fix: wikilinks should allow external links (closes #639)

Jacky Zhao

2023-12-19 b44a79eebaf1664512ef5ff4e3246b3a6a416ff4

 quartz/plugins/transformers/ofm.ts

@@ -1,15 +1,20 @@
import { PluggableList } from "unified"
import { QuartzTransformerPlugin } from "../types"
import { Root, HTML, BlockContent, DefinitionContent, Code } from 'mdast'
import { findAndReplace } from "mdast-util-find-and-replace"
import { slugify } from "../../path"
import { slug as slugAnchor } from 'github-slugger'
import { Root, Html, BlockContent, DefinitionContent, Paragraph, Code } from "mdast"
import { Element, Literal, Root as HtmlRoot } from "hast"
import { ReplaceFunction, findAndReplace as mdastFindReplace } from "mdast-util-find-and-replace"
import { slug as slugAnchor } from "github-slugger"
import rehypeRaw from "rehype-raw"
import { visit } from "unist-util-visit"
import path from "path"
import { JSResource } from "../../resources"
import { JSResource } from "../../util/resources"
// @ts-ignore
import calloutScript from "../../components/scripts/callout.inline.ts"
import { FilePath, pathToRoot, slugTag, slugifyFilePath } from "../../util/path"
import { toHast } from "mdast-util-to-hast"
import { toHtml } from "hast-util-to-html"
import { PhrasingContent } from "mdast-util-find-and-replace/lib"
import { capitalize } from "../../util/lang"
import { PluggableList } from "unified"

export interface Options {
  comments: boolean
@@ -17,6 +22,9 @@
  wikilinks: boolean
  callouts: boolean
  mermaid: boolean
  parseTags: boolean
  parseBlockReferences: boolean
  enableInHtmlEmbed: boolean
}

const defaultOptions: Options = {
@@ -25,6 +33,9 @@
  wikilinks: true,
  callouts: true,
  mermaid: true,
  parseTags: true,
  parseBlockReferences: true,
  enableInHtmlEmbed: false,
}

const icons = {
@@ -43,40 +54,6 @@
  quoteIcon: `<svg xmlns="http://www.w3.org/2000/svg" width="100%" height="100%" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><path d="M3 21c3 0 7-1 7-8V5c0-1.25-.756-2.017-2-2H4c-1.25 0-2 .75-2 1.972V11c0 1.25.75 2 2 2 1 0 1 0 1 1v1c0 1-1 2-2 2s-1 .008-1 1.031V20c0 1 0 1 1 1z"></path><path d="M15 21c3 0 7-1 7-8V5c0-1.25-.757-2.017-2-2h-4c-1.25 0-2 .75-2 1.972V11c0 1.25.75 2 2 2h.75c0 2.25.25 4-2.75 4v3c0 1 0 1 1 1z"></path></svg>`,
}

function canonicalizeCallout(calloutName: string): keyof typeof callouts {
  let callout = calloutName.toLowerCase() as keyof typeof calloutMapping

  const calloutMapping: Record<string, keyof typeof callouts> = {
    note: "note",
    abstract: "abstract",
    info: "info",
    todo: "todo",
    tip: "tip",
    hint: "tip",
    important: "tip",
    success: "success",
    check: "success",
    done: "success",
    question: "question",
    help: "question",
    faq: "question",
    warning: "warning",
    attention: "warning",
    caution: "warning",
    failure: "failure",
    missing: "failure",
    fail: "failure",
    danger: "danger",
    error: "danger",
    bug: "bug",
    example: "example",
    quote: "quote",
    cite: "quote"
  }

  return calloutMapping[callout]
}

const callouts = {
  note: icons.pencilIcon,
  abstract: icons.clipboardListIcon,
@@ -93,135 +70,283 @@
  quote: icons.quoteIcon,
}

const capitalize = (s: string): string => {
  return s.substring(0, 1).toUpperCase() + s.substring(1);
const calloutMapping: Record<string, keyof typeof callouts> = {
  note: "note",
  abstract: "abstract",
  summary: "abstract",
  tldr: "abstract",
  info: "info",
  todo: "todo",
  tip: "tip",
  hint: "tip",
  important: "tip",
  success: "success",
  check: "success",
  done: "success",
  question: "question",
  help: "question",
  faq: "question",
  warning: "warning",
  attention: "warning",
  caution: "warning",
  failure: "failure",
  missing: "failure",
  fail: "failure",
  danger: "danger",
  error: "danger",
  bug: "bug",
  example: "example",
  quote: "quote",
  cite: "quote",
}

// Match wikilinks 
function canonicalizeCallout(calloutName: string): keyof typeof callouts {
  let callout = calloutName.toLowerCase() as keyof typeof calloutMapping
  return calloutMapping[callout] ?? "note"
}

export const externalLinkRegex = /^https?:\/\//i

// !?               -> optional embedding
// \[\[             -> open brace
// ([^\[\]\|\#]+)   -> one or more non-special characters ([,],|, or #) (name)
// (#[^\[\]\|\#]+)? -> # then one or more non-special characters (heading link)
// (|[^\[\]\|\#]+)? -> | then one or more non-special characters (alias)
const wikilinkRegex = new RegExp(/!?\[\[([^\[\]\|\#]+)(#[^\[\]\|\#]+)?(\|[^\[\]\|\#]+)?\]\]/, "g")

// Match highlights 
const highlightRegex = new RegExp(/==(.+)==/, "g")

// Match comments 
export const wikilinkRegex = new RegExp(
  /!?\[\[([^\[\]\|\#]+)?(#+[^\[\]\|\#]+)?(\|[^\[\]\|\#]+)?\]\]/,
  "g",
)
const highlightRegex = new RegExp(/==([^=]+)==/, "g")
const commentRegex = new RegExp(/%%(.+)%%/, "g")

// from https://github.com/escwxyz/remark-obsidian-callout/blob/main/src/index.ts
const calloutRegex = new RegExp(/^\[\!(\w+)\]([+-]?)/)
const calloutLineRegex = new RegExp(/^> *\[\!\w+\][+-]?.*$/, "gm")
// (?:^| )              -> non-capturing group, tag should start be separated by a space or be the start of the line
// #(...)               -> capturing group, tag itself must start with #
// (?:[-_\p{L}\d\p{Z}])+       -> non-capturing group, non-empty string of (Unicode-aware) alpha-numeric characters and symbols, hyphens and/or underscores
// (?:\/[-_\p{L}\d\p{Z}]+)*)   -> non-capturing group, matches an arbitrary number of tag strings separated by "/"
const tagRegex = new RegExp(/(?:^| )#((?:[-_\p{L}\d])+(?:\/[-_\p{L}\d]+)*)/, "gu")
const blockReferenceRegex = new RegExp(/\^([A-Za-z0-9]+)$/, "g")

export const ObsidianFlavoredMarkdown: QuartzTransformerPlugin<Partial<Options> | undefined> = (userOpts) => {
export const ObsidianFlavoredMarkdown: QuartzTransformerPlugin<Partial<Options> | undefined> = (
  userOpts,
) => {
  const opts = { ...defaultOptions, ...userOpts }

  const mdastToHtml = (ast: PhrasingContent | Paragraph) => {
    const hast = toHast(ast, { allowDangerousHtml: true })!
    return toHtml(hast, { allowDangerousHtml: true })
  }

  return {
    name: "ObsidianFlavoredMarkdown",
    textTransform(src) {
    textTransform(_ctx, src) {
      // pre-transform blockquotes
      if (opts.callouts) {
        if (src instanceof Buffer) {
          src = src.toString()
        }

        src = src.replaceAll(calloutLineRegex, (value) => {
          // force newline after title of callout
          return value + "\n> "
        })
      }

      // pre-transform wikilinks (fix anchors to things that may contain illegal syntax e.g. codeblocks, latex)
      if (opts.wikilinks) {
        src = src.toString()
        return src.replaceAll(wikilinkRegex, (value, ...capture) => {
          const [fp, rawHeader, rawAlias] = capture
          const anchor = rawHeader?.trim().slice(1)
          const displayAnchor = anchor ? `#${slugAnchor(anchor)}` : ""
          const displayAlias = rawAlias ?? ""
        if (src instanceof Buffer) {
          src = src.toString()
        }

        src = src.replaceAll(wikilinkRegex, (value, ...capture) => {
          const [rawFp, rawHeader, rawAlias]: (string | undefined)[] = capture

          const fp = rawFp ?? ""
          const anchor = rawHeader?.trim().replace(/^#+/, "")
          const blockRef = Boolean(anchor?.startsWith("^")) ? "^" : ""
          const displayAnchor = anchor ? `#${blockRef}${slugAnchor(anchor)}` : ""
          const displayAlias = rawAlias ?? rawHeader?.replace("#", "|") ?? ""
          const embedDisplay = value.startsWith("!") ? "!" : ""

          if (rawFp?.match(externalLinkRegex)) {
            return `${embedDisplay}[${displayAlias.replace(/^\|/, "")}](${rawFp})`
          }

          return `${embedDisplay}[[${fp}${displayAnchor}${displayAlias}]]`
        })
      }

      return src
    },
    markdownPlugins() {
      const plugins: PluggableList = []
      if (opts.wikilinks) {
        plugins.push(() => {
          return (tree: Root, _file) => {
            findAndReplace(tree, wikilinkRegex, (value: string, ...capture: string[]) => {
              const [fp, rawHeader, rawAlias] = capture
              const anchor = rawHeader?.trim() ?? ""
              const alias = rawAlias?.slice(1).trim()

              // embed cases
              if (value.startsWith("!")) {
                const ext = path.extname(fp).toLowerCase()
                const url = slugify(fp.trim()) + ext
                if ([".png", ".jpg", ".jpeg", ".gif", ".bmp", ".svg"].includes(ext)) {
                  const dims = alias ?? ""
                  let [width, height] = dims.split("x", 2)
                  width ||= "auto"
                  height ||= "auto"
                  return {
                    type: 'image',
                    url,
                    data: {
                      hProperties: {
                        width, height
                      }
      // regex replacements
      plugins.push(() => {
        return (tree: Root, file) => {
          const replacements: [RegExp, string | ReplaceFunction][] = []
          const base = pathToRoot(file.data.slug!)

          if (opts.wikilinks) {
            replacements.push([
              wikilinkRegex,
              (value: string, ...capture: string[]) => {
                let [rawFp, rawHeader, rawAlias] = capture
                const fp = rawFp?.trim() ?? ""
                const anchor = rawHeader?.trim() ?? ""
                const alias = rawAlias?.slice(1).trim()

                // embed cases
                if (value.startsWith("!")) {
                  const ext: string = path.extname(fp).toLowerCase()
                  const url = slugifyFilePath(fp as FilePath)
                  if ([".png", ".jpg", ".jpeg", ".gif", ".bmp", ".svg"].includes(ext)) {
                    const dims = alias ?? ""
                    let [width, height] = dims.split("x", 2)
                    width ||= "auto"
                    height ||= "auto"
                    return {
                      type: "image",
                      url,
                      data: {
                        hProperties: {
                          width,
                          height,
                        },
                      },
                    }
                  } else if ([".mp4", ".webm", ".ogv", ".mov", ".mkv"].includes(ext)) {
                    return {
                      type: "html",
                      value: `<video src="${url}" controls></video>`,
                    }
                  } else if (
                    [".mp3", ".webm", ".wav", ".m4a", ".ogg", ".3gp", ".flac"].includes(ext)
                  ) {
                    return {
                      type: "html",
                      value: `<audio src="${url}" controls></audio>`,
                    }
                  } else if ([".pdf"].includes(ext)) {
                    return {
                      type: "html",
                      value: `<iframe src="${url}"></iframe>`,
                    }
                  } else if (ext === "") {
                    const block = anchor
                    return {
                      type: "html",
                      data: { hProperties: { transclude: true } },
                      value: `<blockquote class="transclude" data-url="${url}" data-block="${block}"><a href="${
                        url + anchor
                      }" class="transclude-inner">Transclude of ${url}${block}</a></blockquote>`,
                    }
                  }
                } else if ([".mp4", ".webm", ".ogv", ".mov", ".mkv"].includes(ext)) {
                  return {
                    type: 'html',
                    value: `<video src="${url}" controls></video>`
                  }
                } else if ([".mp3", ".webm", ".wav", ".m4a", ".ogg", ".3gp", ".flac"].includes(ext)) {
                  return {
                    type: 'html',
                    value: `<audio src="${url}" controls></audio>`
                  }
                } else if ([".pdf"].includes(ext)) {
                  return {
                    type: 'html',
                    value: `<iframe src="${url}"></iframe>`
                  }

                  // otherwise, fall through to regular link
                }
                // otherwise, fall through to regular link
              }

              // internal link
              const url = slugify(fp.trim() + anchor)
              return {
                type: 'link',
                url,
                children: [{
                  type: 'text',
                  value: alias ?? fp
                }]
                // internal link
                const url = fp + anchor
                return {
                  type: "link",
                  url,
                  children: [
                    {
                      type: "text",
                      value: alias ?? fp,
                    },
                  ],
                }
              },
            ])
          }

          if (opts.highlight) {
            replacements.push([
              highlightRegex,
              (_value: string, ...capture: string[]) => {
                const [inner] = capture
                return {
                  type: "html",
                  value: `<span class="text-highlight">${inner}</span>`,
                }
              },
            ])
          }

          if (opts.comments) {
            replacements.push([
              commentRegex,
              (_value: string, ..._capture: string[]) => {
                return {
                  type: "text",
                  value: "",
                }
              },
            ])
          }

          if (opts.parseTags) {
            replacements.push([
              tagRegex,
              (_value: string, tag: string) => {
                // Check if the tag only includes numbers
                if (/^\d+$/.test(tag)) {
                  return false
                }

                tag = slugTag(tag)
                if (file.data.frontmatter && !file.data.frontmatter.tags.includes(tag)) {
                  file.data.frontmatter.tags.push(tag)
                }

                return {
                  type: "link",
                  url: base + `/tags/${tag}`,
                  data: {
                    hProperties: {
                      className: ["tag-link"],
                    },
                  },
                  children: [
                    {
                      type: "text",
                      value: `#${tag}`,
                    },
                  ],
                }
              },
            ])
          }

          if (opts.enableInHtmlEmbed) {
            visit(tree, "html", (node: Html) => {
              for (const [regex, replace] of replacements) {
                if (typeof replace === "string") {
                  node.value = node.value.replace(regex, replace)
                } else {
                  node.value = node.value.replaceAll(regex, (substring: string, ...args) => {
                    const replaceValue = replace(substring, ...args)
                    if (typeof replaceValue === "string") {
                      return replaceValue
                    } else if (Array.isArray(replaceValue)) {
                      return replaceValue.map(mdastToHtml).join("")
                    } else if (typeof replaceValue === "object" && replaceValue !== null) {
                      return mdastToHtml(replaceValue)
                    } else {
                      return substring
                    }
                  })
                }
              }
            })
          }

          mdastFindReplace(tree, replacements)
        }
        )
      }

      if (opts.highlight) {
        plugins.push(() => {
          return (tree: Root, _file) => {
            findAndReplace(tree, highlightRegex, (_value: string, ...capture: string[]) => {
              const [inner] = capture
              return {
                type: 'html',
                value: `<span class="text-highlight">${inner}</span>`
              }
            })
          }
        })
      }
      
      if (opts.comments) {
        plugins.push(() => {
          return (tree: Root, _file) => {
            findAndReplace(tree, commentRegex, (_value: string, ..._capture: string[]) => {
              return {
                type: 'text',
                value: ''
              }
            })
          }
        })
      }
      })

      if (opts.callouts) {
        plugins.push(() => {
@@ -238,41 +363,51 @@
              }

              const text = firstChild.children[0].value
              const restChildren = firstChild.children.splice(1)
              const restChildren = firstChild.children.slice(1)
              const [firstLine, ...remainingLines] = text.split("\n")
              const remainingText = remainingLines.join("\n")

              const match = firstLine.match(calloutRegex)
              if (match && match.input) {
                const [calloutDirective, typeString, collapseChar] = match
                const calloutType = typeString.toLowerCase() as keyof typeof callouts
                const calloutType = canonicalizeCallout(
                  typeString.toLowerCase() as keyof typeof calloutMapping,
                )
                const collapse = collapseChar === "+" || collapseChar === "-"
                const defaultState = collapseChar === "-" ? "collapsed" : "expanded"
                const title = match.input.slice(calloutDirective.length).trim() || capitalize(calloutType)
                const titleContent =
                  match.input.slice(calloutDirective.length).trim() || capitalize(calloutType)
                const titleNode: Paragraph = {
                  type: "paragraph",
                  children: [{ type: "text", value: titleContent + " " }, ...restChildren],
                }
                const title = mdastToHtml(titleNode)

                const toggleIcon = `<svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="fold">
                  <polyline points="6 9 12 15 18 9"></polyline>
                </svg>`

                const titleNode: HTML = {
                const titleHtml: Html = {
                  type: "html",
                  value: `<div 
                  value: `<div
                  class="callout-title"
                >
                  <div class="callout-icon">${callouts[canonicalizeCallout(calloutType)]}</div>
                  <div class="callout-icon">${callouts[calloutType]}</div>
                  <div class="callout-title-inner">${title}</div>
                  ${collapse ? toggleIcon : ""}
                </div>`
                </div>`,
                }

                const blockquoteContent: (BlockContent | DefinitionContent)[] = [titleNode]
                const blockquoteContent: (BlockContent | DefinitionContent)[] = [titleHtml]
                if (remainingText.length > 0) {
                  blockquoteContent.push({
                    type: 'paragraph',
                    children: [{
                      type: 'text',
                      value: remainingText,
                    }, ...restChildren]
                    type: "paragraph",
                    children: [
                      {
                        type: "text",
                        value: remainingText,
                      },
                    ],
                  })
                }

@@ -283,10 +418,12 @@
                node.data = {
                  hProperties: {
                    ...(node.data?.hProperties ?? {}),
                    className: `callout ${collapse ? "is-collapsible" : ""} ${defaultState === "collapsed" ? "is-collapsed" : ""}`,
                    className: `callout ${collapse ? "is-collapsible" : ""} ${
                      defaultState === "collapsed" ? "is-collapsed" : ""
                    }`,
                    "data-callout": calloutType,
                    "data-callout-fold": collapse,
                  }
                  },
                }
              }
            })
@@ -297,12 +434,12 @@
      if (opts.mermaid) {
        plugins.push(() => {
          return (tree: Root, _file) => {
            visit(tree, 'code', (node: Code) => {
              if (node.lang === 'mermaid') {
            visit(tree, "code", (node: Code) => {
              if (node.lang === "mermaid") {
                node.data = {
                  hProperties: {
                    className: 'mermaid'
                  }
                    className: ["mermaid"],
                  },
                }
              }
            })
@@ -313,7 +450,62 @@
      return plugins
    },
    htmlPlugins() {
      return [rehypeRaw]
      const plugins: PluggableList = [rehypeRaw]

      if (opts.parseBlockReferences) {
        plugins.push(() => {
          const inlineTagTypes = new Set(["p", "li"])
          const blockTagTypes = new Set(["blockquote"])
          return (tree: HtmlRoot, file) => {
            file.data.blocks = {}

            visit(tree, "element", (node, index, parent) => {
              if (blockTagTypes.has(node.tagName)) {
                const nextChild = parent?.children.at(index! + 2) as Element
                if (nextChild && nextChild.tagName === "p") {
                  const text = nextChild.children.at(0) as Literal
                  if (text && text.value && text.type === "text") {
                    const matches = text.value.match(blockReferenceRegex)
                    if (matches && matches.length >= 1) {
                      parent!.children.splice(index! + 2, 1)
                      const block = matches[0].slice(1)

                      if (!Object.keys(file.data.blocks!).includes(block)) {
                        node.properties = {
                          ...node.properties,
                          id: block,
                        }
                        file.data.blocks![block] = node
                      }
                    }
                  }
                }
              } else if (inlineTagTypes.has(node.tagName)) {
                const last = node.children.at(-1) as Literal
                if (last && last.value && typeof last.value === "string") {
                  const matches = last.value.match(blockReferenceRegex)
                  if (matches && matches.length >= 1) {
                    last.value = last.value.slice(0, -matches[0].length)
                    const block = matches[0].slice(1)

                    if (!Object.keys(file.data.blocks!).includes(block)) {
                      node.properties = {
                        ...node.properties,
                        id: block,
                      }
                      file.data.blocks![block] = node
                    }
                  }
                }
              }
            })

            file.data.htmlAst = tree
          }
        })
      }

      return plugins
    },
    externalResources() {
      const js: JSResource[] = []
@@ -321,8 +513,8 @@
      if (opts.callouts) {
        js.push({
          script: calloutScript,
          loadTime: 'afterDOMReady',
          contentType: 'inline'
          loadTime: "afterDOMReady",
          contentType: "inline",
        })
      }

@@ -330,15 +522,32 @@
        js.push({
          script: `
          import mermaid from 'https://cdn.jsdelivr.net/npm/mermaid/dist/mermaid.esm.min.mjs';
          mermaid.initialize({ startOnLoad: true });
          const darkMode = document.documentElement.getAttribute('saved-theme') === 'dark'
          mermaid.initialize({
            startOnLoad: false,
            securityLevel: 'loose',
            theme: darkMode ? 'dark' : 'default'
          });
          document.addEventListener('nav', async () => {
            await mermaid.run({
              querySelector: '.mermaid'
            })
          });
          `,
          loadTime: 'afterDOMReady',
          moduleType: 'module',
          contentType: 'inline'
          loadTime: "afterDOMReady",
          moduleType: "module",
          contentType: "inline",
        })
      }

      return { js }
    }
    },
  }
}

declare module "vfile" {
  interface DataMap {
    blocks: Record<string, Element>
    htmlAst: HtmlRoot
  }
}