Jacky Zhao
2023-12-18 78b33fc2fb2d2b30e0cdf6fbd50c74efceac1c56
quartz/plugins/transformers/ofm.ts
@@ -1,7 +1,7 @@
import { PluggableList } from "unified"
import { QuartzTransformerPlugin } from "../types"
import { Root, HTML, BlockContent, DefinitionContent, Code, Paragraph } from "mdast"
import { Element, Literal } from 'hast'
import { Element, Literal, Root as HtmlRoot } from "hast"
import { Replace, findAndReplace as mdastFindReplace } from "mdast-util-find-and-replace"
import { slug as slugAnchor } from "github-slugger"
import rehypeRaw from "rehype-raw"
@@ -14,6 +14,7 @@
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"
export interface Options {
  comments: boolean
@@ -104,16 +105,15 @@
  return calloutMapping[callout] ?? "note"
}
const capitalize = (s: string): string => {
  return s.substring(0, 1).toUpperCase() + s.substring(1)
}
// !?               -> 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")
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
@@ -121,8 +121,8 @@
const calloutLineRegex = new RegExp(/^> *\[\!\w+\][+-]?.*$/, "gm")
// (?:^| )              -> non-capturing group, tag should start be separated by a space or be the start of the line
// #(...)               -> capturing group, tag itself must start with #
// (?:[-_\p{L}])+       -> non-capturing group, non-empty string of (Unicode-aware) alpha-numeric characters, hyphens and/or underscores
// (?:\/[-_\p{L}]+)*)   -> non-capturing group, matches an arbitrary number of tag strings separated by "/"
// (?:[-_\p{L}\d\p{Z}])+       -> non-capturing group, non-empty string of (Unicode-aware) alpha-numeric characters and symbols, hyphens and/or underscores
// (?:\/[-_\p{L}\d\p{Z}]+)*)   -> non-capturing group, matches an arbitrary number of tag strings separated by "/"
const tagRegex = new RegExp(/(?:^| )#((?:[-_\p{L}\d])+(?:\/[-_\p{L}\d]+)*)/, "gu")
const blockReferenceRegex = new RegExp(/\^([A-Za-z0-9]+)$/, "g")
@@ -135,31 +135,32 @@
    const hast = toHast(ast, { allowDangerousHtml: true })!
    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
              }
            })
          }
        })
      }
        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(tree, regex, replace)
      }
    : mdastFindReplace
  return {
@@ -180,8 +181,9 @@
        src = src.replaceAll(wikilinkRegex, (value, ...capture) => {
          const [rawFp, rawHeader, rawAlias] = capture
          const fp = rawFp ?? ""
          const anchor = rawHeader?.trim().slice(1)
          const displayAnchor = anchor ? `#${slugAnchor(anchor)}` : ""
          const anchor = rawHeader?.trim().replace(/^#+/, "")
          const blockRef = Boolean(anchor?.startsWith("^")) ? "^" : ""
          const displayAnchor = anchor ? `#${blockRef}${slugAnchor(anchor)}` : ""
          const displayAlias = rawAlias ?? rawHeader?.replace("#", "|") ?? ""
          const embedDisplay = value.startsWith("!") ? "!" : ""
          return `${embedDisplay}[[${fp}${displayAnchor}${displayAlias}]]`
@@ -238,8 +240,16 @@
                    value: `<iframe src="${url}"></iframe>`,
                  }
                } else if (ext === "") {
                  // TODO: note embed
                  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
              }
@@ -357,8 +367,9 @@
                node.data = {
                  hProperties: {
                    ...(node.data?.hProperties ?? {}),
                    className: `callout ${collapse ? "is-collapsible" : ""} ${defaultState === "collapsed" ? "is-collapsed" : ""
                      }`,
                    className: `callout ${collapse ? "is-collapsible" : ""} ${
                      defaultState === "collapsed" ? "is-collapsed" : ""
                    }`,
                    "data-callout": calloutType,
                    "data-callout-fold": collapse,
                  },
@@ -390,6 +401,11 @@
          return (tree: Root, file) => {
            const base = pathToRoot(file.data.slug!)
            findAndReplace(tree, tagRegex, (_value: string, tag: string) => {
              // Check if the tag only includes numbers
              if (/^\d+$/.test(tag)) {
                return false
              }
              tag = slugTag(tag)
              if (file.data.frontmatter && !file.data.frontmatter.tags.includes(tag)) {
                file.data.frontmatter.tags.push(tag)
@@ -421,26 +437,53 @@
      if (opts.parseBlockReferences) {
        plugins.push(() => {
          const inlineTagTypes = new Set(["p", "li"])
          const blockTagTypes = new Set(["blockquote"])
          return (tree, file) => {
            file.data.blocks = {}
            const validTagTypes = new Set(["blockquote", "p", "li"])
            visit(tree, "element", (node, _index, _parent) => {
              if (validTagTypes.has(node.tagName)) {
            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.value && typeof last.value === 'string') {
                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)
                    node.properties = {
                      ...node.properties,
                      id: block
                    if (!Object.keys(file.data.blocks!).includes(block)) {
                      node.properties = {
                        ...node.properties,
                        id: block,
                      }
                      file.data.blocks![block] = node
                    }
                    file.data.blocks![block] = node
                  }
                }
              }
            })
            file.data.htmlAst = tree
          }
        })
      }
@@ -488,6 +531,6 @@
declare module "vfile" {
  interface DataMap {
    blocks: Record<string, Element>
    htmlAst: HtmlRoot
  }
}