Step 73: Merge the target script files processed by the mapping rule execution engine through the script combiner to generate an target script file, and then interpret the target script file through the script interpreter to generate target models or codes.

The model conversion method that supports the separation of the conversion engine and the mapping rules, wherein, in the step 72, it also includes: when the model element does not have a meta-model attached, the mapping rule query engine uses a characteristic string The step of querying the mapping rules in query mode is specifically: the mapping rule query engine matches a characteristic string passed in by the proxy rule with another characteristic string in the mapping rule and returns when the two are the same. The mapping rules.

The model conversion method that supports the separation of the conversion engine and the mapping rule, wherein, in the step 72, it also includes: when the model element has a meta-model attached, the mapping rule query engine uses a meta-model query The step of querying the mapping rules specifically includes: the mapping rule query engine matches the unary model passed in by the proxy rule with another meta model in the mapping rule and returns the mapping rule when the two are the same.

The model conversion method that supports the separation of the conversion engine and the mapping rules, wherein, in the step 72, it also includes a step of judging whether the current model element contains sub-elements by the model extractor, and if so, returns to The step 71 is executed, otherwise, the step 73 is continued.

Compared with the prior art, the present invention realizes the separation of the model conversion engine and the mapping rules by refining the model conversion rules into proxy rules and mapping rules, enhances the reusability of some common modules in the conversion engine, and greatly simplifies the model The workload and technical difficulty of the conversion. Utilizing the model conversion method proposed in this invention, when realizing the conversion from the source model based on the same meta-model to various target models based on different meta-models, there is no need to re-develop a model conversion engine, and only need to build a new model through the mapping rule editor The mapping rules based on the source metamodel to various target metamodels can be added to the mapping rule base. This model conversion method also greatly improves the reusability of mapping rules.

In addition, the present invention supports a model conversion method in which the conversion engine and the mapping rules are separated, and is applicable to both model-to-model and model-to-code conversions. In the conversion from model to model, the script interpreter will generate the target model; in the conversion from model to code, the code is also abstracted into a model to represent, and then as long as the function of the script interpreter is changed, the script interpreter can correspond to the model code is generated.

The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments, but not as a limitation of the present invention.

Description of drawings

Fig. 1 is a structural diagram of a model conversion framework module of the present invention;

Fig. 2 is a structural diagram of the proxy rule manager of the present invention;

Fig. 3 is that the agent rule of the present invention uses keyword mode to carry out query flowchart;

Fig. 4 is a flow chart of querying the agent rules of the present invention using the meta-model mode;

Fig. 5 is a structural diagram of mapping rules of the present invention;

Fig. 6 is a flow chart of model conversion in the present invention.

Detailed ways

The following is a further detailed description of the model conversion framework and model conversion method of the present invention that supports the separation of the conversion engine and the mapping rules in conjunction with the accompanying drawings:

As shown in Figure 1, it is a structural diagram of the model conversion framework module of the present invention, and Figure 2 is a structural diagram of the proxy rule manager of the present invention, and shown in conjunction with Figures 3, 4, and 5; the model conversion framework consists of a conversion engine and a mapping rule base System composition, including:

The model extractor 101 is responsible for extracting model elements in the source model. The model elements are extracted by layer-by-layer scanning, that is, the root element in the source model is extracted first, and then handed over to the proxy rule manager 103 for processing. After the root element is processed, the sub-elements of the root element are extracted and handed over to the proxy rule manager. Processor 103.

The target platform manager 102, for the same proxy rule 105, may have mapping rules to multiple target platforms corresponding to it, and the target platform manager 102 is responsible for managing the relationship between the proxy rule 105 and the mapping rules of the same target platform, To ensure that when converting to a specific platform, the proxy rule 105 only invokes the mapping rules related to the platform.

The mapping rule belongs to the static mapping rule, which is a static rule representing the mapping relationship between the source model element and the target model element; the proxy rule 105 does not belong to the static mapping rule, it is obtained by querying the proxy rule manager 103, and is mainly used to extract The model elements extracted by the engine 101 are queried by the mapping rule query engine 104 to the mapping rule library 106 to obtain the mapping rules.

The proxy rule manager 103 is used to find the corresponding proxy rule 105 according to the type of the model element extracted by the model extractor 101; one proxy rule manager 103 can manage multiple proxy rules 105 at the same time.

As shown in FIG. 2 , the proxy rule manager 103 maintains a mapping table of model element types 201 and proxy rules 105 , and for each model element type 201 , there is a proxy rule 105 corresponding to it. When a model element is input, the proxy rule manager 103 will check the type of the model element, return the corresponding proxy rule 105, and hand over the model element to the proxy rule 105 for processing.

The mapping rule query engine 104 is used to support the query of model mapping rules, and it also includes: a feature string query module 1041 and a meta-model query module 1042, which respectively support the feature string query mode and the meta-model query mode for mapping rule query:

(1) Feature string query mode: proxy rule 105 passes in a feature string to query the mapping rule, and the feature string query module 1041 matches the feature string with the feature string in the mapping rule, and if they are the same, returns The rule, as shown in Figure 3;

(2) Metamodel query mode: the proxy rule 105 imports the metamodel to query the mapping rules, and the metamodel query module 1042 matches the metamodel with the metamodel in the mapping rules, and if they are the same, returns the rule, as shown in Figure 4 Show.

The query result of the mapping rule query engine 104 is a mapping rule list, and the mapping rule query engine 104 returns the list to the proxy rule 105 .

The proxy rule 105 is used for querying the mapping rule base 106 through the mapping rule query engine 104 based on the model elements extracted by the model extractor 101 to obtain the mapping rules corresponding to the model elements, and submitting the model elements and the mapping rules obtained from the query together to the mapping rule execution engine 108 for processing.

The mapping rule base 106 is mainly used to store mapping rules for query by the conversion engine. In order to realize the reuse of version classes, evaluation and statement modules, the mapping rule base 106 is divided into four packages:

Profile package: profiles are imported from Profile (container of profiles), including some basic information of profiles;

Evaluation package: Evaluation is the main body of mapping rules, which is used to check the attributes or other information of the metamodel;

Statement package: the statement is written in the target script, so the statement itself is also a target script file;

Rule Package: A collection of mapping rules constitutes a rule package.

Basic rule package: general or highly reusable mapping rules are called basic rules, and its collection constitutes a basic rule package. The basic rule package is generally provided by the transformation engine provider.

The rule package, the basic rule package and the rule library are loosely coupled. You only need to specify the path of the rule package and the basic rule package during initialization, and the rule library can dynamically load the rule package and the basic rule package.

As shown in FIG. 5 , the mapping rule structure includes four parts: Stereotype node 501 , evaluation node 502 , branch (Branch) 503 , and sentence node 504 . Version node 501 is a reference to version, similarly, evaluation node 502 and sentence node 504 are references to evaluation and sentence respectively.

Branch 503 is used to connect evaluation node 502 and statement node 504 . A mapping rule can only contain one version node 501 and evaluation node 502 , but can contain multiple branch nodes 503 and statement nodes 504 . Similarly, a category, comment or statement can also be referenced by multiple mapping rules, but branch 503 can only be included in one mapping rule.

The mapping rule editor 107 is configured to edit and generate mapping rules in a visual manner. Express the relationship between each part of the mapping rule with the EMF (Eclipse Modeling Framework, Eclipse Modeling Framework) model on the Eclipse platform, generate a mapping rule editor 107, and add a mapping rule interpreter, so that the mapping rule editor 107 generates Mapping rules can generate Java code through the interpreter.

The mapping rule execution engine 108 is configured to sequentially load and execute the incoming mapping rule list, and return one or more target script files.

After the mapping rule execution engine 108 receives the mapping rules and model elements from the proxy rule 105, the process of executing the mapping rules is specifically as follows:

Obtain the branch 503 corresponding to the current mapping rule, initialize each branch 503, and then transfer the value of the branch 503 to the evaluation node 502, and the evaluation node 502 performs the input of the value of the branch 503 and the model element according to A certain algorithm (the algorithm is defined when defining the mapping rule) performs the operation, and the result of the operation is a Boolean value. When the value is true, the mapping rule execution engine 108 adds the statement corresponding to the branch 503 (the statement itself is a target script file) into the target script list. When all branches are processed, the mapping rule is executed.

After all the mapping rules are processed, the mapping rule execution engine 108 sends the target script list to the script combiner 109 for merging.

Script combiner 109, used to combine multiple target script files into one target script file. For each object model, a set of XML-based scripts is defined to describe the object model, which is called object script (ObjectScript). The script file containing the target model information described by the target script is called the target script file.

In order to describe various model elements in the target model, corresponding nodes are also defined in the target script, so each target script file may include multiple nodes, and the script combiner 109 combines the same nodes in multiple target scripts. Types of nodes are merged, and finally an object script file is generated.

The script interpreter 110 is used to interpret the target script file to generate the target model during the conversion from model to model; to interpret the target file to generate corresponding code during the conversion from model to code;

The public data area 111 mainly stores some global data in the model conversion process, such as information about scanned model elements, data related to target script files, and user-defined data, and is used to support between proxy rules 105 and between mapping rules Sharing of data.

As shown in Fig. 6, it is a flow chart of model conversion in the present invention. Combined with Figures 1, 3, and 4, the processing steps of the model conversion are as follows:

Step 601, the model extractor 101 extracts the model elements in the source model, and transmits them to the proxy rule manager 103;

Before this step, use the mapping rule editor 107 to generate the mapping rule between the source metamodel and the target metamodel, and save it in the mapping rule base 106;

Step 602, the proxy rule manager 103 queries the proxy rule 105 corresponding to the model element according to the type of the model element, and hands the model element to the proxy rule 105 for processing;

Step 603, the proxy rule 105 passes the model element to the mapping rule query engine 104, and the mapping rule query engine 104 queries the mapping rule library 106 according to the model element to obtain the mapping rule corresponding to the model element; the mapping rule query engine 104 queries the obtained mapping The rules are returned to Proxy Rules 105 for processing;

If there is no additional meta-model (that is, version class) on the model element, only the feature string query is used. The query process is shown in Figure 3. Each proxy rule 105 has a query keyword SearchKey, and the SearchKey is used as the feature character The string is given to the feature string query module 1041 of the mapping rule query engine 104, and the feature string query module 1041 returns all keywords Key as the mapping rules of the feature string;

If a meta-model (i.e. version class) is attached to the model element, the meta-model query method is adopted, as shown in Figure 4, the proxy rule 105 hands over the meta-model extracted from the model element to the meta-model query of the mapping rule query engine 104 Module 1042, the metamodel query module 1042 returns all mapping rules that include the metamodel;

Step 604, the agent rule 105 submits the information of the model element and the mapping rule to the mapping rule execution engine 108; the mapping rule execution engine 108 processes the model element and the mapping rule, obtains one or more target script files, and hands them over to the script Merger 109 performs merge processing;

In this step, the mapping rule execution engine 108 sequentially reads and executes the mapping rules in the list, and each mapping rule returns a target script file; all target script files returned are cached in the public data area 111;

Step 605, the model extractor 101 judges whether the current model element contains sub-elements, and if so, returns to step 601 to start execution, that is, extracts the sub-elements of the model element, otherwise, skips to step 606 to merge the target script file;

Step 606, after all model elements have been scanned and processed, all target script files returned are also stored in the public data area 111; the script combiner 109 will merge all target script files cached in the public data area 111, and finally Generate an object script file and interpret it by the script interpreter 110;

In step 607, the script interpreter 110 interprets the target script file to generate target models or codes.

In this step, it is also applicable to the transformation from the model to the code, and the code is also abstracted to be represented by the model, and then the model can be converted into the corresponding code only by changing the function of the script interpreter 110 .

The invention realizes the separation of the model conversion engine and the mapping rules by refining the model conversion rules into proxy rules and mapping rules, enhances the reusability of some common modules in the conversion engine, and greatly simplifies the workload and technical difficulty of model conversion .

Of course, the present invention can also have other various embodiments, and those skilled in the art can make various corresponding changes and deformations according to the present invention without departing from the spirit and essence of the present invention, but these corresponding Changes and deformations should belong to the scope of protection of the appended claims of the present invention.

Claims (10)

1. model conversion framework of supporting that transform engine and mapping ruler are separated, comprise the model extraction apparatus, the model element that is used for the extraction source model, it is characterized in that, also comprise: a target platform manager, is acted on behalf of rule management, a mapping ruler query engine, a mapping ruler storehouse, a mapping ruler editing machine, mapping ruler execution engine, a script combiner, a script interpreter and a public data area;

Relation between the mapping ruler of described target platform manager processes agent rule and same target platform; Described agent rule manager is searched the agent rule corresponding with described model element according to the type of described model element; Described mapping ruler editing machine generates mapping ruler by a visual means editor, and deposits in described mapping ruler storehouse; Described public data area is used to deposit the model conversion desired data;

Described agent rule is passed to described mapping ruler query engine with described model element, described mapping ruler query engine is inquired about described mapping ruler storehouse according to described model element and is obtained the mapping ruler corresponding with described model element, and described agent rule is given described model element and described mapping ruler described mapping ruler again and carried out engine and handle;

The target script file that described mapping ruler execution engine obtains processing is delivered to described script combiner and is merged processing, generates a target script file; Described script interpreter makes an explanation to described target script file and generates object module or code.

2. the model conversion framework that support transform engine according to claim 1 and mapping ruler are separated is characterized in that, described mapping ruler storehouse is divided into again: version class bag, evaluation and test bag, statement bag and rule bag.

3. the model conversion framework that support transform engine according to claim 2 and mapping ruler are separated is characterized in that, described mapping ruler comprises again: version class node, evaluation and test node, branch and statement node; Wherein, described version class node is to the quoting of version class, and described evaluation and test node is to the quoting of evaluation and test, and described statement node is that described branch is used to connect described evaluation and test node and described statement node to the quoting of statement.

4. the model conversion framework that support transform engine according to claim 3 and mapping ruler are separated, it is characterized in that, in the described mapping ruler, the number of described version class node, described evaluation and test node is one, and the number of described branch node, described statement node is one or more.

5. the model conversion framework that is separated according to claim 3 or 4 described support transform engines and mapping ruler, it is characterized in that, same described version class, described evaluation and test or described statement are contained in a plurality of described mapping rulers, and a described branch only is contained in the described mapping ruler.

6. the model conversion framework that is separated according to claim 1,2 or 4 described support transform engines and mapping ruler is characterized in that described mapping ruler query engine also comprises:

One feature string enquiry module, a feature string that is used for that described agent rule is imported into mate with another feature character string of described mapping ruler and return described mapping ruler when both are identical; Or

One meta-model enquiry module, a meta-model that is used for that described agent rule is imported into mate with another meta-model of described mapping ruler and return described mapping ruler when both are identical.

7. one kind is utilized the described model conversion framework of claim 1 to realize the model conversion method that transform engine and mapping ruler are separated, and it is characterized in that, comprising:

Step 71 by the model element in the described model extraction apparatus extraction source model, and is passed to described agent rule manager, and described agent rule manager obtains the agent rule corresponding with described model element according to the type of described model element;

Step 72, described agent rule is passed to described mapping ruler query engine with described model element, described mapping ruler query engine is inquired about described mapping ruler storehouse according to described model element and is obtained the mapping ruler corresponding with described model element, and described agent rule is passed to described model element and described mapping ruler described mapping ruler again and carried out engine and handle; And

Step 73, by described script combiner described mapping ruler execution engine is handled the target script file that obtains and merge target script file of processing generation, explain that by described script interpreter described target script file generates object module or code again.

8. the model conversion method that support transform engine according to claim 7 and mapping ruler are separated, it is characterized in that, in the described step 72, also comprise: when described model element adds meta-model, by the step of described mapping ruler query engine with a feature string inquiry mode query mappings rule, be specially: the feature string that described mapping ruler query engine imports described agent rule into mates with another feature character string in the described mapping ruler and return described mapping ruler when both is identical.

9. the model conversion method that support transform engine according to claim 7 and mapping ruler are separated, it is characterized in that, in the described step 72, also comprise: when described model element adds meta-model, by the step of described mapping ruler query engine with a meta-model inquiry mode query mappings rule, be specially: the meta-model that described mapping ruler query engine imports described agent rule into mates with another meta-model in the described mapping ruler and return described mapping ruler when both is identical.

10. the model conversion method that is separated according to claim 7,8 or 9 described support transform engines and mapping ruler, it is characterized in that, in the described step 72, comprise that also one judges by described model extraction apparatus whether current model element comprises the step of daughter element, if contain, then be back to described step 71 and carry out, otherwise, described step 73 continued.

Images