robot: allow messages to be copypasta but not learned

Fixes #72.

Author: zephyrtronium

channel/channel.go

@@ -18,8 +18,10 @@ type Channel struct {
 	// Learn and Send are the channel tags.
 	Learn, Send string
 	// Block is a regex that matches messages which should not be used for
-	// learning.
+	// learning or copypasta.
 	Block *regexp.Regexp
+	// Meme is a regex that matches messages which bypass Block only for copypasta.
+	Meme *regexp.Regexp
 	// Responses is the probability that a received message will trigger a
 	// random response.
 	Responses float64

config.go

@@ -262,27 +262,35 @@ func (robo *Robot) InitTwitchUsers(ctx context.Context, owner *Privilege, global
 	return nil
 }
 
+func mergere(global, ch string) (*regexp.Regexp, error) {
+	var re string
+	switch {
+	case global != "" && ch != "":
+		re = "(" + global + ")|(" + ch + ")"
+	case global != "":
+		re = global
+	case ch != "":
+		re = ch
+	default:
+		re = "$^"
+	}
+	return regexp.Compile(re)
+}
+
 // SetTwitchChannels initializes Twitch channel configuration.
 // It must be called after SetTMI.
 func (robo *Robot) SetTwitchChannels(ctx context.Context, global Global, channels map[string]*ChannelCfg) error {
 	// TODO(zeph): we can convert this to a SetChannels, where it just adds the
 	// channels for any given service
 	for nm, ch := range channels {
-		var re string
-		switch {
-		case global.Block != "" && ch.Block != "":
-			re = "(" + global.Block + ")|(" + ch.Block + ")"
-		case global.Block != "":
-			re = global.Block
-		case ch.Block != "":
-			re = ch.Block
-		default:
-			re = "$^"
-		}
-		blk, err := regexp.Compile(re)
+		blk, err := mergere(global.Block, ch.Block)
 		if err != nil {
 			return fmt.Errorf("bad global or channel block expression for twitch.%s: %w", nm, err)
 		}
+		meme, err := mergere(global.Meme, ch.Meme)
+		if err != nil {
+			return fmt.Errorf("bad global or channel meme expression for twitch.%s: %w", nm, err)
+		}
 		emotes := pick.New(pick.FromMap(mergemaps(global.Emotes, ch.Emotes)))
 		effects := pick.New(pick.FromMap(mergemaps(global.Effects, ch.Effects)))
 		ign, mod := make(map[string]bool), make(map[string]bool)
@@ -308,6 +316,7 @@ func (robo *Robot) SetTwitchChannels(ctx context.Context, global Global, channel
 				Learn:     ch.Learn,
 				Send:      ch.Send,
 				Block:     blk,
+				Meme:      meme,
 				Responses: ch.Responses,
 				Rate:      rate.NewLimiter(rate.Every(fseconds(ch.Rate.Every)), ch.Rate.Num),
 				Ignore:    ign,
@@ -493,6 +502,9 @@ type ChannelCfg struct {
 	Rate Rate `toml:"rate"`
 	// Copypasta is the configuration for copypasta.
 	Copypasta Copypasta `toml:"copypasta"`
+	// Meme is a regular expression of messages to allow to be copypasta even
+	// if matched by this channel's or the global Block.
+	Meme string `toml:"meme"`
 	// Emotes is the emotes and their weights for the channel.
 	Emotes map[string]int `toml:"emotes"`
 	// Effects is the effects and their weights for the channel.
@@ -505,6 +517,9 @@ type ChannelCfg struct {
 type Global struct {
 	// Block is a regular expression of messages to ignore everywhere.
 	Block string `toml:"block"`
+	// Meme is a regular expression of messages to allow to be copypasta even
+	// if matched by Block.
+	Meme string `toml:"meme"`
 	// Emotes is the emotes and their weights to use everywhere.
 	Emotes map[string]int `toml:"emotes"`
 	// Effects is the effects and their weights to use everywhere.

config_test.go

@@ -55,6 +55,7 @@ func TestExampleConfig(t *testing.T) {
 	eqcase(t, "Twitch[`bocchi`].Rate.Num", cfg.Twitch[`bocchi`].Rate.Num, 2)
 	eqcase(t, "Twitch[`bocchi`].Copypasta.Need", cfg.Twitch[`bocchi`].Copypasta.Need, 2)
 	eqcase(t, "Twitch[`bocchi`].Copypasta.Within", cfg.Twitch[`bocchi`].Copypasta.Within, 30)
+	eqcase(t, "Twitch[`bocchi`].Meme", cfg.Twitch[`bocchi`].Meme, `^\S*$`)
 	eqcase(t, "Twitch[`bocchi`].Privileges[0].Name", cfg.Twitch[`bocchi`].Privileges[0].Name, `zephyrtronium`)
 	eqcase(t, "Twitch[`bocchi`].Privileges[0].Level", cfg.Twitch[`bocchi`].Privileges[0].Level, `moderator`)
 	eqcase(t, "Twitch[`bocchi`].Emotes[`btw`]", cfg.Twitch[`bocchi`].Emotes[`btw make sure to stretch, hydrate, and take care of yourself <3`], 1)

example.toml

@@ -43,9 +43,11 @@ listen = ':4959'
 
 # global includes chat settings that apply to all channels.
 [global]
-# block is a regex that blocks messages from being learned in any channel.
+# block is a regex that blocks messages from being learned or copypastad in any channel.
 # Unlike most string options, it is not expanded with environment variables.
 block = '(?i)bad\s+stuff[^$x]'
+# meme is a regex that overrides block for copypasta only.
+meme = '^\S*$'
 
 # global.emotes is a table of emotes to use in every channel along with their
 # relative weights.
@@ -100,8 +102,9 @@ learn = 'bocchi'
 send = 'bocchi'
 # block is a regex that blocks messages from being learned in this channel. Any
 # message containing text matching this or the global block regex is not used
-# for learning. Unlike most string options, it is not expanded with environment
-# variables.
+# for learning. block also prevents a message from contributing to copypasta
+# unless the message additionally matches meme, below.
+# Unlike most string options, it is not expanded with environment variables.
 block = '(?i)cucumber[^$x]'
 # responses is the probability of generating a random message when a
 # non-command message is received.
@@ -110,6 +113,8 @@ responses = 0.02
 rate = { every = 10.1, num = 2 }
 # copypasta is the configuration of copypastaing.
 copypasta = { need = 2, within = 30 }
+# meme overrides block for copypasta only.
+meme = '^\S*$'
 # Access levels for users.
 # Each entry must have a name or ID and a level. If both a name and ID are
 # given, the name is ignored.

privmsg.go

@@ -39,6 +39,10 @@ func (robo *Robot) tmiMessage(ctx context.Context, group *errgroup.Group, send c
 			log.InfoContext(ctx, "message from ignored user")
 			return
 		}
+		if ch.Block.MatchString(m.Text) && !ch.Meme.MatchString(m.Text) {
+			log.InfoContext(ctx, "blocked message", slog.String("text", m.Text), slog.Bool("meme", false))
+			return
+		}
 		if cmd, ok := parseCommand(robo.tmi.name, m.Text); ok {
 			robo.command(ctx, log, ch, m, from, cmd)
 			return
@@ -239,7 +243,7 @@ func (robo *Robot) learn(ctx context.Context, log *slog.Logger, ch *channel.Chan
 		return
 	}
 	if ch.Block.MatchString(msg.Text) {
-		log.InfoContext(ctx, "blocked message", slog.String("text", msg.Text))
+		log.InfoContext(ctx, "blocked message", slog.String("text", msg.Text), slog.Bool("meme", true))
 		return
 	}
 	if ch.Learn == "" {