htmlnano/dist/_modules/removeComments.mjs

Modular HTML minifier, built on top of the PostHTML

import { isComment, isConditionalComment } from '../helpers.mjs';

const MATCH_EXCERPT_REGEXP = /^\s*more\b/i;
const MATCH_NOINDEX_REGEXP = /^\s*\/?\s*noindex\s*$/i;
const MATCH_SSE_REGEXP = /^\s*\/?\s*sse\s*$/i;
/** Removes HTML comments */ const mod = {
    onNode (_, removeType) {
        removeType = normalizeRemoveType(removeType);
        return (node)=>{
            if (isCommentToRemove(node, removeType)) {
                return '';
            }
            return node;
        };
    },
    onContent (_, removeType) {
        removeType = normalizeRemoveType(removeType);
        return (contents)=>{
            return contents.filter((content)=>!isCommentToRemove(content, removeType));
        };
    }
};
function isCommentToRemove(text, removeType) {
    if (typeof text !== 'string') {
        return false;
    }
    if (!isComment(text)) {
        // Not HTML comment
        return false;
    }
    if (removeType === 'safe') {
        const commentBody = getCommentBody(text);
        // Preserve license comments "<!--! ... -->", the HTML analogue of the
        // "/*! ... */" comments kept by Terser/cssnano.
        if (commentBody && commentBody.startsWith('!')) {
            return false;
        }
        const isNoindex = commentBody ? MATCH_NOINDEX_REGEXP.test(commentBody) : false;
        // Don't remove noindex comments. It was used by some search engines in the past.
        if (isNoindex) {
            return false;
        }
        const isServerSideExclude = commentBody ? MATCH_SSE_REGEXP.test(commentBody) : false;
        // Don't remove sse comments.
        // See: https://support.cloudflare.com/hc/en-us/articles/200170036-What-does-Server-Side-Excludes-SSE-do-
        if (isServerSideExclude) {
            return false;
        }
        // https://en.wikipedia.org/wiki/Conditional_comment
        if (isConditionalComment(text)) {
            return false;
        }
        // Hexo: https://hexo.io/docs/tag-plugins#Post-Excerpt
        // Hugo: https://gohugo.io/content-management/summaries/#manual-summary-splitting
        // WordPress: https://wordpress.com/support/wordpress-editor/blocks/more-block/2/
        // Jekyll: https://jekyllrb.com/docs/posts/#post-excerpts
        const isCMSExcerptComment = commentBody ? MATCH_EXCERPT_REGEXP.test(commentBody) : false;
        if (isCMSExcerptComment) {
            return false;
        }
    }
    if (isMatcher(removeType)) {
        return isMatch(text, removeType);
    }
    return true;
}
function isMatch(input, matcher) {
    if (matcher instanceof RegExp) {
        return matcher.test(input);
    }
    if (typeof matcher === 'function') {
        // eslint-disable-next-line @typescript-eslint/no-unsafe-call -- typescript incorrectly infers Partialled type
        return !!matcher(input);
    }
    return false;
}
function normalizeRemoveType(removeType) {
    if (removeType === 'all' || removeType === 'safe' || isMatcher(removeType)) {
        return removeType;
    }
    if (typeof removeType === 'string') {
        const regexp = parseRegexString(removeType);
        if (regexp) {
            return regexp;
        }
    }
    return 'safe';
}
function isMatcher(matcher) {
    if (matcher instanceof RegExp || typeof matcher === 'function') {
        return true;
    }
    return false;
}
function parseRegexString(value) {
    const literalMatch = value.match(/^\/([\s\S]+)\/([gimsuy]*)$/);
    if (literalMatch) {
        return tryCreateRegExp(literalMatch[1], literalMatch[2]);
    }
    return tryCreateRegExp(value);
}
function tryCreateRegExp(pattern, flags) {
    try {
        return new RegExp(pattern, flags);
    } catch  {
        return null;
    }
}
function getCommentBody(text) {
    const trimmed = text.trim();
    const match = trimmed.match(/^<!--([\s\S]*?)-->$/);
    if (!match) {
        return null;
    }
    return match[1];
}

export { mod as default };