ChordsOverWordsParser fails with TypeError on specific valid-looking inputs

[seiler-emerson]

Hello, and thank you for creating and maintaining the chordsheetjs library! It's a very powerful tool.

I am reporting a persistent issue I'm encountering with the ChordsOverWordsParser. It seems to fail with a TypeError on certain "chords-over-words" text inputs, while successfully parsing others that are structurally very similar.

Environment

• Library Version: "chordsheetjs": "^12.3.0",
• Runtime: Next.js 14 Server Action : "next": "14.2.31",
• Language: TypeScript

Steps to Reproduce

The issue can be reproduced with a simple, isolated function call, independent of my application's framework.

1. Code used for parsing:

import {
    ChordProParser,
    ChordsOverWordsParser,
    TextFormatter,
    ChordProFormatter,
} from 'chordsheetjs';

function textToChordPro(textContent: string): string {
    try {
        const parser = new ChordsOverWordsParser();
        const song = parser.parse(textContent);

        const formatter = new ChordProFormatter();
        return formatter.format(song);
    } catch (error) {
        console.error("Failed to convert text to ChordPro:", error);
        return `{start_of_chorus}\n${textContent}\n{end_of_chorus}`;
    }
}

2. Test Case 1 (Successful): This input text is parsed correctly without any issues.

const successfulText1 = `
F   G     Am   G  F
Renova-me senhor Jesus
F        C        G
Já não quero ser igual
`;
textToChordPro(successfulText1); // This works

---

const successfulText2 = `
          Am   G  F
Renova-me senhor Jesus
        C        G
Já não quero ser igual
`;
textToChordPro(successfulText2); // This works

3. Test Case 2 (Fails): This input text, which appears to follow the same "chords-over-words" pattern, consistently fails.

const failingText1 = `
C     G       Am
Um menino nasceu
         C     G      Am
Como um filho se nos deu
                 C
Ele é o próprio Deus
   G       Am
E vive em mim
`;
textToChordPro(failingText1); // This fails

---

const failingText2 = `
C     G       Am
Um menino nasceu
G      Am
Como um filho se nos deu
C
Ele é o próprio Deus
Am
E vive em mim
`;
textToChordPro(failingText2); // This fails

Expected Behavior

I expect parser.parse(failingText) to succeed and return a valid chordpro object, just as it does for successfulText. The structure of both inputs seems to conform to the "chords-over-words" format.

Actual Behavior The parser.parse(failingText) call throws the following error:

TypeError: Cannot read properties of undefined (reading 'toLowerCase')
    at Key.getNoteForNumber (...)
    at Key.get note (...)
    at Key.toString (...)
    at Chord.toString (...)
    at ChordSheetSerializer.parseChordLyricsPair (...)
    ...

Analysis and Debugging Attempts I have made several attempts to debug this, ruling out common issues:

• Non-Standard Whitespace: My first hypothesis was that hidden characters (like non-breaking spaces, \u00A0) were causing the issue. This was disproven when the error persisted even after I manually typed the entire failing text into the code, ensuring only standard spaces were used.
• Indentation and Alignment: My next hypothesis was that the parser was struggling with the leading whitespace on some lyric lines. I created a pre-processing function to normalize the indentation of the entire text block. This also did not solve the issue.

Given that the error persists even with a hardcoded string and after applying the documented solutions, I believe this might be a bug or a limitation in the parser's logic for handling certain valid chord/lyric alignments.

Is this a known issue, or is there a specific structural rule in the failingText that I am overlooking which makes it invalid?

Thank you for your time and for this great library. I'm happy to provide any more information needed.