CN114327607A - Automatic generation method of BS code - Google Patents

Abstract

A BS code automatic generation method comprises the following steps: step 1, receiving a database table design file; and 2, matching the interface layout rule of the table object in the knowledge base. Searching a table interface design rule of the most similar table in a knowledge base according to the most similar table data; step 3, matching interface design rules of fields in the knowledge base; searching a field interface design rule of the most similar field in a knowledge base according to the most similar field data; step 4, filtering the layout template codes according to the table interface design rule obtained in the step 2 to obtain layout codes; filtering the field template codes according to the field interface design rule obtained in the step 3 to obtain field interface codes; and combining to obtain the output interface code. The invention measures the similarity of new and old data by adding a similarity algorithm, directly migrates the historical storage data of the knowledge base to the current code programming by taking the similarity as a key index, simplifies the code generation process and improves the code matching degree.

Description

Automatic generation method of BS code

Technical Field

The invention belongs to the technical field of computer software, relates to a code automatic generation technology, and particularly relates to a BS code automatic generation method.

Background

The B/S (Browser/Server) architecture is also referred to as a Browser/Server architecture, and such an architecture may be understood to be a change to and facilitate the C/S architecture. The B/S architecture has a wider application range. The BS front-end code is code that is executable under the B/S architecture.

The code generator is an essential tool in the working process of programmers, is mainly used for developing design codes, analyzing a database and a target language and compiling a new program, and can realize batch generation of various codes and new establishment of projects; the existing code generation method has the disadvantages of long generated code data volume, inconvenient running and testing and large subsequent optimization workload due to insufficient constraint conditions.

Disclosure of Invention

In order to overcome the technical defects in the prior art, the invention discloses a BS code automatic generation method.

The BS code automatic generation method comprises the following steps:

step 1, receiving a database table design file;

the database table design file comprises table information and field information;

step 2, matching the interface layout rule of the table object in the knowledge base

Reading various table information in the data table design file as table input parameters; searching historical table information of the same category in a knowledge base as table reference data;

the knowledge base is a database which stores a large number of historical code files, and the historical code files comprise historical table information and historical field information;

giving different weight scores to different types of table information, and matching and screening table input parameters and table reference data according to the similarity, the use frequency and the weight scores to obtain table reference result data;

according to the table reference result data, searching a table interface design rule corresponding to the table reference result data in a knowledge base;

step 3, matching the interface design rule of the field in the knowledge base

Reading various field information in the data table design file as field input parameters; searching historical field information of the same category in a knowledge base as field reference data;

different weight distributions are given to different types of field information, and field input parameters and field reference data are matched and screened according to the similarity, the use frequency and the weight distribution to obtain field reference result data;

searching a field interface design rule corresponding to the field reference result data in a knowledge base according to the field reference result data;

step 4, generating output interface codes

Inputting a layout template code, and filtering the layout template code according to the table interface design rule obtained in the step 2 to obtain a layout code;

inputting a field template code, and filtering the field template code according to the field interface design rule obtained in the step 3 to obtain a field interface code;

and after the table layout code and the field interface code are obtained, combining to obtain an output interface code.

Preferably, the table information of the database table design file comprises a table name, a table code and a table remark; the field information comprises field name, field code, field type, field length, field uniqueness constraint, field filling constraint, field maximum value constraint, field minimum value constraint, field default value, field remark description and field metadata description.

Preferably, the matching screening process of step 2 specifically includes the following steps:

step 2.1

Reading three table information of a table name, a table code and a table remark which are used as table input parameters, sequentially carrying out similarity matching with the table reference data by using a text similarity algorithm, sequentially outputting 3 groups of matching results, and obtaining similarity scores of all the matching results;

each group of matching results are sorted according to the similarity and the use times, firstly sorted by using the similarity, the similarity is the same, and then sorted according to the use times of the table reference data; selecting the first N1 pieces of table reference data of each type of field information as matching results, wherein N1 is a preset first primary screening number;

step 2.2

Assigning different weight distributions to different classes for the matching results obtained in the step 2.1;

giving 1-time weight to the remarks;

endowing 2 times of weight distribution to the table name and the table code;

according to the position of data sequencing in each class, the 1 st position is divided into 3 points, the 2 nd position is divided into 2 points, the descending is carried out in sequence until 1 point is reached, and the data of subsequent ranking is divided into 1 point, so that all the data can participate in scoring.

The final score = position score and weight score of each matching result, and the final score of the data in the classification sorting is obtained;

and for each table reference data in the matching result, performing score accumulation in three scoring standards of table name, table code and table remark, and outputting the table reference data with the highest score as the table reference result data.

Preferably, the matching screening process of step 3 specifically includes the following steps:

step 3.1

Reading 11 field information of field names, field codes, field types, field lengths, field uniqueness constraints, field mandatory filling constraints, field maximum constraints, field minimum constraints, field default values, field remark descriptions and field metadata descriptions serving as field input parameters, sequentially performing similarity matching with the field reference data by using a text similarity algorithm, sequentially outputting 11 groups of matching results, and obtaining similarity scores of all matching results;

each group of matching results are sorted according to the similarity and the use times, firstly sorted by using the similarity, the similarity is the same, and then sorted according to the use times of the table reference data; selecting the first N2 pieces of table reference data of each type of field information as matching results, wherein N2 is the preset second primary screening number;

step 3.2

Giving different weight scores to different classes for the matching results obtained in the step 3.1;

giving 1-time weight to field default values, field remark description and field metadata description;

2 times of weight distribution is given to the field maximum value constraint, the field minimum value constraint and the field length;

giving 3 times weight distribution to field uniqueness constraint and field filling constraint;

4 times of weight distribution is given to the field name and the field code;

giving 8 times weight to the field type;

according to the position of data sorting in each class, the position of the 1 st bit is divided into 10 minutes, the position of the 2 nd bit is divided into 9 minutes, and the 1 st bit is sequentially decreased until 1 minute;

the final score = position score and weight score of each matching result, and the final score of the data in the classification sorting is obtained;

for each field reference data in the matching result, performing score accumulation in eleven scoring standards of field name, field code, field type, field length, field uniqueness constraint, field mandatory filling constraint, field maximum constraint, field minimum constraint, field default, field remark description and field metadata description, outputting the first N3 field reference data with the highest score as recommended candidate field data, and outputting N3 as the preset third primary screening number;

and pushing recommendation candidate field data, and selecting the field data with the highest score by a user or default as field reference result data.

Preferably, the similarity algorithm is one or more weighted mean of Euclidean distance, cosine similarity, Hamming distance, Jaro distance, Manhattan distance, Jaccard similarity coefficient, S Bronsten-Dice coefficient, Jaro-Winkler distance and SimHsah Hamming distance algorithm.

Compared with the prior art, the BS code automatic generation method has the following technical advantages that:

the similarity of new and old data is measured by adding a similarity algorithm, and historical storage data of a knowledge base is directly migrated to the current code programming by taking the similarity as a key index, so that the code generation process is simplified, and the code matching degree is improved.

And secondly, the codes are filtered and simplified by adopting a filtering method, and the number of output result codes is greatly reduced.

And giving different weight scores to different types of data, and screening the codes by comprehensively adopting the use frequency, the weight and the similarity, so that the pertinence and the reliability of the codes are improved.

Detailed Description

The following provides a more detailed description of the present invention.

The code generation method of the invention comprises the following steps:

step 1, receiving a database table design file;

the database table design file is a database table design file designed by PowerDesigner which is general software in the prior industry;

the database table design file generally contains table information such as table names, table codes and table remarks, and field information such as field names, field codes, field types, field lengths, field uniqueness constraints, field filling constraints, field maximum value constraints, field minimum value constraints, field default values, field remark descriptions and field metadata descriptions, and the table information and the field information are all required information for code generation.

Step 2, matching the interface layout rule of the table object in the knowledge base

Reading three field information of a table name, a table code and a table remark in a data table design file as table input parameters; searching historical table names, table codes and table remark information in files stored in a knowledge base to serve as table reference data;

the knowledge base is generally located on a server, such as a cloud-side server.

And matching the table input parameters with the table reference data, and solving a matching result according to the algorithm rule of the step 2.1.

Step 2.1

Reading three field information of a table name, a table code and a table remark which are used as table input parameters, sequentially carrying out similarity matching with the table reference data by using a text similarity algorithm, and sequentially outputting 3 groups of matching results.

The similarity calculation method can be a known algorithm such as Euclidean distance, cosine similarity, Hamming distance and the like.

Each group of matching results are sorted according to the similarity and the use times, firstly sorted by using the similarity, the similarity is the same, and then sorted according to the use times of the table reference data; the first 3 pieces of the table reference data of each type of field information are selected as matching results, namely the first primary screening number N1= 3.

According to the actual use condition, the more the use times of the historical data stored in the knowledge base is, the most suitable historical data is not represented, the similarity between codes is more important due to the problems of individuation and design habits of software design, the similarity compares the differences of naming specifications, terms and industry descriptions of two designs to a certain extent, for example, the raw material is called as a material in the catering industry, the raw material is called as a material in the manufacturing industry, the raw material and the material are the same but are not similar, and the similarity can distinguish industries, terms and habits to a certain extent.

The preference of similarity as a first screening index also represents personalized design learning, for example, if a user uses a code for the first time to generate a first table, although the number of usage times is 1, since the user design habit is preferred, the user design should be preferentially considered, and not other designs with more usage times in history. If the history library can be matched with similar data, the data is the historical design of the user or the design of industries with the same history and habits, so that the similarity is first, and the personalized code design style of the user can be better adapted.

Step 2.2

And (3) giving the matching results obtained in the step (2.1) to different weights to obtain a table weight score ordering:

giving 1-time weight to the remarks;

endowing 2 times of weight distribution to the table name and the table code;

according to the position of data sorting in each class, the 1 st bit position is divided into 3 minutes, the 2 nd bit position is divided into 2 minutes, and the 1 st bit position is decreased in sequence until 1 minute.

The final score = position weight similarity of each bar, resulting in the final score of the bar in the sort order.

And for each table reference data in the matching result, performing score accumulation in three scoring standards of table name, table code and table remark, and outputting the table reference data with the highest score as the table reference result data.

The name of the table represents the main business object, which is the direct object of the software system data storage, the name represents the important business meaning such as the stock table, the table code is the unique English letter representing code for the name, and has the same importance and meaning with the name, therefore, the weight is 2 times of the weight. However, when the names of the tables are the same, the table remark information can be further distinguished, so that the table remark weight is set to be 1 time of the weight division.

Step 2.3

And (3) inquiring the interface design rule aiming at the table stored in the knowledge base according to the table reference result data obtained in the step (2.2).

The interface design rules comprise layout design rules and master-slave design rules;

the layout design rules include: the 5 layouts such as the position middle layout, the position east layout, the position west layout, the position north layout and the position south layout can be assembled into a left-right, up-down and other composite layout modes.

The master-slave design rules include: design rules such as an independent table, a master table of a master-slave table, a slave table of the master-slave table and the like.

The main design rules specifically function as follows:

independent table: the relationships between tables are independent relationships. The form presentation is a separate form, popup.

Master table of master-slave table: when generating the code, some tables may have a master-slave relationship with other tables. For example, a report reimbursement list, the header of which is a master list, and reimbursement details of which are slave lists, 1 master list records a plurality of data corresponding to the reimbursement details of the slave lists. While the style layout of the main form presentation is usually in the position of seeing the screen or west, or north or center.

Slave table of master-slave table: from the reimbursement details of the above example, the interface is usually presented in a tabular form, and there are a plurality of pieces. Such slave tables, typically layout presentation positions, are all bottom or east positions, and are presented in association with the master table. The editing mode of the slave table is generally different from that of the master table, and more is direct editing or popup editing. And a partial table is a natural slave table, such as an XX list. In the actual process, the master-slave relationship combines two objects of a master-slave table to form an interface display form.

And 3, matching interface design rules of fields in the knowledge base.

Reading field type information, taking a field name, a field code, a field type, a field length, a field uniqueness constraint, a field mandatory filling constraint, a field maximum constraint, a field minimum constraint, a field default value, a field remark and a field metadata remark as field input parameters, reading field type information stored in a database in a historical acquisition mode in a knowledge base as field reference data, matching the field input parameters with the field reference data, and solving a matching result according to the algorithm rule of the step 3.1.

And 3.1, carrying out similarity matching on various field information and field reference data by using a text similarity algorithm in sequence, and outputting 11 types of matching results in sequence. And (3) arranging the 11 types of matching result sets according to the similarity and the use times in a reverse order, outputting the first 10 pieces of matching results of each type as the matching results, namely outputting 110 pieces of matching results with the second primary screening number of N2= 10:

step 3.2, the data output by the step 3.1 are subjected to weight scoring and sorting:

assigning 1-time weight to the field default value, the field remark and the metadata remark of the field;

2 times of weight distribution is given to the field maximum value constraint, the field minimum value constraint and the field length;

giving 3 times weight distribution to field uniqueness constraint and field filling constraint;

4 times of weight distribution is given to the field name and the field code;

giving 8 times weight to the field type;

and calculating the score of each piece of data in turn for the 11 types of matching result sets according to the weight distribution table.

Different weights are given to different field class information, so that the key form of the field control display is determined according to the influence of the field class information on the field, such as the type of a field. Such as float decimal type and integer shaped and varchar string type fields, their display controls are very different. The weight is divided into the highest 8 times weight division. The field name and the field code determine a property range determined by the comparison of the field, such as the name, and determine that the field is a name-related property of an object, so that the capacity range of the field can be well determined. The uniqueness constraint on whether the field must be filled is different in different industries. The importance of the two is less than the first two. The default values of the fields, the field description and the metadata description belong to subjective service description, so that the subjectivity is high and the importance is relatively low; however, these description fields often record the actual remark information of the designer during the design process. So it takes 1 times the weight division.

The scoring method comprises the following steps: according to the position of data sorting in each class, the 1 st bit is divided into 10 th points and the 2 nd bit is divided into 9 th points, and the number is decreased sequentially until the 1 st point. Data score = position weight similarity, resulting in the final score of the piece of data in the sort order.

And (4) accumulating scores in each field reference data in the matching result in each category, and outputting the top 3 fields reference data with the highest scores as the recommended candidate field data, namely the third primary screening number N3= 3.

Step 3.3, according to the recommended candidate field data output in the step 3.2, pushing the recommended candidate field data to an optimal field reference result data selected by a designer manually according to experience

And inquiring interface design rules stored in the knowledge base for the field reference result data.

The interface design rule is the design rule information of the field reference result data collected and stored in a history file stored in a knowledge base, and comprises the following 5 parts of contents:

1. a control type of the field;

2. the field layout position in the form, for example, typically includes the following rules

2.1 whether the field is hidden from view;

this entry has a default value, does not require the user to enter and see the field, but must exist in the foreground code,

2.2 whether the field exists in the front end; the item does not need to be input by a user, does not need to be displayed on a front-end interface, and data is completely filled by a back-end server;

2.3 whether the field is read-only; the item is visible to the user and can not be changed;

2.4 field, e.g. only numbers or letters can be entered as input rules,

2.5 field check rule; for example, whether the input identification card, mail, telephone number, etc. is correct; ,

2.6 whether it is a mandatory item;

3. the sequence of fields in the form, the data display style (including font, color and background), whether the fields are hidden or not and the width of field display;

4. whether the field in the toolbar is a query field;

5. whether the field is a foreign key associated field;

step 4 output interface code generation

4.1 inputting layout template code, filtering the layout code of the output table according to the layout design rule obtained in step 2 to obtain table layout code,

4.2 generating field interface codes, namely firstly inputting field template codes, and filtering the interface codes in the field template codes to obtain the field interface codes according to the interface design rules obtained in the step 3 as conditions;

the template code is a mature technology, and dynamic data is filled and the template code is output through keyword matching replacement.

Common template 1:

allowBlank ：{$forceInput},

when the common template 1 adopts a text replacement mode, whether the value of the mandatory item is true or false, the code is output. In fact, when false, the output code is redundant.

Template and filtration methods herein:

firstly, designing the required filling attribute value of the file according to the field, and if the required filling attribute value is required, normally outputting the template code of the attribute. If the attribute is not necessary to be filled, the code of the attribute is not output, and the unnecessary filling is defaulted, so that the redundant attribute code output is reduced.

1) If the mandatory item is true, firstly, according to the context environment of the code, the position of the output attribute code is judged at the head, middle and tail of the position of the code syntax, and the position of the syntax separator is determined.

2) If the attribute code is at the beginning of the grammar position, the code filling is directly output. [ Attribute 1]

3) If the attribute code is in the bits of the syntax position, a syntax delimiter is added before the line is output. To ensure grammatical correctness. Such as: [ Attribute 1, Attribute 2]

4) If the attribute code is at the end of a bit of a syntax position, a syntax separator is added before the line is output. Since it is the last one, the last syntax separator is removed.

5) The first, middle, and last bits may change in the code context, such as assembling an array object, for example, [ line number, box, number, xxxx ] for boxes. If no recommended line number is present in the design rule, the box becomes the head. If there is a row number, the box is in bits. This effect is generally present at both ends. The grammar position record is adopted to ensure that the grammar of the code output is correct.

And after the table layout code and the field interface code are obtained, combining to obtain an output interface code.

One specific code architecture for the front-facing object is:

{

network request code

Toolbar code

Query field code

Form field code

Form field code

Service function code: add, delete, modify, view, copy, export, import

Form code, embedded reference form field code, embedded reference toolbar code

Form code: embedding reference form field code, embedding reference service function code

Popup window code

Layout panel

Reference to "form code"

Introduction of "form code"

}

Returns to the layout panel-to display

}

One specific code architecture for the back-end facing the object is:

control layer code

{

Injection service layer interface

Front-end and back-end communication variable interface

Service function code: data conversion, data verification, object assembly, calling logic layer interface, such as adding, deleting, modifying, checking, copying, exporting and importing

Returning data results

}

Service layer code

{

Injected data persistence layer interface

Service function code: the functions of adding, deleting, modifying, inquiring, copying, exporting, importing and the like are realized by the code

}

Persistent layer code

{

Injection database connection pool interface

Interface code: object save, update, delete, query, multi-object save, multi-object update, delete on object, multi-conditional query, etc.

The foregoing is a description of preferred embodiments of the present invention, and the preferred embodiments in the preferred embodiments may be combined and combined in any combination, if not obviously contradictory or prerequisite to a certain preferred embodiment, and the specific parameters in the examples and the embodiments are only for the purpose of clearly illustrating the inventor's invention verification process and are not intended to limit the patent protection scope of the present invention, which is defined by the claims and the equivalent structural changes made by the content of the description of the present invention are also included in the protection scope of the present invention.

Claims (5)

1. A BS code automatic generation method is characterized by comprising the following steps:

step 1, receiving a database table design file;

the database table design file comprises table information and field information;

step 2, matching the interface layout rule of the table object in the knowledge base

Reading various table information in the data table design file as table input parameters; searching historical table information of the same category in a knowledge base as table reference data;

the knowledge base is a database which stores a large number of historical code files, and the historical code files comprise historical table information and historical field information;

giving different weight scores to different types of table information, and matching and screening table input parameters and table reference data according to the similarity, the use frequency and the weight scores to obtain table reference result data;

according to the table reference result data, searching a table interface design rule corresponding to the table reference result data in a knowledge base;

step 3, matching the interface design rule of the field in the knowledge base

Reading various field information in the data table design file as field input parameters; searching historical field information of the same category in a knowledge base as field reference data;

different weight distributions are given to different types of field information, and field input parameters and field reference data are matched and screened according to the similarity, the use frequency and the weight distribution to obtain field reference result data;

searching a field interface design rule corresponding to the field reference result data in a knowledge base according to the field reference result data;

step 4, generating output interface codes

Inputting a layout template code, and filtering the layout template code according to the table interface design rule obtained in the step 2 to obtain a layout code;

inputting a field template code, and filtering the field template code according to the field interface design rule obtained in the step 3 to obtain a field interface code;

and after the table layout code and the field interface code are obtained, combining to obtain an output interface code.

2. The BS code automatic generation method of claim 1, wherein the table information of the database table design file includes a table name, a table code, a table remark; the field information comprises field name, field code, field type, field length, field uniqueness constraint, field filling constraint, field maximum value constraint, field minimum value constraint, field default value, field remark description and field metadata description.

3. The BS code automatic generation method of claim 1, wherein the matching screening process of step 2 specifically comprises the steps of:

step 2.1

Reading three table information of a table name, a table code and a table remark which are used as table input parameters, sequentially carrying out similarity matching with the table reference data by using a text similarity algorithm, sequentially outputting 3 groups of matching results, and obtaining similarity scores of all the matching results;

each group of matching results are sorted according to the similarity and the use times, firstly sorted by using the similarity, the similarity is the same, and then sorted according to the use times of the table reference data; selecting the first N1 pieces of table reference data of each type of field information as matching results, wherein N1 is a preset first primary screening number;

step 2.2

Assigning different weight distributions to different classes for the matching results obtained in the step 2.1;

giving 1-time weight to the remarks;

endowing 2 times of weight distribution to the table name and the table code;

according to the position of data sequencing in each class, the position of the 1 st bit is divided into 3 points and the position of the 2 nd bit is divided into 2 points, the positions are sequentially decreased until the 1 point is reached, and the data of subsequent ranking are all divided into 1 point, so that all the data can participate in scoring;

the final score = position score and weight score of each matching result, and the final score of the data in the classification sorting is obtained;

and for each table reference data in the matching result, performing score accumulation in three scoring standards of table name, table code and table remark, and outputting the table reference data with the highest score as the table reference result data.

4. The BS code automatic generation method of claim 1, wherein the matching screening process of step 3 specifically comprises the steps of:

step 3.1

Reading 11 field information of field names, field codes, field types, field lengths, field uniqueness constraints, field mandatory filling constraints, field maximum constraints, field minimum constraints, field default values, field remark descriptions and field metadata descriptions serving as field input parameters, sequentially performing similarity matching with the field reference data by using a text similarity algorithm, sequentially outputting 11 groups of matching results, and obtaining similarity scores of all matching results;

each group of matching results are sorted according to the similarity and the use times, firstly sorted by using the similarity, the similarity is the same, and then sorted according to the use times of the table reference data; selecting the first N2 pieces of table reference data of each type of field information as matching results, wherein N2 is the preset second primary screening number;

step 3.2

Giving different weight scores to different classes for the matching results obtained in the step 3.1;

giving 1-time weight to field default values, field remark description and field metadata description;

2 times of weight distribution is given to the field maximum value constraint, the field minimum value constraint and the field length;

giving 3 times weight distribution to field uniqueness constraint and field filling constraint;

4 times of weight distribution is given to the field name and the field code;

giving 8 times weight to the field type;

according to the position of data sorting in each class, the position of the 1 st bit is divided into 10 minutes, the position of the 2 nd bit is divided into 9 minutes, and the 1 st bit is sequentially decreased until 1 minute;

the final score = position score and weight score of each matching result, and the final score of the data in the classification sorting is obtained;

for each field reference data in the matching result, performing score accumulation in eleven scoring standards of field name, field code, field type, field length, field uniqueness constraint, field mandatory filling constraint, field maximum constraint, field minimum constraint, field default, field remark description and field metadata description, outputting the first N3 field reference data with the highest score as recommended candidate field data, and outputting N3 as the preset third primary screening number;

and pushing recommendation candidate field data, and selecting the field data with the highest score by a user or default as field reference result data.

5. The BS code automatic generation method of claim 1, wherein the similarity algorithm is one or several weighted means of euclidean distance, cosine similarity, hamming distance, Jaro distance, manhattan distance, Jaccard similarity coefficient, S red-Dice coefficient, Jaro-Winkler distance, SimHsah hamming distance algorithm.