mdream 0.3.0 → 0.4.0

package/dist/_chunks/readability-BfCjcbbx.mjs

@@ -1,271 +0,0 @@
-import { ELEMENT_NODE, TAG_A, TAG_ADDRESS, TAG_ARTICLE, TAG_ASIDE, TAG_AUDIO, TAG_B, TAG_BLOCKQUOTE, TAG_BODY, TAG_BR, TAG_BUTTON, TAG_CAPTION, TAG_CODE, TAG_DD, TAG_DETAILS, TAG_DIV, TAG_DL, TAG_DT, TAG_EM, TAG_EMBED, TAG_FIELDSET, TAG_FIGCAPTION, TAG_FIGURE, TAG_FOOTER, TAG_FORM, TAG_H1, TAG_H2, TAG_H3, TAG_H4, TAG_H5, TAG_H6, TAG_HEAD, TAG_HEADER, TAG_HR, TAG_HTML, TAG_I, TAG_IFRAME, TAG_IMG, TAG_INPUT, TAG_LI, TAG_MAIN, TAG_META, TAG_NAV, TAG_OBJECT, TAG_OL, TAG_P, TAG_PRE, TAG_SCRIPT, TAG_SECTION, TAG_SELECT, TAG_SPAN, TAG_STRONG, TAG_STYLE, TAG_SUMMARY, TAG_SVG, TAG_TABLE, TAG_TBODY, TAG_TD, TAG_TEXTAREA, TAG_TFOOT, TAG_TH, TAG_THEAD, TAG_TITLE, TAG_TR, TAG_UL, TAG_VIDEO, collectNodeContent, createBufferRegion, createPlugin$1 as createPlugin } from "./plugin-DCJFRZej.mjs";
-
-//#region src/plugins/frontmatter.ts
-/**
-* A plugin that manages frontmatter generation from HTML head elements
-* Extracts metadata from meta tags and title and generates YAML frontmatter
-*/
-function frontmatterPlugin(options = {}) {
-	const additionalFields = options.additionalFields || {};
-	const metaFields = new Set([
-		"description",
-		"keywords",
-		"author",
-		"date",
-		"og:title",
-		"og:description",
-		"twitter:title",
-		"twitter:description",
-		...options.metaFields || []
-	]);
-	const frontmatter = {
-		...additionalFields,
-		meta: {}
-	};
-	let inHead = false;
-	const formatValue = options.formatValue || ((name, value) => {
-		value = value.replace(/"/g, "\\\"");
-		if (value.includes("\n") || value.includes(":") || value.includes("#") || value.includes(" ")) return `"${value}"`;
-		return value;
-	});
-	return createPlugin({
-		onNodeEnter(node) {
-			if (node.tagId === TAG_HEAD) {
-				inHead = true;
-				return;
-			}
-			if (inHead && node.type === ELEMENT_NODE && node.tagId === TAG_TITLE) return;
-			if (inHead && node.type === ELEMENT_NODE && node.tagId === TAG_META) {
-				const elementNode = node;
-				const { name, property, content } = elementNode.attributes || {};
-				const metaName = property || name;
-				if (metaName && content && metaFields.has(metaName)) frontmatter.meta[metaName.includes(":") ? `"${metaName}"` : metaName] = formatValue(metaName, content);
-				return void 0;
-			}
-		},
-		onNodeExit(node, state) {
-			if (node.type === ELEMENT_NODE && node.tagId === TAG_HEAD) {
-				inHead = false;
-				if (Object.keys(frontmatter).length > 0) {
-					const frontmatterContent = generateFrontmatter();
-					collectNodeContent({
-						type: 1,
-						regionId: 0
-					}, frontmatterContent, state);
-				}
-			}
-			return void 0;
-		},
-		processTextNode(node) {
-			if (!inHead) return;
-			const parent = node.parent;
-			if (parent && parent.tagId === TAG_TITLE && node.value) {
-				frontmatter.title = formatValue("title", node.value.trim());
-				return {
-					content: "",
-					skip: true
-				};
-			}
-		}
-	});
-	/**
-	* Generate YAML frontmatter string from collected metadata
-	*/
-	function generateFrontmatter() {
-		if (Object.keys(frontmatter).length === 0) return "";
-		let yamlLines = [];
-		const entries = Object.entries(frontmatter).sort(([a], [b]) => {
-			if (a === "title") return -1;
-			if (b === "title") return 1;
-			if (a === "description") return -1;
-			if (b === "description") return 1;
-			return a.localeCompare(b);
-		});
-		for (const [key, value] of entries) if (key === "meta" && Object.keys(value).length > 0) {
-			yamlLines.push("meta:");
-			const metaEntries = Object.entries(value).sort(([a], [b]) => a.localeCompare(b)).map(([metaKey, metaValue]) => `  ${metaKey}: ${metaValue}`);
-			yamlLines.push(...metaEntries);
-		} else if (key !== "meta" || Object.keys(value).length > 0) yamlLines.push(`${key}: ${value}`);
-		if (Object.keys(frontmatter.meta).length === 0) yamlLines = yamlLines.filter((line) => !line.startsWith("meta:"));
-		return `---\n${yamlLines.join("\n")}\n---\n\n`;
-	}
-}
-
-//#endregion
-//#region src/plugins/readability.ts
-const REGEXPS = {
-	positive: /article|body|content|entry|main|page|post|text|blog|story|recipe|ingredient|instruction|description|docs?|guide|tutorial|reference|manual/i,
-	negative: /ad|banner|combx|comment|disqus|extra|foot|header|menu|meta|nav|promo|related|scroll|share|sidebar|sponsor|social|tags|widget|sitemap|copyright|login|register|subscribe|newsletter|signup|category|author|date|publish|cta|button|apply|trial|likes|views|metrics|stats|breadcrumb|pagination|filter|sort|search/i,
-	commas: /,/g,
-	periodAtEnd: /\.( |$)/,
-	hidden: /hidden|display:\s*none|visibility:\s*hidden/i,
-	advertisement: /^ad-|^ad$|advertisement|sponsor|promo|banner/i,
-	comments: /comment|disqus|replies/i
-};
-const TagScores = {
-	[TAG_ARTICLE]: 15,
-	[TAG_SECTION]: 8,
-	[TAG_MAIN]: 15,
-	[TAG_P]: 5,
-	[TAG_DIV]: 2,
-	[TAG_BLOCKQUOTE]: 5,
-	[TAG_PRE]: 8,
-	[TAG_CODE]: 6,
-	[TAG_IMG]: 3,
-	[TAG_FIGURE]: 4,
-	[TAG_FIGCAPTION]: 3,
-	[TAG_VIDEO]: 3,
-	[TAG_AUDIO]: 3,
-	[TAG_SVG]: 1,
-	[TAG_TABLE]: 0,
-	[TAG_CAPTION]: 2,
-	[TAG_THEAD]: 0,
-	[TAG_TBODY]: 0,
-	[TAG_TFOOT]: 0,
-	[TAG_TR]: -1,
-	[TAG_TH]: -2,
-	[TAG_TD]: 0,
-	[TAG_UL]: -8,
-	[TAG_OL]: -5,
-	[TAG_LI]: -6,
-	[TAG_DL]: 0,
-	[TAG_DT]: 0,
-	[TAG_DD]: 0,
-	[TAG_H1]: 1,
-	[TAG_H2]: 1,
-	[TAG_H3]: 1,
-	[TAG_H4]: 0,
-	[TAG_H5]: 0,
-	[TAG_H6]: 0,
-	[TAG_HEADER]: -15,
-	[TAG_FOOTER]: -25,
-	[TAG_NAV]: -30,
-	[TAG_ASIDE]: -25,
-	[TAG_FORM]: -8,
-	[TAG_BUTTON]: -5,
-	[TAG_INPUT]: -5,
-	[TAG_TEXTAREA]: -5,
-	[TAG_SELECT]: -5,
-	[TAG_FIELDSET]: -5,
-	[TAG_IFRAME]: -3,
-	[TAG_EMBED]: -3,
-	[TAG_OBJECT]: -3,
-	[TAG_A]: -8,
-	[TAG_STRONG]: 1,
-	[TAG_B]: 1,
-	[TAG_EM]: 1,
-	[TAG_I]: 1,
-	[TAG_HR]: 0,
-	[TAG_BR]: 0,
-	[TAG_SPAN]: 0,
-	[TAG_SCRIPT]: -25,
-	[TAG_STYLE]: -25,
-	[TAG_DETAILS]: 2,
-	[TAG_SUMMARY]: 1,
-	[TAG_ADDRESS]: -3
-};
-/**
-* Apply score adjustments based on class and ID names
-*/
-function scoreClassAndId(node) {
-	let scoreAdjustment = 0;
-	if (node.attributes?.class) {
-		const className = node.attributes.class;
-		if (/nav|menu|header|footer|sidebar|ad-|advertisement|banner|promo|cta|button|apply|trial|engagement|sharing|likes|views|metrics|stats|breadcrumb|pagination|filter|sort|search/i.test(className)) scoreAdjustment -= 35;
-		else if (REGEXPS.negative.test(className)) scoreAdjustment -= 15;
-		else if (REGEXPS.positive.test(className)) {
-			scoreAdjustment += 10;
-			if (/docs?|guide|tutorial|reference|manual|article/i.test(className)) scoreAdjustment += 5;
-		}
-	}
-	if (node.attributes?.id) {
-		const id = node.attributes.id;
-		if (/nav|menu|header|footer|sidebar|ad-|advertisement|banner|promo|cta|button|apply|trial|engagement|sharing|likes|views|metrics|stats|breadcrumb|pagination|filter|sort|search/i.test(id)) scoreAdjustment -= 35;
-		else if (REGEXPS.negative.test(id)) scoreAdjustment -= 15;
-		else if (REGEXPS.positive.test(id)) {
-			scoreAdjustment += 10;
-			if (/docs?|guide|tutorial|reference|manual|article/i.test(id)) scoreAdjustment += 5;
-		}
-	}
-	return scoreAdjustment;
-}
-/**
-* Creates a plugin that implements readability.js style heuristics for content quality assessment
-* Controls content inclusion/exclusion using buffer regions
-*/
-function readabilityPlugin() {
-	let inHead = false;
-	return createPlugin({
-		onNodeEnter(node, state) {
-			if (inHead) return;
-			if (!node.context) node.context = {};
-			if (node.tagId === TAG_BODY || node.tagId === TAG_HTML) return;
-			if (node.tagId === TAG_HEAD) {
-				createBufferRegion(node, state, true);
-				inHead = true;
-				return;
-			}
-			const tagScore = TagScores[node.tagId] ?? 0;
-			const classAndIdScore = scoreClassAndId(node);
-			node.context.score = tagScore + classAndIdScore;
-			node.context.tagCount = 1;
-			node.context.linkTextLength = 0;
-			node.context.textLength = 0;
-			const hasStrongNegativePattern = node.name && /nav|header|footer|aside|form|fieldset|button/i.test(node.name) || node.attributes?.class && /nav|menu|header|footer|sidebar|hidden|copyright|ad-|advertisement|banner|promo|related|comment|login|register|subscribe|newsletter|category|meta|tag|cta|button|apply|trial|engagement|sharing|likes|views|metrics|stats|breadcrumb|pagination|filter|sort|search/i.test(node.attributes.class) || node.attributes?.id && /nav|menu|header|footer|sidebar|hidden|copyright|ad-|advertisement|banner|promo|related|comment|login|register|subscribe|newsletter|category|meta|tag|cta|button|apply|trial|engagement|sharing|likes|views|metrics|stats|breadcrumb|pagination|filter|sort|search/i.test(node.attributes.id) || node.attributes?.style && /display:\s*none|visibility:\s*hidden/i.test(node.attributes.style) || node.attributes && Object.keys(node.attributes).some((attr) => attr.startsWith("aria-") && node.attributes[attr] === "true" && /hidden|invisible/i.test(attr));
-			if (hasStrongNegativePattern) createBufferRegion(node, state, false);
-			else if (node.parent && node.parent.context) node.context.score += node.parent.context.score || 0;
-		},
-		processTextNode(node) {
-			if (!node.parent || inHead) return;
-			const textValue = node.value;
-			const len = textValue.length;
-			const commaCount = Math.min(3, (textValue.match(REGEXPS.commas) || []).length);
-			const isInsideLink = !!node.parent.depthMap?.[TAG_A];
-			let parent = node.parent;
-			while (parent) {
-				if (!parent.context) parent.context = {};
-				parent.context.score = (parent.context.score || 0) + commaCount;
-				parent.context.textLength = (parent.context.textLength || 0) + len;
-				if (isInsideLink) parent.context.linkTextLength = (parent.context.linkTextLength || 0) + len;
-				parent = parent.parent;
-			}
-		},
-		onNodeExit(node, state) {
-			if (!node.context) return;
-			if (node.tagId === TAG_BODY || node.tagId === TAG_HTML) return;
-			if (node.tagId === TAG_HEAD) {
-				inHead = false;
-				return;
-			}
-			if (inHead) return;
-			const textLength = node.context.textLength || 0;
-			if (textLength === 0 && node.tagId !== TAG_BODY && !node.childTextNodeIndex) {} else if (textLength > 100) node.context.score += 3;
-			else if (textLength >= 50) node.context.score += 2;
-			else if (textLength >= 25) node.context.score += 1;
-			const linkTextLength = node.context.linkTextLength || 0;
-			if (textLength > 0) {
-				const linkDensity = linkTextLength / textLength;
-				if (linkDensity > .4) if (linkDensity > .6) {
-					node.context.score = node.context.score * .02;
-					if (linkTextLength > 50) node.context.isHighLinkDensity = true;
-				} else node.context.score *= 1 - linkDensity * 2;
-				else if (linkDensity > .2) node.context.score *= 1 - linkDensity * 1;
-			}
-			if (linkTextLength > 0 && textLength > 0) {
-				const linkRatio = linkTextLength / textLength;
-				const hasDocumentationMarkers = node.attributes?.class && /docs?|guide|tutorial|reference|manual|article|content/i.test(node.attributes.class) || node.attributes?.id && /docs?|guide|tutorial|reference|manual|article|content/i.test(node.attributes.id) || node.name && /main|article|section/i.test(node.name);
-				if (linkRatio > .3 && linkTextLength > 30 && !hasDocumentationMarkers) node.context.score -= 10;
-			}
-			const finalScore = node.context.score;
-			if (finalScore <= -12) createBufferRegion(node, state, false);
-			else if (node.context.isHighLinkDensity || linkTextLength > 50 && textLength > 0 && linkTextLength / textLength > .5) createBufferRegion(node, state, false);
-			if (node.tagHandler?.isInline) {
-				const parent = node.parent;
-				if (parent && parent.context) parent.context.score += finalScore - (parent.context.score || 0);
-			}
-		}
-	});
-}
-
-//#endregion
-export { frontmatterPlugin, readabilityPlugin };