using System;
using System.Linq;
using System.Collections.Generic;
using System.Text.RegularExpressions;
using Sentimentinator.Interfaces;

namespace Sentimentinator
{
    class Tokenizer : ITokenizer
    {
        // The regex is carefully constructed as is, so if change needed: be very wary of side-effects.
        private string defaultSentenceSplit = @"(?<! adm| adr| afd| alm| ang| bl|\.a| bla| opg| org| dr| dvs| eks| fx| fhv| hhv| i\.e| ifl| ifb| hr| frk| jf| jvf|www| m| mio| ml| mr| mrs| pga)\.(?!\.|[0-9]|com|eks|dk|e\.|\?|\!)|\?(?!\?|\.|\!)|\!(?!\!|\.|\?)";

        //default regex constructor
        public Tokenizer()
        {
        }

        //constructor that makes user able to change default sentence splitter with own regex
        public Tokenizer(string regexSentenceSplit)
        {
           defaultSentenceSplit = regexSentenceSplit;
        }

        /*
         * Takes a list of raw comments and converts them into a list of TokenizedComments
         * Param: List of raw comments
         * Returns: List<TokenizedComment>
        */

        public List<TokenizedComment> TokenizeComments(List<string> comments)
        {
            var tokenizedComments = new List<TokenizedComment>();

            foreach (string comment in comments) {
                tokenizedComments.Add(TokenizeComment(comment));
            }

            return tokenizedComments;
        }

        /*
            Takes a raw comment and splits it into sentences, which are then tokenized
            Param: A raw comment
            Returns: TokenizedComment
        */

        public TokenizedComment TokenizeComment(string comment)
        {
            var sentences = SplitComment(comment);
            var tokenComment = new List<List<string>>();

            foreach (string sentence in sentences){
                var tokens = TokenizeSentence(sentence);
                tokenComment.Add(tokens);
            }

            return new TokenizedComment(tokenComment);
        }

        /*
         * Splits a sentence into tokens and filters out empty
        */
        private List<string> TokenizeSentence(string sentence)
        {
            var tokens = new List<string>(sentence.Split(" "));
            tokens = tokens.Where(x => x != "").ToList<string>();

            return tokens;

        }

        /*
         Takes a string and seperates at [.?!] and tries to avoid splitting on popular abbreviations
         */
        private List<string> SplitComment(string s)
        {
            var list = new List<string>();

            //replaces newlines with a punct aswell as adds a punct if comment does not have end punct.
            s = s.Replace(Environment.NewLine, ".").ToLower() + ".";

            var sentences = Regex.Split(s, @defaultSentenceSplit);

            foreach (string sentence in sentences)
                list.Add(RemoveNonLetters(sentence).Trim());

            return list;
        }

        /*
         *  Removes non-letters
         *  Side-effect: Removes !? and emojies
        */

        private string RemoveNonLetters(string s)
        {
            var rgx = new Regex(@"[^a-zæøå\s:]");

            s = rgx.Replace(s,"");

            return s;
        }

    }
}