Wrong input lines on error

[agladysh]

long-text.txt

1
2
3
4

Error message:

Error: Syntax error at line 1 col 1:

1  3
   ^
2  4
3  
Unexpected numbers token: "1". Instead, I was expecting to see one of the following:

Note the error message mentions token value "1", but points line with 3. Line numbers are also off.

How to reproduce:

long-text.ne

@preprocessor typescript

@{%
import moo from 'moo';

const lexer = moo.compile({
  newlines: { match: /\n+/, lineBreaks: true },
  numbers: /[0-9 ]+/
});

%}

@lexer lexer

grammar -> chunk:+

chunk -> %newlines | %spaces # no %numbers

long-text.ts

import nearley from 'nearley';

import compiled from './long-text-compiled';

async function main() {
  const grammar = nearley.Grammar.fromCompiled(compiled);
  const parser = new nearley.Parser(grammar);

  for await (const chunk of process.stdin) {
    parser.feed(chunk.toString('utf8'));
  }

  parser.finish();
}

main()
  .catch(async (e) => {
    console.error(e);
    process.exit(1);
  });

Command:

npx nearleyc <long-text.ne >long-text-compiled.ts && npx tsx long-text.ts < long-text.txt

Version:

% npx nearleyc --version
2.20.1

[jeremysf]

I'm seeing something similar using nearley and moo.

In my case, the line number and offset in the error are correct, but the displayed source context in the error is from a completely different part of the input.

In the example below, the error really is a line 66 col 18, but the displayed text of "}" is from the end of the input which is actually line 151.

Syntax error at line 66 col 18:

64      vec2 position = 2;
65      optional string pinReference = 3;    
66  }
                     ^
Unexpected symbol token: "extends". Instead, I was expecting to see one of the following:

A lbrace token based on:
    element → %element %ws %symbol element$ebnf$1 ● %lbrace element$ebnf$2 element$ebnf$3 %rbrace
    declaration →  ● element
    unit$ebnf$1 → unit$ebnf$1 ● declaration
    unit →  ● unit$ebnf$1

In my case, I'm feeding the parser like this:

const parser = new nearley.Parser(nearley.Grammar.fromCompiled(grammar))
parser.feed(sourceText)
const results = parser.finish()

[TheGrandmother]

I am experiencing this as well

[timneedham]

Hi,

Yeh, it's the same for me as well. Line numbers are spot on, but as said elsewhere: the displayed source context in the error is from a completely different part of the input.

[snosenzo]

Also seeing this with our omgidl parsing library that uses nearley and moo here: https://github.com/foxglove/omgidl/tree/main/packages/omgidl-parser Line numbers are correct but source is not.