Bsp.:
Michi spielt oracle ---> "hat die Flexibilität einer Tomate bewiesen"
Cormo spielt Jakiro --> " - üüüüberraaaschung - " + Punchline
Cormo spielt Ogre ---> " ist doch auch nur nen Jakiro "

I feel for doing delicate little changes there should be a straightforward way to create debugging matches to use methods on...

After entering a valid voiceline command, the command message is deleted, however the actual voiceline is not actually delivered. Other behaviour such as the help messages are not affected.

My theory is this has todo with the message deletion...

evtl in Abhängigkeit der Spielleistung
Bsp.: Omniknight: "hat die Gegner in den Boden gehämmert"
Lion: "mehr gefingert als im Puff"
Riki: "hat nochmal versucht, cloak and Dagger zu lernen"
"hat verschwindent impact gehabt"
"die Gegner haben ihren Tod nie kommen sehen"

The DotA2 wiki has migrated to liquipedia (Official announcement).

Retrieving voicelines works by finding the appropriate voiceline on the wiki page of the hero in question, scanning the html of that page for the voiceline text, and then delivering a link to the corresponding audio file. With the new host and possibly changed html layout, changes will be necessary to continue this feature.

Hard-coded host:

Indiviual hero url:

CSS selector for finding voiceline text:

Wrong section

The punline is currently in section for meme constant "1" but describes a positive outcome. It should probably go in section "2".

Ends in period

This doesn't match with how the punlines are integrated into the message, as currently punlines are not supposed to be separate sentences.

Missing capitalization

Start of second sentence is not capitalized.

einfach eine besondere Erwähnung bei einer Rampage!
Wie z.B.:
" hat das ganze gegnerische Team im alleingang in Grund und Boden gestampft! Warum ist er nur Crusader?"
" hat allein alle fünf Gegner zu Kleinholz verarbeitet!"

The method for determining the hero response url fails if the hero name contains a preposition (e.g. Keeper of the Light, Queen of Pain). The problem is that the hero response URL is case sensitive, and with the current method the prepositions are captialized when they shouldn't be.

When the text tag being matched contains a link, the heuristic for getting the voiceline no longer works as the text is broken up by a link element.

Reprodicible example

/voiceline Rubick: Oh. Shitty wizard! (Link)

--> No voiceline will be found.

Proposed solution

Use all text in the tag in question for matching. Slight performance penalty because we have to potentially do this for each voiceline tag of a hero, but necessary I think.

The voiceline command attempts to send a message of the initiating user's telegram username in order to let others know who came up with the masterpiece of comedic timing in question while not polluting the chat with ugly commands.

However the update.message.from_user.username field may be empty when a user has no user name (or possibly is choosing to keep it private). In that case update.message.from_user.username is None, and the operation fails, aborting the whole voiceline operation.

Proposed solution

Have a hierarchy of fallback options for attributing the voiceline, so that if one is unavailable, we can still have the attribution.

I'd go with: username > first name (maybe last name, if available) > User ID > "Unknown User".

Obviously I don't know which of these are available when, so I need to double check the docs. If I did, this problem wouldn't exist.

Für DoDos oder DOMiNas der einem coole meme strats vorschlägt. Die Captains schreibt ihm dann privat und er antwortet mit entsprechender Strategie oder halt öffentlich in der SiliGruppe. Der Bot müsste halt dann auch darauf achten kein hero doppelt zu vergeben oder die strat ein bisschen aufzuweichen

update heroes.json to include Muerta