hexo-text-pipeline 0.2.0

package/lib/presets/obsidian/converters/embed/index.js

@@ -0,0 +1,79 @@
+'use strict';
+
+const { replaceOutsideCode } = require('../../../../core/markdown-guard');
+const { normalizeBase } = require('../../../../core/config');
+const { getPostIndex, resolvePostByTarget, buildPostHref } = require('../../post-index');
+
+const IMAGE_EXTENSIONS = ['png', 'jpg', 'jpeg', 'gif', 'svg', 'webp', 'bmp', 'avif'];
+
+/**
+ * Markdown 阶段处理 Obsidian 嵌入 ![[target|param]]：
+ * - 图片（按扩展名识别）→ 标准 markdown 图片；param 为数字时输出 <img width>（Obsidian 缩放语法）
+ *   src = asset_prefix + target（attachment 目录布局是站点私事，交给配置）
+ * - 能解析为文章的笔记嵌入 → 降级为指向该文章的链接（博客无法内联整篇笔记）
+ * - 其他（pdf、音视频、解析不到的笔记）→ 原样保留
+ *
+ * 必须在 wikilink 之前运行：否则 ![[x]] 的 [[x]] 部分会被替换成链接，
+ * 残留的 ! 使其变成指向文章 URL 的图片。
+ */
+function parseEmbed(raw) {
+  const firstPipe = raw.indexOf('|');
+  const targetAndAnchor = firstPipe >= 0 ? raw.slice(0, firstPipe) : raw;
+  const param = firstPipe >= 0 ? raw.slice(firstPipe + 1).trim() : '';
+
+  const hashIndex = targetAndAnchor.indexOf('#');
+  const target = (hashIndex >= 0 ? targetAndAnchor.slice(0, hashIndex) : targetAndAnchor).trim();
+
+  return { target, param };
+}
+
+function imageExtension(target) {
+  const dot = target.lastIndexOf('.');
+  if (dot === -1) return '';
+  return target.slice(dot + 1).toLowerCase();
+}
+
+function renderImage(target, param, assetPrefix) {
+  const src = encodeURI((assetPrefix ? assetPrefix + '/' : '') + target.replace(/^\/+/, ''));
+  if (/^\d+$/.test(param)) {
+    return '<img src="' + src + '" width="' + param + '">';
+  }
+  const alt = param || target.split('/').pop();
+  return '![' + alt + '](' + src + ')';
+}
+
+function createEmbedReplacer(index, domainPrefix, assetPrefix) {
+  return function replaceEmbeds(segment) {
+    return segment.replace(/!\[\[([^\]]+)\]\]/g, (full, inner) => {
+      const { target, param } = parseEmbed(inner);
+      if (!target) return full;
+
+      if (IMAGE_EXTENSIONS.includes(imageExtension(target))) {
+        return renderImage(target, param, assetPrefix);
+      }
+
+      const post = resolvePostByTarget(index, target);
+      if (!post) return full;
+
+      const href = buildPostHref(post, domainPrefix);
+      if (!href) return full;
+
+      return '[' + (param || target) + '](' + href + ')';
+    });
+  };
+}
+
+module.exports = {
+  name: 'embed',
+  stage: 'before_post_render',
+  test(content) {
+    return content.includes('![[');
+  },
+  convert(content, ctx) {
+    const index = getPostIndex(ctx.hexo);
+    const domainPrefix = normalizeBase(ctx.presetConfig.domain_prefix);
+    const assetPrefix = normalizeBase(ctx.config.asset_prefix);
+    return replaceOutsideCode(content, createEmbedReplacer(index, domainPrefix, assetPrefix));
+  },
+  _internal: { parseEmbed, createEmbedReplacer, renderImage }
+};

package/lib/presets/obsidian/converters/highlight/index.js

@@ -0,0 +1,23 @@
+'use strict';
+
+const { replaceOutsideCode } = require('../../../../core/markdown-guard');
+
+/**
+ * Markdown 阶段把 Obsidian 高亮 ==文本== 改写为 <mark>文本</mark>。
+ * 不跨行、内容非空且不含 =；代码内为字面量。
+ */
+function replaceHighlights(segment) {
+  return segment.replace(/==([^=\n]+)==/g, '<mark>$1</mark>');
+}
+
+module.exports = {
+  name: 'highlight',
+  stage: 'before_post_render',
+  test(content) {
+    return content.includes('==');
+  },
+  convert(content) {
+    return replaceOutsideCode(content, replaceHighlights);
+  },
+  _internal: { replaceHighlights }
+};

package/lib/presets/obsidian/converters/mdlink/index.js

@@ -0,0 +1,63 @@
+'use strict';
+
+const { normalizeBase } = require('../../../../core/config');
+const { getPostIndex, resolvePostByTarget, buildPostHref, safeDecodeURI } = require('../../post-index');
+
+/**
+ * HTML 阶段兜底：把渲染后仍指向 .md 文件的链接（markdown 形式和 href 属性两种残留）
+ * 改写为 abbrlink 永久链接。处理 Obsidian 导出的相对路径和 URI 编码。
+ */
+function isExternalHref(href) {
+  return /^(https?:|mailto:|tel:|\/\/)/i.test(href);
+}
+
+function resolveHref(index, domainPrefix, rawHref) {
+  const hashPos = rawHref.indexOf('#');
+  const hrefWithoutHash = hashPos >= 0 ? rawHref.slice(0, hashPos) : rawHref;
+  const hashRaw = hashPos >= 0 ? rawHref.slice(hashPos + 1) : '';
+
+  const post = resolvePostByTarget(index, safeDecodeURI(hrefWithoutHash));
+  if (!post) return null;
+
+  const href = buildPostHref(post, domainPrefix);
+  if (!href) return null;
+
+  const anchor = hashRaw ? '#' + encodeURIComponent(safeDecodeURI(hashRaw)) : '';
+  return href + anchor;
+}
+
+function createMarkdownMdLinkReplacer(index, domainPrefix) {
+  return function replaceMarkdownMdLinks(content) {
+    return content.replace(/\[([^\]]+)\]\(([^\n]+?\.md(?:#[^\n]+)?)\)/gi, (full, text, rawHref) => {
+      if (isExternalHref(rawHref)) return full;
+      const href = resolveHref(index, domainPrefix, rawHref);
+      return href ? '[' + text + '](' + href + ')' : full;
+    });
+  };
+}
+
+function createHtmlMdHrefReplacer(index, domainPrefix) {
+  return function replaceMdHref(html) {
+    return html.replace(/(href\s*=\s*["'])([^"']+?\.md(?:#[^"']*)?)(["'])/gi, (full, prefix, rawHref, suffix) => {
+      if (isExternalHref(rawHref)) return full;
+      const href = resolveHref(index, domainPrefix, rawHref);
+      return href ? prefix + href + suffix : full;
+    });
+  };
+}
+
+module.exports = {
+  name: 'mdlink',
+  stage: 'after_post_render',
+  test(content) {
+    return content.includes('.md');
+  },
+  convert(content, ctx) {
+    const index = getPostIndex(ctx.hexo);
+    const domainPrefix = normalizeBase(ctx.presetConfig.domain_prefix);
+    const replaceMarkdownMdLinks = createMarkdownMdLinkReplacer(index, domainPrefix);
+    const replaceMdHref = createHtmlMdHrefReplacer(index, domainPrefix);
+    return replaceMdHref(replaceMarkdownMdLinks(content));
+  },
+  _internal: { createMarkdownMdLinkReplacer, createHtmlMdHrefReplacer }
+};

package/lib/presets/obsidian/converters/mermaid/index.js

@@ -0,0 +1,102 @@
+'use strict';
+
+const DEFAULT_SCRIPT_SRC = 'https://cdn.jsdelivr.net/npm/mermaid@11/dist/mermaid.min.js';
+const DEFAULT_CLASS = 'mermaid';
+
+/**
+ * Markdown 阶段把 ```mermaid 围栏块替换为 <pre class="mermaid">，
+ * 让图源码绕过语法高亮（highlight.js / prismjs 会把它渲染成代码而非图）。
+ * 内容做 HTML 转义，浏览器解析后 textContent 还原原文，mermaid.js 正常读取。
+ *
+ * 前端脚本默认按需注入（页面没有 .mermaid 元素时不加载 CDN）：
+ * converters.mermaid 子配置：class / inject_script / script_src / theme。
+ */
+function escapeHtml(text) {
+  return text.replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
+}
+
+function transformMermaidBlocks(content, className) {
+  const lines = content.split('\n');
+  const out = [];
+
+  let inFence = false;
+  let fenceChar = '';
+  let fenceLen = 0;
+
+  for (let i = 0; i < lines.length; i += 1) {
+    const line = lines[i];
+    const match = line.match(/^ {0,3}([`~]{3,})(.*)$/);
+
+    if (!inFence && match && match[2].trim().toLowerCase() === 'mermaid') {
+      const token = match[1];
+      // 收集到配对的闭合围栏；未闭合则原样保留
+      const body = [];
+      let closed = false;
+      let j = i + 1;
+      for (; j < lines.length; j += 1) {
+        const closeMatch = lines[j].match(/^ {0,3}([`~]{3,})\s*$/);
+        if (closeMatch && closeMatch[1][0] === token[0] && closeMatch[1].length >= token.length) {
+          closed = true;
+          break;
+        }
+        body.push(lines[j]);
+      }
+      if (closed) {
+        out.push('<pre class="' + className + '">' + escapeHtml(body.join('\n')) + '</pre>');
+        i = j;
+        continue;
+      }
+      out.push(line);
+      continue;
+    }
+
+    if (match) {
+      const token = match[1];
+      if (!inFence) {
+        inFence = true;
+        fenceChar = token[0];
+        fenceLen = token.length;
+      } else if (token[0] === fenceChar && token.length >= fenceLen && !match[2].trim()) {
+        inFence = false;
+      }
+    }
+
+    out.push(line);
+  }
+
+  return out.join('\n');
+}
+
+function buildLoaderScript(config) {
+  if (config.inject_script === false) return '';
+  const src = typeof config.script_src === 'string' && config.script_src ? config.script_src : DEFAULT_SCRIPT_SRC;
+  const theme = typeof config.theme === 'string' && config.theme ? config.theme : 'default';
+  const className = typeof config.class === 'string' && config.class ? config.class : DEFAULT_CLASS;
+  return (
+    '(function(){' +
+    'if(!document.querySelector("pre.' + className + '"))return;' +
+    'var s=document.createElement("script");' +
+    's.src=' + JSON.stringify(src) + ';' +
+    's.onload=function(){window.mermaid.initialize({startOnLoad:false,theme:' + JSON.stringify(theme) + '});' +
+    'window.mermaid.run({querySelector:"pre.' + className + '"});};' +
+    'document.head.appendChild(s);' +
+    '})();'
+  );
+}
+
+module.exports = {
+  name: 'mermaid',
+  stage: 'before_post_render',
+  js: buildLoaderScript,
+  test(content) {
+    return /^ {0,3}[`~]{3,}\s*mermaid\s*$/im.test(content);
+  },
+  convert(content, ctx) {
+    const className =
+      ctx && ctx.config && typeof ctx.config.class === 'string' && ctx.config.class
+        ? ctx.config.class
+        : DEFAULT_CLASS;
+    return transformMermaidBlocks(content, className);
+  },
+  _internal: { transformMermaidBlocks, buildLoaderScript }
+};

package/lib/presets/obsidian/converters/wikilink/index.js

@@ -0,0 +1,65 @@
+'use strict';
+
+const { replaceOutsideCode } = require('../../../../core/markdown-guard');
+const { normalizeBase } = require('../../../../core/config');
+const { getPostIndex, resolvePostByTarget, buildPostHref } = require('../../post-index');
+
+/**
+ * Markdown 阶段把 Obsidian wiki link [[target#anchor|alias]] 改写为
+ * 指向 abbrlink 永久链接的标准 markdown 链接。无法解析的目标原样保留。
+ *
+ * - ![[...]] 是嵌入，归 embed node 管，这里用 lookbehind 跳过
+ * - #^block-id 块引用锚点在渲染后的 HTML 里不存在，丢弃锚点只留文章链接
+ * - [[#标题]]（无 target 的同页链接）改写为页内锚点
+ */
+function parseWikiLink(raw) {
+  const firstPipe = raw.indexOf('|');
+  const targetAndAnchor = firstPipe >= 0 ? raw.slice(0, firstPipe) : raw;
+  const alias = firstPipe >= 0 ? raw.slice(firstPipe + 1).trim() : '';
+
+  const hashIndex = targetAndAnchor.indexOf('#');
+  const target = hashIndex >= 0 ? targetAndAnchor.slice(0, hashIndex).trim() : targetAndAnchor.trim();
+  const anchor = hashIndex >= 0 ? targetAndAnchor.slice(hashIndex + 1).trim() : '';
+
+  return { target, anchor, alias };
+}
+
+function createWikiLinkReplacer(index, domainPrefix) {
+  return function replaceWikiLinks(segment) {
+    return segment.replace(/(?<!!)\[\[([^\]]+)\]\]/g, (full, inner) => {
+      const parsed = parseWikiLink(inner);
+
+      // 同页链接 [[#标题]]：没有 target，只有锚点
+      if (!parsed.target) {
+        if (!parsed.anchor || parsed.anchor.startsWith('^')) return full;
+        return '[' + (parsed.alias || parsed.anchor) + '](#' + encodeURIComponent(parsed.anchor) + ')';
+      }
+
+      const post = resolvePostByTarget(index, parsed.target);
+      if (!post) return full;
+
+      const href = buildPostHref(post, domainPrefix);
+      if (!href) return full;
+
+      // 块引用锚点（#^id）在 HTML 里没有对应元素，降级为文章链接
+      const anchor = parsed.anchor && !parsed.anchor.startsWith('^') ? '#' + encodeURIComponent(parsed.anchor) : '';
+      const text = parsed.alias || parsed.target;
+
+      return '[' + text + '](' + href + anchor + ')';
+    });
+  };
+}
+
+module.exports = {
+  name: 'wikilink',
+  stage: 'before_post_render',
+  test(content) {
+    return content.includes('[[');
+  },
+  convert(content, ctx) {
+    const index = getPostIndex(ctx.hexo);
+    const replacer = createWikiLinkReplacer(index, normalizeBase(ctx.presetConfig.domain_prefix));
+    return replaceOutsideCode(content, replacer);
+  },
+  _internal: { parseWikiLink, createWikiLinkReplacer }
+};

package/lib/presets/obsidian/index.js

@@ -0,0 +1,37 @@
+'use strict';
+
+const { invalidatePostIndex } = require('./post-index');
+
+/**
+ * Obsidian Flavored Markdown preset —— 本插件自带的"插件的插件"。
+ * nodes 数组顺序就是同 stage 同 priority 下的执行顺序：
+ * comment 必须最先（被注释掉的语法要在其他 node 看到之前消失）。
+ *
+ * 启用与配置（_config.yml）：
+ * text_pipeline:
+ *   presets:
+ *     - name: obsidian
+ *       config:
+ *         domain_prefix: ''                    # wikilink/mdlink 的链接前缀
+ *         callout: { enable: true }            # callout 出厂默认关
+ *         mermaid: { theme: dark }             # 任意 node 级子配置 / priority 覆盖
+ */
+module.exports = {
+  name: 'obsidian',
+  nodes: [
+    require('./converters/comment'),
+    require('./converters/embed'), // 必须在 wikilink 前：![[x]] 不能被当作 ![[x]] 的 [[x]] 部分
+    require('./converters/wikilink'),
+    require('./converters/highlight'),
+    require('./converters/blockid'),
+    require('./converters/mdlink'),
+    require('./converters/mermaid'),
+    require('./converters/callout')
+  ],
+  init(hexo) {
+    // wikilink/mdlink 共享的文章索引：每轮 generate 前失效重建
+    hexo.extend.filter.register('before_generate', () => {
+      invalidatePostIndex(hexo);
+    });
+  }
+};

package/lib/presets/obsidian/post-index.js

@@ -0,0 +1,158 @@
+'use strict';
+
+/**
+ * 文章索引服务：把 hexo 的 Post 集合建成多键索引（title / slug / source 路径），
+ * 供链接类 converter 把 Obsidian 目标解析成 abbrlink 永久链接。
+ * 缓存以 hexo 实例为键，engine 在 before_generate 时调用 invalidatePostIndex。
+ */
+
+function trimSlashes(value) {
+  return String(value || '').replace(/^\/+|\/+$/g, '');
+}
+
+function stripMarkdownExtension(value) {
+  return value.replace(/\.(md|markdown)$/i, '');
+}
+
+function normalizeLookupKey(value) {
+  return String(value || '').trim().replace(/\\/g, '/').replace(/^\/+|\/+$/g, '').toLowerCase();
+}
+
+function stripLeadingDotSegments(value) {
+  return value.replace(/^(\.\.\/|\.\/)+/, '');
+}
+
+function safeDecodeURI(value) {
+  try {
+    return decodeURIComponent(value);
+  } catch (_err) {
+    return value;
+  }
+}
+
+function buildPostIndex(hexo) {
+  let posts = [];
+  if (hexo.model && typeof hexo.model === 'function') {
+    const postModel = hexo.model('Post');
+    if (postModel && typeof postModel.toArray === 'function') {
+      posts = postModel.toArray();
+    }
+  }
+
+  if (!posts.length && hexo.locals && typeof hexo.locals.get === 'function') {
+    const postQuery = hexo.locals.get('posts');
+    posts = postQuery ? postQuery.toArray() : [];
+  }
+
+  const byTitle = new Map();
+  const bySlug = new Map();
+  const bySourcePath = new Map();
+  const bySourceBase = new Map();
+
+  let indexedCount = 0;
+
+  for (const post of posts) {
+    // abbrlink（frontmatter 标签）优先；没有时兜底用 hexo 生成的 post.path
+    if (!post || (!post.abbrlink && !post.path)) continue;
+    indexedCount += 1;
+
+    if (post.title) {
+      byTitle.set(String(post.title).trim().toLowerCase(), post);
+    }
+
+    if (post.slug) {
+      bySlug.set(String(post.slug).trim().toLowerCase(), post);
+    }
+
+    if (post.source) {
+      const normalizedSource = normalizeLookupKey(post.source).replace(/^.*?_posts\//, '');
+      const sourceNoExt = stripMarkdownExtension(normalizedSource);
+      const sourceBase = sourceNoExt.includes('/') ? sourceNoExt.slice(sourceNoExt.lastIndexOf('/') + 1) : sourceNoExt;
+
+      if (sourceNoExt) {
+        bySourcePath.set(sourceNoExt, post);
+      }
+
+      if (sourceBase) {
+        bySourceBase.set(sourceBase, post);
+      }
+    }
+  }
+
+  return { byTitle, bySlug, bySourcePath, bySourceBase, indexedCount };
+}
+
+function buildTargetCandidates(target) {
+  const normalized = normalizeLookupKey(stripLeadingDotSegments(target));
+  if (!normalized) return [];
+
+  const candidates = new Set();
+  const noExt = stripMarkdownExtension(normalized);
+
+  candidates.add(normalized);
+  candidates.add(noExt);
+
+  if (normalized.includes('/')) {
+    const base = normalized.slice(normalized.lastIndexOf('/') + 1);
+    candidates.add(base);
+    candidates.add(stripMarkdownExtension(base));
+  }
+
+  return Array.from(candidates).filter(Boolean);
+}
+
+function resolvePostByTarget(index, target) {
+  const candidates = buildTargetCandidates(target);
+
+  for (const key of candidates) {
+    const post =
+      index.byTitle.get(key) ||
+      index.bySlug.get(key) ||
+      index.bySourcePath.get(key) ||
+      index.bySourceBase.get(key);
+    if (post) return post;
+  }
+
+  return null;
+}
+
+function buildPostHref(post, domainPrefix) {
+  let hrefPath;
+  if (post.abbrlink) {
+    hrefPath = '/posts/' + trimSlashes(post.abbrlink);
+  } else if (post.path) {
+    hrefPath = '/' + trimSlashes(String(post.path));
+  } else {
+    return null;
+  }
+  return domainPrefix ? domainPrefix + hrefPath : hrefPath;
+}
+
+const cache = new WeakMap();
+
+function getPostIndex(hexo) {
+  let index = cache.get(hexo);
+  if (!index || !index.indexedCount) {
+    index = buildPostIndex(hexo);
+    cache.set(hexo, index);
+  }
+  return index;
+}
+
+function invalidatePostIndex(hexo) {
+  cache.delete(hexo);
+}
+
+module.exports = {
+  trimSlashes,
+  stripMarkdownExtension,
+  normalizeLookupKey,
+  stripLeadingDotSegments,
+  safeDecodeURI,
+  buildPostIndex,
+  buildTargetCandidates,
+  resolvePostByTarget,
+  buildPostHref,
+  getPostIndex,
+  invalidatePostIndex
+};

package/package.json

@@ -0,0 +1,41 @@
+{
+  "name": "hexo-text-pipeline",
+  "version": "0.2.0",
+  "description": "A general-purpose hooks bus for Hexo's render pipeline: hang your own scripts/commands on any text stage (text in, text out, edit-and-use), with a checker system as the safety net. Ships an Obsidian Flavored Markdown preset.",
+  "main": "index.js",
+  "files": [
+    "index.js",
+    "lib",
+    "README.md",
+    "README.zh-CN.md"
+  ],
+  "scripts": {
+    "test": "node --test",
+    "prepublishOnly": "npm test"
+  },
+  "keywords": [
+    "hexo",
+    "hexo-plugin",
+    "hooks",
+    "pipeline",
+    "text-transform",
+    "obsidian",
+    "obsidian-flavored-markdown",
+    "wikilink",
+    "mermaid",
+    "callout"
+  ],
+  "author": "Sentixxx",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/moesin-lab/hexo-text-pipeline.git"
+  },
+  "homepage": "https://github.com/moesin-lab/hexo-text-pipeline#readme",
+  "bugs": {
+    "url": "https://github.com/moesin-lab/hexo-text-pipeline/issues"
+  },
+  "engines": {
+    "node": ">=16"
+  }
+}