[cristianchiriac]

Description

Fixes #2634.

RichText.de_json() treated a plain-text string as JSON whenever it looked like a JSON object or array and parsed. A superscript footnote mark like "[56]" (common in Wikipedia-sourced rich messages) parses to the list [56]. It then went on to check_json() and obj.pop('type', None) on that list, and polling failed with:

pop expected at most 1 argument, got 2

Now only a string that parses to a JSON object is handled as a serialized RichText. Anything else ("[56]", "[]", numbers, "null", invalid JSON) is returned as plain text, as the docs describe ("can be either a String for plain text…"). A JSON array string never worked before either: it hit the same pop on a list. Real arrays of rich text arrive as lists and still go through the existing list branch.

Describe your tests

New test_json_richtext_plain_text_that_looks_like_json:

• the issue's case: a URL whose text is a superscript "[56]" deserializes to RichTextUrl → RichTextSuperscript with text "[56]";
• "[56]", "[]", "{}x", "123", "null" and '["a", "b"]' passed directly come back unchanged.

It fails before the change and passes after it. pytest tests/test_types.py: 266 passed.

Python version: 3.12

OS: Windows 11

Checklist:

• I added/edited example on new feature/change (if exists) (not applicable)
• My changes won't break backward compatibility
• I made changes both for sync and async (shared telebot/types.py)

[Badiboy]

I assumed that

https://core.telegram.org/bots/api#richtext

RichText
This object represents a rich formatted text. Currently, it can be either a String for plain text, an Array of [RichText]...(https://core.telegram.org/bots/api#richtext), or any of the following types:

means array in data... Looks you are right: "Array of [RichText]" in fact is also encoded as json...

[Badiboy]

Mmm. I remebered the reason of having this condition.

[{"x": 1}, {"y": 2}]

this is also a valid JSON. Thus due to the problem arized by Telegram who gives us "...it can be either a String for plain text,..." the way to detect if string is JSON or not is quite unsolvable...

Thus I supposed that
Array of [RichText](https://core.telegram.org/bots/api#richtext)
will be represented exactly as

[{richtext},{richtext},{richtext}]

in json string.

I cannot proof this. If this suggestion is wrong - we can apply your fix. If not - we definitely cannot.

[cristianchiriac]

An Array of RichText doesn't reach de_json as a string. The whole update is one JSON document, parsed once, and every caller passes obj['text'] (or obj['credit']) from that already-parsed dict. So an array arrives as a Python list and goes through the list branch, and a nested array like {"type": "bold", "text": ["a", {"type": "italic", "text": "b"}]} still gives ['a', RichTextItalic] with this change. When the value is a str, Telegram sent a JSON string, which is plain text.

A string like '[{"type": "bold", "text": "x"}]' was also never handled on master: it passes the json.loads check, then check_json returns a list and obj.pop('type', None) fails with the same pop expected at most 1 argument error as "[56]" (checked on current master). So the change doesn't drop any case that worked before.

[Badiboy]

Okay, looks you are right. I'll check a bit more and will proceed.

Thank you.