Best way to reference PostgreSQL query in PERN app - reactjs

Question

Summary: PERN app needs to download XLSX that represents the same query that was run to populate a data grid on the client side. For example a user has defined a series of product family prefixes and they're sent to the server endpoint:
const getItemsByFamily = async (like) => {
const items = await fetch('/item/like/' + like);
const result = await items.json();
return result;
};
Where the endpoint uses a call to the associated database function:
app.get('/item/like/:like', async (req, res) => {
const { like } = req.params;
const { rows } = await dbo.getByItemPrefix(like);
res.json(rows);
});
And, the function runs the actual query against the PG Db:
getByItemPrefix: async (like) => await pool.query(`select * from myschema.items where ${like}`);
All is well... But, now stakeholders on this project have requested the ability to download a representative XLSX for the same data set. My idea is to use PG's COPY command to write a unique XLSX on the server and link to it somehow.
I start with a proof of concept, hard-coded address, just to simply download a sample xlsx:
// TODO: Add some sort of reference to represent data-grid query.
const DownloadXLSX = ({/* query or query ref */}) => {
return (
<a href='http://localhost:1234/download?fileName=sample.xlsx' target='_blank' rel='noopener noreferrer' download>
<Button
variant='contained'
size='small'
>
Download
</Button>
</a>
);
};
But, now I'm on SO because I am stuck in an architecture question. What is possible?
Is it time to audit every query and reference it somehow that way? Currently this product has no auditing, but perhaps it is time to spend time to do that? Possibly make an audit entry and send the id down with the results, passing the audit.id down do the download button for its query param?
Or, is there some more simple way to track a data grid's representative query on the client?
I like to do whatever is optimal and I'm hoping to find opinions or some insight. Thanks in advance for any help in getting over this architecture hurdle (missing having a team!).

Answer 1

I thought I should follow up here.
To access the actual text of a query within a PG stored function that could be used with a trigger you can use the function current_query().
However, my whole idea of using triggers went down the drain when I found this in the PG docs:
SELECT does not modify any rows so you cannot create SELECT triggers. Rules and views are more appropriate in such cases.
While the suggested approach might work in some cases it really would not for me as the results of some queries are huge and I really needed a reference.
Instead I wrap pool.query with a function that just takes the passed-in SQL and stores it in a simple, but sufficient for my purposes, "audit" table.
CREATE TABLE myschema.audit
(
id serial,
user_id varchar(50) not null,
full_query text not null,
x_time timestamp not null default now(),
CONSTRAINT audit_pkey PRIMARY KEY (id)
)
TABLESPACE pg_default;
ALTER TABLE myschema.audit
OWNER to postgres;
pool.queryWithAudit = async query => {
const { rows } = await pool.query(SQL`insert into iol.trac_audit (user_id, full_query) values ($1, $2) returning id`, ['dev-trac', query]);
const auditID = rows[0].id;
const queryResult = await pool.query(query);
queryResult.auditID = auditID;
return queryResult;
};
And, finally, the retun from the server changed just a little bit to include the auditID, and the download component had to change (to CSV for now, instead of XLSX):
const DownloadXLSX = ({ queryId, style = {} }) => {
return (
<a style={{ textDecoration: 'none', ...style }} href={`http://localhost:1234/download?queryId=${queryId}`} target='_blank' rel='noopener noreferrer' download>
<Button
endIcon={<GetAppIcon />}
size='small'
>
Download
</Button>
</a>
);
};
Along with corresponding route for lookup of query, then execution:
app.get('/download', async (req, res) => {
const { queryId } = req.query;
// Now grab query from audit and run that.
const fileName = await dbo.makeXLSX(queryId);
res.download(fileName, fileName.split('/').pop(), (err) => {
console.log(err ? 'Error attempting to download: ' + err.message : 'No server-side error in attempt to download.');
if (!err) fs.unlink(fileName, (unlinkErr) => console.log(unlinkErr ? 'Temp folder cleanup error: ' + unlinkErr.message : `File at: [${fileName}] removed from system temp folder.`));
});
});
dbo.js updates:
makeXLSX: async auditId => {
const { rows } = await pool.query(`select full_query from myschema.audit where id = ${auditId}`);
const query = rows[0].full_query;
const fileName = path.join('C:/tmp/', Date.now().toString() + '.csv');
const copyCommand = `COPY (${query}) TO '${fileName}' CSV HEADER;`;
await pool.query(copyCommand);
return fileName;
}
Anyway, this works for me. Maybe somebody will submit improvements? I just include the button if there is a queryId, which I null-out with the associated useState setter function:
const [queryId, setAuditIdRef] = useState(null);
{queryId ? <DownloadXLSX queryId={queryId} /> : null}
Good day.