find: fix bolding when replacing inside a previous replacement

Perform the replacement with an unformatted version of the substitute
string, then create a display version with bolding added only after the
matching line (if any) is found.

Updated the corresponding test case. Also renamed some local vars in the
`find` plugin file to make a bit more sense (but not too dramatic).

sopel/builtins/find.py

@@ -170,7 +170,7 @@ def findandreplace(bot, trigger):
 
     # only clean/format the new string if it's non-empty
     if new:
-        new = bold(escape_sequence_pattern.sub(decode_escape, new))
+        new = escape_sequence_pattern.sub(decode_escape, new)
 
     # If g flag is given, replace all. Otherwise, replace once.
     if 'g' in flags:
@@ -183,42 +183,43 @@ def findandreplace(bot, trigger):
     if 'i' in flags:
         regex = re.compile(re.escape(old), re.U | re.I)
 
-        def repl(s):
-            return re.sub(regex, new, s, count == 1)
+        def repl(line, subst):
+            return re.sub(regex, subst, line, count == 1)
     else:
-        def repl(s):
-            return s.replace(old, new, count)
+        def repl(line, subst):
+            return line.replace(old, subst, count)
 
     # Look back through the user's lines in the channel until you find a line
     # where the replacement works
-    new_phrase = None
+    new_line = new_display = None
     for line in history:
         if line.startswith("\x01ACTION"):
             me = True  # /me command
             line = line[8:]
         else:
             me = False
-        replaced = repl(line)
+        replaced = repl(line, new)
         if replaced != line:  # we are done
-            new_phrase = replaced
+            new_line = replaced
+            new_display = repl(line, bold(new))
             break
 
-    if not new_phrase:
+    if not new_line:
         return  # Didn't find anything
 
     # Save the new "edited" message.
     action = (me and '\x01ACTION ') or ''  # If /me message, prepend \x01ACTION
-    history.appendleft(action + new_phrase)  # history is in most-recent-first order
+    history.appendleft(action + new_line)  # history is in most-recent-first order
 
     # output
     if not me:
-        new_phrase = 'meant to say: %s' % new_phrase
+        new_display = 'meant to say: %s' % new_display
     if trigger.group(1):
-        phrase = '%s thinks %s %s' % (trigger.nick, rnick, new_phrase)
+        msg = '%s thinks %s %s' % (trigger.nick, rnick, new_display)
     else:
-        phrase = '%s %s' % (trigger.nick, new_phrase)
+        msg = '%s %s' % (trigger.nick, new_display)
 
-    bot.say(phrase)
+    bot.say(msg)
 
 
 def decode_escape(match):

test/builtins/test_builtins_find.py

@@ -102,8 +102,6 @@ def test_replace_the_replacement(bot, irc, user, channel):
             channel, user.nick, bold('eggs'),
         ),
         "PRIVMSG %s :%s meant to say: %s" % (
-            channel, user.nick, bold(bold('bacon')),
-            # the test is accurate, even though the behavior here (doubled bold
-            # control characters) is less than ideal
+            channel, user.nick, bold('bacon'),
         ),
     )