Fix for TelegramJsonParser

[galatolofederico]

Messages in my Telegram JSON exports occasionally lack the "from" and "text" attributes. This PR fixes that.

[joweich]

Left one small comment. Thank you for your contribution @galatolofederico!

[joweich]

@galatolofederico Do we need these two additional checks? In which cases would they return false, while the first two checks return true?
If we don't need them, I would drop them for codestyle and performance reasons.

[galatolofederico]

Hi honestly type(mess["text"]) is str is a bit of a workaround because it filters only the pure text messages.
text can be either a string if the message contains only text or a list containing strings and dicts.

For example

    {
      "id": ...,
      "type": "message",
      "date": "...",
      "date_unixtime": "...",
      "from": "...",
      "from_id": "...",
      "text": [
        "This is some text followed by ",
        {
          "type": "link",
          "text": "https://a-link"
        }
      ],
      "text_entities": [
        {
          "type": "plain",
          "text": "This is some text followed by "
        },
        {
          "type": "link",
          "text": "https://a-link"
        }
      ]
    },

The second check len(mess["text"]) > 0 is redundant i guess.
Maybe the best approach would be to use text as is, if it is a string otherwise rebuild the original message from the data structure.
Let me know what you think and i will update the PR

[joweich]

Agree with your proposal of reconstructing the original message in case the text field is not a native string.
Really happy that you're contributing to this project!🤗

[galatolofederico]

Thank you for sharing this amazing project!

[joweich]

You might have a look at #18 if you want to leverage your expertise in telegram exports even more!