mdream 0.1.1

package/dist/shared/mdream.C8ruysN5.mjs

@@ -0,0 +1,291 @@
+import { T as TAG_BODY, a as TAG_HTML, b as TAG_HEAD, c as createBufferRegion, d as TAG_A, e as TAG_ADDRESS, f as TAG_SUMMARY, g as TAG_DETAILS, h as TAG_STYLE, i as TAG_SCRIPT, j as TAG_SPAN, k as TAG_BR, l as TAG_HR, m as TAG_I, n as TAG_EM, o as TAG_B, p as TAG_STRONG, q as TAG_OBJECT, r as TAG_EMBED, s as TAG_IFRAME, t as TAG_FIELDSET, u as TAG_SELECT, v as TAG_TEXTAREA, w as TAG_INPUT, x as TAG_BUTTON, y as TAG_FORM, z as TAG_ASIDE, A as TAG_NAV, B as TAG_FOOTER, C as TAG_HEADER, D as TAG_H6, E as TAG_H5, F as TAG_H4, G as TAG_H3, H as TAG_H2, I as TAG_H1, J as TAG_DD, K as TAG_DT, L as TAG_DL, M as TAG_LI, N as TAG_OL, O as TAG_UL, P as TAG_TD, Q as TAG_TH, R as TAG_TR, S as TAG_TFOOT, U as TAG_TBODY, V as TAG_THEAD, W as TAG_CAPTION, X as TAG_TABLE, Y as TAG_SVG, Z as TAG_AUDIO, _ as TAG_VIDEO, $ as TAG_FIGCAPTION, a0 as TAG_FIGURE, a1 as TAG_IMG, a2 as TAG_CODE, a3 as TAG_PRE, a4 as TAG_BLOCKQUOTE, a5 as TAG_DIV, a6 as TAG_P, a7 as TAG_MAIN, a8 as TAG_SECTION, a9 as TAG_ARTICLE } from './mdream.-hdaPj9a.mjs';
+import { c as createPlugin } from './mdream.cpEmpxyh.mjs';
+
+const REGEXPS = {
+  // Positive patterns that suggest high-quality content
+  positive: /article|body|content|entry|main|page|post|text|blog|story|recipe|ingredient|instruction|description|docs?|guide|tutorial|reference|manual/i,
+  // Negative patterns that suggest low-quality content
+  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,
+  // Used for counting commas to determine complexity
+  commas: /,/g};
+const TagScores = {
+  // Main structural elements
+  [TAG_ARTICLE]: 15,
+  // Explicit content container, highest confidence
+  [TAG_SECTION]: 8,
+  // Designated content section
+  [TAG_MAIN]: 15,
+  // Main content indicator
+  [TAG_P]: 5,
+  // Direct paragraph content
+  [TAG_DIV]: 2,
+  // Generic container, slightly positive
+  [TAG_BLOCKQUOTE]: 5,
+  // Quoted content, usually important
+  // Code and pre-formatted content
+  [TAG_PRE]: 8,
+  // Preformatted text/code, high value for documentation
+  [TAG_CODE]: 6,
+  // Code content, high value for documentation
+  // Media elements
+  [TAG_IMG]: 3,
+  // Images are typically content
+  [TAG_FIGURE]: 4,
+  // Figure with caption, content-focused
+  [TAG_FIGCAPTION]: 3,
+  // Description for a figure
+  [TAG_VIDEO]: 3,
+  // Video content
+  [TAG_AUDIO]: 3,
+  // Audio content
+  [TAG_SVG]: 1,
+  // Vector graphic, slight positive
+  // Table elements
+  [TAG_TABLE]: 0,
+  // Could be data or layout, neutral
+  [TAG_CAPTION]: 2,
+  // Table caption
+  [TAG_THEAD]: 0,
+  // Table structure, neutral
+  [TAG_TBODY]: 0,
+  // Table structure, neutral
+  [TAG_TFOOT]: 0,
+  // Table structure, neutral
+  [TAG_TR]: -1,
+  // Table row, slight negative
+  [TAG_TH]: -2,
+  // Table header, more negative than cells
+  [TAG_TD]: 0,
+  // Table cell, neutral
+  // List elements
+  [TAG_UL]: -8,
+  // Higher penalty as lists are often navigation
+  [TAG_OL]: -5,
+  // Ordered lists still often navigation
+  [TAG_LI]: -6,
+  // Higher penalty for list items to avoid nav lists
+  [TAG_DL]: 0,
+  // Definition lists, neutral
+  [TAG_DT]: 0,
+  // Definition lists, neutral
+  [TAG_DD]: 0,
+  // Definition lists, neutral
+  // Heading elements
+  [TAG_H1]: 1,
+  // Top-level heading (may be site title)
+  [TAG_H2]: 1,
+  // Section headers, slightly positive
+  [TAG_H3]: 1,
+  // Section headers, slightly positive
+  [TAG_H4]: 0,
+  // Minor headers, neutral
+  [TAG_H5]: 0,
+  // Minor headers, neutral
+  [TAG_H6]: 0,
+  // Minor headers, neutral
+  // Navigation and structural elements (negative)
+  [TAG_HEADER]: -15,
+  // Page header, often not content but may contain article headers
+  [TAG_FOOTER]: -25,
+  // Footer, rarely content
+  [TAG_NAV]: -30,
+  // Navigation, not content
+  [TAG_ASIDE]: -25,
+  // Sidebar, usually not main content
+  // Form elements (negative)
+  [TAG_FORM]: -8,
+  // User input, not content
+  [TAG_BUTTON]: -5,
+  // Interactive element, not content
+  [TAG_INPUT]: -5,
+  // Form field, not content
+  [TAG_TEXTAREA]: -5,
+  // Text input, not content
+  [TAG_SELECT]: -5,
+  // Drop-down, not content
+  [TAG_FIELDSET]: -5,
+  // Form field group, not content
+  // Embedded content (mostly negative)
+  [TAG_IFRAME]: -3,
+  // Embedded content, often ads
+  [TAG_EMBED]: -3,
+  // Embedded content, often ads
+  [TAG_OBJECT]: -3,
+  // Embedded content, often ads
+  // Links
+  [TAG_A]: -8,
+  // Higher penalty to avoid navigation-heavy areas
+  // Text formatting
+  [TAG_STRONG]: 1,
+  // Emphasized text, slightly positive
+  [TAG_B]: 1,
+  // Emphasized text, slightly positive
+  [TAG_EM]: 1,
+  // Emphasized text, slightly positive
+  [TAG_I]: 1,
+  // Emphasized text, slightly positive
+  // Miscellaneous elements
+  [TAG_HR]: 0,
+  // Divider, neutral
+  [TAG_BR]: 0,
+  // Line break, neutral
+  [TAG_SPAN]: 0,
+  // Inline container, neutral
+  [TAG_SCRIPT]: -25,
+  // Script, never content
+  [TAG_STYLE]: -25,
+  // Style, never content
+  // Expandable content
+  [TAG_DETAILS]: 2,
+  // Expandable content
+  [TAG_SUMMARY]: 1,
+  // Header for expandable content
+  // Additional tags not explicitly in scoring.md
+  [TAG_ADDRESS]: -3
+  // Similar to footer, rarely content
+};
+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;
+}
+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 > 0.4) {
+          if (linkDensity > 0.6) {
+            node.context.score = node.context.score * 0.02;
+            if (linkTextLength > 50) {
+              node.context.isHighLinkDensity = true;
+            }
+          } else {
+            node.context.score *= 1 - linkDensity * 2;
+          }
+        } else if (linkDensity > 0.2) {
+          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 > 0.3 && 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 > 0.5) {
+        createBufferRegion(node, state, false);
+      }
+      if (node.tagHandler?.isInline) {
+        const parent = node.parent;
+        if (parent && parent.context) {
+          parent.context.score += finalScore - (parent.context.score || 0);
+        }
+      }
+    }
+  });
+}
+
+export { readabilityPlugin as r };