ChatBI系统架构思考—自动生成SQL

I will give you a list of postgresql tables schemas in JSON format, and some instructions.
Then I will ask you some questions about tables provided like Question: 
Query which students in Class 1 with scores greater than 90 where teacher name is zhangSan. 
 You might need to join tables to answer the questions.

Below is the format:
  Table Schema: (JSON array)
  Instructions: (sentences)

Table Schema:
{"sz_school":{"teachers":["teacher_name","teacher_class"],"students":["student_name","student_score","student_class"]}} ;

Instructions:
  * The table in the where clause appear in the tables or temp tables you selected from.
  * Use TO_CHAR(), DO NOT use FORMAT_DATE().
  * Convert TIMESTAMP to DATE using CAST().
  * Use full column name including the table name.
  * You can ONLY read, cannot UPDATE or DELETE or MAKE ANY CHANGES to the data.
  * It is Okay to make assumptions to answer the question.
  * DO NOT use any field not included in schemas.
  * Keep the assumptions concise.
  * You should return assumptions and PLAIN TEXT postgresql query for the question ONLY, NO explanation, NO markdown.
  * Use UNNEST() for ARRAY field.
  * Wrap table name with \`\`
  * NO content after the query.
  * Table name in the query should be without database name.

  Use the following format for response:
    Database: (string)
    Table Name: (string)
    Assumptions: (bullets)
    Query: (query)

Database: sz_school_db

Table Name: sz_school.teachers, sz_school.students

Assumptions:
- Teacher ZhangSan teaches Class 1.
- There is only one teacher per class.

Query:
SELECT sz_school.students.student_name
FROM sz_school.students
JOIN sz_school.teachers ON sz_school.students.student_class = sz_school.teachers.teacher_class
WHERE sz_school.students.student_class = 1 AND sz_school.students.student_score > 90 AND sz_school.teachers.teacher_name = 'ZhangSan';

我将给你一份PostgreSQL数据库的表元数据信息的JSON格式数据以及一些指令。然后，我会根据这些表信息提出一些问题，例如：“查询Class 1中分数大于90的学生，教师名字为zhangSan。”您可能需要通过关联表生成SQL语句来回答这些问题。

以下是数据格式： 
   表元数据信息格式: (JSON数组) ，
   指令: (句子)

表元数据信息格式: {"sz_school":{"teachers":["teacher_name","teacher_class"],"students":["student_name","student_score","student_class"]}} ;

指令:
1、WHERE子查询语句中提到的表应该出现在存在的数据库中或者是查询语句的临时表。
2、支持对应PostgreSQL的方言，例如日期函数使用TO_CHAR()，不使用DATE_FORMAT (); 使用CAST()将TIMESTAMP转换为DATE。
3、应该使用完整的列名，包含表名。
4、你只能生成查询数据的SQL语句，不能生成修改更新或删除数据或对数据进行任何更改的SQL语句。
5、可以做出回答问题的假设，对假设做简明的解释。
6、不能使用表元数据信息中未包含的任何字段。
7、生成的SQL查询语句后不使用任何内容。
8、查询中的表名应不带数据库名。
9、请使用以下格式进行回复： 
  数据库: (字符串) 
  表名: (字符串) 
  假设: (句子) 
  查询: (生成的查询SQL语句)   

{
  "数据库": "sz_school",
  "表名": "students",
  "假设": "在这个查询中，我们假设存在表名为'teachers'和'students'的两个表，它们通过'teacher_class'和'student_class'字段进行关联。同时，我们假设'teacher_name'字段的数据类型为VARCHAR，'teacher_class'字段的数据类型为VARCHAR，'students'表的'student_score'字段的数据类型为INTEGER，'student_class'字段的数据类型为VARCHAR。",
  "查询": "SELECT students.student_name FROM students INNER JOIN teachers ON students.student_class = teachers.teacher_class WHERE teachers.teacher_name = 'zhangSan' AND students.student_score > 90;"
}

parentMessageId = ChatGPTAPI.sendMessage( promptDescribe，null)；
ChatGPTAPI.sendMessage( promptInstructions，parentMessageId )；