New sub/person substitution algorithm. #208
Conversation
…tions when the number of subs is too large
src/brain.coffee
Outdated
|
|
||
| # Convert the placeholders back in. | ||
| # Take the original message with no punctuation | ||
| pattern = msg.replace(/[.?,!;:]/, "") |
There was a problem hiding this comment.
This could probably reuse the @master.unicodePunctuation regexp, and that way if UTF-8 supporting users change the set of punctuation symbols to fit their target language it will apply here too.
src/brain.coffee
Outdated
| msg = msg.replace(new RegExp("\x00<#{cap}>\x00", "g"), result) | ||
| # Look for words/phrases until there is no "spaces" in pattern | ||
| while pattern.indexOf(" ") > -1 | ||
| giveup++ |
There was a problem hiding this comment.
At time of writing, this line is breaking the unit tests because the variable name is tries, not giveup; so rename one of the variables to match.
There was a problem hiding this comment.
The variable name is even giveup, but it was undeclared
src/brain.coffee
Outdated
| giveup++ | ||
| # Give up if there are too many substitutions (for safety) | ||
| if giveup >= 50 | ||
| @warn "Too many loops in substitution placeholders!" |
There was a problem hiding this comment.
The warning message should more accurately reflect what it's doing. It looks like your new algorithm doesn't use the substitution placeholders like the old one did (which is good 😄), but warnings like this might say instead "Too many loops when handling substitutions!" or similar.
src/utils.coffee
Outdated
| @@ -154,3 +154,6 @@ exports.isAPromise = (obj) -> | |||
| typeof obj.then is 'function' and | |||
| typeof obj.catch is 'function' and | |||
| typeof obj.finally is 'function' | |||
|
|
|||
| exports.nIndexOf = (string, match, index) -> | |||
| return string.split(match, index).join(match).length | |||
There was a problem hiding this comment.
Can you add documentation to this function? I had to stare at it for a while to figure out what it does. Something like,
##
# int nIndexOf (string input, string delimiter, int wordCount)
#
# This function splits an input string and then takes the first `wordCount` words from it,
# rejoins them and returns the length of that part of the input.
#
# The return value is the index in the original input string where the first N words end.
##The name of the function, nIndexOf isn't very descriptive either, but I can't think of a better name for it.
src/brain.coffee
Outdated
| else | ||
| # Otherwise Look for substitutions in a subpattern | ||
| while subpattern.indexOf(" ") > -1 | ||
| subgiveup++ |
There was a problem hiding this comment.
I didn't see where this variable is declared; this line would give an error if you don't initialize the variable first.
It Makes a lot of less interactions when the number of subs is too large