A cross-database sql statement uniform analysis processing method
By combining segmentation and feature processor matching with genetic algorithms and caching mechanisms, the problems of low efficiency and insufficient accuracy in cross-database SQL statement conversion are solved, achieving efficient and accurate SQL statement conversion and improved system stability.
Patent Information
- Application Number
- CN202411041195.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2044-07-31
AI Technical Summary
Existing SQL statement conversion methods suffer from inefficiency and inaccuracy in cross-database operations, especially when dealing with complex SQL statements. AST parsing methods are time-consuming and prone to causing statement discrepancies.
By receiving SQL statements submitted by users, the system divides them into multiple segments using a preset statement segmentation strategy, extracts feature information, matches them with feature processors for transformation, combines feature processor association tables to achieve cross-database compatibility, and uses genetic algorithms to optimize population screening and caching mechanisms to improve transformation efficiency and accuracy.
It achieves efficient conversion and improved accuracy of SQL statements, simplifies the development process, reduces errors, improves system stability and availability, and provides detailed execution log support.
Smart Images

Figure CN119106047B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of statement parsing technology, specifically to a unified method for parsing and processing SQL statements across databases. Background Technology
[0002] In the field of information technology, with changing enterprise needs and technological advancements, the demand for database migration and cross-database operations is increasing. These operations typically involve transferring data from one database system to another or executing queries across multiple database systems. However, due to differences in SQL syntax and features between different database systems, directly executing SQL statements often leads to errors. Therefore, developing an effective conversion mechanism is crucial to solving this challenge.
[0003] Existing SQL statement transformation methods typically use Abstract Syntax Tree (AST) parsing to parse SQL statements. While this method solves the SQL statement transformation problem to some extent, it still has some obvious limitations:
[0004] 1. For complex SQL statements, the construction and transformation of AST can be very time-consuming, resulting in low efficiency of the overall transformation process.
[0005] 2. During the conversion process, if the parsing or reconstruction of the AST is inaccurate, the generated SQL statement may not match the original meaning, thus affecting the correctness of the data and consequently making it impossible to guarantee accuracy.
[0006] Therefore, there is an urgent need for a unified method for parsing and processing SQL statements across databases, which can improve the efficiency and accuracy of SQL statement conversion. Summary of the Invention
[0007] One of the objectives of this invention is to provide a unified parsing and processing method for SQL statements across databases, which can improve the efficiency and accuracy of SQL statement conversion.
[0008] To achieve the above objectives, a unified method for parsing and processing SQL statements across databases is provided, including the following steps:
[0009] S1. Receive the current SQL statement submitted by the user;
[0010] S2. Based on a preset statement segmentation strategy, the received current SQL statement is segmented into various segmentation units, where each segmentation unit is a complete SQL statement fragment.
[0011] S3. Based on the formed segmentation units, extract the feature information corresponding to each segmentation unit, and match the processor corresponding to each feature information based on the pre-configured feature processor association table.
[0012] S4. Based on the processor corresponding to each matching feature information, perform statement transformation on each corresponding segmentation unit to generate corresponding transformation result data. The transformation result data is the target SQL statement that can be executed by the target database.
[0013] S5. Send the generated target SQL statement to the target database for execution and generate the corresponding execution results.
[0014] The technical principle and effect of this solution: In this solution, the current SQL statement submitted by the user is first received. Then, the received current SQL statement is segmented into various segmentation units through a preset statement segmentation strategy. Each segmentation unit is a complete SQL statement fragment. Through this reasonable segmentation, the current SQL statement can be divided into multiple parts, which not only achieves statement segmentation but also simplifies the difficulty of statement reading.
[0015] Then, based on each segmentation unit, the corresponding feature information is extracted. Following this, according to a pre-configured feature processor association table, the processor corresponding to each extracted feature information is matched. Based on the matched processors, the corresponding segmentation unit is transformed into a target SQL statement executable by the target database. The generated target SQL statement is then sent to the target database for execution, generating the corresponding execution result. This feature matching and processor matching approach avoids the complexity and inefficiency of traditional parsing methods, achieving efficient SQL statement conversion. Furthermore, directly comparing the features corresponding to the SQL statement and matching the corresponding processor reduces errors that may be introduced by parsing complex SQL statements, improving the accuracy of SQL statement conversion. Of course, the feature processor association table allows for the configuration of features and corresponding processors for SQL statements from various databases, thus achieving compatibility with multiple databases and cross-database compatibility. Throughout the conversion and parsing process, developers do not need to delve into the syntactic differences between SQL statements from different databases; they only need to focus on the correlation between the corresponding features and processors to configure the feature information and processors, thereby simplifying the development and maintenance process.
[0016] Furthermore, the preset statement segmentation strategy is as follows:
[0017] Based on the received current SQL statement, identify the delimiters in the current SQL statement, and perform preliminary segmentation of the current SQL statement based on the delimiters to form several SQL statement sub-fragments;
[0018] Based on the formed SQL statement sub-fragments, several initial populations are randomly generated. The individuals in the initial population are the fragment sets corresponding to multiple SQL statement fragments formed by combining several SQL statement sub-fragments. Based on the constraints, the individuals in the initial population are initially screened and judged. If the constraints are met, the individuals in the corresponding initial population are feasible solutions; otherwise, they are infeasible solutions and are deleted.
[0019] The selected population is subjected to fitness calculation, which includes the calculation of first fitness and second fitness.
[0020] The first fitness is calculated as follows:
[0021]
[0022] In the formula, F1 is the first fitness, and G i X represents the statement completeness corresponding to the i-th SQL statement fragment in the population. i Let K be the keyword order reliability corresponding to the i-th SQL statement fragment in the population, and K be a preset set of SQL statement keywords. i Let M be the keyword corresponding to the i-th SQL statement fragment in the population, and M be the number of keywords corresponding to the i-th SQL statement fragment in the population in the preset set of SQL statement keywords. P represents the actual position of the j-th keyword corresponding to the i-th SQL statement fragment in the SQL statement fragment. j y is the expected position of the j-th keyword in the SQL statement fragment corresponding to the i-th SQL statement fragment in the population; w1, w2, and ∝ are the corresponding weighting coefficients; m is the total number of SQL statement fragments corresponding to the population; E is a randomly generated random number; and bmax is a preset random threshold.
[0023] The second fitness is calculated as follows:
[0024]
[0025] In the formula, F2 is the second fitness, W is the semantic evaluation value corresponding to the population, and r z h is the semantic integrity score corresponding to the z-th SQL statement fragment in the population. z The weighting coefficient corresponding to the semantic integrity score of the z-th SQL statement fragment in the population;
[0026] Within the first preset number of iterations, based on the first fitness corresponding to the population, select the population with a first fitness greater than or equal to the first fitness threshold, and among the populations removed at this time, select the three populations with the smallest second fitness and save them to the backup library.
[0027] When the number of iterations exceeds the first preset number, the current population is combined with the population in the backup pool to form a new population, and the population whose second fitness is less than or equal to the preset second fitness is selected.
[0028] Based on a preset hybridization and mutation strategy, hybridization and mutation operations are performed on the selected population to obtain the corresponding offspring population.
[0029] After obtaining the offspring population, continue to calculate fitness until the preset number of iterations is met;
[0030] Output the offspring population as the set of optimal solutions for multi-objective optimization.
[0031] Beneficial effects: In this scheme, firstly, based on the received current SQL statement, the delimiters corresponding to each delimiter in the current SQL statement are identified, and the current SQL statement is initially segmented based on the delimiters to form several SQL statement sub-segments; based on the formed several SQL statement sub-segments, several initial populations are randomly generated. The individuals in the initial population are the fragment sets corresponding to multiple SQL statement fragments formed by combining several SQL statement sub-segments. In this way, complex SQL statements are simplified. Then, the initial population is filtered by constraints, thereby improving the convergence speed.
[0032] The subsequent fitness calculation of the population is used to achieve a secondary selection of the population. Specifically, in the first fitness calculation, a random number is set for X. i The determination of the calculation method greatly improves the randomness of the algorithm. Within the first preset number of iterations, the first fitness of the population is mainly examined, that is, the first fitness is greater than or equal to the preset first fitness threshold. In this way, it is possible to match the population with relatively small differences in sentence completeness and keyword order.
[0033] Of course, during this process, the eliminated populations will also be selected, and the top three populations with the lowest second fitness will be saved in the backup pool. After completing the first preset number of iterations, the focus will be on semantic evaluation. At this time, the populations stored in the backup pool will be combined with the current population to form a new population. This new population includes the populations selected based on the first fitness and the top three populations with the lowest second fitness from each eliminated population. The offspring population obtained in this way will be better, and will be able to better retain genes with good semantic evaluation, making the subsequent offspring populations even better.
[0034] Furthermore, the following is also included between S2 and S3:
[0035] S230. Based on the formed segmentation units, identify the parameters corresponding to each segmentation unit, and generate the hash cache identifier corresponding to each segmentation unit based on the identified parameters.
[0036] S231. Based on the hash cache identifier corresponding to each generated segmentation unit, query the cache database to see if there is a common hash cache identifier. If so, directly retrieve the conversion result data corresponding to the hash cache identifier from the cache database and execute S4; otherwise, execute S3. The cache database is used to store the hash cache identifier corresponding to the historical segmentation unit and the corresponding conversion result data.
[0037] Beneficial effects: In this solution, the identified hash cache identifiers and their corresponding conversion results are stored in a cache database. Before performing SQL statement conversion, the cache database can be queried to determine if there is a matching hash cache identifier. If so, the corresponding conversion result data can be used directly. Otherwise, subsequent conversions are performed. This caching mechanism greatly improves the conversion efficiency of repetitive SQL statements.
[0038] Furthermore, it also includes S6, which identifies and captures error information in real time during the execution of the target SQL statement in the target database, and generates error message based on the captured error information;
[0039] S7. Record the key operations and transformation results data during the execution process to form the corresponding execution log.
[0040] Beneficial effects: In this solution, by identifying and capturing error information in real time during execution, the present invention can immediately provide error prompts to users, facilitating quick problem location and repair, thereby reducing debugging time and improving system stability and availability.
[0041] By recording key operations and transformation results during the execution process, a detailed execution log is formed, which helps with subsequent audit tracking, performance analysis, and troubleshooting, and also provides a basis for continuous system improvement.
[0042] Furthermore, the preset hybridization mutation strategy is as follows:
[0043] During hybridization, the selected populations are paired up, two hybridization fragments are randomly selected, and two random numbers H1 and H2 between 0 and 1 are assigned to the two hybridization fragments. If both random numbers H1 and H2 are greater than or equal to a preset threshold H, the hybridization process is considered complete. y The hybridization probability of the two hybridization fragments is A. If only one of the random numbers H1 or H2 is greater than or equal to the preset threshold H, then... y The hybridization probability of the two hybridization fragments is B. If both random numbers H1 and H2 are less than the preset threshold H, then... y If the probability of hybridization between the two hybrid fragments is C;
[0044] in;
[0045]
[0046] After determining the corresponding hybridization probability, a random number H3 between 0 and 1 is generated. If the random number H3 is greater than or equal to the hybridization probability, the two corresponding hybridization fragments are hybridized; otherwise, no hybridization operation is performed.
[0047] During mutation, based on the corresponding iteration number and the fitness corresponding to the iteration number, the mutation probability value is configured for each SQL statement fragment in the population. The higher the first fitness, the higher the mutation probability value, and the lower the second fitness, the higher the mutation probability value. The mutation probability value of each SQL statement sub-fragment in the SQL statement fragment is the same as the mutation probability value of the SQL statement fragment.
[0048] Use a roulette wheel to select the corresponding mutated SQL statement sub-fragment, and then perform the mutation operation on the selected SQL statement sub-fragment.
[0049] Beneficial effects: Different combinations of the magnitudes of two random numbers are associated with the crossover probability, and when A+B+C=1 and A>B>C, the probability of crossover is highest when both random numbers are high; if only one is high, the probability is second; and if both are low, the probability is lowest. This crossover strategy helps maintain population diversity and encourages the propagation of high-quality traits.
[0050] The mutation probability is determined by the number of iterations and the fitness of the individuals. As the number of iterations increases, individuals with better fitness have a higher mutation probability. A roulette wheel selection mechanism is used to choose which individuals to mutate, and the mutation operation is performed on the selected SQL statement fragments. This mutation strategy can promote the exploration ability among different individuals in the population, helping the algorithm escape local optima. Attached Figure Description
[0051] Figure 1 This is a logical block diagram of the cross-database SQL statement unified parsing and processing method in Embodiment 1 of the present invention. Detailed Implementation
[0052] The following detailed description illustrates the specific implementation method:
[0053] Example 1
[0054] A unified method for parsing and processing SQL statements across databases, basically as follows: Figure 1 As shown, it includes the following steps:
[0055] S1. Receive the current SQL statement submitted by the user; in this embodiment, the SQL statement submitted by the user is received through API, Web interface, or other means. Perform necessary preprocessing on the received SQL statement, such as removing comments and formatting.
[0056] S2. Based on a preset statement segmentation strategy, the received current SQL statement is segmented into various segmentation units, where each segmentation unit is a complete SQL statement fragment.
[0057] The preset statement segmentation strategy is as follows:
[0058] Based on the received current SQL statement, the delimiters corresponding to the current SQL statement are identified, and the current SQL statement is initially segmented based on the delimiters to form several SQL statement sub-fragments; in this embodiment, the delimiters mainly include semicolons, newlines, etc.
[0059] Based on the formed SQL statement sub-fragments, several initial populations are randomly generated. The individuals in the initial population are the fragment sets corresponding to multiple SQL statement fragments formed by combining several SQL statement sub-fragments. Based on the constraints, the individuals in the initial population are initially screened and judged. If the constraints are met, the individuals in the corresponding initial population are feasible solutions; otherwise, they are infeasible solutions and are deleted.
[0060] The selected population is subjected to fitness calculation, which includes the calculation of first fitness and second fitness.
[0061] The first fitness is calculated as follows:
[0062]
[0063] In the formula, F1 is the first fitness, and G i X represents the statement completeness corresponding to the i-th SQL statement fragment in the population. i Let K be the keyword order reliability corresponding to the i-th SQL statement fragment in the population, and K be a preset set of SQL statement keywords. i Let M be the keyword corresponding to the i-th SQL statement fragment in the population, and M be the number of keywords corresponding to the i-th SQL statement fragment in the population in the preset set of SQL statement keywords. P represents the actual position of the j-th keyword corresponding to the i-th SQL statement fragment in the SQL statement fragment. j y is the expected position of the j-th keyword in the SQL statement fragment corresponding to the i-th SQL statement fragment in the population; w1, w2, and ∝ are the corresponding weighting coefficients; m is the total number of SQL statement fragments corresponding to the population; E is a randomly generated random number; and bmax is a preset random threshold.
[0064] The second fitness is calculated as follows:
[0065]
[0066] In the formula, F2 is the second fitness, W is the semantic evaluation value corresponding to the population, and r z h is the semantic integrity score corresponding to the z-th SQL statement fragment in the population. z The weighting coefficient corresponding to the semantic integrity score of the z-th SQL statement fragment in the population;
[0067] Within the first preset number of iterations, based on the first fitness corresponding to the population, select the population with a first fitness greater than or equal to the first fitness threshold, and among the populations removed at this time, select the three populations with the smallest second fitness and save them to the backup library.
[0068] When the number of iterations exceeds the first preset number, the current population is combined with the population in the backup pool to form a new population, and the population whose second fitness is less than or equal to the preset second fitness is selected.
[0069] Based on a preset hybridization and mutation strategy, hybridization and mutation operations are performed on the selected population to obtain the corresponding offspring population.
[0070] The preset hybridization mutation strategy is as follows:
[0071] During hybridization, the selected populations are paired up, two hybridization fragments are randomly selected, and two random numbers H1 and H2 between 0 and 1 are assigned to the two hybridization fragments. If both random numbers H1 and H2 are greater than or equal to a preset threshold H, the hybridization process is considered complete. y The hybridization probability of the two hybridization fragments is A. If only one of the random numbers H1 or H2 is greater than or equal to the preset threshold H, then... y The hybridization probability of the two hybridization fragments is B. If both random numbers H1 and H2 are less than the preset threshold H, then... y If the probability of hybridization between the two hybrid fragments is C;
[0072] in;
[0073]
[0074] After determining the corresponding hybridization probability, a random number H3 between 0 and 1 is generated. If the random number H3 is greater than or equal to the hybridization probability, the two corresponding hybridization fragments are hybridized; otherwise, no hybridization operation is performed.
[0075] During mutation, based on the corresponding iteration number and the fitness corresponding to the iteration number, the mutation probability value is configured for each SQL statement fragment in the population. The higher the first fitness, the higher the mutation probability value, and the lower the second fitness, the higher the mutation probability value. The mutation probability value of each SQL statement sub-fragment in the SQL statement fragment is the same as the mutation probability value of the SQL statement fragment.
[0076] Use a roulette wheel to select the corresponding mutated SQL statement sub-fragment, and then perform the mutation operation on the selected SQL statement sub-fragment.
[0077] After obtaining the offspring population, continue to calculate fitness until the preset number of iterations is met;
[0078] Output the offspring population as the set of optimal solutions for multi-objective optimization.
[0079] Between S2 and S3, there also includes:
[0080] S230. Based on the formed segmentation units, the parameters corresponding to each segmentation unit are identified, and based on the identified parameters, a hash cache identifier corresponding to each segmentation unit is generated; in this embodiment, specific information such as SQL templates can also be identified.
[0081] S231. Based on the hash cache identifier corresponding to each generated segmentation unit, query the cache database to see if there is a common hash cache identifier. If so, directly retrieve the conversion result data corresponding to the hash cache identifier from the cache database and execute S4; otherwise, execute S3. The cache database is used to store the hash cache identifier corresponding to the historical segmentation unit and the corresponding conversion result data.
[0082] S3. Based on the formed segmentation units, extract the corresponding feature information in each segmentation unit, and match the processor corresponding to each feature information based on the pre-configured feature processor association table; in this embodiment, the corresponding feature information includes table name, field name, operator, etc.
[0083] S4. Based on the processor corresponding to each matching feature information, the corresponding segmentation unit is transformed into a statement to generate corresponding transformation result data. The transformation result data is the target SQL statement that can be executed by the target database. In this embodiment, the processor transforms the original SQL statement into an SQL statement that can be executed by the target database according to a preset transformation protocol.
[0084] S5. Send the generated target SQL statement to the target database for execution and generate the corresponding execution results.
[0085] It also includes S6, which identifies and captures error information in real time during the execution of the target SQL statement in the target database, and generates error message based on the captured error information;
[0086] S7. Record key operations and transformation results during the execution process to form corresponding execution logs. In this embodiment, a log query function is provided to facilitate subsequent auditing and problem tracking.
[0087] The above descriptions are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, based on the guidance provided in this application, improve and implement this solution in combination with their own capabilities. Some typical well-known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.
Claims
1. A unified parsing and processing method for cross-database SQL statements, characterized in that: Includes the following steps: S1. Receive the current SQL statement submitted by the user; S2. Based on a preset statement segmentation strategy, the received current SQL statement is segmented into various segmentation units, where each segmentation unit is a complete SQL statement fragment. S3. Based on the formed segmentation units, extract the feature information corresponding to each segmentation unit, and match the processor corresponding to each feature information based on the pre-configured feature processor association table. S4. Based on the processor corresponding to each matching feature information, perform statement transformation on each corresponding segmentation unit to generate corresponding transformation result data. The transformation result data is the target SQL statement that can be executed by the target database. S5. Send the generated target SQL statement to the target database for execution and generate the corresponding execution results; The preset statement segmentation strategy is as follows: Based on the received current SQL statement, identify the delimiters in the current SQL statement, and perform preliminary segmentation of the current SQL statement based on the delimiters to form several SQL statement sub-fragments; Based on the formed SQL statement sub-fragments, several initial populations are randomly generated. The individuals in the initial population are the fragment sets corresponding to multiple SQL statement fragments formed by combining several SQL statement sub-fragments. Based on the constraints, the individuals in the initial population are initially screened and judged. If the constraints are met, the individuals in the corresponding initial population are feasible solutions; otherwise, they are infeasible solutions and are deleted. The selected population is subjected to fitness calculation, which includes the calculation of first fitness and second fitness. The first fitness is calculated as follows: In the formula, For first fitness, Let represent the statement completeness corresponding to the i-th SQL statement fragment in the population. Let $\frac{i}{i}$ represent the reliability of the keyword order corresponding to the $i$-th SQL statement fragment in the population. For a pre-defined set of SQL statement keywords, Let i be the keyword corresponding to the i-th SQL statement fragment in the population. This represents the number of keywords corresponding to the i-th SQL statement fragment in the population within the preset set of SQL statement keywords. Let be the actual position of the j-th keyword corresponding to the i-th SQL statement fragment in the SQL statement fragment. This represents the expected position of the j-th keyword corresponding to the i-th SQL statement fragment in the SQL statement fragment; , , These are the corresponding weighting coefficients; This represents the total number of SQL statement fragments corresponding to the population. The random number is generated randomly. A preset random threshold is used; The second fitness is calculated as follows: In the formula, For second fitness, The semantic evaluation value corresponding to the population. Let z be the semantic completeness score corresponding to the z-th SQL statement fragment in the population. The weighting coefficient corresponding to the semantic integrity score of the z-th SQL statement fragment in the population; Within the first preset number of iterations, based on the first fitness corresponding to the population, select the population with a first fitness greater than or equal to the first fitness threshold, and among the populations removed at this time, select the three populations with the smallest second fitness and save them to the backup library. When the number of iterations exceeds the first preset number, the current population is combined with the population in the backup pool to form a new population, and the population whose second fitness is less than or equal to the preset second fitness is selected. Based on a preset hybridization and mutation strategy, hybridization and mutation operations are performed on the selected population to obtain the corresponding offspring population. After obtaining the offspring population, continue to calculate fitness until the preset number of iterations is met; Output the offspring population as the set of optimal solutions for multi-objective optimization.
2. The method for unified parsing and processing cross-database SQL statements according to claim 1, characterized in that: Between S2 and S3, there also includes: S230. Based on the formed segmentation units, identify the parameters corresponding to each segmentation unit, and generate the hash cache identifier corresponding to each segmentation unit based on the identified parameters. S231. Based on the hash cache identifier corresponding to each generated segmentation unit, query the cache database to see if there is a common hash cache identifier. If so, directly retrieve the conversion result data corresponding to the hash cache identifier from the cache database and execute S4; otherwise, execute S3. The cache database is used to store the hash cache identifier corresponding to the historical segmentation unit and the corresponding conversion result data.
3. The method for unified parsing and processing of cross-database SQL statements according to claim 2, characterized in that: It also includes S6, which identifies and captures error information in real time during the execution of the target SQL statement in the target database, and generates error message based on the captured error information; S7. Record the key operations and transformation results data during the execution process to form the corresponding execution log.
4. The method for unified parsing and processing of cross-database SQL statements according to claim 3, characterized in that: The preset hybridization mutation strategy is as follows: During hybridization, the selected populations are paired up, two hybridization fragments are randomly selected, and a random number between 0 and 1 is assigned to each of the two hybridization fragments. and If two random numbers and All are greater than or equal to the preset threshold. Then the hybridization probability of the two hybrid fragments is If there is only one random number among them or Greater than or equal to the preset threshold Then the hybridization probability of the two hybrid fragments is If two random numbers and All are less than the preset threshold Then the hybridization probability of the two hybrid fragments is ; in; After determining the corresponding hybridization probability, a random number between 0 and 1 is generated again. If the random number If the probability is greater than or equal to the hybridization probability, then the two corresponding hybridization fragments will be hybridized; otherwise, no hybridization operation will be performed. During mutation, based on the corresponding iteration number and the fitness corresponding to the iteration number, the mutation probability value is configured for each SQL statement fragment in the population. The higher the first fitness, the higher the mutation probability value, and the lower the second fitness, the higher the mutation probability value. The mutation probability value of each SQL statement sub-fragment in the SQL statement fragment is the same as the mutation probability value of the SQL statement fragment. Use a roulette wheel to select the corresponding mutated SQL statement sub-fragment, and then perform the mutation operation on the selected SQL statement sub-fragment.
Citation Information
Patent Citations
Database adaptation method and device, equipment and storage medium
CN117648363A
