Now the last little changes have been implemented, had to modify how we got the user to message since we were previously getting the first user mentioned in the message, whereas we now match against all known users in the guild. This means users actually get messaged by, say, rolebot even if they're not in the channel you're tipping from. Of course you can't directly mention anyone not in the channel for multitip, so that's a non-issue in most use cases.
Also had to ensure certain function calls were sending the right type, because dynamic languages and lack of type safety.
That _should_ be absolutely everything needed to have it fully functional.
All finally operational and tested on MSFT's RVN server, code ported back over after figuring out various JS specific implementation issues. Now all works correctly and doesn't include anyone it shouldn't.
Implements MSFTblockchain's fixes from PR #4 without customisations or version regression.
Now requires Node.js Regex module for testing for user mentions in text with regex, including a fallback if a unit test fails.
Also adds !tipcommands command which gives an up to date list of all commands tipbot accepts with a brief description of each.
The tip message tail about DMing tipbot is now contextualised to give the appropriate !<command> and also points users towards the new !tipcommands command.
Private tip messages are now a bit nicer to read and prv is now a boolean flag to fix some odd errors that were occuring.
Error Messages now delete themselves after a short time for cleanliness.
Help messages are now in Discord Embeds to line up with Wunderbot, though this currently ignores tabs, so help messages don't quite match the requested format. Commands have been made bold to prevent confusion.
Everything should now actually work, but will require testing to be sure.
This is actually not too substantial an addition. it adds 2 additional command exports for tipbot, namely !multitip and !roletip.
!multitip [private] <user>+ <amount>
The more complex new feature, this command will take a list of users, seperated by spaces, and passes through to doTip() if you only list a single user. It tests each word past the ! command and private tag, checking for a user mention using the regex.test() function of the Discord.js USERS_PATTERN, using that to count the number of users mentioned in a row, then takes the first word that is not a match to be the amount. Error checking then returns the appropriate errors to the user, if any. Otherwise it proceeds to send tips to each user individually. this results in messages for every user that receives a tip, which can be messy. It will also message the author once for every user a tip is sent to, if using private mode, this isn't ideal, but would require rewriting the sendLbc() function, which is outside the scope of this commit.
!roletip <role> <amount>
This is a relatively simpler feature, the command taking a single role and an amount, then extracting the userIDs from the role via the Roles.members and GuildMember.user values. It will return seperate errors for a lack of a role in the message and the lack of any users in a role. Like the !multitip command, it will send tips individually to each user, with the same spam of messages either in the channel or the author's private channel.
This one involves a fair amount of reworking of the way functions operate, all checks for being in private or bot sandbox have been removed from functions and that functionality is now applied through privateOrSandboxOnly(), passing the message, channel warning, function to be called and then an array of arguments, sans message, as that's automatically attached at the head as all primary functions require the message anyway. while this makes each case entry a bit longer, it does reduce code reuse and potential for old code being left in, since each function had to previously be updated individually if something happened with the way these tests were carried out.
I've yet to find a way to allow us to set the export description and the help message in one fell swoop, so for now I've manually set the description. The help message itself is now broken down into pairs of subcommand and description, the subcommand will be indented once, while the description will be indented twice.
The way the help function now works is to allow everything to be set at the start, without needing to hunt all the way down to find the duplicate help message. Now everything can be replaced in the main function, though this will require a bit of computation with each request presently, as I don't know if or how the exported functions can access constants defined outside them. This can be optimised later if need be and at the very least shouldn't slow things down very much with the few entries that there are.
Unfortunately changing the way that the help message works does involve parameterising the help message itself for any function that requires it, but this also provides a helpful indicator of what functions might fail and fall back on the help function.
Updated doHelp() to match updated module description in Commit ec48f11 (ec48f114d8 (diff-baad512c11a603dedf1cf7f391e341bd))
Also altered help logic so the help message is always DM'd to user to assist new users who struggle to find how to begin a dm session and to improve general usability.
Further added a self tip check to sendLbc() to prevent duplicate messages being received by the user, so as to make clear that the tip is only being sent once.
Allow for Tipbot to be used privately while still notifying the user that has received the tip, via private message.
Updates the Help message for new usage instructions, fixes misspelling of 'address' and adds a Key for the [ ] and < > syntax, as some users may not understand their usage.
Reworks the subcommand parsing by using a more compact Switch Function. Necessitates the use of 'break;' but is far more compact and legible, presumed the old if else block is from early development with fewer features available. Switch statement should allow for easier expansion.
Modifies doTip() to avoid requiring code duplication, includes two new variables used to indicate that the tip should be done in privacy mode and where the tip value is located in the 'words' array.
Significant change to sendLbc and it's function call, requirement for private message for recipient required User object, thus the function now requires the actual GuildMember object, not only the ID. While this marginally increases overhead, it should also allow access to the send() function for that Guild Member, allowing for private message to be sent. Further the addition of a privacy flag (0 for non-private, 1 for private) was added so that the function can determine whether to send a public or private response. this required updating references to the Guild Member's id to 'member.id', but otherwise operation remains the same until the transaction is completed, following which it will check the privacy flag and if it is set, will privately send the successful tip message to both author and recipient, otherwise it will simply reply via the normal '.reply()' function.
