import * as t from './tokenTypes';

const unescapable = {
    [t.tab]: true,
    [t.newline]: true,
    [t.cr]: true,
    [t.feed]: true,
};
const wordDelimiters = {
    [t.space]: true,
    [t.tab]: true,
    [t.newline]: true,
    [t.cr]: true,
    [t.feed]: true,

    [t.ampersand]: true,
    [t.asterisk]: true,
    [t.bang]: true,
    [t.comma]: true,
    [t.colon]: true,
    [t.semicolon]: true,
    [t.openParenthesis]: true,
    [t.closeParenthesis]: true,
    [t.openSquare]: true,
    [t.closeSquare]: true,
    [t.singleQuote]: true,
    [t.doubleQuote]: true,
    [t.plus]: true,
    [t.pipe]: true,
    [t.tilde]: true,
    [t.greaterThan]: true,
    [t.equals]: true,
    [t.dollar]: true,
    [t.caret]: true,
    [t.slash]: true,
};


const hex = {};
const hexChars = "0123456789abcdefABCDEF";
for (let i = 0; i < hexChars.length; i++) {
    hex[hexChars.charCodeAt(i)] = true;
}

/**
 *  Returns the last index of the bar css word
 * @param {string} css The string in which the word begins
 * @param {number} start The index into the string where word's first letter occurs
 */
function consumeWord (css, start) {
    let next = start;
    let code;
    do {
        code = css.charCodeAt(next);
        if (wordDelimiters[code]) {
            return next - 1;
        } else if (code === t.backslash) {
            next = consumeEscape(css, next) + 1;
        } else {
            // All other characters are part of the word
            next++;
        }
    } while (next < css.length);
    return next - 1;
}

/**
 *  Returns the last index of the escape sequence
 * @param {string} css The string in which the sequence begins
 * @param {number} start The index into the string where escape character (`\`) occurs.
 */
function consumeEscape (css, start) {
    let next = start;
    let code = css.charCodeAt(next + 1);
    if (unescapable[code]) {
        // just consume the escape char
    } else if (hex[code]) {
        let hexDigits = 0;
        // consume up to 6 hex chars
        do {
            next++;
            hexDigits++;
            code = css.charCodeAt(next + 1);
        } while (hex[code] && hexDigits < 6);
        // if fewer than 6 hex chars, a trailing space ends the escape
        if (hexDigits < 6 && code === t.space) {
            next++;
        }
    } else {
        // the next char is part of the current word
        next++;
    }
    return next;
}

export const FIELDS = {
    TYPE: 0,
    START_LINE: 1,
    START_COL: 2,
    END_LINE: 3,
    END_COL: 4,
    START_POS: 5,
    END_POS: 6,
};

export default function tokenize (input) {
    const tokens   = [];
    let css        = input.css.valueOf();
    let {length}   = css;
    let offset     = -1;
    let line       =  1;
    let start      =  0;
    let end        =  0;

    let code,
        content,
        endColumn,
        endLine,
        escaped,
        escapePos,
        last,
        lines,
        next,
        nextLine,
        nextOffset,
        quote,
        tokenType;

    function unclosed (what, fix) {
        if ( input.safe ) { // fyi: this is never set to true.
            css += fix;
            next = css.length - 1;
        } else {
            throw input.error('Unclosed ' + what, line, start - offset, start);
        }
    }

    while ( start < length ) {
        code = css.charCodeAt(start);

        if ( code === t.newline ) {
            offset = start;
            line  += 1;
        }

        switch ( code ) {
        case t.space:
        case t.tab:
        case t.newline:
        case t.cr:
        case t.feed:
            next = start;
            do {
                next += 1;
                code = css.charCodeAt(next);
                if ( code === t.newline ) {
                    offset = next;
                    line  += 1;
                }
            } while (
                code === t.space   ||
                code === t.newline ||
                code === t.tab     ||
                code === t.cr      ||
                code === t.feed
            );

            tokenType = t.space;
            endLine = line;
            endColumn = next - offset - 1;
            end = next;
            break;

        case t.plus:
        case t.greaterThan:
        case t.tilde:
        case t.pipe:
            next = start;
            do {
                next += 1;
                code = css.charCodeAt(next);
            } while (
                code === t.plus        ||
                code === t.greaterThan ||
                code === t.tilde       ||
                code === t.pipe
            );

            tokenType = t.combinator;
            endLine = line;
            endColumn = start - offset;
            end = next;
            break;

        // Consume these characters as single tokens.
        case t.asterisk:
        case t.ampersand:
        case t.bang:
        case t.comma:
        case t.equals:
        case t.dollar:
        case t.caret:
        case t.openSquare:
        case t.closeSquare:
        case t.colon:
        case t.semicolon:
        case t.openParenthesis:
        case t.closeParenthesis:
            next = start;
            tokenType = code;
            endLine = line;
            endColumn = start - offset;
            end = next + 1;
            break;

        case t.singleQuote:
        case t.doubleQuote:
            quote = code === t.singleQuote ? "'" : '"';
            next  = start;
            do {
                escaped = false;
                next    = css.indexOf(quote, next + 1);
                if ( next === -1 ) {
                    unclosed('quote', quote);
                }
                escapePos = next;
                while ( css.charCodeAt(escapePos - 1) === t.backslash ) {
                    escapePos -= 1;
                    escaped = !escaped;
                }
            } while ( escaped );

            tokenType = t.str;
            endLine = line;
            endColumn = start - offset;
            end = next + 1;
            break;

        default:
            if ( code === t.slash && css.charCodeAt(start + 1) === t.asterisk ) {
                next = css.indexOf('*/', start + 2) + 1;
                if ( next === 0 ) {
                    unclosed('comment', '*/');
                }

                content = css.slice(start, next + 1);
                lines   = content.split('\n');
                last    = lines.length - 1;

                if ( last > 0 ) {
                    nextLine   = line + last;
                    nextOffset = next - lines[last].length;
                } else {
                    nextLine   = line;
                    nextOffset = offset;
                }

                tokenType = t.comment;
                line   = nextLine;
                endLine = nextLine;
                endColumn = next - nextOffset;
            } else if (code === t.slash) {
                next = start;
                tokenType = code;
                endLine = line;
                endColumn = start - offset;
                end = next + 1;
            } else {
                next = consumeWord(css, start);
                tokenType = t.word;
                endLine = line;
                endColumn = next - offset;
            }

            end = next + 1;
            break;
        }

        // Ensure that the token structure remains consistent
        tokens.push([
            tokenType,              // [0] Token type
            line,                   // [1] Starting line
            start - offset,         // [2] Starting column
            endLine,                // [3] Ending line
            endColumn,              // [4] Ending column
            start,                  // [5] Start position / Source index
            end,                    // [6] End position
        ]);

        // Reset offset for the next token
        if (nextOffset) {
            offset = nextOffset;
            nextOffset = null;
        }

        start = end;
    }

    return tokens;
}