Creating an instance of Doc with a wrongly-typed word list results in wrong TypeError

[DTWdata]

How to reproduce the behaviour

We forget to cast an object to string, using str(potato()) as an example.

import spacy
from spacy.tokens import Doc
nlp = spacy.load("en_core_web_sm")

class potato:
    def __init__(self):
        pass
    def __str__(self):
        return "potato"

token_texts = ["I", "like", potato(), "!"]
labels = [("O"), ("O"), ("I-FOOD"), ("O")]
whitespaces = [True, True, False, False]
doc = Doc(nlp.vocab, words=token_texts, ents=labels, spaces=whitespaces)

results in the error

doc = Doc(nlp.vocab, words=token_texts, ents=labels, spaces=whitespaces)
File "spacy\tokens\doc.pyx", line 268, in spacy.tokens.doc.Doc.__init__
TypeError: an integer is required

whereas it should result in

doc = Doc(nlp.vocab, words=token_texts, ents=labels, spaces=whitespaces)
File "spacy\tokens\doc.pyx", line 268, in spacy.tokens.doc.Doc.__init__
TypeError: a string is required

The line 268 in spacy.tokens.doc.Doc.init does however not raise this specific error message, unless I am mistaken. I propose adding another elif clause to check if it is not a string, and raising an error. I don't know what the else clause is for, so I am not opening a pull request (although I could, if desired)

Your Environment

• spaCy version: 3.1.1
• Platform: Windows-10-10.0.19041-SP0
• Python version: 3.8.2
• Pipelines: de_core_news_lg (3.1.0), de_core_news_sm (3.1.0), de_dep_news_trf (3.1.0), en_core_web_sm (3.1.0)

[polm]

This error is confusing but what's happening is that when it's not a string it's being treated as potentially an ID/hash. That would be an int type as an argument to get_by_orth, and the mismatch there is raising the error.

We can look at making the error clearer. I'm not sure we actually need to support providing IDs in this context anyway, so that's something else we can check.

[github-actions]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.