# Introduction

Street talk can be really difficult to understand, in particular to programmers, who aren't known to be very streetwise.

It is your job to create an interpreter to help us all survive in the urban environment.

# Challenge

Given an English sentence as input, create a program or a function that determines whether the outcome of the sentence is positive or negative.

The sentence will contain 0 to 2 negative words. As any programmer knows, a double negative results in a positive. Therefore, your code must output or return a truthy/falsey value according to the following rule:

No negative words  -> truthy
One negative word  -> falsey
Two negative words -> truthy


The list of negative words:

• no, not, none
• Anything ending in n't
• never, neither, nor
• nobody, nothing, nowhere

There is one edge case. Whenever a sentence begins with No,, that word isn't treated as a negative word when determining the result (it does count towards the number of negative words so there can be just one more).

The sentence will follow basic grammar rules (capitalization, punctuation) and will only contain words that can be found from a dictionary (luckily, this doesn't invalidate the question title). The sentence won't contain any proper nouns (sorry, Dr. No, you're out).

# Test cases

Truthy:

Yes.
It's noon.
Hello, World!
What is this?
Ain't no thang!
Never say never.
No, it's noon now.
Neither me nor you.
I didn't do nothing!
A non-alcoholic drink.
I can't get no satisfaction.
All your base are belong to us.


Falsey:

No.
No, no!
Not today.
Neither am I.
Don't do that!
That's no moon!
And none survived.
Is this not my car?
No man is an island.
Nosebleeds are no fun.
Nothing compares to you.
That's a no, I'm afraid.
No, I am not your mother.


The irony here, of course, is that some of these should be interpreted differently. But hey, you can't fault the speaker for not conforming to our logic.

# Rules

Standard loopholes are forbidden. This is , so be concise!

• Nobody ain't never been neither nowhere nor nothing. – Magic Octopus Urn Nov 22 '17 at 19:01
• @MagicOctopusUrn: You can lose the been for a 100% negative sentence! – Antti29 Nov 22 '17 at 19:36

## Retina, 63 bytes

No,

Mi\bn(e(ith|v)er|o(|body|ne|r|t|thing|where))\b|n't\b
0|2


Try it online!

### Explanation

No,



Remove No, from the input. Due to the capitalisation rules, this can only appear at the beginning of the input, so we don't need an explicit ^.

Mi\bn(e(ith|v)er|o(|body|ne|r|t|thing|where))\b|n't\b


Count the number of matches of the case-insensitive regex after the . It just matches all the relevant words, where I've extracted common prefixes/suffixes with the alternatives.

0|2


Count 0 or 2s, so we turn even counts into 1 and odd counts into 0.

• do you do the common letter extraction by hand or use a program that finds the optimal solution for you? – Jonah Nov 21 '17 at 14:02
• @Jonah I did that by hand. There are tools for automated regex metagolf, but they usually take two lists, one to match and one to fail, and generate a regex for that. I'm not aware of any tool that generates an optimal regex to match a particular set of substrings in a larger string. – Martin Ender Nov 21 '17 at 14:08
• could make an interesting challenge... – Jonah Nov 21 '17 at 14:15
• You should be able to assume that n't doesn't need \b after it, since the words have to come from a dictionary. Also, I had the same thing, but I didn't have the meat of the answer as concisely, using a couple more bytes. – mbomb007 Nov 21 '17 at 16:50

# Bash, 11510799989795 85 bytes

Uses packages Core Utilities (for wc) and grep. Assume the sentence is given via Standard Input. History expansion is disabled by set +o histexpand.

((~grep -Pio "(?!^no,)\b(no(|t|r|ne|body|thing|where)|ne(v|ith)er|.*n't)\b"|wc -l%2))


Check the result: In Bash 0 is for true, 1 is for false

### How does it work?

((                       )) # Logical evaluation: non-zero to TRUE, zero to FALSE
~                    %2   # C-style arithmetic: Bit-Negate and Modulus 2
$( ) # Output of the program chain grep -Pio "regex" # PCRE match, ignore case, output matching part one-per-line | wc -l # Pipe to wc and count number of lines  18 bytes (115 to 99) saved by inspiration from Qwertiy's answer and Martin Ender's answer. 1 byte thanks to Nahuel Fouilleul. • regex is not correct : matches noon and not That's a no, I'm afraid. – Nahuel Fouilleul Nov 21 '17 at 12:51 • @NahuelFouilleul Fixed. – iBug Nov 21 '17 at 12:54 • to check: tio however couldn't paste tests because comment length limit – Nahuel Fouilleul Nov 21 '17 at 12:58 • this gives right results ((~$(grep -Pio "(?!^no,)\b(no(|t|r|ne|body|thing|where)|ne(v|ith)er)\b|.*n't\b"|wc -l)%2)) – Nahuel Fouilleul Nov 21 '17 at 13:00
• back quotes instead of $(..) save 1 byte – Nahuel Fouilleul Nov 21 '17 at 13:02 # Javascript ES6, 8987 86 chars s=>s.match(/(?!^no,)\bn(o(|t|r|ne|body|thing|where)|e(v|ith)er)\b|n't\b|$/ig).length&1


f=s=>s.match(/(?!^no,)\bn(o(|t|r|ne|body|thing|where)|e(v|ith)er)\b|n't\b|$/ig).length&1 console.log(Yes. It's noon. Hello, World! Never say never. Ain't no thang! No, it's noon now. Neither me nor you. I didn't do nothing! No, I am your father. A non-alcoholic drink. I can't get no satisfaction. All your base are belong to us..split .every(f)) console.log(No. No, no! Not today. Neither am I. Don't do that! That's no moon! And none survived. No man is an island. Nosebleeds are no fun. Nothing compares to you. That's a no, I'm afraid. No, I am not your mother..split .every(s=>!f(s))) ## Perl 5, 74 bytes 73 bytes code + 1 for -p. s/No,//;$_=!(s/(\bn(o(r|t|ne|body|thing|where)?|e(v|ith)er)|n't)\b//gi%2)
`

Try it online!