Improve Cooldown UX (#2412)

This improves the feedback given on `command_on_cool_down`, as well as automatically handling cooldowns under 1 second for the user.

redbot/core/events.py

@@ -4,6 +4,7 @@ import codecs
 import datetime
 import logging
 import traceback
+import asyncio
 from datetime import timedelta
 
 import aiohttp
@@ -15,7 +16,7 @@ from pkg_resources import DistributionNotFound
 from .. import __version__ as red_version, version_info as red_version_info, VersionInfo
 from . import commands
 from .data_manager import storage_type
-from .utils.chat_formatting import inline, bordered, format_perms_list
+from .utils.chat_formatting import inline, bordered, format_perms_list, humanize_timedelta
 from .utils import fuzzy_command_search, format_fuzzy_results
 
 log = logging.getLogger("red")
@@ -213,8 +214,22 @@ def init_events(bot, cli_flags):
         elif isinstance(error, commands.NoPrivateMessage):
             await ctx.send("That command is not available in DMs.")
         elif isinstance(error, commands.CommandOnCooldown):
+            if error.retry_after < 1:
+                async with ctx.typing():
+                    # the sleep here is so that commands using this for ratelimit purposes
+                    # are not made more lenient than intended, while still being
+                    # more convienient for the user than redoing it less than a second later.
+                    await asyncio.sleep(error.retry_after)
+                    await ctx.bot.invoke(ctx)
+                    # done this way so checks still occur if there are other
+                    # failures possible than just cooldown.
+                    # do not change to ctx.reinvoke()
+                    return
+
             await ctx.send(
-                "This command is on cooldown. Try again in {:.2f}s".format(error.retry_after)
+                "This command is on cooldown. Try again in {}.".format(
+                    humanize_timedelta(seconds=error.retry_after)
+                )
             )
         else:
             log.exception(type(error).__name__, exc_info=error)