Backslash causing issues

[OutGame1]

Hello, I encountered a problem with backslashes. I don't know if this is the expected output, but when using a backslash in a string, it somehow consider that the string is not closed.

Expected behavior

import minify from "pg-minify";
const backslash = '\\';
console.log(minify(`SELECT * FROM USER WHERE NAME = '${backslash}'`, {compress: true}));
> "SELECT * FROM USER WHERE NAME='\'"

Actual behavior

SQLParsingError: Error parsing SQL at {line:1,col:33}: Unclosed text block.
    at throwError (...\node_modules\pg-minify\lib\parser.js:186:15)
    at minify (...\node_modules\pg-minify\lib\parser.js:124:17)
    at <anonymous> (...\src\index.ts:214:13)
    at ModuleJob.run (node:internal/modules/esm/module_job:218:25)
    at async ModuleLoader.import (node:internal/modules/esm/loader:329:24)
    at async loadESM (node:internal/process/esm_loader:34:7)
    at async handleMainPromise (node:internal/modules/run_main:113:12) {
  error: 'Unclosed text block.',
  code: 1,
  position: { line: 1, column: 33 }
}

When I replace all the backslashes with a token, then replaces them back, it works :

import minify from "pg-minify";
const bsToken = "__BS__";
const backslash = '\\';
const sql = `SELECT * FROM USER WHERE NAME = '${backslash}'`;
const replaced = sql.replaceAll(backslash, bsToken);
const minified = minify(replaced, {compress: true});
console.log(minified.replaceAll(bsToken, backslash));
> "SELECT * FROM USER WHERE NAME='\'"

Environment

• Version of pg-minify: 1.7.0
• OS type (Linux/Windows/Mac): Windows
• Version of Node.js: 20.10.0

[vitaly-t]

That's because this library automatically checks for escaping sequences in strings (presence of \ in them), and when found, makes sure to use Unicode strings, i.e. making sure E is prepended to the string.

In your case, \\ in JavaScript is in fact a single \ inside a string, which is not a valid string when represented in Unicode:

• E'\' - invalid
• E'\\' - valid

I know this can be confusing. Effectively, you end up with query:

SELECT * FROM USER WHERE NAME = E'\'

which is not valid.

For your case, you should use:

const backslash = '\\\\';

This will result in SQL:

SELECT * FROM USER WHERE NAME = E'\\'

which is a valid SQL, in which E'\\' is Unicode presentation of a single back-slash

To summarize, when this library sees \ inside a string, it makes sure the string is cast as Unicode.

---

Understandably, this may not be the designed behavior in your case, and we might consider a feature to disable that.

[OutGame1]

Hello, thanks for answering me, I understood the problem. Adding a feature that would disable this behaviour could be interesting, I don't know if this is feasible in the current architecture as I didn't look the code but maybe adding some documentation about this could be a bit less confusing and don't take that much time.

[vitaly-t]

I had time for a closer look at the current logic for this thing...

pg-minify/lib/parser.js

Line 131 in 0981186

const hasLB = text.indexOf('\n') > 0;

As you can see from the above, the code checks if the quoted text contains \n or \t, and then makes sure such text block is pre-pended with E. i.e. the information I gave earlier wasn't correct (sorry, I wrote it many years ago).

Basically, you are not supposed to have non-Unicode strings that contain \n or \t, and so the current logic is perfectly valid. I don't think that switching it off, even optionally, would be a good addition. Therefore, I am closing this issue.