Jacky Zhao
2024-01-24 fa6c02d3213dfd4e6da8e78bd3a2e7004555fd01
quartz/plugins/transformers/ofm.ts
@@ -1,11 +1,10 @@
import { PluggableList } from "unified"
import { QuartzTransformerPlugin } from "../types"
import { Root, HTML, BlockContent, DefinitionContent, Code, Paragraph } from "mdast"
import { Root, Html, BlockContent, DefinitionContent, Paragraph, Code } from "mdast"
import { Element, Literal, Root as HtmlRoot } from "hast"
import { Replace, findAndReplace as mdastFindReplace } from "mdast-util-find-and-replace"
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 { SKIP, visit } from "unist-util-visit"
import path from "path"
import { JSResource } from "../../util/resources"
// @ts-ignore
@@ -15,6 +14,7 @@
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
@@ -23,8 +23,11 @@
  callouts: boolean
  mermaid: boolean
  parseTags: boolean
  parseArrows: boolean
  parseBlockReferences: boolean
  enableInHtmlEmbed: boolean
  enableYouTubeEmbed: boolean
  enableVideoEmbed: boolean
}
const defaultOptions: Options = {
@@ -34,8 +37,11 @@
  callouts: true,
  mermaid: true,
  parseTags: true,
  parseArrows: true,
  parseBlockReferences: true,
  enableInHtmlEmbed: false,
  enableYouTubeEmbed: true,
  enableVideoEmbed: true,
}
const icons = {
@@ -105,6 +111,10 @@
  return calloutMapping[callout] ?? "note"
}
export const externalLinkRegex = /^https?:\/\//i
export const arrowRegex = new RegExp(/-{1,2}>/, "g")
// !?               -> optional embedding
// \[\[             -> open brace
// ([^\[\]\|\#]+)   -> one or more non-special characters ([,],|, or #) (name)
@@ -115,7 +125,7 @@
  "g",
)
const highlightRegex = new RegExp(/==([^=]+)==/, "g")
const commentRegex = new RegExp(/%%(.+)%%/, "g")
const commentRegex = new RegExp(/%%[\s\S]*?%%/, "g")
// from https://github.com/escwxyz/remark-obsidian-callout/blob/main/src/index.ts
const calloutRegex = new RegExp(/^\[\!(\w+)\]([+-]?)/)
const calloutLineRegex = new RegExp(/^> *\[\!\w+\][+-]?.*$/, "gm")
@@ -123,8 +133,10 @@
// #(...)               -> capturing group, tag itself must start with #
// (?:[-_\p{L}\d\p{Z}])+       -> non-capturing group, non-empty string of (Unicode-aware) alpha-numeric characters and symbols, hyphens and/or underscores
// (?:\/[-_\p{L}\d\p{Z}]+)*)   -> non-capturing group, matches an arbitrary number of tag strings separated by "/"
const tagRegex = new RegExp(/(?:^| )#((?:[-_\p{L}\d\p{Z}])+(?:\/[-_\p{L}\d\p{Z}]+)*)/, "gu")
const blockReferenceRegex = new RegExp(/\^([A-Za-z0-9]+)$/, "g")
const tagRegex = new RegExp(/(?:^| )#((?:[-_\p{L}\p{Emoji}\d])+(?:\/[-_\p{L}\p{Emoji}\d]+)*)/, "gu")
const blockReferenceRegex = new RegExp(/\^([-_A-Za-z0-9]+)$/, "g")
const ytLinkRegex = /^.*(youtu.be\/|v\/|u\/\w\/|embed\/|watch\?v=|\&v=)([^#\&\?]*).*/
const videoExtensionRegex = new RegExp(/\.(mp4|webm|ogg|avi|mov|flv|wmv|mkv|mpg|mpeg|3gp|m4v)$/)
export const ObsidianFlavoredMarkdown: QuartzTransformerPlugin<Partial<Options> | undefined> = (
  userOpts,
@@ -136,40 +148,25 @@
    return toHtml(hast, { allowDangerousHtml: true })
  }
  const findAndReplace = opts.enableInHtmlEmbed
    ? (tree: Root, regex: RegExp, replace?: Replace | null | undefined) => {
        if (replace) {
          visit(tree, "html", (node: HTML) => {
            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, regex, replace)
      }
    : mdastFindReplace
  return {
    name: "ObsidianFlavoredMarkdown",
    textTransform(_ctx, src) {
      // do comments at text level
      if (opts.comments) {
        if (src instanceof Buffer) {
          src = src.toString()
        }
        src = src.replace(commentRegex, "")
      }
      // pre-transform blockquotes
      if (opts.callouts) {
        src = src.toString()
        src = src.replaceAll(calloutLineRegex, (value) => {
        if (src instanceof Buffer) {
          src = src.toString()
        }
        src = src.replace(calloutLineRegex, (value) => {
          // force newline after title of callout
          return value + "\n> "
        })
@@ -177,15 +174,24 @@
      // pre-transform wikilinks (fix anchors to things that may contain illegal syntax e.g. codeblocks, latex)
      if (opts.wikilinks) {
        src = src.toString()
        src = src.replaceAll(wikilinkRegex, (value, ...capture) => {
          const [rawFp, rawHeader, rawAlias] = capture
        if (src instanceof Buffer) {
          src = src.toString()
        }
        src = src.replace(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}]]`
        })
      }
@@ -194,103 +200,184 @@
    },
    markdownPlugins() {
      const plugins: PluggableList = []
      if (opts.wikilinks) {
        plugins.push(() => {
          return (tree: Root, _file) => {
            findAndReplace(tree, 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,
      // 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", ".webp"].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 {
                    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="${url}"></iframe>`,
                  }
                } else if (ext === "") {
                  const block = anchor
                  return {
                    type: "html",
                    data: { hProperties: { transclude: true } },
                    value: `<blockquote class="transclude" data-url="${url}" data-block="${block}"><a href="${
                      url + anchor
                    }" class="transclude-inner">Transclude of ${url}${block}</a></blockquote>`,
                  }
                  // otherwise, fall through to regular link
                }
                // otherwise, fall through to regular link
              }
                // internal link
                const url = fp + 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">${inner}</span>`,
                }
              },
            ])
          }
          if (opts.parseArrows) {
            replacements.push([
              arrowRegex,
              (_value: string, ..._capture: string[]) => {
                return {
                  type: "html",
                  value: `<span>&rarr;</span>`,
                }
              },
            ])
          }
          if (opts.parseTags) {
            replacements.push([
              tagRegex,
              (_value: string, tag: string) => {
                // Check if the tag only includes numbers
                if (/^\d+$/.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.replace(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) {
      if (opts.enableVideoEmbed) {
        plugins.push(() => {
          return (tree: Root, _file) => {
            findAndReplace(tree, highlightRegex, (_value: string, ...capture: string[]) => {
              const [inner] = capture
              return {
                type: "html",
                value: `<span class="text-highlight">${inner}</span>`,
              }
            })
          }
        })
      }
            visit(tree, "image", (node, index, parent) => {
              if (parent && index != undefined && videoExtensionRegex.test(node.url)) {
                const newNode: Html = {
                  type: "html",
                  value: `<video controls src="${node.url}"></video>`,
                }
      if (opts.comments) {
        plugins.push(() => {
          return (tree: Root, _file) => {
            findAndReplace(tree, commentRegex, (_value: string, ..._capture: string[]) => {
              return {
                type: "text",
                value: "",
                parent.children.splice(index, 1, newNode)
                return SKIP
              }
            })
          }
@@ -312,7 +399,7 @@
              }
              const text = firstChild.children[0].value
              const restChildren = firstChild.children.slice(1)
              const restOfTitle = firstChild.children.slice(1)
              const [firstLine, ...remainingLines] = text.split("\n")
              const remainingText = remainingLines.join("\n")
@@ -328,7 +415,10 @@
                  match.input.slice(calloutDirective.length).trim() || capitalize(calloutType)
                const titleNode: Paragraph = {
                  type: "paragraph",
                  children: [{ type: "text", value: titleContent + " " }, ...restChildren],
                  children:
                    restOfTitle.length === 0
                      ? [{ type: "text", value: titleContent + " " }]
                      : restOfTitle,
                }
                const title = mdastToHtml(titleNode)
@@ -336,7 +426,7 @@
                  <polyline points="6 9 12 15 18 9"></polyline>
                </svg>`
                const titleHtml: HTML = {
                const titleHtml: Html = {
                  type: "html",
                  value: `<div
                  class="callout-title"
@@ -396,50 +486,16 @@
        })
      }
      if (opts.parseTags) {
        plugins.push(() => {
          return (tree: Root, file) => {
            const base = pathToRoot(file.data.slug!)
            findAndReplace(tree, 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: `#${tag}`,
                  },
                ],
              }
            })
          }
        })
      }
      return plugins
    },
    htmlPlugins() {
      const plugins = [rehypeRaw]
      const plugins: PluggableList = [rehypeRaw]
      if (opts.parseBlockReferences) {
        plugins.push(() => {
          const inlineTagTypes = new Set(["p", "li"])
          const blockTagTypes = new Set(["blockquote"])
          return (tree, file) => {
          return (tree: HtmlRoot, file) => {
            file.data.blocks = {}
            visit(tree, "element", (node, index, parent) => {
@@ -488,6 +544,30 @@
        })
      }
      if (opts.enableYouTubeEmbed) {
        plugins.push(() => {
          return (tree: HtmlRoot) => {
            visit(tree, "element", (node) => {
              if (node.tagName === "img" && typeof node.properties.src === "string") {
                const match = node.properties.src.match(ytLinkRegex)
                const videoId = match && match[2].length == 11 ? match[2] : null
                if (videoId) {
                  node.tagName = "iframe"
                  node.properties = {
                    class: "external-embed",
                    allow: "fullscreen",
                    frameborder: 0,
                    width: "600px",
                    height: "350px",
                    src: `https://www.youtube.com/embed/${videoId}`,
                  }
                }
              }
            })
          }
        })
      }
      return plugins
    },
    externalResources() {