No way to "reverse" parse from `SQL_DML_Command` back to SQL (`text`)

[steven-contreras-datapipertech]

This will be needed on rare occasions, at least in my project it is. I am currently working on a solution.

[steven-contreras-datapipertech]

So far, I am thinking a "reverse parse" grammar will be needed in order to do this the "right" way. What are your thoughts?

[steven-contreras-datapipertech]

For further context, here is my particular use-case:

• first, I have developed an API that "enforces" referential integrity - i.e. I have documents that refer to other document id's as a foreign key
• and when a user attempts to delete a document with a dependent "foreign key" that still exists, I hook my api call to delete the PK document
• if there are any documents that depend on the document to be deleted, an exception is thrown indicating that the PK document has dependencies and cannot be deleted
• to search for dependent documents, I need to parse the delete SQL and convert it to select SQL to first find the document ids that would be selected by the delete SQL
• then I query the dependent collections for any occurrences of the primary document id as a FK
• if any exist, the exception (as previously) described is thrown
• this depends on:
  • parsing the delete SQL - to a structured SQL_DML_Command - and using it's "from" and "where" clause to craft/convert to a select SQL statement, specifying the docid as column in the select SQL statement
  • I have built a preliminary tree traversal algorithm (in-order, typical) but it is getting messy
  • I would like to utilize the parse-tree semantic currently in-place if possible

Thus, I am thinking the best/cleanest way to do this would be to build out a "reverse parse" grammar from SQL_DML_Command to SQL (text).

Your thoughts?

[steven-contreras-datapipertech]

I've found an alternate method to accomplish this.

[steven-contreras-datapipertech]

After successfully implementing this in my API, I wonder if this feature would be desirable in PyFireSQL itself - that is, would enforcing Referential Integrity restraints be a desirable feature in PyFireSQL? If so, we ought to have a more thorough design discussion. Let me know what you think.