INCHIKEY field gets filled with html content

[hechth]

In an annotated MSP, there are some INCHIKEY's which are html texts instead of an inchi, see below attached file.

I don't know with which options the file was annotated, so currently can't reproduce the error.

bug_report.txt

[hechth]

This error occurs when running the CIR: smiles -> inchikey conversion

[xtrojak]

Can you also provide the input file? I cannot reproduce this error.

[xtrojak]

There are some cases when CIR shows help page instead of the result. Compare the following cases:

• Correct smiles shows inchikey: https://cactus.nci.nih.gov/chemical/structure/CC(NC(C)=O)C/stdinchikey
• Probably incorrect smiles shows the help page: https://cactus.nci.nih.gov/chemical/structure/CC(NC(C)=O)C#N/stdinchikey
• Unknown smiles returns 404 correctly: https://cactus.nci.nih.gov/chemical/structure/XXXXX/stdinchikey

[xtrojak]

This should be generalised to validate whether, for example, obtained inchi or smiles has valid format. For that, matchms validation functions can be used.

It seems like it detects some issues even for simple molecules (e.g. according to matchms, PubChem's Methane record contains invalid smiles).